A method and system for determining an operation parameter adjustment strategy of a hydrogen production system by electrolysis of water, and a storage medium

Abstract

The present application relates to the technical field of hydrogen production by water electrolysis, and particularly relates to a method and system for determining an operating parameter adjustment strategy of a hydrogen production system by water electrolysis and a storage medium, comprising: obtaining multi-source operating data of the hydrogen production system by water electrolysis, and establishing a corresponding relationship between electrochemical, temperature and bubble dynamic characteristics according to electrode positions and time after feature extraction; performing multi-feature fusion on the electrochemical characteristics, temperature characteristics and bubble dynamic characteristics with the corresponding relationship to obtain a multi-field collaborative feature set; performing multi-scale prediction based on the multi-field collaborative feature set to obtain electrode short-term state prediction results and electrode long-term trend prediction results, and determining a short-term operating parameter adjustment strategy and a long-term operating parameter adjustment strategy of the hydrogen production system by water electrolysis. Through the fusion of electrochemical, temperature and bubble dynamic characteristics and the multi-scale prediction, the present application can simultaneously output electrode short-term state and long-term trend prediction results, and provide corresponding basis for operating parameter adjustment at different time scales.

Description

Technical Field

[0001] This invention relates to the technical field of hydrogen production by water electrolysis, and in particular to a method, system, and storage medium for determining the adjustment strategy of operating parameters of a water electrolysis hydrogen production system. Background Technology

[0002] Hydrogen production systems via water electrolysis produce hydrogen through electrochemical reactions within an electrolyzer. During system operation, the electrode operating status is affected by factors such as electrochemical reactions, thermal conditions, and gas evolution. To monitor the operating status of the electrodes and electrolyzer, existing technologies collect operational data from the water electrolysis hydrogen production system and generate corresponding status information through monitoring, statistical analysis, or trend analysis, serving as a reference for adjusting system operating parameters.

[0003] Different operational data reflect varying operational factors and their change cycles. In processing methods that generate state information based on specific operational factors or a single time scale, the resulting state information primarily reflects the electrode's operational state within that corresponding operational factor or time scale, making it difficult to consider the correlations between different operational factors and the changes in electrode operational state across different time scales. Therefore, existing operational data processing methods for water electrolysis hydrogen production systems struggle to generate electrode operational state characterization and prediction results that take into account multiple factor correlations and different time scales, and the resulting state information is insufficient to provide a basis for adjusting operating parameters at different time scales. Summary of the Invention

[0004] This invention provides a method, system, and storage medium for determining the adjustment strategy of operating parameters in a water electrolysis hydrogen production system, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system includes: Acquire multi-source operating data of the water electrolysis hydrogen production system, including electrochemical parameters, temperature data, and bubble imaging data; Feature extraction is performed on the multi-source operating data to obtain electrochemical features, temperature features, and bubble dynamic features. The correspondence between the electrochemical features, temperature features, and bubble dynamic features is established according to the electrode monitoring position and monitoring time. The electrochemical features, temperature features, and bubble dynamic features with the aforementioned correspondence are fused to obtain a multi-field synergistic feature set; Based on the multi-field collaborative feature set, multi-scale prediction is performed to obtain the short-term state prediction result and the long-term trend prediction result of the electrode. Based on the short-term state prediction results of the electrodes, a short-term operating parameter adjustment strategy for the water electrolysis hydrogen production system is determined, and based on the long-term trend prediction results of the electrodes, a long-term operating parameter adjustment strategy for the water electrolysis hydrogen production system is determined.

[0006] Furthermore, the extraction of the bubble dynamic features includes: The bubble imaging data is denoised to obtain denoised image data; The denoised image data is segmented to obtain binarized image data for distinguishing bubble regions and non-bubble regions; Using the imaging device and imaging parameters used to acquire the bubble imaging data, a calibration image of a calibration plate with known physical dimensions is acquired, and the pixel physical conversion coefficient is determined based on the physical dimensions of the calibration plate and the pixel dimensions in the calibration image. The dynamic features of the bubble are extracted based on the binarized image data and the pixel physical conversion coefficients.

[0007] Furthermore, the electrode monitoring position corresponds to a local area of ​​a preset electrode, wherein: The electrochemical characteristics include local current density and unit hydrogen production energy consumption. The local current density is determined based on the ratio of the current corresponding to the local area of ​​the pre-positioned electrode to the effective reaction area of ​​the local area of ​​the pre-positioned electrode. The unit hydrogen production energy consumption is determined based on the ratio of the product of the current and the voltage of the local area of ​​the pre-positioned electrode to the amount of hydrogen produced per unit time. The temperature characteristics include the temperature of the local area of ​​the electrode and the average temperature of the electrode. The bubble dynamic characteristics include bubble coverage and bubble residence time. The bubble coverage is determined by the ratio of the sum of the projected areas of all bubbles in the local area of ​​the preset electrode to the physical area of ​​the local area of ​​the preset electrode. The bubble residence time is determined by the difference between the time when the same bubble leaves the electrode surface and the time when it is generated.

[0008] Furthermore, the multi-field collaborative feature set includes: Bubble-current density correlation coefficient, used to characterize the degree of linear correlation between the bubble coverage and the local current density; Bubble energy consumption increment coefficient, used to characterize the proportional relationship between bubble residence time deviation and unit hydrogen production energy consumption deviation; Bubble-temperature hotspot matching degree is used to characterize the degree of spatial overlap between bubble coverage state and temperature distribution state. The electrode multi-field health index is used to comprehensively characterize the electrode operating state reflected by the electrochemical characteristics, temperature characteristics, and bubble dynamic characteristics.

[0009] Furthermore, the correlation coefficient between the bubble and the current density is a Pearson correlation coefficient of multiple sets of bubble coverage and local current density corresponding to the same local area of ​​the same preset electrode within the same monitoring period, according to the monitoring time.

[0010] Furthermore, the bubble energy consumption increment coefficient is determined based on the ratio of the deviation in unit hydrogen production energy consumption to the deviation in bubble residence time, wherein: The deviation in unit hydrogen production energy consumption is the difference between the unit hydrogen production energy consumption in the current monitoring period and the benchmark unit hydrogen production energy consumption. The bubble residence time deviation is the difference between the average bubble residence time of each bubble in the current monitoring period and the reference bubble residence time. The benchmark unit hydrogen production energy consumption and the benchmark bubble residence time are respectively the arithmetic average of the unit hydrogen production energy consumption and bubble residence time of the water electrolysis hydrogen production system under preset benchmark operating conditions for multiple monitoring cycles. When the bubble residence time deviation is zero, the bubble energy consumption increment coefficient is determined to be zero.

[0011] Furthermore, determining the matching degree between the bubble and the temperature hotspot includes: The local area of ​​the pre-set electrode is divided into multiple regional units, and the bubble coverage rate and temperature of each regional unit are determined respectively. Regions with bubble coverage rates greater than a preset coverage threshold are classified into a bubble coverage over-threshold region set, and regions with temperatures greater than a preset temperature threshold are classified into a temperature over-threshold region set. When the union of the bubble-covered over-threshold region unit set and the temperature over-threshold region unit set is not empty, the matching degree between the bubble and the temperature hotspot is determined by using the number of region units that simultaneously belong to both the bubble-covered over-threshold region unit set and the temperature over-threshold region unit set as the numerator and the number of region units in the union as the denominator.

[0012] Furthermore, the determination of the electrode multi-field health index includes: The electrochemical search bond is determined based on the combination of the numerical range to which the local current density belongs and the numerical range to which the unit hydrogen production energy consumption belongs. The temperature search key is determined based on the combination of the numerical range to which the temperature of the local area of ​​the electrode belongs and the numerical range to which the average temperature of the electrode belongs. The bubble status retrieval key is determined based on the combination of the numerical range to which the bubble coverage rate belongs and the numerical range to which the bubble dwell time belongs. Based on the electrochemical search key, the temperature search key, and the bubble state search key, a preset feature scoring key value table is retrieved to obtain the electrochemical score, temperature score, and bubble state score. The preset feature scoring key value table records the mapping relationship between each search key and the corresponding score. The electrochemical score, temperature score, and bubble state score are weighted and summed according to the weighting coefficients corresponding to the electrochemical score, temperature score, and bubble state score, respectively, to obtain the electrode multi-field health index.

[0013] Furthermore, the short-term state prediction of the electrode includes: The regional temperature difference is determined based on the difference between the temperature of the local area of ​​the electrode and the average temperature of the electrode; The short-term observation time series data, consisting of the regional temperature difference, the bubble coverage rate, and the correlation coefficient between the bubble and the current density obtained from multiple consecutive monitoring periods, are input into the short-term state prediction model. The predicted values ​​of short-term local current density decay rate and short-term bubble coverage rate are obtained through the short-term state prediction model and used as the short-term state prediction results of the electrode.

[0014] Furthermore, the long-term trend prediction of the electrode includes: The long-term observation time series data, consisting of the bubble energy consumption increment coefficient, the matching degree between the bubble and the temperature hotspot, and the electrode multi-field health index at multiple consecutive monitoring times, is input into the long-term trend prediction model. The remaining effective lifespan of the electrode, the failure risk rate of the electrode edge region, and the predicted long-term unit hydrogen production energy consumption are obtained through the long-term trend prediction model and used as the long-term trend prediction results of the electrode. The remaining effective lifespan of the electrode is the predicted time from the predicted time until the electrode reaches the preset electrode failure judgment condition. The failure risk rate of the electrode edge region is the predicted probability that the electrode edge region will reach the preset edge region failure judgment condition within the prediction time range of the long-term trend prediction of the electrode.

[0015] Further, based on the short-term state prediction results of the electrodes, the short-term operating parameter adjustment strategy is determined, including: A short-term prediction combination feature is formed based on the short-term local current density decay rate and the short-term bubble coverage prediction value; Based on the pre-defined correspondence between short-term prediction combination features and short-term operating parameter adjustment strategies, determine the short-term operating parameter adjustment strategy corresponding to the short-term prediction combination features; The short-term operating parameter adjustment strategy includes at least one of the following: electrolyte flow rate adjustment, tank voltage adjustment, and stirring rate adjustment.

[0016] Furthermore, based on the long-term trend prediction results of the electrode, the long-term operating parameter adjustment strategy is determined, including: A long-term prediction combination feature is formed based on the remaining effective lifespan of the electrode, the failure risk rate of the electrode edge region, and the long-term unit hydrogen production energy consumption prediction value. Based on the pre-defined correspondence between long-term prediction combination features and long-term operating parameter adjustment strategies, determine the long-term operating parameter adjustment strategy corresponding to the long-term prediction combination features. The long-term operating parameter adjustment strategy includes at least one of the following: adjustment of operating current density, adjustment of electrolyte concentration, and adjustment of operating temperature.

[0017] Furthermore, the short-term observation time series data includes regional temperature difference, bubble coverage, and bubble-current density correlation coefficient obtained at multiple electrode monitoring locations in multiple consecutive monitoring cycles, and the data at the multiple electrode monitoring locations constitute spatial distribution data at each monitoring time.

[0018] This invention also provides a system for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system, comprising: The data acquisition module is used to acquire multi-source operating data of the water electrolysis hydrogen production system, including electrochemical parameters, temperature data and bubble imaging data. The feature extraction module is used to extract features from the multi-source operating data to obtain electrochemical features, temperature features and bubble dynamic features, and to establish the correspondence between the electrochemical features, the temperature features and the bubble dynamic features according to the electrode monitoring position and monitoring time; The feature fusion module is used to perform multi-feature fusion on the electrochemical features, temperature features and bubble dynamic features that have the corresponding relationship to obtain a multi-field synergistic feature set; The state prediction module is used to perform multi-scale prediction based on the multi-field collaborative feature set to obtain the short-term state prediction result and the long-term trend prediction result of the electrode. The strategy determination module is used to determine the short-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the short-term state prediction results of the electrodes, and to determine the long-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the long-term trend prediction results of the electrodes.

[0019] The present invention also provides a storage medium storing a computer program thereon, wherein when the computer program is executed by a processor, it implements the method for determining the adjustment strategy of the operating parameters of the water electrolysis hydrogen production system as described in any one of the claims.

[0020] The technical solution of this invention can achieve the following technical effects: This invention achieves comprehensive characterization of multiple operating factors, including electrochemical reactions, thermal state, and gas evolution, by simultaneously acquiring electrochemical parameters, temperature data, and bubble imaging data, and establishing a correspondence among these three data according to electrode monitoring location and monitoring time. Based on this, a multi-field collaborative feature set is obtained through multi-feature fusion, enabling the mining of correlation information between different operating factors. Furthermore, multi-scale prediction is performed based on this feature set, simultaneously outputting short-term electrode state prediction results and long-term trend prediction results, taking into account the changing patterns of operating status at different time scales. Therefore, this invention can provide differentiated decision-making basis for matching short-term operational adjustments and long-term maintenance planning for water electrolysis hydrogen production systems, improving the comprehensiveness of system operating status assessment and the pertinence of operating parameter adjustments, thus helping to improve hydrogen production efficiency, extend electrode life, and reduce operational risks.

[0021] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the method for determining the adjustment strategy of the operating parameters of the water electrolysis hydrogen production system in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the short-term state prediction model in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the long-term trend prediction model in an embodiment of the present invention; Figure 4 This is a schematic diagram of the system for determining the operating parameter adjustment strategy of the water electrolysis hydrogen production system in an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] like Figure 1 As shown, the method for determining the adjustment strategy of the operating parameters of the water electrolysis hydrogen production system of the present invention specifically includes the following steps: Step S100: Obtain multi-source operating data of the water electrolysis hydrogen production system. The multi-source operating data includes electrochemical parameters, temperature data, and bubble imaging data. Step S200: Extract features from multi-source operating data to obtain electrochemical features, temperature features and bubble dynamic features, and establish the correspondence between electrochemical features, temperature features and bubble dynamic features according to electrode monitoring position and monitoring time; Step S300: Perform multi-feature fusion on the corresponding electrochemical features, temperature features and bubble dynamic features to obtain a multi-field synergistic feature set; Step S400: Perform multi-scale prediction based on multi-field collaborative feature set to obtain the short-term state prediction result and the long-term trend prediction result of the electrode. Step S500: Determine the short-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the short-term state prediction results of the electrodes, and determine the long-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the long-term trend prediction results of the electrodes.

[0027] In this embodiment, electrochemical parameters include parameters reflecting the characteristics of the electrochemical reaction, such as local current and voltage in the electrode region. These parameters directly determine the rate and efficiency of the electrochemical reaction. Temperature data refers to the temperature values ​​of the electrode surface and various parts of the electrolyte, which affect the kinetics of the electrochemical reaction and thus impact hydrogen production efficiency and system stability. Bubble imaging data is image sequence data acquired by an imaging device, used to reflect the generation, growth, detachment, and movement of bubbles on the electrode surface. The electrochemical parameters and temperature data are numerical time series, while the bubble imaging data is an image sequence. Feature extraction allows for the analysis of the electrochemical reaction within the same numerical space. The correlation analysis of chemical reaction, temperature distribution and bubble behavior is carried out; the actual operating state of the electrode surface is the result of the coupling of electrochemical field, temperature field and bubble field. By fusing multiple features, the characteristics of different physical fields are incorporated into the same system, so that the operating state of the electrode can exist in the numerical form of multi-field coupling; the operating state of the electrode shows different change patterns at different time scales. By making predictions at different time scales, the different prediction results are matched with corresponding adjustment measures and transformed into specific adjustment schemes that can be executed. This allows the closed loop of the complete information chain formed by data acquisition, feature extraction, feature fusion and multi-scale prediction to be fed back to the system operation control.

[0028] In one specific embodiment, electrochemical parameters are collected by current and voltage sensors installed in the electrolytic cell, temperature data is collected by thermocouples or infrared thermal imagers installed in the electrolytic cell, and bubble imaging data is collected by high-speed cameras or industrial cameras.

[0029] Due to inherent limitations in sensor accuracy, external electromagnetic interference, and momentary equipment malfunctions, the collected electrochemical parameters and temperature data are prone to contain random noise, abnormal jump values, and missing data points. Before feature extraction from multi-source operating data, data cleaning operations are performed on the electrochemical parameters and temperature data, including: using a sliding window filtering method to identify and remove abnormal jump points in the electrochemical parameters and temperature data, and using linear interpolation to fill in the missing data.

[0030] Bubble imaging data is susceptible to fluctuations in lighting conditions, mechanical vibrations of equipment, and electromagnetic interference during acquisition. Common noise signals are unrelated to bubble morphology and can obscure the true outline, size, and spatial distribution of the bubble. Wavelet transform denoising is employed to process the bubble imaging data, including: decomposing the noisy image data into multi-scale, multi-frequency wavelet coefficients; setting a threshold and eliminating or suppressing wavelet coefficients corresponding to noise components while retaining effective coefficients characterizing bubble morphology and dynamic features; and reconstructing the image from the retained and suppressed wavelet coefficients to obtain denoised image data where noise is suppressed and the bubble outline and dynamic information are clearly presented.

[0031] After the above data cleaning and wavelet transform denoising process, outliers and missing values ​​in electrochemical parameters and temperature data are eliminated, and interference noise in bubble imaging data is suppressed. The cleaned electrochemical parameters, cleaned temperature data and denoised image data reach a unified data quality benchmark, meeting the input quality requirements for analysis and modeling.

[0032] In one specific embodiment, the electrochemical characteristics include local current density and energy consumption per unit of hydrogen production; local current density refers to the magnitude of the current passing through a unit effective reaction area of ​​the electrode, used to characterize the rate of electrochemical reaction and the utilization rate of the electrode material; the formula for calculating local current density J is: J=I / A Where I represents the local current of the electrode, the uniformity of its distribution affects the local reaction efficiency; A represents the effective reaction area of ​​the electrode, the value of which is positively correlated with the reaction rate; both I and A are directly extracted from the electrochemical parameters after cleaning or obtained through statistical calculations. Energy consumption per unit of hydrogen production refers to the electrical energy consumed to produce a unit volume or unit mass of hydrogen, used to measure the energy efficiency of a system; the formula for calculating energy consumption per unit of hydrogen production, E, is: E=(I×U) / Q Wherein, U is the voltage of the local region of the electrode, the value of which determines the driving force of the electrochemical reaction in that region; Q is the amount of hydrogen produced per unit time, the value of which characterizes the hydrogen production efficiency of the system; both U and Q are directly extracted from the electrochemical parameters after cleaning or obtained through statistical calculations.

[0033] Temperature characteristics include local electrode temperature and average electrode temperature. Local electrode temperature refers to the temperature value at a specific local location on the electrode surface. An abnormal increase in local temperature may lead to a decrease in local reaction efficiency or accelerated aging of the electrode material. Average electrode temperature refers to the average temperature at each monitoring location on the electrode surface, reflecting the overall thermal environment of the electrode. Local electrode temperature is directly extracted from the temperature data after cleaning, while average electrode temperature is obtained by calculating the arithmetic mean of the temperature data after cleaning according to the monitoring locations.

[0034] The dynamic characteristics of bubbles include bubble coverage and bubble residence time. Bubble coverage is determined by the ratio of the sum of the projected areas of all bubbles in the local area of ​​the pre-set electrode to the physical area of ​​the local area of ​​the pre-set electrode. Its value can reflect the degree of shading of the local reaction area of ​​the electrode by the bubbles. Bubble residence time is determined by the difference between the time when the same bubble leaves the electrode surface and the time when it is generated. Its length directly affects the duration for which the active sites on the electrode surface are blocked.

[0035] To obtain the bubble coverage and bubble residence time, features such as bubble shape, area, and number need to be extracted from the denoised image data. Since the denoised image contains non-bubble areas such as the electrode substrate and electrolyte background, image segmentation is required to clearly distinguish between bubble and non-bubble areas. This image segmentation involves inputting the denoised image data into a preset image segmentation model for processing, outputting a pixel-by-pixel probability map of the bubble area, and then binarizing the pixel-by-pixel probability map to obtain binarized image data.

[0036] Specifically, the image segmentation model employs a U-Net semantic segmentation network based on an encoder-decoder structure. This U-Net semantic segmentation network includes an encoder path that progressively downsamples to extract multi-scale features, a decoder path that progressively upsamples to restore spatial resolution, and skip connections connecting features at the same level between the encoder and decoder. The network's endpoint outputs a pixel-wise probability map of the bubble region through a sigmoid activation function. The training samples for this image segmentation model are bubble imaging data collected during the actual operation of the water electrolysis hydrogen production system and denoised using wavelet transform. Each frame of the image is manually annotated at the pixel level by annotators. The pixels containing bubbles on the electrode surface are labeled as foreground with a label value of 1, while non-bubble pixels such as the background and electrode substrate are labeled as background with a label value of 0. The labels are then verified by a reviewer to ensure consistency, resulting in a bubble segmentation label map. The labeled samples are divided into training, validation, and test sets in a 7:2:1 ratio. The Adam optimizer is used, with the sum of binary cross-entropy loss and Dice loss as the loss function. The initial learning rate is set to 0.001 and decays with each iteration. The batch size is set to 16, and the training run is 100 epochs. During training, early stopping is performed based on the cross-union ratio (CUI) of the validation set, and the optimal model is saved.

[0037] When processing the probability map output by the model, the pixel-by-pixel probability map is binarized according to a preset probability threshold. The preset probability threshold is preferably 0.5, which is the optimal decision threshold for binary classification based on the maximum a posteriori probability criterion. Then, morphological opening operation is performed in sequence to remove isolated noise points, morphological closing operation is performed to fill the voids inside the bubbles, and connected components with pixel areas smaller than the preset connected component area threshold are removed to obtain binarized image data. This binarized image data contains only two types of pixel values, for example, pixel value 255 represents the bubble region, and another pixel value such as 0 represents the non-bubble region.

[0038] Furthermore, since the bubble features in the binarized image data are in pixels, they cannot directly reflect the actual physical size of the bubble. Therefore, it is necessary to establish a correspondence between pixels and physical sizes. To this end, the same imaging equipment, shooting angle, and distance as the original bubble image are used to photograph a calibration board of a preset size, resulting in a calibration image. This calibration board has feature patterns with known physical sizes, such as square grids with known side lengths or dots with known diameters. Based on the pixel size corresponding to the known physical size features in the calibration board in the calibration image, the pixel physical conversion coefficient is calculated, which is the actual physical length corresponding to a single pixel. For example, if a grid with a side length of 1mm on the calibration board corresponds to 20 pixels in the image, then the pixel physical conversion coefficient is 0.05mm / pixel. This coefficient is used to convert the pixel area and pixel diameter of the bubble in the binarized image into the actual physical area and physical diameter.

[0039] Electrode monitoring location specifically refers to a pre-defined local area on the electrode surface during the design or parameter monitoring phase of an electrolytic water hydrogen production system. This area is used to focus on monitoring the electrochemical reaction state, and its physical area is a known parameter that can be obtained through design drawings or actual measurement.

[0040] To extract the area of ​​a single bubble, firstly, a contour detection algorithm is used to identify the bubble contour in the binary image data, thereby determining the boundary and shape range of the bubble relative to the background. Then, the number of white pixels within the contour of a single bubble is counted to obtain the total number of white pixels in that bubble. Next, the total number of white pixels is multiplied by the square of the pixel physical conversion coefficient to obtain the actual physical area of ​​a single bubble. The formula for calculating the area S of a single bubble is: S=N×k 2 Where N represents the total number of white pixels in the bubble, and k represents the pixel physical conversion coefficient.

[0041] The formula for calculating the bubble coverage rate R based on the area S of a single bubble is:

[0042] Where N1 is the total number of bubbles in the local area of ​​the electrode, obtained by counting all the bubbles identified in that area; S i Let A1 be the projected area of ​​the i-th bubble; A1 is the known physical area of ​​the local region of the electrode.

[0043] The bubble generation and detachment times are obtained by bubble tracking of the binarized image sequence. This process uses image processing algorithms to identify the position of the same bubble in different frames and establish a correlation to achieve continuous tracking of its motion trajectory. When the bubble first appears on the electrode surface, the timestamp of the corresponding image frame is recorded as the bubble generation time. When the bubble outline completely detaches from the electrode surface outline and there is no longer any overlapping area, the timestamp of the corresponding image frame is recorded as the bubble detachment time.

[0044] The formula for calculating the bubble residence time T, based on the bubble formation and detachment times, is as follows: T=t2-t1 Where t2 represents the time when the bubble detaches and t1 represents the time when the bubble is generated.

[0045] After the above processing, the local current density, unit hydrogen production energy consumption, local electrode temperature, average electrode temperature, bubble coverage, and bubble residence time are obtained. All of the above features are established according to the electrode monitoring location and monitoring time. That is, the electrochemical features, temperature features, and bubble dynamic features of the same local electrode region at the same monitoring time correspond to each other, so that subsequent multi-feature fusion can perform correlation analysis on the features of different physical fields at the same spatial location and time reference.

[0046] In one specific embodiment, a multi-field collaborative feature set is used to characterize the correlation between electrochemical features, temperature features, and bubble dynamics features; the multi-field collaborative feature set includes: a) A bubble-current density correlation coefficient used to characterize the degree of linear correlation between bubble coverage and local current density; this coefficient is a Pearson correlation coefficient for multiple sets of bubble coverage and local current density corresponding to the same local area of ​​the same pre-set electrode within the same monitoring period, calculated using the following formula:

[0047] in, denoted by , which represents the correlation coefficient between bubbles and current density; n represents the number of sets of matching data on bubble coverage and local current density collected synchronously within the same monitoring period in the local area of ​​the preset electrode, with each set of data corresponding to the monitoring value at the same moment; the value range of this coefficient is [-1, 1], with positive values ​​indicating that the local current density increases as the bubble coverage increases, and negative values ​​indicating that the two change in opposite directions; b. A bubble energy consumption increment coefficient used to characterize the proportional relationship between bubble residence time deviation and unit hydrogen production energy consumption deviation; this coefficient is determined based on the ratio of unit hydrogen production energy consumption deviation to bubble residence time deviation, where unit hydrogen production energy consumption deviation is the difference between unit hydrogen production energy consumption in the current monitoring period and the benchmark unit hydrogen production energy consumption; bubble residence time deviation is the difference between the average bubble residence time of each bubble in the current monitoring period and the benchmark bubble residence time; the calculation formula is:

[0048] in, This represents the bubble energy consumption increment coefficient; ΔE represents the energy consumption deviation, i.e. E represents the unit hydrogen production energy consumption during the current monitoring period. The reference unit is the energy consumption for hydrogen production; ΔT represents the residence time deviation, i.e. T represents the average residence time of each bubble within the current monitoring period. Reference bubble residence time; reference energy consumption per unit of hydrogen production and reference bubble residence time These are the arithmetic averages of the unit hydrogen production energy consumption and bubble residence time of the water electrolysis hydrogen production system over multiple monitoring cycles under preset baseline operating conditions. The baseline operating conditions are the normal operating conditions where the system is at rated current density, rated temperature, and stable bubble behavior. When the bubble residence time deviation ΔT is zero, to avoid the denominator being zero, the bubble energy consumption increment coefficient is... A value of zero indicates that the bubble residence time has not changed and has no additional contribution to the increase in energy consumption per unit of hydrogen production. c. The degree of spatial overlap between bubble coverage and temperature distribution, used to characterize the degree of overlap between bubble coverage and temperature distribution; the determination of this degree of overlap includes: The local area of ​​the pre-placed electrode is divided into multiple non-overlapping regional units of equal area, and the bubble coverage rate and temperature of each regional unit are determined respectively. Regions with bubble coverage rates greater than a preset coverage rate threshold are classified into a set of regions with bubble coverage exceeding the threshold. The preset coverage rate threshold is determined by statistically analyzing the bubble coverage rate distribution of each grid region during the historical stable operation of the water electrolysis hydrogen production system, taking the upper quartile of the bubble coverage rate under normal operating conditions, and can be adjusted within the range of 20%-40% according to the specific electrolyzer structure, current density, and process requirements. Regions whose temperature exceeds the preset temperature threshold are classified into the temperature exceeding threshold region set. The preset temperature threshold is based on the tolerance temperature of the electrolytic cell electrode material and diaphragm, as well as the upper limit of the normal reaction temperature range allowed by the process. The upper limit of the normal reaction temperature range is taken as the critical value, and can be adjusted accordingly according to the normal reaction temperature range of different electrolysis processes. When the union of the bubble-covered and temperature-over-threshold region sets is not empty, the matching degree between the bubble and temperature hotspots is determined by using the number of region units belonging to both sets as the numerator and the number of region units in the union as the denominator; the calculation formula is as follows:

[0049] in, This indicates the degree of matching between bubbles and temperature hotspots, with a value range of [0,1]. The closer the value is to 1, the higher the degree of spatial overlap between the dense bubble region and the high-temperature region. This indicates the number of overlapping grids between high-bubbling and high-temperature areas; Indicates the number of grid cells in the high bubble region; Indicates the number of grid cells in the high-temperature zone; d. An electrode multi-field health index used to comprehensively characterize the electrode's operating state as reflected by electrochemical characteristics, temperature characteristics, and bubble dynamic characteristics; the determination of this health index includes: The electrochemical search bond is determined by the combination of the numerical range of the local current density and the numerical range of the energy consumption per unit of hydrogen production. The temperature lookup key is determined based on the combination of the numerical range to which the local temperature of the electrode belongs and the numerical range to which the average temperature of the electrode belongs. The bubble status retrieval key is determined based on the combination of the numerical range to which the bubble coverage rate belongs and the numerical range to which the bubble dwell time belongs. Based on the electrochemical search key, temperature search key, and bubble state search key, the preset feature scoring key value table is searched to obtain the electrochemical score, temperature score, and bubble state score. The preset feature scoring key value table records the mapping relationship between each search key and the corresponding score. It was established by experts in the field based on long-term experience in water electrolysis for hydrogen production and industry electrode performance standards, and the scoring range was pre-defined for different numerical ranges of each search key. Specifically, the local current density, unit hydrogen production energy consumption, local electrode temperature, average electrode temperature, bubble coverage, and bubble residence time are pre-divided into multiple numerical intervals, and each numerical interval is assigned a corresponding interval number; the boundary of each numerical interval is determined based on at least one of the following: historical operating data obtained by the water electrolysis hydrogen production system under preset benchmark conditions, the allowable operating range of the equipment, and electrode performance evaluation requirements. The electrochemical search key is formed by combining the interval number corresponding to the local current density and the interval number corresponding to the unit hydrogen production energy consumption. For example, the local current density is divided into a first current density interval, a second current density interval, and a third current density interval, and the unit hydrogen production energy consumption is divided into a first energy consumption interval, a second energy consumption interval, and a third energy consumption interval. When the local current density is in the second current density interval and the unit hydrogen production energy consumption is in the first energy consumption interval, the corresponding electrochemical search key is recorded as: EC-2-1. The temperature lookup key is formed by combining the interval number corresponding to the local temperature of the electrode and the interval number corresponding to the average temperature of the electrode; for example, when the local temperature of the electrode is in the third local temperature interval and the average temperature of the electrode is in the second average temperature interval, the corresponding temperature lookup key is recorded as: TEMP-3-2. The bubble status retrieval key is formed by combining the interval number corresponding to the bubble coverage rate and the interval number corresponding to the bubble dwell time; for example, when the bubble coverage rate is in the second coverage rate interval and the bubble dwell time is in the third dwell time interval, the corresponding bubble status retrieval key is recorded as: BUB-2-3. The preset feature scoring key value table records the correspondence between electrochemical search keys and electrochemical scores, temperature search keys and temperature scores, and bubble state search keys and bubble state scores, respectively. For example, the preset feature scoring key value table records the mapping relationship between EC-2-1 and the first preset electrochemical score, the mapping relationship between TEMP-3-2 and the first preset temperature score, and the mapping relationship between BUB-2-3 and the first preset bubble state score. When the above search keys are formed based on the current running data, the scores corresponding to the formed search keys are read from the corresponding preset feature scoring key value table to obtain the electrochemical score, temperature score, and bubble state score, respectively. For example, when the scoring range of the electrode multifield health index is 0 to 10, the preset feature scoring key value table can record an electrochemical score of 8 for EC-2-1, a temperature score of 5 for TEMP-3-2, and a bubble state score of 4 for BUB-2-3. When the three search keys formed by the current running data are EC-2-1, TEMP-3-2, and BUB-2-3, the retrieved electrochemical score, temperature score, and bubble state score are 8, 5, and 4, respectively.

[0050] To further illustrate the mapping relationship between various search keys and their corresponding scores, some examples of the mapping relationships in the preset feature scoring key-value table are shown in Table 1:

[0051] Table 1. Example of partial mapping relationships in the preset feature scoring key-value table The search keys and corresponding scores shown in Table 1 are only used to illustrate the mapping and retrieval process of the preset feature score key value table. The actual numerical interval boundaries and corresponding scores should be determined based on the electrolyzer structure, electrode material, rated operating parameters, and historical operating data of the water electrolysis hydrogen production system. The above-mentioned search key encoding format, interval number, and corresponding scores are only used to illustrate the mapping process between search keys and scores, and do not constitute a limitation on the search key encoding format, interval number, and corresponding scores.

[0052] The electrochemical score, temperature score, and bubble state score are weighted and summed according to their respective weighting coefficients to obtain the electrode multi-field health index; the calculation formula is as follows: HI = w1×S1 + w2×S2 + w3×S3 Wherein, HI represents the electrode multi-field health index, with a value ranging from 0 to 10, and a higher score indicates a better health status; w1, w2, and w3 are the weighting coefficients corresponding to the electrochemical score S1, temperature score S2, and bubble state score S3, respectively, and w1+w2+w3=1; each weighting coefficient is calculated using the analytic hierarchy process (AHP) or the entropy weighting method, based on the degree of influence of electrochemical characteristics, temperature characteristics, and bubble dynamic characteristics on the overall performance of the electrode and hydrogen production efficiency; the AHP method involves experts constructing a judgment matrix based on the relative importance of the three types of characteristics to the electrode health status, and then normalizing and assigning weights after consistency testing; the entropy weighting method objectively assigns values ​​based on the dispersion of each score in historical data, with higher dispersion resulting in greater weight.

[0053] In this embodiment, the obtained bubble-current density correlation coefficient, bubble energy consumption increment coefficient, bubble-temperature hot spot matching degree, and electrode multi-field health index are used as a multi-field collaborative feature set. This feature set integrates the originally independent electrochemical features, temperature features, and bubble dynamic features into a unified feature system through calculation relationships such as correlation functions, proportional coefficients, spatial overlap, and weighted scoring, so that the electrode operating status can be received and utilized by the prediction model in a multi-field coupled numerical form.

[0054] In one specific embodiment, the prediction time range of the long-term trend prediction of the electrode is longer than that of the short-term state prediction. The short-term state prediction is used to reflect the state change trend of the electrode within a shorter time scale, providing a basis for immediate operational adjustments. The long-term trend prediction is used to reflect the performance degradation law and failure risk of the electrode over a longer time scale, providing a reference for maintenance plans and operational strategy optimization. The two have different time scales, and the feature inputs and prediction models used are also different, as detailed below: Step S410: Determine the regional temperature difference based on the difference between the local temperature of the electrode and the average temperature of the electrode; the calculation formula is as follows:

[0055] in, Indicates regional temperature difference. This indicates the temperature of a localized area of ​​the electrode. This represents the average electrode temperature; this temperature difference reflects the degree of unevenness in the temperature distribution on the electrode surface. Step S420: The regional temperature difference, bubble coverage, and bubble-current density correlation coefficient obtained from multiple consecutive monitoring cycles constitute short-term observation time-series data; the short-term observation time-series data includes the above three variables obtained from multiple consecutive monitoring cycles at multiple electrode monitoring locations, and the data from multiple electrode monitoring locations constitute spatial distribution data at each monitoring time. Step S430: Input the short-term observation time series data into the short-term state prediction model, and obtain the predicted values ​​of short-term local current density decay rate and short-term bubble coverage rate through the short-term state prediction model, which are used as the short-term state prediction results of the electrode. The short-term state prediction model is a spatiotemporal attention electrolysis efficiency prediction model. Its input consists of short-term observation time series data and its spatial distribution data, which are composed of regional temperature difference, bubble coverage rate and bubble-current density correlation coefficient. The output is the short-term local current density decay rate and the short-term bubble coverage rate prediction. The short-term local current density decay rate refers to the decrease of local current density per unit time in the future predetermined short-term period. The short-term bubble coverage rate prediction is the estimate of the proportion of the electrode surface covered by bubbles in the future predetermined short-term period.

[0056] like Figure 2 As shown, the spatiotemporal attention electrolysis efficiency prediction model includes: The spatial feature extraction module is used to extract spatial features from spatially distributed data using a convolutional network. The time feature extraction module is used to extract time variation features from short-term observation time series data using a long short-term memory network. The attention mechanism layer is used to employ a multi-head attention mechanism to weight spatial and temporal variation features along the monitoring time dimension and the electrode monitoring location dimension. And an output layer, used to output short-term state prediction results of the electrodes based on the weighted features.

[0057] The model expression is:

[0058] in, This indicates the short-term forecast results, including the short-term local current density decay rate and the short-term bubble coverage rate forecast. This represents a spatiotemporal attention electrolysis efficiency prediction model; This represents the input time-series feature matrix, containing regional temperature differences. Bubble coverage Correlation coefficient between bubbles and current density In time series Observations and spatial distribution information; This represents the learnable parameter matrix of the model, which is obtained through optimization during the training process.

[0059] The training data for this model comes from short-cycle time-series operational data collected from a water electrolysis hydrogen production system under various operating conditions. This includes observations and spatial distribution information of regional temperature difference, bubble coverage, and the correlation coefficient between bubbles and current density at continuous monitoring times. After preprocessing and sliding window segmentation, a time-series feature matrix sample is formed. Labels are defined as the measured short-term local current density decay rate and short-term bubble coverage for the next time period (1 to 24 hours) corresponding to each time-series sample. The model's loss function uses the mean squared error between the predicted value and the true label. The convolutional layer kernel size is 3×3, and the temporal feature extraction uses a long short-term memory network with 128 hidden layer units, 4 attention heads, a batch size of 32, an initial learning rate of 0.0005, and the Adam optimizer. The training iterations are 150 epochs combined with an early stopping strategy. Validation metrics include root mean square error, mean absolute error, and coefficient of determination. The training and validation sets are divided in a 7:3 ratio, and the model with the highest coefficient of determination and the lowest root mean square error is selected as the final prediction model.

[0060] Step S440: Input the long-term observation time series data, which consists of the bubble energy consumption increment coefficient, the matching degree between the bubble and the temperature hot spot, and the electrode multi-field health index at multiple consecutive monitoring times, as well as the preset electrode local area identifier corresponding to the long-term observation time series data, into the long-term trend prediction model. The remaining effective life of the electrode, the failure risk rate of the electrode edge area, and the long-term unit hydrogen production energy consumption prediction value are obtained through the long-term trend prediction model, which are used as the long-term trend prediction results of the electrode. The remaining effective lifespan of the electrode is the predicted time from the predicted time until the electrode reaches the preset electrode failure judgment condition. The preset electrode failure judgment condition is, for example, a certain percentage increase in unit hydrogen production energy consumption exceeding the rated value or a local current density decay below a threshold. The electrode edge region failure risk rate is the predicted probability that the preset electrode local region corresponding to the preset electrode local region identifier reaches the preset region failure judgment condition within the predicted time range of the long-term trend prediction of the electrode. The long-term unit hydrogen production energy consumption prediction value is the predicted value of unit hydrogen production energy consumption over a relatively long period of time in the future.

[0061] like Figure 3 As shown, this long-term trend prediction model is an electrode health and energy consumption trend prediction model, which includes: The input layer is used to receive long-term observation time-series data and the corresponding preset electrode local region identifiers. The feature fusion module is used to fuse features between long-term observation time-series data and preset local region identifiers of electrodes. The long-term trend extraction layer is used to extract long-term temporal dependency features from the fused features using a temporal convolutional network. And an output layer, used to output the long-term trend prediction results of the electrodes based on long-term time-dependent characteristics.

[0062] The model expression is:

[0063] in, This represents the long-term forecast results, including the remaining effective life of the electrode, the failure risk rate of the electrode edge region, and the long-term predicted energy consumption per unit of hydrogen production. This represents a model for predicting electrode health and energy consumption trends. This represents the long-term feature matrix of the input, containing the bubble energy consumption increment coefficients. Bubble-temperature hot spot matching degree Electrode multi-field health index In time series Cumulative observations on; The learnable parameter matrix of the model is obtained through training and optimization using long-term running data.

[0064] The training data for this model comes from the time-series data of the cumulative bubble energy consumption increment coefficient, bubble-temperature hot spot matching degree, and electrode multi-field health index accumulated during the long-term continuous operation of the water electrolysis hydrogen production system, combined with the actual operating status and maintenance records of the electrodes recorded during the same period. Its lifespan and failure risk labels are defined as follows: the remaining effective lifespan of the electrode is calibrated by the time elapsed from its commissioning to performance degradation to the failure judgment standard; the failure risk rate of the electrode edge area is calibrated by the statistical frequency of actual failures occurring in the electrode edge area; and the long-term unit hydrogen production energy consumption is calibrated by the long-term measured unit hydrogen production energy consumption. The training cycle consists of 200 rounds of iterative training on the accumulated long-term time-series samples, using the Adam optimizer. The loss function adopts the mean square error loss of the regression task, and a weighted summation of the three types of outputs: lifespan, failure risk, and energy consumption. The evaluation indicators are root mean square error, mean absolute percentage error, and coefficient of determination. Its validation method uses time-series cross-validation, that is, dividing the training interval and validation interval according to the time sequence to avoid future data leakage, and reserving independent later operating data as a test set to verify the model's generalization ability in long-term trend prediction.

[0065] After the above processing, the short-term state prediction results and long-term trend prediction results of the electrode form quantitative predictions of the electrode operating state at different time scales. The prediction results are output in numerical form. By converting them into specific operating parameter adjustment instructions, they can provide guidance for the actual operation of the water electrolysis hydrogen production system. Different combinations of prediction results correspond to different system operating states, and corresponding adjustment measures need to be matched to optimize system performance, extend electrode life and reduce energy consumption.

[0066] In one specific embodiment, a short-term operating parameter adjustment strategy is determined based on the electrode short-term state prediction results, including: Short-term prediction combination features are formed based on the predicted values ​​of short-term local current density decay rate and short-term bubble coverage rate. Based on the pre-defined correspondence between short-term forecast combination characteristics and short-term operating parameter adjustment strategies, determine the short-term operating parameter adjustment strategy corresponding to the short-term forecast combination characteristics. Among them, the short-term prediction combination feature is formed by dividing the predicted values ​​of short-term local current density attenuation rate and short-term bubble coverage rate into multiple numerical intervals, such as "attenuation rate greater than 5mA / (cm)". 2 •h) and coverage greater than 30%" or "attenuation rate less than or equal to 2mA / (cm) 2 •h) and coverage less than or equal to 15%; the correspondence between the preset short-term prediction combination features and the short-term operating parameter adjustment strategies is stored in a preset short-term operating strategy library, which is generated based on a large amount of historical operating data; the short-term operating parameter adjustment strategies include at least one selected from electrolyte flow rate adjustment, tank voltage adjustment, and stirring rate adjustment; the strategy library is retrieved according to the currently obtained short-term prediction combination features, and the corresponding short-term operating parameter adjustment strategy is matched. Each identifier corresponds to a unique adjustment scheme in the preset short-term strategy library; for example, when the combination feature is "attenuation rate greater than 5mA / (cm)", the corresponding short-term operating parameter adjustment strategy is obtained. 2 When “·h) and the coverage is greater than 30%”, the matching strategy is “increase the electrolyte flow rate to 18L / min and adjust the stirring rate to 700r / min”.

[0067] Based on the long-term trend prediction results of the electrodes, a long-term operating parameter adjustment strategy is determined, including: Long-term forecast combination characteristics are formed based on the remaining effective life of the electrode, the failure risk rate of the electrode edge area, and the long-term unit hydrogen production energy consumption forecast. Based on the pre-defined correspondence between long-term forecast combination characteristics and long-term operating parameter adjustment strategies, determine the long-term operating parameter adjustment strategy corresponding to the long-term forecast combination characteristics. The long-term prediction combination features are jointly composed of the remaining effective life of the electrode, the failure risk rate of the electrode edge region, and the predicted long-term unit hydrogen production energy consumption. The correspondence between the preset long-term prediction combination features and the long-term operating parameter adjustment strategies is stored in a preset long-term operating strategy library, which is generated based on a large amount of historical operating data. This long-term operating strategy library contains a three-level strategy system: maintenance strategies, such as electrode polishing cycle and catalyst replenishment amount; structural optimization strategies, such as adjusting the radius of the electrode edge rounded corners and improving the flow channel design; and long-term setting of process parameters, such as basic current density and electrolyte concentration. The strategy library is retrieved based on the currently obtained long-term prediction combination features, and the corresponding long-term operating parameter adjustment strategy is matched. Each identifier corresponds to a unique adjustment scheme in the preset long-term operating strategy library. For example, when the predicted long-term unit hydrogen production energy consumption is greater than 5.5 kWh / m³, the strategy will be adjusted accordingly. 3 At that time, the parameter adjustment strategy is matched with "the electrolyte concentration is adjusted from 0.5 mol / L to 0.8 mol / L and the operating temperature is increased to 80℃".

[0068] After the above processing, short-term and long-term operating parameter adjustment strategies are obtained, which together constitute the operating parameter optimization strategy for the water electrolysis hydrogen production system. The short-term operating parameter adjustment strategy is used to address the current decay and excessive bubble coverage of the electrode in the short term by adjusting the electrolyte flow rate, cell voltage, or stirring rate in real time to stabilize the system operation. The long-term operating parameter adjustment strategy is used to address the performance decay and energy consumption increase trend of the electrode in long-term operation by adjusting the operating current density, electrolyte concentration, or operating temperature to extend the electrode life and reduce long-term energy consumption. This processing transforms the prediction results into specific, executable adjustment schemes corresponding to the actual operating parameters through a preset strategy library, so that the information chain formed by data acquisition, feature extraction, feature fusion, and multi-scale prediction can ultimately be fed back to the system operation control in a closed loop.

[0069] Based on the same inventive concept as the method for determining the adjustment strategy of operating parameters of a water electrolysis hydrogen production system in the foregoing embodiments, this invention also provides a system for determining the adjustment strategy of operating parameters of a water electrolysis hydrogen production system, such as... Figure 4 As shown, the system includes: The data acquisition module is used to acquire multi-source operating data of the water electrolysis hydrogen production system, including electrochemical parameters, temperature data, and bubble imaging data. The feature extraction module is used to extract features from multi-source operating data to obtain electrochemical features, temperature features, and bubble dynamic features, and to establish the correspondence between electrochemical features, temperature features, and bubble dynamic features according to electrode monitoring location and monitoring time; The feature fusion module is used to fuse corresponding electrochemical features, temperature features, and bubble dynamic features to obtain a multi-field synergistic feature set. The state prediction module is used to perform multi-scale prediction based on a multi-field collaborative feature set to obtain short-term state prediction results and long-term trend prediction results of the electrode. The strategy determination module is used to determine the short-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the short-term state prediction results of the electrodes, and to determine the long-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the long-term trend prediction results of the electrodes.

[0070] The system described above in this invention can effectively realize a method for determining the adjustment strategy of operating parameters of an electrolytic water hydrogen production system, and the technical effects it can achieve are as described in the above embodiments, which will not be repeated here.

[0071] The present invention also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program may include some or all of the steps in various embodiments of the method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system provided by the present invention. The storage medium may be read-only memory (ROM), random access memory (RAM), compact-disc read-only memory (CD-ROM), magnetic tape, floppy disk, or optical data storage device, etc.

[0072] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for determining the adjustment strategy of operating parameters in a water electrolysis hydrogen production system, characterized in that, include: Acquire multi-source operating data of the water electrolysis hydrogen production system, including electrochemical parameters, temperature data, and bubble imaging data; Feature extraction is performed on the multi-source operating data to obtain electrochemical features, temperature features, and bubble dynamic features. The correspondence between the electrochemical features, temperature features, and bubble dynamic features is established according to the electrode monitoring position and monitoring time. The electrochemical features, temperature features, and bubble dynamic features with the aforementioned correspondence are fused to obtain a multi-field synergistic feature set; Based on the multi-field collaborative feature set, multi-scale prediction is performed to obtain the short-term state prediction result and the long-term trend prediction result of the electrode. Based on the short-term state prediction results of the electrodes, a short-term operating parameter adjustment strategy for the water electrolysis hydrogen production system is determined, and based on the long-term trend prediction results of the electrodes, a long-term operating parameter adjustment strategy for the water electrolysis hydrogen production system is determined.

2. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 1, characterized in that, The extraction of the bubble dynamic features includes: The bubble imaging data is denoised to obtain denoised image data; The denoised image data is segmented to obtain binarized image data for distinguishing bubble regions and non-bubble regions; Using the imaging device and imaging parameters used to acquire the bubble imaging data, a calibration image of a calibration plate with known physical dimensions is acquired, and the pixel physical conversion coefficient is determined based on the physical dimensions of the calibration plate and the pixel dimensions in the calibration image. The dynamic features of the bubble are extracted based on the binarized image data and the pixel physical conversion coefficients.

3. The method for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system according to claim 1, characterized in that, The electrode monitoring position corresponds to a local area of ​​a preset electrode, wherein: The electrochemical characteristics include local current density and unit hydrogen production energy consumption. The local current density is determined based on the ratio of the current corresponding to the local area of ​​the pre-positioned electrode to the effective reaction area of ​​the local area of ​​the pre-positioned electrode. The unit hydrogen production energy consumption is determined based on the ratio of the product of the current and the voltage of the local area of ​​the pre-positioned electrode to the amount of hydrogen produced per unit time. The temperature characteristics include the temperature of the local area of ​​the electrode and the average temperature of the electrode. The bubble dynamic characteristics include bubble coverage and bubble residence time. The bubble coverage is determined by the ratio of the sum of the projected areas of all bubbles in the local area of ​​the preset electrode to the physical area of ​​the local area of ​​the preset electrode. The bubble residence time is determined by the difference between the time when the same bubble leaves the electrode surface and the time when it is generated.

4. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 3, characterized in that, The multi-field collaborative feature set includes: Bubble-current density correlation coefficient, used to characterize the degree of linear correlation between the bubble coverage and the local current density; Bubble energy consumption increment coefficient, used to characterize the proportional relationship between bubble residence time deviation and unit hydrogen production energy consumption deviation; Bubble-temperature hotspot matching degree is used to characterize the degree of spatial overlap between bubble coverage state and temperature distribution state. The electrode multi-field health index is used to comprehensively characterize the electrode operating state reflected by the electrochemical characteristics, temperature characteristics, and bubble dynamic characteristics.

5. The method for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system according to claim 4, characterized in that, The correlation coefficient between bubbles and current density is a Pearson correlation coefficient between multiple sets of bubble coverage and local current density corresponding to the same local area of ​​the same preset electrode within the same monitoring period, according to the monitoring time.

6. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 4, characterized in that, The bubble energy consumption increment coefficient is determined based on the ratio of the deviation in unit hydrogen production energy consumption to the deviation in bubble residence time, wherein: The deviation in unit hydrogen production energy consumption is the difference between the unit hydrogen production energy consumption in the current monitoring period and the benchmark unit hydrogen production energy consumption. The bubble residence time deviation is the difference between the average bubble residence time of each bubble in the current monitoring period and the reference bubble residence time. The benchmark unit hydrogen production energy consumption and the benchmark bubble residence time are respectively the arithmetic average of the unit hydrogen production energy consumption and bubble residence time of the water electrolysis hydrogen production system under preset benchmark operating conditions for multiple monitoring cycles. When the bubble residence time deviation is zero, the bubble energy consumption increment coefficient is determined to be zero.

7. The method for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system according to claim 4, characterized in that, The determination of the matching degree between the bubble and the temperature hotspot includes: The local area of ​​the pre-set electrode is divided into multiple regional units, and the bubble coverage rate and temperature of each regional unit are determined respectively. Regions with bubble coverage rates greater than a preset coverage threshold are classified into a bubble coverage over-threshold region set, and regions with temperatures greater than a preset temperature threshold are classified into a temperature over-threshold region set. When the union of the bubble-covered over-threshold region unit set and the temperature over-threshold region unit set is not empty, the matching degree between the bubble and the temperature hotspot is determined by using the number of region units that simultaneously belong to both the bubble-covered over-threshold region unit set and the temperature over-threshold region unit set as the numerator and the number of region units in the union as the denominator.

8. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 4, characterized in that, The determination of the electrode multi-field health index includes: The electrochemical search bond is determined based on the combination of the numerical range to which the local current density belongs and the numerical range to which the unit hydrogen production energy consumption belongs. The temperature search key is determined based on the combination of the numerical range to which the temperature of the local area of ​​the electrode belongs and the numerical range to which the average temperature of the electrode belongs. The bubble status retrieval key is determined based on the combination of the numerical range to which the bubble coverage rate belongs and the numerical range to which the bubble dwell time belongs; Based on the electrochemical search key, the temperature search key, and the bubble state search key, a preset feature scoring key value table is retrieved to obtain the electrochemical score, temperature score, and bubble state score. The preset feature scoring key value table records the mapping relationship between each search key and the corresponding score. The electrochemical score, temperature score, and bubble state score are weighted and summed according to the weighting coefficients corresponding to the electrochemical score, temperature score, and bubble state score, respectively, to obtain the electrode multi-field health index.

9. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 4, characterized in that, Electrode short-term state prediction includes: The regional temperature difference is determined based on the difference between the temperature of the local area of ​​the electrode and the average temperature of the electrode; The short-term observation time series data, consisting of the regional temperature difference, the bubble coverage rate, and the correlation coefficient between the bubble and the current density obtained from multiple consecutive monitoring periods, are input into the short-term state prediction model. The predicted values ​​of short-term local current density decay rate and short-term bubble coverage rate are obtained through the short-term state prediction model and used as the short-term state prediction results of the electrode.

10. The method for determining the adjustment strategy of operating parameters for a water electrolysis hydrogen production system according to claim 4, characterized in that, Long-term trend forecasts for electrodes include: The long-term observation time series data, consisting of the bubble energy consumption increment coefficient, the matching degree between the bubble and the temperature hotspot, and the electrode multi-field health index at multiple consecutive monitoring times, is input into the long-term trend prediction model. The remaining effective lifespan of the electrode, the failure risk rate of the electrode edge region, and the predicted long-term unit hydrogen production energy consumption are obtained through the long-term trend prediction model and used as the long-term trend prediction results of the electrode. The remaining effective lifespan of the electrode is the predicted time from the predicted time until the electrode reaches the preset electrode failure judgment condition. The failure risk rate of the electrode edge region is the predicted probability that the electrode edge region will reach the preset edge region failure judgment condition within the prediction time range of the long-term trend prediction of the electrode.

11. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 9, characterized in that, The short-term operating parameter adjustment strategy is determined based on the electrode short-term state prediction results, including: A short-term prediction combination feature is formed based on the short-term local current density decay rate and the short-term bubble coverage prediction value; Based on the pre-defined correspondence between short-term prediction combination features and short-term operating parameter adjustment strategies, determine the short-term operating parameter adjustment strategy corresponding to the short-term prediction combination features; The short-term operating parameter adjustment strategy includes at least one of the following: electrolyte flow rate adjustment, tank voltage adjustment, and stirring rate adjustment.

12. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 10, characterized in that, The long-term operating parameter adjustment strategy is determined based on the long-term trend prediction results of the electrode, including: A long-term prediction combination feature is formed based on the remaining effective lifespan of the electrode, the failure risk rate of the electrode edge region, and the long-term unit hydrogen production energy consumption prediction value. Based on the pre-defined correspondence between long-term prediction combination features and long-term operating parameter adjustment strategies, determine the long-term operating parameter adjustment strategy corresponding to the long-term prediction combination features. The long-term operating parameter adjustment strategy includes at least one of the following: adjustment of operating current density, adjustment of electrolyte concentration, and adjustment of operating temperature.

13. The method for determining the adjustment strategy of operating parameters of the water electrolysis hydrogen production system according to claim 9, characterized in that, The short-term observation time series data includes regional temperature difference, bubble coverage, and bubble-current density correlation coefficient obtained at multiple electrode monitoring locations in multiple consecutive monitoring cycles. The data from the multiple electrode monitoring locations constitute spatial distribution data at each monitoring time.

14. A system for determining the adjustment strategy of operating parameters in a water electrolysis hydrogen production system, characterized in that, include: The data acquisition module is used to acquire multi-source operating data of the water electrolysis hydrogen production system, including electrochemical parameters, temperature data and bubble imaging data. The feature extraction module is used to extract features from the multi-source operating data to obtain electrochemical features, temperature features and bubble dynamic features, and to establish the correspondence between the electrochemical features, the temperature features and the bubble dynamic features according to the electrode monitoring position and monitoring time; The feature fusion module is used to perform multi-feature fusion on the electrochemical features, temperature features and bubble dynamic features that have the corresponding relationship to obtain a multi-field synergistic feature set; The state prediction module is used to perform multi-scale prediction based on the multi-field collaborative feature set to obtain the short-term state prediction result and the long-term trend prediction result of the electrode. The strategy determination module is used to determine the short-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the short-term state prediction results of the electrodes, and to determine the long-term operating parameter adjustment strategy of the water electrolysis hydrogen production system based on the long-term trend prediction results of the electrodes.

15. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for determining the adjustment strategy of the operating parameters of the water electrolysis hydrogen production system as described in any one of claims 1 to 13.

Images