Fuel cell vehicle energy management strategy optimization method oriented to working condition requirements

Abstract

The invention relates to a fuel cell vehicle energy management strategy optimization method facing working condition demands. The method comprises an equivalent factor iteration updating module, a current moment working condition identification module, a future time domain working condition demand prediction module and a power system execution module. Based on current working condition identification classification and demand working condition prediction in a future time domain, equivalent factors are updated from historical and future dimensions, optimal control sequence distribution of the hybrid power system power is obtained by using an F-AECMS algorithm based on the instantaneous local optimization advantage of a Pontryagin minimum principle, and the optimal control sequence distribution of the hybrid power system power is obtained. Energy management of the fuel cell automobile is achieved, and the optimality and working condition adaptability of an energy management strategy are effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of fuel cell vehicle energy management, and in particular to a fuel cell vehicle energy management strategy optimization method oriented to working condition requirements. Background Art

[0002] Fuel cell hybrid electric vehicles have two power sources, and the power demand of the whole vehicle depends on the sum of the output power of the fuel cell and the output power of the power battery. The goal of the energy management control strategy is to adjust the output power of these two power sources to meet the driving needs of the vehicle. Energy management strategy is a key technology that affects the energy efficiency and economy of fuel cell hybrid electric vehicles. Considering the diversity and complexity of vehicle driving conditions, as well as the time-varying, nonlinear and multi-physical characteristics of fuel cell and power battery hybrid systems, if the global driving conditions are unknown and the system constraints are met, it is of great significance for hybrid electric vehicles to design efficient and reasonable energy management strategies if the driving condition information can be mined from historical and future multi-dimensional dimensions.

[0003] At present, energy management strategies based on optimization are mainly divided into two types: global optimization and instantaneous optimization. Global optimization methods include dynamic programming (DP) and Pontryagin minimum principle (PMP), which can find the global optimal solution for energy management control. However, these methods require knowing the future driving conditions in advance and have a huge amount of calculation, so they are difficult to directly apply to the online control of actual vehicles and are usually only used as a reference for the theoretical optimal solution.

[0004] The energy management strategy based on instantaneous optimization is not adaptable to working conditions. For example, in ECMS, the equivalent factor is fixed, or the best equivalent factor is calculated in advance based on the driving cycle using global optimization. However, the equivalent factor is very sensitive to working conditions. In this case, the equivalent factor obtained cannot adapt to solving the optimization problem under unknown working conditions, and thus deviates from the optimal solution.

[0005] For the adaptive update of equivalent factors in ECMS, some studies have designed SOC The feedback PI controller updates the equivalent factor, which improves the robustness of the optimization algorithm to a certain extent, but it is not sufficient to simply use the current actual SOC As a post-compensation mechanism, the deviation-corrected equivalent factor from the reference value is bound to have certain limitations, and it is necessary to comprehensively consider the impact of future operating conditions on the equivalent factor.

[0006] Based on the above problems, the present invention proposes an optimization method for an energy management strategy with minimum energy consumption for working condition demand prediction. Summary of the invention

[0007] The present invention currently provides a fuel cell vehicle energy management strategy optimization method oriented to working condition requirements to solve the technical problems mentioned in the background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions: Terminology explanation: Some English abbreviations involved in the embodiments of the present invention are explained as follows: V2V, vehicle-to-vehicle communication.

[0009] V2I, vehicle-to-infrastructure communication.

[0010] PEMFC, Proton Exchange Membrane Fuel Cell.

[0011] PMP, Pontryagin's Minimum Principle.

[0012] FCEV, Fuel Cell Electric Vehicle.

[0013] EMS, Energy Management System.

[0014] DCDC stands for Direct Current to Direct Current Convertor.

[0015] A fuel cell vehicle energy management strategy optimization method oriented to working condition requirements, the method comprising: The equivalent factor iteration update module first uses the PMP algorithm to solve the global optimal solution under different driving conditions. The global optimal solution is used as a data set for training and testing the neural network co-state variable adjustment model. The model is used to predict the optimal initial co-state variables of the online part in real time. Its input includes the power battery SOC , the vehicle's current required power, current vehicle speed information, current vehicle acceleration information, and future time domain speed prediction information; then, through the equivalent relationship between PMP and ECMS, the PMP co-state variables are converted into equivalent factor values ​​of ECMS, integrating the working condition identification module, the predicted working condition and the battery SOC Adaptively modify the equivalent factor to obtain the optimal equivalent factor; The working condition identification module determines the working condition type through cluster analysis of the current driving condition, and solves the optimal data set under different working conditions through PMP. It can analyze the optimal value range of the equivalent factor under different working condition types, thereby determining the upper and lower limits of the equivalent factor to prevent the adaptive adjustment of the equivalent factor from deviating too much from the optimal value; The working condition demand prediction module uses the improved RBF algorithm to predict the working condition demand in the future time domain based on the historical vehicle working condition data, and designs a suitable prediction step size to obtain the vehicle speed prediction and power prediction within the prediction time domain step size by analyzing the impact of different prediction step sizes on the prediction accuracy and the impact of prediction error on the selection of co-state variables; The power system execution module obtains the optimal equivalent factor value through the equivalent factor iterative update module, takes the minimum instantaneous equivalent fuel hydrogen consumption as the optimization goal, considers the influence of fuel cell power fluctuations and the satisfaction of relevant constraints, and uses the ECMS algorithm to realize power distribution control of fuel cell vehicle energy management.

[0016] As a further technical solution of the present invention, the equivalent factor iterative update module includes: Equivalent Factor Training: The present invention is based on the goal of improving vehicle economy, takes the minimum total hydrogen consumption as the optimization goal, and uses the global optimization algorithm based on the PMP principle to optimize each group of different operating conditions (including but not limited to standard operating conditions) under the premise of satisfying the constraints: ; t 0 is the starting time of the working condition; t f is the end time of the working condition, and the cumulative term is the hydrogen consumption at each moment, which is obtained through the curve of PEMFC output power and hydrogen consumption, Δ t is the time step; The state variable of the optimization target is the power battery SOC , the control variable is the fuel cell power P fc , the state equation and constraints are established as follows: ; Where: U oc is the open circuit voltage of the power battery, which can be obtained by SOC Functional representation of R b is the internal resistance of the power battery, which can also be obtained by SOC Functional representation of Q b is the battery capacity of the power battery, P ref The power required by the main vehicle is SOClow and SOC high They are SOC The upper and lower limits allowed; P fcmin and P fcmax are the upper and lower limits of the fuel cell power, P bmin and P bmax is the upper and lower limits of the power of the power battery, Δ P fc is the fuel cell power fluctuation value, Δ P fc ( k )= P fc ( t )- P fc ( t- 1), SOC 0 and SOC tf are the initial conditions and terminal values ​​of the main vehicle; According to the PMP principle, the necessary condition for solving the optimal control problem is δJ=0. The Hamiltonian H is introduced and defined as: ; is the state quantity, the state of charge value of the power battery in the current state; Choose to use the shooting method for iterative solution; After the global optimal solution under different driving conditions is solved by using the PMP algorithm, the global optimal solution is used as a data set for training and testing the neural network co-state variable adjustment model. SOC , the vehicle's current required power, current vehicle speed information, current vehicle acceleration information, and future time domain vehicle speed prediction information are used as inputs, and the output is obtained through neural network training λ ( t ); ECMS is derived from engineering experience, but its essence is derived from the PMP method. It is a local real-time optimization method based on PMP. Therefore, ECMS is essentially equivalent to PMP. The equivalence relationship between the two is as follows: ; Where: The low calorific value of the fuel cell is 120MJ / Kg; is the open circuit voltage value; According to the equivalence relationship between the two, the co-state variables of PMP are converted into equivalent factor values ​​of ECMS to prepare for the subsequent adaptive update; As a further technical solution of the present invention, the equivalent factor iterative update module further includes: Equivalence factor adaptive update: Optimal Equivalence Factor S It is sensitive to working conditions and needs to make corresponding adaptive adjustments according to unknown working condition scenarios. The present invention is based on future working condition power prediction and SOC feedback, in which the power ratio in the future section is used as a feedforward adjustment of the equivalent factor, and SOC feedback is used as a post-compensation mechanism to combine the effects of the two and adjust the equivalent factor. S Perform adaptive update, and the update formula is as follows: ; ; ; ; Among them, the definition C p_n is the adjustment coefficient, which reflects the changing trend of future working conditions to a certain extent. K cp is the proportionality coefficient; P std To predict the power standard deviation in the time domain (t+T), P ave To predict the average power in the time domain (t+T), the equivalent factor is based on C p_n The value of should be adjusted in real time to meet the changing working conditions, which is specifically reflected in: C p_n When the value is large, C p_n From the definition of , it can be seen that the vehicle requires a large power in the prediction time domain, and the battery discharge provides energy to drive the vehicle. In order to maintain the SOC balance as much as possible, it is necessary to increase the equivalent factor in the prediction time domain; when C p_n When the value is small, it indicates that the vehicle's required power in the prediction time domain is small, and the equivalent factor in the prediction time domain should be appropriately reduced at this time; is the updated equivalent factor; In addition, the upper and lower limits of the equivalent factor are determined through the working condition identification module, and the optimal equivalent factor is output to the F-AECMS algorithm.

[0017] As a further technical solution of the present invention, the improved RBF algorithm architecture mainly includes four parts: input layer, hidden layer, receiving layer and output layer; the connection between the input layer, hidden layer and output layer is similar to that of a feedforward network, the units of the input layer only play a role in signal transmission, and the input includes historical vehicle speed information and historical acceleration information; the output layer units play a weighted role and mainly output the predicted speed value in the (t+T) time domain, while the receiving layer plays the role of a temporary variable, which is used to memorize the output value of the hidden layer unit at the previous moment, and can be considered as a delay operator with a one-step delay, so that the entire network structure has the ability to adapt to the time series.

[0018] As a further technical solution of the present invention, the power system execution module includes: 1) Establish the longitudinal dynamics model of the whole vehicle: The vehicle longitudinal dynamics model is constructed through the vehicle driving balance equation, and the mathematical formula is: ; m veh The quality of the main vehicle, f r is the friction coefficient, α is the slope angle, C D is the drag coefficient, A f is the frontal area, ρ is the air density, and δ is the rotation mass conversion factor; 2) Establish the optimal control problem: Based on the strategy of minimizing equivalent hydrogen consumption, the state variable is the power battery. SOC , the control variable is the fuel cell power P fc ， Establish the optimal control problem under certain constraints: ; ; is the actual fuel consumption before; is the actual power consumption before; The low heating value of hydrogen; is the updated equivalent factor; is the upper limit threshold of the power battery SOC operation, The lower limit of the SOC working of the power battery; the upper and lower limits of the output power in the high-efficiency working range of the fuel cell and ; is the fuel cell power value at the current moment; The power value of the power battery at the current moment; is the fuel cell power change rate; and The upper and lower limits of the fuel cell power variation range; is the initial value of the power battery state of charge; It is the terminal value of the state of charge of the power battery; is the power battery capacity; is the internal resistance of the power battery; is the required power; The current state of charge of the power battery; is the state of charge value of the power battery at time t+1; When solving the problem, the control variables are discretized, and the power output combination of the power source with the minimum equivalent hydrogen consumption is found in the feasible domain as the optimal control variable. The working torque of each power source is allocated, and the various working parameters of the vehicle model during driving are tracked.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. The optimization method of the energy management strategy for minimum energy consumption for working condition demand prediction described in the present invention mines driving condition information from historical and future multi-dimensional dimensions, uses the working condition training set to obtain the optimal SOC trajectory data set and the corresponding optimal equivalent factor under different working conditions based on the PMP principle, and then combines the future working condition prediction data to use the neural network algorithm to more accurately and quickly obtain the initial equivalent factor value.

[0020] 2. The present invention uses an improved RBF algorithm to obtain the demand condition prediction in the future time domain based on the historical vehicle operating condition data, and combines the SOC feedback and operating condition identification module to adaptively correct the equivalent factor. This adjustment method that comprehensively considers current and future operating conditions can better adapt to vehicle driving in unknown scenarios, thereby reducing energy consumption and improving the economy of the entire vehicle.

[0021] 3. The control strategy based on the F-AECMS algorithm provided by the present invention can better maintain the charge state balance of the power battery during vehicle driving, reduce the start and stop of the fuel cell, and take into account the influence of power fluctuations, so that the vehicle is more economical during driving, the drivability and riding comfort of the vehicle are improved, and the service life of the fuel cell is greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a diagram of the powertrain of a fuel cell vehicle in an embodiment of the present invention.

[0023] Figure 2 This is a structural diagram of a fuel cell vehicle energy management strategy optimization method oriented to operating conditions.

[0024] Figure 3It is a flow chart of the PMP targeting method in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0026] The present invention is based on the current working condition identification and classification and the prediction of the required working condition in the future time domain, updates the equivalent factors from the historical and future dimensions, and uses the instantaneous local optimization advantage of the Pontryagin minimum principle to obtain the optimal control sequence allocation of the hybrid power system power using the F-AECMS algorithm, thereby realizing the energy management of fuel cell vehicles and effectively improving the optimality and working condition adaptability of the energy management strategy.

[0027] Figure 1 It refers to the powertrain of a hydrogen fuel cell vehicle, which is mainly composed of a hydrogen system, a fuel cell system, a power battery system, a drive motor, a main reducer, a DCDC and related controllers.

[0028] See also Figure 2 The present invention provides a fuel cell vehicle energy management strategy optimization method oriented to working condition requirements, the method comprising: The equivalent factor iteration update module first uses the PMP algorithm to solve the global optimal solution under different driving conditions. The global optimal solution is used as a data set for training and testing the neural network co-state variable adjustment model. The model is used to predict the optimal initial co-state variables of the online part in real time. Its input includes the power battery SOC , the vehicle's current required power, current vehicle speed information, current vehicle acceleration information, and future time domain speed prediction information; then, through the equivalent relationship between PMP and ECMS, the PMP co-state variables are converted into equivalent factor values ​​of ECMS, integrating the working condition identification module, the predicted working condition and the battery SOC Adaptively modify the equivalent factor to obtain the optimal equivalent factor; a. Equivalent factor training: The present invention is based on the goal of improving vehicle economy, takes the minimum total hydrogen consumption as the optimization goal, and uses the global optimization algorithm based on the PMP principle to optimize each group of different operating conditions (including but not limited to standard operating conditions) under the premise of satisfying the constraints: ; t 0 is the starting time of the working condition; t fis the end time of the working condition, and the cumulative term is the hydrogen consumption at each moment, which is obtained through the curve of PEMFC output power and hydrogen consumption, Δ t is the time step; The state variable of the optimization target is the power battery SOC , the control variable is the fuel cell power P fc , the state equation and constraints are established as follows: ; Where: U oc is the open circuit voltage of the power battery, which can be obtained by SOC Functional representation of R b is the internal resistance of the power battery, which can also be obtained by SOC Functional representation of Q b is the battery capacity of the power battery, P ref The power required by the main vehicle is SOC low and SOC high They are SOC The upper and lower limits allowed; P fcmin and P fcmax are the upper and lower limits of the fuel cell power, P bmin and P bmax is the upper and lower limits of the power of the power battery, Δ P fc is the fuel cell power fluctuation value, Δ P fc ( k )= P fc ( t )- P fc ( t- 1), SOC 0 and SOC tf are the initial conditions and terminal values ​​of the main vehicle; According to the PMP principle, the necessary condition for solving the optimal control problem is δJ=0. If we introduce the Hamiltonian H, it is defined as: ; is the state quantity, the state of charge value of the power battery in the current state; The shooting method is used for iterative solution. The specific solution process is as follows: Figure 3 As shown in the figure, this method "shoots" a solution curve by adjusting the initial conditions to make it meet specific boundary conditions. The specific steps are: select the initial guess value; solve the differential equation; calculate the error; adjust the initial conditions; check convergence; output the results; After the global optimal solution under different driving conditions is solved by using the PMP algorithm, the global optimal solution is used as a data set for training and testing the neural network co-state variable adjustment model. SOC , the vehicle's current required power, current vehicle speed information, current vehicle acceleration information, and future time domain vehicle speed prediction information are used as inputs, and the output is obtained through neural network training λ ( t ); ECMS is derived from engineering experience, but its essence is derived from the PMP method. It is a local real-time optimization method based on PMP. Therefore, ECMS is essentially equivalent to PMP. The equivalence relationship between the two is as follows: ; Where: LHV H2 The low calorific value of the fuel cell is 120MJ / Kg; is the open circuit voltage value; According to the equivalence relationship between the two, the co-state variables of PMP are converted into equivalent factor values ​​of ECMS to prepare for the subsequent adaptive update; b. Equivalent factor adaptive update: Optimal Equivalence Factor S It is sensitive to working conditions and needs to make corresponding adaptive adjustments according to unknown working condition scenarios. The present invention is based on future working condition power prediction and SOC feedback, in which the power ratio in the future section is used as a feedforward adjustment of the equivalent factor, and SOC feedback is used as a post-compensation mechanism to combine the effects of the two and adjust the equivalent factor. S Perform adaptive update, and the update formula is as follows:

[0029] Among them, the definition C p_n is the adjustment coefficient, which reflects the changing trend of future working conditions to a certain extent. K cp is the proportionality coefficient; P std To predict the power standard deviation in the time domain (t+T), P ave To predict the average power in the time domain (t+T), the equivalence factor should be calculated based on C p_nThe value of should be adjusted in real time to meet the changing working conditions, which is specifically reflected in: C p_n When the value is large, C p_n From the definition of , it can be seen that the vehicle requires a large power in the prediction time domain, and the battery discharge provides energy to drive the vehicle. In order to maintain the SOC balance as much as possible, it is necessary to increase the equivalent factor in the prediction time domain; when C p_n When the value is small, it indicates that the vehicle's required power in the prediction time domain is small, and the equivalent factor in the prediction time domain should be appropriately reduced at this time; is the updated equivalent factor; In addition, the upper and lower limits of the equivalent factor are determined through the working condition identification module, and the optimal equivalent factor is output to the F-AECMS algorithm.

[0030] The working condition identification module determines the working condition type through cluster analysis of the current driving working condition, and solves the optimal data set under different working conditions through PMP. It can analyze the optimal value range of the equivalent factor under different working condition types, so as to determine the upper and lower limits of the equivalent factor and avoid the adaptive adjustment of the equivalent factor from deviating too much from the optimal value. The working condition is the relationship curve between time and speed, the current driving working condition is the vehicle speed at the current moment, and the working condition types include high-speed and high-load working condition, low-speed and low-load working condition and comprehensive working condition. The working condition demand prediction module uses the improved RBF algorithm to predict the working condition demand in the future time domain based on the historical working condition data of the vehicle, and designs a suitable prediction step to obtain the vehicle speed prediction and power prediction within the prediction time domain step by analyzing the influence of different prediction steps on the prediction accuracy and the influence of prediction error on the selection of co-state variables; the improvement of the improved RBF algorithm here refers to the improvement of the weight threshold of the RBF algorithm, which can obtain the weight threshold more accurately and quickly, and improve the prediction performance of the RBF algorithm; the historical working condition data of the vehicle is the working condition data that occurred before, such as NEDC working condition, UDDS working condition, CTCL working condition, etc. The algorithm architecture mainly includes four parts: input layer, hidden layer, receiving layer and output layer; the connection between the input layer, hidden layer and output layer is similar to that of a feedforward network, and the units in the input layer only play a role in signal transmission. The input includes historical vehicle speed information and historical acceleration information; the units in the output layer play a weighted role and mainly output the predicted speed value in the (t+T) time domain, while the receiving layer plays the role of a temporary variable, which is used to memorize the output value of the hidden layer unit at the previous moment. It can be considered as a delay operator with a one-step delay, which enables the entire network structure to have the ability to adapt to time series.

[0031] The power system execution module obtains the optimal equivalent factor value through the equivalent factor iteration update module, takes the minimum instantaneous equivalent fuel hydrogen consumption as the optimization goal, considers the influence of fuel cell power fluctuation and the satisfaction of related constraints, and uses the ECMS algorithm to realize the power distribution control of fuel cell vehicle energy management. The specific contents are as follows: 1) Establish the longitudinal dynamics model of the whole vehicle: The longitudinal dynamics model of the whole vehicle is constructed by the vehicle driving balance equation, and the mathematical formula is:

[0032] m veh The quality of the main vehicle, f r is the friction coefficient, α is the slope angle, C D is the drag coefficient, A f is the frontal area, ρ is the air density, δ is the rotation mass conversion coefficient, g is the gravitational acceleration, and V(t) is the vehicle speed at time t; 2) Establish the optimal control problem: Based on the strategy of minimizing equivalent hydrogen consumption, the state variable is the power battery. SOC , the control variable is the fuel cell power P fc ， Establish the optimal control problem under certain constraints:

[0033] is the actual fuel consumption before; is the actual power consumption before; The low heating value of hydrogen; is the updated equivalent factor; is the upper limit threshold of the power battery SOC operation, The lower limit of the SOC working of the power battery; the upper and lower limits of the output power in the high-efficiency working range of the fuel cell and ; is the fuel cell power value at the current moment; The power value of the power battery at the current moment; is the fuel cell power change rate; and The upper and lower limits of the fuel cell power variation range; is the initial value of the power battery state of charge; It is the terminal value of the state of charge of the power battery; is the power battery capacity; is the internal resistance of the power battery; is the required power; The current state of charge of the power battery; is the state of charge value of the power battery at time t+1; When solving the problem, the control variables are discretized, and the power output combination of the power source with the minimum equivalent hydrogen consumption is found in the feasible domain as the optimal control variable. The working torque of each power source is allocated, and the various working parameters of the vehicle model during driving are tracked.

[0034] It should be noted that, in this article, the term "comprises" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0035] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A fuel cell vehicle energy management strategy optimization method oriented to working condition requirements, characterized in that: The method comprises: S1, the equivalent factor iterative update module first uses the PMP algorithm to solve the global optimal solution of multiple working conditions, as the data set for training and testing the neural network co-state variable adjustment model, which is used to predict the optimal initial co-state variables of the online part in real time, and then converts the co-state variables into the equivalent factor value of ECMS through the equivalent relationship between PMP and ECMS, integrating the working condition identification module, the predicted working condition and the battery SOC Adaptively modify the equivalent factor to obtain the optimal equivalent factor; S2, the working condition identification module analyzes the optimal value range of the equivalent factor under different working conditions by clustering analysis of the current driving conditions and solving the optimal data set under different working conditions through PMP; S3, the working condition demand prediction module uses the improved RBF algorithm to predict the working condition demand in the future time domain, and designs the prediction step length to obtain the vehicle speed prediction and power prediction within the prediction time domain step length by analyzing the influence of different prediction step lengths on the prediction accuracy and the influence of prediction error on the selection of co-state variables; S4, the power system execution module obtains the optimal equivalent factor value through the equivalent factor iteration update module, takes the minimization of instantaneous equivalent fuel hydrogen consumption as the optimization goal, and uses the ECMS algorithm to realize the power distribution control of the fuel cell vehicle energy management.

2. The method for optimizing the energy management strategy of a fuel cell vehicle based on working condition requirements according to claim 1, characterized in that: Step S1 includes: Equivalent Factor Training: Based on the goal of improving vehicle economy, the total hydrogen consumption is minimized as the optimization goal. Under the premise of satisfying the constraints, the global optimization algorithm based on the PMP principle is used to optimize each group of different working conditions: ； J is a performance indicator; is the mass flow rate of hydrogen; is the fuel cell power; t 0 is the starting time of the working condition; t f is the end time of the working condition, and the cumulative term is the hydrogen consumption at each moment, which is obtained through the curve of PEMFC output power and hydrogen consumption, Δ t is the time step; The state variable of the optimization target is the power battery SOC , the control variable is the fuel cell power P fc , the state equation and constraints are established as follows: ； Where: U oc is the open circuit voltage of the power battery, R b is the internal resistance of the power battery, Q b is the battery capacity of the power battery, P ref The power required by the main vehicle is SOC low and SOC high They are SOC The upper and lower limits allowed; P fcmin and P fcmax are the upper and lower limits of the fuel cell power, P bmin and P bmax is the upper and lower limits of the power of the power battery, Δ P fc is the fuel cell power fluctuation value, Δ P fc ( k )= P fc ( t )- P fc ( t- 1), SOC 0 and SOC tf are the initial conditions and terminal values ​​of the main vehicle; According to the PMP principle, the necessary condition for solving the optimal control problem is δJ=0. The Hamiltonian H is introduced and defined as: ； is the state quantity, the state of charge value of the power battery in the current state; Choose to use the shooting method for iterative solution; After the global optimal solution under different driving conditions is solved by using the PMP algorithm, the global optimal solution is used as a data set for training and testing the neural network co-state variable adjustment model. SOC , the vehicle's current required power, current vehicle speed information, current vehicle acceleration information, and future time domain vehicle speed prediction information are used as inputs, and the output is obtained through neural network training λ ( t ); ECMS is a local real-time optimization method based on PMP. Therefore, ECMS is essentially equivalent to PMP. The equivalence relationship between the two is as follows: ； Where: The low calorific value of the fuel cell is 120MJ / Kg; is the open circuit voltage value; According to the equivalence relationship between the two, the covariates of PMP are converted into equivalent factor values ​​of ECMS.

3. The method for optimizing the energy management strategy of a fuel cell vehicle based on working condition requirements according to claim 2 is characterized in that: The step S1 further comprises: Equivalence factor adaptive update: Optimal Equivalence Factor S It is sensitive to working conditions and needs to make corresponding adaptive adjustments according to unknown working conditions. It is based on the power prediction of future working conditions and SOC feedback. The power ratio in the future section is used as the feedforward adjustment of the equivalent factor, and the SOC feedback is used as a post-compensation mechanism to combine the influence of the two and adjust the equivalent factor. S Perform adaptive update, and the update formula is as follows: ； ； ； ； Among them, the definition C p_n is the adjustment factor, K cp is the proportionality coefficient; P std To predict the power standard deviation in the time domain, P ave To predict the average power in the time domain, the equivalent factor is based on C p_n The value of should be adjusted in real time to meet the changing working conditions. C p_n When the value is large, the equivalent factor in the prediction time domain is increased; when C p_n When the value is small, the equivalent factor in the prediction time domain is reduced; is the updated equivalent factor; In addition, the upper and lower limits of the equivalent factor are determined through the working condition identification module, and the optimal equivalent factor is output to the F-AECMS algorithm.

4. The method for optimizing the energy management strategy of a fuel cell vehicle based on working condition requirements according to claim 1, characterized in that: The improved RBF algorithm architecture mainly includes four parts: input layer, hidden layer, receiving layer and output layer; the units in the input layer only play a role in signal transmission, and the input includes historical vehicle speed information and historical acceleration information; the output layer units play a weighted role and mainly output the speed value in the predicted time domain, while the receiving layer plays the role of a temporary variable to memorize the output value of the hidden layer unit at the previous moment.

5. The method for optimizing the energy management strategy of a fuel cell vehicle based on working condition requirements according to claim 1, characterized in that: Step S4 includes: 1) Establish the longitudinal dynamics model of the whole vehicle: The longitudinal dynamics model of the whole vehicle is constructed by the vehicle driving balance equation, and the mathematical formula is: ； m veh The quality of the main vehicle, f r is the friction coefficient, α is the slope angle, C D is the drag coefficient, A f is the frontal area, ρ is the air density, δ is the rotation mass conversion coefficient, g is the gravitational acceleration, and V(t) is the vehicle speed at time t; 2) Establish the optimal control problem: Based on the strategy of minimizing equivalent hydrogen consumption, the state variable is the power battery. SOC , the control variable is the fuel cell power P fc ， Establish the optimal control problem under the constraints: ； ； is the actual fuel consumption before; is the actual power consumption before; The low heating value of hydrogen; is the updated equivalent factor; is the upper limit threshold of the power battery SOC operation, The lower limit of the SOC working of the power battery; the upper and lower limits of the output power in the high-efficiency working range of the fuel cell and ; is the fuel cell power value at the current moment; The power value of the power battery at the current moment; is the fuel cell power change rate; and The upper and lower limits of the fuel cell power variation range; is the initial value of the power battery state of charge; It is the terminal value of the state of charge of the power battery; is the power battery capacity; is the internal resistance of the power battery; is the required power; The current state of charge of the power battery; is the state of charge value of the power battery at time t+1; When solving the problem, the control variables are discretized, and the power output combination of the power source with the minimum equivalent hydrogen consumption is found in the feasible domain as the optimal control variable. The working torque of each power source is allocated, and the various working parameters of the vehicle model during driving are tracked.

Images