Performance Evaluation of a Hybrid Power System for Unmanned Aerial Vehicles Applications

Abstract

While electric unmanned aerial vehicles (UAVs) offer advantages in noise reduction, safety, and operational efficiency, their endurance is limited by current battery technology. Extending flight autonomy without compromising performance is a critical challenge in UAV system development. Previous studies introduced hybrid micro-turbogenerator architectures, but limitations in control stability and output power constrained their practical implementation. This study aimed to finalize the design and experimental validation of an optimized hybrid power system featuring a micro-turboprop engine mechanically coupled to an upgraded electric generator. A fuzzy logic-based control algorithm was implemented on a single-board computer to enable autonomous voltage regulation. The test bench architecture was reinforced and instrumented to allow stable multi-stage testing across increasing power levels. Results demonstrated stable voltage control at 48 VDC and electrical power outputs up to 3 kW, with an estimated maximum of 3.5 kW at full throttle. Efficiency was calculated at approximately 67%, and analysis of the generator’s KV constant revealed that using a lower KV variant (KV80) could reduce required rotational speed (RPM) and improve performance. These findings underscore the value of adaptive hybridization in UAVs and suggest that tuning generator electromechanical parameters can significantly enhance overall energy efficiency and platform autonomy.

1. Introduction

In recent years, electric unmanned aerial vehicles (UAVs) have garnered significant interest due to their capacity to perform demanding and high-risk missions while offering advantages in operational flexibility, safety, and cost-efficiency [1]. To enhance efficiency and reliability, hybrid power systems integrating energy sources such as solar arrays, lithium batteries, and fuel cells have been developed. These systems enable improved energy utilization, flexible power control, and stable operation through real-time monitoring, thereby substantially extending flight endurance and optimizing energy management. Previous studies [2,3] have established design methodologies for such hybrid configurations. In [2], a hybrid power system was designed to enhance UAV endurance, supported by a proposed energy management strategy ensuring efficient system operation. Further research [3] focused on hybrid fuel cell and battery systems, emphasizing their efficiency in extending flight autonomy and reliability in responding to demand peaks. Optimal component sizing was determined, recommending a 2000 W fuel cell paired with a 3000 mAh battery. These advancements collectively contribute to improved UAV endurance, energy efficiency, and operational reliability.

The adoption of electric propulsion systems in UAVs has grown significantly, driven by their notable benefits such as dependable performance [4], low noise output, resistance to disturbances [5], and reduced thermal visibility [6,7]. These systems also offer high operational efficiency [8], produce no harmful emissions, and feature autonomous starting and advanced control capabilities that allow for precise maneuvering. Such attributes make electric UAVs particularly well-suited for use in noise-sensitive and environmentally regulated urban settings, supporting tasks like traffic observation, medical logistics, and aerial imaging with minimal environmental or societal impact. Nevertheless, UAVs equipped with internal combustion engines (ICEs) continue to be favored for missions demanding longer flight durations, benefiting from the superior power and energy densities characteristic of ICE technology. Additionally, during cruise operations, thermal engines are often favored over batteries due to their greater payload capacity, shorter refueling times, and longer operational range [9].

Alongside the growing adoption of electric propulsion, significant advancements have been made in internal combustion engine (ICE) technologies for UAVs. Innovations such as direct injection (DI) and self-pressurized injectors have improved fuel economy and endurance by facilitating stratified charge formation, addressing cold start limitations, and reducing scavenging losses [10,11]. Further improvements in fuel injection timing, combustion precision, and the incorporation of alternative fuels have contributed to enhanced endurance and payload capacity [10]. Recent developments include a newly designed 100 kW opposed-piston diesel engine optimized for UAV applications; however, practical efficiency remains influenced by mechanical loads from auxiliary subsystems, including supercharging, lubrication, and onboard power generation requirements [12]. Additionally, research on two-stroke spark-ignition UAV engines has demonstrated that optimized fuel injection strategies can increase power efficiency by 5% to over 10% at rated speeds, further improving operational performance [13].

Fossil fuel-based power sources for UAVs fall into two main categories: piston engines (primarily internal combustion engines) and turboprop engines. Piston engines perform best at low altitudes and speeds [14], while turboprop engines exhibit superior performance at higher altitudes. Additionally, turboprop engines consume fuel at a higher rate compared to internal combustion engines, which are more fuel-efficient [15]. The noise profiles of these engines also differ significantly due to their distinct designs and operating principles. Turboprop engines generate high-frequency noise, mainly from the compressor [16], whereas piston engines produce lower-frequency noise due to their mechanical operation [15]. The power output of a gas turbine engine suitable for powering a UAV typically ranges from 3 to 200 kW [17], whereas the power output of the micro-piston engine for UAV applications varies between 10 W [18] and 10 kW [12]. However, purely fuel-based systems are not without drawbacks, including increased UAV weight, higher emissions, and slower response times to flight controller commands. To address the limitations inherent to purely ICE-based or electric systems, hybrid architectures combining ICEs and electric propulsion have been explored, offering improved fuel consumption rates and extended endurance for remotely piloted aircraft systems (RPASs), despite a lack of comprehensive efficiency benchmarks [19]. More advanced hybrid configurations integrate multiple energy sources, such as fuel cells, lithium batteries, and supercapacitors.

By combining the strengths of both fuel and electric power systems, hybrid energy solutions improve operational efficiency, endurance, and flight range. The high energy density of fossil fuels [20], paired with the high-power density of lithium batteries, extends UAV flight duration [21]. Additionally, the electric component ensures rapid response to onboard computer commands, while the fuel-based unit provides sustained endurance, enhancing overall performance. The configuration of hybrid power systems varies depending on the UAV’s mission. Solar or fuel cells are well-suited for urban and rural applications, whereas military and long-range surveillance missions typically favor fossil fuel-based systems, such as internal combustion or micro turboprop engines. It should be noted that hybrid systems are most efficient in larger UAVs, as the added mass from these power sources makes them less practical for smaller drones.

These systems utilize hierarchical control strategies that balance long-term efficiency with real-time responsiveness. For example, a model predictive control (MPC) layer employing the equivalent consumption minimization strategy (ECMS) governs power distribution, while a subordinate control layer regulates voltage and current outputs to maintain stable operation during demanding flight conditions [22]. Another prominent example of widely implemented energy management controllers for UAV applications is the fuzzy logic-based control system [23]. As demonstrated in the research by L.L.P. Aguilar et al. [24], employing a fuzzy logic-based energy management strategy improves both power regulation and fossil fuel consumption control during critical flight phases, including take-off and cruise operations. This enhancement is achieved through the improved stability and dynamic responsiveness of the UAV’s power system [25]. Furthermore, the Adaptive Neuro-Fuzzy Inference System (ANFIS) utilizing Takagi–Sugeno architecture [26] proves particularly suitable for applications requiring adaptive control, as it dynamically adjusts membership functions and fuzzy rules to establish a more responsive, data-driven control framework [27]. An alternative approach involves the implementation of Proportional–Integral–Derivative (PID) controllers, which are frequently utilized in energy management systems for hybrid power sources [28]. However, while its structure is typically straightforward, this approach proves less suitable for systems exhibiting nonlinear characteristics due to the inherent challenges in controller design and parameter tuning [29]. Furthermore, the PID and PI controllers exhibit a comparatively longer response time relative to the fuzzy logic controller [30,31,32]. Consequently, this study employs the fuzzy logic controller due to its superior operational efficiency, which translates to enhanced fuel economy and more rapid command execution.

Beyond UAV propulsion, hybrid energy generation systems, notably those employing pressurized solid oxide fuel cells (SOFC) coupled with gas turbines or ICEs operating on SOFC tail gases, have emerged as promising solutions for clean energy generation. Experimental evaluations of spark-ignition (SI) engines fueled by SOFC anode gas have achieved brake thermal efficiencies up to 31.4% under optimized operating parameters while maintaining low nitrogen oxide (NOx) emissions and compliance with evolving regulatory standards [33].

Simultaneously, improvements in UAV communication and power management frameworks have been pursued. A novel cell-free network architecture for UAV base stations (UBS), incorporating high-altitude platform stations (HAPSs) as central processing units, has demonstrated the potential to enhance user equipment quality of service (QoS), reduce onboard energy consumption, and extend communication durations. Leveraging millimeter wave (mmWave) frequencies for access and backhaul links, this architecture employs advanced optimization algorithms for user association, bandwidth distribution, and power allocation, surpassing conventional techniques in both efficiency and computational performance [34].

While fixed-wing UAVs and RPASs benefit significantly from propulsion and hybrid energy system advancements, multirotor platforms face distinct challenges primarily related to limited flight autonomy. Their reliance on lithium polymer batteries, characterized by relatively low energy density coupled with high power demand, necessitates frequent recharging, restricting their operational range and duration. To address these limitations, recent studies have explored the feasibility of integrating hybrid internal combustion engine (ICE) and electric generator sets as alternative power sources for multi-rotors. Evaluations conducted on theoretical designs accommodating mid-to-heavy payloads (15–35 kg maximum take-off mass) have demonstrated that hybrid configurations substantially improve endurance and extend aerial coverage compared to battery-only systems. This increased autonomy expands multirotor application potential across domains such as aerial imaging, surveying, and industrial inspection [35].

Further innovations in energy storage have focused on comparing conventional energy storage systems (CESSs) with hybrid energy storage systems (HESSs), particularly for UAV applications requiring efficient power delivery and thermal management. Experimental results indicate that HESS configurations, by incorporating supercapacitors, enable batteries to complete additional discharge cycles relative to the CESS while reducing initial current response and lowering power delivery demands by approximately 30 W. Moreover, HESSs demonstrate improved thermal performance, with surface temperature increases delayed and reduced compared to CESSs, differences averaging 1.2 °C in sustained conditions and peaking at 1.6 °C. These improvements not only enhance energy consumption efficiency but also support more stable temperature profiles during extended discharge cycles, highlighting the promise of HESSs in augmenting UAV endurance and reliability [36].

2. Materials and Methods

Despite these advancements, several challenges remain in the practical realization of hybrid power systems for UAVs, particularly in maximizing power output while ensuring system stability, efficiency, and adaptability to evolving mission profiles. Previous studies [14,37] established the feasibility of integrating a micro-turbogenerator (MTG) system into UAV platforms, developed a test bench for performance evaluation, and implemented a fuzzy logic-based control strategy to enhance dynamic power allocation and voltage stability. However, limitations such as capped power delivery and the need for upgraded system components constrained the full realization of the system’s potential. Building upon this foundational work, the present study introduces a redesigned hybrid power system featuring an MTG, a micro-turboprop engine coupled to an electric generator. The redesigned architecture incorporates modifications to both mechanical and electrical subsystems and implements an enhanced control strategy that addresses the prior power limitations. Experimental testing confirms the system’s capability to achieve the desired power output and operational reliability, representing the final validation of the proposed hybrid power solution.

2.1. Test Bench Optimization

In study [37], a test bench was presented for a hybrid power generation system composed of a micro-turboprop engine mechanically coupled to an electric generator via a flexible mechanical coupling. The experimental results highlighted the need for improvements to both the electrical subsystem and the control strategy governing the overall system.

These issues were addressed in [14] through the integration of additional electronic components, including diodes and a sensor for monitoring electrical parameters such as voltage and current. Furthermore, the control-related shortcomings were resolved by implementing a fuzzy logic-based algorithm. Following its implementation and testing, the algorithm achieved improved voltage stabilization by refining the membership function constraints of the fuzzy logic controller. This advancement facilitated the automation of the throttle control for the system’s primary load, the Electric Ducted Fan (EDF), manufactured by JP Hobby (Bourbonne-les-Bains, France).

Following the experimental campaign reported in [21], structural issues were identified in the test bench, primarily due to vibrations generated by the micro-turboprop engine and the positioning of the EDF. Specifically, the EDF was mounted in front of a vertical crossbeam, which, when intersected by the airflow, induced oscillations that propagated through the entire structure. Additionally, repeated testing revealed that the electric generator used in studies [14,37] was unable to deliver the required electrical power output. Consequently, the generator was replaced with a higher-performance unit. This upgrade introduced an additional electrical load on the EDF, which risked destabilizing system operation. To mitigate this issue, a second EDF was integrated into the test bench architecture. The addition of this second EDF necessitated modifications to the electrical configuration, including the incorporation of an additional electronic speed controller (ESC) and a diode. The updated electrical architecture is illustrated in Figure 1.

Figure 1 illustrates the updated electrical architecture of the hybrid power generation test bench. The connections between components are represented by distinct colored lines, where each hue signifies the type of linkage, whether power supply (red and black), signal transmission (blue), mechanical coupling (black), or magnetic-based connection (green). The system is based on a micro-turboprop engine, manufactured by KingTech Turbines (Kaohsiung City, Taiwan), mechanically coupled to an electric generator via a flexible coupling. The generator produces AC power, which is converted to DC using a rectifier bridge. The DC power is then routed to supply two EDFs (EDF 1 and EDF 2), each controlled by an individual ESC (ESC 1 and ESC 2).

A 48 VDC battery is integrated into the system, not as the primary power source, but as a buffer stabilizing the bus voltage and supplying additional power where needed, including to the micro-turboprop engine’s support systems (e.g., fuel pump, ECU). Diodes are employed to manage current flow and prevent backfeeding between components. A Hall sensor monitors generator output, while a tachometer tracks engine speed. A single-board computer (SBC) interfaces with the ESCs and the Electronic Control Unit (ECU), facilitating real-time control and coordination of system components.

The revised configuration, presented in Figure 2, enhances power stability, supports dual EDF operation, and allows for dynamic power sharing between the generator and the energy buffer, contributing to improved efficiency and robustness under variable load conditions.

The updated structure of the test bench reflects a design approach focused on rigidity, modularity, and dimensional precision. Constructed primarily from extruded aluminum profiles with standardized cross-sections and V-slot fastening channels, the configuration provides an optimal framework for integrating functional components within a robust experimental setup. Each element is positioned to meet both axial alignment requirements and operational considerations such as accessibility, maintainability, and in situ adjustability during testing.

The modular construction principle enables rapid replacement of individual components without compromising the overall structural integrity, while multiple junctions using corner brackets and metal plates ensure effective load transmission and structural continuity. Reinforcements and anchoring solutions in critical interface zones, such as the generator mount and engine support area, are specifically engineered to minimize deformation under load, thereby enhancing the system’s stability under dynamic operating conditions.

2.2. Optimized Control Algorithm

Before addressing the optimization of the control algorithm for the micro-engine, the custom-developed control interface must be described. This interface incorporates five functional buttons, five display zones dedicated to real-time monitoring of key parameters recorded during testing, and an input field for user-defined parameter specification. The functionalities assigned to the five buttons are as follows:

• The START button initializes all control systems and brings the micro-turboprop engine to idle operating conditions.
• The STOP button terminates both the engine and associated control systems while automatically saving the recorded test data to an Excel file.
• The IDLE button acts as a safety mechanism, allowing the operator to immediately disengage the automatic control of the MTG. It sets the turbine throttle to idle and simultaneously disables the test bench’s power consumers, namely the two EDFs.
• The AUTO button activates the autonomous control program for the MTG.
• The POWER button transmits the user-defined power value, entered in the field labeled ‘Enter MAX POWER [W]:’, directly to the control algorithm.

Additionally, the interface includes five display zones that provide real-time feedback on key system parameters:

• The status field indicates the current operational state of the MTG.
• The voltage display shows the DC voltage value, measured at the output of the system’s rectifier bridge and expressed in volts (V).
• The current display presents the direct current (DC) measured at the same output point, expressed in amperes (A).
• The power field calculates and displays the electrical power, expressed in watts (W), as the product of the measured voltage and current.
• The speed indicator refers to the rotational speed of the electrical generator, expressed in revolutions per minute (RPM).

The interface is illustrated in Figure 3.

The improved control program developed for the MTG test bench is structured around six dedicated control cores responsible for the autonomous operation of the micro-turboprop engine. The central core, serving as the main execution thread, manages communication with the graphical control interface and coordinates the execution of five additional subroutines categorized into three functional domains: control, sensing, and data logging.

The two sensing subroutines are linked to a Hall-effect-based sensor that measures voltage and current at the rectifier output and a tachometer that records the rotational speed of the electric generator. Both sensors operate at a sampling frequency of 10 Hz. The REGISTER subroutine is responsible for continuously logging all relevant parameters until the STOP command is issued, at which point the collected data is automatically stored and consolidated in an Excel file.

The control of the test bench is executed through two additional subroutines: PWM, which governs the power delivery to the system’s primary consumers (e.g., EDFs), and MIND, which dynamically regulates the operating state of the micro-turboprop engine.

The MIND subroutine ensures the autonomous control of the micro-turboprop engine by maintaining the supply voltage at a constant setpoint of 48 VDC. The autonomous control subroutine continuously monitors and adjusts the turboprop engine’s throttle position to maintain system voltage stability while responding dynamically to power demand fluctuations. This intelligent regulation system serves as the core energy management mechanism, carefully balancing generator output with consumer requirements. During operation, the system implements a graduated power allocation strategy for the EDFs on the test bench. When the power demand value, inserted by the operator in the “Enter MAX POWER” label of the interface, remains below 1000 W, the control algorithm selectively activates only the primary EDF, gradually increasing its throttle position by 1% every four seconds. This measured approach prevents sudden load changes that could destabilize the system. As the demand escalates beyond the 1000 W threshold toward the 2000 W maximum, the control scheme demonstrates its adaptive capabilities. The primary EDF maintains its current throttle setting while the secondary EDF comes online, mirroring the same deliberate 1% per four seconds ramp-up rate. Thus, the throttle settings of the EDFs are incrementally adjusted in a controlled manner, ensuring a smooth power transition that aligns precisely with the operator’s specified demand.

The control algorithm implemented within the MIND subroutine is based on the operational principles of a fuzzy logic controller, similar in structure to the one employed in [14]. The mathematical model underlying the fuzzy logic control strategy follows the standard three-stage framework, which includes the following steps:

• Fuzzification involves the transformation of input parameters into fuzzy linguistic variables, enabling the system to interpret continuous numerical values in qualitative terms [38].
• Fuzzy rule formulation consists of establishing a set of cause–effect rules, based on predefined membership functions, to govern how the controller adjusts or corrects the target parameter [39,40].
• Defuzzification refers to the process of converting the fuzzy linguistic outputs back into crisp numerical values with real physical meaning, suitable for direct application in the control process [41,42,43].

The optimized subroutine introduces the fuzzification of three key parameters: the direct current (DC) voltage, expressed in volts (V); the voltage derivative, measured in volts per second (V/s); and the PWM (Pulse Width Modulation) control signal. Based on the error between the measured voltage and the target reference value of 48 VDC, the system associates the input with one of the following fuzzy linguistic categories: VERY LOW, LOW, NEGATIVE OK, OK, POSITIVE OK, HIGH, and VERY HIGH. The corresponding fuzzy cognitive map, presented in Figure 4, illustrates the value ranges for each category. Notably, the interval widths decrease progressively as the voltage approaches the nominal value of 48 VDC, reflecting a finer granularity near the setpoint.

To dampen transient voltage components and improve the dynamic response time, the voltage derivative is introduced as an additional fuzzy input. Depending on its magnitude and direction, it is classified into one of the following linguistic terms: NEGATIVE FAST, NEGATIVE SLOW, ZERO, POSITIVE SLOW, or POSITIVE FAST. The corrective action of the fuzzy system is executed through a PWM control signal generated by the SBC. As such, this output parameter is also integrated into the fuzzy control scheme and mapped into the following categories: VERY LOW, LOW, NEGATIVE MEDIUM, MEDIUM, POSITIVE MEDIUM, HIGH, and VERY HIGH. The intervals corresponding to each linguistic term were derived based on experimental analysis.

The fuzzy rules that govern the logical relationships among the fuzzy linguistic categories assigned to the three analyzed parameters, DC voltage, voltage derivative, and PWM control signal, are illustrated in the previously introduced Figure 4. Given the significant influence of voltage inertia on system dynamics, the derivative of the voltage is assigned the highest weight in the fuzzy decision-making process. For example, if the measured voltage falls within the NEGATIVE OK category and its derivative is classified as NEGATIVE SLOW, the MIND subroutine responds by adjusting the PWM control signal to a value corresponding to the HIGH range. Based on this decision logic, a total of 35 regulation rules were formulated and implemented within the MIND control core, all of which are schematically represented in Figure 4.

In addition to the fuzzy regulation rules, the MIND control core incorporates active protection mechanisms designed to prevent damage to the electrical system under extreme operating conditions. Automatic shutdown procedures are triggered under the following circumstances:

• If, during engine startup, the DC voltage fails to reach at least 15 VDC within 80 s, it is an indication of a potential malfunction in the generator or the electrical system. Under nominal idle conditions (generator speed ~2800 RPM), the system typically produces approximately 20 VDC.
• If the generator speed exceeds 7000 RPM, it signals that the micro-turboprop engine is producing substantially more power than is being consumed, which may lead to overloading or system damage.
• If the measured voltage exceeds 55 VDC, it is a level that poses a risk of overvoltage stress to the MTG’s electrical architecture.
• If the system is in an active state but the measured voltage drops below 15 VDC, which may indicate that the generator is no longer rotating, the sensor is malfunctioning, or a fault has occurred within the electrical system.
• If the generator speed drops to 0 RPM, it suggests either a mechanical failure or a sensor malfunction.

Furthermore, if the measured voltage exceeds 53 VDC, the MIND control core automatically initiates a transition from AUTO mode to IDLE mode, independently of manual input via the IDLE button. Elevated potential differences in this magnitude indicate that the automatic control sequence within the MIND core is no longer capable of maintaining voltage stability. In such cases, the system reverts to IDLE mode as a protective measure to prevent overstressing the electrical subsystem.

2.3. Experimental Campaign Strategy

Following the optimization of both the test bench architecture and the control program, the MTG system was subjected to a series of experimental tests aimed at validating the performance enhancements. The testing protocol was designed to evaluate the system’s ability to stabilize the output voltage across various levels of electrical power demand while simultaneously ensuring the integrity and operational safety of the overall setup. The final configuration of the optimized testing procedure is outlined below:

• Initiation of the MTG at idle operating conditions;
• System verification and preliminary functional checks;
• Activation of the AUTO control mode within the MIND core;
• Incremental increase in the electrical power output to 500 W, following voltage stabilization, and maintaining steady-state operation at this level;
• Subsequent increase in electrical power output to 1000 W;
• Further increase in electrical power to 2000 W;
• Raising the electrical power level to 2500 W;
• Final increase to 3000 W;
• Engagement of the IDLE mode through the control system;
• Complete shutdown of the MTG.

These tests aim to monitor key parameters that characterize both the power generation subsystem of the test bench and its associated control logic. Accordingly, performance analysis requires the simultaneous recording of two categories of data. The first includes the fundamental electrical parameters measured at the output of the rectifier bridge, namely voltage, current, and the resulting electrical power, calculated as the product of the former two. The second category comprises the mechanical parameters associated with the power delivery subsystem, including the rotational speed of the electric generator and the throttle positions of the two EDFs, as well as that of the micro-engine. In addition, to enable time-resolved analysis of system dynamics, the total duration of each test is also recorded.

3. Results

To evaluate the performance of the optimized control algorithm, a series of four tests were conducted, each targeting a specific electrical power output. In the initial phase, testing was performed using only one of the two available power consumers, with the objective of reaching an electrical power level of 600 W. The corresponding experimental results are presented in Figure 5. As shown in the figure, which illustrates the variation in key test parameters, the MTG reaches its idle regime approximately 50 s after the activation of the control program, at which point the system voltage stabilizes around 20 VDC. Seeing as the power demand was on the lower side of the spectrum attainable with this test bench, only one of the EDFs was active during this test. From this point onward, the autonomous control sequence within the MIND core becomes active, progressively adjusting the throttle of the micro-turboprop engine to raise the system voltage toward the target value of approximately 48 VDC. As depicted in the graph, the control algorithm maintains a stable potential difference by incrementally increasing the throttle of the power consumer, the active EDF, by 1% every 4 s until the desired electrical power output of 600 W is achieved. At this power level, the system operated under the following conditions:

• Micro-turbine throttle position: 42%;
• Generator rotational speed: approximately 5300 RPM;
• EDF throttle position: 28%.

The second test followed a procedure similar to that of the first, as can be observed by comparing Figure 5 and Figure 6. In this case, however, the system reached an electrical power output of 1 kW. Similarly to the first test presented, only one EDF was active. The corresponding operating parameters at this load level were as follows:

• Micro-turbine throttle position: 50%;
• Generator rotational speed: approximately 5500 RPM;
• EDF throttle position: 35%.

The next experimental test, with parameter variations illustrated in Figure 7, targeted a total electrical power output of 2 kW. The procedure followed a similar sequence to the previous tests until a power level of approximately 650 W was reached. At that point, an operator error occurred, specifically, the unintended activation of the second power consumer (EDF 2). To correct this, the IDLE button was triggered, halting the autonomous control sequence and reducing the system’s electrical power consumption to 0 W. The autonomous control mode was then re-engaged, and the system resumed operation, reaching 1 kW following the previously established control logic.

To reach higher power levels, the second EDF was intentionally activated. Its throttle was increased by 1% every 4 s, mirroring the control strategy used for the first EDF, until the total electrical power consumption reached 2 kW, distributed approximately equally between the two consumers. The recorded operating parameters at this stage were as follows:

• Micro-turbine throttle position: 75%;
• Generator rotational speed: approximately 6000 RPM;
• EDF 1 throttle position: 55%;
• EDF 2 throttle position: 52%.

The final experimental test investigated the system’s ability to achieve a total electrical power output of 3 kW, with the evolution of the relevant parameters presented in Figure 8. The testing methodology followed a sequence similar to that of the previous experiment up to the 2 kW threshold. In the range between 2 and 2.5 kW, the throttle of the first power consumer (EDF 1) was gradually increased. Beyond 2.5 kW, the adjustment of EDF 1 was halted, and control was transitioned to EDF 2, whose throttle was incrementally raised to guide the system toward the target output in a stable and controlled manner. The recorded operating parameters at the moment the 3 kW output was reached were as follows:

• Micro-turbine throttle position: 88%;
• Generator rotational speed: approximately 6300 RPM;
• EDF 1 throttle position: 59%;
• EDF 2 throttle position: 64%.

For enhanced clarity and accessibility, we have compiled all key experimental results in

Table 1. Key experimental results.

Power [W]	Turbine Engine Throttle [%]	Generator Rotational Speed [rpm]	Consumer Throttle
600	42	5300	EDF_1 = 28%
			EDF_2 = 0%
1000	50	5500	EDF_1 = 35%
			EDF_2 = 0%
2000	75	6000	EDF_1 = 55%
			EDF_2 = 52%
3000	88	6300	EDF_1 = 59%
			EDF_2 = 64%

below.

4. Discussion

The results of the experimental testing confirm that the developed system is capable of fully autonomous operation. Specifically, the MTG’s SBC independently regulates the throttle of the turbine engine to maintain the generator output voltage in the vicinity of the target value of 48 VDC. Moreover, due to the structural and control optimizations of the test bench, the system successfully achieved a power output of 3000 W at a turbine throttle position of approximately 88%. Based on the overall experimental campaign, it is estimated that the current configuration is capable of delivering up to a maximum of 3500 W of electrical power when the micro-turboprop engine operates at full throttle.

However, a performance analysis reveals that while the turbine engine produces a mechanical power output of approximately 5.2 kW under these conditions, the corresponding electrical power delivered is limited to around 3.5 kW, resulting in an overall conversion efficiency of approximately 67%. The observed reduction in efficiency can be attributed to two primary factors: the selection of an electric generator with an excessively high KV rating and the specific electrical power requirements of the designed UAV on which the hybrid power sources will be mounted. The UAV was designed to achieve optimal performance at an electrical power output of 3 kW. Consequently, generating power beyond this threshold was deemed unnecessary for meeting the operational objectives. The UAV platform designed to incorporate the proposed hybrid power system is presented in Figure 9, with its principal technical specifications detailed in

Table 2. Main characteristics of the hybrid-system-powered UAV.

Parameter	Specification
UAV Type	Quadcopter
Propulsion System Layout	Coaxial
MTOW	40 kg
Maximum Load Capacity	3 kg
Electric Power	3 kW
Flight Autonomy (full electric)	10.6 min
Flight Autonomy (hybrid power source)	124 min

.

Furthermore, the decision to limit power output was influenced by operational constraints of the micro-turbine engine, which had already reached its maximum safe throttle setting. Any additional power enhancement would consequently require substitution of the turbine engine itself, as the current configuration operates at its designed performance threshold.

To improve energy conversion efficiency, a viable approach would be to optimize the generator’s KV factor in order to enhance mechanical-to-electrical energy conversion.

Given that generator speed can be estimated as the product of the generator voltage and the KV factor, the U15II KV100 generator is theoretically capable of producing 48 VDC at an unloaded rotational speed of 4800 RPM. However, under load conditions, the effective KV factor increases progressively, reaching a value of approximately 131 at a measured electrical power output of 3100 W, as illustrated in Figure 10.

The $K_v$ constant, expressed in RPM per volt (RPM/V), defines the relationship between the rotational speed of a motor or generator and the voltage it produces or requires, representing the number of revolutions per minute generated per applied volt under no-load conditions; its mathematical expression is provided below.

$$K_v={\displaystyle \frac{n}{V}}$$

(1)

Ohm’s Law in the context of electrical power is as follows:

$$P=I·V$$

(2)

The voltage expression from Equation (1) can now be substituted into Equation (2) to yield the following formulation:

$$P=I·{\displaystyle \frac{n}{K_v}}$$

(3)

Based on Equation (3), it follows that for a constant rotational speed and current, the electrical power output increases as the voltage decreases, which implicitly corresponds to a lower $K_v$ value.

Based on all of the above, by reducing the $K_v$ factor from 100 to 80, the $K_v$–power characteristic undergoes a significant shift, resulting in a displacement of the operating point. Under this new configuration, the electrical power output of 3100 W is no longer achieved at 6300 RPM but instead at approximately 5560 RPM. This adjustment effectively lowers the rotational speed required to reach the target output voltage of 48 VDC, reducing the necessary speed by roughly 1000 RPM. Consequently, by appropriately tuning the $K_v$ factor, the overall performance of the system can be improved, enhancing the conversion efficiency and maximizing the output of useful electrical energy.

Conclusively, this study has made several key contributions to the development and experimental validation of the usage of a micro-turboprop engine as a hybrid power source on a UAV:

• Structural and Electrical Optimization—The test bench configuration presented in [14] was enhanced through improvements to both its mechanical architecture and power electronics components;
• Control Algorithm Refinement—The governing control logic was optimized to achieve more precise operational management of the turbogenerator system;
• Performance Characterization—Comprehensive experimental testing was conducted to evaluate and quantify the system’s operational parameters under controlled conditions.

As part of future experimental work, the feasibility of replacing the U15II KV100 generator with the U15II KV80 model, both of which are manufactured by T-Motors (Hong Kong RM C, 6/F, Nanchang, China), will be evaluated, with the primary objective of improving the energy efficiency of the MTG test bench.

5. Conclusions

This study presents the final stage of development and experimental validation of a hybrid power generation system designed for integration into multirole unmanned aerial vehicles (UAVs). Building upon previous iterations, the system now incorporates a micro-turboprop engine coupled with an upgraded electric generator, a dual-consumer electrical load configuration, and a refined fuzzy logic-based control algorithm. Structural improvements to the test bench and the implementation of autonomous control capabilities enabled stable operation at electrical power levels up to 3 kW, with projections suggesting a maximum deliverable output of approximately 3.5 kW when the turbine engine is operated at full throttle.

The test results presented confirm that the single-board computer (SBC) successfully regulates the turbine throttle to maintain the output voltage at the nominal value of 48 VDC under variable load conditions. Despite the micro-turboprop engine delivering a mechanical power output of 5.2 kW, the observed electrical conversion efficiency reached approximately 67%, highlighting areas for improvement in the generator’s electromechanical characteristics.

Overall, the results strongly support the integration of hybrid power systems in UAV platforms, contributing to extended flight endurance, improved power management, and greater operational autonomy.