Hydrogen fuel cell performance prediction model construction method and device and hydrogen fuel cell performance analysis method and device

Abstract

The invention relates to a hydrogen fuel cell performance prediction model construction and performance analysis method and device, and is applied to the technical field of hydrogen fuel cells, and the cell performance prediction model construction method comprises the following steps: obtaining operation data of a hydrogen fuel cell at a plurality of moments, and calculating the total voltage loss at the moment according to the operation data at the single moment; carrying out independent component analysis on the total voltage loss at multiple moments to obtain the voltage loss of three components at each moment; taking the voltage loss of the continuous N moment components as input data, and taking the voltage loss of the next moment component of the continuous N moments as label data to construct training data corresponding to the components; constructing a transform prediction model corresponding to each component, and training the transform prediction model according to the training data corresponding to the component; and when a termination condition is satisfied, ending the training to obtain a hydrogen fuel cell performance prediction model. The accuracy of model training can be improved.

Description

Technical Field

[0001] The present application relates to the field of hydrogen fuel cell technology, and in particular to a hydrogen fuel cell performance prediction model construction, performance analysis method and device. Background Art

[0002] Hydrogen fuel cells are clean, environmentally friendly energy conversion devices that are widely used in automobiles, portable energy storage devices, and other fields with high energy efficiency requirements. The performance of hydrogen fuel cells is affected by many factors, and how to effectively evaluate and analyze the performance of hydrogen fuel cells has become an important research topic in this field.

[0003] The performance analysis methods of hydrogen fuel cells mainly include mechanism models based on physical principles and data-driven machine learning models. The mechanism model predicts the performance of the battery under different working conditions by describing phenomena such as reaction processes, heat transfer and current flow. The machine learning model uses data-driven and deep learning methods to train it with a large amount of battery operation data to achieve the performance calculation of hydrogen fuel cells. However, the existing machine learning models have low prediction accuracy for hydrogen fuel cell performance, and the prediction results of machine learning algorithms are poorly interpretable. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a method and device for constructing a hydrogen fuel cell performance prediction model, a method and device for analyzing hydrogen fuel cell performance based on an AI large model, an electronic device and a storage medium.

[0005] According to a first aspect of the present application, a method for constructing a hydrogen fuel cell performance prediction model is provided, comprising: Acquiring operating data of the hydrogen fuel cell at multiple moments, and calculating the total voltage loss at a single moment according to the operating data at the moment; Based on the physical mechanism of hydrogen fuel cell output voltage, independent component analysis is performed on the total voltage loss at multiple moments to obtain the voltage loss of three components at each moment; the three components include polarization, ohm and concentration difference; For each component, the voltage loss of the component at N consecutive moments is used as input data, and the voltage loss of the component at the next moment of the N consecutive moments is used as label data to construct training data corresponding to the component; N is a positive integer; Constructing a transformer prediction model corresponding to each component, and training the transformer prediction model according to the training data corresponding to the component; The training ends when the termination condition is met, and a hydrogen fuel cell performance prediction model is obtained.

[0006] Optionally, based on the physical mechanism of the hydrogen fuel cell output voltage, independent component analysis is performed on the total voltage loss at multiple moments to obtain voltage losses of three components at each moment, including: The multiple moments are divided into M cycles, and based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle, multiple constant parameters corresponding to the cycle and used to calculate the voltage loss of the three components are determined; M is a positive integer; Selecting any moment from each cycle, and calculating the reference voltage loss of the three components at the moment according to the operation data at the moment and a plurality of constant parameters; According to the reference voltage losses of the three components at M moments, independent component analysis is performed on the total voltage loss at multiple moments to obtain the voltage losses of the three components at each moment.

[0007] Optionally, performing independent component analysis on the total voltage loss at multiple moments according to the reference voltage losses of the three components at M moments to obtain the voltage losses of the three components at each moment includes: Take the objective function: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S ; in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient.

[0008] Optionally, the method further comprises: In the process of training the transformer prediction model according to the training data corresponding to the components, the hyperparameters of the transformer prediction model are optimized using an optimization algorithm.

[0009] According to the second aspect of the present application, a hydrogen fuel cell performance analysis method based on an AI large model is provided, comprising: Obtaining user instructions, and using a pre-built large language model for hydrogen fuel cells to identify the user instructions and determine whether they are instructions for hydrogen fuel cell performance analysis; If the user instruction does not belong to the hydrogen fuel cell performance analysis instruction, the corresponding query result is output; If the user instruction is a hydrogen fuel cell performance analysis instruction, the hydrogen fuel cell operation data corresponding to the user instruction is obtained, and the model corresponding to the user instruction is called from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain a performance analysis result; Wherein, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the method described in the first aspect.

[0010] Optionally, the method for constructing the hydrogen fuel cell professional large language model includes: Acquire hydrogen fuel cell knowledge base corpus, and convert the hydrogen fuel cell knowledge base corpus into question-answer pair data; A general large language model is obtained, and the general large language model is fine-tuned using the question-answer pair data to obtain the hydrogen fuel cell professional large language model.

[0011] According to a third aspect of the present application, a hydrogen fuel cell performance prediction model construction device is provided, comprising: A total voltage loss calculation module, used to obtain the operating data of the hydrogen fuel cell at multiple moments, and calculate the total voltage loss at a single moment based on the operating data at the moment; An independent component analysis module is used to perform independent component analysis on the total voltage loss at multiple moments based on the physical mechanism of the hydrogen fuel cell output voltage, and obtain the voltage loss of three components at each moment; the three components include polarization, ohm and concentration difference; A training data construction module is used to construct training data corresponding to each component by taking the voltage loss of the component at N consecutive moments as input data and the voltage loss of the component at the next moment of the N consecutive moments as label data; N is a positive integer; A transformer prediction model training module, used to construct a transformer prediction model corresponding to each component, and train the transformer prediction model according to the training data corresponding to the component; The hydrogen fuel cell performance prediction model generation module is used to terminate the training when the termination condition is met to obtain the hydrogen fuel cell performance prediction model.

[0012] Optionally, the independent component analysis module is specifically used to divide the multiple moments into M cycles, and determine a plurality of constant parameters corresponding to the cycle for calculating the voltage loss of the three components based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle; M is a positive integer; select any moment from each cycle, and calculate the reference voltage loss of the three components at the moment according to the operating data at the moment and a plurality of constant parameters; perform independent component analysis on the total voltage loss at multiple moments according to the reference voltage loss of the three components at M moments, and obtain the voltage loss of the three components at each moment.

[0013] Optionally, the independent component analysis module is specifically used to perform independent component analysis on the total voltage loss at multiple moments according to the reference voltage loss of three components at M moments through the following steps to obtain the voltage loss of three components at each moment: Take the objective function: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S ; in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient.

[0014] Optionally, the transformer prediction model training module is also used to optimize the hyperparameters of the transformer prediction model using an optimization algorithm during the process of training the transformer prediction model according to the training data corresponding to the components.

[0015] According to a fourth aspect of the present application, a hydrogen fuel cell performance analysis device based on an AI large model is provided, comprising: A user instruction acquisition module, used to acquire user instructions; A user instruction recognition module is used to use a pre-built hydrogen fuel cell professional large language model to recognize the user instruction and determine whether it is a hydrogen fuel cell performance analysis instruction; A query result output module, used for outputting a corresponding query result if the user instruction recognition module determines that the user instruction does not belong to a hydrogen fuel cell performance analysis instruction; a performance analysis structure output module, for obtaining hydrogen fuel cell operation data corresponding to the user instruction if the user instruction recognition module determines that the user instruction belongs to a hydrogen fuel cell performance analysis instruction, and calling the model corresponding to the user instruction from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain a performance analysis result; Wherein, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the method described in the first aspect.

[0016] Optionally, the hydrogen fuel cell performance analysis device based on the AI ​​large model also includes: The hydrogen fuel cell professional large language model construction module is used to obtain the hydrogen fuel cell knowledge base corpus and convert the hydrogen fuel cell knowledge base corpus into question-answer pair data; obtain a general large language model and use the question-answer pair data to fine-tune the general large language model to obtain the hydrogen fuel cell professional large language model.

[0017] According to a fifth aspect of the present application, an electronic device is provided, comprising: a processor, wherein the processor is used to execute a computer program stored in a memory, wherein the computer program implements the method described in the first aspect or the second aspect when executed by the processor.

[0018] According to a sixth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first aspect or the second aspect is implemented.

[0019] According to a seventh aspect of the present application, a computer program product is provided. When the computer program product is run on a computer, the computer is caused to execute the method described in the first aspect or the second aspect.

[0020] Compared with the prior art, the technical solution provided by the embodiments of the present application has the following advantages: Based on the physical mechanism of hydrogen fuel cell voltage loss, that is, the actual output voltage of the hydrogen fuel cell is the difference between the theoretical voltage of the hydrogen fuel cell and the activation voltage loss, ohmic voltage loss, and concentration voltage loss, it can be seen that the total voltage loss calculated by collecting the actual output voltage of the hydrogen fuel cell includes three components: activation voltage loss, ohmic voltage loss, and concentration voltage loss. Therefore, the three voltage losses can be separated from the total voltage loss by independent component analysis. Then, a transformer prediction model of the three components is constructed to analyze the changing trend of the loss of each component with higher accuracy. At the same time, the loss changes of each component can be corresponded in the result analysis to enhance the interpretability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0023] Figure 1 This is an architecture diagram of a hydrogen fuel cell performance analysis system based on an AI big model in an embodiment of the present application; Figure 2 A flow chart of a method for constructing a hydrogen fuel cell performance prediction model in an embodiment of the present application; Figure 3 Schematic diagram of an independent component analysis method based on physical information constraints in an embodiment of the present application; Figure 4 A schematic diagram of a method for constructing a hydrogen fuel cell performance prediction model in an embodiment of the present application; Figure 5 A flow chart of a gas turbine performance analysis method based on an AI large model in an embodiment of the present application; Figure 6 A schematic diagram of a structure of a device for constructing a hydrogen fuel cell performance prediction model in an embodiment of the present application; Figure 7 A schematic diagram of the structure of a gas turbine performance analysis device based on an AI large model in an embodiment of the present application; Figure 8 A schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the scheme of the present application will be further described below. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0025] In the following description, many specific details are set forth to facilitate a full understanding of the present application, but the present application may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only part of the embodiments of the present application, rather than all of the embodiments.

[0026] As a natural language processing technology based on deep learning, the big language model has powerful text understanding and generation capabilities. With the transformer algorithm as the core, it realizes complex language generation, question-answering and other tasks through pre-training of large-scale text data. In recent years, with the improvement of computing power and the accumulation of large amounts of data, related technologies based on the big language model have developed rapidly and have achieved significant application results in many fields. Through natural language interaction, it can effectively improve user experience and operational efficiency and provide more intelligent and automated problem solutions.

[0027] However, although the general large language model has made significant progress in the field of natural language processing, there are still some deviations in the understanding of hydrogen fuel cell expertise. Its understanding of specific questions and the generation of specific answers are often not targeted, resulting in insufficient depth and accuracy in the explanation of specific questions.

[0028] Based on this, the embodiment of the present application proposes a professional large language model based on a professional knowledge base of hydrogen fuel cells, namely, a professional large language model of hydrogen fuel cells. The professional large language model of hydrogen fuel cells can receive operating data of hydrogen fuel cells under different working conditions, and realize efficient current performance calculation, future performance prediction and result analysis under complex working conditions of hydrogen fuel cells through natural language interaction with user instructions.

[0029] Figure 1 This is the architecture diagram of the hydrogen fuel cell performance analysis system based on the AI ​​big model in the embodiment of this application. The system is based on the hydrogen fuel cell professional big language model as the core, based on intelligent agent technology, integrating the two functional modules of hydrogen fuel cell knowledge base and hydrogen fuel cell performance analysis model, to achieve efficient natural language interaction for hydrogen fuel cell performance analysis.

[0030] Among them, the hydrogen fuel cell knowledge base corpus includes: basic theoretical knowledge of hydrogen fuel cell composition structure, operating principle and catalyst, proton exchange membrane, gas diffusion layer, bipolar plate and other key components of hydrogen fuel cells, basic scientific knowledge in related fields such as reaction kinetics, fluid mechanics, thermodynamics, heat transfer, and hydrogen fuel cell performance calculation methods, equipment failure types, cause analysis, control schemes and other related technical information. The corpus sources include authoritative professional books, technical specifications and high-level papers and documents to ensure the accuracy of the knowledge base content. The collected corpus is quality evaluated and data cleaned, and the characters in the corpus are extracted through OCR (optical character recognition) technology to provide high-quality text data for large language model input.

[0031] Optionally, the method for constructing a large language model for hydrogen fuel cells includes: obtaining a hydrogen fuel cell knowledge base corpus, converting the hydrogen fuel cell knowledge base corpus into question-answer pair data, obtaining a general large language model, using the general large language model as a base, and fine-tuning the general large language model using the question-answer pair data to obtain a large language model for hydrogen fuel cells.

[0032] The hydrogen fuel cell performance analysis model includes: hydrogen fuel cell power generation hydrogen consumption model, hydrogen fuel cell efficiency calculation model, hydrogen fuel cell rated power average decay rate model and hydrogen fuel cell performance prediction model. The hydrogen fuel cell professional large language model uses intelligent agent technology to select the appropriate performance calculation model for analysis and calculation according to user instruction requirements, and outputs the analysis results through natural language.

[0033] Among them, the hydrogen consumption model of hydrogen fuel cell power generation can be expressed as the formula: ; H is the hydrogen consumption of hydrogen fuel cell power generation at rated power, in kg / kWh. is the mass of hydrogen consumed at the rated power of the hydrogen fuel cell, in kg / s, and P is the power generation power of the hydrogen fuel cell, in kW.

[0034] The hydrogen fuel cell efficiency calculation model can be expressed as the formula: ; For hydrogen fuel cell efficiency, , are the hydrogen inlet and outlet flow rates, in kg / s, P a is the power consumption of the battery-assisted system, P s For the battery stack, V cell is the voltage of the hydrogen fuel cell unit, E 0 is the theoretical voltage of the hydrogen fuel cell, and the default voltage is 1.23V.

[0035] The average attenuation rate model of the rated power of hydrogen fuel cells can be expressed as the formula: ; S is the rated power attenuation rate of the hydrogen fuel cell, U 0 is the voltage at the rated current of a new hydrogen fuel cell. U 1 is the voltage at rated current after the hydrogen fuel cell operates to the current state according to the power generation condition.

[0036] The main function of the hydrogen fuel cell performance prediction model is to accurately predict the remaining life of the hydrogen fuel cell. Based on the physical mechanism of hydrogen fuel cell voltage loss, a supervised decomposition analysis of the voltage loss is performed based on the independent component analysis method. The different independent components decomposed are input into the transformer prediction model for training to achieve hydrogen fuel cell performance prediction.

[0037] See also Figure 2 , Figure 2 A flow chart of a method for constructing a hydrogen fuel cell performance prediction model in an embodiment of the present application may include the following steps: Step S202, obtaining the operating data of the hydrogen fuel cell at multiple moments, and calculating the total voltage loss at a single moment according to the operating data at that moment.

[0038] The operating data of the hydrogen fuel cell during operation includes: hydrogen fuel cell voltage, current and operating time parameters, etc.

[0039] According to the formula: ; Calculate the theoretical voltage of hydrogen fuel cells E 0, in, is the change in Gibbs free energy of the chemical reaction of the hydrogen fuel cell, the unit is: J; n is the number of electrons transferred in the chemical reaction, n The value of is 2; F The Faraday constant is 96485 C / mol.

[0040] According to the formula: E loss = mE 0- E a , calculate the total voltage loss E loss .

[0041] in, m is the number of single cells in the hydrogen fuel cell stack, E a is the actual output voltage of the hydrogen fuel cell.

[0042] Step S204 , based on the physical mechanism of the hydrogen fuel cell output voltage, an independent component analysis is performed on the total voltage loss at multiple moments to obtain the voltage loss of three components at each moment.

[0043] Based on the physical mechanism of hydrogen fuel cell output voltage: ; in, V , V act , V ohm , V trans They are hydrogen fuel cell output voltage, activation voltage loss, ohmic voltage loss, and concentration voltage loss.

[0044] It can be seen that the total voltage loss of a hydrogen fuel cell includes three parts: activation voltage loss, ohmic voltage loss, and concentration voltage loss. Therefore, independent component analysis of the total voltage loss can be decomposed into activation voltage loss, ohmic voltage loss, and concentration voltage loss. Subsequently, corresponding transformer prediction models can be constructed for each of the three components.

[0045] In some embodiments, in order to improve the accuracy of blind source separation of independent component analysis, one or more target moments can be selected from multiple moments to calculate the activation voltage loss, ohmic voltage loss, and concentration voltage loss at the target moment. The activation voltage loss, ohmic voltage loss, and concentration voltage loss at the target moment are used as supervisory constraints for the characteristic decomposition of the total voltage loss, so that the characteristic decomposition result of the total voltage loss of the hydrogen fuel cell is more consistent with the physical characteristics of the three voltage losses of polarization, ohm, and concentration.

[0046] For hydrogen fuel cells, the theoretical calculation formula for activation voltage loss is: ; The theoretical calculation formula for ohmic voltage loss is: ; The theoretical calculation formula for concentration voltage loss is: ; Where A is the Tafel slope, R is the gas constant 8.314 J / (mol·K), T is the operating temperature of the hydrogen fuel cell, α is the electrochemical reaction transfer coefficient, which is 0.2-0.5 by default, and i is the operating current density of the hydrogen fuel cell. i loss is the leakage current density of the hydrogen fuel cell, i 0 is the exchange current density of the hydrogen fuel cell. r is the internal resistance of the hydrogen fuel cell, B is the diffusion constant in the battery, i l is the limiting current density of hydrogen fuel cells.

[0047] When the relevant parameters are available, the activation voltage loss, ohmic voltage loss, and concentration voltage loss can be directly calculated according to the above formula. However, the electrochemical reaction transfer coefficient α and leakage current density in the above formula i loss , exchange current density i 0. Limiting current density i l , hydrogen fuel cell internal resistance r, Tafel slope A, battery diffusion constant B and other parameters are all constant parameters related to the battery structure state and operation time. These constant parameters cannot be directly obtained, and activation voltage loss, ohmic voltage loss, and concentration voltage loss cannot be calculated. Therefore, the above constant parameters can be obtained by calculation, and then the activation voltage loss, ohmic voltage loss, and concentration voltage loss can be calculated. The calculated activation voltage loss, ohmic voltage loss, and concentration voltage loss are used as constraints for independent component analysis.

[0048] Independent component analysis methods based on physical information constraints such as Figure 3 As shown. In the embodiment of the present application, it can be considered that the value of the constant parameter is fixed in a short time. Multiple moments are divided into M cycles, and based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle, multiple constant parameters corresponding to the cycle for calculating the voltage loss of the three components are determined; M is a positive integer.

[0049] For example, you can is the objective function, i.e., minimizing the error between the actual voltage and the theoretical voltage, and the above constant parameters are calculated by the least squares method.

[0050] Select any moment from each cycle, and calculate the reference voltage loss of the three components at the moment according to the operating data and multiple constant parameters at the moment. According to the reference voltage loss of the three components at M moments, perform independent component analysis on the total voltage loss at multiple moments to obtain the voltage loss of the three components at each moment. That is, in the process of independent component analysis, the reference voltage loss of the three components at M moments is simultaneously referred to to improve the accuracy of the decomposition result.

[0051] Optionally, the objective function can be: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S .in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient, which can be 1.

[0052] Step S206, for each component, the voltage loss of the component at N consecutive moments is used as input data, and the voltage loss of the component at the next moment after N consecutive moments is used as label data to construct training data corresponding to the component.

[0053] For each component, N is the number of voltage losses of the component input to the transformer prediction model, and N is a positive integer. According to the voltage losses of the component at multiple moments obtained by decomposition, the voltage losses of the component at N consecutive moments can be used as a set of input data, and the voltage loss of the component at the next moment can be used as label data to obtain multiple sets of training data.

[0054] Step S208, constructing a transformer prediction model corresponding to each component, and training the transformer prediction model according to the training data corresponding to the component.

[0055] The settings of the hyperparameters (number of transformer layers, hidden layer dimensions, prediction step length, etc.) of the transformer prediction model are generally based on the designer's experience. In some embodiments, in order to obtain the optimal hyperparameters, in the process of training the transformer prediction model according to the training data corresponding to each component, an optimization algorithm (such as a particle swarm optimization algorithm, a GA genetic algorithm, an ABC bee colony algorithm, etc.) can be used to optimize the hyperparameters of the transformer prediction model.

[0056] Figure 4A schematic diagram of a method for constructing a hydrogen fuel cell performance prediction model in an embodiment of the present application. After the voltage losses of the three components are decomposed by an independent component analysis algorithm based on physical information constraints, the voltage loss of each component is input into the corresponding transformer prediction model. The hyperparameters of the transformer prediction model (transformer layer number 2~6, hidden layer dimension 2~8, prediction step 1~10) are cyclically optimized by the particle swarm optimization algorithm. The particle swarm population size can be set to 18, the maximum number of iterations can be set to 80, and the particle swarm fitness function is set to the MSE (mean square error) of the voltage loss prediction. When the particle swarm algorithm reaches the maximum number of iterations, the optimal hyperparameter combination of the transformer prediction model is output.

[0057] Step S210, the training ends when the termination condition is met, and a hydrogen fuel cell performance prediction model is obtained.

[0058] The method for constructing a hydrogen fuel cell performance prediction model in the embodiment of the present application is based on the physical mechanism of hydrogen fuel cell voltage loss, that is, the actual output voltage of the hydrogen fuel cell is the difference between the theoretical voltage of the hydrogen fuel cell and the activation voltage loss, ohmic voltage loss, and concentration voltage loss. It can be seen that the total voltage loss calculated by collecting the actual output voltage of the hydrogen fuel cell includes three components: activation voltage loss, ohmic voltage loss, and concentration voltage loss. The three voltage losses can be more accurately separated from the total voltage loss by the independent component analysis method based on physical information. Then, a transformer prediction model of three components is constructed, and the change trend of the loss of each component is analyzed to make the accuracy of each transformer prediction model obtained by training higher. At the same time, the loss changes of each component can be corresponding to the result analysis to enhance the interpretability of the model. The optimization algorithm is used to realize the automatic search and optimization of the hyperparameters of the prediction model, so as to obtain the optimal hydrogen fuel cell performance prediction model, avoid the uncertainty of artificially setting hyperparameters, and further improve the model prediction accuracy. Overall, through this architecture design, the hydrogen fuel cell performance prediction model is not only improved in terms of performance prediction accuracy, but also has made progress in model interpretability and optimization efficiency.

[0059] See also Figure 5 , Figure 5 A flow chart of a gas turbine performance analysis method based on an AI large model in an embodiment of the present application includes: Step S502, obtaining user instructions.

[0060] Step S504, using the pre-built hydrogen fuel cell professional large language model, the user instruction is identified to determine whether it is a hydrogen fuel cell performance analysis instruction.

[0061] The hydrogen fuel cell professional language model can identify the user instruction based on its own understanding ability and determine whether the user instruction belongs to the hydrogen fuel cell performance analysis instruction. If the user instruction does not belong to the hydrogen fuel cell performance analysis instruction, execute step S506; if the user instruction belongs to the hydrogen fuel cell performance analysis instruction, execute step S508.

[0062] For example, if the user instruction is "What is the activation of a hydrogen fuel cell?", the hydrogen fuel cell professional language model can identify the user instruction as a general knowledge answer instruction and directly output the query result corresponding to the user instruction. If the user instruction is "How long is the remaining life of a hydrogen fuel cell?", it can be identified that the user instruction belongs to a hydrogen fuel cell performance analysis instruction.

[0063] Step S506: output the corresponding query result.

[0064] Step S508, obtaining hydrogen fuel cell operation data corresponding to the user instruction, and calling the model corresponding to the user instruction from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain a performance analysis result.

[0065] Among them, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the above-mentioned hydrogen fuel cell performance prediction model construction method.

[0066] For example, in response to the user instruction "How long is the remaining life of the hydrogen fuel cell?", the corresponding operating data can be obtained, such as the current operating voltage, operating current and operating time of the hydrogen fuel cell. Then the total voltage loss of the hydrogen fuel cell is calculated, and the voltage losses of the three components are obtained through independent component analysis. The voltage losses of the three components are respectively input into the corresponding transformer prediction model to obtain the predicted voltage losses of the three components. The predicted voltage of the hydrogen fuel cell is calculated based on the predicted voltage losses of the three components, and the remaining life of the hydrogen fuel cell is calculated based on the standard that the battery failure is considered to be 90% below the standard voltage.

[0067] The gas turbine performance analysis method based on the AI ​​large model in the embodiment of the present application improves the ability of the professional large language model for hydrogen fuel cells to understand problems, solve problems, and analyze results by pre-building a professional knowledge base of hydrogen fuel cells. The hydrogen consumption model for hydrogen fuel cell power generation, the hydrogen fuel cell efficiency calculation model, the hydrogen fuel cell rated power average decay rate model, and the hydrogen fuel cell performance prediction model are integrated. Based on intelligent agent technology, it can automatically select the most suitable calculation model according to the user input instructions and provide detailed natural language explanations, thereby solving the selection difficulties and analysis complexity of the current performance calculation model.

[0068] Corresponding to the above method embodiment, the present application embodiment also provides a hydrogen fuel cell performance prediction model construction device, see Figure 6 , the hydrogen fuel cell performance prediction model building device 600 includes: The total voltage loss calculation module 602 is used to obtain the operation data of the hydrogen fuel cell at multiple moments, and calculate the total voltage loss at a moment according to the operation data at a single moment; The independent component analysis module 604 is used to perform independent component analysis on the total voltage loss at multiple moments based on the physical mechanism of the hydrogen fuel cell output voltage, and obtain the voltage loss of three components at each moment; the three components include polarization, ohm and concentration difference; The training data construction module 606 is used to construct the training data corresponding to each component by taking the voltage loss of the component at N consecutive moments as input data and the voltage loss of the component at the next moment of N consecutive moments as label data; N is a positive integer; A transformer prediction model training module 608, for constructing a transformer prediction model corresponding to each component, and training the transformer prediction model according to the training data corresponding to the component; The hydrogen fuel cell performance prediction model generation module 610 is used to terminate the training when the termination condition is met to obtain the hydrogen fuel cell performance prediction model.

[0069] Optionally, the independent component analysis module 604 is specifically used to divide multiple moments into M cycles, and determine multiple constant parameters corresponding to the cycle for calculating the voltage loss of three components based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle; M is a positive integer; select any moment from each cycle, and calculate the reference voltage loss of the three components at the moment according to the operating data at the moment and multiple constant parameters; perform independent component analysis on the total voltage loss of multiple moments based on the reference voltage loss of the three components at M moments to obtain the voltage loss of the three components at each moment.

[0070] Optionally, the independent component analysis module 604 is specifically used to perform independent component analysis on the total voltage loss at multiple moments according to the reference voltage loss of three components at M moments through the following steps to obtain the voltage loss of three components at each moment: Take the objective function: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S ; in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient.

[0071] Optionally, the transformer prediction model training module 608 is further used to optimize the hyperparameters of the transformer prediction model using an optimization algorithm during the process of training the transformer prediction model according to the training data corresponding to the components.

[0072] The present application also provides a hydrogen fuel cell performance analysis device based on an AI large model, see Figure 7 , the hydrogen fuel cell performance analysis device 700 based on the AI ​​large model includes: A user instruction acquisition module 702, used to acquire user instructions; The user instruction recognition module 704 is used to use the pre-built hydrogen fuel cell professional language model to recognize the user instruction and determine whether it is a hydrogen fuel cell performance analysis instruction; The query result output module 706 is used to output the corresponding query result if the user instruction recognition module 704 determines that the user instruction does not belong to the hydrogen fuel cell performance analysis instruction; The performance analysis structure output module 708 is used to obtain the hydrogen fuel cell operation data corresponding to the user instruction if the user instruction recognition module 704 determines that the user instruction belongs to the hydrogen fuel cell performance analysis instruction, and call the model corresponding to the user instruction from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain the performance analysis result; Among them, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the hydrogen fuel cell performance prediction model construction method.

[0073] Optionally, the hydrogen fuel cell performance analysis device 700 based on the AI ​​large model further includes: The hydrogen fuel cell professional large language model construction module is used to obtain the hydrogen fuel cell knowledge base corpus and convert the hydrogen fuel cell knowledge base corpus into question-answer pair data; obtain the general large language model and use the question-answer pair data to fine-tune the general large language model to obtain the hydrogen fuel cell professional large language model.

[0074] The specific details of each module or unit in the above device have been described in detail in the corresponding method, so they will not be repeated here.

[0075] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0076] An embodiment of the present application also provides an electronic device, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the above-mentioned hydrogen fuel cell performance prediction model construction or the gas turbine performance analysis method based on the AI ​​large model in this example embodiment.

[0077] Reference Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. The specific embodiment of the present application does not limit the specific implementation of the electronic device.

[0078] like Figure 8 As shown, the electronic device may include: a processor 802 , a communication interface 804 , a memory 806 , and a communication bus 808 .

[0079] The processor 802 , the communication interface 804 , and the memory 806 communicate with each other via a communication bus 808 .

[0080] The communication interface 804 is used to communicate with other electronic devices or servers.

[0081] The processor 802 is used to execute the program 810, and specifically can execute the relevant steps in the above method embodiment.

[0082] Specifically, the program 810 may include program codes, which include computer operation instructions.

[0083] The processor 802 may be a central processing unit, or a specific integrated circuit, or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0084] The memory 806 is used to store the program 810. The memory 806 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0085] Program 810 can be specifically used to enable processor 802 to execute the steps in the above-mentioned gas turbine performance analysis method embodiment based on AI large model.

[0086] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and modules can refer to the corresponding process descriptions in the aforementioned method embodiments and will not be repeated here.

[0087] A computer-readable storage medium is also provided in an embodiment of the present application, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned hydrogen fuel cell performance prediction model construction method or the gas turbine performance analysis method based on the AI ​​large model is implemented.

[0088] It should be noted that the computer-readable storage medium shown in the present application may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, radio frequency, etc., or any suitable combination of the above.

[0089] In an embodiment of the present application, a computer program product is also provided. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned hydrogen fuel cell performance prediction model construction method or the gas turbine performance analysis method based on the AI ​​large model.

[0090] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0091] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a hydrogen fuel cell performance prediction model, characterized in that: include: Acquiring operating data of the hydrogen fuel cell at multiple moments, and calculating the total voltage loss at a single moment according to the operating data at the moment; Based on the physical mechanism of hydrogen fuel cell output voltage, independent component analysis is performed on the total voltage loss at multiple moments to obtain the voltage loss of three components at each moment; the three components include polarization, ohm and concentration difference; For each component, the voltage loss of the component at N consecutive moments is used as input data, and the voltage loss of the component at the next moment of the N consecutive moments is used as label data to construct training data corresponding to the component; N is a positive integer; Constructing a transformer prediction model corresponding to each component, and training the transformer prediction model according to the training data corresponding to the component; The training ends when the termination condition is met, and a hydrogen fuel cell performance prediction model is obtained.

2. The method according to claim 1, characterized in that Based on the physical mechanism of the output voltage of the hydrogen fuel cell, the total voltage loss at multiple moments is analyzed by independent components to obtain the voltage loss of three components at each moment, including: The multiple moments are divided into M cycles, and based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle, multiple constant parameters corresponding to the cycle and used to calculate the voltage loss of the three components are determined; M is a positive integer; Selecting any moment from each cycle, and calculating the reference voltage loss of the three components at the moment according to the operation data at the moment and a plurality of constant parameters; According to the reference voltage losses of the three components at M moments, independent component analysis is performed on the total voltage loss at multiple moments to obtain the voltage losses of the three components at each moment.

3. The method according to claim 2, characterized in that The method of performing independent component analysis on the total voltage loss at multiple moments according to the reference voltage losses of the three components at M moments to obtain the voltage losses of the three components at each moment includes: Take the objective function: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S ; in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient.

4. The method according to claim 1, characterized in that The method further comprises: In the process of training the transformer prediction model according to the training data corresponding to the components, the hyperparameters of the transformer prediction model are optimized using an optimization algorithm.

5. A hydrogen fuel cell performance analysis method based on AI large model, characterized in that: include: Obtaining user instructions, and using a pre-built large language model for hydrogen fuel cells to identify the user instructions and determine whether they are instructions for hydrogen fuel cell performance analysis; If the user instruction does not belong to the hydrogen fuel cell performance analysis instruction, the corresponding query result is output; If the user instruction is a hydrogen fuel cell performance analysis instruction, the hydrogen fuel cell operation data corresponding to the user instruction is obtained, and the model corresponding to the user instruction is called from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain a performance analysis result; Wherein, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the method described in any one of claims 1 to 4.

6. The method according to claim 5, characterized in that The method for constructing the hydrogen fuel cell professional large language model includes: Acquire hydrogen fuel cell knowledge base corpus, and convert the hydrogen fuel cell knowledge base corpus into question-answer pair data; A general large language model is obtained, and the general large language model is fine-tuned using the question-answer pair data to obtain the hydrogen fuel cell professional large language model.

7. A hydrogen fuel cell performance prediction model construction device, characterized in that: include: A total voltage loss calculation module, used to obtain the operating data of the hydrogen fuel cell at multiple moments, and calculate the total voltage loss at a single moment based on the operating data at the moment; An independent component analysis module is used to perform independent component analysis on the total voltage loss at multiple moments based on the physical mechanism of the hydrogen fuel cell output voltage, and obtain the voltage loss of three components at each moment; the three components include polarization, ohm and concentration difference; A training data construction module is used to construct training data corresponding to each component by taking the voltage loss of the component at N consecutive moments as input data and the voltage loss of the component at the next moment of the N consecutive moments as label data; N is a positive integer; A transformer prediction model training module, used to construct a transformer prediction model corresponding to each component, and train the transformer prediction model according to the training data corresponding to the component; The hydrogen fuel cell performance prediction model generation module is used to terminate the training when the termination condition is met to obtain the hydrogen fuel cell performance prediction model.

8. The device according to claim 7, characterized in that The independent component analysis module is specifically used to divide the multiple moments into M cycles, and determine multiple constant parameters corresponding to the cycle and used to calculate the voltage loss of the three components based on the physical mechanism of the hydrogen fuel cell output voltage and the actual output voltage of the hydrogen fuel cell at multiple moments in each cycle; M is a positive integer; Select any moment from each cycle, and calculate the reference voltage loss of the three components at the moment according to the operating data at the moment and multiple constant parameters; perform independent component analysis on the total voltage loss at multiple moments based on the reference voltage losses of the three components at M moments to obtain the voltage loss of the three components at each moment.

9. The device according to claim 8, characterized in that The independent component analysis module is specifically used to perform independent component analysis on the total voltage loss at multiple moments according to the reference voltage loss of the three components at M moments through the following steps to obtain the voltage loss of the three components at each moment: Take the objective function: To constrain the total voltage loss at multiple times, independent component analysis is performed to obtain the voltage loss of three components: S ; in, X Represents the total voltage loss at a single moment, A represents the signal fusion matrix calculated by independent component analysis, Represents the reference voltage loss of the three components, Represents the weight coefficient.

10. A hydrogen fuel cell performance analysis device based on AI large model, characterized in that: include: A user instruction acquisition module, used to acquire user instructions; A user instruction recognition module is used to use a pre-built hydrogen fuel cell professional large language model to recognize the user instruction and determine whether it is a hydrogen fuel cell performance analysis instruction; A query result output module, used for outputting a corresponding query result if the user instruction recognition module determines that the user instruction does not belong to a hydrogen fuel cell performance analysis instruction; a performance analysis structure output module, for obtaining hydrogen fuel cell operation data corresponding to the user instruction if the user instruction recognition module determines that the user instruction belongs to a hydrogen fuel cell performance analysis instruction, and calling the model corresponding to the user instruction from the pre-built hydrogen fuel cell performance analysis model to process the hydrogen fuel cell operation data to obtain a performance analysis result; Wherein, the hydrogen fuel cell performance analysis model includes a hydrogen fuel cell performance prediction model, and the hydrogen fuel cell performance prediction model is constructed based on the method described in any one of claims 1 to 4.

Images