A test method for determining the dryness and humidity within a fuel cell stack based on inspection sampling

Abstract

The present invention relates to the field of fuel cell stack testing, and in particular to a test method for a fuel cell stack that determines the dryness and wetness of the surface based on patrol sampling. The method comprises the following steps: installing automatically adjustable patrol devices at corresponding patrol points to collect data; establishing a response strategy model by screening out features that contribute to the generation of a response strategy, and generating a specific response strategy based on the prediction or classification results of the response strategy model; establishing a prediction model for the wetness of the surface layer inside a PEMFC stack; and the above-mentioned prediction model is used to evaluate the output performance of the PEMFC stack under different operating conditions, and to obtain possible fault types and their development trends. The present invention achieves accurate judgment of the wetness and dryness of the surface layer inside a fuel cell stack, and provides detailed data support for operators to make corresponding adjustments in a timely manner, thereby optimizing the performance and life of the stack. This helps to improve the performance monitoring and fault prevention capabilities of the fuel cell stack, and ensure that the stack continues to operate in an optimal state.

Description

Technical Field

[0001] The present invention relates to the field of fuel cell stack testing, and in particular to a fuel cell stack testing method for determining surface dryness and wetness based on inspection sampling. Background Art

[0002] When a fuel cell stack is operating, proper humidity maintains the normal chemical reactions within the cell, thereby extending its lifespan. Excessively high humidity can easily cause changes in the battery's internal electrolytes, which can accelerate the aging process and shorten its lifespan. Conversely, excessively low humidity can cause electrolyte dehydration, similarly accelerating battery aging. Therefore, testing the humidity and dryness of fuel cell stacks is essential. However, in actual testing, it is difficult to accurately measure the humidity and dryness of the internal surface layer of a fuel cell stack, resulting in insufficient performance monitoring and fault prevention capabilities for existing stacks. Summary of the Invention

[0003] In response to the problems existing in the background technology, a test method for fuel cell stacks is proposed to determine the dryness and wetness of the surface based on inspection sampling. By installing automatically adjustable inspection devices on the four corners of the dye cell stack, the dryness and wetness of the internal surface layer of the fuel cell stack can be accurately determined, providing detailed data support for operators to make corresponding adjustments in a timely manner, thereby optimizing the performance and life of the fuel cell stack.

[0004] The present invention proposes a test method for a fuel cell stack to determine the dryness and humidity within the surface based on inspection sampling, the steps are as follows:

[0005] S1. First, select four inspection points on the dye battery stack and install the automatically adjustable inspection device at the corresponding inspection points to collect data from each battery in the stack;

[0006] S2. Then, the data is processed to establish a response strategy model by screening out features that contribute to the generation of response strategies, and a specific response strategy is generated based on the prediction or classification results of the response strategy model;

[0007] S3. Finally, based on experimental data and theoretical analysis, a prediction model for the wetness and dryness of the internal surface layer of the PEMFC stack is established; the above prediction model is used to evaluate the output performance of the PEMFC stack under different operating conditions and to obtain possible fault types and their development trends.

[0008] Preferably, in S1, laser positioning technology is used to measure the fuel cell stack, four inspection points are selected according to the actual structure of the stack, and an automatically adjustable inspection device is installed.

[0009] Preferably, the four inspection points are located at four vertex corners, including inspection group one, inspection group two, inspection group three and inspection group four.

[0010] Preferably, the specific steps of S2 are as follows:

[0011] S21, through data processing, merge all data into one data set;

[0012] S22. Filter out features that contribute to response strategy generation from the raw data;

[0013] S23. Select an AI algorithm model based on business needs and data characteristics, use historical data as a training dataset, and train the selected model;

[0014] S24, adjusting model parameters to optimize model performance;

[0015] S25. Design a response strategy template based on business needs, predict or classify real-time data, and generate a specific response strategy based on the prediction or classification results;

[0016] S26. Evaluate the generated response strategy using a test data set or simulation environment to verify its effectiveness and accuracy, and adjust and optimize the response strategy based on the evaluation results;

[0017] S27. Deploy the optimized response strategy to the actual business system;

[0018] S28. Monitor the system operation in real time and continuously adjust and optimize the response strategy based on the feedback results;

[0019] S29. Use the generated dynamic response strategy to load, and combine the dynamic changes of the sampling data at the four inspection points to immediately process and analyze the data collected by the inspection device; once abnormal data is found, immediately trigger the early warning mechanism, and automatically adjust the loading strategy or take other remedial measures.

[0020] Preferably, the specific steps of S3 are as follows:

[0021] S31. Design the experiment according to the research purpose;

[0022] S32. Collect key parameter data of the PEMFC stack under various operating conditions through specialized sensors and measurement equipment;

[0023] S33, establishing the VI curve of the PEMFC stack, analyzing the VI curve, and extracting key parameters;

[0024] S34. Use electrochemical impedance spectroscopy to measure the impedance characteristics of the battery stack at different frequencies and extract the response signals at key frequency points;

[0025] S35. Analyze the variation of VI curve with current, temperature and humidity, and evaluate the output performance of the battery stack under different conditions;

[0026] S36. Based on the data in S34, draw a frequency impedance spectrum characteristic curve, analyze the changes in parameters, and determine the changes in water content in the proton exchange membrane;

[0027] S37. Based on experimental data and theoretical analysis, a prediction model for the wetness and dryness of the internal surface layer of the PEMFC stack was established;

[0028] S38. Based on the data analysis results, explain the wet and dry status of the internal surface layer of the PEMFC stack, evaluate the output performance of the PEMFC stack under different operating conditions, and obtain the possible fault types and their development trends.

[0029] Compared with the existing technology, the present invention has the following beneficial technical effects: First, automatically adjustable inspection devices are installed at the four corners of the fuel cell stack, ensuring maximum accuracy and comprehensiveness of data collection. This enables the four-terminal inspection sampling to capture the voltage information of each cell in the stack with high precision, and also ensures consistent data collection accuracy for all inspection devices. Next, the data is processed, and a response strategy model is established by screening out features that contribute to the generation of the response strategy. A specific response strategy is generated based on the prediction or classification results of the response strategy model. Finally, based on experimental data and theoretical analysis, a prediction model for the wetness of the internal surface layer of the PEMFC stack is established. This prediction model is used to evaluate the output performance of the PEMFC stack under different operating conditions and to determine the possible types of faults and their development trends. This allows for accurate judgment of the wetness of the internal surface layer of the fuel cell stack, providing operators with detailed data support so that they can make corresponding adjustments in a timely manner, thereby optimizing the performance and life of the stack. This helps to improve the performance monitoring and fault prevention capabilities of the fuel cell stack, ensuring that the stack continues to operate in an optimal state. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Schematic diagram of the installation position of the inspection device in the present invention.

[0031] Figure numerals: 1, inspection group one; 2, inspection group two; 3, inspection group three; 4, inspection group four. DETAILED DESCRIPTION

[0032] The present invention proposes a test method for determining the dryness and humidity within a fuel cell stack based on inspection sampling, the steps of which are as follows:

[0033] S1. First, select four inspection points on the dye battery stack and install the automatically adjustable inspection device at the corresponding inspection points to collect data from each battery in the stack;

[0034] It should be further explained that in S1, laser positioning technology is used to measure the fuel cell stack, four inspection points are selected according to the actual structure of the stack, and an automatically adjustable inspection device is installed.

[0035] It needs to be further explained that, Figure 1 As shown, the four inspection points are located at the four corners, including inspection group 1 1, inspection group 2 2, inspection group 3 3 and inspection group 4 4.

[0036] In addition to collecting voltage, the inspection device can also monitor temperature, humidity, and gas flow, and automatically perform a calibration procedure before each test. This layout strategy ensures maximum accuracy and comprehensiveness of data collection, enabling four-terminal inspection sampling to capture the voltage information of each battery in the battery stack with high precision and ensuring consistent data collection accuracy for all inspection devices.

[0037] S2. Then, the data is processed to establish a response strategy model by screening out features that contribute to the generation of response strategies, and a specific response strategy is generated based on the prediction or classification results of the response strategy model;

[0038] It should be further explained that the specific steps of S2 are as follows:

[0039] S21. Combine historical test data based on requirements and perform data processing, such as data comparison, deleting duplicate data, and handling missing values ​​using interpolation or default value filling. Merge all data into a single dataset to ensure data integrity and consistency.

[0040] S22. Filter out features that contribute to the generation of response strategies from the raw data, such as the performance of the fuel cell stack and the degree of influence;

[0041] S23. Based on business needs and data characteristics, select appropriate AI algorithm models, such as decision trees, random forests, and neural networks in machine learning, or recurrent neural network models such as LSTM and GRU in deep learning. Use historical data as the training dataset to train the selected model.

[0042] S24, adjusting model parameters to optimize model performance;

[0043] S25. Design a response strategy template based on business needs, including response conditions and actions. Leverage the trained model to predict or classify real-time data and generate a specific response strategy based on the prediction or classification results. The response strategy may include various measures such as resource allocation, load balancing, and fault recovery.

[0044] S26. Evaluate the generated response strategy using a test data set or simulation environment to verify its effectiveness and accuracy. Based on the evaluation results, adjust and optimize the response strategy to improve its adaptability and effectiveness.

[0045] S27. Deploy the optimized response strategy to the actual business system to ensure that the system can automatically execute the corresponding response actions according to the strategy;

[0046] S28. Monitor the system operation in real time and continuously adjust and optimize the response strategy based on the feedback results;

[0047] S29. Use the generated dynamic response strategy to load, and combine the dynamic changes of the sampling data at the four inspection points to immediately process and analyze the data collected by the inspection device; once abnormal data is found (such as voltage drop, abnormal temperature increase, etc.), immediately trigger the early warning mechanism, and automatically adjust the loading strategy (adjust the operating parameters) or take other remedial measures to ensure the safety and effectiveness of the test process, so as to facilitate subsequent in-depth analysis.

[0048] S3. Finally, based on experimental data and theoretical analysis, a prediction model for the wetness and dryness of the internal surface layer of the PEMFC stack is established; the above prediction model is used to evaluate the output performance of the PEMFC stack under different operating conditions and to obtain possible fault types and their development trends.

[0049] It should be further explained that the specific steps of S3 are as follows:

[0050] S31. Design experiments based on the research objectives, including experiments under different current, temperature, gas humidification relative humidity and other conditions;

[0051] S32. Use specialized sensors and measuring equipment to collect data on key parameters of the PEMFC stack, such as voltage, current, temperature, and humidity, under various operating conditions. Ensure data accuracy and completeness, and record experimental conditions and data collection time in detail. Check and remove outliers, missing values, or erroneous data to ensure data reliability.

[0052] S33, such as converting the time series data into frequency domain data, establishing a voltage-current (VI) curve of the PEMFC stack, analyzing the voltage-current (VI) curve of the PEMFC stack, and extracting key parameters such as open circuit voltage and maximum power point;

[0053] S34. Use electrochemical impedance spectroscopy to measure the impedance characteristics of the battery stack at different frequencies and extract the response signals at key frequency points such as high frequency and transition frequency;

[0054] S35. Analyze the variation of VI curve with current, temperature and humidity, and evaluate the output performance of the battery stack under different conditions;

[0055] S36. Based on the data in S34, draw a frequency impedance spectrum characteristic curve, analyze the changes in parameters such as the secant angle a, and determine the changes in the water content in the proton exchange membrane;

[0056] S37. Based on experimental data and theoretical analysis, a prediction model for the wetness and dryness of the internal surface layer of the PEMFC stack was established;

[0057] S38. Based on the data analysis results, explain the dryness and wetness status of the internal surface layer of the PEMFC stack, including failure phenomena such as membrane drying and water flooding. Evaluate the output performance of the PEMFC stack under different operating conditions, including key indicators such as voltage, power, and efficiency, to determine the possible types of failures and their development trends. When evaluating the dryness and wetness status of a cell stack (especially a fuel cell stack), calculating the deviation (difference from the mean) of the voltage of each cell from the average voltage, the weighted deviation, and the standard deviation can provide us with important information about the performance consistency and health of the cell stack.

[0058] Weighted Deviation: Considering that different battery locations or performance may have different impacts on overall performance, the deviation of each battery cell can be assigned different weights. The weights can be set based on the battery's position in the stack, historical performance data, or expert experience.

[0059] Weighted Bias Among them, V i is the voltage of the ith battery, is the average value of all battery voltages, w i is the weight of the i-th battery.

[0060] Standard Deviation: To gain a more comprehensive understanding of the dispersion of the voltage distribution, the standard deviation of the voltage data can be calculated. Although this is not directly applicable to the deviation calculation of each cell, it provides important information for understanding the stability of the voltage of the entire stack.

[0061] When determining the wetness or dryness of a battery stack, the weighted deviation is typically not used alone (because it is the same as the deviation from the mean when weighted equally). Instead, the deviation from the mean and standard deviation are used in combination. This is because the deviation from the mean can help identify cells whose voltages deviate significantly from the average, potentially indicating localized over-wetting or over-drying. The standard deviation provides an overall measure of the voltage distribution across the entire battery stack. A high standard deviation may indicate significant inconsistency within the battery stack, which may also be related to the wetness or dryness of the battery stack.

[0062]

[0063] A reasonable threshold needs to be set based on experimental data, battery stack specifications and performance requirements. For example, a maximum allowable standard deviation σ can be set. max And a maximum allowable deviation from the mean absolute value |Deviationmax|. If the absolute value of the deviation from the mean of any battery exceeds |Deviationmax|, or the standard deviation of the entire battery stack exceeds σ max , it can be determined that there may be a problem with the dry and wet status of the battery stack.

[0064] Comprehensive judgment: In some cases, the combined impact of the two may need to be considered. For example, even if the standard deviation is within an acceptable range, if the mean deviation of multiple cells approaches or exceeds the threshold, it may indicate that the dryness and wetness of the battery stack need to be adjusted.

[0065] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A fuel cell stack testing method for determining surface dryness and wetness based on inspection sampling, characterized in that: Here are the steps: S1. First, select four inspection points on the fuel cell stack and install the automatically adjustable inspection device at the corresponding inspection points to collect data from each battery in the fuel cell stack; S2. Then, the data is processed to establish a response strategy model by screening out features that contribute to the generation of response strategies, and a specific response strategy is generated based on the prediction or classification results of the response strategy model; The specific steps of S2 are as follows: S21, through data processing, merge all data into one data set; S22. Filter out features that contribute to response strategy generation from the raw data; S23. Select an AI algorithm model based on business needs and data characteristics, use historical data as a training dataset, and train the selected model; S24, adjusting model parameters to optimize model performance; S25. Design a response strategy template based on business needs, predict or classify real-time data, and generate a specific response strategy based on the prediction or classification results; S26. Evaluate the generated response strategy using a test data set or simulation environment to verify its effectiveness and accuracy, and adjust and optimize the response strategy based on the evaluation results; S27. Deploy the optimized response strategy to the actual business system; S28. Monitor the system operation in real time and continuously adjust and optimize the response strategy based on the feedback results; S29, using the generated dynamic response strategy to load, combining the dynamic changes of the sampling data at the four inspection points, and immediately processing and analyzing the data collected by the inspection device; Once abnormal data is detected, the early warning mechanism is triggered immediately, and the loading strategy is automatically adjusted or other remedial measures are taken; S3. Finally, based on experimental data and theoretical analysis, a prediction model for the wetness of the internal surface layer of the PEMFC stack was established. This prediction model was used to evaluate the output performance of the PEMFC stack under different operating conditions and to identify possible fault types and their development trends. The specific steps for S3 are as follows: S31. Design the experiment according to the research purpose; S32. Collect key parameter data of the PEMFC stack under various operating conditions through specialized sensors and measurement equipment; S33, establishing the VI curve of the PEMFC stack, analyzing the VI curve, and extracting key parameters; S34. Use electrochemical impedance spectroscopy to measure the impedance characteristics of the battery stack at different frequencies and extract the response signals at key frequency points; S35. Analyze the variation of VI curve with current, temperature and humidity, and evaluate the output performance of the battery stack under different conditions; S36. Based on the data in S34, draw a frequency impedance spectrum characteristic curve, analyze the changes in parameters, and determine the changes in water content in the proton exchange membrane; S37. Based on experimental data and theoretical analysis, a prediction model for the wetness and dryness of the internal surface layer of the PEMFC stack was established; S38. Based on the data analysis results, explain the wet and dry status of the internal surface layer of the PEMFC stack, evaluate the output performance of the PEMFC stack under different operating conditions, and obtain the possible fault types and their development trends.

2. The test method for determining the dryness and wetness of a fuel cell stack surface by inspection sampling according to claim 1, characterized in that: In S1, laser positioning technology is used to measure the fuel cell stack, four inspection points are selected according to the actual structure of the stack, and an automatically adjustable inspection device is installed.

3. The test method for determining the dryness and wetness of a fuel cell stack surface by inspection sampling according to claim 1, characterized in that: The four inspection points are located at the four corners, including inspection group one (1), inspection group two (2), inspection group three (3) and inspection group four (4).

Images