Self-adaptive load intelligent energy supply platform of fuel cell extended range power system

Abstract

The invention relates to the technical field of energy efficiency optimization, in particular to a self-adaptive load intelligent energy supply platform of a fuel cell range extending power system. Multi-source load data is collected through the environment perception and multi-source load monitoring module, after preprocessing and feature extraction, the dynamic power optimization decision module constructs a real-time rolling optimization model, the multi-constraint nonlinear optimization problem is solved in real time based on an equivalent hydrogen consumption minimization strategy, the optimal fuel cell output power is calculated, and the fuel cell output power is optimized. Precise energy supply is achieved, and the energy utilization efficiency is improved; accurate power following under complex working conditions is realized through multivariable decoupling control, cooling fan and air inlet valve optimization calculation is performed based on the optimal fuel cell output power, fault detection decision, fault category determination and corresponding optimization operation execution are performed by using an SVM support vector machine, and stable operation of the system is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of energy efficiency optimization, and in particular to an adaptive load intelligent energy supply platform for a fuel cell range-extending power system. Background Art

[0002] A fuel cell is a power generation device that converts the chemical energy of a fuel and an oxidant directly into electricity through an electrochemical reaction. It is considered the fourth generation technology after hydropower, thermal power, and nuclear power. Its core advantages lie in its high energy conversion efficiency, environmental friendliness, and widespread availability.

[0003] The fuel cell range extender system combines zero emissions with long driving range. Compared with traditional fuel range extenders, it has less energy loss and higher energy utilization rate.

[0004] However, existing energy efficiency optimization systems for fuel cells have many shortcomings.

[0005] First, the pure fuel cell system needs to bear the entire load power, resulting in a dynamic response lag, a sudden drop in output voltage, and energy loss.

[0006] Fuel cell systems rely on auxiliary equipment such as air compressors, hydrogen circulation pumps, and humidifiers to maintain operation, with their energy consumption accounting for 15% to 20% of the system's total output. Existing systems often use open-loop control, which is unable to dynamically adjust the auxiliary equipment speed based on operating conditions.

[0007] In response to the above problems, it is necessary to propose an adaptive load intelligent energy supply platform for fuel cell range-extended power systems. Summary of the Invention

[0008] The purpose of the present invention is to solve the problems existing in the background technology and to propose an adaptive load intelligent energy supply platform for a fuel cell range extender power system.

[0009] The purpose of the present invention can be achieved through the following technical solutions:

[0010] An adaptive load intelligent energy supply platform for a fuel cell range-extended power system includes an environmental perception and multi-source load monitoring module, a dynamic power optimization decision module, a multi-physics field collaborative control module, an actuator drive and energy conversion module, and a life prediction module.

[0011] The environmental perception and multi-source load monitoring module accesses a distributed sensor network to collect multi-source load data from each fuel cell, building the foundational data layer for the digital twin. This multi-source load data is preprocessed using a drift compensation algorithm to remove noise and environmental errors. Feature extraction is performed on the preprocessed signals to monitor for sudden load changes and voltage ripple.

[0012] The collected multi-source load data includes:

[0013] The load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance and mechanical vibration frequency of each single cell in the fuel cell.

[0014] As a preferred embodiment of the present invention, the fuel cell stack temperature is drift compensated based on the ambient temperature data, and the thermocouple is nonlinearly corrected. The drift compensation formula is:

[0015] T cor =T raw +α×(T amb -T ref )×e -β×Δt

[0016] Where T cor is the correction value of the fuel cell stack temperature, where T raw is the raw reading of the fuel cell stack temperature, where T amb is the ambient temperature, T ref is the preset reference temperature standard value; wherein α is the preset material thermal expansion coefficient, β is the preset time decay constant, and Δt is the acquisition time difference between the ambient temperature and the fuel cell stack temperature.

[0017] The calculated correction value of the fuel cell stack temperature is sent to the multi-physics field coordinated control module.

[0018] As a preferred embodiment of the present invention, feature extraction is performed on the pre-processed signal to monitor load mutations and voltage fluctuations. The specific process is as follows:

[0019] For the battery load, the battery load data is converted from the time domain to the frequency domain through wavelet transform to obtain the decomposed wavelet coefficient W j,k , where j is the scale factor, i.e. the number of layers where the wavelet coefficients are located, j = 1, 2, ..., N, where N is the total number of layers of wavelet coefficients; k is the offset factor, i.e. the node number where the wavelet is located, k = 1, 2, ..., n, where n is the total number of node numbers contained in each layer of wavelet coefficients.

[0020] By preset formula Calculate the energy characteristic value E of the battery load at scale factor j j and wavelet packet energy entropy change criterion S entropy .

[0021] For the DC bus voltage, its harmonic voltage is obtained by Fourier transform, including the harmonic voltage B when the harmonic order k=2, 2, ..., 50 1,rms , B 2,rms ,...,B 50,rms .

[0022] By default formula: Calculate the voltage ripple ratio RF.

[0023] The multi-source load data is sent to the dynamic power optimization decision module and the multi-physics field collaborative control module.

[0024] The wavelet packet energy entropy change criterion and voltage ripple rate are sent to the life prediction module.

[0025] The dynamic power optimization decision module builds a real-time rolling optimization model based on collected multi-source load data and introduces a dynamic penalty factor mechanism. Based on the equivalent hydrogen consumption minimization strategy, it solves the multi-constrained nonlinear optimization problem in real time and calculates the optimal fuel cell output power.

[0026] Build a real-time rolling optimization model, including objective functions, dynamic penalty factors, and constraints.

[0027] The objective function is:

[0028]

[0029] Among them, P fc is the fuel cell output power, where t is time, T is the preset rolling optimization time domain length. Where mH2 is the hydrogen mass flow rate, where N cell is the number of cells contained in the fuel cell stack, where F is the Faraday constant, V cell is the average voltage of the single battery, η faraday Faraday efficiency.

[0030] Where τ is the time integral variable and λ(τ) is the dynamic penalty factor.

[0031] The dynamic penalty factor is as follows:

[0032]

[0033] Where k1 and k2 are the preset basic penalty coefficient and load sensitivity coefficient respectively, ΔI load is the overall load current change rate of the fuel cell, which is obtained by differentiating the load current of each single cell in the fuel cell in the multi-source load data and then calculating the average value.

[0034] The constraints are:

[0035]

[0036] Among them, P fc min and P fc max are the preset minimum and maximum values ​​of the fuel cell output power respectively. is the open circuit voltage of the fuel cell, 4R int is the internal resistance of the fuel cell.

[0037] Based on the objective function, dynamic penalty factor and constraint conditions, the optimization results are calculated to obtain the fuel cell output power when the objective function value is the minimum. It is recorded as the optimal fuel cell output power and sent to the multi-physics field collaborative control module.

[0038] The multi-physics collaborative control module achieves precise power tracking under complex operating conditions through multivariable decoupling control, optimizing the cooling fan and intake valve calculations based on the optimal fuel cell output power. It further analyzes the collected multi-source load data using pre-set fault detection rules, and matches the pre-set fault detection results through numerical analysis.

[0039] Receive and record the optimal fuel cell output power and multi-source load data, and input them into the multivariable decoupling feedforward compensator. The specific process is as follows:

[0040] Cooling fan power u fan and intake valve opening u value is the dependent variable, with the optimal fuel cell output power For known quantities, by preset formula:

[0041]

[0042] Solve for the cooling fan power u fan and intake valve opening u value .

[0043] Where kp is the preset proportional gain, ki is the preset integral gain, t is the time, T error is the correction value of the fuel cell stack temperature T cor The difference from the preset standard temperature; P error The current fuel cell output power and the optimal fuel cell output power difference.

[0044] The calculated cooling fan power u fan and intake valve opening u value Sent to the actuator drive and energy conversion module.

[0045] As a preferred embodiment of the present invention, fault detection decision-making is performed based on an SVM support vector machine to determine the fault category. The specific process is as follows:

[0046] Obtain multi-source load data: the load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency of each single cell in the fuel cell, generate a state data set X(t), and input it into the preset SVM formula:

[0047]

[0048] Calculate the support vector machine's label eigenvalue f[X(t)], where αp is the trained Lagrange multiplier and yp is the sample label value. A value of -1 indicates an abnormality in the multi-source load data component p, and a value of 1 indicates normal data. p is the component label for the multi-source load data component p, and p = 1, 2, 3, 4, 5, 6, 7, and 8 correspond to load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency, respectively. K(X(t), X) is the RBF kernel function, and b is the preset decision bias.

[0049] The output label feature value f[X(t)] has a numerical range of 1 to 100. The preset fault category labels are matched according to the specific numerical values ​​of the label feature value f[X(t)] of the support vector machine, including normal use, hydrogen permeation, high temperature operation, membrane dehydration and air starvation.

[0050] Each fault category label has a unique corresponding preset confidence interval in the numerical range of 1 to 100. At every preset time interval, the collected multi-source load data is input into the preset SVM formula to calculate the label feature value f[X(t)]. When U consecutive label feature values ​​f[X(t)] fall into the confidence interval of the same fault category label, the corresponding fault category label is output.

[0051] The generated fault category label is sent to the actuator drive and energy conversion module.

[0052] The actuator drive and energy conversion module obtains the fault category label and performs optimization operations. It also obtains the cooling fan power and intake valve opening and performs heat dissipation optimization operations.

[0053] Get the fault category label and perform the preset optimization operation:

[0054] When the fault category label is identified as normal use, no other unnecessary actions are taken;

[0055] When the fault category label is identified as hydrogen permeation, the output power of the fuel cell is limited to 50% of the current power for five minutes, and the hydrogen circulation pump is controlled to reduce the speed until the hydrogen supply pressure drops to 70% of the normal value.

[0056] When the fault category label is identified as high temperature operation, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the fuel cell cooling system, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value .

[0057] When the fault category label is identified as membrane dehydration, the output power is gradually reduced at a rate of 5% every 15 minutes until the fault category label disappears. Then, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the cooling system of the fuel cell, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value .

[0058] When the fault category label is identified as air starvation, the management personnel are reminded to clean the air filter and increase the air compressor speed to 1.2 times the rated value.

[0059] The life prediction module predicts the fuel cell life based on the average value of the fuel cell load current, fuel cell stack temperature and battery internal resistance, and determines whether the remaining life of the fuel cell can meet the usage conditions.

[0060] Obtain the load current I and extract the effective value of the current, that is, the harmonic current I when the harmonic order k=1 after Fourier transform of the load current I rms .

[0061] Obtain the average value of the battery internal resistance R and the correction value of the fuel cell stack temperature T cor And the load current I, through the preset formula:

[0062] Establish an electrical degradation model to calculate the estimated decay rate of the fuel cell internal resistance

[0063] The service life is evaluated based on the estimated attenuation rate of the battery internal resistance and the average value of the battery internal resistance:

[0064] If the average value R of the battery internal resistance is greater than the first preset threshold value and the estimated decay rate If the value is greater than a preset second threshold, it is determined that the fuel cell is about to reach the end of its service life and needs to be replaced in time;

[0065] If the average value R of the battery internal resistance is less than or equal to the first preset threshold value, and the estimated decay rate If the value is less than or equal to the preset second threshold, it is determined that the remaining service life of the fuel cell meets the requirements of the current working environment and does not need to be replaced.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] 1. The present invention collects multi-source load data through the environmental perception and multi-source load monitoring module. After pre-processing and feature extraction, the dynamic power optimization decision module constructs a real-time rolling optimization model. Based on the equivalent hydrogen consumption minimization strategy, it solves the multi-constrained nonlinear optimization problem in real time and calculates the optimal fuel cell output power, achieving precise energy supply and improving energy utilization efficiency.

[0068] 2. This invention achieves precise power tracking under complex operating conditions through multivariable decoupling control. It optimizes the cooling fan and intake valve based on the optimal fuel cell output power. It also uses a support vector machine (SVM) for fault detection and decision-making, determines the fault type, and executes corresponding optimization operations to ensure stable system operation.

[0069] 3. The present invention predicts the fuel cell life based on the average value of the fuel cell load current, fuel cell stack temperature and battery internal resistance, and determines whether the remaining life of the fuel cell can meet the usage conditions, providing a scientific basis for the maintenance and replacement of the fuel cell and reducing the cost of use. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings:

[0071] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0072] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0073] See also Figure 1 As shown, an adaptive load intelligent energy supply platform for a fuel cell range-extended power system includes an environmental perception and multi-source load monitoring module, a dynamic power optimization decision module, a multi-physical field collaborative control module, an actuator drive and energy conversion module, and a life prediction module.

[0074] The environmental perception and multi-source load monitoring module accesses a distributed sensor network to collect multi-source load data from each fuel cell, building the foundational data layer for the digital twin. This multi-source load data is preprocessed using a drift compensation algorithm to remove noise and environmental errors. Feature extraction is performed on the preprocessed signals to monitor for sudden load changes and voltage ripple.

[0075] The collected multi-source load data includes:

[0076] The load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance and mechanical vibration frequency of each single cell in the fuel cell.

[0077] Furthermore, the fuel cell stack temperature is drift compensated based on the ambient temperature data, and the thermocouple is nonlinearly corrected. The drift compensation formula is:

[0078] T cor =T raw +α×(T amb -T ref )×e -β×Δt

[0079] Where T cor is the correction value of the fuel cell stack temperature, where T raw is the raw reading of the fuel cell stack temperature, where T amb is the ambient temperature, T ref is the preset reference temperature standard value; wherein α is the preset material thermal expansion coefficient, β is the preset time decay constant, and Δt is the acquisition time difference between the ambient temperature and the fuel cell stack temperature.

[0080] The calculated correction value of the fuel cell stack temperature is sent to the multi-physics field coordinated control module.

[0081] Furthermore, feature extraction is performed on the pre-processed signal to monitor load mutations and voltage fluctuations. The specific process is as follows:

[0082] For the battery load, the battery load data is converted from the time domain to the frequency domain through wavelet transform to obtain the decomposed wavelet coefficient W j,k , where j is the scale factor, i.e. the number of layers where the wavelet coefficients are located, j = 1, 2, ..., N, where N is the total number of layers of wavelet coefficients; k is the offset factor, i.e. the node number where the wavelet is located, k = 1, 2, ..., n, where n is the total number of node numbers contained in each layer of wavelet coefficients.

[0083] By preset formula Calculate the energy characteristic value E of the battery load at scale factor j j and wavelet packet energy entropy change criterion S entropy .

[0084] For the DC bus voltage, its harmonic voltage is obtained by Fourier transform, including the harmonic voltage B when the harmonic order k=2, 2, ..., 50 1,rms , B 2,rms ,...,B 50,rms.

[0085] By default formula: Calculate the voltage ripple ratio RF.

[0086] The multi-source load data is sent to the dynamic power optimization decision module and the multi-physics field collaborative control module.

[0087] The wavelet packet energy entropy change criterion and voltage ripple rate are sent to the life prediction module.

[0088] The dynamic power optimization decision module builds a real-time rolling optimization model based on collected multi-source load data and introduces a dynamic penalty factor mechanism. Based on the equivalent hydrogen consumption minimization strategy, it solves the multi-constrained nonlinear optimization problem in real time and calculates the optimal fuel cell output power.

[0089] Build a real-time rolling optimization model, including objective functions, dynamic penalty factors, and constraints.

[0090] The objective function is:

[0091]

[0092] Among them, P fc is the fuel cell output power, where t is time, T is the preset rolling optimization time domain length. Where mH2 is the hydrogen mass flow rate, where N cell is the number of cells contained in the fuel cell stack, where F is the Faraday constant, V cell is the average voltage of the single battery, η faraday Faraday efficiency.

[0093] Where τ is the time integral variable and λ(τ) is the dynamic penalty factor.

[0094] The dynamic penalty factor is as follows:

[0095]

[0096] Where k1 and k2 are the preset basic penalty coefficient and load sensitivity coefficient respectively, ΔI load is the overall load current change rate of the fuel cell, which is obtained by differentiating the load current of each single cell in the fuel cell in the multi-source load data and then calculating the average value.

[0097] The constraints are:

[0098]

[0099] Among them, P fc min and P fc max are the preset minimum and maximum values ​​of the fuel cell output power respectively. is the open circuit voltage of the fuel cell, 4R int is the internal resistance of the fuel cell.

[0100] Based on the objective function, dynamic penalty factor and constraint conditions, the optimization results are calculated to obtain the fuel cell output power when the objective function value is the minimum. It is recorded as the optimal fuel cell output power and sent to the multi-physics field collaborative control module.

[0101] The multi-physics collaborative control module achieves precise power tracking under complex operating conditions through multivariable decoupling control, optimizing the cooling fan and intake valve calculations based on the optimal fuel cell output power. It further analyzes the collected multi-source load data using pre-set fault detection rules, and matches the pre-set fault detection results through numerical analysis.

[0102] Receive and record the optimal fuel cell output power and multi-source load data, and input them into the multivariable decoupling feedforward compensator. The specific process is as follows:

[0103] Cooling fan power u fan and intake valve opening u value is the dependent variable, with the optimal fuel cell output power For known quantities, by preset formula:

[0104]

[0105] Solve for the cooling fan power u fan and intake valve opening u value .

[0106] Where kp is the preset proportional gain, ki is the preset integral gain, t is the time, T error is the correction value of the fuel cell stack temperature T cor The difference from the preset standard temperature; P error The current fuel cell output power and the optimal fuel cell output power difference.

[0107] The calculated cooling fan power u fan and intake valve opening u value Sent to the actuator drive and energy conversion module.

[0108] Furthermore, the fault detection decision is made based on the SVM support vector machine to determine the fault category. The specific process is as follows:

[0109] Obtain multi-source load data: the load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency of each single cell in the fuel cell, generate a state data set X(t), and input it into the preset SVM formula:

[0110]

[0111] Calculate the support vector machine's label eigenvalue f[X(t)], where αp is the trained Lagrange multiplier and yp is the sample label value. A value of -1 indicates an abnormality in the multi-source load data component p, and a value of 1 indicates normal data. p is the component label for the multi-source load data component p, and p = 1, 2, 3, 4, 5, 6, 7, and 8 correspond to load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency, respectively. K(X(t), X) is the RBF kernel function, and b is the preset decision bias.

[0112] The output label feature value f[X(t)] has a numerical range of 1 to 100. The preset fault category labels are matched according to the specific numerical values ​​of the label feature value f[X(t)] of the support vector machine, including normal use, hydrogen permeation, high temperature operation, membrane dehydration and air starvation.

[0113] Each fault category label has a unique corresponding preset confidence interval in the numerical range of 1 to 100. At every preset time interval, the collected multi-source load data is input into the preset SVM formula to calculate the label feature value f[X(t)]. When U consecutive label feature values ​​f[X(t)] fall into the confidence interval of the same fault category label, the corresponding fault category label is output.

[0114] The generated fault category label is sent to the actuator drive and energy conversion module.

[0115] The actuator drive and energy conversion module obtains the fault category label and performs optimization operations. It also obtains the cooling fan power and intake valve opening and performs heat dissipation optimization operations.

[0116] Get the fault category label and perform the preset optimization operation:

[0117] When the fault category label is identified as normal use, no other unnecessary actions are taken;

[0118] When the fault category label is identified as hydrogen permeation, the output power of the fuel cell is limited to 50% of the current power for five minutes, and the hydrogen circulation pump is controlled to reduce the speed until the hydrogen supply pressure drops to 70% of the normal value.

[0119] When the fault category label is identified as high temperature operation, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the fuel cell cooling system, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value .

[0120] When the fault category label is identified as membrane dehydration, the output power is gradually reduced at a rate of 5% every 15 minutes until the fault category label disappears. Then, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the cooling system of the fuel cell, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value .

[0121] When the fault category label is identified as air starvation, the management personnel are reminded to clean the air filter and increase the air compressor speed to 1.2 times the rated value.

[0122] The life prediction module predicts the fuel cell life based on the average value of the fuel cell load current, fuel cell stack temperature and battery internal resistance, and determines whether the remaining life of the fuel cell can meet the usage conditions.

[0123] Obtain the load current I and extract the effective value of the current, that is, the harmonic current I when the harmonic order k=1 after Fourier transform of the load current I rms .

[0124] Obtain the average value of the battery internal resistance R and the correction value of the fuel cell stack temperature T cor And the load current I, through the preset formula:

[0125] Establish an electrical degradation model to calculate the estimated decay rate of the fuel cell internal resistance

[0126] The service life is evaluated based on the estimated attenuation rate of the battery internal resistance and the average value of the battery internal resistance:

[0127] If the average value R of the battery internal resistance is greater than the first preset threshold value and the estimated decay rate If the value is greater than a preset second threshold, it is determined that the fuel cell is about to reach the end of its service life and needs to be replaced in time;

[0128] If the average value R of the battery internal resistance is less than or equal to the first preset threshold value, and the estimated decay rate If the value is less than or equal to the preset second threshold, it is determined that the remaining service life of the fuel cell meets the requirements of the current working environment and does not need to be replaced.

[0129] It should be understood that the terms “include” and “comprising” used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0130] It should also be understood that the terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and the claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations;

[0131] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An adaptive load intelligent energy supply platform for a fuel cell range-extended power system, comprising an environmental perception and multi-source load monitoring module, a dynamic power optimization decision module, and a multi-physics field collaborative control module, characterized by: The environmental perception and multi-source load monitoring module accesses the distributed sensor network to collect multi-source load data from each cell in the fuel cell, including the load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, cell internal resistance, and mechanical vibration frequency. It pre-processes the multi-source load data using a drift compensation algorithm to remove noise and environmental errors. It then performs feature extraction on the pre-processed signals to monitor for sudden load changes and voltage fluctuations. The dynamic power optimization decision module builds a real-time rolling optimization model based on collected multi-source load data and introduces a dynamic penalty factor mechanism. It solves multi-constrained nonlinear optimization problems in real time based on the equivalent hydrogen consumption minimization strategy to calculate the optimal fuel cell output power. The multi-physics collaborative control module achieves precise power tracking under complex working conditions through multi-variable decoupling control, and optimizes the calculation of the cooling fan and intake valve based on the optimal fuel cell output power. It further analyzes the collected multi-source load data through preset fault detection rules, and matches the preset fault detection results through numerical analysis based on the SVM support vector machine.

2. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: It also includes actuator drive and energy conversion module and life prediction module: The actuator drive and energy conversion module obtains the fault category label and performs optimization operations; obtains the cooling fan power and intake valve opening, and performs heat dissipation optimization operations; The life prediction module predicts the fuel cell life based on the average value of the fuel cell load current, fuel cell stack temperature and battery internal resistance, and determines whether the remaining life of the fuel cell can meet the usage conditions.

3. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: The specific process of preprocessing multi-source load data through the drift compensation algorithm is as follows: Based on the ambient temperature data, the fuel cell stack temperature is drift compensated and the thermocouple is nonlinearly corrected. The drift compensation formula is: T cor =T raw +α×(T amb -T ref )×e -β×Δt Where T cor is the correction value of the fuel cell stack temperature, where T raw is the raw reading of the fuel cell stack temperature, where T amb is the ambient temperature, T ref is the preset reference temperature standard value; wherein α is the preset material thermal expansion coefficient, β is the preset time decay constant, and Δt is the acquisition time difference between the ambient temperature and the fuel cell stack temperature; The calculated correction value of the fuel cell stack temperature is sent to the multi-physics field coordinated control module.

4. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: The specific process of feature extraction and monitoring load mutation and voltage fluctuation is as follows: For the battery load, the battery load data is converted from the time domain to the frequency domain through wavelet transform to obtain the decomposed wavelet coefficient W j,k , where j is the scale factor, i.e. the number of layers where the wavelet coefficients are located, j = 1, 2, ..., N, where N is the total number of layers of wavelet coefficients; k is the offset factor, i.e. the node number where the wavelet is located, k = 1, 2, ..., n, where n is the total number of node numbers contained in each layer of wavelet coefficients; By preset formula Calculate the energy characteristic value E of the battery load at scale factor j j and wavelet packet energy entropy change criterion S entropy ; For the DC bus voltage, its harmonic voltage is obtained by Fourier transform, including the harmonic voltage B when the harmonic order k=2, 2, ..., 50 1,rms , B 2,rms ,...,B 50,rms ; By default formula: Calculate the voltage ripple ratio RF; Send multi-source load data to the dynamic power optimization decision module and the multi-physics field collaborative control module; The wavelet packet energy entropy change criterion and voltage ripple rate are sent to the life prediction module.

5. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: The real-time rolling optimization model includes objective function, dynamic penalty factor and constraints; The objective function is: Among them, P fc is the fuel cell output power, where t is time, T is the preset rolling optimization time domain length; where mH2 is the hydrogen mass flow rate, where N cell is the number of cells contained in the fuel cell stack, where F is the Faraday constant, V cell is the average voltage of the single battery, η faraday Faraday efficiency; Where τ is the time integral variable and λ(τ) is the dynamic penalty factor; The dynamic penalty factor is as follows: Where k1 and k2 are the preset basic penalty coefficient and load sensitivity coefficient respectively, ΔI load is the overall load current change rate of the fuel cell, which is obtained by differentiating the load current of each single cell in the fuel cell in the multi-source load data and then averaging it; The constraints are: Among them, P fc min and P fc max are the preset minimum and maximum values ​​of the fuel cell output power respectively; is the open circuit voltage of the fuel cell, 4R int is the internal resistance of the fuel cell; Based on the objective function, dynamic penalty factor and constraint conditions, the optimization results are calculated to obtain the fuel cell output power when the objective function value is the minimum. It is recorded as the optimal fuel cell output power and sent to the multi-physics field collaborative control module.

6. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: The specific process of optimizing the cooling fan and intake valve based on the optimal fuel cell output power is as follows: Cooling fan power u fan and intake valve opening u value is the dependent variable, with the optimal fuel cell output power For known quantities, by preset formula: Solve for the cooling fan power u fan and intake valve opening u value ; Where kp is the preset proportional gain, ki is the preset integral gain, t is the time, T error is the correction value of the fuel cell stack temperature T cor The difference from the preset standard temperature; P error The current fuel cell output power and the optimal fuel cell output power difference; The calculated cooling fan power u fan and intake valve opening u value Sent to the actuator drive and energy conversion module.

7. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 1, characterized in that: The specific process of matching the preset fault detection results through numerical analysis based on SVM support vector machine is as follows: Obtain multi-source load data: the load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency of each single cell in the fuel cell, generate a state data set X(t), and input it into the preset SVM formula: Calculate the label eigenvalue f[X(t)] of the support vector machine, where αp is the trained Lagrange multiplier, yp is the sample label value, -1 indicates that the multi-source load data component p is abnormal, and 1 indicates that the multi-source load data component p is normal; p is the component label of the multi-source load data component p, and p=1, 2, 3, 4, 5, 6, 7, 8 correspond to load current, load voltage, ambient temperature, intake pressure, fuel cell stack temperature, DC bus voltage, battery internal resistance, and mechanical vibration frequency, respectively; K(X(t), X) is the RBF kernel function, and b is the preset decision bias; The output label feature value f[X(t)] has a numerical range of 1 to 100. The specific numerical value of the label feature value f[X(t)] of the support vector machine is matched with the preset fault category label, including normal use, hydrogen permeation, high temperature operation, membrane dehydration and air starvation; Each fault category label has a unique corresponding preset confidence interval in the numerical range of 1 to 100; At every preset time interval, the collected multi-source load data is input into the preset SVM formula to calculate the label feature value f[X(t)]. When U consecutive label feature values ​​f[X(t)] fall into the confidence interval of the same fault category label, the corresponding fault category label is output; The generated fault category label is sent to the actuator drive and energy conversion module.

8. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 2, characterized in that: The specific process of performing thermal optimization operations is as follows: Get the fault category label and perform the preset optimization operation: When the fault category label is identified as normal use, no other unnecessary actions are taken; When the fault category label is identified as hydrogen permeation, the output power of the fuel cell is limited to 50% of the current power for five minutes, and the speed of the hydrogen circulation pump is controlled to decrease until the hydrogen supply pressure drops to 70% of the normal value; When the fault category label is identified as high temperature operation, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the fuel cell cooling system, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value ; When the fault category label is identified as membrane dehydration, the output power is gradually reduced at a rate of 5% every 15 minutes until the fault category label disappears. Then, the cooling fan power and intake valve opening are obtained, converted into target control signals for the cooling fan power and intake valve opening, and output to the cooling system of the fuel cell, so that the cooling fan power and intake valve opening are adjusted to the output value u of the multivariable decoupling feedforward compensator respectively. fan and u value ; When the fault category label is identified as air starvation, the management personnel are reminded to clean the air filter and increase the air compressor speed to 1.2 times the rated value.

9. The adaptive load intelligent energy supply platform of a fuel cell range-extending power system according to claim 2, characterized in that: The specific process of predicting fuel cell life is as follows: Obtain the load current I and extract the effective value of the current, that is, the harmonic current I when the harmonic order k=1 after Fourier transform of the load current I rms ; Obtain the average value of the battery internal resistance R and the correction value of the fuel cell stack temperature T cor And the load current I, through the preset formula: Establish an electrical degradation model to calculate the estimated decay rate of the fuel cell internal resistance The service life is evaluated based on the estimated attenuation rate of the battery internal resistance and the average value of the battery internal resistance: If the average value R of the battery internal resistance is greater than the first preset threshold value and the estimated decay rate If the value is greater than a preset second threshold, it is determined that the fuel cell is about to reach the end of its service life and needs to be replaced in time; If the average value R of the battery internal resistance is less than or equal to the first preset threshold value, and the estimated decay rate If the value is less than or equal to the preset second threshold, it is determined that the remaining service life of the fuel cell meets the requirements of the current working environment and does not need to be replaced.

Images