Study and Verification of a Fuzzy-Following Energy Management Strategy for Hybrid Tractors

Abstract

Tractors operate under varying and unpredictable conditions, making energy management strategies insufficient for maintaining system power dynamics, which often leads to reduced traction power and overall efficiency. To overcome this challenge, a fuzzy-following energy management strategy was developed. This approach utilizes fuzzy control based on energy following to optimize the tractor’s energy output, ensuring more stable power delivery. A target tractor model was constructed using CRUISE, and joint simulations were carried out via the CRUISE-Simulink interface. The results demonstrated that the fuzzy-following strategy stabilized the battery’s state of charge (SoC) and improved fuel economy. The strategy was implemented for controlling a hybrid tractor, and its effectiveness and stability were validated through drivetrain system tests and real vehicle trials under light load, plowing, and power harrowing conditions, successfully achieving power balance under these diverse operating scenarios. Comparative tests between the hybrid tractor using the fuzzy-following strategy and a powershift tractor revealed that the hybrid tractor exhibited superior plowing efficiency and fuel economy under plowing and power-harrowing conditions.

1. Introduction

Against the backdrop of increasingly acute contradictions in the supply and demand of agricultural diesel and the worsening environmental pollution problems caused by tractors, the “Made in China 2025” implementation outline has explicitly listed intelligent agricultural machinery and equipment as one of the most important development areas [1]. As the main agricultural machinery equipment in production, the development of new energy for tractors is of great significance for the transformation of agricultural machinery equipment towards new energy power, leading to the application of more clean energy in agricultural mechanization and promoting the green and sustainable development of agricultural production [2]. In the tractor industry, global manufacturers, such as John Deere and Fendt, have introduced fully electric tractors at trade shows. In China, institutions such as the National Intelligent Agricultural Machinery Innovation Center have launched pure electric tractors. However, the performance of pure electric tractors is limited by the capacity of high-voltage battery packs, preventing them from meeting the energy demands of high-load, continuous field operations. Hybrid tractors, which combine a diesel engine and a battery pack as dual power sources, can fulfill the energy requirements for high-load, prolonged fieldwork. By employing a well-designed energy management strategy, hybrid tractors can significantly reduce energy consumption and enhance fuel efficiency.

An energy management strategy is crucial for the performance of series hybrid tractors. There are two main types of energy-saving control strategies: rule-based control strategies and optimization-based control strategies [3,4]. In the research on rule-based control strategies, one study [5] proposed a fuzzy control-based energy-saving control strategy for electric vehicles and electric tractors based on a series of hybrid power tractors, which was verified through simulation. The results showed that when using the fuzzy control-based energy-saving control strategy, the State of Charge (SoC) curve of the power battery changed smoothly. In another study [6], a fuzzy reasoning-based energy-saving control strategy was proposed for a series of diesel-electric hybrid tractors, with the engine demand power determined according to self-set fuzzy reasoning rules. The simulation results indicated a 12.3% improvement in fuel economy compared to the power-following control strategy. In terms of optimization-based control strategies, one study [7] gave an energy control strategy based on Pontryagin’s Minimum Principle, which, through simulation, reduced fuel consumption by 11.20% compared to the energy-saving control strategy based on optimal economy. Study [8] using the dynamic programming model and the fundamental theory of dynamic programming and, simulated by MATLAB/Simulink 2020b, when compared to the power-following control strategy, fuel consumption is reduced by approximately 6.3%, and the economy is significantly improved. Most of the studies are limited to simulation analysis, lacking field test data to support their findings. Furthermore, some energy management strategies, especially those based on optimization control strategies that rely on complex algorithms and models, are challenging to implement in real-time during operations. A simple and effective energy management strategy that can be implemented on a microcontroller for new energy tractors is required. The energy management strategy should be verified in field tests to demonstrate its robustness and effectiveness.

In this paper, an energy management strategy based on fuzzy following, which is a rule-based strategy, was developed. Its feasibility was initially validated through joint simulations using Simulink and CRUISE to compare the strategy with the traditional power-following strategy. The Simulink model was subsequently converted to C code and implemented in the controller of a 220-horsepower (hp) hybrid tractor powertrain. The strategy’s effectiveness was confirmed through powertrain bench tests and field tests. Continuous optimization was performed based on feedback from field operations, ultimately resulting in a reliable energy management strategy for hybrid tractors.

2. Energy Management Strategy Development

2.1. Tractor System Architecture

In China, high-power, heavy-duty tractors are mainly used for plowing, power harrowing, or combined operations involving both traction and harrowing. Traction capability is a key performance metric for tractors. Figure 1 illustrates the traction output and engine power of a 220-hp tractor during plowing obtained on the YTO LF2204 power-shift tractor. For hybrid tractors to deliver equivalent traction and power output characteristics, their powertrain design and parameters must be tailored to meet these performance requirements.

By incorporating an electric motor into the powertrain, the parallel hybrid power system can enhance the engine’s economic performance during acceleration. However, the powertrain still requires a complex gearbox to achieve the operational performance of the tractor, and it cannot achieve decoupling between the power take-off (PTO) and the driving speed. A series-parallel hybrid power system also increases the complexity of the tractor’s powertrain system, which is not conducive to simplifying the tractor’s powertrain system. Therefore, a series hybrid power structure was chosen as the hybrid structure applied in this paper, and the established hybrid structure is shown in Figure 2. The generator set, consisting of an engine and generator, functions as the primary energy source, supplying power to the tractor’s electric drive system. The power battery system acts as a supplementary energy source, enhancing the drive system’s transient response and providing power feedback. The e-motor has a hollow axle so that the power take-off (PTO) axle, which is driven by an engine directly, can pass through the e-motor. The e-motor drives the gearbox and transfers the power to the tractor’s wheels. In the architecture, PTO speed and vehicle speed are decoupled. It is more convenient for operators to make the PTO speed and vehicle speed match because operators can control both speeds independently, which is more efficient for farming operations. The width speed range of the e-motor and its power output character can also make the tractor powertrain structure simpler than those of traditional tractors, in which only one or two gear shifts can cover most farming operations. In addition, a small capacity high voltage (HV) battery pack is used to maintain the high voltage of the system, which can supply and absorb power when the load changes.

As depicted in Figure 1, the average traction force of 220 hp during plowing is approximately 45 kN. Given an average plowing speed of 10 km/h, the tractor’s tractive power is roughly 127 kW. Factoring in a 10% slip loss and a transmission efficiency of 95%, the hybrid tractor’s drive motor power requirement is calculated to be 147 kW. Therefore, selecting a motor with a rated power of around 150 kW is sufficient to meet the driving power demands of the 220-hp tractor.

The generator must be capable of supporting the motor’s rated power while considering the operating efficiency of both the generator and motor. The generator’s rated power, P_gen, is determined by the following formula:

P_gen = P_motor/(η_motor × η_gen)

(1)

where P_motor is set at 147 kW, and η_motor and η_gen are both set at 0.95. This results in a generator rated power of 162 kW. A diesel engine with a rated power of 162 kW can be selected to fulfill the PTO output requirements.

The main parameters for the hybrid tractor’s powertrain are detailed in

Table 1. Powertrain parameters of the hybrid tractor.

Item	Parameter	Value
Diesel Engine	Rated power (RPM)	162 kW@2000 rpm
HV battery pack	Capacity	13 kWh
	Rated voltage	540 V
	Battery type	NMC
Drive motor	Rate power	150 kW@2000 rpm
	Maximum speed	6000 rpm
Generator	Rate power	162 kW@2000 rpm
	Maximum speed	4000 rpm

.

2.2. Energy Management Strategy

The tractor developed in this project employs a typical series hybrid architecture. Unlike automobiles, hybrid tractors are equipped with relatively small battery packs that cannot sustain continuous power output for extended periods under high loads. Additionally, the operational demands of tractors differ significantly from those of automobiles, which require short bursts of power for acceleration and hill climbing. In contrast, tractors must operate under full load for prolonged durations. As a result, the hybrid power strategies detailed in [9,10,11] that gave some energy management strategies for road vehicles are unsuitable for tractor applications. Hybrid tractors and series or range extender hybrid vehicles thus necessitate distinct energy management strategies.

The core concept of the energy management strategy for hybrid tractors is to maintain the battery pack’s state of charge (SoC) within a specified range of charging and discharge it at low currents during operation, which enhances battery longevity. A balance between the power generated by the generator and the power required by the drive motor is maintained, with the battery pack supplying power to manage transient load fluctuations and ensure a dynamic power output.

The energy management control system for the hybrid tractor employs a fuzzy-following strategy. In this approach, the generator’s power output adjusts in response to changes in the motor’s power demands. The battery pack serves as an auxiliary power source, compensating for power deficits or absorbing excess energy from the generator while also providing feedback to the motor. The generator’s output is determined based on the operating conditions of both the drive motor and the battery pack, and fuzzy control is applied to optimize power management.

During towing operations, the maximum available power for the drive motor is expressed as:

P_motor = P_gen + P_bat

(2)

where P_motor is the drive motor’s available power, P_gen is the generator’s current power output, and P_bat represents the permissible discharge power of the battery pack. The latter serves as a power reserve to regulate the tractor’s power output when encountering sudden load changes.

The generator’s power output is given by:

P_gen = P_motor + P_bat + P_com

(3)

where P_gen is the power required from the generator, P_motor is the motor’s drive power, and P_bat is the battery pack’s required charging power, which is non-zero when the SoC falls below a set threshold and zero when above the threshold. P_com represents the compensation power calculated on the battery pack’s SoC and current.

Power compensation is managed using fuzzy control, which generates control variables through fuzzy logic and reasoning [12]. Fuzzy control is particularly well-suited to scenarios where an exact mathematical model is unavailable due to its robustness and ease of construction [13,14]. To facilitate code generation and execution within a microcontroller, simple triangular or trapezoidal functions were used as the membership functions.

The power compensation range was set between −10 and 10 kW and processed through fuzzy logic to determine the power compensation membership function. The high-voltage battery pack’s current range was set from −20 to 20 A, with out-of-range values capped at boundary limits. The range of the battery pack SoC was [0, 100]%. Figure 3, Figure 4 and Figure 5 show the membership functions for power compensation, battery current, and SoC after fuzzy processing, respectively. Based on these membership values, fuzzy rules were established and summarized in

Table 2. The fuzzy control rules of the power compensation.

Compensated Power		Battery Pack Current
		BL	L	M	H	BH
SoC	BL	BH	BH	H	M	M
	L	BH	H	M	M	M
	M	H	M	M	M	M
	H	H	M	L	L	L
	BH	M	M	BL	L	BL

.

The fuzzy inference was processed using the MAX-MIN method, and the final power compensation value was calculated using the centroid defuzzication method. This optimized the battery pack’s charging and discharging performance. The final value for P_gen was used to control the generation’s output, powering the tractor’s electric drive system.

3. Simulation and Experimental Study

3.1. CRUISE Vehicle Modeling

CRUISE is primarily designed to simulate the forward performance of vehicles under various driving conditions [15,16,17]. It employs a flexible modular modeling approach [18,19], allowing for a rapid establishment of vehicle models with diverse drivetrain configurations. Based on the system architecture and specification of the hybrid tractor, the individual components were selected within the CRUISE 2015 software. The powertrain and high-voltage electrical systems were interconnected through mechanical and electrical connections to create a comprehensive vehicle powertrain model. Figure 6 illustrates the simulation platform for the series hybrid tractor.

CRUISE offers a data interface with MATLAB/Simulink, enabling the execution of control models developed in MATLAB/Simulink within the CRUISE environment. This project utilized MATLAB DLL [20,21] for the simulation analysis.

3.2. Definition of Simulation Test Operating Conditions and Result Analysis

Series hybrid tractors must maintain a power balance between power generation and propulsion to prevent overcharging or excessive discharging of the power battery pack. This study primarily focused on heavy load driving conditions in the field to assess whether the control strategy can effectively achieve power balancing during such operations.

Plowing is one of the most common and demanding tasks for tractors; thus, it was selected as the operational condition for the simulation calculations. The empirical formula for determining the traction force required for plowing is:

F_T = z × b1 × h_k × k (N)

(4)

where z represents the number of plowshares, b1 is the width of a single plowshare (in cm), h_k denotes the depth of tillage (in cm), and k is the soil-specific resistance (in N/cm²).

For 220-hp farming tractors, a 5-share plow is commonly utilized. Using a single plowshare width of 40 cm, a typical plowing depth of 30 cm, and a soil-specific resistance of 8 N/cm², the plowing resistance calculated using the formula yields 45,000 N. This resistance was employed as the driving resistance for the 220-hp hybrid tractor during plowing.

The vehicle traction model in CRUISE was configured to operate in a vehicle-independent function mode using the equation:

Fb = 45,000 + (0.46 × V²) (N)

(5)

This equation represents a velocity-dependent second-order resistance model. The simulation of the tractor’s plowing condition was executed using this resistance model.

Within CRUISE, the plowing condition was established, incorporating a cyclic running condition designated as “Farming”. Typically, tractors operate at speeds of 8–10 km/h while plowing; therefore, the maximum operating speed was set to 10 km/h with a permissible speed error of ±1 km/h. The duration of a single cycle was set at 300 s, with a corresponding running distance of approximately 900 m for one plowing operation.

The initial battery level was established at 60%, and operation termination conditions were defined at a low battery level of 20% to prevent potential malfunctions in the energy management strategy due to continued operation at low SoC, as well as considering the maximum capacity of the engine’s fuel tank. The SoC at which the power battery pack begins charging was set at 40%, and charging ceases at 80%. Simulations were conducted using both the fuzzy-following strategy and the energy-following strategy. The evaluation of the fuzzy-following control strategy’s impact on the battery system and fuel economy was conducted through analysis of the SoC, fuel consumption, and current curves. Relevant operational results were obtained from simulation calculations performed in CRUISE.

By simulating and comparing the battery SoC and fuel consumption under the fuzzy-following strategy versus the energy-following strategy, the economic benefits and battery longevity of the tractor under different energy management strategies were assessed. Under the fuzzy-following strategy, the battery pack’s SoC declined at a slower rate due to compensation based on the battery pack’s current and SoC. This alignment kept the average current drawn from the battery pack closer to the target value, resulting in fewer charge and discharge cycles, thus effectively extending the battery pack’s service life. Furthermore, with identical fuel consumption of 420 L, the tractor using the fuzzy-following strategy was able to operate for approximately 1200 s longer than under the energy-following strategy, representing a nearly 5% improvement in the economy, which is shown in Figure 7.

The comparison of the current curves, which is shown in Figure 8, indicated that the fuzzy-following strategy effectively minimized both the charging and discharging currents of the battery in the same load condition. Particularly when the load remained stable, the output current from the battery system approached 0 A, aligning with the vehicle’s energy control objectives. The reduction in current not only decreases heat generation but also enhances the longevity of the battery system.

Figure 9 illustrates the speed, power, and efficiency of the electric motor over three cycles during the simulation. At maximum plowing speed, the electric motor operated at approximately 3200 rpm, achieving a peak efficiency of 95.63%. It functioned within the high-efficiency range of the electric drive system, delivering around 135 kW of power. During acceleration from a standstill, the electric motor could produce up to 145 kW. Selecting an electric motor with a rated power of 150 kW could adequately support this acceleration phase. In steady-state conditions, the electric motor maintained a power reserve of about 10% to counter sudden increases in external resistance and sustain speed. Under plowing conditions, the electric motor operated within its most efficient range, thereby minimizing energy loss and enhancing the tractor’s overall economy during plowing tasks.

Figure 10 presents the speed, power, and efficiency curves of the generator across three cycles, highlighting changes before and after the battery charging power was integrated during the simulation. When the battery’s SoC fell below 40%, charging power was allocated to the generator as part of the energy management strategy, leading to an increase in the battery’s SoC. The generator maintained an operational speed of around 2000 rpm throughout the simulation, achieving an efficiency of approximately 95% during stable operations, which is indicative of a high-efficiency range. During vehicle acceleration, the generator’s maximum power output reached 151 kW, which increased to 159 kW after incorporating the battery charging power. The chosen generator met the tractor’s power demands even under extreme operating conditions. The energy management strategy effectively regulated the battery pack, the generator’s speed, and the electric motor, aligning with design expectations and maintaining relatively high power generation efficiency.

3.3. Hybrid Powertrain Bench Test

We integrated the energy management strategy model with other vehicle control modules, converting them into the vehicle controller’s software and generating code using TargetLink 5.3. This code was then integrated with the basic software and downloaded to the vehicle controller hardware, utilizing the NXP MPC5744p as the master controller. A bench test of the hybrid tractor’s powertrain was conducted to further validate the effectiveness of the energy management strategy and the reliability of the simulation results obtained from the CRUISE software.

Figure 11 illustrates the bench test for the hybrid tractor’s powertrain. The traction electric motor served as the power source, replacing the diesel engine, to evaluate the energy management strategy of the powertrain. During the test, CAN bus monitor software CANoe 11 was used to monitor CAN bus data, allowing for the assessment of the system’s operational status. The resulting curves for drive motor speed, power battery pack current, generator current, and electric motor current are presented in Figure 12.

In Figure 12, the curves demonstrate that rapid changes in motor power due to acceleration and deceleration lead to corresponding fluctuations in the power battery pack’s current. The generator was able to quickly respond to these power changes, supplying the necessary current to drive the motor. Any insufficient or excessive was either supplemented or absorbed by the battery pack. The sudden current changes in the battery pack closely aligned with those observed in the CRUISE simulation, with the maximum discharge current not exceeding 25 A and the maximum pulse charging current limited to 20 A. Once the motor reached a steady power output state, the battery pack’s current typically fluctuated by about ±1 A. The energy management strategy implemented during the bench test effectively managed the output states of the motor, generator, and battery pack, aligning with the design objectives and the outcomes of the CRUISE simulation.

3.4. Hybrid Tractor Field Test

In 2023 and 2024, we applied one prototype in the northwest region of China and another prototype in the northeast region of China for a 5-month validation of field test. The two prototypes completed nearly 1000 h of work and 900 hectares(ha) of plowing and power harrowing operations, achieving excellent results.

Table 3. The working conditions of the hybrid tractors.

Item	Prototype 1	Prototype 2
Plowing area (ha)	555	437
Power harrowing area (ha)	464	443
Working time (h)	964	937
Soil type	sand	clay
Plowing speed (km/h)	9~13	9~12
Power harrowing speed (km/h)	8~10	8~10

shows the conditions of the test.

During the field test period, we collected the working data of the hybrid tractor through CAN bus to assess the effectiveness of the fuzzy-following energy management strategy and the reliability of the tractor’s energy management system during heavy-duty traction and PTO operations, specifically in plowing and power-harrowing tasks.

As illustrated in Figure 13, during the plowing operation, when the load was stable, the battery pack experienced a slight discharge, with current levels ranging from 2 to 3 A. Following a sudden drop in the motor load, a pulse charge of 86 A occurred, which the battery pack absorbed. According to the charging parameters of the battery pack, this pulse charging current was below 112 A, ensuring that the battery pack’s service life was not adversely affected. The tractor maintained a plowing speed of 9 to 10 km/h, with motor drive power hovering around its maximum output of 150 kW. As the battery pack gradually discharged, the minimum voltage of an individual battery cell dropped to below 3.75 V, resulting in a reduction of approximately 5 kW in drive power. Some of the current produced by the generator was used for battery charging, keeping the charging current within acceptable limits. The speed remained within the high-efficiency range for farming. The energy management strategy effectively controlled the energy flow between the battery pack, generator, and motor during the plowing operation. The primary objectives included maximizing the efficiency of both the motor and generator, managing the SoC of the battery pack, and enhancing the energy utilization efficiency of the hybrid tractor to achieve energy savings.

Figure 14 depicts the operating status during the PTO’s power-harrowing operation. The PTO operated at 540 rpm throughout the task. When the engine load approached 100% of its rated power, the electric motor’s output power decreased, leading to a reduction in the vehicle’s driving speed. However, this did not affect the PTO’s rotational speed, which remained constant at 540 rpm, ensuring its operational effectiveness. During the entire operation, the instantaneous discharge current from the battery pack was kept below 10 A, while the instantaneous charging current did not exceed 20 A. Overall, the charging and discharging levels of the batteries were maintained at a lower threshold, promoting the longevity of the battery pack.

The field tests demonstrated that the energy management strategy successfully facilitated energy distribution and regulation for the battery pack, generator, motor, and PTO across various conditions of heavy-duty traction and PTO operations with specific traction loads. By optimizing energy management parameters, the strategy effectively improved operational efficiency in diverse environments.

3.5. Operation Comparison with Powershift Tractor

The performance of the hybrid tractor can be assessed through comparative testing with a powershift tractor of equivalent hp. By operating both tractors in a standardized field with identical farming implements, characteristics related to power, fuel economy, and work efficiency can be evaluated.

The comparative tests shown in Figure 15 and their results, given in

Table 4. The result of comparison between the hybrid tractor and the powershift tractor.

Item	Plowing		Power Harrowing
Tractor Type	Hybrid	Power Shift	Hybrid	Power Shift
Operation depth (cm)	32	32	20	20
Operation width (cm)	200	200	400	400
Operation speed (km/h)	8~12	8~11	6.5~7.0	6.0~6.5
Operation time (h)	0.56	0.59	0.81	0.72
Operation Area (ha)	0.894	0.924	1.848	1.4347
Fuel consumption (kg)	16.23	17.78	17.87	15.95
Initial battery voltage (V)	576	\	575	\
End battery voltage (V)	575	\	575	\
Fuel consumption per hour (kg)	28.98	30.14	22.06	22.15
Fuel consumption per hectare (kg)	18.16	19.37	9.67	11.11
Operation efficiency (ha/h)	1.596	1.566	2.281	1.993

, revealed that the plowing efficiency of the hybrid tractor was comparable to that of the powershift tractor, with the hybrid exhibiting approximately 20% lower fuel consumption per unit area. In the power harrowing operation, the hybrid tractor demonstrated nearly 14% greater efficiency than the powershift tractor, alongside a reduction of about 14% in fuel consumption per unit area. The final voltage of the battery is essentially the same as the initial voltage, indicating that the battery pack undergoes a slight discharging and charging process during operation, and the charging and discharging of the battery pack have virtually no impact on the diesel engine. Overall, the hybrid tractor exhibited clear economic advantages over the powershift tractor.

4. Discussion

The fuzzy-following energy management strategy is a rule-based energy management method. Studies [3,4] used fuzzy control to establish fuzzy rules for battery State of Charge (SoC) and engine required power. The control method in study [3] improved fuel economy by 4.35% compared to the energy-following strategy, and by using a fuzzy-following approach, fuel economy was enhanced by 5% in simulations; study [4] established fuzzy rules for tractor required power, resulting in a fuel economy improvement of about 12.3%. This method did not consider the thermal model of the battery system, and continuous charging and discharging of the battery pack could lead to battery overheating, affecting operational performance or necessitating an additional thermal management system to maintain the temperature of the battery system. Studies [5,6] applied optimization-based methods. Study [5] proposed an energy control strategy using Pontryagin’s Minimum Principle, achieving an energy saving of 11.20%, while study [6] used dynamic programming algorithms to improve fuel consumption by 6.3% compared to the energy-following strategy. In the simulations of studies [5,6], it can be observed that the battery pack is continuously charged and discharged, with a 10Ah power battery often operating in a discharge range of 40–80 A, leading to severe heat generation and a reduction in battery life due to excessive charging and discharging power. The method used in this paper is a balanced approach that takes into account various practical operational factors, improving the fuel economy of diesel engines and maintaining the charging and discharging of the battery pack at a lower level, demonstrating high stability after long-term field testing. Therefore, as operational data accumulates, we will attempt to use machine learning to optimize control parameters, identify working modes, and further enhance work efficiency.

5. Conclusions

This paper analyzed the traction force and power requirements of high-hp tractors and the drivetrain structure and operating characteristics of hybrid tractors. A fuzzy-following energy management strategy was developed, and a hybrid tractor simulation model was constructed in CRUISE software. By interfacing CRUISE with Simulink, the fuzzy-following strategy is compared with a traditional energy-following strategy. Results showed that the fuzzy-following strategy was 5% more economical and extended the power battery’s service life. Bench tests validated the strategy’s regulation of power balance under light loads, confirming the accuracy of the simulation results. In the field operation tests, the strategy successfully managed energy distribution among the engine, generator, motor, and battery during plowing and harrowing. Performance comparison tests demonstrated that the hybrid tractor showed improvements in operating efficiency and economy over the powershift tractor. The fuzzy-following energy management strategy is a control method suitable for hybrid tractor applications. This simple and reliable control method has been verified through field tests, effectively improving the energy response and work efficiency of hybrid tractors and providing an excellent control solution for the marketization of hybrid tractors.