Energy management method of fuel cell hybrid electric vehicle based on II-type fuzzy decision framework

Abstract

The invention provides an energy management method of a fuel cell hybrid electric vehicle based on an II-type fuzzy decision framework. According to the method, a two-stage framework combining offline optimization and online application is adopted. In the off-line stage, historical data of different driving modes are acquired to train a driving mode recognition model, and a multi-objective optimization algorithm is utilized to optimize parameters of the II-type fuzzy logic controller for different driving modes to generate a corresponding optimization parameter set. In the online stage, the current driving mode is recognized in real time, a corresponding parameter set is called to configure a fuzzy controller, a fuel cell power instruction is output according to the vehicle demand power and the battery charge state, and efficient energy distribution is achieved through battery compensation residual power. According to the invention, the adaptability, robustness and multi-target collaborative optimization capability of energy management are improved.

Description

Technical Field

[0001] This invention belongs to the field of new energy vehicle energy management technology, and in particular relates to an energy management method for fuel cell hybrid electric vehicles based on a type II fuzzy decision framework. Background Technology

[0002] In recent years, with the transformation of transportation systems towards "intelligentization" and "sustainability," energy management strategies for fuel cell hybrid electric vehicles have become crucial for improving vehicle economy, durability, and safety. Traditional energy management methods are mainly divided into three categories: rule-based, optimization-based, and learning-based. While rule-based methods offer good real-time performance, they struggle to adapt to complex dynamic conditions and are prone to suboptimal decisions at state boundaries. Optimization-based methods, such as dynamic programming, can achieve global optima but have a high computational burden, while methods like model predictive control heavily rely on accurate system models, making practical deployment difficult. Learning-based methods, while possessing good adaptability, are still insufficient in handling high-order uncertainties such as fuel cell system parameter fluctuations and sensor measurement errors, and struggle to effectively coordinate multiple conflicting objectives such as hydrogen economy, component lifespan, and system safety. Therefore, constructing an energy management framework that can adapt to real-time driving condition changes, effectively handle system uncertainties, and simultaneously achieve multi-objective optimization has become a pressing technical problem in this field. Summary of the Invention

[0003] This invention proposes an energy management method for fuel cell hybrid electric vehicles based on a type II fuzzy decision framework to address the problems existing in the prior art.

[0004] To achieve the above objectives, this invention provides an energy management method for fuel cell hybrid electric vehicles based on a Type II fuzzy decision framework, employing a two-stage architecture combining offline optimization and online application, including:

[0005] During the offline phase, historical driving data under different driving modes are acquired, and feature extraction is performed on the historical driving data to train the driving mode recognition model. Based on a multi-objective optimization algorithm, the parameters of the Type II fuzzy logic controller are optimized for at least two driving modes to obtain an optimized parameter set corresponding to each driving mode. The optimized parameter set is used to configure the controller during the online phase.

[0006] During the online phase, vehicle driving data is collected in real time, and the current driving mode is identified using the driving mode recognition model. Based on the identified current driving mode, the corresponding optimized parameter set obtained in the offline phase is called to configure the Type II fuzzy logic controller. Based on the configured Type II fuzzy logic controller, the fuel cell power command is output according to the vehicle's power demand and the battery's state of charge. Based on the fuel cell power command and the vehicle's power demand, the remaining power demand is compensated by the battery to complete the power allocation.

[0007] Optionally, feature extraction of historical driving data during the offline phase includes: extracting segments of historical driving data within a preset time window and extracting vehicle speed and acceleration statistical features from those segments.

[0008] Optionally, the driving mode recognition model is a classification model built based on an ensemble learning algorithm, used to output driving mode classification results based on vehicle speed statistical features and acceleration statistical features.

[0009] Optionally, the Type II fuzzy logic controller uses the vehicle's power demand and the battery's state of charge as input variables, and the fuel cell power command as the output variable.

[0010] Optionally, the parameters of the Type II fuzzy logic controller include membership function parameters and fuzzy rule parameters.

[0011] Optionally, the multi-objective optimization algorithm optimizes the parameters of the Type II fuzzy logic controller with fuel cell efficiency, equivalent hydrogen consumption, and component aging rate as optimization objectives.

[0012] Optionally, the optimized parameter set obtained in the offline phase includes parameter sets corresponding to the high-speed driving mode, smooth driving mode, and congested driving mode, respectively.

[0013] Optionally, in the online phase, power allocation based on the configured Type II fuzzy logic controller includes: obtaining fuel cell power commands through fuzzy inference based on the input vehicle power demand and battery state of charge.

[0014] Optionally, the offline phase of parameter optimization based on a multi-objective optimization algorithm includes: performing simulations based on a constructed system model that includes a fuel cell model, a battery model, and a vehicle dynamics model, and evaluating the optimization objectives on the simulation results to drive the parameter optimization process.

[0015] Optionally, the operating data of the fuel cell and battery are recorded and used as reference data for subsequent offline optimization of the driving mode recognition model and the parameters of the Type II fuzzy logic controller.

[0016] Compared with the prior art, the present invention has the following advantages and technical effects:

[0017] This invention integrates driving mode recognition, Type II fuzzy logic control, and multi-objective optimization to construct a two-stage adaptive architecture that combines offline optimization with online application. This effectively overcomes the shortcomings of traditional strategies in handling high-order uncertainties such as sudden changes in driving conditions, system parameter fluctuations, and measurement errors. Compared to existing methods, the proposed strategy exhibits superior overall performance across various test cycles. While ensuring system safety, it significantly improves energy economy and the lifespan of key components, while also possessing good real-time performance and robustness. This provides a feasible and effective solution for achieving intelligent and efficient energy management in fuel cell hybrid vehicles. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 This is a schematic diagram of the method according to an embodiment of the present invention;

[0020] Figure 2 The oc terminal voltage of this embodiment of the invention ( The graph shows the variation of internal resistance with SOC.

[0021] Figure 3 This is a schematic diagram of the driving cycle for training data in an embodiment of the present invention. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0024] Example 1

[0025] like Figure 1 As shown, this embodiment provides an energy management method for fuel cell hybrid electric vehicles based on a type II fuzzy decision framework. It employs a two-stage architecture combining offline optimization and online application, including:

[0026] During the offline phase, historical driving data under different driving modes are acquired, and feature extraction is performed on the historical driving data to train the driving mode recognition model. Based on a multi-objective optimization algorithm, the parameters of the Type II fuzzy logic controller are optimized for at least two driving modes to obtain an optimized parameter set corresponding to each driving mode. The optimized parameter set is used to configure the controller during the online phase.

[0027] During the online phase, vehicle driving data is collected in real time, and the current driving mode is identified using the driving mode recognition model. Based on the identified current driving mode, the corresponding optimized parameter set obtained in the offline phase is called to configure the Type II fuzzy logic controller. Based on the configured Type II fuzzy logic controller, the fuel cell power command is output according to the vehicle's power demand and the battery's state of charge. Based on the fuel cell power command and the vehicle's power demand, the remaining power demand is compensated by the battery to complete the power allocation.

[0028] Specifically, the following steps are included:

[0029] Step 1: Hardware platform setup and data acquisition;

[0030] This includes a fuel cell system platform, a battery testing platform, and data acquisition. The fuel cell system platform comprises a fuel cell, an air compressor, a hydrogen circulation pump, a unidirectional DC / DC converter, a host computer, a temperature chamber, and a NewareMHW-200 testing equipment. The battery testing platform consists of a battery pack of 206 10Ah / 3.3V lithium batteries and an NBT5V200AC8-T battery testing system. Data acquisition includes recording the fuel cell's power, hydrogen consumption, efficiency, and key parameters such as voltage, current, and SOC.

[0031] Step Two: System Model Construction;

[0032] Vehicle dynamics model:

[0033] ;

[0034] Where μ represents the rolling resistance coefficient, M represents the vehicle mass, g is the acceleration due to gravity, θ represents the road slope, and A is the vehicle's frontal area. It is air density. This represents the air drag coefficient, where v is the vehicle speed. It is the rotational mass correction factor, and 'a' represents the vehicle acceleration.

[0035] Fuel cell model: Calculating hydrogen consumption rate based on Faraday's law Fitting net power using experimental data The relationship between efficiency and hydrogen consumption was investigated, and a lookup table was created for rapid calculation.

[0036] ;

[0037] The efficiency of a fuel cell system is defined as the ratio of net power output to the energy density of the hydrogen consumed, and can be expressed as:

[0038] ;

[0039] LHV represents the lower heating value of hydrogen.

[0040] Lithium battery model: The Rint model was used, and the oc terminal voltage was calibrated experimentally. ) and internal resistance ( The curve showing the change of SOC is as follows: Figure 2 As shown.

[0041] Calculation of SOC change rate based on ampere-hour integral method:

[0042] ;

[0043] in, P represents the rated capacity of the battery system, while P represents the output power of the battery.

[0044] Step 3: Model training and parameter optimization (offline stage);

[0045] 3.1 Model training for Driving Pattern Recognition (DPR);

[0046] Data source: Three standard driving cycles (HWFET - highway, UDDS - smooth city, MANHATTAN - congested) were selected as training data, such as... Figure 3 As shown.

[0047] Data processing: Extract historical driving data segments within a preset time window. Use a window to randomly extract samples, starting at time T), [0,1]. The vehicle speed and acceleration statistical features of this segment were extracted through feature extraction: four features were calculated for each sample—average vehicle speed, idle speed ratio, average absolute acceleration, and acceleration standard deviation.

[0048] Model Training: The driving pattern recognition model is a classification model built on an ensemble learning algorithm, used to output driving pattern classification results based on vehicle speed and acceleration statistical features. Specifically, the AdaBoostM2 ensemble learning algorithm is used, which minimizes prediction error during training. The sample weights are iteratively updated (the weights of misclassified samples are increased), and a weighted pseudo-loss is calculated. and weak learner weights Finally, the classification result is output through weighted majority voting. Then, iterative optimization is performed, updating the sample weights and calculating the weighted pseudo-loss. The classification results are output through weighted majority voting. Finally, the hyperparameters are determined, and the model hyperparameters are optimized using a 5-fold crossover experiment.

[0049] 3.2 Multi-objective optimization of T2FLC parameters using MOGWO;

[0050] Simulations are performed based on a system model comprising a fuel cell model, a battery model, and a vehicle dynamics model. The optimization objectives are evaluated based on the simulation results to drive the parameter optimization process. A Type II fuzzy logic controller uses vehicle power demand and battery state of charge as input variables and fuel cell power command as output variable. The parameters of the Type II fuzzy logic controller include membership function parameters and fuzzy rule parameters. A multi-objective optimization algorithm optimizes the parameters of the Type II fuzzy logic controller with fuel cell efficiency, equivalent hydrogen consumption, and component aging rate as optimization objectives. The optimized parameter set obtained in the offline stage includes parameter sets corresponding to high-speed driving mode, smooth driving mode, and congested driving mode, respectively.

[0051] Optimization Parameters: The Multi-Objective Grey Wolf Optimizer (MOGWO) optimizes the T2FLC parameters for each of the three driving modes. A total of 85 parameters are optimized, including 60 membership function parameters (central). Standard deviation Lag factor Scale factor It includes 25 fuzzy rule-adjustable parameters. Among them, MOGWO generates the Pareto optimal frontier through population iteration and selects the parameter set that balances performance.

[0052] Optimization objectives: fuel cell effective efficiency, with penalty weight set; total equivalent hydrogen consumption; weighted component aging rate, with the aging weight ratio of FC to battery pack set to 4:1.

[0053] The T2FLC serves as the core of the EMS's decision-making and execution. Vehicle power demand and battery state of charge (SOC) are taken as inputs, and the output is the fuel cell power command (Pfc). The input and output are divided into five levels (VL / L / M / H / VH). The interval membership function is based on a type-1 Gaussian function, and the fuzzy inference process uses interval type-2 fuzzy sets to handle the input variables. A scaling factor is introduced to accommodate fluctuations in input variables under different operating conditions, and a lag factor is used to compensate for potential measurement errors in the FCHEV model.

[0054] The specific steps of the T2FLC algorithm are: (1) Initialize the population size N=100, the maximum number of iterations M=600, and the file size. , parameter boundary[ (2). Iterative optimization, for Update control parameters Choose leader Update the population position and constrain the parameters to be inside the boundary. (3) Generate the Pareto optimal frontier, select the parameter set for balancing performance, and store it according to the three driving modes (MH / MU / MC parameter library).

[0055] The T2FLC algorithm based on MOGWO parameter optimization is shown in Table 1.

[0056] Table 1

[0057]

[0058] Step 4: Real-time deployment and verification (online phase);

[0059] Based on the input vehicle power demand and battery state of charge, the fuel cell power command is obtained through fuzzy reasoning, which is mainly divided into three parts: (1) real-time data collection and processing of vehicle speed, SOC, fuel cell state, etc.; (2) driving mode recognition; (3) implementation decision and power allocation using MOGWO optimized T2FLC: based on the DPR results, the corresponding T2FLC parameters are retrieved from the offline optimization parameter library; fuzzy reasoning; power allocation.

[0060] Step 5: Testing, Verification, and Performance Evaluation;

[0061] The test cycles used were US06_HWY (highway), INDIA_URBAN_SAMPLE (smooth city), and NYCC (congestion) segments; CLTC and WLTC segments.

[0062] The performance indicator is SOC deviation. Total equivalent hydrogen consumption, aging rate of FC and battery packs, and total operating costs.

[0063] The comparative experimental group consisted of six other energy management strategies: EMS I used MOGWO optimized parameters to implement only the highway mode, EMS II implemented only the urban smooth mode, EMS III implemented only the traffic congestion mode, EMS IV used unoptimized parameters with T2FLC, EMS V used wavelet transform, and EMS VI used MPC. Tests were conducted under the same test cycles to verify the advantages of the proposed scheme. In terms of economic performance, this strategy achieved the lowest total operating cost in all test cycles, with a cost reduction of at least 0.083% compared to the optimal baseline scheme. The proposed strategy exhibited excellent charging continuity, with its state of charge change value (… The efficiency of the proposed EMS is the closest to zero among all methods. Furthermore, the proposed EMS also demonstrates outstanding performance in component protection and operational safety: both the fuel cell aging rate and the battery aging rate remain at extremely low levels. Simultaneously, this strategy virtually eliminates high-rate current operation of the battery, thereby reducing the risks associated with lithium deposition and thermal stress, and significantly improving the system's safety and lifespan. Specific comparative experimental data are shown in Table 2.

[0064] Table 2

[0065]

[0066] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An energy management method for fuel cell hybrid electric vehicles based on a type II fuzzy decision framework, characterized in that, A two-stage architecture combining offline optimization and online application is adopted, including: During the offline phase, historical driving data under different driving modes are acquired, and feature extraction is performed on the historical driving data to train the driving mode recognition model. Based on a multi-objective optimization algorithm, the parameters of the Type II fuzzy logic controller are optimized for at least two driving modes to obtain an optimized parameter set corresponding to each driving mode. The optimized parameter set is used to configure the controller during the online phase. During the online phase, vehicle driving data is collected in real time, and the current driving mode is identified using the driving mode recognition model. Based on the identified current driving mode, the corresponding optimized parameter set obtained in the offline phase is called to configure the Type II fuzzy logic controller. Based on the configured Type II fuzzy logic controller, the fuel cell power command is output according to the vehicle's power demand and the battery's state of charge. Based on the fuel cell power command and the vehicle's power demand, the remaining power demand is compensated by the battery to complete the power allocation.

2. The energy management method according to claim 1, characterized in that, Feature extraction of historical driving data in the offline phase includes: extracting segments of historical driving data within a preset time window and extracting vehicle speed and acceleration statistical features from those segments.

3. The energy management method according to claim 1, characterized in that, The driving mode recognition model is a classification model built based on an ensemble learning algorithm, which is used to output driving mode classification results based on vehicle speed statistical features and acceleration statistical features.

4. The energy management method according to claim 1, characterized in that, The Type II fuzzy logic controller uses the vehicle's power demand and the battery's state of charge as input variables, and the fuel cell power command as the output variable.

5. The energy management method according to claim 4, characterized in that, The parameters of the Type II fuzzy logic controller include membership function parameters and fuzzy rule parameters.

6. The energy management method according to claim 1, characterized in that, The multi-objective optimization algorithm optimizes the parameters of the Type II fuzzy logic controller with fuel cell efficiency, equivalent hydrogen consumption, and component aging rate as optimization objectives.

7. The energy management method according to claim 1, characterized in that, The optimized parameter set obtained in the offline phase includes parameter sets corresponding to high-speed driving mode, smooth driving mode, and congested driving mode, respectively.

8. The energy management method according to claim 1, characterized in that, During the online phase, power allocation based on the configured Type II fuzzy logic controller includes: obtaining fuel cell power commands through fuzzy inference based on the input vehicle power demand and battery state of charge.

9. The energy management method according to claim 1, characterized in that, The offline phase of parameter optimization based on multi-objective optimization algorithms includes: performing simulations based on a constructed system model that includes a fuel cell model, a battery model, and a vehicle dynamics model, and evaluating the optimization objectives based on the simulation results to drive the parameter optimization process.

10. The energy management method according to claim 1, characterized in that, The operating data of the fuel cell and battery are recorded and used as reference data for subsequent offline optimization of the driving mode recognition model and the parameters of the Type II fuzzy logic controller.

Images