An online adaptive anti-surge control method for fuel cell air path

Abstract

The present invention relates to an online adaptive anti-surge control method for fuel cell air circuit, the method comprising: S1, designing a dual PI control loop with feedforward correction for fuel cell air circuit flow-pressure; S2, designing fuzzy adaptive logic to adaptively correct the dual PI control; S3, extracting flow change characteristics by a normalized method; S4, online detection of air circuit surge phenomenon based on a time window overlap determination method; S5, quantifying the severity of surge based on a characteristic energy method within a time window, feeding back surge information and characteristic energy information for online adjustment of air compressor speed and back pressure valve opening, and implanting a dual feedforward fuzzy PI overoxygen ratio and pressure control framework to perform online adaptive anti-surge control of fuel cell supply. Compared with the prior art, the present invention is conducive to improving the safety and reliability of fuel cell air supply and extending the life of the fuel cell system.

Description

Technical Field

[0001] The invention relates to the technical field of fuel cells, and in particular to an online adaptive anti-surge control method for an air circuit of a fuel cell. Background Art

[0002] Fuel cells have attracted much attention due to their zero pollution, low noise, high energy conversion rate and other advantages. Among them, proton exchange membrane fuel has the advantages of light weight, high power density and good stability, which is very suitable for vehicle engines and is the development trend of future automotive fuel cells. However, the shortcomings of fuel cells in terms of life, efficiency, cost and reliability have restricted the rapid development of the industry to a certain extent.

[0003] Among them, fast, accurate, and stable effective control of the air supply system is the key to ensure the efficiency and reliability of the fuel cell system. The centrifugal compressor currently used in the fuel cell air subsystem has the characteristics of compactness and high efficiency. However, due to its high speed characteristics, it is prone to surge under certain working conditions, causing irreversible damage to the engine and the stack. And because the operating point of the current fuel cell system compressor is often close to the surge line, the probability of surge occurring during the operation of the fuel cell engine is greatly increased, which deteriorates the stable control of the fuel cell control supply and will seriously reduce the life cycle of the system. Therefore, it is necessary to consider preventing the occurrence of surge while quickly and accurately controlling the fuel cell control circuit supply, so as to further stabilize the efficient response of the fuel cell engine and improve the system life.

[0004] At present, most of the research focuses on the relevant control methods of the air excess coefficient and pressure of the fuel cell system. There is still little research on the online monitoring of system surge and surge protection strategies during the control process. The possible surge phenomenon during the control process is ignored, making the system unable to identify the current surge fault in the air path, and it is even more difficult to implement surge protection.

[0005] Therefore, in order to ensure the accuracy and stability of the fuel cell air supply and extend the life of the fuel cell system, it is necessary to design a control technology that can provide the system with real-time rapid online monitoring of surge and timely surge protection. Summary of the invention

[0006] The purpose of the present invention is to provide an online adaptive anti-surge control method for a fuel cell air circuit in order to overcome the defects of the above-mentioned prior art.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A fuel cell air path online adaptive anti-surge control method, the method comprising:

[0009] S1. Design a dual PI control loop for fuel cell air flow-pressure with feedforward correction;

[0010] S2. Design fuzzy adaptive logic to adapt and correct the dual PI control;

[0011] S3, using normalization method to extract traffic change characteristics;

[0012] S4, online detection of air path surge phenomenon based on a time window overlap determination method;

[0013] S5. The severity of surge is quantified based on the characteristic energy method within the time window, and the surge information and characteristic energy information are fed back for online adjustment of the air compressor speed and back pressure valve opening. A double feedforward fuzzy PI excess oxygen ratio and pressure control framework is implanted to perform online adaptive anti-surge control of the fuel cell supply.

[0014] Preferably, in step S1, the flow rate and pressure are controlled respectively by using PI controllers, and the feedforward compensation value is determined according to the excess oxygen ratio and pressure at the working point under different load current conditions in actual conditions.

[0015] Preferably, step S2 specifically includes:

[0016] S21. Design PI controllers for controlling the compressor and the back pressure valve respectively, adjust the parameters, and determine the appropriate range of the parameters;

[0017] S22, respectively designing fuzzy adaptive logic for online adjustment of PI controller parameters of the compressor and the back pressure valve, and determining the input and output range of the fuzzy logic parameters;

[0018] S23, fuzzy logic outputs the corresponding PI controller Kp, Ki parameter values ​​according to the flow-pressure control error and error change rate at this time, to achieve online adaptation of parameters.

[0019] Preferably, in step S2, the compressor speed and the back pressure valve opening are calculated by the following formula:

[0020]

[0021] Where k is the control time, k ≥ 0, w(k) is the compressor speed at time k, α(k) is the back pressure valve opening at time k, e OER and e P are the excess oxygen ratio and pressure error, Determine the compressor speed PI control parameter K i，ω , K P，ω Adaptive fuzzy logic, To determine the back pressure valve opening control PI parameter K i，α , K p，α Adaptive fuzzy logic, FFω (I st (k)) is the load current feed-forward correction law for the compressor.

[0022] Preferably, in step S3, a sigmoid function is used to extract the normalized features of the flow rate change, and the flow rate feature change at time k is for:

[0023]

[0024] Among them, CR W is the rate of change of flow rate;

[0025]

[0026] Among them, CR W (k+1) is the rate of change of flow at time k+1, W c (k) is the air mass flow rate at time k, W c (k+1) is the air mass flow rate at time k+1, and T is the time interval between adjacent moments.

[0027] Preferably, in step S4, a time window is designed to be triggered according to the flow time domain change characteristics. When the flow change characteristics exceed the upper and lower limits of surge determination, the upper and lower time windows will be triggered respectively. When the above two time windows overlap, surge is determined to have occurred.

[0028] Preferably, step S4 is specifically:

[0029] S41, based on the time domain moving window monitoring and judgment method to capture the abnormal fluctuation characteristics of traffic;

[0030] S42, confirming system surge according to the overlap of the upper and lower time windows;

[0031] S43: Output the current status information of the system and issue a surge alarm at the initial stage of surge.

[0032] Preferably, step S41 is specifically:

[0033] Design a time domain moving window monitoring and judgment method to perform surge judgment. For the control step length k:

[0034]

[0035]

[0036] Among them, F Wc (k) is the flow characteristic change at time k, F W，hb For the upper boundary of surge determination, F W，lb To determine the lower boundary of surge, TR h(k) is the upper boundary trigger mark at time k, TR l (k) is the lower boundary trigger mark at time k;

[0037] To indicate that in T s，i Abnormal fluctuations have occurred during the window time. The upper and lower limit breakthrough judgment results are respectively brought into the following moving time windows:

[0038]

[0039] CT(k+1) represents the time from time k+1 to window closing;

[0040] Step S42 is specifically as follows:

[0041] The occurrence of surge is confirmed by the overlap of the upper and lower time windows, as described below:

[0042]

[0043] Among them, Sig surge (k) is the surge trigger signal, ξ(x) is the judgment function, which is expressed as follows:

[0044]

[0045] Step S43 is specifically as follows:

[0046] After the system confirms surge, in the next control time T s，c Within this time, the system will issue a surge alarm signal, which is called the surge alarm time window and is expressed as:

[0047]

[0048] Among them, T s，c is the surge alarm window time constant,

[0049] After the above alarm timing, the current surge state pointer SS of the system is determined by the following formula:

[0050] SS(k)=ξ(SW(k))

[0051] For time k, when SS(k)=0, it indicates that the system is in a non-surge state. On the contrary, when SS(k)=1, it indicates that the current system is in a surge alarm state.

[0052] Preferably, step S5 is specifically:

[0053] S51, based on the surge alarm, quantify the surge intensity using the characteristic energy method;

[0054] S52, feeding back the surge energy to the compressor and back pressure valve control to suppress the surge in time;

[0055] S53. At the same time, the surge signal is used to correct the pressure reference value, so that the system operating point is quickly moved away from the surge area at the initial stage of surge.

[0056] Preferably, step S51 is specifically:

[0057] The method of accumulating flow characteristic changes is used to quantify the severity of surge, and is defined as the characteristic energy index of surge flow fluctuation. For the control time k, the surge energy iteration is as follows:

[0058] CE surge (k+1)=(CE surge (k)+F Wc (k+1) 2 ) · SS(k), CE surge (0) = 0

[0059] F Wc (k+1) is the traffic characteristic change at time k+1, CE surge (k+1), CE surge (k) are the surge energies at time k+1 and time k respectively, SS(k) is the current surge state pointer of the system at time k, when SS(k)=0, it means the system is in a non-surge state, when SS(k)=1, it means the current system is in a surge alarm state;

[0060] Step S52 is specifically as follows:

[0061] The following adaptive control law for the compressor and back pressure valve in the case of surge burst is added to the excess oxygen ratio pressure control equation:

[0062]

[0063] in,

[0064]

[0065]

[0066] T s，e K is the time constant for removing the lag in real-time intervention control of surge after the surge alarm is eliminated. w，f and K α，f is the surge energy feedback coefficient. After adding surge adaptive control, the final compressor and back pressure valve control equation is determined by the following formula:

[0067]

[0068] ω fix (k) is the compressor speed control value during surge adaptive control at time k, α fix(k) is the back pressure valve opening control value during surge adaptive control at time k, ω nol (k) is the compressor speed control value during normal control at time k, α fix (k) is the back pressure valve opening control value under normal control at time k, ω out (k) is the final control value of the compressor speed at time k, α fix (k) is the final control value of the back pressure valve opening at time k;

[0069] Step S53 is specifically as follows:

[0070] The following adaptive corrections to the pressure reference for surge conditions are designed:

[0071] P sm，ref (k) = OP pre (f(k))+K P，ref SW(k)

[0072] SW(k) is the surge alarm time window, OP pre (I(k)) is the pressure reference before correction, P sm,ref (k) is the corrected pressure reference, K P,ref is the proportionality factor for pressure reference correction.

[0073] Compared with the prior art, the present invention has the following advantages:

[0074] (1) The present invention mainly proposes an online adaptive anti-surge control method for fuel cell air supply. This method can realize real-time online detection of system surge while accurately controlling the oxygen ratio and pressure of the air path, thereby improving the working efficiency and system stability of the fuel cell, preventing further occurrence of surge, and extending the life of the fuel cell.

[0075] (2) The time window method adopted by the present invention can quickly monitor surge at the initial stage of surge (within the 0.2s window time) and feed back the signal to the system in real time, thus realizing online rapid detection of surge.

[0076] (3) The surge characteristic energy method adopted in the present invention can quantify the surge intensity in real time after the surge occurs and use it for feedback correction, thereby enhancing the adaptability of surge control and better suppressing surge according to actual conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 The present invention is a flow chart of the method.

[0078] Figure 2 It is the control principle diagram of the present invention.

[0079] Figure 3It is a schematic diagram of time window surge detection and characteristic energy used for adaptive regulation in an example of the present invention. DETAILED DESCRIPTION

[0080] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the following embodiments are merely illustrative in nature, and the present invention is not intended to limit its applicable objects or uses, and the present invention is not limited to the following embodiments.

[0081] Example

[0082] like Figure 1 As shown, the present invention relates to an online adaptive anti-surge control method for a fuel cell air path, comprising the following steps:

[0083] S1. Design a dual PI control loop for fuel cell air flow-pressure with feedforward correction;

[0084] S2. Design fuzzy adaptive logic to adapt and correct the dual PI control;

[0085] S3, using normalization method to extract flow change features;

[0086] S4, online detection of air path surge phenomenon based on a time window overlap determination method;

[0087] S5. The severity of surge is quantified based on the characteristic energy method within the time window, and the surge information and characteristic energy information are fed back for online adjustment of the air compressor speed and back pressure valve opening. A double feedforward fuzzy PI excess oxygen ratio and pressure control framework is implanted to achieve online adaptive anti-surge control of the fuel cell supply.

[0088] In step S1, the flow rate and pressure are controlled by PI controllers respectively, and the feedforward compensation value is determined according to the working point (current oxygen ratio and pressure) under different load current conditions in actual conditions. The working point selection is based on the calibration results under experimental tests, and is based on the highest stack voltage output measured under steady-state conditions.

[0089] In step S2, the corresponding fuzzy logic is designed to adaptively adjust the parameters Ki and Kp in the compressor-back pressure valve dual PI controller to enhance the adaptability of control tracking to variable working conditions. Step S2 is specifically as follows:

[0090] S21. Design PI controllers for controlling the compressor and the back pressure valve respectively, adjust the parameters, and determine the appropriate range of the parameters;

[0091] S22, respectively designing fuzzy adaptive logic for online adjustment of PI controller parameters of the compressor and the back pressure valve, and determining the input and output range of the fuzzy logic parameters;

[0092] S23, fuzzy logic outputs the corresponding PI controller Kp, Ki parameter values ​​according to the flow-pressure control error and error change rate at this time, to achieve online adaptation of parameters.

[0093] In step S4, a time window is designed to be triggered according to the flow time domain change characteristics. When the flow change characteristics exceed the upper and lower limits of surge determination, the upper and lower time windows will be triggered respectively. When the above two time windows overlap, surge is determined to have occurred. Step S4 is specifically as follows:

[0094] S41, based on the time domain moving window monitoring and judgment method to capture the abnormal fluctuation characteristics of traffic;

[0095] S42, confirming system surge according to the overlap of the upper and lower time windows;

[0096] S43: Output the current status information of the system and issue a surge alarm at the initial stage of surge.

[0097] In the step S5, characteristic energy of surge is calculated according to flow characteristics when surge is triggered, and the characteristic energy is used as feedback information in flow-pressure surge control.

[0098] Step S5 is specifically as follows:

[0099] S51, based on the surge alarm, quantify the surge intensity using the characteristic energy method;

[0100] S52, feeding back the surge energy to the compressor and back pressure valve control to suppress the surge in time;

[0101] S53. At the same time, the surge signal is used to correct the pressure reference value, so that the system operating point is quickly moved away from the surge area at the initial stage of surge.

[0102] The overall control framework of the present invention is as follows Figure 2 As shown, the control strategy is designed according to the following ideas:

[0103] First, a dual PI control loop with feedforward correction for fuel cell air flow-pressure is designed, and then fuzzy logic is used to adjust and control the PI parameters in real time to ensure the tracking ability of the fuel cell excess oxygen ratio and pressure reference. Without surge adaptive correction, the excess oxygen ratio and pressure control problem is described in discrete form as follows:

[0104]

[0105] Among them, k is the current control time, Determine the compressor speed PI control parameter K i，ω , K P，ω Adaptive fuzzy logic, To determine the back pressure valve opening control PI parameter Ki，α , K p，α Adaptive fuzzy logic. ω (I st (k)) is the load current feed-forward correction law for the compressor, which is mainly obtained through experiments to speed up the system response. OER and e P are the excess oxygen ratio and pressure error, respectively, and are defined as:

[0106]

[0107]

[0108] OP OER (I(k)) and OP pre (I(k)) is the appropriate operating condition of the fuel cell system obtained in a steady state, and its output is determined by the current load current value.

[0109] In order to realize real-time detection of surge in air supply control, a detection method based on time window overlap is adopted, and the flow fluctuation during the surge alarm period is quantified with characteristic energy and used for system adaptive correction. The principle is as follows: Figure 3 The specific design is as follows:

[0110] The sigmoid function is used to normalize the flow change between [-0.5, 0.5] to achieve normalized feature extraction under different flow conditions, and square enhancement is used to extract features. The flow fluctuation feature is calculated as follows:

[0111]

[0112] In the above formula, CR W is the rate of change of flow, that is:

[0113]

[0114] When the system is surging, the flow characteristics will be significantly different from the usual ones, which is manifested as the flow fluctuations, and the change rate hovers between positive and negative. This is significantly different from the flow changes after normal loading and unloading. Therefore, a time domain moving window monitoring and judgment method is designed to judge surge. For the control step length k:

[0115]

[0116]

[0117] In the above formula, F W，hb For the upper boundary of surge determination, F W，lb Determine the lower boundary of surge.

[0118] In order to express s，i Abnormal fluctuations have occurred during the window time. The upper and lower limit breakthrough judgment results are respectively brought into the following moving time windows:

[0119]

[0120] In the above formula, CT(k+1) represents the time from time k+1 to the closing of the window. When CT=0, it means the time window is closed. The window time T in this study is s，i Set to 0.2s to cover the complete cycle of flow fluctuations during surge. Then the occurrence of surge is confirmed by the overlap of the upper and lower time windows, as described below:

[0121]

[0122] The judgment function is defined as follows:

[0123]

[0124] After the above surge judgment, in the next control time T s，c Within this time, the system will issue a surge alarm signal, which is called the surge alarm time window and is expressed as:

[0125]

[0126] Among them, T s，c is the surge alarm window time constant, which indicates the alarm time interval from the last surge alarm triggering to the system canceling the surge. According to actual experience, in order to improve the redundancy of the system safety, it is set to 2s here. After the above alarm timing, the current surge state pointer SS of the system is determined by the following formula:

[0127] SS(k)=ξ(SW(k))

[0128] For time k, when SS(k) = 0, it means that the system is in a non-surge alarm state. On the contrary, when SS(k) = 1, it means that the current system is in a surge alarm state. Through the above surge judgment process, the occurrence of system surge can be quickly identified and a surge alarm is given.

[0129] The characteristic energy method is used to quantify the severity of surge and as a quantitative index, for the control time k, the surge energy iteration is as follows:

[0130] CE surge (k+1)=(CE surge (k)+F Wc (k+1) 2 ) · SS(k), CE surge (0) = 0

[0131] The greater the characteristic energy accumulated during the surge alarm period, the greater the current surge degree. The characteristic energy feedback is used to timely correct the control output. The correction amount is adaptive according to the energy size, which can reduce the impact of surge while ensuring the robustness of the system in the face of surge. Based on this, on the basis of the excess oxygen ratio pressure control equation, the following adaptive control law for the compressor and back pressure valve in the case of surge burst is added:

[0132] in,

[0133]

[0134]

[0135] T s，e After the surge alarm is eliminated, intervention control removes the lag time constant. K w，f and K α，f is the surge energy feedback coefficient. After the control intervention, the surge alarm disappears after the system surges to dangerous conditions. At this time, the Gaussian attenuation form is used to remove the intervention control and increase a certain control redundancy to increase the robustness of the surge intervention control and prevent the surge from happening again. Therefore, after adding the surge adaptive control, the final compressor and back pressure valve control equations are determined by the following formula:

[0136]

[0137] At the same time, due to the fast response of the back pressure valve control, in order to guide the system out of the surge area, in addition to the output intervention of the inner loop control tracker, the outer loop reference value should also immediately reduce the reference pressure at the time of surge, and be able to gradually return to the steady-state operating conditions to maintain stability as the surge is eliminated. Based on this, the following adaptive correction of the pressure reference for surge conditions is designed:

[0138] P sm，ref (k) = OP pre (I(k))+K P，ref SW(k)

[0139] SW(k) is the surge alarm time window, OP pre (I(k)) is the pressure reference before correction, P sm,ref (k) is the corrected pressure reference, K P,ref is the proportionality factor for pressure reference correction.

[0140] Through the above design, the air path control system can achieve precise flow-pressure control under dynamic conditions, while enabling the system to have fault detection and self-emergency repair capabilities, avoiding surge conditions, effectively increasing the stability and safety of the fuel cell, and thereby achieving a longer system life.

[0141] The above embodiments are merely examples and do not limit the scope of the present invention. These embodiments can also be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the technical concept of the present invention.

Claims

1. A fuel cell air circuit online adaptive anti-surge control method, characterized in that: The method includes: S1. Design a dual PI control loop for fuel cell air flow-pressure with feedforward correction; S2. Design fuzzy adaptive logic to adapt and correct the dual PI control; S3, using normalization method to extract traffic change characteristics; S4, online detection of air path surge phenomenon based on a time window overlap determination method; S5. Quantify the severity of surge based on the characteristic energy method within the time window, use the surge information and characteristic energy information feedback for online adjustment of the air compressor speed and back pressure valve opening, and implant a double feedforward fuzzy PI excess oxygen ratio and pressure control framework to perform online adaptive anti-surge control of the fuel cell supply; In step S4, a time window is designed to be triggered according to the flow time domain variation characteristics. When the flow variation characteristics exceed the upper and lower limits of surge determination, the upper and lower time windows will be triggered respectively. When the above two time windows overlap, it is determined that surge occurs. Step S4 is specifically as follows: S41, based on the time domain moving window monitoring and judgment method to capture the abnormal fluctuation characteristics of traffic; S42, confirming system surge according to the overlap of the upper and lower time windows; S43, outputting the current status information of the system, and issuing a surge alarm at the initial stage of surge; Step S41 is specifically as follows: Design a time domain moving window monitoring and judgment method to perform surge judgment. For the control step length k: Among them, F Wc (k) is the flow characteristic change at time k, F W，hb For the upper boundary of surge determination, F W，lb To determine the lower boundary of surge, TR h (k) is the upper boundary trigger mark at time k, TR l (k) is the lower boundary trigger mark at time k; To indicate that in T s，i Abnormal fluctuations have occurred during the window time. The upper and lower limit breakthrough judgment results are respectively brought into the following moving time windows: CT(k+1) represents the time from time k+1 to window closing; Step S42 is specifically as follows: The occurrence of surge is confirmed by the overlap of the upper and lower time windows, as described below: Among them, Sig surge (k) is the surge trigger signal, ξ(x) is the judgment function, which is expressed as follows: Step S43 is specifically as follows: After the system confirms surge, in the next control time T s，c Within this time, the system will issue a surge alarm signal, which is called the surge alarm time window and is expressed as: Among them, T s，c is the surge alarm window time constant, After the above alarm timing, the current surge state pointer SS of the system is determined by the following formula: SS(k)=ξ(SW(k)) For time k, when SS(k) = 0, it means that the system is in a non-surge state. On the contrary, when SS(k) = 1, it means that the current system is in a surge alarm state. Step S5 is specifically as follows: S51, based on the surge alarm, quantify the surge intensity using the characteristic energy method; S52, feeding back the surge energy to the compressor and back pressure valve control to suppress the surge in time; S53, at the same time, the surge signal is used to correct the pressure reference value, so that the system operating point is quickly moved away from the surge area at the early stage of the surge; Step S51 is specifically as follows: The method of accumulating flow characteristic changes is used to quantify the severity of surge, and is defined as the characteristic energy index of surge flow fluctuation. For the control time k, the surge energy iteration is as follows: WHAT surge (k+1)=(CE surge (k)+F Wc (k+1) 2 )·SS(k),CE surge (0)=0 F Wc (k+1) is the traffic characteristic change at time k+1, CE surge (k+1), CE surge (k) are the surge energies at time k+1 and time k respectively, SS(k) is the current surge state pointer of the system at time k, when SS(k)=0, it means the system is in a non-surge state, when SS(k)=1, it means the current system is in a surge alarm state; Step S52 is specifically as follows: The following adaptive control law for the compressor and back pressure valve in the case of surge burst is added to the excess oxygen ratio pressure control equation: in, T s，e K is the time constant for removing the lag in real-time intervention control of surge after the surge alarm is eliminated. w，f and K α，f is the surge energy feedback coefficient. After adding surge adaptive control, the final compressor and back pressure valve control equation is determined by the following formula: ω fix (k) is the compressor speed control value during surge adaptive control at time k, α fix (k) is the back pressure valve opening control value during surge adaptive control at time k, ω nol (k) is the compressor speed control value during normal control at time k, α fix (k) is the back pressure valve opening control amount during normal control at time k, ω out (k) is the final control value of the compressor speed at time k, α fix (k) is the final control value of the back pressure valve opening at time k; Step S53 is specifically as follows: The following adaptive corrections to the pressure reference for surge conditions are designed: P sm，ref (k)=OP pre (I(k))+K P，ref ·SW(k) SW(k) is the surge alarm time window, OP pre (I(k)) is the pressure reference before correction, P sm,ref (k) is the corrected pressure reference, K P,ref is the proportionality factor for pressure reference correction.

2. A fuel cell air circuit online adaptive anti-surge control method according to claim 1, characterized in that: In step S1, the flow rate and pressure are controlled respectively by using PI controllers, and the feedforward compensation value is determined according to the excess oxygen ratio and pressure at the working point under different load current conditions in actual conditions.

3. The method for online adaptive anti-surge control of fuel cell air circuit according to claim 1, characterized in that: Step S2 is specifically as follows: S21. Design PI controllers for controlling the compressor and the back pressure valve respectively, adjust the parameters, and determine the appropriate range of the parameters; S22, respectively designing fuzzy adaptive logic for online adjustment of PI controller parameters of the compressor and the back pressure valve, and determining the input and output range of the fuzzy logic parameters; S23, fuzzy logic outputs the corresponding PI controller Kp, Ki parameter values ​​according to the flow-pressure control error and error change rate at this time, to achieve online adaptation of parameters.

4. The method for online adaptive anti-surge control of fuel cell air circuit according to claim 1, characterized in that: In step S2, the compressor speed and the back pressure valve opening are calculated by the following formula: Where k is the control time, k ≥ 0, w(k) is the compressor speed at time k, α(k) is the back pressure valve opening at time k, e OER and e P are the excess oxygen ratio and pressure error, Determine the compressor speed PI control parameter K i，ω , K P，ω Adaptive fuzzy logic, To determine the back pressure valve opening control PI parameter K i，α , K p，α Adaptive fuzzy logic, FF ω (I st (k)) is the load current feed-forward correction law for the compressor.

5. The method for online adaptive anti-surge control of fuel cell air circuit according to claim 1, characterized in that: In step S3, the sigmoid function is used to extract the normalized features of the flow change. The flow feature change at time k for: Among them, CR W is the rate of change of flow rate; Among them, CR W (k+1) is the rate of change of flow at time k+1, W c (k) is the air mass flow rate at time k, W c (k+1) is the air mass flow rate at time k+1, and T is the time interval between adjacent moments.

Images