Decoupling method for advance control over air system of fuel cell, and device and medium

Abstract

The present invention belongs to the technical field of fuel cells, and relates to a decoupling method for advance control over an air system of a fuel cell, and a device and a medium. The method comprises: predicting a short-term power of a fuel cell engine by means of a time series prediction algorithm; testing required rotational speeds of an air compressor and required opening degrees of a back pressure valve under different pressures and flow rates; by means of bench testing, acquiring change data of the pressure and flow rate of an air system under a certain range of back pressure valve opening degrees and air compressor rotational speeds; under a data-driven method, identifying a transfer function of the air system, and then designing a corresponding decoupling controller by using a diagonal matrix decoupling method; obtaining a required air compressor rotational speed and a required back pressure valve opening degree from a predicted value of the short-term power by means of a look-up table method; and analyzing a decoupling effect of the air system by means of relative control errors in the pressure and flow rate, and analyzing the fluctuation change condition of an actuator by means of a total variation value of the actuator. The present invention can further improve the stability and system service life of an air system of a fuel cell.

Description

Decoupling method, device and medium for advance control of fuel cell air system

[0001] The present application claims priority to the Chinese patent application No. 202410714764.0, filed on June 4, 2024, and entitled "Decoupling method for advance control of fuel cell air system", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application belongs to the technical field of fuel cells, and relates to a decoupling method for advance control of fuel cell air system, device and medium. BACKGROUND

[0003] The air system of a fuel cell can be regarded as its "lungs", in which the key control parameters include the inlet air flow and the inlet air pressure. Effective control of the inlet air flow helps to prevent oxygen deficiency, while stable inlet air pressure can alleviate the pressure fluctuation in the stack, thereby prolonging the service life of the proton exchange membrane. In some cases, it is difficult to meet the high-precision control requirements of the fuel cell for the cathode inlet air flow and pressure, as well as the strict requirements for environmental adaptability, and the advance control of the inlet air flow and pressure in the time scale.

[0004] SUMMARY

[0005] Therefore, the present application aims to provide a decoupling method for advance control of fuel cell air system, device and medium, which can realize the advance control of the inlet air flow and pressure, and further improve the stability of the fuel cell air system and the service life of the system.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a decoupling method for advance control of fuel cell air system, comprising the following steps:

[0008] S1: designing a time series prediction algorithm to realize short-time power prediction of a fuel cell engine; using sliding window analysis to perform feature engineering on the time series data of power demand, combining with the theory of phase space reconstruction to construct the input and output vector space of the recursive prediction model based on recursive least squares support vector machine, and adopting autocorrelation analysis to determine the embedding dimension of the phase space reconstruction, and proposing a short-term power prediction algorithm;

[0009] S2: obtaining the demand speed of the air compressor and the demand opening of the back pressure valve under different pressures and flow rates according to the test bench, to provide data support for the feedforward control;

[0010] S3: obtaining the change data of the air system pressure and flow under the opening of the back pressure valve and the speed of the air compressor within a certain range through bench test, to provide data for decoupling control;

[0011] S4: Using a data-driven approach, identify the transfer function of the air system, and then design the corresponding decoupling device by combining the diagonal matrix decoupling method.

[0012] S5: The predicted short-term power is used to obtain the required air compressor speed and back pressure valve opening by looking up a table;

[0013] S6: The decoupling effect of the air system can be analyzed by the relative control error of pressure and flow, and the fluctuation of the actuator can be analyzed by the total change value of the actuator.

[0014] In an exemplary embodiment, step S1 specifically includes the following steps:

[0015] S11: Let the time series of power demand of a fuel cell vehicle containing N data points be [p1, p2, p3, ..., p...]. N ] T Choose the embedding dimension m and the time delay τ to construct the input vector x. t =[p t-(m-1)τ ,…,p t-2τ ,p t-τ ,p t ]∈R m Embedding dimension refers to the number of historical time series data points input, and time delay refers to the time interval between inputting historical time series data points; the corresponding predicted output variable is y. t+h =[p t+h ]∈R 1 Where h is the prediction step size, this is called h-step ahead prediction. The recursive prediction model for the demand power time series is as follows:

[0016] Where f(·) is a generalized smooth nonlinear mapping;

[0017] S12: Based on the phase space reconstruction method, the input-output sample pairs for demand power time series prediction are established when the sliding window width L=9, the embedding dimension m=5, the time delay τ=2, and the sliding period T=4, as shown in the following formula:

[0018] In the formula, These are the input matrix and the desired output matrix after reconstructing the phase space, respectively. The demand power time series is represented by [p1, p2, p3, ..., p...]. N ] T This means that the input sample matrix after reconstructing the phase space is used as the input of the prediction model F(·) to estimate the predicted data values ​​at future times;

[0019] S13: The autocorrelation coefficient calculation formula is defined as:

[0020] In the formula, the lag number k = 1, 2, …; the autocorrelation coefficient is the correlation degree of the two sample point sequences y N ] T constructed according to the lag number k, y t = [y 1+k , y 2+k , y 3+k , …, y N ] T and y t-k = [y N-k , y T .

[0021] In an exemplary embodiment, an iterative learning framework is constructed, and the way of completing a large time span prediction at one time is replaced by several iterations of shorter time steps. Based on the theory of reconstructing embedding phase space, the training data sets of learning elements for predicting power and power change rate in the xth iteration are respectively constructed as follows: and as shown in the following formula:

[0022] P(t+β);P(t+2β);…;P(t+xβ);P′(t+β);P′(t+2β);…;P′(t+xβ)]=f(P(t),P′(t),P″(t))

[0023] In the formula, f(·) represents the abstract nonlinear mapping of the trained ILF iterative learning framework.

[0024] In an exemplary embodiment, step S3 specifically comprises the following steps:

[0025] S31: Building and debugging the test bench;

[0026] S32: Setting the amplitude range of the air compressor speed M sequence and the back pressure valve opening;

[0027] S33: Considering the identification algorithm, the sampling period of the control system, and the sensitivity of the sensor, selecting the sampling period of the M sequence, and setting the sampling time.

[0028] In an exemplary embodiment, the building and debugging of the test bench in step S31 specifically comprises:

[0029] (1) Determining the topological structure of the test bench, determining the element selection and pipeline design;

[0030] (2) According to the topological structure, mechanical and electrical connections are made and tested;

[0031] (3) Calibration of each sensor;

[0032] (4) Test the status of the air compressor and back pressure valve;

[0033] (5) According to the pressure drop of the stack, intercooler and humidifier, respectively, the opening of each ball valve is calibrated;

[0034] (6) Test the system bench under full range operating conditions.

[0035] In an exemplary embodiment, in step S33, the sampling time is set as:

[0036] T0 is the sampling period, T 95 is the response time of the controlled object to reach 95% of the target value, and the denominator is set to any constant within the range of 5-15 according to the actual situation.

[0037] In an exemplary embodiment, step S4 specifically includes the following steps:

[0038] S41: Within a certain range of compressor speed, select multiple balanced operating points for system identification;

[0039] S42: Describe the input-output relationship of the system as a first-order inertia link, and the model structure of the air system is represented by the following formula:

[0040] Where is the increment of air flow, Δp is the increment of pressure, Δn is the increment of compressor speed, and Δγ is the increment of back pressure valve opening;

[0041] S43: Through a data-driven method, identify the transfer function of the air system with double inputs and double outputs;

[0042] S44: By adding a matrix in the control system, construct a diagonal matrix decoupling controller, specifically including: matrix The product of the matrix The generalized object matrix formed by the multiplication becomes a diagonal matrix, thereby realizing system decoupling:

[0043] When the mathematical model matrix of the controlled object is non-singular, The decoupling compensation matrix is written as:

[0044] In an exemplary embodiment, step S6 specifically includes:

[0045] The square of the average absolute error of flow and pressure is used, and the total change of the actuators including the air compressor and the back pressure valve is analyzed, as shown in the following formula:

[0046] Wherein Pressure(t), Flow(t) represent the pressure and flow at a certain time, Pressure_MAE, Flow_MAE, Pressure_MAE represent the average value of pressure and flow. The better the decoupling effect of the decoupling strategy is, the smaller Flow_MAE and Pressure_MAE are, the smaller Tv is, and the better stability and anti-interference are.

[0047] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the decoupling method for the fuel cell air system advance control.

[0048] In a third aspect, the present application provides a non-volatile computer readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement the decoupling method for the fuel cell air system advance control.

[0049] The fuel cell air system advance decoupling method, device and medium provided by the present application have the beneficial effects that: the timing prediction algorithm is designed, the short-time power prediction of the fuel cell engine can be realized, the advance control of the pressure and flow can be realized by acquiring the air system flow pressure, air compressor speed and back pressure valve opening data, the coupling of the fuel cell air system can be relieved based on the diagonal matrix decoupling method, and the above method can further improve the stability of the fuel cell air system and the service life of the system.

[0050] Other advantages, objects, and features of the present application will be apparent to those skilled in the art from the following specification, and will be learned from the practice of the present application. The objects and other advantages of the present application can be realized and obtained by the following specification.

[0051] Drawings of the specification

[0052] In order to make the objects, technical solutions and advantages of the present application clearer, the preferred detailed description of the present application will be combined with the drawings to describe the present application, wherein:

[0053] Fig. 1 is a feedforward control chart indicating intention;

[0054] Fig. 2 is a schematic diagram of a short-term power prediction method;

[0055] Fig. 3 is a schematic diagram of the structure of the core part;

[0056] Figure 4 is a decoupling design schematic. DETAILED DESCRIPTION

[0057] The present application is described below by way of specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the disclosure. The present application can also be implemented or applied by different specific embodiments, and various modifications or changes can be made to the details in the specification based on different views and applications without departing from the spirit of the present application. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0058] The drawings are only used for exemplary illustration, and the representation is only a schematic diagram, not a physical diagram, and cannot be interpreted as a limitation on the present application; in order to better illustrate the embodiments of the present application, some components in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0059] The same or similar reference numerals in the drawings of the embodiments of the present application correspond to the same or similar components; in the description of the present application, it should be understood that if the terms "upper", "lower", "left", "right", "front", "back" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore the terms describing the positional relationship in the drawings are only used for exemplary illustration, and cannot be interpreted as a limitation on the present application, for those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0060] The present application provides a decoupling method for advance control of a fuel cell air system, comprising the following steps:

[0061] S1: The fuel cell air system mainly includes an air filter, an air compressor, a intercooler, a humidifier and a back pressure valve, and the above elements are connected in sequence by pipelines. The advance control is realized by designing a time series prediction algorithm to realize short-time power prediction of the fuel cell engine. First, the time series data of power demand is analyzed by the sliding window method, and the input and output vector space of the recursive prediction model based on the machine learning algorithm (recursive least squares support vector machine) is constructed combined with the phase space reconstruction theory, and finally the embedding dimension of the phase space reconstruction is determined by the autocorrelation analysis, and a short-term power prediction algorithm is proposed; comprising the following steps:

[0062] S11: Let the fuel cell vehicle demand power time series with N data points be [p1, p2, p3, …, pN] ∈ R N ] T , select the embedding dimension m and time delay τ, the input vector x t = [p t-(m-1)τ ,…,p t-2τ ,p t-τ ,p t ] ∈ R m can be constructed. The embedding dimension refers to the number of historical time series data points input, and the time delay refers to the time interval of input historical time series data points. The corresponding prediction output variable is y t+h = [p t+h ] ∈ R 1 , where h is the prediction step, which is usually referred to as h-step ahead prediction. The shown is the demand power time series recursive prediction model for h-step ahead prediction.

[0063] where f(·) is a generalized smooth nonlinear mapping.

[0064] S12: Based on the phase space reconstruction method, the demand power time series prediction input and output sample pairs can be established when the sliding window width L = 9, the embedding dimension m = 5, the time delay τ = 2 and the sliding period T = 4, as shown in the following formula.

[0065] In the formula, are the input matrix and expected output matrix after reconstruction of the phase space, respectively. The input sample matrix after reconstruction of the phase space is used as the input of the prediction model F(·) to estimate the predicted data value at the future time (i.e. the predicted value of power).

[0066] S13: In order to evaluate the effectiveness of the prediction method, autocorrelation coefficient calculation can be used. The autocorrelation coefficient can be understood as the correlation degree of the two columns of sample point sequences y N ] T and y t = [y 1+k ,y 2+k ,y 3+k ,…,y N ] T and y t-k = [y1,y2,y3,…,y N-k ] T constructed according to the lag number k. The autocorrelation coefficient calculation formula can be defined as:

[0067] In the formula, the lag number k is It is the average value of the time series.

[0068] Based on the autocorrelation analysis results, when it is necessary to predict the future power demand p h sampling points ahead of the current time t... t+h and power change rate p' t+h At this time, the phase space reconstruction embedding dimension *m* of the power demand and power change rate time series data should not exceed 30-h and 50-h, respectively. The smaller the time range Δt for one-step prediction, the smaller the error term μ. p The smaller (t, Δt) is, the less severe the resulting temporal mismatch. Based on this, this invention proposes an iterative learning framework that replaces the previous method of completing a large time span prediction in one go with several iterative predictions of shorter time steps. For example, the one-step prediction from point a to point f is broken down into multiple iterative predictions: first, the data at point b is predicted; then, the predicted data at point b is combined with the original input data at point a as a new input to predict the next data at point c, and so on. Based on the theory of reconstructed embedding phase space, training datasets for the learning elements used to predict power and the rate of power change in the x-th iteration can be constructed respectively. and As shown in the following formula.

[0069] [(t+β);P(t+2β);…;P(t+xβ);P′(t+β);P′(t+2β);…;P′(t+xβ)]=f(P(t),P′(t),P″(t))

[0070] In the formula, f(·) represents the abstract nonlinear mapping of the trained ILF iterative learning framework, P(t+xβ) represents the power predicted at step x, and P'(t+xβ) represents the slope of the power predicted at step x.

[0071] S2: Based on the test bench, the required speed of the air compressor and the required opening degree of the back pressure valve under different pressures and flow rates are obtained, providing data support for feedforward control.

[0072] S3: Through bench testing, obtain data on the changes in air system pressure and flow rate under the conditions of back pressure valve opening and air compressor speed within a certain range, providing data for decoupling control; Step S3 specifically includes:

[0073] S31: The specific process of test bench construction and commissioning is as follows: (1) Determine the topology of the test bench, determine the component selection and piping design; (2) Make mechanical and electrical connections according to the topology, and test the mechanical and electrical connections; (3) Calibrate each sensor; (4) Test the status of the air compressor and back pressure valve; (5) Calibrate the opening degree of each ball valve according to the pressure drop of the fuel cell stack, intercooler and humidifier; (6) Test the system bench under full range operating conditions.

[0074] S32: To ensure that the identification input has a certain signal-to-noise ratio, the amplitude should not be too small. In combination with the actual situation, the amplitude range of the air compressor speed M sequence is ±1000 RPM, and the back pressure valve opening is ±10%.

[0075] S33: For the selection of the M sequence sampling period, the identification algorithm, the sampling period of the control system, and the sensitivity of the sensor should be considered. And the sampling time should be consistent with the actual application of the control algorithm as much as possible. The following empirical formula is generally used in engineering:

[0076] T0 is the sampling period, T 95 is the response time of the control object to reach 95% of the target value. The sampling time of the existing air supply system controller is 0.01s, which satisfies the above formula. Therefore, in order to maintain consistency, the sampling time of the system identification test is determined to be 0.01s.

[0077] S4: In the data-driven method, the transfer function of the air system is identified, and the corresponding decoupler is designed in combination with the diagonal matrix decoupling method; step S4 specifically includes:

[0078] S41: In order to enhance the adaptability of the controller and obtain more accurate decoupling coefficients, 10 equilibrium operating points are selected for system identification in the range of 0 to 90000 RPM of the compressor speed.

[0079] S42: Considering the dynamic characteristics of the proton exchange membrane fuel cell air supply system and the minimum order selection principle, the input and output relationship of the system is described as a first-order inertia link. Therefore, the model structure of the air system is represented by the following formula:

[0080] Where is the increment of air flow, Δp is the increment of pressure, Δn is the increment of compressor speed, and Δγ is the increment of back pressure valve opening.

[0081] Alternatively, the model structure of the air system is represented as:

[0082] Where s is the identity of the traditional function, k ij (i=1,2,j=1,2) represents the gain, and T ij (i=1,2,j=1,2) represents the time constant.

[0083] S43: Through the data-driven method, the identification of the double-input double-output transfer function of the air system is carried out.

[0084] S44: By adding a matrix in the control system, a diagonal matrix decoupling controller (i.e. decoupler) is proposed. The matrix with the object characteristic matrix product The formed generalized object matrix becomes a diagonal matrix, thereby realizing system decoupling.

[0085] Assuming that the mathematical model matrix of the controlled object is non-singular, Thus, the decoupling compensation matrix is rewritten as:

[0086] where G ij (s) (i = 1, 2, j = 1, 2) represents the transfer function of the system, G pij (s) (i = 1, 2, j = 1, 2) represents the transfer function of the controller.

[0087] S5: The predicted value of short-term power is obtained by table lookup method, and the required air compressor speed and back pressure valve opening degree are obtained.

[0088] Specifically, the predicted value of short-term power is obtained by table lookup method, and the required air system flow and pressure are obtained. The flow and pressure are controlled by air compressor speed and back pressure valve opening degree PID, and the dynamic response of pressure and flow can be obtained. This method can improve the response rate of the air system and prevent the shortage of reaction gas.

[0089] S6: The decoupling effect of the air system can be analyzed by the relative control error of pressure and flow. The fluctuation change of the actuator can be analyzed by the total change value of the actuator (air compressor speed and back pressure valve opening degree).

[0090] In order to quantitatively analyze the simulation results, the square of the average absolute error of flow and pressure is used. At the same time, the total change (Tv) of the actuator including the air compressor and the back pressure valve is analyzed, as shown in the following formula.

[0091] Where Pressure(t), Flow(t) represent the pressure and flow at a certain time, Pressure_MAE, Flow_MAE represent the average value of pressure and flow, C T represents the numerical value of the actuator opening (air compressor speed and back pressure valve opening); the better the decoupling effect of the decoupling strategy, the smaller the Flow_MAE and Pressure_MAE, the smaller the Tv, and the better the stability and anti-interference.

[0092] Or,

[0093] Wherein Pressure(t), Flow(t) represent the pressure and flow at a certain time, Pressure_MAE, Flow_MAE, Pressure_MAE represent the variance of the control error of pressure and flow. The better the decoupling effect of the decoupling strategy, the smaller Flow_MAE and Pressure_MAE, the smaller Tv, the better stability and anti-interference.

[0094] In the present application, for the coupling of air system pressure and flow, according to the pressure-flow-speed table and the pressure-flow-opening table obtained based on the above-mentioned test, the required air compressor speed and back pressure valve opening under a specific demand power can be obtained, and the specific process can be seen from the feedforward control table, as shown in FIG. 1, the speed of the air compressor and the opening of the back pressure valve under different powers can be obtained, which is conducive to improving the dynamic response of the system, and 1-DT(u) in FIG. 1 represents a one-dimensional data table; through the time sequence prediction algorithm as shown in FIG. 2, a new iterative learning framework is developed, and a large time span regression prediction is divided into multiple small time span regression predictions to achieve the purpose of suppressing time phase mismatch and improving prediction accuracy, and short-time power prediction of the fuel cell engine can be realized, which is conducive to realizing the advance control of the air system and can effectively avoid the phenomenon of fuel shortage, and the structure of the core part in FIG. 2 is shown in FIG. 3; through the bench test data, the change data of the air system pressure and flow under the back pressure valve opening and the air compressor speed in a certain range are obtained, and then the double-input double-output transfer function of the air system is identified, and the designed decoupling controller is shown in FIG. 4, and the scheme is to add a decoupling matrix in the controller, and the closed-loop transfer function becomes a diagonal matrix, and the above method can realize the pressure and flow decoupling control of the air system.

[0095] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store power demand time series data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a decoupling method for advance control of a fuel cell air system.

[0096] In an example embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0097] In an example embodiment, a non-volatile computer readable storage medium is provided, storing a computer program, and the computer program implementing the steps in the above method embodiments when executed by a processor.

[0098] A person of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above method embodiments. Any reference to a memory, database or other medium used in the embodiments provided in the present application can include at least one of a non-volatile and volatile memory. The non-volatile memory can include a read-only memory (ROM), a magnetic tape, a floppy disk, a flash memory, an optical storage, a high-density embedded non-volatile memory, a resistive memory (ReRAM), a magnetoresistive random access memory (MRAM), a ferroelectric random access memory (FRAM), a phase change memory (PCM), a graphene memory, etc. The volatile memory can include a random access memory (RAM) or an external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as a static random access memory (SRAM) or a dynamic random access memory (DRAM), etc.

[0099] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, etc., without being limited thereto.

[0100] Finally, it is to be explained that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit the present application. Although the present application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or equivalently replaced without departing from the purpose and scope of the technical solutions, and all should be covered in the scope of the claims of the present application.

Claims

1. A decoupling method for advance control of a fuel cell air system, characterized by: The method comprises the following steps: S1: design a time series prediction algorithm to realize short-time power prediction of the fuel cell engine; S2: according to the test bench test, the demand speed of the air compressor and the demand opening of the back pressure valve under different pressures and flow rates are obtained to provide data support for the feedforward control; S3: through the bench test, the change data of the air system pressure and flow under the back pressure valve opening and the air compressor speed in a certain range are obtained to provide data for decoupling control; S4: under the data-driven method, the transfer function of the air system is identified, and then a corresponding decoupler is designed by combining the diagonal matrix decoupling method; S5: the predicted value of the short-term power is obtained by table lookup method, and the required air compressor speed and back pressure valve opening are obtained; S6: the decoupling effect of the air system is analyzed through the relative control error of pressure and flow, and the fluctuation change of the actuator is analyzed through the total change value of the actuator. Step S1 comprises the following steps:

2. The decoupling method of fuel cell air system advance control according to claim 1, characterized in that: Wherein, f(·) is a generalized smooth nonlinear mapping; S11: Set the fuel cell vehicle demand power time series with N data points as [p1, p2, p3, …, pN] ∈ R N ] T , select the embedding dimension m and time delay τ, construct the input vector x t = [p t-(m-1)τ ,…,p t-2τ ,p t-τ ,p t ] ∈ R m ; the embedding dimension refers to the number of input historical time series data points, and the time delay refers to the time interval of input historical time series data points; the corresponding prediction output variable is y t+h = [p t+h ] ∈ R 1 , where h is the prediction step, referred to as h-step forward prediction, and the demand power time series recursive prediction model for h-step forward prediction is as follows: In the formula, f(·) represents the abstract nonlinear mapping of the trained ILF iterative learning framework. S12: Based on the phase space reconstruction method, the demand power time series prediction input-output sample pairs are established when the sliding window width L=9, the embedding dimension m=5, the time delay τ=2, and the sliding period T=4, as shown in the following formula: In the formulae, are the reconstructed phase space input matrix and the desired output matrix, respectively, and the demand power time series is given by [p1, p2, p3, …, pT]T. N ] T The reconstructed phase space input matrix is used as the input of the prediction model F(·) to estimate the predicted data value at the future time. S13: The autocorrelation coefficient calculation formula is defined as: wherein the lag number k = 1, 2, …; the autocorrelation coefficient is the correlation degree of two sample point sequences y N ] T constructed according to the lag number k. t 1+k 2+k 3+k N T t-k N-k T ​​​​​​​​​ 3. The decoupling method of advance control of a fuel cell air system according to claim 2, characterized in that: An iterative learning framework is constructed, which replaces the way of completing the prediction of a large time span at one time with several iterations of shorter time steps, and based on the theory of reconstructed embedding phase space, training data sets of learning elements for predicting power and power change rate in the xth iteration are constructed respectively and As shown in the following formula: [P(t+β); P(t+2β); …; P(t+xβ); P'(t+β); P'(t+2β); …; P'(t+xβ)] = f(P(t), P'(t), P''(t)) Step S3 comprises the following steps:

4. The decoupling method of advance control of a fuel cell air system according to claim 1, characterized in that: S31: build and debug the test bench; S32: set the amplitude range of the air compressor speed M sequence and the back pressure valve opening; S33: considering the identification algorithm, the sampling period of the control system and the sensitivity of the sensor, the sampling period of the M sequence is selected, and the sampling time is set. The step S31 of building and debugging the test bench comprises:

5. The decoupling method of advance control of a fuel cell air system according to claim 4, characterized in that: (1) determine the topology structure of the test bench, determine the element selection and pipeline design; (2) according to the topology structure, the mechanical and electrical connection is carried out, and the mechanical and electrical connection is tested; (3) calibrate each sensor; (4) test the state of the air compressor and the back pressure valve; (5) according to the pressure drop of the stack, intercooler and humidifier, the opening of each ball valve is calibrated; (6) test the system bench under the full range operating conditions. Step S4 comprises the following steps:

6. The decoupling method of advance control of a fuel cell air system according to claim 4, characterized in that: In step S33, the sampling time is set as: T0 is a sampling period, T 95 is a response time for the control object to reach 95% of the target value 95, and the denominator is set to any constant in the range of 5 to 15 according to the actual situation.

7. The decoupling method of advance control of a fuel cell air system according to claim 1, characterized by: S41: select multiple balance working points for system identification within a certain range of compressor speed; The increment of air flow is Δp, the increment of pressure is Δn, the increment of compressor speed is Δγ, and the increment of back pressure valve opening is Δγ; S42: The input and output relationship of the system is described as a first-order inertia link, and the model structure of the air system is expressed by the following formula: wherein S43: through the data-driven method, the transfer function of the air system with double inputs and double outputs is identified; The formed generalized object matrix becomes a diagonal matrix, so that the generalized object matrix formed by the product of the matrix and the object characteristic matrix becomes S44: constructing a diagonal matrix decoupling controller by adding a matrix in the control system, including: a matrix Product with object feature matrix Step S6 comprises: Diagonal matrix, thus decoupling the system: when the mathematical model matrix of the controlled object is non-singular, The decoupling compensation matrix is written as:

8. The decoupling method of advance control of a fuel cell air system according to claim 1, characterized in that: Wherein, Pressure(t), Flow(t) represents the pressure and flow at a certain time, Pressure_MAE, Flow_MAE represents the average value of pressure and flow; the better the decoupling effect of the decoupling strategy is, the smaller Flow_MAE and Pressure_MAE are, the smaller Tv is, and the better the stability and anti-interference performance is. The square of the average absolute error of flow and pressure is used, while analyzing the total change of the actuators including the air compressor and back pressure valve, as shown in the following equation: ​ 9. A computer device comprising: A memory, a processor, and a computer program stored on the memory and loadable on the processor, characterized in that the processor executes the computer program to implement the decoupling method of the fuel cell air system advance control according to any one of claims 1-8.

10. A non-transitory computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the decoupling method of the fuel cell air system advance control according to any one of claims 1-8.

Images