Fuel cell hydrogen circulation system BSMC control method based on RBF neural network

Abstract

The invention relates to a fuel cell hydrogen circulation system BSMC control method based on an RBF neural network, and belongs to the field of fuel cell system control. According to the method, a fuel cell hydrogen supply circulation anode kinetic model of a single ejector is established, and a radial basis function neural network is adopted to carry out system modeling and uncertainty compensation, so that stable control over the hydrogen supply pressure of the fuel cell is realized. According to the method, the robustness of the system is effectively improved by adopting a reverse thrust sliding mode control strategy, good control performance can be kept under various working conditions, the dynamic response speed and the steady-state precision of the FCHV of the fuel cell hybrid electric vehicle are remarkably improved especially under the conditions of load change and external interference, and the stability of the FCHV of the fuel cell hybrid electric vehicle is improved. And important technical support is provided for reliable operation, energy conservation and emission reduction of the fuel cell automobile.

Description

Technical Field

[0001] The invention belongs to the field of fuel cell system control and relates to a BSMC control method of a fuel cell hydrogen circulation system based on a RBF neural network. Background Art

[0002] With the continuous development of the automobile industry, proton exchange membrane fuel cells (PEMFC) have attracted widespread attention as a promising new power technology. Due to its high energy conversion efficiency, clean and pollution-free, and short charging time, PEMFC has shown broad development prospects in the automotive field and has become the best candidate to replace traditional internal combustion engines. In the actual operation of the proton exchange membrane fuel cell system, the anode side usually provides excess hydrogen, and the unreacted hydrogen is often directly discharged into the environment, which not only causes hydrogen waste, but also brings safety risks.

[0003] To address these issues, fuel cell systems are typically operated in dead-end and recirculation modes: in dead-end mode, the lack of forced convection and excess hydrogen supply may lead to overflow and fuel starvation within the fuel cell stack; in recirculation mode, on the contrary, the unused hydrogen from the anode is redirected back to the fuel cell through the hydrogen recirculation device, allowing it to re-enter the reaction process. This approach not only reduces the possibility of overflow, but also ensures an adequate supply of hydrogen. Currently, mainstream hydrogen recirculation equipment is mainly based on ejector and hydrogen circulation pump design. However, the use of hydrogen circulation pumps will lead to additional parasitic power consumption and reduce system efficiency. At the same time, its control and maintenance costs are high, and operating noise and vibration will also have an adverse effect on the fuel cell system; in contrast, ejectors have received significant attention due to their small size, zero energy consumption, reliable operation and low cost.

[0004] This study proposed for the first time a backstepping sliding mode controller (BSMC) to control the ejector-based fuel cell hydrogen supply system, and combined it with a radial basis function RBF neural network for fitting compensation, showing good stability and robustness advantages. Under real vehicle load conditions, compared with the traditional proportional-integral-derivative control (PID) controller, the BSMC controller combined with the RBF neural network has better stability and target tracking performance, with a maximum tracking error of 8.173 kPa, an average error of 0.17 kPa, and a root mean square error of 0.53 kPa. Summary of the invention

[0005] In view of this, an object of the present invention is to provide a BSMC control method for a fuel cell hydrogen circulation system based on a RBF neural network.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A BSMC control method for a fuel cell hydrogen circulation system based on a RBF neural network comprises the following steps:

[0008] a) Establish a fuel cell hydrogen supply cycle anode kinetic model based on a single ejector;

[0009] First, the dynamic response of hydrogen pressure in the control volume is established as:

[0010]

[0011] Where P represents the hydrogen pressure of the control volume, W in represents the mass flow rate into the volume, W out represents the mass flow rate of the outflow volume; R, T and V represent the gas constant, volume and temperature of the gas inside it respectively;

[0012] Secondly, the control flow of the proportional valve is linearly related to the control signal. By changing the flow area of ​​the gas, the mass flow out of the proportional valve is controlled, which is specifically expressed as:

[0013] W fcv =u v W v_max

[0014] where u v is the proportional valve opening control signal ranging from 0 to 1, W v_max is the maximum mass flow rate when the valve is fully open;

[0015] Then, the ejector model is established, and its secondary flow rate W s It is expressed as:

[0016]

[0017] R p and R s Respectively represent the flow radius of the primary flow and the secondary flow, V s is the secondary flow velocity, represents the average density of the secondary flow fluid, N V is an exponential function of the hydrogen velocity;

[0018] When the ejector works in critical mode, its primary flow rate W p It is expressed as:

[0019]

[0020] Where P em is the ejector manifold pressure, A ej represents the throat area of ​​the nozzle, η p is the isentropic coefficient of the primary flow, T ej represents the fluid temperature in the ejector, and γ is the specific heat ratio of the gas; when the ejector works in the subcritical mode, its primary flow rate is expressed as:

[0021]

[0022] η s is the isentropic coefficient of the secondary flow, P s is the secondary flow manifold pressure;

[0023] b) Verify the flow and pressure characteristics of the ejector model based on actual bench data;

[0024] c) A back-stepping sliding mode controller is proposed to stabilize the anode pressure of the fuel cell;

[0025] First, set the actual state of the system anode pressure to x1 and the ideal reference pressure state to x ref At this time, the tracking error e1 of the anode pressure is defined as:

[0026] e1=x ref -x1

[0027] The derivative of e1 with respect to time Given by:

[0028]

[0029] Introduce a steady-state auxiliary variable α and design α as:

[0030] α=-k1e1

[0031] Where k1 is a constant greater than zero, and the auxiliary tracking error variable e2 is defined as:

[0032]

[0033] Define the first Lyapunov function as:

[0034]

[0035] Define the second Lyapunov function as:

[0036]

[0037] Where s is the sliding surface; the sliding surface is defined as:

[0038] s=k2e1+e2

[0039] k2 is also a constant greater than 0, and its derivative is obtained:

[0040]

[0041] If the sliding surface appears In the case of , the expression of the control input u of the backstepping sliding mode controller is:

[0042]

[0043] d) The radial basis function (RBF) neural network is used for fitting compensation, and combined with the originally designed backstepping sliding mode controller, the control effect is verified under the current step condition and the actual vehicle load condition;

[0044] Further, in d), the RBF neural network is used to estimate the nonlinear disturbance caused by the purge action and to calibrate the hydrogen supply system of the fuel cell, and the method steps are specifically as follows:

[0045] d1) The relationship between the input layer, hidden layer and output layer is:

[0046] Input layer: x = [x1, x2, ... x n ] T Represents the input of the grid, where n represents the input dimension of the grid;

[0047] Hidden layer: The Gaussian basis function is selected as the membership function of the input layer, and the output of the jth hidden layer neuron is expressed as:

[0048]

[0049] Among them, c j is the coordinate of the center point of the Gaussian basis function of the jth hidden layer neuron, b j represents the width of the Gaussian function of the jth hidden layer neuron;

[0050] Output layer: The output of the RBF neural network is y = W T h=W1h1+W2h2+....+W m h m ;

[0051] d2) Using RBF neural network to approximate and compensate for the unknown disturbance f(x) caused by purge;

[0052] At this time, the dynamic equation of the fuel cell anode pressure is rewritten as:

[0053]

[0054] In the formula is the estimated weight of the neural network, and the theoretical output of the network is f(x) = W *T h(x)+ε, where W * is the ideal weight of the neural network, ε is the approximation error of the grid;

[0055] d3) Combining RBF neural network with backstepping sliding mode controller;

[0056] Combining the RBF neural network with the backstepping sliding mode controller yields a new control rate:

[0057]

[0058] The beneficial effects of the present invention are:

[0059] (1) Improved fuel cell stability. Through the innovative backstepping sliding mode controller (BSMC) combined with the radial basis function (RBF) neural network, precise control of the fuel cell hydrogen supply system is achieved, thereby ensuring the adequacy of fuel supply on the anode side and the stability of the hydrogen supply pressure. This precise management strategy helps to maintain the efficient and stable operation of the fuel cell under various working conditions and effectively extends the battery's service life.

[0060] (2) Enhanced robustness. The invention specifically considers the unknown disturbances that the fuel cell may encounter during the purge process, and uses the powerful nonlinear mapping capability of the RBF neural network to compensate for them, significantly improving the robustness of the system. This means that even in the face of complex and changeable actual operating environments, the system can maintain stable performance and reduce performance fluctuations caused by disturbances.

[0061] (3) The control system is optimized. The control strategy proposed in this invention patent can adaptively adjust the hydrogen supply strategy under different operating conditions to ensure the dynamic response and output performance of the fuel cell system, which not only improves the response speed of the fuel cell, but also improves the overall control accuracy.

[0062] (4) It can reduce energy consumption and maintenance costs. Compared with traditional hydrogen circulation pumps, the ejector design adopted by this technology does not require additional energy consumption, thereby reducing the total energy consumption of the system, helping to improve the energy efficiency of fuel cell vehicles and reduce operating costs. Due to the simple structure and low failure rate of the ejector, the application of this technology has greatly reduced the maintenance cost of the fuel cell system; in addition, due to the high efficiency of the system control strategy, it also reduces the additional maintenance costs caused by control system failures.

[0063] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0065] Figure 1 Performance comparison of three controllers: (a) is load current; (b) is control performance; (c) is startup performance; (d) is current drop; (e) is current increase; (f) is current increase and target pressure increase; (g) is current drop and target pressure decrease;

[0066] Figure 2 Performance comparison of three controllers: (a) is load current; (b) is purge signal; (c) is control performance; (d) is purge condition; (e) is current drop and purge condition; (f) is current drop and purge condition;

[0067] Figure 3 The control results of fuel cell hybrid vehicle (FCHV) under load current; (a) is load current; (b) is purge signal; (c) is control performance;

[0068] Figure 4 is the anode hydrogen excess coefficient under nonlinear battery stack control NNBSMC;

[0069] Figure 5 It is an evaluation index of control effect. DETAILED DESCRIPTION

[0070] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0071] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0072] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0073] Figure 1 Performance comparison of three controllers: (a) is load current; (b) is control performance; (c) is startup performance; (d) is current drop; (e) is current increase; (f) is current increase and target pressure increase; (g) is current drop and target pressure decrease;

[0074] Figure 2 Performance comparison of three controllers: (a) is load current; (b) is purge signal; (c) is control performance; (d) is purge condition; (e) is current drop and purge condition; (f) is current drop and purge condition;

[0075] Figure 3 The control results of fuel cell hybrid vehicle (FCHV) under load current; (a) is load current; (b) is purge signal; (c) is control performance;

[0076] Figure 4 is the anode hydrogen excess coefficient under nonlinear battery stack control NNBSMC;

[0077] Figure 5 It is an evaluation index of control effect.

[0078] Proton exchange membrane fuel cell systems usually operate in a recirculation mode. The mainstream hydrogen recirculation equipment on the market is mainly based on the design of ejectors and hydrogen circulation pumps. However, the use of hydrogen circulation pumps will lead to additional parasitic power consumption and reduce system efficiency. At the same time, its control and maintenance costs are high, and operating noise and vibration will also have an adverse effect on the fuel cell system. In contrast, ejectors have received significant attention due to their small size, zero energy consumption, reliable operation and low cost.

[0079] Therefore, this study first established a fuel cell hydrogen supply cycle anode kinetic model based on a single ejector, verified the flow and pressure characteristics of the ejector model based on actual bench data, and then proposed a back-stepping sliding mode controller to stabilize the fuel cell anode pressure. Considering that the necessary purge action during the operation of the fuel cell stack will have a significant impact on the dynamic characteristics of the anode pressure, the RBF neural network was further used for fitting compensation, and the control effect was verified under the current step condition and actual vehicle load condition in combination with the back-stepping sliding mode controller of the original design.

[0080] This study focuses on the disturbance correction capability based on radial basis function (RBF) neural network, because during the purge process of the fuel cell system, the flow characteristics of some components change, which will affect the state of the entire system and introduce unknown disturbances. The original backstepping sliding mode controller (BSMC) cannot effectively manage the system under these conditions. To solve this problem, the RBF neural network is used to estimate the nonlinear disturbance caused by the purge action and correct the hydrogen supply system of the fuel cell. Through this data-driven approach, the robustness of the system in the face of unknown disturbances can be improved, thereby improving the performance and stability of the fuel cell during the purge period.

[0081] First, this study compared the control effect without a purge signal. The system input a load current step condition that fluctuates within an appropriate current range, that is, from 250A to 320A, such as Figure 1 As shown in (a), the simulation results of three control algorithms of the fuel cell hydrogen supply system under ideal conditions are obtained. Figure 1 (b) shows the tracking curve of the anode pressure of the fuel cell system under the control of PID, SMC and BSMC controllers.

[0082] Compared with the current step condition, it can be seen that when the load current increases, the hydrogen consumption of the fuel cell increases, and the anode pressure will drop briefly due to the hysteresis of the system response; when the load current decreases, the hydrogen consumption of the fuel cell decreases, and the anode pressure will rise briefly; when the current is stable, all three controllers can stabilize the anode pressure well. Figure 1It can be observed in (c) that compared with the PID controller, the BSMC controller and the SMC controller both show faster response speeds and can achieve the control target in a short time without overshoot, and their performance is significantly better than that of the PID controller. Figure 1 Middle (d) and Figure 1 It can be seen in (e) that during the increase or decrease of the external load current, the BSMC controller has the best pressure fluctuation suppression effect, the system adjustment time is shorter, and the anode pressure can recover to the set target value faster.

[0083] There is also a certain relationship between the hydrogen supply pressure of the fuel cell anode and the load current. When the load current demand increases, the fuel cell needs more hydrogen flow to support the electrochemical reaction. Increasing the anode pressure can increase the diffusion rate of hydrogen and ensure that enough hydrogen can be quickly delivered to the electrode surface. At the same time, high load current will cause the polarization loss of the fuel cell electrode to increase. By increasing the pressure of the anode gas, the polarization loss of the electrode can be reduced and the rate of the electrochemical reaction can be increased. Conversely, when the load current demand decreases, the anode pressure can also be reduced accordingly to improve the utilization of hydrogen. Therefore, in order to better simulate the actual working state of the hydrogen supply to the anode of the fuel cell, at the 40th second of the input step current condition, the load current increases upward and the theoretical reference pressure also increases accordingly. At 60 seconds, the load current decreases downward, and the theoretical reference pressure also decreases accordingly. Therefore, it is obtained. Figure 1 Middle (f) and Figure 1 (g) The control effects of the three controllers.

[0084] Among them, the control effect of BSMC controller is obviously better than that of SMC and PID controller.

[0085] Subsequently, in order to further verify the control effect of the designed neural network-based backstepping sliding mode controller (NNBSMC), a periodic purge signal was added under the above step current condition. The purge action was performed every 15 seconds, and the purge time was 1 second.

[0086] Purge signal Figure 2 As shown in (b), the control effects of PID, BSMC and NNBSMC controllers are compared under working conditions.

[0087] Figure 2 (c) shows the control effect of PID, BSMC and NNBSMC controllers on the pressure of the fuel cell anode hydrogen supply system under load current fluctuations and periodic purge disturbances.

[0088] from Figure 2As can be seen in (d), the NNBSMC controller is able to compensate the pressure fluctuations generated during the purge process very well, while the BSMC controller without compensation will produce a large pressure difference during the purge period, which is not conducive to the stable operation of the fuel cell.

[0089] Figure 2 (e) and Figure 2 (f) in the middle shows the control effects of the three controllers on the anode pressure when there are upward step current purge and downward step current purge, that is, when there are two external disturbances at the same time.

[0090] Due to the coupling relationship between load current and hydrogen flow rate, it is found that the step of load current will have a certain compensation or amplification effect on the fluctuation of anode pressure during the purge process, but overall, the control performance of the NNBSMC controller is significantly better than that of the other two controllers.

[0091] In order to more realistically reflect the output characteristics of the fuel cell system during actual vehicle driving, this study used the collected load current condition data of the fuel cell hybrid electric vehicle during actual operation to verify the control effect of the controller. These condition data are the total current of the fuel cell and the power battery, and the actual current change rate of the vehicle fuel cell system will be much lower than the DC bus current. Under this working condition, the current is scaled and limited to more than 120A to keep the current within the normal operating range of the fuel cell system model established in this paper. At the same time, the ejector operates in the recirculation mode, resulting in insufficient hydrogen supply to the reactor, limiting its output performance.

[0092] Figure 3 (a) shows the actual vehicle DC bus current conditions. Under this working condition, the control results of the system anode pressure by PID, SMC and NNBSMC controllers are shown in Figure 2. Figure 3 As shown in (c), from the overall trend of the figure, it can be seen that under actual vehicle conditions, large current fluctuations and purge interference, the PID controller failed to achieve good results in controlling the anode hydrogen pressure, and frequent pressure fluctuations are not conducive to the stability of the fuel cell stack output performance. At the same time, the excessive pressure difference between the two electrodes will cause irreversible damage to the anode exchange membrane and the reactor.

[0093] Figure 3 Middle (d) and Figure 3 (e) is Figure 3 From the enlarged view of (c), we can see more clearly that the control effects of SMC and NNBSMC controllers are significantly better than those of PID controller. In addition, under external disturbance, NNBSMC controller shows better performance and can effectively maintain the stability of hydrogen supply pressure of fuel cell, with the best robustness.

[0094] The hydrogen excess coefficient in the anode system under the NNBSMC controller is as follows: Figure 4 It can be seen that during the entire operation process, the hydrogen excess coefficient is basically maintained between 2 and 4, which can fully ensure that the anode system has sufficient hydrogen flow supply to meet the normal operation requirements of the fuel cell.

[0095] By comprehensively evaluating the control effects of the three controllers under actual vehicle conditions, the dynamic performance indicators of the fuel cell anode hydrogen supply system are demonstrated. Figure 5 As shown in the figure, under actual vehicle conditions, the three performance indicators of the NNBSMC controller are significantly better than those of the other two controllers, showing excellent control performance and robustness, which is of great significance for the pressure control of full-power fuel cell vehicles in actual application scenarios.

[0096] Finally, the conclusion is:

[0097] (1) In the scenario where purge operation is not performed, the backstepping sliding mode control (BSMC) scheme shows better control effect, better performance and robustness compared with the PID controller and sliding mode controller commonly used in industry.

[0098] (2) After adding the periodic purge action, the neural network-based backstepping sliding mode controller (NNBSMC) can accurately compensate for the pressure fluctuations caused by the purge process. The BSMC controller without compensation will cause a large pressure difference at the anode of the fuel cell stack when the purge valve is working, which is not conducive to the stable operation of the fuel cell.

[0099] (3) Under actual vehicle load conditions, the BSMC controller combined with radial basis function (RBF) neural network has better stability and target tracking performance than the traditional PID controller. The maximum tracking error is 8.173 kPa, the average error is 0.17 kPa, and the root mean square error is 0.53 kPa.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. BSMC control method for fuel cell hydrogen circulation system based on RBF neural network, characterized by: The method comprises the following steps: a) Establish a fuel cell hydrogen supply cycle anode kinetic model based on a single ejector; First, the dynamic response of hydrogen pressure in the control volume is established as: Where P represents the hydrogen pressure of the control volume, W in represents the mass flow rate into the volume, W out represents the mass flow rate of the outflow volume; R, T and V represent the gas constant, volume and temperature of the gas inside it respectively; Secondly, the control flow of the proportional valve is linearly related to the control signal. By changing the flow area of ​​the gas, the mass flow out of the proportional valve is controlled, which is specifically expressed as: W fcv =u v W v_max where u v is the proportional valve opening control signal ranging from 0 to 1, W v_max is the maximum mass flow rate when the valve is fully open; Then, the ejector model is established, and its secondary flow rate W s It is expressed as: R p and R s Respectively represent the flow radius of the primary flow and the secondary flow, V s is the secondary flow velocity, represents the average density of the secondary flow fluid, N V is an exponential function of the hydrogen velocity; When the ejector works in critical mode, its primary flow rate W p It is expressed as: Where P em is the ejector manifold pressure, A ej represents the throat area of ​​the nozzle, η p is the isentropic coefficient of the primary flow, T ej represents the fluid temperature in the ejector, and γ is the specific heat ratio of the gas; when the ejector works in the subcritical mode, its primary flow rate is expressed as: η s is the isentropic coefficient of the secondary flow, P s is the secondary flow manifold pressure; b) Verify the flow and pressure characteristics of the ejector model based on actual bench data; c) A back-stepping sliding mode controller is proposed to stabilize the anode pressure of the fuel cell; First, set the actual state of the system anode pressure to x1 and the ideal reference pressure state to x ref At this time, the tracking error e1 of the anode pressure is defined as: e1=x ref -x1 The derivative of e1 with respect to time Given by: Introduce a steady-state auxiliary variable α and design α as: α=-k1e1 Where k1 is a constant greater than zero, and the auxiliary tracking error variable e2 is defined as: Define the first Lyapunov function as: Define the second Lyapunov function as: Where s is the sliding surface; the sliding surface is defined as: s=k2e1+e2 k2 is also a constant greater than 0, and its derivative is obtained: If the sliding surface appears In the case of , the expression of the control input u of the backstepping sliding mode controller is: d) A radial basis function (RBF) neural network is used for fitting compensation, and combined with the originally designed backstepping sliding mode controller, the control effect is verified under the current step condition and actual vehicle load condition.

2. The BSMC control method for a fuel cell hydrogen circulation system based on a RBF neural network according to claim 1 is characterized in that: In the above d), the RBF neural network is used to estimate the nonlinear disturbance caused by the purge action and to calibrate the hydrogen supply system of the fuel cell. The specific steps of the method are as follows: d1) The relationship between the input layer, hidden layer and output layer is: Input layer: x = [x1, x2, ... x n ] T Represents the input of the grid, where n represents the input dimension of the grid; Hidden layer: The Gaussian basis function is selected as the membership function of the input layer, and the output of the jth hidden layer neuron is expressed as: Among them, c j is the coordinate of the center point of the Gaussian basis function of the jth hidden layer neuron, b j Represents the width of the Gaussian function of the jth hidden layer neuron; Output layer: The output of the RBF neural network is y = W T h=W1h1+W2h2+....+W m h m ; d2) Using RBF neural network to approximate and compensate for the unknown disturbance f(x) caused by purge; At this time, the dynamic equation of the fuel cell anode pressure is rewritten as: In the formula is the estimated weight of the neural network, and the theoretical output of the network is f(x) = W *T h(x)+ε, where W * is the ideal weight of the neural network, ε is the approximation error of the grid; d3) Combining RBF neural network with backstepping sliding mode controller; Combining the RBF neural network with the backstepping sliding mode controller yields a new control rate:

Images