Method and device for determining attenuation state of fuel cell and fuel cell system

Abstract

The invention discloses a fuel cell attenuation state determination method and device and a fuel cell system. The method comprises the following steps: on the basis of operation data of a fuel cell, obtaining characteristic statistical data, the characteristic statistical data being used for representing a performance change in an operation process of the fuel cell; determining a clustering result of the feature statistical data; fitting is carried out according to the clustering result, and a fitting result of the feature statistical data is obtained; and processing based on the fitting result to obtain a performance degradation result of the fuel cell. The technical problem that the real-time attenuation state analysis effect of the fuel cell is not ideal is solved.

Description

Technical Field

[0001] The present application relates to the field of fuel cell operation analysis, and more specifically, to a method and device for determining a fuel cell attenuation state, and a fuel cell system. Background Art

[0002] The performance degradation characteristics of fuel cells are important characteristics related to the application of fuel cells. In order to ensure the efficient, safe, stable and long-life operation of the fuel cell system, it is necessary to continue to pay attention to the operation degradation of the fuel cell in actual work.

[0003] Fuel cell power generation involves complex electrochemical coupling processes, and it is difficult to fully grasp its reaction mechanism. The regulation of hydrothermal and hydrogen-air systems is an important factor affecting the performance and life of fuel cells. In addition, with the continuous expansion and diversification of hydrogen energy product lines, offline analysis methods have gradually exposed the problems of long cycles, high costs, and low data utilization in dealing with fuel cell performance evaluation. Related technologies have large differences between the collected data and the real data, and no cluster analysis is performed on strategy changes and attenuation inflection points. Expert experience and repeated verification methods are difficult to meet the needs of complex product lines, resulting in low data utilization and unsatisfactory performance attenuation analysis. Technical problems.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet.

[0005] Application Contents

[0006] The embodiments of the present application provide a method and device for determining a fuel cell attenuation state, and a fuel cell system, so as to at least solve the technical problem that the real-time attenuation state analysis effect of the fuel cell is not ideal.

[0007] According to one aspect of an embodiment of the present application, a method for determining a fuel cell attenuation state is provided, comprising: obtaining characteristic statistical data based on operating data of the fuel cell, wherein the characteristic statistical data is used to characterize performance changes during the operation of the fuel cell; determining clustering results of the characteristic statistical data; performing fitting according to the clustering results to obtain fitting results of the characteristic statistical data; and performing processing based on the fitting results to obtain a performance attenuation result of the fuel cell.

[0008] Optionally, multiple data points in the feature statistical data come from multiple Gaussian distributions, and determining the clustering results of the feature statistical data includes: using a Gaussian mixture model to determine the cluster centers in the feature statistical data based on the probabilities that the multiple data points respectively belong to corresponding Gaussian distributions; and performing clustering processing on the feature statistical data according to the cluster centers to obtain the clustering results.

[0009] Optionally, the method also includes: determining multiple candidate total numbers of the multiple Gaussian distributions; determining penalty function values ​​corresponding to the multiple candidate total numbers, respectively, wherein the penalty function values ​​are used to control the fit of the Gaussian mixture model; and determining the candidate total number with the smallest penalty function value among the multiple candidate total numbers as the target total number of the multiple Gaussian distributions.

[0010] Optionally, the fuel cell-based operating data obtains characteristic statistical data, including: using a time window of predetermined length to intercept the operating data to obtain window data; according to the window data, obtaining characteristic data corresponding to the predetermined characteristics respectively; using a predetermined statistical method to process the characteristic data to obtain the characteristic statistical data.

[0011] Optionally, before obtaining characteristic statistical data based on the fuel cell operation data, the method also includes: acquiring structured data of the fuel cell; performing data cleaning on the structured data based on subsystem calibration parameter information, hardware performance boundaries, detection instrument ranges, and physical response characteristics in the fuel cell to obtain cleaned data; and obtaining the operation data based on the cleaned data.

[0012] Optionally, obtaining the operating data based on the cleaning data includes: determining data features of the cleaning data and data missing points where data is missing; when the data features indicate discrete feature data, determining neighboring data points of the data missing points; based on the neighboring data points, performing gap filling processing on the data missing points to obtain gap-filled cleaning data as the operating data; or, when the data features indicate continuous feature data, performing regression processing on the cleaning data to obtain regression results; based on the regression results, performing gap filling processing on the data missing points to obtain gap-filled cleaning data as the operating data.

[0013] Optionally, obtaining the structured data of the fuel cell includes: receiving a communication message of the fuel cell; parsing the communication message to obtain initial data of the fuel cell; and classifying and storing the initial data according to the identification code, maintenance component information, and associated components of the fuel cell to obtain the structured data.

[0014] According to another aspect of an embodiment of the present application, a fuel cell decay state determination device is provided, comprising: a feature data acquisition module, for obtaining feature statistical data based on operating data of the fuel cell, wherein the feature statistical data is used to characterize performance changes during the operation of the fuel cell; a clustering result determination module, for determining clustering results of the feature statistical data; a fitting result acquisition module, for performing fitting according to the clustering results to obtain fitting results of the feature statistical data; and an attenuation result acquisition module, for performing processing based on the fitting results to obtain a performance attenuation result of the fuel cell.

[0015] According to another aspect of an embodiment of the present application, a fuel cell system is provided, comprising a detection unit that applies any one of the methods for determining a fuel cell attenuation state.

[0016] According to another aspect of an embodiment of the present application, a non-volatile storage medium is provided, wherein the non-volatile storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing any one of the methods for determining a fuel cell attenuation state.

[0017] In the embodiment of the present application, characteristic statistical data is obtained based on the operation data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance changes during the operation of the fuel cell; the clustering results of the characteristic statistical data are determined; fitting is performed according to the clustering results to obtain the fitting results of the characteristic statistical data; and processing is performed based on the fitting results to obtain the performance decay results of the fuel cell. The purpose of improving data utilization and real-time performance is achieved, and the technical effect of improving the real-time decay state analysis effect of the fuel cell is achieved, thereby solving the technical problem that the real-time decay state analysis effect of the fuel cell is not ideal. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 is a flow chart of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application;

[0020] Figure 2 is a first schematic diagram of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application;

[0021] Figure 3 is a second schematic diagram of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application;

[0022] Figure 4 It is a schematic diagram of an optional fuel cell decay state determination device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] According to an embodiment of the present application, an embodiment of a method for determining a fuel cell attenuation state is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0026] Figure 1 is a flow chart of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0027] Step S102, obtaining characteristic statistical data based on the operation data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance change during the operation of the fuel cell;

[0028] It can be understood that based on the operating data of the fuel cell, characteristic statistical data that can characterize the performance changes during the operation of the fuel cell are obtained, and the performance indicators of the fuel cell are monitored in real time through the characteristic statistical data to improve the accuracy of operation maintenance and attenuation state determination.

[0029] In an optional embodiment, characteristic statistical data is obtained based on the operating data of the fuel cell, including: using a time window of predetermined length to intercept the operating data to obtain window data; according to the window data, obtaining characteristic data corresponding to predetermined characteristics; using a predetermined statistical method to process the characteristic data to obtain characteristic statistical data.

[0030] It can be understood that a time window of a predetermined length is used to intercept the operation data to obtain window data, and characteristic data corresponding to predetermined characteristics are obtained based on the window data. The characteristic data is processed using a predetermined statistical method to obtain characteristic statistical data. By setting a time window, data can be collected and analyzed in real time to ensure that decisions or detection are based on the latest data, thereby improving the real-time response capability of the system and the accuracy of data analysis.

[0031] Optionally, the above-mentioned operating data can be of multiple types, for example, obtaining factory data information of the fuel cell system, continuously receiving real-time operating data of the fuel cell, and structuring the real-time operating data based on R&D and production data, wherein the real-time operating data includes the fuel cell output current, output voltage and a set of response output parameters of each subsystem.

[0032] In an optional embodiment, before obtaining characteristic statistical data based on the operating data of the fuel cell, the method also includes: acquiring structured data of the fuel cell; performing data cleaning on the structured data based on subsystem calibration parameter information, hardware performance boundaries, detection instrument ranges, and physical response characteristics in the fuel cell to obtain cleaned data; and obtaining operating data based on the cleaned data.

[0033] It can be understood that before obtaining characteristic statistical data, the structured data of the fuel cell is obtained. The above structured data can be of multiple types. For example, the structured reorganization tool of the operating data reorganizes the real-time operating data in the database based on the time series, product model, factory information characteristics and system maintenance records to form a structured data set. Based on the subsystem calibration parameter information, hardware performance boundaries, detection instrument range, and physical response characteristics in the fuel cell, the structured data is cleaned to obtain cleaned data, and finally the operating data is calculated. The data cleaning process can effectively remove noise, inconsistency and redundant data to ensure that the cleaned data used for analysis has high accuracy and integrity.

[0034] Optionally, the above data cleaning method can be multiple, for example, combining the subsystem component calibration information, the physical boundaries of the system output parameters based on expert knowledge, and the characteristic parameters such as the range, hysteresis and measurement accuracy of the system-side measuring instruments, to clean abnormal data that does not truly reflect the system operation status. Using data cleaning and completion tools, determine data cleaning rules based on hardware performance boundaries, physical response characteristics and output boundaries of sensing instrument equipment, clean abnormal data that does not truly reflect the system operation status, and complete it.

[0035] In an optional embodiment, operating data is obtained based on the cleaned data, including: determining data features of the cleaned data and data missing points where data is missing; when the data features indicate discrete feature data, determining neighboring data points of the data missing points; based on the neighboring data points, filling in the missing data points to obtain cleaned data after filling in the gaps as operating data; or, when the data features indicate continuous feature data, performing regression processing on the cleaned data to obtain regression results; based on the regression results, filling in the missing data points to obtain cleaned data after filling in the gaps as operating data.

[0036] It can be understood that the data features of the cleaned data are determined. The data features may be discrete feature data or continuous feature data, and there may be data missing points. When the data features are discrete feature data, the neighboring data points of the data missing points are determined, and the data missing points are filled with the neighboring data points. When the data features are continuous feature data, the data regression method is used to perform regression processing on the cleaned data to obtain the regression results, and the data missing points are filled with the regression results. The cleaned data after filling is obtained as the running data. By effectively filling in the missing data points, the integrity of the data set is ensured, and the analysis deviation and model performance degradation caused by data missing are avoided.

[0037] Optionally, the above data completion method can be multiple, for example, using neighboring values, linear regression and other methods to complete missing values. For discrete feature data of the operating status of the recording system, neighboring values ​​are used for completion, and for continuously changing feature data such as voltage, linear regression is used for completion.

[0038] In an optional embodiment, obtaining structured data of a fuel cell includes: receiving a communication message of the fuel cell; parsing the communication message to obtain initial data of the fuel cell; and classifying and storing the initial data according to the identification code of the fuel cell, maintenance component information, and associated components to obtain structured data.

[0039] It can be understood that by receiving and parsing the communication message of the fuel cell, converting it into initial data, and then finely classifying and storing the data according to the identification code of the fuel cell, maintenance component information and related components, structured data is formed. Real-time data collection and analysis are achieved, ensuring the timeliness of information, and through classified storage, rapid retrieval and analysis of data are facilitated.

[0040] Optionally, the above-mentioned methods for fine classification and storage of data can be multiple, for example, by obtaining the research and development and production process records of the scientific research data system and the production line data acquisition system, the static parameters of the fuel cell system generated in the research and development and production links are stored in a relational database; the communication message sent by the fuel cell system data transmission terminal in real time is received, and after verification based on the agreed rules, it is decoded into real-time operation data according to the established data analysis protocol and stored in the time series database. Through the unique identification code of the product and its components, the real-time operation data of the fuel cell system is associated with the replacement data of the research and development, production and after-sales maintenance parts, and the real-time operation data is marked and distinguished by model, material parameters, supplier information and component association.

[0041] Step S104, determining the clustering result of the feature statistical data;

[0042] It can be understood that based on the real-time operation data of the fuel cell, the clustering results of the characteristic statistical data are determined by using the aggregation calculation method. Through cluster analysis, the inherent structure and similarity groups in the data are automatically identified, and the potential patterns and abnormal behaviors of the operation status can be characterized.

[0043] Optionally, the above-mentioned aggregation calculation method can be multiple, for example, the windowed aggregation calculation tool for fuel cell real-time operation data performs aggregation calculation of general statistics of single cell voltage during the operation of the fuel cell system, including mean calculation, discrete point calculation and data abnormal mutation point calculation. For the continuous change process of fuel cell real-time operation data, the time window method is used for partition calculation, and the time window is slid along the time series direction and aggregated. The input parameters of the aggregation calculation include the loading condition, the inlet and outlet temperature and pressure of the hydrogen, air and water three chambers, the single-chip voltage of the fuel cell inspection, and the hydrogen concentration in the shell. The output parameters include the average value, deviation from the mean and standard deviation of parameters such as temperature, pressure, voltage and hydrogen concentration in each time window under the loading condition with a predetermined ranking (such as the top three) as the time length. The aggregation calculation results are written into the data table for retrieval, viewing and data analysis. The length of the time window can be set by itself. In this embodiment, the preferred window length is set to 1h (hour).

[0044] In an optional embodiment, multiple data points in the feature statistical data come from multiple Gaussian distributions, and the clustering results of the feature statistical data are determined, including: using a Gaussian mixture model to determine the cluster centers in the feature statistical data based on the probabilities that the multiple data points respectively belong to the corresponding Gaussian distributions; performing clustering processing on the feature statistical data according to the cluster centers to obtain clustering results.

[0045] It can be understood that the Gaussian mixture model is used to cluster the characteristic statistical data, and the cluster center is determined based on the probability that each data point belongs to a different Gaussian distribution to obtain the clustering result. By determining the cluster center, the robustness of the model and the reliability of the analysis are further improved, so that the system can realize refined fuel cell real-time attenuation analysis based on more accurate clustering results.

[0046] Optionally, in a Gaussian mixture model (GMM), each Gaussian component has a mean vector, a covariance matrix, and a mixture weight, so the total number of model parameters k is related to the number of Gaussian components c. The number of Gaussian mixture components c is the number of Gaussian distributions set in the GMM model, and the optimal c value is determined by minimizing the penalty function BIC_reconfig, selecting the number of Gaussian distributions that can best fit the data while avoiding overfitting.

[0047] In an optional embodiment, the method also includes: determining multiple candidate total numbers for multiple Gaussian distributions; determining penalty function values ​​corresponding to the multiple candidate total numbers, wherein the penalty function values ​​are used to control the fit of the Gaussian mixture model; and selecting the candidate total number with the smallest penalty function value among the multiple candidate total numbers as the target total number for the multiple Gaussian distributions.

[0048] It can be understood that the total number of multiple candidates for multiple Gaussian distributions is determined, and based on the penalty function values ​​corresponding to the multiple candidate total numbers, the total number of candidates with the smallest penalty function value is selected as the target total number of multiple Gaussian distributions. While ensuring that the model can capture the complex structure of the data, the penalty function is used to limit the complexity of the model to prevent the model from overfitting to noise or outliers.

[0049] Optionally, the above-mentioned penalty function can be multiple, for example, using a Gaussian mixture model, unsupervised clustering is performed on the aggregated statistical results of each type of feature parameters, and a penalty function is constructed based on the Bayesian Information Criterion in this process. The Bayesian Information Criterion (BIC) is a model selection criterion in statistics, which is used to measure the degree of fit of different models to the data, while penalizing the complexity of the model to avoid overfitting. In the Gaussian mixture model cluster analysis, the Bayesian Information Criterion is used to determine the optimal number of model parameters, such as the number of mixed components. In this embodiment, the penalty function is constructed as:

[0050]

[0051] Among them, BIC_reconfig represents the penalty function, k is the total number of model parameters (that is, the number of sub-models in the Gaussian mixture model), n is the number of samples, L is the likelihood function, Re is the correction coefficient based on the actual scenario, c is the number of Gaussian mixture components, and the optimal correction coefficient is 0.985. The c that makes the function take the minimum value is the optimal number of categories under this criterion, which is regarded as the total number of trends that exist in the evolution process of this type of characteristic parameters over time.

[0052] The correction coefficient Re in the BIC_reconfi formula is used to adjust the strength of the penalty term to ensure that the model does not overfit due to too many Gaussian components. In fuel cell decay analysis, overfitting can lead to false trend identification, thus affecting the accuracy of the analysis. The process of minimizing BIC_reconfi is actually a model selection process, which adjusts the c value to find the best Gaussian mixture model that can both fit the data well and maintain the simplicity of the model. Adaptive segmented fitting, after determining the optimal c value, divide the data into c segments according to the clustering results, and then perform linear fitting on each segment. The above processing can dynamically identify the decay trends of different stages, and can smoothly transition between different stages, avoiding the inadequacy of fitting a single model over the entire time range.

[0053] In the fuel cell real-time decay analysis system, the role of the c value is to determine the total number of trends of the characteristic parameters evolving over time. This helps to divide the data into different trend segments, so as to more accurately analyze and understand the decay process of the fuel cell. For example, if the c value is 3, it means that the system divides the aggregated statistical results of the characteristic parameters into 3 different trend segments, each of which may correspond to a different decay stage or state change.

[0054] By minimizing the penalty function BIC_reconfig, the optimal c value can be found to optimize the clustering results. The optimal c value can ensure that the clustering algorithm finds the most appropriate trend segment in the data and reduces the impact of noise and outliers. Specifically, the c value corresponding to the minimum value of the penalty function is the optimal number of categories, which helps to improve the accuracy and reliability of clustering.

[0055] It should be noted that the determination of the c value is adaptive, that is, the system can automatically adjust the number of trend segments according to the changes in actual data. This adaptability enables the system to maintain high analysis accuracy under different conditions (such as different operating conditions or different fuel cell models). By adaptively determining the c value, the system can more flexibly respond to various complex data features and improve the real-time and accuracy of data analysis.

[0056] Step S106, performing fitting according to the clustering result to obtain a fitting result of characteristic statistical data;

[0057] It can be understood that the fitting results of characteristic statistical data are obtained by fitting based on clustering. Through the technology of combining clustering and fitting, the inherent distribution pattern of the data can be accurately identified and described, the data characteristics of fuel cells under different operating conditions can be revealed, and more detailed and accurate data support can be provided for the real-time attenuation analysis of fuel cells.

[0058] Optionally, the above-mentioned fitting method can be multiple, for example, based on the windowed aggregation calculation results, high-order statistics calculation and trend fitting analysis of the system operation data aggregation calculation results are performed, including mean change trend fitting analysis and discrete change trend analysis, and Gaussian mixture model clustering and piecewise linear fitting are combined to adaptively complete multi-segment fitting of feature parameter aggregation calculation result data. Using the Gaussian mixture model, unsupervised clustering is performed on the aggregated statistical results of each type of feature parameter. In the process, a penalty function is constructed based on the Bayesian information criterion to determine the optimal number of categories that meet the criterion. This number is the total number of trends that exist in the evolution process of this type of feature parameter over time. According to the clustering results and the total number of trends, each type of feature data pair is piecewise linearly fitted. The above two algorithms are used to adaptively complete multi-segment fitting of aggregated statistical data, and finally the statistical values ​​and fitting analysis results are stored in the database by the regular update program.

[0059] It should be noted that the attenuation process of fuel cells is not linear and may show different attenuation rates and patterns due to different operating conditions, maintenance conditions and other factors. The multi-segment fitting method can better capture these nonlinear changes, improve the accuracy of fitting, and make the data analysis results more accurately reflect the actual situation. Through the combination of Gaussian mixture model (GMM) clustering and piecewise linear fitting, the system can adaptively identify the trend change points of characteristic parameters (i.e., attenuation inflection points) in different operating stages. Accurately identifying these change points helps to promptly detect abnormal attenuation or accelerated attenuation, providing an important basis for fault diagnosis and performance optimization.

[0060] Compared with a single fitting method, multi-segment fitting can provide more detailed analysis results, showing the attenuation characteristics of different operating stages, which helps to gain a deeper understanding of the performance changes of fuel cells under different conditions and provide support for subsequent design optimization and maintenance strategies. The multi-segment fitting method can better handle complex nonlinear data patterns and improve the robustness of data analysis by dividing the data into multiple sub-segments and fitting them separately.

[0061] Step S108: Processing is performed based on the fitting result to obtain the performance attenuation result of the fuel cell.

[0062] It can be understood that the performance degradation results of the fuel cell are obtained based on the fitting results. In-depth analysis of the fitting results and identification of performance degradation patterns and trends can provide early warning of potential performance degradation.

[0063] Optionally, unsupervised cluster analysis of feature statistics is performed using a Gaussian mixture model (GMM), through which data points can be grouped into different Gaussian distributions, so that patterns of different states or stages in the data can be identified, such as normal operation, slight decay, severe decay, etc. The clustering results are determined according to the Bayesian Information Criterion (BIC), and the optimal number of Gaussian components is selected based on the clustering results to ensure a balance between the complexity of the model and the accuracy of data fitting.

[0064] The system performs data analysis based on the fitting results to identify the starting point, acceleration point, and stabilization point of performance degradation, as well as other important performance indicator changes. These analysis results will be used to evaluate the health status of the fuel cell, predict the remaining service life, and provide a basis for the maintenance and optimization of the fuel cell. The visual display of performance degradation results, such as through line graphs or bar graphs, makes it easy to intuitively understand the changing trend of battery performance over time.

[0065] Through the above step S102, characteristic statistical data is obtained based on the operation data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance changes during the operation of the fuel cell; step S104, the clustering results of the characteristic statistical data are determined; step S106, fitting is performed according to the clustering results to obtain the fitting results of the characteristic statistical data; step S108, processing is performed based on the fitting results to obtain the performance decay results of the fuel cell. The purpose of improving data utilization and real-time performance can be achieved, and the technical effect of improving the real-time decay state analysis effect of the fuel cell is achieved, thereby solving the technical problem that the real-time decay state analysis effect of the fuel cell is not ideal.

[0066] Based on the above embodiments and optional embodiments, the present application proposes an optional implementation mode: Figure 2 is a first schematic diagram of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application, such as Figure 2 As shown in the figure, the system consists of four functional modules: integrated data storage module, data cleaning and completion module, real-time data analysis module and data visualization module. The integrated data storage module stores R&D and production data, receives and stores real-time fuel cell operation data, and establishes the correlation between R&D, production, operation and after-sales data; the data cleaning and completion module cleans the real-time data based on the performance range of components, the physical boundaries of characteristic parameters and the measurement capabilities of instruments, and completes the missing values ​​based on the data characteristics; the real-time data analysis module integrates a variety of data aggregation methods and data statistical tools; the data visualization module includes the front-end display page, data retrieval tools and visualization charts.

[0067] Step S1, obtain the factory data information of the fuel cell system, continuously receive the real-time operation data of the fuel cell, and structure the real-time operation data based on the research and development and production data, wherein the real-time operation data includes the fuel cell output current, output voltage and the response output parameter set of each subsystem.

[0068] The integrated data storage module is used to store the parameters of each subsystem of the fuel cell system, the design parameters of the fuel cell stack and the factory parameters, and to receive and store the real-time operation data sent by the fuel cell system; the structured reorganization tool of the operation data reorganizes the real-time operation data in the database based on the time sequence, product model, factory information characteristics and system maintenance records to form a structured data set.

[0069] Based on the construction conditions of the scientific research data system and the production line data acquisition system, the static parameters of the fuel cell system generated during the R&D and production data process are stored in the integrated data storage module; the communication message is sent through the fuel cell system transmission terminal, and the real-time operation data is stored in the module after verification, analysis and matching by the data platform. Through the unique identification code of the product and its components, the blood relationship between the R&D and production data and the operation data of the fuel cell system is established. Combined with the system maintenance and component replacement records, based on the time series relationship of the real-time data, the real-time operation data is labeled and distinguished by model, material parameters, usage time, supplier information and component association, and the structured organization of the real-time data based on association is completed.

[0070] Step S2, data cleaning and completion based on operating data thresholds and trends; data cleaning and completion tools determine data cleaning rules based on hardware performance boundaries, physical response characteristics, and output boundaries of sensing instrument equipment, clean abnormal data that does not truly reflect the system operating status, and complete the data.

[0071] Step S3, aggregating multiple statistics of the fuel cell system and high-order statistics information in real time by using a fuel cell real-time operation data aggregation calculation method;

[0072] The windowed aggregation calculation tool of the real-time operation data of the fuel cell is used to perform aggregation calculations of general statistics of the single cell voltage during the operation of the fuel cell system, including mean calculation, discrete point calculation and data abnormal mutation point calculation. Based on the windowed aggregation calculation results, high-order statistical calculations and trend fitting analysis of the system operation data aggregation calculation results are performed, including mean change trend fitting analysis and discreteness change trend analysis. The Gaussian mixture model clustering and piecewise linear fitting are combined to adaptively complete the multi-segment fitting of the characteristic parameter aggregation calculation result data.

[0073] Clustering can separate different trends or patterns in the data set, so that in the subsequent piecewise linear fitting, each clustering result can be fitted separately, avoiding the problem of poor fitting effect caused by mixing data with different trends. Clustering can automatically identify the inflection points of performance decay in the data, that is, the key turning points of performance change. These inflection points correspond to different stages of fuel cell decay. Through clustering, the inflection points can be identified more accurately, which is conducive to analyzing and predicting the decay process of fuel cells. The piecewise linear fitting is performed within each cluster cluster, which can more accurately capture the decay characteristics of different stages. Compared with a single linear fitting, the piecewise linear fitting can better adapt to the nonlinear changes and multi-stage characteristics of the data, thereby improving the accuracy of the fitting model. The operating data of the fuel cell may be affected by many factors, resulting in uneven data distribution and multimodality. The method of clustering first and then fitting can effectively deal with this variability and provide a more reliable basis for subsequent decay analysis and fault diagnosis. The above-mentioned method of clustering first and then fitting can improve the accuracy and efficiency of data processing, provide a more powerful tool for real-time decay analysis of fuel cells, and help to more accurately diagnose the health status of fuel cells and predict their life.

[0074] Figure 3 is a second schematic diagram of an optional method for determining a fuel cell attenuation state according to an embodiment of the present application, such as Figure 3 As shown, the algorithm flow for running data decay analysis includes integrated data storage, data cleaning and completion, real-time data analysis and data visualization.

[0075] Integrated data storage includes: by acquiring the research and development and production process records of the scientific research data system and the production line data acquisition system, the static parameters of the fuel cell system generated in the research and development and production links are stored in the relational database; receiving the communication messages sent by the fuel cell system data transmission terminal in real time, verifying them based on the agreed rules, and decoding them into real-time operation data according to the established data analysis protocol and storing them in the time series database. Through the unique identification code of the product and its components, the real-time operation data of the fuel cell system is associated with the replacement data of the research and development, production and after-sales maintenance parts, and the real-time operation data is marked and distinguished by model, material parameters, supplier information and component association.

[0076] Data cleaning and completion include: cleaning abnormal data that does not truly reflect the system's operating status based on the calibration parameter information of subsystem components, the physical boundaries of system output parameters, and characteristic parameters such as the range, hysteresis characteristics, and measurement accuracy of system-side measuring instruments. Taking the voltage of a single cell as an example, the theoretical maximum open-circuit voltage of a proton exchange membrane fuel cell is 1.23V (volts), and based on this, data outside the range of -1.23V to 1.23V is removed; missing values ​​are completed based on data features, and the neighboring value method is used to complete discrete feature data that reflects the system's operating status, such as engine status, subsystem status, and control command issuance. For continuously changing feature data such as temperature, gas pressure, hydrogen concentration, and voltage, linear regression is used to complete them.

[0077] Real-time data analysis includes: for the continuous change process of the real-time operation data of the fuel cell, partition calculation is performed in the form of time windows, the time window is slid along the time series direction and aggregated, the input parameters of the aggregate calculation include the loading condition, the inlet and outlet temperature and pressure of the three chambers of hydrogen, air and water, the single-chip voltage of the fuel cell inspection, and the hydrogen concentration in the shell, and the output parameters include the average value, deviation from the mean and standard deviation of parameters such as temperature, pressure, voltage and hydrogen concentration in each time window under the loading conditions with the top three durations, and the aggregate calculation results are written into the data table for retrieval and data analysis. Optionally, the time window length can be set by itself. In this embodiment, the window length is set to 1h. A Gaussian mixture model is used to perform unsupervised clustering on the aggregated statistical results of each type of characteristic parameters. In this process, a penalty function is constructed based on the Bayesian information criterion. In this embodiment, the penalty function is constructed as follows:

[0078]

[0079] Among them, BIC_reconfig represents the penalty function, k is the total number of model parameters, n is the number of samples, L is the likelihood function, Re is the correction coefficient based on the actual scenario, c is the number of Gaussian mixture components, and the optimal correction coefficient in this embodiment is 0.985. The c that makes the function take the minimum value is the optimal number of categories under this criterion, which is regarded as the total number of trends that exist in the evolution process of this type of feature parameters over time. A piecewise linear regression model is used to perform piecewise linear fitting on each type of feature data according to the clustering results and the total number of trends. Based on the linkage mode of the two algorithms, multi-segment fitting of aggregated statistical data is adaptively completed, and the statistical values ​​and fitting calculation results are stored in the database statistical analysis table through a timed update program.

[0080] Data visualization includes: developing front-end interactive pages, embedding SQL (Structured Query Language) functions and query statements into visualization modules to form interactions between visualization pages and databases. Business personnel can filter characteristic parameters based on equipment number, factory parameter characteristics, operating condition range, and operating time. Embedded query statements filter from database production record tables, real-time data tables, after-sales operation and maintenance record tables, and statistical analysis tables based on the filtering conditions, extract corresponding data that meets the conditions, and visualize the current status of the fuel-electric system and fuel cell and the attenuation evolution trend during the operating cycle in the front-end display interface through scatter plots, line graphs, and bar graphs. Charts and statistical data support downloading and exporting.

[0081] Step S4, based on the real-time operation data of the fuel cell and its aggregated statistical characteristics, determine the attenuation of the fuel cell under different states and display them visually.

[0082] The system operation attenuation trend is visualized on the front-end page in the form of statistical charts. The front-end display page has the function of filtering according to the battery stack model, factory parameter characteristics, operating conditions and operating time, and outputs statistical charts based on the filtering conditions for visual display and export.

[0083] Develop front-end interactive pages. By embedding SQL functions into visualization modules, business personnel can filter characteristic parameters based on equipment number, factory parameter characteristics, operating condition range, and operating time. Embedded statements extract corresponding data from the database statistical data table based on the filtering conditions. The current status of the fuel-electric system and fuel cell and the attenuation evolution trend during the operating cycle can be visualized on the front-end interface through scatter plots, line graphs, and bar graphs. Charts and statistical data can be downloaded and exported.

[0084] The above optional implementation achieves at least the following effects: constructs an adaptive fuel cell attenuation trend fitting analysis method, adaptively finds the inflection point of the characteristic parameter trend and fits it through the Gaussian mixture model clustering algorithm and the piecewise linear fitting algorithm, and displays the parameter change trend in an intuitive screening manner through a visual interface, quickly locates the time when abnormal attenuation or accelerated attenuation occurs, provides efficiency support for R&D and after-sales guarantee, and highlights the value of data.

[0085] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0086] A fuel cell system is also provided in an embodiment of the present application. The fuel cell system provided in the embodiment of the present application is introduced below.

[0087] The system includes a detection unit of any optional fuel cell attenuation state determination method provided in the above embodiments. Through the above processing, an integrated fuel cell data integrated collection, storage, analysis and visualization analysis system is constructed to collect the operation data of the fuel cell and its subsystems, and based on the physical characteristics of the fuel cell and the response characteristics of the data acquisition device, a generalized data cleaning method is formed to improve the data quality.

[0088] The purpose of improving data utilization and real-time performance was achieved, and the technical effect of improving the real-time attenuation state analysis effect of the fuel cell was realized, thereby solving the technical problem of unsatisfactory real-time attenuation state analysis effect of the fuel cell.

[0089] In this embodiment, a fuel cell decay state determination device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the terms "module" and "device" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0090] According to an embodiment of the present application, a device embodiment for implementing a method for determining a fuel cell attenuation state is also provided. Figure 4 is a schematic diagram of an optional fuel cell decay state determination device provided according to an embodiment of the present application, such as Figure 4 As shown, the above fuel cell decay state determination device includes a feature data acquisition module 402, a clustering result determination module 404, a fitting result acquisition module 406, and a decay result acquisition module 408. The device is described below.

[0091] A characteristic data acquisition module 402 is used to obtain characteristic statistical data based on the operation data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance change during the operation of the fuel cell;

[0092] A clustering result determination module 404, connected to the feature data acquisition module 402, is used to determine the clustering result of the feature statistical data;

[0093] The fitting result acquisition module 406 is connected to the clustering result determination module 404 and is used to perform fitting according to the clustering result to obtain the fitting result of the characteristic statistical data;

[0094] The attenuation result acquisition module 408 is connected to the fitting result acquisition module 406 and is used to process based on the fitting result to obtain the performance attenuation result of the fuel cell.

[0095] In a fuel cell decay state determination device provided by an embodiment of the present application, a feature data acquisition module 402 is provided to obtain feature statistical data based on the operation data of the fuel cell, wherein the feature statistical data is used to characterize the performance changes during the operation of the fuel cell; a clustering result determination module 404 is connected to the feature data acquisition module 402 and is used to determine the clustering results of the feature statistical data; a fitting result acquisition module 406 is connected to the clustering result determination module 404 and is used to perform fitting according to the clustering results to obtain the fitting results of the feature statistical data; and a decay result acquisition module 408 is connected to the fitting result acquisition module 406 and is used to perform processing based on the fitting results to obtain the performance decay results of the fuel cell. The purpose of improving data utilization and real-time performance is achieved, and the technical effect of improving the real-time decay state analysis effect of the fuel cell is achieved, thereby solving the technical problem that the real-time decay state analysis effect of the fuel cell is not ideal.

[0096] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0097] It should be noted that the feature data acquisition module 402, clustering result determination module 404, fitting result acquisition module 406, and attenuation result acquisition module 408 correspond to steps S102 to S108 in the embodiment, and the examples and application scenarios implemented by the modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments. It should be noted that the modules as part of the device can be run in a computer terminal.

[0098] It should be noted that the optional or preferred implementation of this embodiment can refer to the relevant description in the embodiment, which will not be repeated here.

[0099] The above-mentioned fuel cell decay state determination device may also include a processor and a memory. The characteristic data acquisition module 402, the clustering result determination module 404, the fitting result acquisition module 406, the decay result acquisition module 408, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions.

[0100] The processor includes a kernel, which retrieves the corresponding program unit from the memory. There can be one or more kernels. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0101] An embodiment of the present application provides a non-volatile storage medium on which a program is stored. When the program is executed by a processor, a method for determining a fuel cell attenuation state is implemented.

[0102] The embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program stored in the memory and can be run on the processor. When the processor executes the program, the following steps are implemented: based on the operation data of the fuel cell, characteristic statistical data is obtained, wherein the characteristic statistical data is used to characterize the performance change during the operation of the fuel cell; the clustering result of the characteristic statistical data is determined; fitting is performed according to the clustering result to obtain the fitting result of the characteristic statistical data; processing is performed based on the fitting result to obtain the performance attenuation result of the fuel cell. The device in this article can be a server, a PC, etc.

[0103] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that is initialized with the following method steps: obtaining characteristic statistical data based on the operating data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance changes during the operation of the fuel cell; determining the clustering results of the characteristic statistical data; performing fitting according to the clustering results to obtain the fitting results of the characteristic statistical data; and performing processing based on the fitting results to obtain the performance attenuation results of the fuel cell.

[0104] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0105] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0106] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0107] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0108] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0109] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0110] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0111] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0112] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0113] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for determining a fuel cell attenuation state, characterized in that: include: Based on the operation data of the fuel cell, characteristic statistical data are obtained, wherein the characteristic statistical data are used to characterize the performance change during the operation of the fuel cell; Determining clustering results of the feature statistical data; Perform fitting according to the clustering result to obtain the fitting result of the characteristic statistical data; Processing is performed based on the fitting result to obtain the performance attenuation result of the fuel cell.

2. The method according to claim 1, characterized in that The plurality of data points in the feature statistical data come from a plurality of Gaussian distributions, and the determining of a clustering result of the feature statistical data comprises: Using a Gaussian mixture model, based on the probability that the plurality of data points belong to corresponding Gaussian distributions, a cluster center in the feature statistical data is determined; According to the cluster center, clustering processing is performed on the feature statistical data to obtain the clustering result.

3. The method according to claim 2, characterized in that The method further comprises: Determining a total number of multiple candidates for the multiple Gaussian distributions; Determine penalty function values ​​corresponding to the total number of the multiple candidates, respectively, wherein the penalty function values ​​are used to control the fit of the Gaussian mixture model; The candidate total number with the smallest penalty function value among the multiple candidate total numbers is used as the target total numbers of the multiple Gaussian distributions.

4. The method according to claim 1, characterized in that The characteristic statistical data is obtained based on the operation data of the fuel cell, including: Using a time window of a predetermined length, intercepting the operation data to obtain window data; According to the window data, obtaining feature data corresponding to the predetermined features respectively; The characteristic data are processed using a predetermined statistical method to obtain the characteristic statistical data.

5. The method according to claim 1, characterized in that Before obtaining characteristic statistical data based on the operation data of the fuel cell, the method further includes: Acquiring structured data of the fuel cell; Based on the subsystem calibration parameter information, hardware performance boundary, detection instrument range, and physical response characteristics in the fuel cell, the structured data is cleaned to obtain cleaned data; The operation data is obtained based on the cleaning data.

6. The method according to claim 5, characterized in that The obtaining the operation data based on the cleaning data includes: Determining data features of the cleaned data and data missing points where data is missing; In the case where the data feature indication is discrete feature data, determining the neighboring data points of the data missing point; based on the neighboring data points, performing gap filling processing on the data missing point to obtain the gap-filled cleaned data as the operating data; or, In the case where the data feature indicates continuous feature data, regression processing is performed on the cleaned data to obtain a regression result; based on the regression result, the data missing points are filled to obtain the cleaned data after filling as the operation data.

7. The method according to claim 5, characterized in that The obtaining of structured data of the fuel cell comprises: Receiving a communication message from the fuel cell; Parsing the communication message to obtain initial data of the fuel cell; The initial data is classified and stored according to the identification code of the fuel cell, maintenance component information, and associated components to obtain the structured data.

8. A fuel cell system, characterized in that: A detection unit comprising the method for determining the attenuation state of a fuel cell according to any one of claims 1 to 7.

9. A fuel cell decay state determination device, characterized in that: include: A characteristic data acquisition module, used to obtain characteristic statistical data based on the operation data of the fuel cell, wherein the characteristic statistical data is used to characterize the performance change during the operation of the fuel cell; A clustering result determination module, used to determine the clustering result of the feature statistical data; A fitting result acquisition module, used for performing fitting according to the clustering result to obtain the fitting result of the characteristic statistical data; The attenuation result acquisition module is used to process based on the fitting result to obtain the performance attenuation result of the fuel cell.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the fuel cell decay state determination method according to any one of claims 1 to 7.

Images