Selection and Placement of Sensors for Electric Motors: A Review and Preliminary Investigation

Abstract

This review explores sensor selection and placement strategies for electric motor monitoring in industrial settings. A wide range of sensor types including temperature, vibration, current, and position sensors—are evaluated in terms of their technical features and application constraints. Preliminary experimental data on vibration sensors highlight how signal amplitude varies with sensor placement, reinforcing the importance of correct positioning. However, this study stops short of applying AI/ML techniques to optimize placement. Accordingly, this paper serves as a foundational step toward developing intelligent sensor deployment frameworks. Future work will build on this review by integrating supervised learning, dimensionality reduction, and reinforcement learning techniques to automate sensor placement and improve condition monitoring in electric motors.

1. Introduction

Electric motors are fundamental components across a wide range of applications, from industrial automation and electric vehicles to offshore platforms and consumer electronics. In critical environments such as offshore oil and gas facilities, where reliability, safety, and operational continuity are paramount, effective monitoring of electric motor systems is essential. Sensors play a key role in ensuring the performance, efficiency, and longevity of these motors by providing real-time data for fault detection, predictive maintenance, and adaptive control.

To achieve these goals, a variety of sensors are integrated into electric motor systems, measuring parameters such as temperature, vibration, current, and rotational speed. These sensors enable timely interventions that prevent equipment failures and reduce downtime. Strategic selection and accurate placement of sensors within the motor system are crucial to ensure reliable data acquisition and optimized motor performance. Proper sensor deployment not only enhances operational efficiency but also minimizes maintenance costs and extends system lifespan.

As highlighted in [1], “A sensor in a system that continuously monitors, detects, and provides real-time data to enhance performance, safety, and automation, while enabling data collection, system protection, predictive maintenance, and safety compliance, ensuring efficiency and preventing failures.” This underscores the vital, multifaceted role of sensors in electric motor systems.

This paper presents a comprehensive review of sensor technologies and deployment strategies for electric motors, with particular attention given to sensor types, selection criteria, placement considerations, and environmental factors. The review also explores recent advancements in sensor technology, including wireless sensing, self-powered devices, and integration with artificial intelligence (AI) and machine learning (ML) for intelligent monitoring.

This work serves as a foundation for ongoing research aimed at developing AI-driven sensor placement frameworks for complex environments such as offshore platforms. While preliminary experimental results on vibration sensor placement are included, future publications will extend the experimental scope to cover additional sensor types (e.g., current and temperature sensors) and advanced data-driven diagnostic systems.

This review aims to investigate methodologies, advancements, and best practices in the deployment of electric motor sensor systems, exploring how cutting-edge technologies can contribute to more efficient, reliable, and cost-effective motor operation [2].

2. Sensor Overview and Placement Summary

Electric motors require a variety of strategically placed sensors to monitor critical operational parameters. Correct sensor type and placement directly influence the system’s ability to detect faults early, optimize performance, and enable predictive maintenance. Note that the sensor placements shown in Figure 1 are conceptual and included to support the review discussion. They do not represent the actual experimental setup, which is detailed later in Section 9.

Table 1. Summary of sensor types, placement zones, and monitoring objectives [3].

Sensor Type	Typical Placement Zone	Monitoring Purpose/Operational Parameter
Vibration Sensor (Accelerometer)	Bearing housings, motor casing	Detects imbalance, misalignment, bearing wear, and mechanical looseness
Temperature Sensor (RTD, Thermocouple)	Embedded in stator windings or motor housing	Monitors thermal behavior to prevent overheating and assess thermal loading
Current Sensor (Rogowski coil, Hall-effect)	Supply lines, terminal box	Measures current flow to detect overload, unbalanced phase currents, and harmonics
Position/Speed Sensor (Encoder, Resolver)	Attached to rotor shaft or motor back end	Provides rotor position and speed feedback for closed-loop control systems

summarizes the major sensor types, typical placement zones within the motor, and their operational functions or ranges.

AI and Machine Learning Approaches to Sensor Placement

The increasing complexity of electric motor systems and the need for predictive maintenance have driven the application of artificial intelligence (AI) and machine learning (ML) techniques to sensor placement strategies. These approaches aim to enhance monitoring accuracy, reduce data redundancy, and optimize sensor utilization by learning from historical or simulated data.

Several studies have demonstrated the use of ML algorithms for optimizing the number and location of sensors, particularly for vibration-based fault detection. For example, clustering methods such as k-means have been used to identify sensor positions that maximize sensitivity to specific fault signatures, such as imbalance or bearing defects. In [4], the authors applied a genetic algorithm to optimize accelerometer locations based on fault classification performance using vibration signals.

Deep learning models, including convolutional neural networks (CNNs), have also been employed to automate the feature extraction process using time–frequency representations of sensor data, enabling more intelligent placement strategies. In [5], CNNs were used to evaluate sensor importance based on classification accuracy, guiding optimal placement on induction motors.

Other approaches use dimensionality reduction techniques such as Principal Component Analysis (PCA) or mutual information analysis to identify redundant sensor channels and propose minimal yet informative sensor configurations [6]. These methods help reduce hardware costs, while maintaining fault diagnostic reliability.

Moreover, reinforcement learning and adaptive placement strategies have been proposed in dynamic environments, where sensor positions are adjusted in real-time based on changing operating conditions or fault progression [7].

These studies underscore the potential of AI/ML to guide sensor placement decisions, shifting the process from experience-based heuristics to data-driven optimization. While most existing works focused on vibration sensors, the same principles are being extended to current, temperature, and magnetic field sensing for holistic motor health monitoring.

This review lays the groundwork for future research that will implement and compare these techniques in the context of electric machines operating in industrial and offshore environments.

3. Sensors for Electric Motors

Sensors are essential for optimizing electric motor performance by monitoring temperature, vibration, current, and position in real time. Temperature sensors prevent overheating, vibration sensors detect mechanical problems, and current sensors identify electrical faults, while position sensors ensure precise motion control. These sensors improve efficiency, reduce energy consumption, and extend the useful life of the motor [8]. Their integration in industries such as automation, electric vehicles, and renewable energy allows predictive maintenance, minimizing downtime and operational costs. With advancements in IoT-enabled smart sensors, remote diagnostics and real-time optimization can further improve reliability and safety, driving innovation in motor-driven applications. In this research, various sensor types are carefully considered [9].

3.1. Temperature Sensors

Temperature sensors in electrical motors monitor and protect against overheating, ensuring efficiency and safety. They help prevent failures, optimize performance, enable predictive maintenance, improve safety, and support adaptive control in variable-speed drives (VSDs) [10].

Common types include Resistance Temperature Detectors (RTDs), thermistors, thermocouples, and embedded sensors. Each type is chosen based on accuracy, response time, operating conditions, and application requirements [11]. These temperature sensors include RTDs, thermocouples (TCs), thermistors, infrared (IR) sensors, and embedded temperature sensors (ETS), each of which is explained below.

3.1.1. Resistance Temperature Detectors (RTDs)

RTDs are sensors that function based on the idea that their electrical resistance changes with temperature. As the temperature increases, the resistance of the RTDs increases proportionally, enabling an accurate and reliable temperature reading. RTDs measure temperature by changing electrical resistance (typically using platinum, for example, PT100, PT1000) and are highly accurate and stable over time, making them ideal for critical applications that require precision [11].

3.1.2. Thermocouples (TCs)

TCs are temperature sensors made of two different metals that generate a voltage proportional to the temperature differences. They are widely used in high-temperature applications due to their durability, fast response times, and ability to operate in extreme environments. With a broad temperature range and reliable performance, thermocouples are ideal for industrial, scientific, and automotive applications. Common types include K, J, T, and E, each of which offers specific characteristics suited for different operating conditions [12].

Thermocouples are widely used due to their low cost, simplicity, quick response, broad temperature range, and durability. They operate without an external power source but have drawbacks such as limited accuracy, non-linearity, and low voltage output, requiring signal amplification and linearization. A thermocouple consists of two different metal wires joined at one end (measuring junction) and separated at the other (reference end). The thermocouple operates based on the Seebeck effect, discovered by Thomas Johann Seebeck (1770–1831). The Seebeck coefficient for materials A and B, also referred to as the thermoelectric power, is defined by Equation (1) [13].

$$S_{AB}=lim\Delta {\epsilon }_{AB}/\Delta T=d{\epsilon }_{AB}/dT$$

(1)

where T is the temperature and ${\epsilon }_{AB}$ is the Seebeck EMF.

Research indicates that, following Jean Peltier (Jean Charles Athanase Peltier (1785–1845)), heat is either emitted or absorbed when an electric current passes through the junction of two dissimilar conductors. The direction of current flow influences whether heat is released or absorbed, and this process is reversible, unlike $I^{2}R$ electrical dissipation, which is irreversible and does not alter its sign when the current flow is reversed. The rate of thermal exchange at a junction is directly proportional to the intensity of current traversing the junction, as delineated by Equation (2).

$${\Pi }_{AB}I=dQ/dt$$

(2)

where $dQ/dt$ is the rate of change in the heat content (not the electric current), I is the electric current, and $\Pi AB$ is the Peltier coefficient for the two materials, A and B.

A voltage is generated at the reference end on the basis of the temperature difference between the two ends, making it a temperature-to-voltage transducer. When the end of the copper wire is heated, electrons move toward the cooler end, creating a temperature gradient along the wire. This movement of electrons converts heat into electrical energy [14] (Figure 2).

The temperature vs. voltage relationship is given by Equation (3).

$$Emf={\int }_{T_1}^{T_2}{S}_{12}·dT={\int }_{T_1}^{T_2}\left(S_1-S_2\right)·dT$$

(3)

In this context, $Emf$ refers to the electromotive force, or voltage, generated by the thermocouple at the tail end. $T_1$ and $T_2$ denote the temperatures at the reference and measurement ends, respectively. The term S12 refers to the Seebeck coefficient of the thermocouple, while $S_1$ and $S_2$ denote the Seebeck coefficients of the individual thermoelements. It is important to note that the Seebeck coefficient is contingent on the material used for the thermoelements. The Seebeck coefficients exhibit a dependence on temperature.

3.1.3. Thermistors

Thermistors are highly sensitive temperature sensors that change their electrical resistance in response to temperature variations. They come in two main types; negative temperature coefficient (NTC) thermistors, where the resistance decreases as the temperature rises, and positive temperature coefficient (PTC) thermistors, where the resistance increases with temperature. Due to their high precision and fast response, thermistors are commonly used in medical devices, household appliances, automotive systems, and industrial applications for temperature monitoring and control. Their small size, affordability, and ability to detect even minor temperature changes make them an excellent choice for applications that require precision and reliability [15].

According to [16], these types of temperature sensors have a fast response time, which is ideal for thermal protection. Thermistors are made from semiconductor materials and are widely used for temperature measurement and control in electric motors, being placed in motor windings and bearings for overheating protection. In addition, Ref. [11] states that thermistors are ideal for precise temperature sensing due to their high sensitivity. In Positive Temperature Coefficient (PTC) Thermistors, resistance increases as temperature increases. They are often used as self-resetting overcurrent protectors and motor thermal cutoffs.

Thermistors have several limitations [17], their parameters can vary even within the same production batch, their thermoelectric properties can change over time during operation, and they exhibit a nonlinear relationship between temperature and electrical resistance, which is described by the Steinhart–Hart Equation (4) [18].

$${\displaystyle \frac{1}{T}}=\mathrm{A}+BlnR+C{ln}^{3}R$$

(4)

where T is the absolute temperature of the semiconductor resistor; coefficients $A,B,C$ depend on the parameters of the thermistor and the range of its operating temperature, and R is the resistance of the thermistor.

3.1.4. Infrared (IR) Sensors

Infrared (IR) sensors are non-contact temperature sensors that detect heat emitted from an object as infrared radiation, allowing temperature measurement without direct physical contact. This makes them particularly useful for applications where traditional contact-based sensors are impractical or inefficient, such as monitoring the surface temperature of electric motors, detecting overheating in electrical equipment, and ensuring safety in high-temperature industrial environments. IR sensors are also widely used in medical thermometers, HVAC systems, food processing, and automotive diagnostics. Their ability to provide fast, accurate and reliable temperature readings from a distance makes them essential for preventive maintenance, quality control, and thermal imaging applications [19]. They are useful for external monitoring of motor housings and bearings.

3.1.5. Embedded Temperature Sensors (ETS)

Embedded temperature sensors (ETS) are installed inside critical motor components, such as bearings and windings, to provide direct and real-time monitoring of internal temperatures. By continuously monitoring temperature fluctuations, these sensors play a crucial role in preventing overheating, optimizing motor performance, and extending the useful life of industrial motors. They are widely used in applications where precise thermal management is essential, such as manufacturing, power generation, and heavy machinery. Embedded sensors help detect potential failures early, allowing for predictive maintenance and reducing the risk of costly downtime. In addition, they contribute to energy efficiency by ensuring that motors operate within optimal temperature ranges, ultimately improving reliability, safety, and overall system performance [20,21].

According to [22], temperature sensors vary in their characteristics and are suitable for different applications (

Table 2. Comparison of temperature sensor characteristics [22].

Characteristic	Thermistor	RTD	Thermocouples	IR Sensor	ETS
Temp Range (°C)	−55 to 125	−200 to 850	−600 to 2000	−50 to 1000	−40 to 250
Linearity	Exponential	Fairly linear	Fairly linear	Linear	Linear
Sensitivity	High	Low	Medium	Medium	High
Response Time	Fast	Moderate	Fast	Very fast	Fast
Contact Required	Yes	Yes	Yes	No	Yes
Stability	Low	High	Moderate	Moderate	High

). Thermistors offer high sensitivity and fast response, but have a limited temperature range and low stability. RTDs provide high stability and accuracy over a wider range, although they are slower and more costly, especially wire-wound types. Thermocouples cover the broadest temperature range and do not require power but have medium sensitivity and cost. Semiconductor sensors boast the best linearity and highest sensitivity, which is ideal for narrow temperature ranges, although they respond slowly and have medium stability.

3.2. Vibration Sensors

Vibration sensors are used in various motors (synchronous, asynchronous, and DC) to detect mechanical issues such as misalignment, imbalance, bearing wear, and structural faults. These sensors measure vibrations and convert them into electrical signals that can be analyzed to assess motor health [23].

A sensor detects these vibrations and converts them into an electrical signal, which is further processed using filtering and amplification to eliminate noise and improve accuracy [24]. According to [25], advanced vibration sensors use Fast Fourier Transform (FFT) analysis to break down vibration frequencies and identify fault patterns.

Table 3. Overview of primary benefits and drawbacks of frequency domain analysis [32].

Frequency Domain Analysis	Benefits	Drawbacks
Fast Fourier Transform	Simple to execute.	Failure detection is inefficient when the machine’s normal operating frequency and amplitude signals are not known.
Cepstrum Analysis	Simple method, effective for identifying harmonics, sidebands, or echoes.	Susceptible to noise in the vibration signals.
Envelope Analysis	Timely identification of bearing issues.	Identifying the optimal frequency band for this approach represents a hurdle.
Power Spectral Density	A clearly defined frequency domain representation of the signal allows for the identification of specific frequency components linked to faults or irregularities in rotating machinery.	Graphical analysis of the signal requires the support of a specialized service.

summarizes the advantages and disadvantages of frequency domain analysis. The Fast Fourier Transform (FFT) is a fast algorithm used to convert time-domain signals into the frequency domain, aiding in vibration analysis of rotating machinery by identifying characteristic frequencies associated with faults such as imbalance or misalignment. Cepstrum analysis, which involves taking the inverse Fourier transform of the logarithmic spectrum, is effective in detecting periodic structures and low-frequency fault indicators, such as bearing or gear problems, though it can be sensitive to noise. Envelope analysis isolates low-frequency signals from noise, which is particularly useful for detecting early-stage bearing faults, but its effectiveness depends on selecting the right frequency band. Power Spectral Density (PSD) measures signal power across frequencies, offering clear insights into energy distribution and facilitating fault detection and condition monitoring, though it assumes signal stationarity. Each method has its strengths and limitations in frequency domain vibration analysis [26,27,28,29,30]. Cepstrum analysis is achieved by applying the inverse Fourier transform to the logarithmic spectrum of a signal, as defined in Equation (5) [31].

$$C\left(x\left(t\right)\right)={\mathcal{F}}^{-1}(log\left(X\left(\omega \right)\right))$$

(5)

where ${\mathcal{F}}^{-1}$ is the inverse of the Fourier transform, $x\left(t\right)$ is the signal in the time domain, and $X\left(\omega \right)$ is the signal in the frequency domain.

Furthermore, the processed signal is sent to a monitoring system (PLC, IoT device, or AI-based system).

The system analyzes the vibration data, detects abnormalities, and triggers alerts if necessary. Vibration sensors can be categorized into the following types.

• Accelerometers are among the most widely used vibration sensors for monitoring electric motors, as they measure the acceleration forces caused by vibrations and convert them into electrical signals for analysis. These sensors can detect issues such as misalignment and bearing wear, aiding in fault detection, predictive maintenance, and efficiency optimization. Piezoelectric and Micro-Electro-Mechanical System (MEMS) accelerometers are commonly used for their accuracy and real-time monitoring capabilities, helping reduce downtime and improve motor reliability [3,33].
• Velocity sensors are used to measure the speed of vibration movement, which helps in detecting mechanical faults in electric motors. They provide a balanced response between displacement and acceleration, making them ideal for monitoring the medium- to high-frequency vibrations commonly found in rotating machinery [34].
• Piezoelectric sensors detect vibrations and mechanical stress in electric motors by converting physical motion into an electrical signal. They are widely used for vibration monitoring, fault detection, and predictive maintenance. These sensors contain piezoelectric materials (e.g., quartz, ceramic) that generate an electrical charge when subjected to mechanical stress or vibrations [34].
  When the motor vibrates, the piezoelectric element deforms, producing an electrical signal proportional to the intensity of the vibration. The generated charge is converted into a voltage signal, amplified, and analyzed to detect motor health issues such as misalignment, imbalance, and bearing failures [35].
  Microelectromechanical system (MEMS) sensors are compact and highly sensitive devices used in electrical machines to monitor vibration, motion, and condition. These sensors integrate mechanical and electronic components on a microscale, making them cost-effective and efficient for real-time diagnostics [36]. MEMs detect excessive motor vibrations, indicating an imbalance, misalignment, or bearing wear. This helps in predictive maintenance by identifying faults early [37,38].
  MEMs track rotor movement and alignment for precision control in servo motors and robotics. They are also used in speed sensing for feedback control in motor drives. Some MEMS sensors measure temperature fluctuations in electrical machines to prevent overheating [39]. Pressure sensors help to monitor fluid cooling systems. In addition, MEMS sensors can be connected to monitoring systems wirelessly, enabling remote diagnostics and reducing maintenance costs [40].
  The mechanical monitoring and vibration analysis process for fault diagnosis is mainly divided into four steps: data acquisition, data transmission, signal processing, and fault detection. Figure 3 shows the main steps of vibration analysis for determining the failure of rotating machinery.

  Figure 3. The main steps for diagnosing faults in rotating machinery by analyzing vibrations [41].

  Figure 3. The main steps for diagnosing faults in rotating machinery by analyzing vibrations [41].
  Data acquisition can be carried out using many available vibration measurement devices. These devices can use different types of transducers to measure. Among the types of sensors used to capture vibration signals, accelerometers are the most commonly used [42]. Signal processing consists of manipulation, filtering, digitalization, and analysis of raw data, in order to extract relevant information. This is an important aspect of vibration analysis because it can extract patterns and insights from large quantities of vibration data that are otherwise difficult to interpret [43].
  Fault detection is the last step in the vibration analysis process. At this stage, the vibration signal is recorded in the time domain or frequency domain, then an expert interprets the signal and determines the type and location of the error [41].

In the practical experiment, the sensors used were calibrated using an available handheld calibrator with the following sensitivity: $98.0$ mV/g (g = 9.81 m/s² gravity) (Figure 4).

A vibration sensor comparison is shown in

Table 4. Vibration sensor comparison.

Type	Measurement Range	Frequency Response	Typical Mounting	Applications
Accelerometer	±1–500 g	1 Hz–10 kHz	Housing, Bearings	Fault Detection, Condition Monitoring
Velocity	0.01–100 mm/s	5 Hz–1 kHz	Structural Frame	General Monitoring
Piezoelectric	±50 g	0.5 Hz–12 kHz	Direct on Motor	Early Faults, High Sensitivity
MEMS	±2 to ±16 g	0.1 Hz–2 kHz	Embedded	Cost-Efficient, IoT Systems

.

3.3. Current Sensors

Current sensors are critical components in monitoring and controlling electrical machines and systems. They measure the current flowing through a circuit and convert it into readable signals—either analog or digital—for applications such as fault detection, efficiency optimization, overload protection, and power quality analysis [44,45]. Current can either be sensed through the magnetic field it generates or by directly measuring the voltage drop across a known resistor. The processed signal is typically forwarded to a motor controller, protection circuit, or monitoring system.

3.3.1. Rogowski Coils

Rogowski coils are air-cored, non-intrusive current sensors primarily used to measure alternating current (AC), especially in high-current and transient applications, where conventional current transformers (CTs) may not be suitable [46]. They consist of a flexible toroidal coil wound around a conductor. Due to the absence of a magnetic core, Rogowski coils do not suffer from saturation, enabling accurate measurement across a wide dynamic range.

The output voltage of a Rogowski coil is proportional to the time derivative of the current being measured:

$$V_{out}\propto {\displaystyle \frac{dI}{dt}}$$

(6)

To reconstruct the actual current waveform, the output must be passed through an integration circuit.

Rogowski coils offer numerous advantages: high bandwidth for capturing fast transients, immunity to saturation, minimal insertion loss, and flexible, lightweight construction. These attributes make them well suited for embedded motor monitoring, fault diagnostics, and harmonic analysis in power systems [47,48].

3.3.2. Shunt Resistors

Shunt resistors are precision, low-resistance elements connected in series with the current path. They operate on Ohm’s law to convert current into a measurable voltage drop:

$$V=IR$$

(7)

This voltage is then sensed and interpreted by control electronics to determine the current magnitude [49]. Shunt resistors are known for their high measurement accuracy in low-current applications but may introduce power loss and heating at higher currents.

3.3.3. Current Transformers (CTs)

Current transformers are widely used in industrial applications to step down high AC currents to manageable levels for measurement and protection. They provide galvanic isolation and are suitable for long-term monitoring of steady-state currents in power distribution systems [50].

3.3.4. Hall Effect Sensors

Hall effect sensors detect magnetic fields generated by current-carrying conductors and are capable of measuring both AC and DC currents. These sensors are versatile, non-contact devices commonly used for current sensing in power electronics, automotive systems, and motor drives [48]. They are particularly effective in applications requiring isolation and compact form factors.

Table 5. Comparison of current sensor types.

Sensor Type	Contact Type	Frequency Range	Accuracy	Typical Applications
Rogowski Coil	Non-contact	Wide (AC)	Medium	High-current motors, transient monitoring, harmonic analysis
Shunt Resistor	Contact	Low to Medium	High	Low-current circuits, embedded control systems
Current Transformer (CT)	Contact	AC only	High	Industrial monitoring, protection relays
Hall Effect Sensor	Non-contact	AC/DC	Medium to High	Motor control, automotive systems, general-purpose current sensing

presents a comparison of different types of current sensors.

3.4. Speed and Position Sensors

Speed and position sensors play a crucial role in determining rotor velocity and angular position in electric motors. These sensors enhance control accuracy and system efficiency in applications such as robotics, aerospace, and industrial automation [51].

Encoders (optical, magnetic, incremental, and absolute) generate pulse signals corresponding to the rotor position or speed. Tachometers convert rotational speed into electrical signals and are available in both analog and digital forms. Resolvers use electromagnetic induction to detect angular position with high accuracy, particularly in harsh industrial or aerospace environments.

Rotary encoders and LVDTs (Linear Variable Differential Transformers) provide precise measurements for rotary and linear displacement, respectively, and are used in CNC machines and actuators [52].

Hall Effect Sensors, when used for detecting changes in magnetic fields of a rotor, provide reliable speed and position feedback, especially in BLDC motors. Their durability and non-contact design make them ideal for continuous feedback in dynamic conditions [53,54].

Inductive Proximity Sensors generate an electromagnetic field to detect nearby metal components without contact. They are widely used for position detection of motor shafts, gears, and actuators. Their high durability, fast response, and resistance to harsh environments make them valuable in speed monitoring, fault detection, and automation systems [55,56,57,58].

4. Factors Influencing Sensor Selection and Placement

The selection and placement of the right sensor for an electrical machine depend on several factors, including the application, environmental conditions, performance requirements, and the nature of the data to be collected. For vibration sensors, in particular, such as accelerometers and piezoelectric sensors, both the type and the mounting location play critical roles in ensuring accurate diagnostics and effective monitoring. Below are the key factors to consider:

4.1. Measurement Types

Different types of measurements require specific sensors to ensure accurate data collection. For position measurement, sensors such as encoders, resolvers, and Hall effect sensors are commonly used. Speed measurement relies on tachometers, encoders, and Hall effect sensors. Current measurement is typically performed using Rogowski coils, shunt resistors, and current transformers (CTs) [59].

Temperature measurement involves the use of thermocouples and resistance temperature detectors (RTDs). For vibration measurement, accelerometers and piezoelectric sensors are utilized, especially for motor health monitoring. The effectiveness of vibration monitoring depends not only on sensor type but also on sensor placement—typically on the housing near bearings or other critical mechanical interfaces where vibrations are most indicative of faults [60].

4.2. Accuracy and Resolution

Accuracy and resolution are critical factors in sensor selection, especially for high-precision applications such as servo motors, robotics, and CNC machines, which require high-resolution encoders or resolvers to ensure precise control and positioning [61].

In contrast, general-purpose speed monitoring applications can function effectively with low-resolution encoders or Hall effect sensors, where extreme precision is not a primary requirement. The choice of sensor depends on the level of accuracy needed for the specific application [62].

4.3. Response Time

Response time is an important consideration when selecting sensors for different applications. Fast response sensors, such as optical encoders and Hall effect sensors, are essential for high-speed motor control in applications such as servomotors and brushless DC (BLDC) motors, where real-time feedback is crucial for precise operation [63].

However, slower-response sensors, such as the thermocouples used for temperature monitoring, are suitable for applications where rapid changes are not critical, as temperature variations typically occur gradually. The choice of sensor depends on the speed and responsiveness required for effective system performance [64].

4.4. Environmental Conditions

• Temperature: The selection of sensors for electrical motors is significantly influenced by the operating temperature of the environment. In high-temperature settings, such as industrial furnaces or motors in heavy-duty machines, sensors must be able to withstand extreme heat, while maintaining accurate performance [65].
  Resistance Temperature Detectors (RTDs) and thermocouples are commonly used in these environments because of their precision in temperature measurement. In addition, resolvers are often preferred in high-temperature conditions due to their durability and ability to operate reliably in intense heat [66].
• Vibration and Shock Resistance: Industrial environments with heavy machinery often expose sensors to high levels of vibration and shock, which makes durability a key factor in sensor selection [3].
  Resolvers are particularly well suited for such conditions because of their mechanically robust design. Similarly, Hall effect sensors, which do not have moving parts, offer improved reliability and resistance to mechanical wear and tear [67]. For vibration sensors themselves, proper placement is essential. Accelerometers should be mounted as close as possible to the source of vibration—typically on bearing housings or motor end shields—to ensure accurate detection of faults such as imbalance, misalignment, and bearing wear. Misplaced sensors may underreport or miss critical vibration signatures, leading to inaccurate diagnostics.
• Dust and Moisture Resistance: Sensors used in outdoor or harsh industrial environments must be resistant to dust and moisture, to ensure long-term functionality. Magnetic sensors are commonly used in such conditions due to their ability to operate effectively even in the presence of contaminants [68].
  Furthermore, IP-rated encoders provide additional protection against dust and moisture ingress, making them ideal for applications where exposure to these elements is unavoidable. Selecting sensors with appropriate protective features enhances their useful life and ensures consistent performance in challenging environments [69].

4.5. Contact vs. Non-Contact Sensors

Non-contact sensors, such as optical encoders, Hall effect sensors, and Rogowski coils, offer several advantages, including no wear and tear, a longer lifespan, and the ability to perform high-speed measurements. These features make them ideal for applications that require durability and precision over extended periods [70].

In contrast, contact-based sensors, such as potentiometers and mechanical tachometers, are more prone to wear due to physical contact, but are often a more cost-effective option. The choice between non-contact and contact-based sensors depends on factors such as longevity, accuracy, and budget constraints [71].

4.6. Power Supply and Signal Compatibility

When selecting a sensor, it is important to consider its power requirements and the compatibility of the output signal with the control system. Some sensors, such as active Hall effect sensors, require an external power supply to function properly. In addition, the sensor output signal must match the input requirements of the control system. Analog output sensors, including thermocouples, tachometers, and linear variable differential transformers (LVDTs), provide signals in the form of voltage or current. However, digital output sensors, such as encoders and Hall effect sensors, transmit data using pulse-width modulation (PWM), pulse signals, or quadrature encoding. Ensuring proper compatibility between the sensor and the control system is essential for accurate data acquisition and efficient performance [72].

4.7. Cost vs. Performance Trade-Off

The choice of sensors often involves a trade-off between cost and performance. In high-end applications, such as precision robotics and aerospace, the higher cost of resolvers or high-resolution encoders is justified by their superior accuracy and reliability. However, for general industrial motors, where extreme precision is not required, more cost-effective options such as basic Hall effect sensors or tachometers are sufficient. Balancing performance needs with budget constraints is the key to selecting the right sensor for a given application [73].

5. Integration with Control System

The integration of sensors with control systems in electrical machines is crucial to achieve efficient monitoring, protection, and optimization of motor performance. To ensure seamless operation, the sensor must be able to interface with motor controllers, programmable logic controllers (PLCs), or microcontrollers [74] (Figure 5). This integration enables real-time data acquisition, which allows operators to monitor key parameters such as temperature, vibration, and current. By feeding this information into the control system, it becomes possible to make informed decisions, implement corrective actions, and dynamically adjust system settings, thus improving the overall efficiency and minimizing the risk of equipment failure [75].

In some cases, certain sensors require additional signal conditioning to ensure compatibility with the control system. For example, sensors such as Rogowski coils, which are used to measure current, require integration circuits to convert their output into a usable signal for the control system. Signal conditioning ensures that the data are properly processed and scaled, allowing the control system to accurately interpret and act on the sensor’s readings. This step is vital to ensure that the sensor output is consistent and reliable, which ultimately improves the performance and protection of the electrical machine [76,77].

6. Sensor Installation and System Compatibility

The installation of sensors in electric motor systems must consider mechanical accessibility, environmental factors, data transmission infrastructure, and signal integrity. Improper installation can lead to degraded measurements, reduced fault detectability, or system instability.

For example, vibration sensors such as accelerometers must be securely mounted to rigid surfaces like bearing housings or motor casings using adhesives, bolts, or magnetic bases to ensure accurate signal transfer. Improper surface preparation or flexible mounting can attenuate the high-frequency components critical for detecting early-stage bearing wear [78]. Similarly, temperature sensors embedded in stator windings require insulation and thermal contact optimization to accurately reflect hotspot behavior.

Current sensor installation depends on whether the application uses contact-based (e.g., shunt resistor) or non-contact (e.g., Hall effect) sensors. Shielding and grounding are important in electrically noisy environments, especially when long cable runs are involved. Wireless or self-powered sensor systems have emerged as alternatives to reduce wiring complexity, but they introduce trade-offs in signal reliability and latency [79].

Some studies have begun to explore fault-informed placement techniques, where historical fault data guide the manual positioning of sensors near historically problematic zones (e.g., end-windings, inverters). In more advanced cases, unsupervised clustering or reinforcement learning models have been proposed to iteratively reposition sensors to maximize fault signal detectability over time [80].

Ultimately, effective sensor installation must balance mechanical feasibility with data quality, and ensure compatibility with downstream signal processing and AI/ML diagnostic pipelines.

7. Sensor Installation and Safety Considerations

When installing sensors in electric motor systems, both mechanical constraints and safety standards must be addressed to ensure optimal performance and compliance.

7.1. Mounting, Size, and Environmental Constraints

Mounting requirements and physical size are critical factors in sensor selection. In applications with limited space, compact sensors such as small Hall effect sensors or mini encoders are preferred. The type of motion being measured also determines the appropriate sensor: rotary encoders are ideal for shaft rotation, while linear variable differential transformers (LVDTs) are more suited for linear displacement. Additionally, proper sensor mounting—ensuring secure attachment, appropriate alignment, and minimal exposure to mechanical stress or vibration—is essential to maintain data accuracy and sensor longevity [81].

7.2. Safety, Compliance, and System Compatibility

Safety and compliance are especially important in high-voltage or hazardous environments. Sensors offering galvanic isolation, such as Rogowski coils, are preferred over shunt resistors to minimize electrical hazards. The selected sensors must also conform to industry standards such as ISO, IEC, or UL, to ensure reliability, durability, and regulatory compliance [82]. Compatibility with the broader system is also crucial; sensors must deliver signals (analog or digital) that integrate seamlessly with the motor control hardware and communication protocols, facilitating effective condition monitoring and feedback control.

8. Optimal Sensor Placement Strategies

Proper placement of sensors is crucial to ensure accurate measurements and effective monitoring of electrical motor performance. Strategic positioning helps to detect potential problems early, improving reliability and efficiency. The key considerations for sensor placement include factors such as heat generation, mechanical stress, and electrical characteristics.

8.1. Temperature Sensors

Temperature sensors are typically placed near critical heat-generating components such as windings, bearings, or hot spots in the motor housing. These areas are prone to excessive heat buildup, which can lead to overheating and eventual motor failure. By placing RTDs or thermocouples in these locations, operators can monitor temperature fluctuations and take preventive action to avoid thermal damage.

In the research carried out by [83] using a Lumped-Parameter Thermal Network (LPTN) (also known as a lumped capacitance network, which is a simplified approach to model heat transfer in physical systems, where the temperature is assumed to be uniform throughout the object and varies only with time), the LPTN was modeled as the temperature of the stator windings and the rotor block in an induction motor.

Figure 6 presents the model, with ${\theta }_{cu}\left(t\right)$ and ${\theta }_{rotor}\left(t\right)$ denoting the temperatures of the stator windings and the rotor. Thermal capacitances are given by $C_{cu}$ and $C_{rotor}$, while heat losses are represented by $P_{cu}$ (stator) and $P_{rotor}$ (rotor). The thermal resistances $R_1$ and $R_2$ model the heat transfer to ambient air at ${\theta }_{room}$ and between the rotor and stator windings, respectively [84].

The thermal response is described by the subsequent differential Equation (8).

$${\left[C\right]}_{2\times 2}{\left[{\displaystyle \frac{d\theta }{dt}}\right]}_{2\times 1}={\left[Y\right]}_{2\times 2}{\left[\theta \right]}_{2\times 1}+{\left[P\right]}_{2\times 1}$$

(8)

In this expression, ${\left[\theta \right]}_{2\times 1}$ denotes the temperature vector that represents the stator windings and the rotor. The term ${\left[{\displaystyle \frac{d\theta }{dt}}\right]}_{2\times 1}$ corresponds to its time derivative. The matrix ${\left[C\right]}_{2\times 2}$ characterizes the thermal capacitance, while ${\left[Y\right]}_{2\times 2}$ defines the thermal admittance. Finally, ${\left[P\right]}_{2\times 1}$ is the vector that accounts for internal heat sources. Equation (8) can be written in its full form as (9) [85]

$$\begin{array}{c}\hfill \left[\begin{array}{cc}{C}_{\mathrm{Cu}}& 0\\ 0& C_{\mathrm{rotor}}\end{array}\right]\left[\begin{array}{c}{\displaystyle \frac{d{\theta }_{\mathrm{Cu}}}{dt}}\\ {\displaystyle \frac{d{\theta }_{\mathrm{rotor}}}{dt}}\end{array}\right]=\left[\begin{array}{cc}-{\displaystyle \frac{1}{R_1}}-{\displaystyle \frac{1}{R_2(\Omega )}}& {\displaystyle \frac{1}{R_2(\Omega )}}\\ {\displaystyle \frac{1}{R_2(\Omega )}}& -{\displaystyle \frac{1}{R_2(\Omega )}}\end{array}\right]\left[\begin{array}{c}{\theta }_{\mathrm{Cu}}\\ {\theta }_{\mathrm{rotor}}\end{array}\right]+\left[\begin{array}{ccc}{\displaystyle \frac{1}{R_1}}& 1& 0\\ 0& 0& 1\end{array}\right]\left[\begin{array}{c}{\theta }_{\mathrm{room}}\\ P_{\mathrm{Cu}}\left(T\right)\\ P_{\mathrm{rotor}}(T,\Omega )\end{array}\right]\end{array}$$

(9)

Here, T and $\Omega$ represent the motor torque and angular speed, respectively. Thermal resistance $R_2(\Omega )$ is highly dependent on motor speed $\omega$, as it accounts for the convective heat transfer effects between the rotor, the air gap, and the stator. The heat generation terms $P_{\mathrm{Cu}}\left(T\right)$ and $P_{\mathrm{rotor}}(T,\Omega )$ corresponding to copper (Joule) losses and rotor iron losses are functions of the motor’s operating conditions, namely torque and speed. These quantities can be explicitly expressed in terms of T and $\Omega$, as shown in Equations (10)–(12) [83].

$$R_2(\Omega )=p_{R2,0}+p_{R2,1}\Omega +p_{R2,2}{\Omega }^2$$

(10)

$$P_{\mathrm{Cu}}\left(T\right)=p_{\mathrm{Cu},0}+p_{\mathrm{Cu},1}T+p_{\mathrm{Cu},2}{T}^2$$

(11)

$$\begin{array}{c}\hfill p_{\mathrm{rotor}}(T,\Omega )=p_{\mathrm{rotor},00}+p_{\mathrm{rotor},10}T+p_{\mathrm{rotor},01}\Omega +\\ \hfill p_{\mathrm{rotor},20}{T}^2+p_{\mathrm{rotor},11}T\Omega +p_{\mathrm{rotor},02}{\Omega }^2\end{array}$$

(12)

The coefficients p correspond to empirical fitting parameters that must be identified through experimental calibration. Equation (8) can be reformulated into the canonical state-space representation, as presented in Equations (13) through (17) [83]:

$$\dot{\mathit{\theta }}\left(t\right)=4\mathit{\theta }\left(t\right)+B\mathit{u}$$

(13)

$$\mathit{\theta }\left(t\right)=\left[\begin{array}{c}{\theta }_{\mathrm{Cu}}\left(t\right)\\ {\theta }_{\mathrm{rotor}}\left(t\right)\end{array}\right]$$

(14)

$$A=C^{-1}Y$$

(15)

$$B=C^{-1}\left[\begin{array}{ccc}{\displaystyle \frac{1}{R_1}}& 1& 0\\ 0& 0& 1\end{array}\right]$$

(16)

$$\mathit{u}=\left[\begin{array}{c}{\theta }_{\mathrm{room}}\\ P_{\mathrm{Cu}}\left(T\right)\\ P_{\mathrm{rotor}}(T,\Omega )\end{array}\right]$$

(17)

The complete time-domain responses, ${\theta }_{\mathrm{Cu}}\left(t\right)$ and ${\theta }_{\mathrm{rotor}}\left(t\right)$, are obtained by superimposing the zero-input response ${\Theta }_{0-1}\left(t\right)$ with the zero-state response ${\Theta }_{0-s}\left(t\right)$, as given in Equation (18):

$$\mathit{\theta }\left(t\right)=\left[\begin{array}{c}{\theta }_{\mathrm{Cu}}\left(t\right)\\ {\theta }_{\mathrm{rotor}}\left(t\right)\end{array}\right]={\theta }_{0-\mathbf{I}}\left(t\right)+{\theta }_{0-\mathbf{s}}\left(t\right)$$

(18)

8.2. Current Sensors

Current sensors are installed along motor power supply lines to monitor electrical characteristics such as current flow and load fluctuations. These sensors help detect anomalies such as overcurrent conditions, phase imbalances, or power supply disturbances, which can affect motor performance. Proper placement of current sensors ensures accurate detection of electrical problems, allowing timely corrective measures [86].

8.3. Position and Speed Sensors

Position and speed sensors are commonly mounted on the rotor shaft or embedded within the motor housing to enable high-precision motion control. These sensors deliver essential feedback for closed-loop control systems in applications demanding precise speed regulation and positional accuracy, such as robotics and industrial automation. Proper placement ensures optimal motor efficiency, improves control accuracy, and prevents operational errors. By carefully considering sensor placement, engineers can maximize the effectiveness of condition monitoring systems, improve motor performance, and prevent unexpected failures [87].

8.4. Vibration Sensors

Vibration sensors are mounted in the motor casing or near the bearings to detect mechanical imbalances, misalignments, and potential faults. Bearings are among the most vulnerable components of a motor, and excessive vibrations can indicate wear or imminent failure [88].

The best location for a vibration sensor is directly on a bearing, as this ensures the most accurate detection of the high-frequency signals generated during early bearing failure. In this research, a preliminary test was carried out by placing three sensors in three different locations on the motor chosen randomly to examine the accuracy, which will lead to the final conclusion after a number of tests and examining different parameters in addition to the vibration. Most vibration sensors are single-axis and best detect vibrations perpendicular to their mounting surface. The direction of measurement—vertical, horizontal, or axial—affects the types of defects detected [89]:

• Horizontal direction: better for identifying imbalances and electrical problems.
• Vertical direction: generally better for detecting looseness and bearing issues.
• Axial direction: more effective in detecting angular misalignment.

If direct placement on a bearing is not possible (for example, due to a fan cover), the sensor should be mounted as close to the sensor as possible, understanding that signal strength may decrease with distance. By continuously monitoring vibration levels, maintenance teams can identify and address problems before they escalate, reducing downtime and extending motor lifespan [89].

9. Experimental Setup and Preliminary Results (Vibration)

This review is part of an ongoing project focused on the integration of artificial intelligence (AI) in electrical machines. Sensors, which serve as the primary means for data acquisition in machine learning applications, are initially being used for vibration analysis [90]. The experiments are being carried out in the Vibration Research and Testing Center (VRTC) laboratory and will later be extended to include additional parameters such as temperature, current, and others.

The current experimental setup involves a 3 kW, two-pole, three-phase induction motor. The motor nameplate is shown in Figure 7.

The placement of the vibration sensors on the motor is shown in Figure 8 where S1, S2, and S3 stand for sensors 1, 2, and 3, respectively.

The three sensors produced different results, the experiment was carried out at 3 speeds (750, 1500, and 3000), and the speed was varied by means of a variable frequency drive (VFD) (Figure 9).

The average voltages measured by sensors 1, 2, and 3 were 0.0027 V, 0.0149 V, and 0.0211 V, respectively (Figure 10). These results suggest that sensors placed closer to a bearing experience higher vibration amplitudes, which can lead to more accurate readings and improved fault detection capabilities. The induction motor used in the experiment was assumed to be in good working condition, as subsequent phases of the research will focus on integrating machine learning techniques for fault prediction. Sensor placement was carried out on the basis of the factors that influence sensor selection, as discussed in Section 3. It is important to note that the sensor placements used in this experimental setup were selected manually based on general mechanical intuition and common industry practices. No AI-driven models, optimization algorithms, or historical datasets were used to guide placement. The objective of this preliminary experiment was to provide a baseline understanding of how vibration signals vary with position on the motor surface. Future work will utilize techniques such as Fast Fourier Transform (FFT), order tracking, and machine-learning-based signal classification to extract deeper insights from the sensor data. These methods will allow for the detection of characteristic fault frequencies and the evaluation of placement effectiveness relative to mechanical conditions such as imbalance, misalignment, or bearing wear.

Note: The vibration signals presented in this section, including the data in Figure 10, are shown in voltage as recorded directly from the sensors. Although a handheld calibrator with a known sensitivity of 98.0 mV/g was used, the conversion to physical acceleration units (g or m/s²) was not applied due to missing or inconsistent calibration records during the measurements. This limitation will be addressed in future experiments by applying proper calibration to ensure unit-consistent and physically interpretable results.

Frequency Domain Analysis of Vibration Signals

To evaluate how sensor location impacts the diagnostic quality of vibration measurements, we performed Fast Fourier Transform (FFT) analysis on the signals collected from all three accelerometer positions. The resulting frequency spectra revealed distinct amplitude patterns based on proximity to motor components.

Sensors placed near the bearing housings captured high-amplitude peaks in the 100–300 Hz range, corresponding to fundamental frequencies of mechanical rotation and potential early-stage bearing wear. In contrast, the sensor mounted on the motor casing showed an attenuated high-frequency content, suggesting reduced sensitivity to localized mechanical faults. This implies that signal strength and diagnostic value vary significantly with placement.

These results support the argument that frequency content can act as a proxy for placement effectiveness, offering a metric for intelligent sensor deployment. In future work, such spectral features could be used to train AI/ML models—such as reinforcement learning agents or supervised classifiers—to guide sensor positioning toward locations that maximize fault visibility or information gain.

Although this study used manually selected positions, the results demonstrate how signal quality is affected by placement, reinforcing the need for data-driven optimization strategies in real-world systems.

As shown in Figure 11, the frequency response varied significantly with sensor placement, with higher spectral amplitudes observed at positions closer to the motor’s mechanical components.

10. Emerging Trends and Future Directions

As technology advances, the selection and placement of sensors in electric motors undergoes significant improvements. The focus is on improving efficiency, reducing maintenance costs, and strengthening real-time monitoring capabilities. Innovations in sensor technology are shaping the future of motor condition monitoring, with the aims of higher reliability and performance [91].

Wireless sensor networks are changing how sensors are deployed by offering greater flexibility and reducing installation costs. By eliminating the need for extensive cabling, wireless sensors are easier to install and move, especially in complex industrial environments, where access can be difficult. Wireless technology also supports scalable solutions, making it easy to add more sensors when needed [92].

Artificial intelligence and machine learning are playing vital roles in the use of sensor data for predictive maintenance and anomaly detection. By analyzing large amounts of real-time data from temperature, vibration, and current sensors, AI systems can detect patterns that signal potential failures. This proactive approach can reduce downtime and improve maintenance efficiency. Furthermore, AI can help optimize sensor placement by identifying the best monitoring points based on past performance [93].

Energy harvesting sensors offer a sustainable solution by deriving power from ambient energy sources such as mechanical vibrations, thermal gradients, or electromagnetic fields. This methodology is particularly advantageous in remote or inaccessible environments, where conventional power supply options are unfeasible. By removing the reliance on batteries or wired connections, these sensors facilitate efficient, low-maintenance monitoring systems [94].

The integration of sensors with Internet of Things (IoT) infrastructure enables remote monitoring capabilities and facilitates real-time data acquisition and analysis. By transmitting data to cloud platforms, operators can access critical motor performance information from any location. This connectivity enables quick decision-making, early fault detection, and remote troubleshooting. IoT integration also supports automated alerts and advanced analytics, improving motor efficiency and reliability [79].

Together, these trends are transforming sensor selection and placement in electric motors, leading to improvements in predictive maintenance, operational efficiency, and cost savings. As sensor technology continues to progress, future advancements will likely emphasize greater automation, enhanced connectivity, and more sustainable solutions in motor monitoring systems [95] (Figure 12).

One of the most promising directions for future research involves the use of AI and machine learning techniques to guide optimal sensor placement. Supervised learning algorithms can be trained on labeled datasets containing vibration signals from various fault conditions and sensor positions to identify high-sensitivity zones for specific fault types. Dimensionality reduction techniques such as Principal Component Analysis (PCA) or t-distributed Stochastic Neighbor Embedding (t-SNE) may be used to evaluate signal redundancy and identify the minimal set of placement locations needed for accurate fault detection. Reinforcement learning may also be explored for adaptive placement strategies in dynamic environments, such as offshore oil platforms, where operating conditions and failure modes can change over time. These approaches aim to reduce trial-and-error placement, improve diagnostic accuracy, and support predictive maintenance.

Challenges and Path Forward for AI/ML Integration

Despite the promise of artificial intelligence (AI) and machine learning (ML) in electric motor monitoring, their practical integration with sensor placement strategies remains limited. Most reviewed studies applied AI/ML for fault classification or signal analysis rather than for optimizing the sensor deployment itself.

One core challenge is the lack of annotated datasets that link sensor position with fault detection performance. Effective ML models require labeled data that capture how placement affects the visibility of specific fault signatures—something that current datasets do not adequately provide. Additionally, the dynamic nature of operating environments, such as those in offshore platforms, adds complexity to training robust, generalizable models.

Recent works such as [6,96] have demonstrated the potential of ML in vibration-based condition monitoring and dimensionality reduction for sensor selection. However, few works have explored adaptive placement frameworks using reinforcement learning or uncertainty-based exploration.

This paper identifies these gaps and proposes future work to develop a placement optimization framework using supervised learning on fault-labeled vibration data, followed by reinforcement learning for adaptive strategies in dynamic environments. These approaches will be supported by our forthcoming experimental dataset.

11. Conclusions

This paper presents a comprehensive review of the selection and placement of sensors for electric motor monitoring, with particular emphasis on applications in industrial and offshore environments. The study covers various sensor types—such as temperature, current, position, and vibration sensors—and evaluates them based on key selection factors including accuracy, response time, environmental compatibility, and signal integration.

The review is complemented by a preliminary experimental setup involving three accelerometers placed on a low-voltage motor to observe vibration signal variations. While the experimental sensor placement was not guided by an AI- or data-driven strategy, it served to illustrate the baseline signal behavior at different locations and demonstrated the need for more structured approaches.

This work is intended as the first step toward a more advanced framework for intelligent sensor deployment. Future research will incorporate AI/ML techniques—such as classification models, dimensionality reduction, and reinforcement learning—to develop optimized and adaptive sensor placement strategies based on historical fault data. These efforts aim to improve fault detection, reduce sensor redundancy, and enhance the reliability and performance of electric machines, especially in mission-critical environments like offshore platforms.