New energy vehicle multi-mode energy management control method and system based on event triggering

Abstract

The invention discloses a new energy vehicle multi-mode energy management control method and system based on event triggering. The method comprises the steps that the total demand power of a fuel cell vehicle is obtained, and vehicle speed information in the short-term future is predicted; performing feature analysis on the short-time future vehicle speed information, and determining a short-time future working condition type and a corresponding output mode of the fuel cell vehicle; the optimal control output of the fuel cell vehicle is obtained based on a model prediction control method in combination with the output mode of the fuel cell vehicle and the total demand power of the fuel cell vehicle; and performing prediction judgment on the optimal control output of the fuel cell vehicle, and determining an optimal control sequence of the fuel cell vehicle. The method can improve the adaptability and economy of the fuel cell in the fuel cell vehicle, reduce the dynamic load pressure of the fuel cell, and improve the utilization rate of computing resources. The new energy vehicle multi-mode energy management control method and system based on event triggering can be widely applied to the technical field of fuel cell energy management.

Description

Technical Field

[0001] The present invention relates to the field of fuel cell energy management technology, and in particular to an event-triggered multi-modal energy management control method and system for new energy vehicles. Background Art

[0002] Fuel cell vehicles can achieve completely zero carbon emissions and have advantages such as short hydrogen refueling time and long driving range. They usually use hydrogen fuel cells as the main power source and are equipped with one or two auxiliary power sources (such as lithium-ion batteries or capacitors). During driving, multiple power sources work together to provide power to the drive motor. However, fuel cells face severe challenges in durability, which hinders the industrial development of fuel cell vehicles. The essence of durability is the degradation of the performance of key internal components (such as proton exchange membranes). Operating conditions are one of the factors that affect the durability performance of fuel cells, such as dynamic loading, start-stop, idling and high-power conditions. Therefore, more and more research is devoted to improving the energy management system based on the characteristics of the operating conditions and their impact on the fuel cell, so as to effectively slow down the performance degradation of the fuel cell.

[0003] From the perspective of optimization strategies and real-time performance, existing energy management strategies can generally be divided into three categories: rule-based strategies, optimization-centric strategies, and learning-based strategies. Rule-based strategies aim to ensure that the fuel cell system always operates at the appropriate efficiency point during driving. Common strategies include state machine methods, power tracking methods, and power consumption-power maintenance methods. For energy balance and allocation, optimization-based strategies comprehensively consider factors such as hydrogen energy consumption, system cost, and power source lifespan, and establish corresponding objective functions and boundary constraints. Currently, optimization-centric strategies are mainly based on global optimization and instantaneous optimization algorithms. Global optimization includes Pontryagin's minimum principle and dynamic programming. Learning-based strategies typically utilize advanced data-driven techniques, combining historical data and traffic information to calculate optimal control. Khalatbarisoltani uses federated reinforcement learning to find the optimal energy management strategy in the time domain, providing scalability and adaptability for applications. However, improving the real-time performance of model predictive control and reducing computational burden require further research. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a multimodal energy management control method and system for new energy vehicles based on event triggering, which can improve the adaptability and economy of fuel cells in fuel cell vehicles, reduce the dynamic load pressure of fuel cells, and reduce the calculation amount of the vehicle control unit, thereby improving the utilization rate of computing resources.

[0005] The first technical solution adopted by the present invention is: a multi-modal energy management control method for new energy vehicles based on event triggering, comprising the following steps:

[0006] Obtain the total required power of the fuel cell vehicle and predict the vehicle speed information of the fuel cell vehicle in the short term;

[0007] Perform characteristic analysis on the fuel cell vehicle's speed information in the short term future to determine the fuel cell vehicle's short term future operating condition type and corresponding output mode;

[0008] Combining the output mode of the fuel cell vehicle with the total power demand of the fuel cell vehicle, the optimal control output of the fuel cell vehicle is obtained through a model-based predictive control method;

[0009] Based on the event triggering mechanism, the optimal control output of the fuel cell vehicle is predicted and judged, the optimal control sequence of the fuel cell vehicle is determined, and the energy management of the fuel cell vehicle is realized.

[0010] Furthermore, the step of obtaining the total required power of the fuel cell vehicle and predicting the vehicle speed information of the fuel cell vehicle in the short term includes:

[0011] Determining driving information of the fuel cell vehicle according to the driving resistance of the fuel cell vehicle;

[0012] Determine the output power of the fuel cell vehicle drive motor based on the output torque, speed and efficiency of the fuel cell vehicle drive motor;

[0013] Obtain the output power of the fuel cell stack of the fuel cell vehicle, and determine the total power requirement of the fuel cell vehicle by combining the driving information of the fuel cell vehicle and the output power of the fuel cell vehicle drive motor;

[0014] Based on the RBFNN model, the historical speed data of fuel cell vehicles is obtained, and the speed information of fuel cell vehicles in the short term in the future is predicted through a sliding window.

[0015] Furthermore, the step of obtaining historical speed data of the fuel cell vehicle based on the RBFNN model and predicting the short-term future speed information of the fuel cell vehicle through a sliding window specifically includes:

[0016] Determine the activation function and output of the RBFNN model, and obtain the diffusion factor and the number of hidden layer neurons of the RBFNN model;

[0017] The diffusion factor and the number of hidden layer neurons of the RBFNN model are optimized by the Bayesian optimization algorithm to obtain the optimized RBFNN model;

[0018] Based on the optimized RBFNN model, the historical speed data of fuel cell vehicles is input and the speed information of fuel cell vehicles in the short term in the future is predicted through a sliding window.

[0019] Furthermore, the step of performing feature analysis on the vehicle speed information of the fuel cell vehicle in the short-term future to determine the fuel cell vehicle's short-term future operating condition type and the corresponding output mode specifically includes:

[0020] Using a preset future time in the short-term future fuel cell vehicle speed information as a dynamic sliding window, extracting four types of features from each window, the four types of features including average speed, maximum speed, maximum acceleration, and minimum acceleration;

[0021] determining whether the four characteristics exceed preset thresholds and whether the duration of the current fuel cell vehicle driving condition meets a minimum duration, and marking the future window, wherein the fuel cell vehicle driving condition includes low speed driving, medium speed driving, medium-high speed driving, and high speed driving;

[0022] The output mode of the fuel cell vehicle is determined according to the driving conditions of the fuel cell vehicle, where:

[0023] The output modes of the low-speed and medium-speed driving are that the fuel cell provides a first output power, and the remaining power is provided by the lithium-ion battery;

[0024] The output mode for medium and high speed driving is that the output power of the lithium-ion battery remains unchanged, and the fuel cell provides a second output power, which is greater than the first output power;

[0025] The output mode of high-speed driving is that the fuel cell provides a third output power, and the remaining power is provided by the lithium-ion battery. The third output power is greater than the second output power.

[0026] Furthermore, the step of combining the output mode of the fuel cell vehicle with the total required power of the fuel cell vehicle and obtaining the optimal control output of the fuel cell vehicle through a model-based predictive control method specifically includes:

[0027] Obtain the SoC and instantaneous equivalent hydrogen consumption equation of fuel cell vehicles;

[0028] Determine the discrete equation of the SoC of the fuel cell vehicle based on the SoC of the fuel cell vehicle, and determine the discrete equation of the instantaneous equivalent hydrogen consumption based on the instantaneous equivalent hydrogen consumption equation;

[0029] The discrete state space equation is constructed by combining the discrete equation of the SoC of the fuel cell vehicle with the discrete equation of the instantaneous equivalent hydrogen consumption;

[0030] Based on the total power demand of the fuel cell vehicle as the disturbance variable and the change in the output power of the fuel cell as the control variable, the discrete state space equation is constrained, and the discrete state transfer equation and observation variables are constructed;

[0031] Based on the estimation of fuel cell vehicle economy and SoC change, a constrained cost function for fuel cell vehicles is constructed;

[0032] The constrained cost function of the fuel cell vehicle is converted into a quadratic programming form, and combined with the output mode of the fuel cell vehicle, multi-objective optimization is performed to obtain the optimal control output of the fuel cell vehicle.

[0033] Furthermore, the expression of the optimal control output of the fuel cell vehicle is:

[0034]

[0035] In the above formula, represents the optimal control output of the fuel cell vehicle, J(k) represents the constrained cost function of the fuel cell vehicle, represents the weight matrix, U(k) represents the calculated optimal control sequence, represents the quadratic coefficient matrix of the objective function, represents the linear term vector, M and E represent the inequality constraint matrix, U min Indicates the lower limit of the control quantity, U max represents the upper limit of the control quantity, Ξ, Ψ, Ω all represent the reconstructed state matrix, ρ represents the relaxation factor, x(k) represents the state quantity, W represents the disturbance quantity matrix, Indicates the reference value of the output quantity.

[0036] Furthermore, the step of predicting and judging the optimal control output of the fuel cell vehicle based on the event triggering mechanism, determining the optimal control sequence of the fuel cell vehicle, and realizing energy management of the fuel cell vehicle specifically includes:

[0037] Design event trigger conditions and event trigger thresholds based on disturbance boundaries, penalty matrices, and prediction time domains;

[0038] Obtaining the difference between the ideal state output of the fuel cell vehicle and the optimal control output of the fuel cell vehicle and making a judgment;

[0039] If the event trigger threshold is exceeded, the model predictive control is triggered at the next moment;

[0040] If the event trigger threshold is not exceeded, the model predictive control will not be triggered at the next moment, and the control output at time k+1 in the prediction time domain will be used as the optimal control sequence for the fuel cell vehicle to achieve energy management of the fuel cell vehicle.

[0041] The second technical solution adopted by the present invention is: a multi-modal energy management and control system for new energy vehicles based on event triggering, comprising:

[0042] The first module is used to obtain the total required power of the fuel cell vehicle and predict the vehicle speed information of the fuel cell vehicle in the short term;

[0043] The second module is used to perform feature analysis on the fuel cell vehicle's speed information in the short term future, and determine the fuel cell vehicle's short term future operating condition type and the corresponding output mode;

[0044] The third module is used to combine the output mode of the fuel cell vehicle with the total required power of the fuel cell vehicle and obtain the optimal control output of the fuel cell vehicle through a model-based predictive control method;

[0045] The fourth module is used to predict and judge the optimal control output of the fuel cell vehicle based on the event trigger mechanism, determine the optimal control sequence of the fuel cell vehicle, and realize energy management of the fuel cell vehicle.

[0046] The beneficial effects of the method and system of the present invention are as follows: the present invention obtains the total power demand of the fuel cell vehicle and predicts the vehicle speed information of the fuel cell vehicle in the short term future, further performs feature analysis on the vehicle speed information of the fuel cell vehicle in the short term future, determines the short term future operating condition type and the corresponding output mode of the fuel cell vehicle, adopts the Bayesian optimization method to optimize the core parameters of the radial basis function neural network, uses the optimized radial basis function neural network to predict the future vehicle speed, effectively improves the vehicle speed prediction accuracy, further combines the output mode of the fuel cell vehicle with the total power demand of the fuel cell vehicle, obtains the optimal control output of the fuel cell vehicle through the model predictive control method, reasonably allocates power, adjusts SoC, reduces hydrogen consumption, and avoids drastic power fluctuations of the fuel cell, finally, based on the event trigger mechanism, predicts and judges the optimal control output of the fuel cell vehicle, determines the optimal control sequence of the fuel cell vehicle, can improve the adaptability and economy of the fuel cell in the fuel cell vehicle, reduce the dynamic load pressure of the fuel cell, and reduce the calculation amount of the vehicle control unit, thereby improving the utilization rate of computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flowchart of the steps of the event-triggered multi-modal energy management control method for new energy vehicles of the present invention;

[0048] Figure 2 This is a structural block diagram of the event-triggered multi-modal energy management and control system for new energy vehicles of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The step numbers in the following embodiments are provided for ease of description only and do not limit the order of the steps. The order of execution of the steps in the embodiments can be adaptively adjusted based on the understanding of those skilled in the art.

[0050] First of all, it should be noted that the durability and adaptability of fuel cells are poor, which hinders the further promotion and application of fuel cell electric vehicles under complex conditions. By predicting future short-term information during driving, such as vehicle speed, route information and traffic signals, prior knowledge for optimizing energy management can be obtained, effectively enhancing the adaptability of fuel cells under different driving conditions. Zendegan combines altitude and speed predictions to approximate offline power distribution, providing a reference for online power distribution strategies for heavy-duty fuel cell vehicles. Liu Yonggang of Chongqing University proposed a short-term vehicle speed prediction model that integrates random prediction and machine learning. The model achieved good results and improved energy economy. Therefore, considering traffic information in the energy management strategy system is an effective solution to optimize economy and adaptability under different driving conditions.

[0051] However, existing fuel cell vehicle energy management strategies have the following problems:

[0052] Rule-based energy management strategies are easy to implement and do not require prior knowledge of driving conditions. However, their drawback is that they rely on expert experience and have poor adaptability. Although global optimization energy management methods can improve global optimization results, they are difficult to apply to energy management systems in real time. They provide a reference for instantaneous optimization algorithms. The latter include equivalent consumption minimization strategies and model predictive control. Learning-based strategies do not rely on models, but have the disadvantages of complexity and difficulty in real-time application. Instantaneous optimization energy management strategies based on model predictive control are a promising and widely used solution. Their main advantages lie in multi-objective optimization and robustness, such as energy economy and power supply durability as optimization objectives. However, model predictive control is computationally intensive. How to improve the real-time performance of model predictive control and reduce computational pressure is a problem that requires further research.

[0053] Furthermore, while traffic information can provide predictive prior knowledge for fuel cell management systems, the electrochemical reaction environment within the fuel cell can sometimes become more complex and harsh under changing driving conditions, accelerating fuel cell performance degradation. Therefore, capacity management strategies must consider the fuel cell's adaptability and dynamic load pressure in light of complex driving conditions.

[0054] Therefore, the embodiments of the present invention aim to improve the adaptability and economy of fuel cells in fuel cell vehicles, reduce the dynamic load pressure of fuel cells, and reduce the calculation amount of the vehicle control unit, thereby improving the utilization rate of computing resources.

[0055] It should also be noted that the embodiments of the present invention are specific to a fuel cell vehicle configuration. This configuration can be described as follows: the output voltage of the lithium-ion battery is consistent with the bus voltage; the fuel cell is connected in series with a DC / DC converter and in parallel with the lithium-ion battery, either jointly or independently supplying power to the drive motor. Power losses in intermediate components are ignored; real-time control calculations and energy distribution between the fuel cell and lithium-ion battery are dependent on the vehicle controller.

[0056] Reference Figure 1 The present invention provides a multi-modal energy management control method for new energy vehicles based on event triggering, the method comprising the following steps:

[0057] S100, obtaining the total required power of the fuel cell vehicle and predicting the vehicle speed information of the fuel cell vehicle in the short term;

[0058] S110, determining driving information of the fuel cell vehicle according to the driving resistance of the fuel cell vehicle;

[0059] In this embodiment, the longitudinal external forces on the vehicle during driving include rolling resistance, air resistance, slope resistance, acceleration resistance, etc.

[0060]

[0061] Where m is the vehicle mass, C D is the drag coefficient, ρ is the air density, f is the rolling resistance coefficient, δ is the rotation conversion coefficient, α is the slope, A f is the frontal area.

[0062] S120, determining the output power of the fuel cell vehicle drive motor according to the output torque, speed, and efficiency of the fuel cell vehicle drive motor;

[0063] In this embodiment, the driving torque is provided by the driving motor, and after passing through the reducer, part of it is lost in the transmission system. m and speed w m Expressed as:

[0064]

[0065] w m =i0v / r

[0066] Where r is the wheel radius, η fd is the transmission efficiency, and i0 is the final reducer transmission ratio.

[0067] Next, the output power of the drive motor is calculated from its motor torque, motor speed, and motor efficiency:

[0068]

[0069] In the above formula, P m Indicates the output power of the drive motor.

[0070] S130, obtaining the output power of the fuel cell stack of the fuel cell vehicle, and determining the total required power of the fuel cell vehicle in combination with the driving information of the fuel cell vehicle and the output power of the fuel cell vehicle drive motor;

[0071] In this embodiment, the total power is provided by two power supplies, and the output power of the fuel cell stack is calculated as follows:

[0072] P fc =(P d -P bat ) / η DC / DC

[0073] P fc_idl ≤P fc ≤P fc_max

[0074] where η DC / DC Indicates the efficiency of the DC / DC converter; P d =P m +P aux , P aux is the auxiliary load of the accessory, such as a lamp; P fc_max and P fc_idl They are maximum power and idle power respectively.

[0075] S140 , based on the RBFNN model, obtain historical speed data of the fuel cell vehicle, and predict the speed information of the fuel cell vehicle in the short term in the future through a sliding window.

[0076] Specifically, the short-term future vehicle speed information prediction technology selects the appropriate number of neurons based on historical speed data and uses a sliding window to predict the time domain speed in the next 5 seconds.

[0077] S141, determining the activation function of the RBFNN model and the output of the RBFNN model, and obtaining the diffusion factor and the number of hidden layer neurons of the RBFNN model;

[0078] In this embodiment, first, the activation function of the RBFNN model is set to a Gaussian function:

[0079]

[0080] Among them, x speed is the input vector of historical velocity; c j is the center vector of the hidden layer neurons; σ jis the width of the basis function (given by the diffusion factor σ d control);||x speed -c j || 2 is the square of the Euclidean distance.

[0081] Next, the output of RBFNN is designed as:

[0082]

[0083] Among them, N h is the number of neurons in the hidden layer; w kj is the weight from the jth hidden neuron to the kth output neuron; b k is the bias of the output neuron.

[0084] S142. Optimizing the diffusion factor and the number of hidden layer neurons of the RBFNN model using a Bayesian optimization algorithm to obtain an optimized RBFNN model;

[0085] In this embodiment, the accuracy of RBFNN is affected by N h and σ d In order to improve the real-time performance and adaptability, the Bayesian optimization algorithm is used to optimize the two parameters.

[0086] First, the objective function is modeled using Gaussian process (GP). Assume that the objective function f(x speed ) obeys a multivariate Gaussian distribution:

[0087]

[0088] Among them, m(x speed ) is the average function; σ f is the signal variance; l is the length scale.

[0089]

[0090] e(x speed )=fspeed min

[0091] Among them, f min is the currently observed f(x speed )’s minimum target value; Φ b and φ b They are The cumulative distribution function and probability density function of . By obtaining the function update Gaussian process (GP) to select the next sampling point, minimize the objective function, and thus obtain the appropriate N h and σ d .

[0092] S143. Based on the optimized RBFNN model, historical speed data of the fuel cell vehicle is input, and the speed information of the fuel cell vehicle in the short term in the future is predicted through a sliding window.

[0093] In this embodiment, the network is pre-trained using HWFET and MANHATTAN data, and simulations are performed under both WLTC and NEDC operating conditions. The proposed method can effectively predict vehicle speed and provide accurate future short-term speed information for energy management of fuel cell electric vehicles.

[0094] S200, performing feature analysis on the vehicle speed information of the fuel cell vehicle in the short-term future to determine the fuel cell vehicle's short-term future operating condition type and the corresponding output mode;

[0095] S210, using a preset future time in the short-term future speed information of the fuel cell vehicle as a dynamic sliding window, and extracting four types of features from each window, the four types of features including average speed, maximum speed, maximum acceleration, and minimum acceleration;

[0096] S220: determining whether the four characteristics exceed preset thresholds and whether the duration of the current fuel cell vehicle driving condition meets a minimum duration, and marking the future window, wherein the fuel cell vehicle driving condition includes low-speed driving, medium-speed driving, medium-high-speed driving, and high-speed driving;

[0097] In this example, the next five-second window is used as a dynamic sliding window. Four features are extracted from each window: average speed, maximum speed, maximum acceleration, and minimum acceleration. The system then determines whether the features exceed corresponding thresholds and whether the current driving condition duration meets a minimum duration. Future windows are then marked. To prevent frequent switching between driving conditions, the minimum duration is set to 60 seconds.

[0098] Next, the working conditions are divided into four categories:

[0099] 1) Low speed, which means driving at low speed on urban roads with frequent acceleration and deceleration;

[0100] 2) Medium speed, indicating driving on urban roads with frequent acceleration and deceleration;

[0101] 3) Medium to high speed, indicating driving on suburban roads or expressways with relatively high speeds;

[0102] 4) High-speed driving, that is, on the highway, the average vehicle speed is very high and the acceleration and deceleration changes are large.

[0103] S230: Determine the output mode of the fuel cell vehicle according to the driving condition of the fuel cell vehicle, wherein:

[0104] The output modes of the low-speed and medium-speed driving are that the fuel cell provides a first output power, and the remaining power is provided by the lithium-ion battery;

[0105] The output mode for medium and high speed driving is that the output power of the lithium-ion battery remains unchanged, and the fuel cell provides a second output power, which is greater than the first output power;

[0106] The output mode of high-speed driving is that the fuel cell provides a third output power, and the remaining power is provided by the lithium-ion battery. The third output power is greater than the second output power.

[0107] In this embodiment, for each driving condition, the proposed mechanism sets the corresponding output mode for the fuel cell:

[0108] 1) When driving at low and medium speeds, when acceleration and deceleration are frequent, the fuel cell power P fc The remaining power should be provided by the lithium-ion battery.

[0109] At low speed:

[0110] P fc-ref =1kw

[0111] At medium speed:

[0112] P fc-ref =10kw

[0113] 2) When driving at medium and high speeds, the power of the lithium-ion battery does not change much, and the fuel cell provides greater power.

[0114] P fc-ref =max(12kw,min(0.75P d ,20kw))

[0115] 3) When driving at high speed, due to the slow response of the fuel cell, P fc Should maintain a high level, the remaining power is provided by the lithium-ion battery,

[0116] P fc_ref =25kw

[0117] Among them, P fc_ref Indicates the output power of the fuel cell.

[0118] S300, combining the output mode of the fuel cell vehicle and the total required power of the fuel cell vehicle, obtaining an optimal control output of the fuel cell vehicle through a model-based predictive control method;

[0119] First, it's important to note that in this model-predictive control-based energy management strategy, the required power during driving is provided by both the fuel cell and the lithium-ion battery. Using model predictive control, the total power demand is rationally allocated, ensuring that both the fuel cell and the lithium-ion battery operate within the desired range.

[0120] Specifically, the SoC and instantaneous equivalent hydrogen consumption equation of the fuel cell vehicle are obtained; the discrete equation of the SoC of the fuel cell vehicle is determined based on the SoC of the fuel cell vehicle, and the discrete equation of the instantaneous equivalent hydrogen consumption is determined based on the instantaneous equivalent hydrogen consumption equation; the discrete equation of the SoC of the fuel cell vehicle and the discrete equation of the instantaneous equivalent hydrogen consumption are combined to construct a discrete state space equation; based on the total power demand of the fuel cell vehicle as the disturbance and the change in the output power of the fuel cell as the control variable, the discrete state space equation is constrained, and a discrete state transfer equation and observation variables are constructed; based on the economy of the fuel cell vehicle and the estimation of the SoC change, a constrained cost function of the fuel cell vehicle is constructed; the constrained cost function of the fuel cell vehicle is converted into a quadratic programming form, and combined with the output mode of the fuel cell vehicle, multi-objective optimization is performed to obtain the optimal control output of the fuel cell vehicle.

[0121] In this embodiment, first, a discrete state space equation is constructed by combining SoC and the instantaneous equivalent hydrogen consumption equation. The discrete equation of SoC can be expressed as:

[0122]

[0123] The instantaneous equivalent hydrogen consumption discrete equation can be expressed as:

[0124]

[0125] Where ΔT is the sampling time and λ(k) is the equivalence factor.

[0126] The equivalent factor can be adjusted according to SoC to achieve regulatory adaptability, with an initial value of λ0 = 1.9.

[0127] λ(k)=λ0+K p (SoC ref -SoC(k))

[0128] The discrete state space equation can be expressed as:

[0129]

[0130] It should be noted that A(k), B u (k), B w (k) contains an element η that changes with time fc and V bat(SoC) to ensure accuracy. However, to simplify calculations, these matrices are treated as constant matrices within the prediction horizon and are set to their values at time k. These matrices are only updated at time k.

[0131] Among them, the required power P d (k) is the disturbance, and the change in fuel cell output power ΔP fc (k) is used as a control variable. This method is beneficial to P fc (k) to ensure the smooth output.

[0132]

[0133] The discrete state transfer equation is:

[0134]

[0135]

[0136] W(k)=[P d (k),P d (k+1),...,P d (k+N p -1)]

[0137] U(k)=[ΔP fc (k),ΔP fc (k+1),...,ΔP fc (k+N c -1)]

[0138] Among them, Φ, Θ, and Γ are the reconstructed state matrices.

[0139] The observed variables are reconstructed as:

[0140]

[0141]

[0142] Among them, N p Represents the prediction time domain; N c Indicates the control time domain.

[0143] For fuel cell vehicles, the objective function is established based on the estimation of economic efficiency and SoC changes. Solving the optimization problem requires satisfying the constraints of each power source and SoC. The constrained cost function is

[0144]

[0145] Among them, Q, R, and F are weight matrices; ρ is the relaxation factor.

[0146] Convert the objective function into quadratic programming form

[0147]

[0148] Among them, Q 11 , Q 22 、 is the weight matrix.

[0149] From this, we can get the multi-objective optimization result, and the optimal control output expression is:

[0150]

[0151] In the above formula, represents the optimal control output of the fuel cell vehicle, J(k) represents the constrained cost function of the fuel cell vehicle, represents the weight matrix, U(k) represents the calculated optimal control sequence, represents the quadratic coefficient matrix of the objective function, represents the linear term vector, M and E represent the inequality constraint matrix, U min Indicates the lower limit of the control quantity, U max represents the upper limit of the control quantity, Ξ, Ψ, Ω all represent the reconstructed state matrix, ρ represents the relaxation factor, x(k) represents the state quantity, W represents the disturbance quantity matrix, Indicates the reference value of the output quantity.

[0152] S400: Based on the event triggering mechanism, the optimal control output of the fuel cell vehicle is predicted and judged, the optimal control sequence of the fuel cell vehicle is determined, and the energy management of the fuel cell vehicle is realized.

[0153] First, it should be noted that the event trigger mechanism and the multi-mode control mechanism can adjust the reference value of the fuel cell as the driving conditions change. fc-ref When a certain operating condition remains constant, continuously triggering the controller may lead to redundant computational load. Considering the limited computational resources of the fuel cell vehicle control unit, event-triggered model predictive control based on state errors can significantly reduce the computational burden and is very suitable for application in the energy management system of fuel cell vehicles.

[0154] Specifically, event triggering conditions and event triggering thresholds are designed based on the disturbance boundary, penalty matrix and prediction time domain; the difference between the ideal state output of the fuel cell vehicle and the optimal control output of the fuel cell vehicle is obtained and judged; if the event triggering threshold is exceeded, the model predictive control is triggered at the next moment; if the event triggering threshold is not exceeded, the model predictive control is not triggered at the next moment, and the control output at time k+1 in the prediction time domain is used as the optimal control sequence of the fuel cell vehicle to realize energy management of the fuel cell vehicle.

[0155] In this embodiment, first, the event triggering condition is designed. Assume that x * (τ; t k ) represents the predicted state obtained based on the optimal control calculation, x(τ; t k ) represents the actual state. Due to external disturbances, such as the influence of the internal temperature of the battery and the response delay of the fuel cell, x(τ; t k ) gradually deviates from x * (τ; t k ). Given a time series t representing the triggering moment k , the event trigger conditions are designed as follows:

[0156]

[0157] in represents the trigger threshold; P represents the terminal penalty matrix; τ=[t k+1 ,t k+1 ].

[0158]

[0159] Among them, T p =N p ΔT. This constraint ensures that the interval between two consecutive triggering moments does not exceed the prediction range.

[0160]

[0161] where ρ w is the perturbation limit of the system; β(t) is an adjustable constant; L represents the Lipschitz constant.

[0162]

[0163] Where β0=0.5; K s , K f is the proportionality coefficient.

[0164] Next, the stability of the optimization problem is derived.

[0165] First, assume the following assumptions: Given two matrices R>0 and Can get where V(x(t))=||x(t)|| P , In order to ensure the feasibility of the algorithm, three conditions are given. The execution interval time has an upper limit: Applying the triangle inequality, Then using the Gronwall-Bellman inequality, According to the triggering conditions, t k+1 -t k ≥β(t)T p

[0166] Prediction range T p Meet the following two conditions: 1. 2.

[0167] External disturbances are bounded: Where 1-γ<β(t).

[0168] Then the stability of the algorithm is proved: First, assume that τ∈[t k+1 ,t k +T], according to the trigger condition and the triangle inequality, we can get Then, using the Gronwall-Bellman inequality, the formula is established: ||x(τ; t k+1 )|| P ≤||x * (τ; t k )|| P +σexp(A(t k +Tt k+1 )); when τ=t k +T, since there is an upper limit on the execution interval, we can get ||x(t k +T;t k+1 )|| P ≤γε+σexp(A(T-βT)). Considering that the perturbation is bounded, we get ||x(t k +T;t k+1 )|| P ≤ε. Secondly, when τ∈(t k +T,t k+1 +T], because Can get When τ=t k+1 +T, we can get V(x(t k+1 +T;t k+1 ))≤γε 2 It can be concluded that it will enter the invariant set x(t k+1 +T;t k+1 )∈Ω(ε). Therefore, the algorithm stability can be proved.

[0169] Next, determine whether the square of the state difference norm exceeds the event trigger threshold. If it does, the model predictive control is triggered at the next moment. If it does not, the model predictive control is not triggered at the next moment, and the control output at time k+1 in the prediction domain is used.

[0170] The calculated output power of the fuel cell and lithium-ion battery is transmitted to the battery management system via the CAN bus, and the desired power is output according to the control instructions to provide power for the drive motor and control the longitudinal movement of the fuel cell vehicle.

[0171] In summary, the embodiment of the present invention first calculates the total power demand, calculates the total power demand of the vehicle at the current moment according to the vehicle speed, efficiency and related vehicle information, and calculates the SoC and equivalent hydrogen consumption at the current moment according to the lithium-ion battery SoC and the instantaneous equivalent hydrogen consumption rate equation; then predicts the short-term future vehicle speed information, combines the historical vehicle speed information, and predicts the vehicle speed in the next 5 seconds; further, the energy management multi-mode selection mechanism performs feature analysis on the predicted vehicle speed information, and identifies the short-term future working condition type according to the predicted vehicle speed characteristics in the next 5 seconds and the current working condition type and duration; selects the corresponding output mode according to the identified short-term future working condition type, and stores the working condition duration; then, based on the energy management strategy of model predictive control, according to the output mode , determine a reference value for the fuel cell output power; then, based on the total required power of the vehicle, combined with the lithium-ion battery charge state SoC and the instantaneous equivalent hydrogen consumption rate, the optimal control output at time k, that is, the output power of the fuel cell and lithium-ion battery, is solved based on the model predictive control method; finally, through the event triggering mechanism, according to the optimal control sequence solved by the model predictive control method, the ideal state (SoC, equivalent hydrogen consumption and fuel cell output power) in the prediction time domain is obtained, and the difference between the actual state and the ideal state caused by the disturbance is calculated to determine whether it exceeds the event triggering threshold. If so, the model predictive control is triggered at the next moment. If not, the model predictive control is not triggered at the next moment, and the control output at time k+1 in the prediction time domain is used.

[0172] The embodiments of the present invention differ from the prior art in the following ways:

[0173] 1) The present invention designs a future short-term vehicle speed prediction technology, adopts the Bayesian optimization method to optimize the core parameters of the radial basis function neural network, and uses the optimized radial basis function neural network to predict the future vehicle speed, effectively improving the vehicle speed prediction accuracy.

[0174] 2) The present invention designs a multimodal energy management method based on model predictive control. According to the vehicle speed characteristics in the next 5-second window, the operating conditions are divided into four categories, and the output mode of the fuel cell is determined for each corresponding operating condition; the discrete state space equation is constructed by combining SoC and the instantaneous equivalent hydrogen consumption equation, and the rolling optimization solution is obtained in the prediction time domain to obtain the optimal solution control sequence, reasonably allocate power, adjust SoC, reduce hydrogen consumption, and avoid drastic power fluctuations of the fuel cell.

[0175] 3) The present invention designs an event triggering mechanism, which triggers the model predictive control to solve the optimal control sequence when the state difference exceeds the trigger threshold. It also proves the stability of the energy management strategy under the event triggering mechanism, which not only ensures the efficiency of energy management control, but also significantly reduces the calculation amount of the control unit.

[0176] Therefore, the embodiment of the present invention provides a priori information by establishing a vehicle speed prediction model, switches the corresponding control mode for different driving conditions, and allocates reasonable power to the fuel cell and lithium-ion battery to achieve multi-objective optimization and adaptability. Its comprehensive performance is better than the comparison method:

[0177] 1) Stabilize SoC trajectory and reduce hydrogen consumption;

[0178] 2) Avoid severe dynamic loads on fuel cell output power and extend fuel cell service life;

[0179] 3) Improve energy management efficiency;

[0180] 4) Compared with the algorithm without event triggering mechanism, the computational workload is reduced by 56.31%.

[0181] Reference Figure 2 , a multi-modal energy management and control system for new energy vehicles based on event triggering, including:

[0182] The first module is used to obtain the total required power of the fuel cell vehicle and predict the vehicle speed information of the fuel cell vehicle in the short term;

[0183] The second module is used to perform feature analysis on the fuel cell vehicle's speed information in the short term future, and determine the fuel cell vehicle's short term future operating condition type and the corresponding output mode;

[0184] The third module is used to combine the output mode of the fuel cell vehicle with the total required power of the fuel cell vehicle and obtain the optimal control output of the fuel cell vehicle through a model-based predictive control method;

[0185] The fourth module is used to predict and judge the optimal control output of the fuel cell vehicle based on the event trigger mechanism, determine the optimal control sequence of the fuel cell vehicle, and realize energy management of the fuel cell vehicle.

[0186] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0187] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A multi-modal energy management control method for new energy vehicles based on event triggering, characterized in that: The following steps are involved: Obtain the total required power of the fuel cell vehicle and predict the vehicle speed information of the fuel cell vehicle in the short term; Perform characteristic analysis on the fuel cell vehicle's speed information in the short term future to determine the fuel cell vehicle's short term future operating condition type and corresponding output mode; Combining the output mode of the fuel cell vehicle with the total power demand of the fuel cell vehicle, the optimal control output of the fuel cell vehicle is obtained through a model-based predictive control method; Based on the event triggering mechanism, the optimal control output of the fuel cell vehicle is predicted and judged, the optimal control sequence of the fuel cell vehicle is determined, and the energy management of the fuel cell vehicle is realized.

2. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 1 is characterized in that: The step of obtaining the total required power of the fuel cell vehicle and predicting the vehicle speed information of the fuel cell vehicle in the short term includes: Determining driving information of the fuel cell vehicle according to the driving resistance of the fuel cell vehicle; Determine the output power of the fuel cell vehicle drive motor based on the output torque, speed and efficiency of the fuel cell vehicle drive motor; Obtain the output power of the fuel cell stack of the fuel cell vehicle, and determine the total power requirement of the fuel cell vehicle by combining the driving information of the fuel cell vehicle and the output power of the fuel cell vehicle drive motor; Based on the RBFNN model, the historical speed data of fuel cell vehicles is obtained, and the speed information of fuel cell vehicles in the short term in the future is predicted through a sliding window.

3. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 2 is characterized in that: The step of obtaining historical speed data of fuel cell vehicles based on the RBFNN model and predicting the short-term future speed information of fuel cell vehicles through a sliding window specifically includes: Determine the activation function and output of the RBFNN model, and obtain the diffusion factor and the number of hidden layer neurons of the RBFNN model; The diffusion factor and the number of hidden layer neurons of the RBFNN model are optimized by the Bayesian optimization algorithm to obtain the optimized RBFNN model; Based on the optimized RBFNN model, the historical speed data of fuel cell vehicles is input and the speed information of fuel cell vehicles in the short term in the future is predicted through a sliding window.

4. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 3 is characterized in that: The step of performing feature analysis on the fuel cell vehicle's speed information in the short-term future to determine the fuel cell vehicle's short-term future operating condition type and the corresponding output mode specifically includes: Using a preset future time in the short-term future fuel cell vehicle speed information as a dynamic sliding window, extracting four types of features from each window, the four types of features including average speed, maximum speed, maximum acceleration, and minimum acceleration; determining whether the four characteristics exceed preset thresholds and whether the duration of the current fuel cell vehicle driving condition meets a minimum duration, and marking the future window, wherein the fuel cell vehicle driving condition includes low speed driving, medium speed driving, medium-high speed driving, and high speed driving; The output mode of the fuel cell vehicle is determined according to the driving conditions of the fuel cell vehicle, where: The output modes of the low-speed and medium-speed driving are that the fuel cell provides a first output power, and the remaining power is provided by the lithium-ion battery; The output mode for medium and high speed driving is that the output power of the lithium-ion battery remains unchanged, and the fuel cell provides a second output power, which is greater than the first output power; The output mode of high-speed driving is that the fuel cell provides a third output power, and the remaining power is provided by the lithium-ion battery. The third output power is greater than the second output power.

5. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 4 is characterized in that: The step of combining the output mode of the fuel cell vehicle with the total required power of the fuel cell vehicle and obtaining the optimal control output of the fuel cell vehicle through a model-based predictive control method specifically includes: Obtain the SoC and instantaneous equivalent hydrogen consumption equation of fuel cell vehicles; Determine the discrete equation of the SoC of the fuel cell vehicle based on the SoC of the fuel cell vehicle, and determine the discrete equation of the instantaneous equivalent hydrogen consumption based on the instantaneous equivalent hydrogen consumption equation; The discrete state space equation is constructed by combining the discrete equation of the SoC of the fuel cell vehicle with the discrete equation of the instantaneous equivalent hydrogen consumption; Based on the total power demand of the fuel cell vehicle as the disturbance variable and the change in the output power of the fuel cell as the control variable, the discrete state space equation is constrained, and the discrete state transfer equation and observation variables are constructed; Based on the estimation of fuel cell vehicle economy and SoC change, a constrained cost function for fuel cell vehicles is constructed; The constrained cost function of the fuel cell vehicle is converted into a quadratic programming form, and combined with the output mode of the fuel cell vehicle, multi-objective optimization is performed to obtain the optimal control output of the fuel cell vehicle.

6. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 5 is characterized in that: The expression of the optimal control output of the fuel cell vehicle is: In the above formula, represents the optimal control output of the fuel cell vehicle, J(k) represents the constrained cost function of the fuel cell vehicle, represents the weight matrix, U(k) represents the calculated optimal control sequence, represents the quadratic coefficient matrix of the objective function, represents the linear term vector, M and E represent the inequality constraint matrix, U min Indicates the lower limit of the control quantity, U max represents the upper limit of the control quantity, Ξ, Ψ, Ω all represent the reconstructed state matrix, ρ represents the relaxation factor, x(k) represents the state quantity, W represents the disturbance quantity matrix, Indicates the reference value of the output quantity.

7. The event-triggered multi-modal energy management control method for new energy vehicles according to claim 6 is characterized in that: The step of predicting and judging the optimal control output of the fuel cell vehicle based on the event triggering mechanism, determining the optimal control sequence of the fuel cell vehicle, and realizing energy management of the fuel cell vehicle specifically includes: Design event trigger conditions and event trigger thresholds based on disturbance boundaries, penalty matrices, and prediction time domains; Obtaining the difference between the ideal state output of the fuel cell vehicle and the optimal control output of the fuel cell vehicle and making a judgment; If the event trigger threshold is exceeded, the model predictive control is triggered at the next moment; If the event trigger threshold is not exceeded, the model predictive control will not be triggered at the next moment, and the control output at time k+1 in the prediction time domain will be used as the optimal control sequence for the fuel cell vehicle to achieve energy management of the fuel cell vehicle.

8. The event-triggered multi-modal energy management and control system for new energy vehicles is characterized by: Includes the following modules: The first module is used to obtain the total required power of the fuel cell vehicle and predict the vehicle speed information of the fuel cell vehicle in the short term; The second module is used to perform feature analysis on the fuel cell vehicle's speed information in the short term future, and determine the fuel cell vehicle's short term future operating condition type and the corresponding output mode; The third module is used to combine the output mode of the fuel cell vehicle with the total required power of the fuel cell vehicle and obtain the optimal control output of the fuel cell vehicle through a model-based predictive control method; The fourth module is used to predict and judge the optimal control output of the fuel cell vehicle based on the event trigger mechanism, determine the optimal control sequence of the fuel cell vehicle, and realize energy management of the fuel cell vehicle.

Images