Multi-target control method for fuel cell hydrogen supply system based on switching type model prediction

Abstract

The invention relates to a multi-target control method for a fuel cell hydrogen supply system based on switching type model prediction. The method comprises the following steps: S1, constructing a high-order nonlinear model of the fuel cell hydrogen supply system; s2, obtaining a high-load linear sub-model and a low-load linear sub-model; s3, converting the high-load linear sub-model and the low-load linear sub-model into a high-load model prediction controller and a low-load model prediction controller respectively; and S4, acquiring the actual load current of the electric pile, judging the adopted model prediction controller based on a switching strategy judged by a load current threshold, outputting a control variable by the adopted model prediction controller, controlling the fuel cell hydrogen supply system based on the control variable, and repeating the step S4 until the control is finished. Compared with the prior art, the hydrogen supply system has the advantages that the combined precise adjustment of the anode pressure and the hydrogen passing ratio is realized, so that the dynamic response performance of the hydrogen supply system and the overall operation stability of the system are improved, and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fuel cell, in particular to a switching type model prediction fuel cell hydrogen supply system multi-objective control method. BACKGROUND

[0002] Proton exchange membrane fuel cell is widely concerned in the field of new energy vehicles, especially heavy commercial vehicles, due to its high energy conversion efficiency, low working temperature, fast start and other advantages. As an important part of the fuel cell system, the control strategy of the hydrogen supply system will directly affect the dynamic response performance of the stack, the hydrogen utilization rate, and the stability and life of the system.

[0003] The anode side of the fuel cell stack needs to maintain a certain pressure difference with the cathode side, which is determined by the pressure limit of the proton exchange membrane. By adjusting the anode flow field pressure, the pressure difference on both sides of the membrane can be effectively controlled within a suitable range, thereby prolonging the service life of the proton exchange membrane. On the other hand, a good hydrogen supply system control strategy can ensure that the hydrogen concentration in the anode flow field of the stack is appropriate, thereby improving the electrochemical reaction rate and avoiding the "hydrogen starvation" problem of the stack.

[0004] In actual operation, rapid changes in vehicle operating conditions (rapid acceleration, regenerative braking) will cause the fuel cell load current to fluctuate greatly, thereby requiring higher response speed and adjustment accuracy of the hydrogen supply system. The current mainstream hydrogen supply system control strategy mostly uses open-loop control or single variable closed-loop control, which only adjusts the proportional valve opening or circulating pump speed based on empirical rules, resulting in insufficient adjustment accuracy of anode pressure and hydrogen over ratio. SUMMARY

[0005] The purpose of the present application is to achieve joint precise adjustment of anode pressure and hydrogen over ratio, thereby improving the dynamic response performance of the hydrogen supply system and the overall operation stability of the system, and a switching type model prediction fuel cell hydrogen supply system multi-objective control method is provided.

[0006] The purpose of the present application can be achieved by the following technical solutions:

[0007] A switching type model prediction fuel cell hydrogen supply system multi-objective control method, the method comprising the following steps:

[0008] S1, a high-order nonlinear model of a fuel cell hydrogen supply system is constructed;

[0009] S2, the nonlinear model is linearized at high load and low load to obtain a high load linear submodel and a low load linear submodel, and the linear submodel is in the form of a state space equation;

[0010] S3, discretize the high-load linear sub-model and the low-load linear sub-model respectively to form a high-load model predictive controller and a low-load model predictive controller;

[0011] S4, obtain an actual load current of the stack, determine a model predictive controller to be used based on a switching strategy determined by a load current threshold, output a control variable by the model predictive controller to be used, control the hydrogen supply system of the fuel cell based on the control variable, and repeat S4 until the control is ended.

[0012] Further, the state space equation is:

[0013]

[0014] wherein X(t) is a state vector, U(t) is a control vector, d(t) is a measurable disturbance quantity, A is a state matrix, B u is a control input matrix, B d is a disturbance input matrix, δ i is a disturbance unmeasurable by the system, C is an output matrix, D u is a direct transmission matrix of the control, D d is a direct transmission matrix of the disturbance, and Y(t) is an output vector.

[0015] Further, the state vector, the control vector, the measurable disturbance quantity, and the output vector of the state space equation are:

[0016]

[0017] wherein, is an anode hydrogen pressure, is an anode supply manifold hydrogen pressure, is an anode return manifold hydrogen pressure, is an anode water vapor pressure, is an anode supply manifold water vapor pressure, is an anode return manifold water vapor pressure, ω pump is an angular velocity of a hydrogen circulation pump; P anode represents an anode pressure, represents a hydrogen excess ratio, u fcv and u pump respectively represent control voltages of proportional valves and circulation pumps, I st represents a fuel cell stack current.

[0018] Further, the state space equation of the model discretization is:

[0019]

[0020] wherein ΔX represents an increment of a state variable, A dB ud B dd They are matrices A and B respectively. u B d The discretized matrix representation is as follows: ΔU(k) ​​represents the column vector composed of control increments in the next m steps, and Δd(k) represents the column vector composed of disturbance increments in the next p steps.

[0021] Furthermore, the state matrix, control input matrix, disturbance input matrix, output matrix, control direct transfer matrix, and disturbance direct transfer matrix of the spatial state equations corresponding to the high-load linear sub-model and the low-load linear sub-model are all different.

[0022] Furthermore, the specific steps for determining the model predictive controller used in the switching strategy based on the load current threshold are as follows:

[0023] If the actual load current of the fuel cell stack is I stack If I0 > I0 + β, where I0 is the preset threshold current and β is an additional threshold, then a high-load model predictive controller is used.

[0024] If the actual load current of the fuel cell stack is I stack If <I0+β, then a low-load model predictive controller is used;

[0025] If the actual load current of the fuel cell stack is within I stack In the interval [I0-β, I0+β], then:

[0026] Calculate the slope of the short-time current change. If the slope of the short-time current change is greater than a small positive threshold, a high-load model predictive controller is used; otherwise, a low-load model predictive controller is used.

[0027] Furthermore, the slope of the short-time current change is:

[0028]

[0029] Where slope represents the slope of the short-time current change, n represents the nth sampling point, N represents the sampling interval, and T s Indicates the sampling period.

[0030] Furthermore, the specific steps for using the model to predict the output control variables of the controller are as follows:

[0031] The prediction equation is derived from the discretized state-space equation of the model. Based on the prediction equation, the objective function of the multi-objective model predictive control is obtained. The control variables are obtained by solving the objective function of the multi-objective model predictive control.

[0032] Furthermore, the objective function is:

[0033]

[0034] wherein, γ, κ are weight factors, Y P is the matrix of future p steps controlled variables, R is the matrix containing reference values of anode pressure and hydrogen over ratio, ΔU represents the control input column vector, p is the prediction horizon, and the control horizon is m.

[0035] Further, the matrix of future p steps controlled variables is:

[0036] Y p (k+1|k)=P X ΔX(k)+P U ΔU(k)+P d Δd(k)+EY c (k)

[0037] wherein, P X , P U , P d respectively represent the prediction matrix of state vector, control vector and disturbance vector, and E represents the unit matrix.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] The present application designs MPC controllers for high and low load conditions respectively, and introduces a switching mechanism, effectively avoiding the performance decline of a single controller under non-design conditions, and improving the adaptability and robustness of the control system under complex dynamic environment. The multi-model switching control mode can better adjust the anode hydrogen pressure and hydrogen metering ratio under different loads, realize fast response and accurate tracking, reduce steady-state error and overshoot, and improve the efficiency and output stability of the fuel cell system.

[0040] Since the optimization problem of the standard MPC is established based on a linear system model, linearization operation needs to be performed on the current nonlinear hydrogen supply system model. The core idea is to use a linear model to approximate the original nonlinear model near a certain working point, so as to simplify the controller design. The working point is a set of variable values of the hydrogen supply system under a certain stable operating condition, usually including state variables, input variables, disturbance variables, etc. When linearizing, the linearization model can be obtained near the current working point by using the numerical perturbation method, combined with the model state equation:

[0041]

[0042] The load current is selected to be constant at 200A, 250A and 330A respectively for linearization operation of the hydrogen supply system. The specific information of the working point of the above hydrogen supply system is shown in Table 1:

[0043] Table 1 Working point of hydrogen supply system

[0044]

[0045] where x1 to x6 represent the partial pressure of hydrogen and water vapor in the supply manifold, anode flow channel and return manifold respectively (Pa) under the current load current when the system is running stably, x7 represents the rotating speed of the hydrogen circulation pump (rad / s), and u1 and u2 represent the opening degree of the current proportional valve (dimensionless 0-1) and the circulation pump voltage signal (V) respectively.

[0046] Since the fuel cell hydrogen supply system presents significant nonlinear characteristics in actual operation, a single linear MPC controller often cannot effectively cover all working conditions. When the system working conditions deviate far from the linearization working point, the control precision and stability will significantly decrease. In order to overcome this problem, a switching MPC control strategy based on piecewise linearization model is proposed to effectively cope with the changes of the nonlinear system in a wide range. The piecewise linearization multi-model predictive control is suitable for processing systems with a larger working range. In this paper, the normal working current range of the fuel cell is approximately 100 to 400 A. During operation, the change of the current will cause a huge change of the internal parameters of the stack. The piecewise linearization divides the working range of the entire system into high load condition and low load condition according to the load current. The information in the table is set, the working point 1 is selected as the linearization working point under the low load condition, and the working point 2 is selected as the linearization working point under the high load condition; the linear state space model of the hydrogen supply system is established for the two working points respectively, and the corresponding MPC controller is designed to cover the working range of the system. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 is the control principle diagram of the fuel cell hydrogen supply system of the present application;

[0048] Figure 2 is the switching determination strategy principle diagram of the model predictive controller under high and low conditions in the present application;

[0049] Figure 3 is the anode pressure and over hydrogen ratio control comparison diagram of the fuel cell system under the load step current condition in the present application. DETAILED DESCRIPTION

[0050] The present application will be described in detail below in combination with the drawings and specific embodiments. The embodiments are implemented on the premise of the technical scheme of the present application, and detailed implementation methods and specific operation processes are given, but the protection scope of the present application is not limited to the following embodiments.

[0051] Vehicle working condition changes rapidly (rapid acceleration, regenerative braking) can cause fuel cell load current fluctuation, and then put forward higher response speed and adjustment accuracy requirements for hydrogen supply system. At present, the control strategy of the mainstream hydrogen supply system mostly uses open loop control or single variable closed loop control, which only adjusts the proportional valve opening or circulating pump speed based on empirical rules. Therefore, the present application proposes a fast and accurate control method with multi-variable coupling coordination ability to realize the joint control of anode pressure and over hydrogen ratio, and improve the stability, safety and economy of system operation.

[0052] The purpose of the present application is to provide a switching type multi-objective model predictive control method for proton exchange membrane fuel cell hydrogen supply system, to realize the joint precise adjustment of anode pressure and over hydrogen ratio, and to improve the dynamic response performance of hydrogen supply system and the overall stability of system operation.

[0053] To achieve the above purpose, the present application proposes the following technical solutions:

[0054] A fuel cell hydrogen supply system anode pressure and over hydrogen ratio control strategy based on circulating pump, comprising the following steps:

[0055] Step S1: According to the characteristics of vehicle fuel cell proton exchange membrane fuel cell hydrogen supply system, a high-order nonlinear model of fuel cell hydrogen supply system including proportional valve, supply manifold, return manifold, hydrogen circulating pump and anode flow channel and other key components is constructed, and it is assumed that the oxygen supply on the cathode side of the stack meets the needs;

[0056] Step S2: The above nonlinear model is linearized at representative working points (high load and low load) to obtain two typical linear submodels, which are used to describe the local dynamic behavior of hydrogen supply system under high load and low load conditions respectively, and the linearized model is converted into the form of state space equation;

[0057] Step S3: For the high and low load models after linearization, a model predictive control algorithm is used to design a controller, which takes anode pressure and over hydrogen ratio as controlled output variables, and proportional valve opening and circulating pump driving voltage as control input variables. By constructing optimization objective function and constraint conditions, the future control sequence is predicted and optimized to realize multi-objective dynamic adjustment, and the corresponding physical constraints of the controlled object and the reference output of the system are added;

[0058] Step S4: For the two model predictive controllers designed in step S3, a switching strategy based on load current threshold is designed to realize the efficient switching of multiple MPC controllers under different operating conditions. This strategy can select the corresponding sub-model and its MPC controller according to the current load level of the fuel cell stack, realize smooth switching between controllers, avoid system jitter or control discontinuity, and ensure that the system has good control performance in the full operating range.

[0059] Further, the high-order nonlinear model of the hydrogen supply system of the fuel cell established in S1 is:

[0060]

[0061] wherein is a state variable, specifically is the pressure of hydrogen in the anode supply manifold, is the gas constant of hydrogen, T sm is the supply manifold temperature, V sm is the supply manifold volume, is the hydrogen flow rate into the supply manifold, is the hydrogen flow rate out of the supply manifold, is the anode hydrogen pressure, T an is the anode temperature, V an is the anode flow channel volume, is the hydrogen flow rate into the anode, is the hydrogen flow rate out of the anode, is the hydrogen flow rate consumed in the electrochemical reaction, is the hydrogen pressure in the anode return manifold, T rm is the return manifold temperature, V rm is the return manifold volume, is the hydrogen flow rate into the return manifold, is the hydrogen flow rate out of the return manifold, is the water vapor pressure in the anode supply manifold, is the gas constant of water vapor, is the water vapor flow rate into the supply manifold, is the water vapor flow rate out of the supply manifold, is the anode water vapor pressure, is the water vapor flow rate into the anode, is the water vapor flow rate out of the anode, is the water vapor diffusion rate from the anode to the cathode, is the water vapor pressure in the anode return manifold, is the water vapor flow rate into the return manifold, is the water vapor flow rate out of the return manifold, ωpump is the angular velocity of the hydrogen circulation pump, J pump is the motor rotational inertia, τ cm is the circulation pump driving torque, τ cp is the motor torque.

[0062] Specifically, when the liquid water generation condition is not met in the supply manifold, the dynamic process of hydrogen and water vapor in the supply manifold is:

[0063]

[0064] Where the hydrogen flow rate flowing into the supply manifold is equal to the sum of the hydrogen flow rate flowing out of the proportional valve and the hydrogen flow rate flowing out of the circulation pump. Since the pressure difference between the supply manifold and the anode of the stack is small, the fluid density does not change much, so the mass flow rate at the outlet of the pipeline and the pressure difference between the two ends of the pipeline can be approximately proportional, which can be expressed as:

[0065]

[0066] Where u fcv is the control voltage of the proportional valve, ranging from 0 to 1, W fcv，full is the maximum mass flow rate of the proportional valve, W pump is the gas mass flow rate of the circulation pump, is the mass fraction of hydrogen in the return manifold, which can be expressed as:

[0067]

[0068] In the formula, δ is the correction coefficient, λ pres , λ temp are the pressure correction and temperature correction coefficients, respectively, ρ rm is the gas density of the return manifold, d pump is the diameter of the circulation pump blade, R g,rm is the gas constant in the manifold. Similarly, the hydrogen mass flow rates flowing into the anode and flowing out of the anode the hydrogen mass flow rate flowing into the return manifold the water vapor mass flow rate flowing out of the supply manifold the water vapor mass flow rates flowing into and out of the anode and the water vapor mass flow rate flowing into the return manifold can all be expressed in approximately proportional relationships:

[0069]

[0070] In the formula, k sm , k rm are the mass flow rate coefficients at the outlets of the supply manifold and the return manifold, respectively.

[0071] Furthermore, the hydrogen permeability ratio of the vehicle fuel cell refers to the flow rate of hydrogen entering the anode channel of the battery. Hydrogen flow rate consumed in the battery electrochemical reaction The ratio can be expressed as:

[0072]

[0073] In the formula N cell It refers to the number of individual fuel cell cells. It is the molar mass of hydrogen, I st It is the fuel cell stack current, and F is the Faraday constant.

[0074] Furthermore, the specific method for obtaining the state parameter data of fuel cells under high and low load conditions based on a high-order nonlinear model is as follows:

[0075] 1. First, establish the corresponding nonlinear model of the hydrogen supply system in the Matlab / Simulink simulation software environment, and record the required state parameters in the model using the corresponding observation module;

[0076] 2. Next, input constant high and low operating conditions corresponding to the load current to simulate the load current of the fuel cell system under steady-state conditions. Specifically, select a constant high operating condition load current of 330A and a constant low operating condition load current of 200A.

[0077] 3. Observe and record the internal parameter data of the hydrogen supply system model under high and low load currents.

[0078] Furthermore, step S2 specifically involves: performing local linearization processing on the nonlinear system model at the rated operating point based on the numerical perturbation method; introducing small perturbations sequentially to the system state variables, control input, and perturbation input; calculating the finite difference between the system state derivative function and the output function; thereby constructing the Jacobian matrix of the system; and thus obtaining the linear state-space model.

[0079] Furthermore, step S3 yields the system state equations based on the linearization method described above, as shown below:

[0080]

[0081] in

[0082]

[0083] In the formula, X(t) is the state vector, U(t) is the control vector, d(t) is the measurable disturbance, A is the state matrix, and B is the control vector. u To control the input matrix, B d Let δ be the perturbation input matrix. iC is the disturbance directly transmitted to the system, D is the output matrix, and u C is the disturbance directly transmitted to the system, D is the output matrix, and d C is the disturbance directly transmitted to the system, D is the output matrix, and

[0084]

[0085] Based on the above model, the prediction equation of the system is derived:

[0086] Y p (k+1|k)=P X ΔX(k)+P U ΔU(k)+P d Δd(k)+EY c (k)

[0087] Y p (k+1|k) is the output of the system calculated by the model predictive control at time k, p is the prediction horizon, ΔU(k) is the control column vector applied, the control horizon is m, and P X ,P U ,P d is the prediction matrix. Based on the prediction equation of the output variable within the prediction horizon, the objective function of the multi-objective model predictive control is established:

[0088]

[0089] where γ and κ are weight factors, and R is a matrix containing the reference values of the anode pressure and the over-hydrogen ratio. Finally, the model predictive controller of the fuel cell system under high and low load conditions is obtained.

[0090] Y p (k+i|k) is the column vector of the future P-step controlled variables (anode pressure and over-hydrogen ratio) calculated at time k, i.e., the prediction equation is:

[0091] Y p (k+1|k)=P X ΔX(k)+P U ΔU(k)+P d Δd(k)+EY c (k)

[0092] where ΔX, ΔU, Δd, are the same as listed in the state equation:

[0093]

[0094] Substituting the objective function

[0095] Further, since the model predictive controller is designed based on the linearization of the hydrogen supply system at a single nominal operating point, the prediction model error will be large when deviating from the operating point. To ensure that the system has good control performance in the entire operating range, a switching strategy based on load current threshold is designed to realize efficient switching of multiple MPC controllers in different operating conditions. The strategy can select the corresponding sub-model and its MPC controller according to the current load level of the fuel cell stack, realize smooth switching between controllers, and avoid system jitter or control discontinuity. The step S4 is specifically:

[0096] First, the operating condition is determined, and the load current I of the fuel cell stack is measured in real time stack , and compared with the preset threshold current I0. If I stack >I0, the system enters the high load mode, otherwise it enters the low load mode;

[0097] Further, to prevent the controller from switching back and forth due to slight jitter of the current near the boundary, a buffer band [I0-β, I0+β] is introduced. If I stack >I0+β, the high load operating condition model predictive controller is selected, if I stack <I0+β, the low load operating condition model predictive controller is selected, and if I stack falls within the interval [I0-β, I0+β], the current trend is determined;

[0098] In the buffer interval, the short-term current change slope is calculated:

[0099]

[0100] Compare with the preset small positive threshold ε, if slope>ε (indicating that the current is rising significantly), switch or remain as the high load operating condition model predictive controller, if slope<ε (indicating that the current is decreasing significantly or changing slowly), switch or remain as the low load operating condition model predictive controller;

[0101] Finally, according to the above determination results, the tasks of adjusting the valve opening and the circulating pump speed are handed over to the selected model predictive controller, and each model predictive controller internally performs rolling optimization based on its corresponding linear model to generate optimal control increments and output to the actuator.

[0102] Compared with the prior art, the present application has the following beneficial effects:

[0103] 1. The present application adopts nonlinear modeling and segmented linearization for different load conditions, which can more truly reflect the dynamic characteristics of the fuel cell hydrogen supply system in different operating states, thereby improving the modeling accuracy.

[0104] 2. MPC controllers are designed for high and low load conditions respectively, and a switching mechanism is introduced to effectively avoid the performance degradation of a single controller under non-design conditions, thereby improving the adaptability and robustness of the control system in complex dynamic environments.

[0105] 3. The multi-model switching control mode can better adjust the anode hydrogen pressure and hydrogen metering ratio under different loads, achieve fast response and accurate tracking, reduce steady-state error and overshoot, and improve the efficiency and output stability of the fuel cell system.

[0106] 4. The switchable MPC framework has good modularity and feasibility, can be easily embedded into the controller, has good engineering application prospects, and is suitable for intelligent hydrogen supply control in commercial fuel cell systems.

[0107] Reference Figure 1 This invention provides a switching multi-objective model predictive control method for a proton exchange membrane fuel cell hydrogen supply system, comprising the following steps:

[0108] Step S1: Based on the characteristics of the hydrogen supply system for a proton exchange membrane fuel cell in a vehicle...

[0109] To construct a high-order nonlinear model of a fuel cell hydrogen supply system, including key components such as proportional valves, supply manifolds, return manifolds, hydrogen circulation pumps, and anode flow channels, it is assumed that the oxygen supply on the cathode side of the stack meets the requirements.

[0110] Step S2: Apply the above nonlinear model to representative operating points (high load and low load).

[0111] Piecewise linearization is performed to obtain two typical linear sub-models, which are used to describe the local dynamic behavior of the hydrogen supply system under high load and low load conditions, respectively, and the linearized model is converted into the form of state-space equations.

[0112] Step S3: For the linearized high and low load models, a model predictive control algorithm is adopted.

[0113] A separate controller is designed, with anode pressure and hydrogen ratio as the controlled output variables and proportional valve opening and circulating pump drive voltage as the control input variables. By constructing an optimization objective function and constraints, the controller predicts and optimizes the future control sequence to achieve dynamic adjustment of multiple objectives. The corresponding physical constraints of the controlled object and the reference output of the system are also added.

[0114] Step S4: For the two model predictive controllers designed in step S3, a switching strategy based on load current threshold is designed to realize the efficient switching of multiple MPC controllers under different operating conditions. This strategy can select the corresponding sub-model and its MPC controller according to the current load level of the fuel cell stack, realize smooth switching between controllers, avoid system jitter or control discontinuity, and ensure that the system has good control performance in the full operating range.

[0115] In this embodiment, the established model of the hydrogen supply system of the fuel cell includes a proportional valve model, a supply manifold and a return manifold model, and a circulating pump model, which are described as follows:

[0116] The proportional valve is a variable opening nozzle, and the expression of the control input signal of the model under steady state is:

[0117] W fcv =u fcv W fcv,full

[0118] Where u fcv is the control voltage of the proportional valve, ranging from 0 to 1, W fcv，full is the maximum mass flow rate of the proportional valve;

[0119] In the manifold model, it is assumed that there is no liquid water in the manifold under ideal conditions, and the dynamic process can be described as:

[0120]

[0121] is the pressure of hydrogen in the anode supply manifold, is the gas constant of hydrogen, T sm is the supply manifold temperature, V sm is the supply manifold volume, is the hydrogen flow rate into the supply manifold, is the hydrogen flow rate out of the supply manifold, is the hydrogen flow rate consumed in the electrochemical reaction, is the pressure of hydrogen in the anode return manifold, T rm is the return manifold temperature, V rm is the return manifold volume, is the hydrogen flow rate into the return manifold, is the hydrogen flow rate out of the return manifold, is the pressure of water vapor in the anode supply manifold, is the gas constant of water vapor, is the water vapor flow rate into the supply manifold, is the water vapor flow rate out of the supply manifold, is the pressure of water vapor in the anode return manifold, is the water vapor flow rate flowing into the return manifold, is the water vapor flow rate flowing out of the return manifold;

[0122] The dynamic process of anode hydrogen and water vapor is:

[0123]

[0124] wherein is the anode hydrogen pressure, T an is the anode temperature, V an is the anode flow channel volume, is the hydrogen flow rate flowing into the anode, is the hydrogen flow rate flowing out of the anode, is the hydrogen flow rate consumed by the electrochemical reaction of the battery, N cell is the number of fuel cell monomers, is the hydrogen molar mass, I st is the fuel cell stack current, F is the Faraday constant, is the anode water vapor pressure, is the water vapor flow rate flowing into the anode, is the water vapor flow rate flowing out of the anode, is the water vapor diffusion rate from the anode to the cathode;

[0125] The model of the hydrogen circulation pump is:

[0126]

[0127] wherein ω pump is the angular velocity of the hydrogen circulation pump, J pumpcm is the motor rotational inertia, τ cm is the circulation pump driving torque, τ cp is the motor torque, δ is the correction coefficient, λ pres , λ temp are the pressure correction and temperature correction coefficients, respectively, ρ rm is the gas density of the return manifold, d pump is the circulation pump blade diameter.

[0128] In the present embodiment, the specific method for obtaining the state parameter data of the fuel cell under high and low load working conditions based on the above model is:

[0129] 1. First, establish the corresponding hydrogen supply system nonlinear model in the Matlab / Simulink simulation software environment, and use the corresponding observation module to record the required state parameters in the model;

[0130] 2. Secondly, the constant high and low load current corresponding to the working condition is input to simulate the load current of the fuel cell system under steady state working condition, specifically, the constant high load current is selected as 330A, and the constant low load current is selected as 200A.

[0131] 3. The internal parameter data of the hydrogen supply system model under high and low load currents are observed and recorded.

[0132] In the embodiment, the numerical perturbation method is used to perform local linearization processing on the nonlinear system model, the system state variable, the control input and the disturbance input are sequentially introduced into a small disturbance, the finite difference of the system state derivative function and the output function is calculated, the Jacobian matrix of the system is constructed, and then the linear state space model is obtained:

[0133]

[0134] Among them

[0135]

[0136] In the formula, X(t) is a state vector, U(t) is a control vector, d(t) is a measurable disturbance, A is a state matrix, B u is a control input matrix, B d is a disturbance input matrix, δ i is a system unmeasurable disturbance, C is an output matrix, D u is a control direct transfer matrix, D d is a disturbance direct transfer matrix.

[0137] In the embodiment, the system continuous state space equation is discretized, and the model predictive control algorithm is used to design the controller for the high and low load working conditions:

[0138]

[0139] Based on the above model, the prediction equation of the system is derived:

[0140] Y p (k+1|k)=P X ΔX(k)+P U ΔU(k)+P d Δd(k)+EY c (k)

[0141] In the formula, Y p (k+1|k) is the output of the system calculated by the model predictive control at time k, p is the prediction time domain, ΔU(k) is the applied control column vector, the control time domain is m, P X , P U , P dThe prediction matrix is used to predict the matrix. The objective function of the multi-objective model predictive control is established based on the prediction equation of the output variable in the prediction time domain:

[0142]

[0143] wherein γ and κ are weight factors, and R is a matrix containing the reference value of the anode pressure and the over-hydrogen ratio. Finally, the model predictive controller of the fuel cell system under high and low load conditions is obtained.

[0144] In the embodiment, to ensure that the system has good control performance in the full range of working conditions and to realize efficient switching of multiple MPC controllers under different working conditions, a switching strategy based on load current threshold judgment is designed. The specific judgment rule expression is:

[0145]

[0146] wherein I st is the load current of the fuel cell stack, I0 is a preset threshold current, β is an additional threshold, the buffer band is [I0-β, I0+β], n represents the nth sampling point, N represents the sampling interval, T s represents the sampling period. Specifically, the controller switching logic principle is shown in Figure 2 .

[0147] Using this method, the corresponding sub-model and its model predictive controller can be selected in real time according to the current load current of the stack, smooth switching between controllers is realized, system jitter or control discontinuity is avoided, and the overall control strategy principle diagram is shown in Figure 1 .

[0148] The novel multi-objective model predictive control strategy of the hydrogen supply system of the above invention can realize the control of the anode pressure and the over-hydrogen ratio, as shown in Figure 3 , and the embodiment gives the anode pressure and over-hydrogen ratio control comparison diagram of the fuel cell under full-range step current conditions (the compared controller is a traditional PID controller).

[0149] The above detailed the preferred embodiments of the invention. It should be understood that those skilled in the art can make many modifications and changes without creative labor according to the concept of the invention. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment based on the existing technology according to the concept of the invention shall be within the protection scope determined by the claims.

Claims

1. A multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction, characterized in that, The method includes the following steps: S1. Construct a high-order nonlinear model of the fuel cell hydrogen supply system; S2. The nonlinear model is piecewise linearized at high load and low load to obtain a high-load linear sub-model and a low-load linear sub-model. The linear sub-model is in the form of a state-space equation. S3. Discretize the high-load linear sub-model and the low-load linear sub-model respectively, and transform them into a high-load model predictive controller and a low-load model predictive controller. S4. Obtain the actual load current of the fuel cell stack, determine the model predictive controller to be used based on the switching strategy based on the load current threshold, output the control variable of the model predictive controller, control the fuel cell hydrogen supply system based on the control variable, and repeat S4 until the control ends.

2. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 1, characterized in that, The state-space equation is: Where X(t) is the state vector, U(t) is the control vector, d(t) is the measurable disturbance, A is the state matrix, and B is the control vector. u To control the input matrix, B d Let δ be the perturbation input matrix. i Let C be the unmeasurable disturbance of the system, and D be the output matrix. u Control direct transfer matrix, D d Y(t) is the direct transfer matrix of the perturbation, and Y(t) is the output vector.

3. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 2, characterized in that, The state vector, control vector, measurable disturbance, and output vector in the state-space equations are as follows: in, It is the hydrogen pressure at the anode. It is the pressure of hydrogen in the anode supply manifold. It is the pressure of hydrogen gas returning to the anode manifold. It is the anode water vapor pressure. It is the pressure of water vapor in the anode supply manifold. It is the pressure of water vapor returning to the manifold from the anode, ω pump It is the angular velocity of the hydrogen circulation pump; P anode Indicates anode pressure. Indicates the hydrogen peroxide ratio, u fcv and u pump I represents the control voltage of the proportional valve and the circulating pump, respectively. st This indicates the current in the fuel cell stack.

4. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 3, characterized in that, The state-space equations of the discretized model are as follows: Where ΔX represents the increment of the state variable, A d B ud B dd They are matrices A and B respectively. u B d The discretized matrix representation is as follows: ΔU(k) ​​represents the column vector composed of control increments in the next m steps, and Δd(k) represents the column vector composed of disturbance increments in the next p steps.

5. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 4, characterized in that, The state matrix, control input matrix, disturbance input matrix, output matrix, control direct transfer matrix, and disturbance direct transfer matrix of the spatial state equations corresponding to the high-load linear sub-model and the low-load linear sub-model are all different.

6. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 5, characterized in that, The specific steps for determining the model predictive controller used in the switching strategy based on load current threshold judgment are as follows: If the actual load current of the fuel cell stack is I stack If I0 > I0 + β, where I0 is the preset threshold current and β is an additional threshold, then a high-load model predictive controller is used. If the actual load current of the fuel cell stack is I stack If <I0+β, then a low-load model predictive controller is used; If the actual load current of the fuel cell stack is within I stack In the interval [I0-β, I0+β], then: Calculate the slope of the short-time current change. If the slope of the short-time current change is greater than a small positive threshold, a high-load model predictive controller is used; otherwise, a low-load model predictive controller is used.

7. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 6, characterized in that, The slope of the short-time current change is: Where slope represents the slope of the short-time current change, n represents the nth sampling point, N represents the sampling interval, and T s Indicates the sampling period.

8. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 3, characterized in that, The specific steps for using the model predictive controller to output control variables are as follows: The prediction equation is derived from the discretized state-space equation of the model. Based on the prediction equation, the objective function of the multi-objective model predictive control is obtained. The control variables are obtained by solving the objective function of the multi-objective model predictive control.

9. The multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 8, characterized in that, The objective function is: Where γ and κ are weighting factors, Y P Let R be a matrix consisting of the controlled variables for the next p steps, R be a matrix containing the reference values ​​of anode pressure and hydrogen peroxide ratio, ΔU be the control input column vector, p be the prediction time domain, and m be the control time domain.

10. A multi-objective control method for a fuel cell hydrogen supply system based on switching model prediction according to claim 9, characterized in that, The matrix consisting of the controlled variables in the next p steps is: Y p (k+1|k)=P X ΔX(k)+P U ΔU(k)+P d Δd(k)+EY c (k) Among them, P X P U P d Let E represent the prediction matrices for the state vector, control vector, and disturbance vector, respectively, and let E represent the identity matrix.

Images