Method and system for constructing electrolysis energy consumption model and storage medium

Abstract

The invention relates to the technical field of model construction, in particular to a construction method and system of an electrolysis energy consumption model and a storage medium. The method comprises the following steps: collecting electrolysis process parameters by using a multi-mode sensor system, wherein the electrolysis process parameters comprise a current value, a voltage value, an electrolyte temperature and electrolysis time; constructing an electrolysis energy consumption initial model based on the electrolysis process parameters, and inputting the electrolysis process parameters into the electrolysis energy consumption initial model to output a unit product energy consumption value and a current-voltage efficiency curve; electrolytic bubble acoustic monitoring is conducted on the side wall of the electrolytic cell to obtain bubble breakage acoustic wave signals; and analyzing the sound intensity and the rupture frequency of the bubble rupture sound wave signal. According to the method, through the multi-modal sensor, multi-dimensional signal analysis and dynamic parameter correction, the defects of single data and static state of a traditional electrolysis energy consumption model are broken through, and high-precision dynamic modeling and optimization of energy consumption in the electrolysis process are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of model construction, and in particular to a construction method, system and storage medium of an electrolysis energy consumption model. Background Art

[0002] Energy consumption models for electrolysis processes are often based on empirical formulas and lack an in-depth description of electrochemical reaction mechanisms and mass transfer processes, resulting in limited model accuracy and applicability. With the advancement of computing power and the development of electrochemical theory, numerical models based on physical and chemical principles have gradually emerged. These models achieve more accurate predictions of electrolysis energy consumption by coupling electrode reaction kinetics, electrolyte transport, and thermodynamic processes. However, these models are often computationally complex and highly dependent on input parameters, limiting their practical application. In recent years, with the development of big data and machine learning technologies, hybrid modeling approaches that integrate experimental data with theoretical models have gradually become a trend. These approaches improve the model's generalization ability and prediction accuracy through data-driven optimization of model parameters. At the same time, more adaptable energy consumption models are constantly being proposed for different electrolysis systems (such as alkaline electrolysis, water electrolysis, and proton exchange membrane electrolysis) to meet the needs of energy efficiency assessment under industrial-scale and dynamic operating conditions. However, the accumulation of bubbles during the electrolysis process can cause electrode shielding and reduce current efficiency. Existing models generally lack a quantitative description of the abnormal behavior of bubbles. At the same time, electrode corrosion and surface roughness changes directly affect the electrolysis efficiency, but most traditional models do not incorporate such influencing factors, resulting in low accuracy in the construction of electrolysis energy consumption models. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, system and storage medium for constructing an electrolysis energy consumption model to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for constructing an electrolysis energy consumption model is provided, the method comprising the following steps:

[0005] Step S1: using a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature, and electrolysis time; constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, and inputting the electrolysis process parameters into the initial electrolysis energy consumption model to output a unit product energy consumption value and a current-voltage efficiency curve;

[0006] Step S2: Performing electrolytic bubble acoustic monitoring on the side wall of the electrolytic cell to obtain a bubble burst acoustic wave signal; analyzing the sound intensity and burst frequency of the bubble burst acoustic wave signal; performing abnormal bubble accumulation judgment on the bubble burst acoustic wave signal based on the sound intensity and burst frequency, and generating a bubble shielding correction coefficient;

[0007] Step S3: using an industrial endoscope to obtain a microscopic image of the electrode surface; performing grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identifying the electrode corrosion pit density in the grayscale image of the electrode surface, and performing surface roughness corrosion discrimination on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor;

[0008] Step S4: Input the bubble shielding correction coefficient and the surface roughness attenuation factor into the current efficiency parameter of the dynamic correction current-voltage efficiency curve in the initial electrolysis energy consumption model to generate an electrolysis energy consumption optimization model; perform simulation energy consumption verification on the electrolysis energy consumption optimization model, and when the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, generate the final electrolysis energy consumption model.

[0009] The present invention constructs a basic energy consumption assessment framework by inputting core process parameters such as current, voltage, temperature, and electrolysis time into the initial energy consumption model, and then introduces microscopic influencing factors such as bubbles and corrosion through acoustic monitoring and image recognition to achieve more accurate modeling of the complex coupling relationship of the electrolysis process. By means of the sound intensity and rupture frequency analysis of the bubble burst acoustic signal, the abnormal accumulation of bubbles in the electrolysis process is recognized in real time, and a bubble shielding correction coefficient is generated to effectively correct the influence of bubbles on current efficiency and improve the credibility of the electrolysis efficiency parameters. Microscopic images are obtained through an industrial endoscope, and combined with grayscale gradient analysis technology, the density of electrode corrosion pits is identified, and then the surface roughness of the electrode is judged, and a surface roughness attenuation factor is generated to achieve modeling feedback on the effect of electrode aging on energy consumption changes. The bubble shielding correction coefficient and the surface roughness attenuation factor are introduced into the initial model for dynamic correction to construct an electrolysis energy consumption optimization model, so that the model has the ability to respond to changes in process status in real time, thereby improving the dynamic adaptability and predictive performance of the modeling. By verifying the deviation between the unit product energy consumption value output by the model and the actual value through simulation, and outputting the final energy consumption model when the deviation falls below a set threshold, this model can provide a reliable basis for energy-saving control and optimized operation of the electrolysis process, and has promising industrial application prospects. Therefore, through multimodal sensors, multidimensional signal analysis, and dynamic parameter correction, this invention overcomes the data-single and static limitations of traditional electrolysis energy consumption models, achieving high-precision dynamic modeling and optimization of electrolysis process energy consumption.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: using a current sensor to collect the current value of the electrolysis process;

[0012] Step S12: collecting the voltage value of the electrolysis process through the voltage collection module;

[0013] Step S13: using a thermosensitive element to obtain the electrolyte temperature during the electrolysis process;

[0014] Step S14: obtaining the electrolysis time of the electrolysis process based on the timing module;

[0015] Step S15: synchronously scheduling and controlling the current value, voltage value, electrolyte temperature, and electrolysis time to obtain the original parameters of the electrolysis process; performing data preprocessing on the original parameters of the electrolysis process to generate electrolysis process parameters, wherein the data preprocessing includes data denoising, outlier removal, time synchronization calibration, and data standardization;

[0016] Step S16: constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, inputting the electrolysis process parameters into the initial electrolysis energy consumption model to calculate energy consumption, and outputting the unit product energy consumption value and the current-voltage efficiency curve.

[0017] The present invention uses a current sensor, a voltage acquisition module, a thermistor and a timing module to independently collect key physical quantities of the electrolysis process (S11-S14), thereby improving the accuracy and stability of parameter acquisition and laying a high-quality data foundation for subsequent energy consumption modeling. In step S15, a synchronous scheduling control mechanism is introduced to ensure that the data collected by different sensors are time-synchronous, and the credibility and versatility of the electrolysis process parameters are improved through pre-processing means such as data denoising, outlier removal, time synchronization calibration, and data standardization. In step S16, an initial model of electrolysis energy consumption is constructed based on high-quality electrolysis process parameters, and the unit product energy consumption value and current-voltage efficiency curve are output to provide preliminary energy consumption evaluation results for subsequent model correction and energy efficiency optimization. Through a step-by-step structured design, the system has better modular expansion capabilities, which is convenient for the subsequent integration of other sensors or optimization algorithms, and also provides technical support for the realization of automatic modeling-real-time correction-intelligent control. By refining the collection process and data processing logic, the energy consumption characteristics of the electrolysis process can be captured and modeled quickly, accurately and in a fine-grained manner, providing more accurate energy consumption prediction capabilities for high-frequency industrial electrolysis systems.

[0018] Preferably, analyzing the sound intensity and burst frequency of the bubble burst sound wave signal in step S2 includes:

[0019] Extract the original bubble burst acoustic wave signal collected during the electrolysis process;

[0020] Calculate the instantaneous sound pressure value of the bubble sound wave signal and obtain the sound pressure envelope curve;

[0021] Identify the peak area in the sound pressure envelope curve and extract potential bubble collapse event fragments;

[0022] Detect the average sound intensity value and the maximum sound intensity value of each potential bubble burst event segment to generate sound intensity characteristic parameters;

[0023] Calculate the main frequency energy distribution in the rupture event fragment, extract the main frequency points and generate the frequency spectrum;

[0024] Identify the area with continuous resonance peaks in the frequency spectrum and determine it as the effective rupture frequency range;

[0025] The sound intensity characteristic parameters of each rupture event and the effective rupture frequency range are integrated to obtain the sound intensity and rupture frequency of the complete bubble rupture acoustic wave signal.

[0026] The present invention effectively distinguishes between real bubble burst events and background noise interference by extracting the sound pressure envelope curve from the original sound wave and identifying the peak area, thereby enhancing the robustness and accuracy of bubble signal recognition and laying a solid foundation for subsequent feature extraction. By extracting the average sound intensity value and the maximum sound intensity value of each burst segment respectively, a sound intensity characteristic parameter set is formed, which can accurately quantify the acoustic energy of bubble burst and provide quantitative support for judging the degree of bubble shielding. The main frequency energy distribution analysis and frequency spectrum construction are introduced, and combined with the resonance peak continuity judgment method, the effective frequency range of the burst event is accurately identified, realizing a multi-dimensional frequency domain characterization of the burst behavior. The sound intensity characteristic parameters and the effective burst frequency range are integrated and analyzed to fully restore the energy and spectral characteristics of the bubble burst process, providing high-dimensional input features for subsequent abnormal bubble accumulation discrimination and shielding correction coefficient generation, thereby enhancing the discrimination ability of the model. With the help of the non-contact characteristics of the acoustic signal, dynamic monitoring of the bubble behavior inside the electrolytic cell is achieved without interfering with the electrolysis process itself, thereby improving the intelligent monitoring capability and maintenance response speed of the electrolysis system. Through a standardized analysis process for acoustic wave events, a joint sound intensity-frequency feature database is established to provide an algorithmic foundation and training data support for the construction of future AI-based abnormal bubble aggregation warning systems and acoustic control feedback mechanisms.

[0027] Preferably, in step S2, judging abnormal bubble accumulation on the bubble burst sound wave signal according to the sound intensity and the burst frequency includes:

[0028] The bubble burst acoustic wave signal is used to identify abnormal bubble accumulation based on the sound intensity and burst frequency. When any of the following conditions occurs, it is determined to be an abnormal local high-density bubble accumulation, and the abnormal local bubble accumulation data is obtained: the sound intensity of the bubble burst acoustic wave signal is greater than 85dB, the bubble burst frequency exceeds 200 times / second, and the burst event duration per unit time exceeds 30 seconds;

[0029] The bubble shielding effect is considered significant and abnormal bubble shielding effect data is obtained when the following conditions occur simultaneously: the amplitude of the sound intensity variation exceeds ±25dB, and the variation period is less than 2 seconds; there is overlapping interference between multiple bubble burst frequencies, resulting in an identification overlap rate exceeding 40%; the detected effective sound signal attenuation rate exceeds 60%, and the duration exceeds 15 minutes;

[0030] When the following conditions occur simultaneously, it is determined to be a bubble accumulation stability imbalance and the bubble accumulation stability imbalance data is obtained: the average burst frequency fluctuates by more than ±35 times / second within 10 minutes, the temperature deviates from the set range by more than ±5°C, the average bubble diameter distribution deviation in the monitoring area exceeds 30%, and the center of gravity shift rate exceeds 0.5mm / s, where the temperature deviation from the set range is 35°C to 55°C;

[0031] Integrate the data of abnormal local bubble accumulation, abnormal bubble shielding effect, and imbalanced bubble accumulation stability to calculate and obtain the abnormal bubble accumulation judgment results;

[0032] The abnormal bubble accumulation judgment result is corrected based on the bubble shielding correction formula to generate a bubble shielding correction coefficient, where the bubble shielding correction formula is as follows:

[0033] C bub =γ1·ΔI+γ2·Δf;

[0034] Where C bub is the bubble shielding correction coefficient, ΔI is the sound intensity deviation, Δf is the frequency shielding amplitude, γ1 and γ2 are the sound intensity and frequency correction weight coefficients respectively.

[0035] The present invention sets up a multi-condition, hierarchical bubble anomaly identification mechanism, and performs multi-dimensional judgment on the sound intensity and rupture frequency data collected during the electrolysis process. It can accurately judge three typical abnormal phenomena, including local high-density bubble aggregation anomaly (such as sound intensity greater than 85dB, frequency greater than 200 times / second, and duration greater than 30 seconds), bubble shielding effect anomaly (such as sound intensity variation greater than ±25dB and period less than 2 seconds, frequency overlap rate greater than 40%, acoustic signal attenuation rate greater than 60%, and duration greater than 15 minutes), and bubble accumulation stability imbalance (such as average rupture frequency fluctuations exceeding ±35 times / second within 10 minutes, temperature deviation from the set range of 35℃~55℃ exceeding ±5℃, average bubble diameter distribution deviation exceeding 30%, and center of gravity offset rate exceeding 0.5mm / s), and generate corresponding abnormal data respectively. On this basis, the above-mentioned multi-source bubble anomaly data are integrated to form a complete bubble abnormal accumulation discrimination result, which solves the problem of single dimension and poor real-time performance in the traditional acoustic analysis of complex bubble behavior recognition. Furthermore, by constructing a correction model formula C based on the sound intensity deviation (ΔI) and the frequency masking amplitude (Δf), bub=γ1·ΔI+γ2·Δf, combining the two-dimensional influence weights of sound intensity and frequency (γ1, γ2) to generate a bubble shielding correction coefficient. This coefficient has good adjustability and physical interpretability, and can provide dynamic correction support at the microscopic level for the current efficiency parameter in the electrolysis energy consumption model, effectively improving the accuracy of energy consumption simulation and robustness in practical applications. In addition, through acoustic non-contact anomaly detection, this method can provide real-time warning of the efficiency loss risk caused by bubble accumulation, providing a solid data foundation and algorithmic framework for subsequent intelligent control systems and bubble behavior prediction models.

[0036] Preferably, in step S3, identifying the electrode corrosion pit density of the electrode surface grayscale image and performing surface roughness corrosion discrimination on the electrode surface grayscale image according to the electrode corrosion pit density includes:

[0037] Calculate the local grayscale variance and texture gradient value of the grayscale image to form an initial corrosion feature response map;

[0038] Detect local low-grayscale concave areas in the corrosion feature response image and mark potential corrosion pit candidate areas;

[0039] Identify the boundary continuity and morphological closure of potential corrosion pit candidate areas and screen out effective corrosion pit areas;

[0040] Extract the number of center points of the effective corrosion pit area and calculate the ratio of the center points to the unit area of ​​the image; obtain the corrosion pit density value data within the unit area;

[0041] The surface roughness corrosion is judged on the grayscale image of the electrode surface according to the density of the electrode corrosion pits, and the surface roughness attenuation factor is generated.

[0042] This method generates a corrosion feature response map by calculating the local grayscale variance and texture gradient of the electrode surface grayscale image. This map effectively highlights the local structural changes in microscopic corrosion regions on the electrode surface. Local low-grayscale depressions are then detected within this feature map, initially marking potential corrosion pit candidates. Further analysis of the boundary continuity and morphological closure of these regions accurately identifies valid corrosion pit regions, effectively eliminating artifacts and background interference, and improving recognition accuracy. Furthermore, the number of center points of all valid corrosion pit regions is extracted, and the corrosion pit density data is calculated based on the image unit area to form a quantifiable corrosion degree index. Finally, the surface roughness corrosion is identified in the electrode surface grayscale image based on this corrosion pit density value, generating a surface roughness attenuation factor that provides a physical basis for subsequent fine-tuning of electrolysis efficiency. This method achieves an objective and standardized representation of electrode corrosion status at the microscale, significantly improving the automation level and evaluation accuracy of corrosion monitoring while avoiding the subjective errors associated with manual interpretation. This method provides key support for the construction of a highly robust and adaptable energy consumption optimization model.

[0043] Preferably, performing surface roughness corrosion determination on the grayscale image of the electrode surface according to the electrode corrosion pit density includes:

[0044] The surface roughness corrosion is judged by the grayscale image of the electrode surface according to the density of the electrode corrosion pits. When the density of the corrosion pits in the grayscale image of the electrode surface exceeds 0.15 / mm 2 , and the area of ​​a single corrosion pit exceeds 0.02mm 2 When , it is judged as light corrosion, and light corrosion discrimination data is obtained;

[0045] When the corrosion pit density in the grayscale image of the electrode surface is between 0.15 and 0.35 per mm 2 , and when the average depth of the corrosion pit is greater than 20 μm, it is judged to be moderate corrosion, and the moderate corrosion discrimination data is obtained;

[0046] When the corrosion pit density on the electrode surface exceeds 0.35 / mm 2 , and the depth of the corrosion pit exceeds 50μm, and the total area of ​​the corrosion pit accounts for more than 10% of the surface area, it is judged as severe corrosion and the severe corrosion discrimination data is obtained;

[0047] Integrate the light corrosion discrimination data, moderate corrosion discrimination data and severe corrosion discrimination data into the corrosion grade discrimination result;

[0048] The corrosion pit density and corrosion grade discrimination results are calculated based on the surface roughness attenuation quantification formula to generate the surface roughness attenuation factor. The surface roughness attenuation quantification formula is as follows:

[0049] D r =α·ρ c +β·σ g ;

[0050] Where D r is the surface roughness attenuation factor, ρ c is the corrosion pit density, σ g is the grayscale standard deviation of the grayscale image on the electrode surface, α and β are the empirical weight coefficients of the corrosion pit density and the grayscale standard deviation, respectively.

[0051] The present invention uses the electrode surface grayscale image to identify surface roughness and corrosion based on the electrode corrosion pit density, constructs a multi-level corrosion recognition mechanism based on image features, and realizes quantitative evaluation of the electrode surface state through parameterized modeling. Specifically, when the corrosion pit density in the electrode surface grayscale image exceeds 0.15 / mm 2 And the area of ​​a single corrosion pit exceeds 0.02mm 2 When the corrosion pit density is between 0.15 and 0.35 per mm, it is judged as mild corrosion and mild corrosion discrimination data is generated; when the corrosion pit density is between 0.15 and 0.35 per mm2 If the average depth of the corrosion pits is greater than 20 μm, it is judged as moderate corrosion and the moderate corrosion discrimination data is obtained; if the corrosion pit density exceeds 0.35 / mm 2 , if the depth of the corrosion pit is greater than 50μm and the total area of ​​the corrosion pit accounts for more than 10% of the electrode surface area, it is judged as severe corrosion and severe corrosion discrimination data is generated. The above multi-level corrosion discrimination standard is based on quantitative threshold setting, which can effectively cover different corrosion stages from initial local pitting to extensive deep erosion. Subsequently, the light, moderate and severe corrosion discrimination data are integrated to form a complete corrosion grade discrimination result. On this basis, the surface roughness attenuation quantification formula is introduced: D r =α·ρ c +β·σ g , where D r is the surface roughness attenuation factor, ρ c is the corrosion pit density, σ g is the grayscale standard deviation of the electrode surface grayscale image, and α and β are empirical weighting coefficients for the corrosion pit density and grayscale standard deviation, respectively. This formula couples density information representing the electrode's microscopic corrosion structure with information about grayscale fluctuations in the macroscopic image, constructing a quantification mechanism for corrosion influencing factors for energy consumption model correction. Overall, this method integrates the grading of corrosion severity, quantitative grayscale image analysis, and generation of surface attenuation factors, providing refined support for energy efficiency assessment and dynamic optimization of electrolysis systems.

[0052] Preferably, step S4 includes the following steps:

[0053] Step S41: normalizing the bubble shielding correction coefficient data and the surface roughness attenuation factor data to generate normalized correction factor data; inputting the normalized correction factor data into the electrolysis energy consumption initial model, dynamically adjusting the current efficiency parameter in the current-voltage efficiency curve, and generating an electrolysis energy consumption optimization model;

[0054] Step S42: performing multi-operating condition simulation energy consumption verification based on the electrolysis energy consumption optimization model to generate simulation energy consumption data;

[0055] Step S43: Calculate the deviation between the simulated energy consumption data and the actual unit product energy consumption value to obtain energy consumption deviation data;

[0056] Step S44: Compare the energy consumption deviation data with the preset deviation threshold. If the energy consumption deviation data is lower than the preset deviation threshold, confirm that the electrolysis energy consumption optimization model is the final electrolysis energy consumption model; if the energy consumption deviation data is higher than or equal to the preset deviation threshold, cyclically adjust the current efficiency parameter in the current-voltage efficiency curve until the energy consumption deviation data is lower than the preset deviation threshold.

[0057] The present invention generates normalized correction factor data of a unified scale by normalizing the bubble shielding correction coefficient data and the surface roughness attenuation factor data; then, the normalized correction factor data is input into the initial model of electrolysis energy consumption, and the current efficiency parameters in the current-voltage efficiency curve are dynamically adjusted to construct an electrolysis energy consumption optimization model. Based on this optimization model, multi-condition simulation energy consumption verification is carried out to generate simulation energy consumption data; then, the deviation between the simulation energy consumption data and the actual unit product energy consumption value is calculated to obtain energy consumption deviation data. Finally, the energy consumption deviation data is compared with a preset deviation threshold. If the deviation is lower than the threshold, the current electrolysis energy consumption optimization model is confirmed to be the final electrolysis energy consumption model; if the deviation is higher than or equal to the threshold, the current efficiency parameters in the current-voltage efficiency curve are cyclically adjusted, and the model is repeatedly optimized until the energy consumption deviation meets the preset threshold requirement. This step effectively realizes the dynamic adaptive correction and accuracy improvement of the model, and significantly improves the reliability and practicality of the electrolysis energy consumption model.

[0058] Preferably, step S42 includes the following steps:

[0059] Step S421: Combining and setting the key variable parameters input into the electrolysis energy consumption optimization model, including five parameters: electrolyte concentration, electrolysis temperature, electrode spacing, anode material conductivity, and electrolytic cell gas pressure intensity, setting three valid interval values ​​for each parameter, and using full factor permutation and combination to generate fixed-dimensional multi-operating condition parameter combination set data;

[0060] Step S422: sequentially importing the multiple operating condition parameter combination set data into the electrolysis energy consumption optimization model, performing simulation calculations group by group, outputting the current-voltage operating response data under each group of operating conditions, and generating a simulation response data set;

[0061] Step S423: performing energy consumption integration processing on each set of current-voltage response curves in the simulation response data set to generate unit product energy consumption data corresponding to each set of working conditions;

[0062] Step S424: Perform a one-to-one mapping between the unit product energy consumption data and the multi-operating condition parameter combination set, and combine the parameter group and the corresponding energy consumption value into energy consumption mapping matrix data in a five-dimensional parameter space; perform a multi-dimensional offset rate analysis on the operating condition energy consumption mapping matrix data, identify abnormal operating condition parameter combinations whose unit product energy consumption value exceeds two standard deviations of the overall operating condition average deviation, and extract them as high-risk energy consumption operating condition data sets;

[0063] Step S425: Classify and archive the unit product energy consumption data and the high-risk energy consumption condition data set to generate simulated energy consumption data.

[0064] The present invention combines and sets the key variable parameters of the electrolysis energy consumption optimization model, involving five parameters: electrolyte concentration, electrolysis temperature, electrode spacing, anode material conductivity, and electrolytic cell gas pressure. Each parameter is assigned three valid intervals, and a full factor permutation and combination method is used to generate a fixed-dimensional multi-condition parameter combination data set. Subsequently, the multi-condition parameter combination data set is sequentially input into the electrolysis energy consumption optimization model. Simulation calculations are performed group by group, outputting current-voltage operating response data for each operating condition to form a simulation response data set. Next, energy consumption integration is performed on each set of current-voltage response curves in the simulation response data set to generate unit product energy consumption data for the corresponding operating condition. The unit product energy consumption data is then mapped one-to-one with the multi-condition parameter combination set to form an energy consumption mapping matrix in a five-dimensional parameter space. Multidimensional offset rate analysis is used to identify abnormal operating condition parameter combinations whose unit product energy consumption exceeds two standard deviations of the overall operating condition mean and extract them as a high-risk energy consumption condition data set. Finally, the unit product energy consumption data and the high-risk energy consumption condition data set are classified and archived to generate a complete simulation energy consumption data set. This step achieves comprehensive simulation verification of the energy consumption performance under multivariable conditions of the electrolysis process and identification of risk conditions, which helps to accurately evaluate and optimize the model.

[0065] In this specification, a system for constructing an electrolysis energy consumption model is provided, which is used to execute the above-mentioned method for constructing an electrolysis energy consumption model. The system for constructing an electrolysis energy consumption model includes:

[0066] A model building module is used to use a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature and electrolysis time; construct an initial electrolysis energy consumption model based on the electrolysis process parameters, and input the electrolysis process parameters into the initial electrolysis energy consumption model to output unit product energy consumption value and current-voltage efficiency curve;

[0067] The bubble analysis module is used to acoustically monitor the electrolytic bubbles on the side wall of the electrolytic cell to obtain the bubble burst sound wave signal; analyze the sound intensity and burst frequency of the bubble burst sound wave signal; and identify abnormal bubble accumulation based on the sound intensity and burst frequency of the bubble burst sound wave signal to generate a bubble shielding correction coefficient;

[0068] The corrosion analysis module is used to obtain a microscopic image of the electrode surface using an industrial endoscope; perform grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identify the electrode corrosion pit density in the grayscale image of the electrode surface, and perform surface roughness corrosion judgment on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor;

[0069] The model optimization module is used to input the bubble shielding correction coefficient and the surface roughness attenuation factor into the initial electrolysis energy consumption model to dynamically correct the current efficiency parameters of the current-voltage efficiency curve to generate an electrolysis energy consumption optimization model; the electrolysis energy consumption optimization model is simulated and verified for energy consumption. When the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, the final electrolysis energy consumption model is generated.

[0070] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for constructing the electrolysis energy consumption model as described above is implemented.

[0071] The beneficial effect of the present invention is that by collecting key electrolysis process parameters such as current, voltage, electrolyte temperature and electrolysis time through multimodal sensors, comprehensive monitoring of the electrolysis process is achieved, and the data is rich and accurate, providing a solid foundation for subsequent model construction. By combining the actual collected parameters to construct an initial electrolysis energy consumption model, the unit product energy consumption and current-voltage efficiency curve can be output in real time, facilitating preliminary evaluation and analysis of the energy consumption performance of the electrolysis process. By analyzing the sound intensity and rupture frequency of the bubble bursting acoustic signal on the side wall of the electrolytic cell, the abnormal bubble accumulation is accurately judged, and a bubble shielding correction coefficient is generated to achieve dynamic compensation for the efficiency loss caused by bubble shielding in the electrolysis process. Using an industrial endoscope to obtain electrode microscopic images, grayscale gradient analysis and corrosion pit density identification are performed to quantitatively judge the rough corrosion state of the electrode surface, generate a surface roughness attenuation factor, and achieve quantitative correction of the impact of electrode surface corrosion on electrolysis efficiency. By inputting the bubble shielding correction coefficient and surface roughness attenuation factor into the initial energy consumption model, the current efficiency parameters are dynamically corrected, achieving real-time optimization of the electrolysis energy consumption model. The error between the simulated energy consumption and the actual unit product energy consumption is controlled within a preset range, ensuring that the model has high prediction accuracy and applicability. The final electrolysis energy consumption model provided by the system can more accurately reflect the energy consumption changes in the actual electrolysis process, help to achieve precise regulation of electrolysis process parameters, thereby effectively reducing energy consumption and improving production efficiency, with good economic and environmental benefits. Real-time monitoring of abnormal bubble accumulation and electrode corrosion status provides an important basis for equipment maintenance, detects potential anomalies in advance, reduces equipment failure rate, extends electrode life, and improves the stability and reliability of the overall operation of the electrolysis system. Therefore, the present invention breaks through the single data and static defects of the traditional electrolysis energy consumption model through multimodal sensors, multidimensional signal analysis and dynamic parameter correction, and realizes high-precision dynamic modeling and optimization of energy consumption in the electrolysis process. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 A schematic flow chart of a method for constructing an electrolysis energy consumption model;

[0073] Figure 2 for Figure 1Detailed implementation steps of step S4 in FIG.

[0074] Figure 3 for Figure 2 Detailed implementation steps of step S42 are shown in the flowchart;

[0075] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0076] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0077] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0078] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0079] To achieve this, please refer to Figures 1 to 3 , a method for constructing an electrolysis energy consumption model, the method comprising the following steps:

[0080] Step S1: using a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature, and electrolysis time; constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, and inputting the electrolysis process parameters into the initial electrolysis energy consumption model to output a unit product energy consumption value and a current-voltage efficiency curve;

[0081] Step S2: Performing electrolytic bubble acoustic monitoring on the side wall of the electrolytic cell to obtain a bubble burst acoustic wave signal; analyzing the sound intensity and burst frequency of the bubble burst acoustic wave signal; performing abnormal bubble accumulation judgment on the bubble burst acoustic wave signal based on the sound intensity and burst frequency, and generating a bubble shielding correction coefficient;

[0082] Step S3: using an industrial endoscope to obtain a microscopic image of the electrode surface; performing grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identifying the electrode corrosion pit density in the grayscale image of the electrode surface, and performing surface roughness corrosion discrimination on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor;

[0083] Step S4: Input the bubble shielding correction coefficient and the surface roughness attenuation factor into the current efficiency parameter of the dynamic correction current-voltage efficiency curve in the initial electrolysis energy consumption model to generate an electrolysis energy consumption optimization model; perform simulation energy consumption verification on the electrolysis energy consumption optimization model, and when the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, generate the final electrolysis energy consumption model.

[0084] The present invention constructs a basic energy consumption assessment framework by inputting core process parameters such as current, voltage, temperature, and electrolysis time into the initial energy consumption model, and then introduces microscopic influencing factors such as bubbles and corrosion through acoustic monitoring and image recognition to achieve more accurate modeling of the complex coupling relationship of the electrolysis process. By means of the sound intensity and rupture frequency analysis of the bubble burst acoustic signal, the abnormal accumulation of bubbles in the electrolysis process is recognized in real time, and a bubble shielding correction coefficient is generated to effectively correct the influence of bubbles on current efficiency and improve the credibility of the electrolysis efficiency parameters. Microscopic images are obtained through an industrial endoscope, and combined with grayscale gradient analysis technology, the density of electrode corrosion pits is identified, and then the surface roughness of the electrode is judged, and a surface roughness attenuation factor is generated to achieve modeling feedback on the effect of electrode aging on energy consumption changes. The bubble shielding correction coefficient and the surface roughness attenuation factor are introduced into the initial model for dynamic correction to construct an electrolysis energy consumption optimization model, so that the model has the ability to respond to changes in process status in real time, thereby improving the dynamic adaptability and predictive performance of the modeling. By verifying the deviation between the unit product energy consumption value output by the model and the actual value through simulation, and outputting the final energy consumption model when the deviation falls below a set threshold, this model can provide a reliable basis for energy-saving control and optimized operation of the electrolysis process, and has promising industrial application prospects. Therefore, through multimodal sensors, multidimensional signal analysis, and dynamic parameter correction, this invention overcomes the data-single and static limitations of traditional electrolysis energy consumption models, achieving high-precision dynamic modeling and optimization of electrolysis process energy consumption.

[0085] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart showing the steps of a method for constructing an electrolysis energy consumption model according to the present invention. In this example, the method for constructing an electrolysis energy consumption model includes the following steps:

[0086] Step S1: using a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature, and electrolysis time; constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, and inputting the electrolysis process parameters into the initial electrolysis energy consumption model to output a unit product energy consumption value and a current-voltage efficiency curve;

[0087] Step S2: Performing electrolytic bubble acoustic monitoring on the side wall of the electrolytic cell to obtain a bubble burst acoustic wave signal; analyzing the sound intensity and burst frequency of the bubble burst acoustic wave signal; performing abnormal bubble accumulation judgment on the bubble burst acoustic wave signal based on the sound intensity and burst frequency, and generating a bubble shielding correction coefficient;

[0088] Step S3: using an industrial endoscope to obtain a microscopic image of the electrode surface; performing grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identifying the electrode corrosion pit density in the grayscale image of the electrode surface, and performing surface roughness corrosion discrimination on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor;

[0089] Step S4: Input the bubble shielding correction coefficient and the surface roughness attenuation factor into the current efficiency parameter of the dynamic correction current-voltage efficiency curve in the initial electrolysis energy consumption model to generate an electrolysis energy consumption optimization model; perform simulation energy consumption verification on the electrolysis energy consumption optimization model, and when the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, generate the final electrolysis energy consumption model.

[0090] In an embodiment of the present invention, a multimodal sensor system is used to simultaneously collect key parameters in the electrolysis process, including electrolysis current, electrolysis voltage, electrolyte temperature and electrolysis time. The current value is measured in real time by a current transformer with a measurement range of 0 to 1000A and a sampling frequency set to 1kHz to ensure that subtle fluctuations in current changes are captured. The voltage value is collected using a high-precision voltage sensor with a measurement range of 0 to 30V, a resolution of 0.01V, and a sampling frequency of 1kHz. The electrolyte temperature is collected by a corrosion-resistant PT100 platinum resistance temperature sensor with a measurement accuracy of ±0.1°C. It is installed on the side wall of the electrolytic cell near the electrode and has a sampling frequency of 1Hz. The electrolysis time is recorded by the system clock, from the start to the end of electrolysis, with units accurate to seconds. The collected parameter data is centrally transmitted to the industrial control computing platform via a high-speed data acquisition card. Based on the collected current, voltage, temperature and time data, an initial model of electrolysis energy consumption is established, and the energy consumption value per unit product is calculated using the formula: Where U(t) is the time-varying voltage, I(t) is the time-varying current, the integration time is the electrolysis time T, and M is the mass of product produced per unit time (in kg). The mass of product is measured using a flow meter and a weight sensor. The current-voltage efficiency curve is generated by fitting discretely sampled current and voltage data points using the least squares method, with a point spacing of 0.1 A. All sensor data are time-stamped to ensure accurate data correspondence. A highly sensitive ultrasonic sensor array is fixed to the sidewall of the electrolytic cell to capture the acoustic signal generated by bubble collapse. The ultrasonic sensor has a frequency response range of 20 kHz to 200 kHz, and the sampling rate is set to 500 kHz to ensure that the complete spectral information of the bubble collapse sound wave is captured. The collected acoustic signal is filtered through a bandpass filter to remove noise below 20 kHz and above 200 kHz. The time-frequency characteristics of the signal are further analyzed using short-time Fourier transform (STFT). The sound intensity index of the rupture sound wave is extracted from the sound wave signal. The signal envelope peak detection method is used to calculate the instantaneous sound intensity of the rupture sound wave in dB SPL (sound pressure level). The sound intensity threshold is set to 80dB SPL to eliminate background noise. The rupture frequency is obtained by analyzing the peak occurrence frequency of the short-time energy envelope. The calculation window length is 100ms and the step length is 10ms. The sound intensity and rupture frequency are used as input to judge the abnormal accumulation of bubbles. The abnormal judgment threshold is set to an abnormal accumulation state when the rupture frequency exceeds 50 times / second and the sound intensity exceeds 95dB SPL. Based on the abnormal accumulation judgment result, the bubble shielding correction coefficient is calculated using the linear mapping method: C bub =γ1·ΔI+γ2·Δf; where, C bub is the bubble masking correction factor, ΔI is the sound intensity deviation, Δf is the frequency masking amplitude, and γ1 and γ2 are the sound intensity and frequency correction weighting coefficients, respectively. Microscopic images of the electrode surface were acquired using an industrial endoscope with at least 500x optical magnification and a resolution of 1920 × 1080 pixels. The endoscope was equipped with an LED ring illumination to ensure uniform, shadow-free imaging. Multiple image sequences covering the entire electrode surface were acquired at a perpendicular angle to the electrode surface. The images were formatted as lossless TIFFs. The acquired images were preprocessed, including grayscale conversion and denoising. A Gaussian filter (standard deviation σ = 1.5) was used to remove imaging noise. The gradient of the grayscale image was calculated using the Sobel operator for both horizontal and vertical gradients, resulting in a gradient magnitude image. The gradient magnitude was used to enhance the edges of electrode corrosion pits in the image. The corrosion pit region was extracted using a threshold segmentation method with a threshold of 0.6 times the maximum grayscale gradient. The corrosion pit density was calculated based on a connected domain analysis of the corrosion pit region, expressed as the number of pits per square millimeter. The corrosion pit density exceeds 30 / mm 2 The surface roughness corrosion phenomenon is determined when the surface roughness attenuation factor is calculated based on the corrosion pit density, using the exponential attenuation function: Dr =α·ρ c +β·σ g Where, D r is the surface roughness attenuation factor, ρ c is the corrosion pit density, σ g is the grayscale standard deviation of the grayscale image of the electrode surface, and α and β are the empirical weight coefficients of the corrosion pit density and the grayscale standard deviation, respectively. The bubble shielding correction coefficient obtained in step S2 and the surface roughness attenuation factor obtained in step S3 are used as input parameters to introduce the dynamic correction of the current efficiency parameters of the initial model of electrolysis energy consumption. The unit product energy consumption value of the electrolysis process is calculated based on the corrected current efficiency parameters through the simulation platform. The simulation adopts the numerical integration method and the time step is set to 1 second. The simulation results are compared with the actual measured unit product energy consumption value, and the calculated error is the absolute value of the difference between the two. When the error is less than 0.5%, it is considered to meet the preset standard. When the error meets the conditions, the current electrolysis energy consumption model is confirmed to be the final model; if the error exceeds 0.5%, the correction coefficient weight is readjusted and the correction process is repeated until the error standard is met to ensure that the parameters of the electrolysis energy consumption model accurately reflect the actual process.

[0091] Preferably, step S1 includes the following steps:

[0092] Step S11: using a current sensor to collect the current value of the electrolysis process;

[0093] Step S12: collecting the voltage value of the electrolysis process through the voltage collection module;

[0094] Step S13: using a thermosensitive element to obtain the electrolyte temperature during the electrolysis process;

[0095] Step S14: obtaining the electrolysis time of the electrolysis process based on the timing module;

[0096] Step S15: synchronously scheduling and controlling the current value, voltage value, electrolyte temperature, and electrolysis time to obtain the original parameters of the electrolysis process; performing data preprocessing on the original parameters of the electrolysis process to generate electrolysis process parameters, wherein the data preprocessing includes data denoising, outlier removal, time synchronization calibration, and data standardization;

[0097] Step S16: constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, inputting the electrolysis process parameters into the initial electrolysis energy consumption model to calculate energy consumption, and outputting the unit product energy consumption value and the current-voltage efficiency curve.

[0098] In this embodiment of the present invention, a high-precision current sensor is used to collect current during the electrolysis process. The current sensor used is a Hall-effect current sensor with a measurement range of 0 to 1000 amperes and an accuracy of 0.5%. The sensor is fixedly mounted on the current loop of the electrolysis equipment to ensure accurate sensing of the current passing through the sensor. The sampling frequency is set to 1000 times per second (1kHz) to capture instantaneous current changes in real time. The collected current signal is transmitted to the data acquisition system via a shielded cable to ensure signal integrity and interference-free. During the acquisition process, the sensor and circuitry must be calibrated using a known standard current source to ensure the accuracy and stability of the measured data. Voltage acquisition uses a high-precision voltage sensor module with a range of 0 to 30 volts and a resolution of 0.01 volt. The sensor is directly connected to the two ends of the electrolytic cell, and an isolated acquisition circuit ensures signal security and isolation. The sampling frequency is also set to 1kHz to ensure that the collected data is synchronized with the current sampling frequency. The voltage sensor is calibrated using a standard voltage source before installation to ensure that the output voltage value is consistent with the actual value. The collected signal uses differential sampling to reduce environmental electromagnetic interference. Voltage data is transmitted to the control unit via a high-speed data acquisition card for real-time monitoring. Electrolyte temperature is acquired using a corrosion-resistant platinum resistance temperature sensor (PT100). The sensor is placed in the electrolytic cell liquid area, avoiding areas where bubbles are generated, to ensure accurate temperature measurement. The sensor's measurement range is -10 to 100 degrees Celsius, with an accuracy of ±0.1 degrees Celsius. Temperature data acquisition is set to once per second to ensure stable temperature changes. The sensor signal is connected to the temperature acquisition module via a four-wire connection to reduce errors caused by cable resistance. The temperature acquisition system is regularly calibrated using an ice-water mixing tank and a constant-temperature water bath for both low and high temperature calibration. Electrolysis time is captured using an embedded timing module, which utilizes a high-precision real-time clock chip with millisecond-level accuracy. The timing module initiates the timing signal at the start of the electrolysis process and stops at its end. The timing module is synchronously connected to the acquisition system, ensuring that the current, voltage, and temperature data are time-stamped, facilitating subsequent unified data management and processing. The timing module's time base is regularly synchronized with an external standard time source to ensure accurate and consistent time recording. Current, voltage, temperature, and time data are centrally accessed and synchronized by a real-time data acquisition system. A central controller aligns the sensor data streams according to their timestamps, ensuring that all parameters correspond to the same time point. The raw data is preprocessed by first using a low-pass filter to remove high-frequency noise, with the filter cutoff frequency set to 200 Hz. Statistical methods are then used to detect outliers. Specifically, a sliding window method is used to calculate the mean and standard deviation of the data. Outliers are defined as data points exceeding three times the standard deviation from the mean. These outliers are removed and replaced with the mean of the adjacent normal data. Data time synchronization calibration ensures consistent timing between all data by aligning the timestamps of all data.Finally, a standardization method is applied to map the current, voltage, temperature and time data to the same numerical interval, and the standardization interval is set to 0 to 1 for subsequent model processing. Using the pre-processed current, voltage, temperature and time parameters, combined with product output data, an initial calculation framework for electrolysis energy consumption is constructed. Energy consumption is calculated by integrating the product of current and voltage over time, and combining the output data to calculate the power consumption per unit product. After smoothing the current and voltage data, a current-voltage efficiency curve is drawn. The efficiency curve uses a discrete point fitting method to form an efficiency change trend through multi-point sampling, and the sampling interval is set to one set of data every 0.1 second during the electrolysis process. This efficiency curve reflects the energy conversion efficiency under different current and voltage conditions and is output as the basic parameter of the electrolysis energy consumption model. During the calculation process, time synchronization accuracy and data integrity are ensured, and real-time performance and accuracy are guaranteed through multi-threaded data processing.

[0099] Preferably, analyzing the sound intensity and burst frequency of the bubble burst sound wave signal in step S2 includes:

[0100] Extract the original bubble burst acoustic wave signal collected during the electrolysis process;

[0101] Calculate the instantaneous sound pressure value of the bubble sound wave signal and obtain the sound pressure envelope curve;

[0102] Identify the peak area in the sound pressure envelope curve and extract potential bubble collapse event fragments;

[0103] Detect the average sound intensity value and the maximum sound intensity value of each potential bubble burst event segment to generate sound intensity characteristic parameters;

[0104] Calculate the main frequency energy distribution in the rupture event fragment, extract the main frequency points and generate the frequency spectrum;

[0105] Identify the area with continuous resonance peaks in the frequency spectrum and determine it as the effective rupture frequency range;

[0106] The sound intensity characteristic parameters of each rupture event and the effective rupture frequency range are integrated to obtain the sound intensity and rupture frequency of the complete bubble rupture acoustic wave signal.

[0107] In an embodiment of the present invention, raw sound pressure signals over continuous time periods are extracted from raw bubble burst acoustic signals collected by a high-sensitivity underwater acoustic sensor (such as a piezoelectric sensor) mounted on the sidewall of the electrolytic cell. The acoustic sensor sampling frequency is set to 200 kHz to ensure that the high-frequency transient acoustic wave characteristics of bubble burst are captured. The collected raw signal is preprocessed with an anti-aliasing filter to remove noise components above the sensor bandwidth before being transmitted to a digital signal processing unit. Within the digital signal processing unit, short-term envelope detection technology is used to calculate the instantaneous sound pressure value of the bubble acoustic signal. Specifically, a Hilbert transform is used to extract the envelope of the acoustic signal to obtain a sound pressure envelope curve. This envelope curve reflects the temporal trend of acoustic wave energy and can clearly locate the instantaneous high-energy region of bubble burst. By setting a threshold for the sound pressure envelope curve (for example, 20% of the envelope maximum value), peak regions in the envelope curve are automatically identified. Each peak region corresponds to a potential bubble burst event. Peak regions are divided into independent burst event segments based on peak width and height. For each event segment, the average sound intensity (i.e., the mean of the sound pressure envelope within that time period) and the maximum sound intensity (envelope peak) are calculated to form a set of sound intensity characteristic parameters. Furthermore, the acoustic wave signal of each potential rupture event segment is subjected to frequency domain analysis. Using a fast Fourier transform (FFT), the frequency spectrum of the sound wave is calculated within the event segment. The frequency resolution is set to 1kHz to capture the main energy distribution. The main frequency point in the spectrum graph, i.e., the frequency with the highest energy peak in the spectrum, is extracted as the main frequency of the rupture event. The frequency spectrum graph is analyzed for resonant peaks within a continuous frequency range, and multiple continuous peak regions in the frequency spectrum are identified. These continuous peak regions are determined to be valid rupture frequency ranges by filtering out invalid noise peaks by setting an energy threshold (e.g., peak energy greater than 15% of the maximum energy of the spectrum). Finally, the sound intensity characteristic parameters (average sound intensity, maximum sound intensity) of each rupture event are integrated with the corresponding valid rupture frequency range to form a complete description of the bubble rupture acoustic wave signal sound intensity and rupture frequency. This data is used in subsequent discriminant analysis of abnormal bubble accumulation to ensure that the time-frequency characteristics of the rupture event are accurately captured and quantified.

[0108] Of particular importance is the average and maximum sound intensity values ​​for each segment detecting potential bubble collapse events, including:

[0109] The raw waveform data collected from the acoustic sensor network deployed in the bubble burst area is processed by window sliding to extract the multi-channel acoustic wave signals within the 5ms time window before and after the perturbation, generating the perturbation window acoustic wave signal data.

[0110] Perform multi-channel waveform inversion based on time inversion on the perturbation window acoustic wave signal data to determine the initial propagation path structure of the sound source;

[0111] The perturbation window acoustic wave signal data is mapped to the frequency domain according to the initial propagation path structure of the sound source. The frequency distribution corresponding to each sound source point is subjected to low-pass band guided filtering and tensor flux aggregation projection reconstruction using the tensor-flux bitmap superposition algorithm to generate the bubble burst main frequency spatial reconstruction data.

[0112] The main frequency peak is extracted and the main lobe angle is estimated for the spatial reconstruction data of the main frequency of bubble bursting to generate the main frequency of bubble bursting and its spatial beam pattern data;

[0113] The average sound intensity and maximum sound intensity of each potential bubble burst event segment within the main frequency range are calculated using the bubble burst main frequency and its spatial beam pattern data.

[0114] In an embodiment of the present invention, the original multi-channel acoustic signal data collected by the acoustic sensor network deployed inside the bubble burst reaction area is used. For each potential bubble burst event segment identified by the sound pressure envelope peak, the sliding window technology is used to perform segmented extraction processing on the time period in which it is located. With the sound pressure perturbation mutation point as the center, it is extended forward and backward by 5ms respectively to form a perturbation time window with a total duration of 10ms, and the acoustic wave signals of all channels in the time window are synchronously extracted to generate perturbation window acoustic wave signal data. Subsequently, the time-reversal acoustic wave inversion algorithm (Time-Reversal Acoustic Inversion) is applied to the perturbation window acoustic wave signal data. This method utilizes the reverse propagation characteristics of each channel signal on the time axis, and propagates it back to the sound source point area through numerical simulation, thereby constructing the sound source topological flow field model data. This model not only describes the initial position of each sound source and its sound energy diffusion path, but also can characterize the wavefront interference and reflection behavior, thereby accurately characterizing the actual spatial propagation structure of the bubble burst sound source. On the basis of establishing the sound source propagation path structure, the perturbation window acoustic wave signal data is frequency-domain mapped. The original time-domain waveform data is spectrally expanded using a multi-dimensional Fourier transform, and the spectrum information is directional filtered and energy normalized using the Tensor-Flux Overlay Mapping Algorithm. Specifically, after filtering out high-frequency disturbances and background noise, an energy focusing operation is performed on the frequency distribution of each sound source point in the 0-20kHz low-pass frequency band, and the flux superposition process is used to cumulatively reconstruct the energy field intensity in the same frequency band to generate the spatial reconstruction data of the bubble burst main frequency. Based on the above spatial reconstruction results, the main frequency peak extraction process is further performed to identify the main frequency interval with the highest energy concentration, and the main lobe angle (Main Lobe Angle) in this interval is calculated to generate spatial beam diagram data containing the spatial energy distribution pattern and frequency pointing characteristics of the bubble burst event. This beam diagram is used to reflect the sound intensity distribution trend of each burst sound source in different directions. Finally, within the identified main frequency interval, the spatial beam pattern data for each potential bubble burst event segment is integrated and peak-extracted to calculate the average and maximum sound intensity values ​​for the burst event within the main frequency interval. The average sound intensity is calculated through weighted integration of all channels within the main frequency interval, reflecting the overall energy distribution trend of the event. The maximum sound intensity is directly derived from the maximum energy peak in the main lobe direction of the spatial beam pattern, reflecting the instantaneous extreme sound pressure of the bubble burst. These sound intensity parameters, as core acoustic indicators, will be used in subsequent steps to accurately determine the bubble aggregation state and conduct modeling analysis.

[0115] Of particular importance is the multi-channel waveform inversion based on time reversal of the perturbation window acoustic wave signal data to determine the initial propagation path structure of the sound source, which also includes:

[0116] Perform multi-channel time reversal on the perturbation window acoustic wave signal data to generate inverted propagation wavefield data;

[0117] Perform wavefront focusing detection on the inverted propagation wavefield data to identify the center of acoustic wave energy convergence and generate candidate sound source point data;

[0118] The time series propagation trajectory of the candidate sound source point data is traced back, the initial propagation path is reconstructed, and the initial propagation path structure data of the sound source is generated;

[0119] The path stability of the initial sound source propagation path structure data is verified, the credible path segments are screened, and the backbone data of the sound source propagation path is generated.

[0120] In an embodiment of the present invention, a multi-channel synchronous time reversal operation is performed on the extracted perturbation window acoustic wave signal data. Specifically, the acoustic wave signals collected by each acoustic sensor are reversed along the time axis to form a time-reversed signal sequence. Subsequently, the inverted multi-channel signals are input into a preset acoustic field propagation model. Using the finite-difference time-domain method (FDTD) or finite element simulation, the signals are reversely propagated in a reconstructed medium environment to generate inversion propagation wavefield data. This wavefield data reflects the dynamic behavior of the focusing evolution of the sound source energy during the spatiotemporal inversion process. A three-dimensional wavefront focusing analysis is performed on the generated inversion propagation wavefield data to detect the spatial locations where the acoustic wave energy converges. By calculating the rate of change of the energy density gradient at each spatial node in the wavefield, the local energy peak regions formed during the inversion propagation process are identified. The locations of these energy convergence regions are identified as potential sound source points and output as sound source candidate point data. For each sound source candidate point, the time series trajectory of its wavefront expansion is traced back. The specific method is to track the coherent wavefront propagation path in the inverted wave field, combine the geometric layout of the original sensor network and the time-reversed propagation velocity parameters, and gradually reconstruct the spatial path of the sound wave propagating from the candidate point to the sensor array. Each backtracking path uses phase consistency and wavefront coherence as constraints to generate the initial propagation path structure data corresponding to the candidate point. The path stability of all generated initial propagation path structure data is verified. By statistically analyzing indicators such as the overlap, consistent phase delay, and propagation direction stability between the paths of different candidate sound source points, propagation path segments with high stability and high matching are screened out. These verified path segments are extracted and integrated into the sound source propagation path backbone data, which is the final credible sound source initial propagation path structure for subsequent frequency domain mapping and main frequency space reconstruction.

[0121] Preferably, in step S2, judging abnormal bubble accumulation on the bubble burst sound wave signal according to the sound intensity and the burst frequency includes:

[0122] The bubble burst acoustic wave signal is used to identify abnormal bubble accumulation based on the sound intensity and burst frequency. When any of the following conditions occurs, it is determined to be an abnormal local high-density bubble accumulation, and the abnormal local bubble accumulation data is obtained: the sound intensity of the bubble burst acoustic wave signal is greater than 85dB, the bubble burst frequency exceeds 200 times / second, and the burst event duration per unit time exceeds 30 seconds;

[0123] The bubble shielding effect is considered significant and abnormal bubble shielding effect data is obtained when the following conditions occur simultaneously: the amplitude of the sound intensity variation exceeds ±25dB, and the variation period is less than 2 seconds; there is overlapping interference between multiple bubble burst frequencies, resulting in an identification overlap rate exceeding 40%; the detected effective sound signal attenuation rate exceeds 60%, and the duration exceeds 15 minutes;

[0124] When the following conditions occur simultaneously, it is determined to be a bubble accumulation stability imbalance and the bubble accumulation stability imbalance data is obtained: the average burst frequency fluctuates by more than ±35 times / second within 10 minutes, the temperature deviates from the set range by more than ±5°C, the average bubble diameter distribution deviation in the monitoring area exceeds 30%, and the center of gravity shift rate exceeds 0.5mm / s, where the temperature deviation from the set range is 35°C to 55°C;

[0125] Integrate the data of abnormal local bubble accumulation, abnormal bubble shielding effect, and imbalanced bubble accumulation stability to calculate and obtain the abnormal bubble accumulation judgment results;

[0126] The abnormal bubble accumulation judgment result is corrected based on the bubble shielding correction formula to generate a bubble shielding correction coefficient, where the bubble shielding correction formula is as follows:

[0127] C bub =γ1·ΔI+γ2·Δf;

[0128] Where C bub is the bubble shielding correction coefficient, ΔI is the sound intensity deviation, Δf is the frequency shielding amplitude, γ1 and γ2 are the sound intensity and frequency correction weight coefficients respectively.

[0129] In an embodiment of the present invention, the peak sound intensity and rupture frequency of the bubble burst event are counted second by second. If the instantaneous sound intensity of the bubble burst sound wave signal exceeds 85 decibels, and the number of bubble bursts per unit time (1 second) exceeds 200 times, and the duration of the continuous detection of the rupture event exceeds 30 seconds, it is determined that a local high-density accumulation of bubbles occurs in the area. The abnormal data is recorded in the form of information such as timestamp, maximum sound intensity value, maximum rupture frequency, and duration during the abnormal period to generate bubble local accumulation abnormal data. The amplitude of the sound intensity change over time is monitored, and its maximum amplitude of change within the sliding time window (2 seconds) is calculated. If the amplitude of the sound intensity change exceeds plus or minus 25 decibels, and the change period is shorter than 2 seconds, and the overlap rate of multiple bubble burst frequencies detected at the same time exceeds 40% (that is, the frequency spectrum overlap of different rupture events accounts for more than 40%), accompanied by an effective sound signal attenuation rate of more than 60%, and the duration of this state reaches or exceeds 15 minutes, it is determined that the bubble shielding effect is significant. The effective sound signal attenuation rate is calculated by comparing the signal energy attenuation before and after shielding. The abnormal state information is organized into bubble shielding effect abnormal data, including indicators such as time period, maximum sound intensity change amplitude, overlap rate, and signal attenuation rate. With a 10-minute observation window, the average value and fluctuation range of the bubble burst frequency are statistically calculated. If the average burst frequency fluctuates by more than plus or minus 35 times per second, and the regional temperature deviates from the preset safety range (35 degrees Celsius to 55 degrees Celsius) by more than plus or minus 5 degrees Celsius, the bubble diameter distribution deviation in the monitoring area exceeds 30%, and the bubble center of gravity offset rate exceeds 0.5 mm per second, then the bubble accumulation stability is judged to be unbalanced. The temperature is collected in real time by thermistors, and the bubble diameter distribution and center of gravity offset are obtained by the acoustic image analysis system. The abnormal judgment result is stored as bubble accumulation stability imbalance data. The above-mentioned bubble local aggregation abnormal data, bubble shielding effect abnormal data, and bubble accumulation stability imbalance data are time-series integrated, and the temporal and spatial overlapping parts of the abnormal events are normalized. A weighted fusion algorithm is used to calculate the comprehensive bubble abnormal accumulation judgment result. The weight can be set according to the degree of influence of the anomaly type on the model, for example, the local aggregation weight is 0.4, the shielding effect weight is 0.35, and the stability imbalance weight is 0.25. The fusion result includes the anomaly intensity level, the anomaly time period and the spatial distribution. Based on the fused anomaly accumulation judgment result, the bubble shielding correction formula is used to calculate the correction coefficient. In the specific operation, the calculated sound intensity deviation is the difference between the current detection sound intensity and the normal reference sound intensity (unit: decibel), and the frequency shielding amplitude is the degree of contraction of the effective rupture frequency range, both of which are used as input variables. The preset sound intensity correction weight coefficient γ1 and the frequency correction weight coefficient γ2 are 0.6 and 0.4 respectively to ensure that the impact of the sound intensity change is more significant. Through weighted calculation, the final bubble shielding correction coefficient is obtained, which is used for the dynamic correction of the subsequent electrolysis energy consumption model.

[0130] Preferably, in step S3, identifying the electrode corrosion pit density of the electrode surface grayscale image and performing surface roughness corrosion discrimination on the electrode surface grayscale image according to the electrode corrosion pit density includes:

[0131] Calculate the local grayscale variance and texture gradient value of the grayscale image to form an initial corrosion feature response map;

[0132] Detect local low-grayscale concave areas in the corrosion feature response image and mark potential corrosion pit candidate areas;

[0133] Identify the boundary continuity and morphological closure of potential corrosion pit candidate areas and screen out effective corrosion pit areas;

[0134] Extract the number of center points of the effective corrosion pit area and calculate the ratio of the center points to the unit area of ​​the image; obtain the corrosion pit density value data within the unit area;

[0135] The surface roughness corrosion is judged on the grayscale image of the electrode surface according to the density of the electrode corrosion pits, and the surface roughness attenuation factor is generated.

[0136] In an embodiment of the present invention, an industrial endoscope is used to collect a microscopic grayscale image of the electrode surface, and the image resolution is set to at least 2048×2048 pixels to ensure that the details are fully displayed. The image is stored in grayscale format, and the grayscale range is 0 to 255. For the collected grayscale image, a sliding window method is used, and the window size is set to 31×31 pixels. The local grayscale variance is calculated for the grayscale values ​​in the window. This variance is used to reflect the degree of discreteness of the local grayscale. The area with higher local variance usually corresponds to the surface roughness or corrosion pit location. Secondly, the texture gradient calculation based on the Sobel operator is used to obtain the directionality and intensity of the grayscale change in the image. The texture gradient amplitude is used to describe the edge characteristics of the surface texture. The local grayscale variance map is fused with the texture gradient map, and the initial corrosion feature response map is generated according to the weighted average method. The weights are 0.6 for the local grayscale variance weight and 0.4 for the texture gradient weight. A region growing algorithm is applied to the initial corrosion feature response map, focusing on detecting concave areas with low grayscale response. The grayscale threshold is set to less than 60 (within the grayscale range of 0-255) to locate possible corrosion pits. Morphological closing operations are used to fill small holes in the detected regions, maintaining regional continuity. The area of ​​each connected domain is calculated, and noisy regions smaller than 50 pixels are removed. The remaining regions are marked as potential corrosion pit candidates. Edge detection algorithms (such as the Canny operator) are used to extract the boundaries of the candidate regions and evaluate their continuity. The continuity threshold is set to 90% or more continuous edges. The morphological closure of the regions is also calculated, defined as the ratio of the region boundary perimeter to its contained area. The closure threshold is set to less than 1.3. Only corrosion pits that meet the boundary continuity and morphological closure requirements are retained, while non-corrosion morphological depressions are excluded. The center points of valid corrosion pit regions are counted, obtained by calculating the region centroid. The unit area is defined as per million pixels (1024×1024 pixels), and the number of corrosion pits is normalized to the corrosion pit density per unit area. The corrosion pit density is expressed as the number of corrosion pits per million pixels, and the value range is usually between 0 and 500. Based on the corrosion pit density data, surface roughness corrosion judgment is performed. The corrosion pit density judgment threshold is set to 200 corrosion pits per million pixels. Grayscale images with a density higher than this threshold are judged as severely corroded, and the corresponding surface roughness attenuation factor is generated. The attenuation factor value is determined according to the linear mapping relationship of the corrosion pit density, and the value range is 0.3 to 1.0. The specific mapping is: when the corrosion pit density is less than 50, the attenuation factor is 1.0 (no corrosion effect); when the density is between 50 and 200, the attenuation factor decreases linearly in proportion; when the density exceeds 200, the attenuation factor is fixed at 0.3.

[0137] The resulting surface roughness attenuation factor is used as a dynamic adjustment parameter for the electrode surface state in the electrolysis energy consumption model.

[0138] Preferably, performing surface roughness corrosion determination on the grayscale image of the electrode surface according to the electrode corrosion pit density includes:

[0139] The surface roughness corrosion is judged by the grayscale image of the electrode surface according to the density of the electrode corrosion pits. When the density of the corrosion pits in the grayscale image of the electrode surface exceeds 0.15 / mm 2 , and the area of ​​a single corrosion pit exceeds 0.02mm 2 When , it is judged as light corrosion, and light corrosion discrimination data is obtained;

[0140] When the corrosion pit density in the grayscale image of the electrode surface is between 0.15 and 0.35 per mm 2 , and when the average depth of the corrosion pit is greater than 20 μm, it is judged to be moderate corrosion, and the moderate corrosion discrimination data is obtained;

[0141] When the corrosion pit density on the electrode surface exceeds 0.35 / mm 2 , and the depth of the corrosion pit exceeds 50μm, and the total area of ​​the corrosion pit accounts for more than 10% of the surface area, it is judged as severe corrosion and the severe corrosion discrimination data is obtained;

[0142] Integrate the light corrosion discrimination data, moderate corrosion discrimination data and severe corrosion discrimination data into the corrosion grade discrimination result;

[0143] The corrosion pit density and corrosion grade discrimination results are calculated based on the surface roughness attenuation quantification formula to generate the surface roughness attenuation factor. The surface roughness attenuation quantification formula is as follows:

[0144] D r =α·ρ c +β·σ g ;

[0145] Where D r is the surface roughness attenuation factor, ρ c is the corrosion pit density, σ g is the grayscale standard deviation of the grayscale image on the electrode surface, α and β are the empirical weight coefficients of the corrosion pit density and the grayscale standard deviation, respectively.

[0146] In the embodiment of the present invention, a high-resolution industrial microendoscope is used to collect grayscale images of the electrode surface, and the image spatial resolution is at least 1 μm / pixel. The image is processed by dedicated image processing software to identify and segment the corrosion pits, and the two-dimensional area of ​​each corrosion pit (unit: mm) is obtained. 2 ), and count the number of corrosion pits per unit area and calculate the corrosion pit density in units of pieces / mm 2 The corrosion pit area measurement is based on the pixel count multiplied by the pixel area, and the single corrosion pit area threshold is set to 0.02mm 2, pits below this value are not included in the corrosion judgment to exclude tiny noise and invalid corrosion points. With the help of three-dimensional microscopic scanning technology (such as confocal microscope or laser scanning microscope), the depth of the corrosion pit is measured, and the data is in micrometers (μm). The average depth of the corrosion pit is obtained by the arithmetic average of all valid corrosion pit depth values. The depth threshold of moderate corrosion is set to 20μm, and the depth threshold of severe corrosion is set to 50μm. When the density of corrosion pits per unit area exceeds 0.15 / mm 2 And there is a single corrosion pit with an area exceeding 0.02mm 2 When the corrosion pit density is between 0.15 and 0.35 per mm, it is judged as mild corrosion and the mild corrosion judgment data is recorded. 2 When the average depth of the corrosion pits is greater than 20 μm, it is judged as moderate corrosion and the moderate corrosion judgment data is recorded. 2 When the depth of the corrosion pit exceeds 50 μm and the total area of ​​the corrosion pits exceeds 10% of the total electrode surface area, it is judged as severe corrosion, and the severe corrosion discrimination data is recorded. The mild, moderate, and severe corrosion discrimination data are integrated to form a comprehensive corrosion grade discrimination result based on the corrosion area ratio and corrosion density data. This is used as input for the subsequent calculation of the attenuation factor. The grayscale standard deviation of the electrode surface grayscale image is calculated as a whole. The standard deviation reflects the degree of dispersion of the image grayscale distribution, reflecting the surface grayscale non-uniformity and roughness. The calculation process uses statistical analysis methods of the image grayscale matrix and outputs a dimensionless value. Based on the corrosion pit density and corrosion grade discrimination results, combined with the grayscale standard deviation, the attenuation factor is calculated using the surface roughness attenuation quantification formula: the attenuation factor is a linear combination of the corrosion pit density multiplied by an empirical weight coefficient and the grayscale standard deviation multiplied by another empirical weight coefficient. The empirical weight coefficients α and β are determined based on historical experimental data. Typically, α is 0.6 and β is 0.4 to ensure a reasonable balance between the contributions of corrosion pit density and grayscale distribution to the attenuation factor. The final generated attenuation factor is used to adjust the relevant parameters in the electrolysis energy consumption model to achieve dynamic reflection of the electrode surface state.

[0147] As an example of the present invention, refer to Figure 2 As shown, in this example, step S4 includes:

[0148] Step S41: normalizing the bubble shielding correction coefficient data and the surface roughness attenuation factor data to generate normalized correction factor data; inputting the normalized correction factor data into the electrolysis energy consumption initial model, dynamically adjusting the current efficiency parameter in the current-voltage efficiency curve, and generating an electrolysis energy consumption optimization model;

[0149] Step S42: performing multi-operating condition simulation energy consumption verification based on the electrolysis energy consumption optimization model to generate simulation energy consumption data;

[0150] Step S43: Calculate the deviation between the simulated energy consumption data and the actual unit product energy consumption value to obtain energy consumption deviation data;

[0151] Step S44: Compare the energy consumption deviation data with the preset deviation threshold. If the energy consumption deviation data is lower than the preset deviation threshold, confirm that the electrolysis energy consumption optimization model is the final electrolysis energy consumption model; if the energy consumption deviation data is higher than or equal to the preset deviation threshold, cyclically adjust the current efficiency parameter in the current-voltage efficiency curve until the energy consumption deviation data is lower than the preset deviation threshold.

[0152] In an embodiment of the present invention, the bubble shielding correction coefficient data and the surface roughness attenuation factor data are normalized separately. The normalization adopts the Min-Max method to linearly map the data to the interval of 0 to 1. The specific operations are as follows: calculate the minimum and maximum values ​​in the data set; apply the normalization formula to each data point, convert it into a relative ratio, and eliminate the influence of different data dimensions and ranges. The normalized bubble shielding correction coefficient and the normalized surface roughness attenuation factor are weighted and fused to obtain a unified normalized correction factor data. The weight ratio is set based on experimental experience and model fitting requirements to ensure that the correction factor can reasonably reflect the two influencing factors. The normalized correction factor data is input into the initial model of electrolysis energy consumption, and is mainly used to dynamically adjust the current efficiency parameters in the current-voltage efficiency curve. The adjustment process is: adjust the current efficiency parameters according to the correction factor so that it reflects the influence of the actual electrode surface and the bubble shielding state on the current utilization efficiency; the corrected current efficiency parameters replace the default parameters in the original model to form an electrolysis energy consumption optimization model. Design multiple sets of different operating parameters such as current, voltage, temperature and electrolysis time to cover typical and extreme operating conditions in actual production. Based on the electrolysis energy consumption optimization model generated in step S41, perform multi-operating condition simulation calculations to simulate the energy consumption performance under different operating conditions and output simulated energy consumption data. Organize and archive the energy consumption data obtained by simulation for subsequent deviation analysis and model optimization. Obtain the actual measured value of unit product energy consumption from the actual electrolysis production process. Calculate the deviation between the simulated energy consumption data and the actual unit product energy consumption value, and use absolute error or relative error indicators to obtain energy consumption deviation data. The absolute error is the absolute value of the difference between the two values; the relative error is the ratio of the absolute error to the actual value. Compare the energy consumption deviation data obtained in step S43 with the preset deviation threshold. The preset deviation threshold is set according to the production accuracy requirements and the model tolerance. If the energy consumption deviation data is lower than the preset threshold, confirm that the current electrolysis energy consumption optimization model is the final electrolysis energy consumption model, and end the optimization process. If the energy consumption deviation data is higher than or equal to the preset threshold, a loop adjustment mechanism is initiated: the current efficiency parameter in the current-voltage efficiency curve is adjusted based on the deviation size and trend; the simulation and deviation calculation in steps S42 to S43 are rerun; and the loop iterates until the energy consumption deviation data falls below the preset threshold. To prevent infinite loops, a maximum number of iterations or a time limit is set. If the limit is exceeded and the result is not met, the current optimal model and a warning message are output.

[0153] As an example of the present invention, refer to Figure 3 As shown, in this example, step S42 includes:

[0154] Step S421: Combining and setting the key variable parameters input into the electrolysis energy consumption optimization model, including five parameters: electrolyte concentration, electrolysis temperature, electrode spacing, anode material conductivity, and electrolytic cell gas pressure intensity, setting three valid interval values ​​for each parameter, and using full factor permutation and combination to generate fixed-dimensional multi-operating condition parameter combination set data;

[0155] Step S422: sequentially importing the multiple operating condition parameter combination set data into the electrolysis energy consumption optimization model, performing simulation calculations group by group, outputting the current-voltage operating response data under each group of operating conditions, and generating a simulation response data set;

[0156] Step S423: performing energy consumption integration processing on each set of current-voltage response curves in the simulation response data set to generate unit product energy consumption data corresponding to each set of working conditions;

[0157] Step S424: Perform a one-to-one mapping between the unit product energy consumption data and the multi-operating condition parameter combination set, and combine the parameter group and the corresponding energy consumption value into energy consumption mapping matrix data in a five-dimensional parameter space; perform a multi-dimensional offset rate analysis on the operating condition energy consumption mapping matrix data, identify abnormal operating condition parameter combinations whose unit product energy consumption value exceeds two standard deviations of the overall operating condition average deviation, and extract them as high-risk energy consumption operating condition data sets;

[0158] Step S425: Classify and archive the unit product energy consumption data and the high-risk energy consumption condition data set to generate simulated energy consumption data.

[0159] In an embodiment of the present invention, by selecting five key parameters in the electrolysis energy consumption optimization model, namely electrolyte concentration, electrolysis temperature, electrode spacing, anode material conductivity and electrolytic cell gas pressure intensity, and setting three representative effective interval values ​​for each parameter, covering the typical operating range; then using the full factor design method, the three values ​​of the five parameters are fully combined to generate a total of 243 groups of multi-operating condition parameter combination sets, and saved in matrix or table form for easy batch simulation calls. Then, the parameter combinations are input into the electrolysis energy consumption optimization model in order, and simulation calculations are performed to simulate the current-voltage working response under the corresponding working conditions, and the response curve data is extracted to form a simulation response data set, which is saved in order for subsequent processing. Subsequently, the energy consumption integral calculation is performed on each group of response curves, and the energy consumption in the corresponding time period is calculated based on the current and voltage curves, and converted into a unit product energy consumption value to form a complete energy consumption data set. Based on this, the unit product energy consumption data is mapped one-to-one with multiple operating condition parameter combinations, and an energy consumption mapping matrix is ​​constructed in a five-dimensional parameter space. The energy consumption values ​​of all operating conditions are counted, the overall average and standard deviation are calculated, and abnormal operating condition parameter combinations with energy consumption exceeding twice the standard deviation of the average are identified and extracted as high-risk energy consumption condition data sets. Finally, the normal operating condition energy consumption data and high-risk operating condition data are classified and archived separately, and the multiple operating condition parameters, corresponding energy consumption, and abnormality identification are stored in a structured data format to ensure orderly data management and secure backup. Through systematic multi-operating condition design and scientific abnormality identification, the simulation accuracy and risk warning capabilities of the electrolysis energy consumption optimization model are guaranteed, providing a solid data foundation for subsequent model optimization and process adjustment.

[0160] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0161] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing an electrolysis energy consumption model, characterized in that: The following steps are involved: Step S1: using a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature and electrolysis time; Constructing an initial electrolysis energy consumption model based on electrolysis process parameters, and inputting the electrolysis process parameters into the initial electrolysis energy consumption model to output unit product energy consumption value and current-voltage efficiency curve; Step S2: performing electrolytic bubble acoustic monitoring on the side wall of the electrolytic cell to obtain bubble burst acoustic wave signals; Analyze the sound intensity and burst frequency of the bubble burst acoustic wave signal; identify abnormal bubble accumulation based on the sound intensity and burst frequency, and generate a bubble shielding correction coefficient; Step S3: using an industrial endoscope to obtain a microscopic image of the electrode surface; performing grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identifying the electrode corrosion pit density in the grayscale image of the electrode surface, and performing surface roughness corrosion discrimination on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor; Step S4: inputting the bubble shielding correction coefficient and the surface roughness attenuation factor into the current efficiency parameter of the dynamic correction current-voltage efficiency curve in the initial electrolysis energy consumption model to generate an electrolysis energy consumption optimization model; The electrolysis energy consumption optimization model is simulated and verified. When the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, the final electrolysis energy consumption model is generated.

2. The method for constructing an electrolysis energy consumption model according to claim 1, wherein: Step S1 includes the following steps: Step S11: using a current sensor to collect the current value of the electrolysis process; Step S12: collecting the voltage value of the electrolysis process through the voltage collection module; Step S13: using a thermosensitive element to obtain the electrolyte temperature during the electrolysis process; Step S14: obtaining the electrolysis time of the electrolysis process based on the timing module; Step S15: synchronously scheduling and controlling the current value, voltage value, electrolyte temperature, and electrolysis time to obtain the original parameters of the electrolysis process; performing data preprocessing on the original parameters of the electrolysis process to generate electrolysis process parameters, wherein the data preprocessing includes data denoising, outlier removal, time synchronization calibration, and data standardization; Step S16: constructing an initial electrolysis energy consumption model based on the electrolysis process parameters, inputting the electrolysis process parameters into the initial electrolysis energy consumption model to calculate energy consumption, and outputting the unit product energy consumption value and the current-voltage efficiency curve.

3. The method for constructing an electrolysis energy consumption model according to claim 1, wherein: Analyzing the sound intensity and burst frequency of the bubble burst sound wave signal in step S2 includes: Extract the original bubble burst acoustic wave signal collected during the electrolysis process; Calculate the instantaneous sound pressure value of the bubble sound wave signal and obtain the sound pressure envelope curve; Identify the peak area in the sound pressure envelope curve and extract potential bubble collapse event fragments; Detect the average sound intensity value and the maximum sound intensity value of each potential bubble burst event segment to generate sound intensity characteristic parameters; Calculate the main frequency energy distribution in the rupture event fragment, extract the main frequency points and generate the frequency spectrum; Identify the area with continuous resonance peaks in the frequency spectrum and determine it as the effective rupture frequency range; The sound intensity characteristic parameters of each rupture event and the effective rupture frequency range are integrated to obtain the sound intensity and rupture frequency of the complete bubble rupture acoustic wave signal.

4. The method for constructing an electrolysis energy consumption model according to claim 1, wherein: In step S2, the abnormal bubble accumulation judgment of the bubble burst sound wave signal according to the sound intensity and the burst frequency includes: The bubble burst acoustic wave signal is used to identify abnormal bubble accumulation based on the sound intensity and burst frequency. When any of the following conditions occurs, it is determined to be an abnormal local high-density bubble accumulation, and the abnormal local bubble accumulation data is obtained: the sound intensity of the bubble burst acoustic wave signal is greater than 85dB, the bubble burst frequency exceeds 200 times / second, and the burst event duration per unit time exceeds 30 seconds; The bubble shielding effect is considered significant and abnormal bubble shielding effect data is obtained when the following conditions occur simultaneously: the amplitude of the sound intensity variation exceeds ±25dB, and the variation period is less than 2 seconds; there is overlapping interference between multiple bubble burst frequencies, resulting in an identification overlap rate exceeding 40%; the detected effective sound signal attenuation rate exceeds 60%, and the duration exceeds 15 minutes; When the following conditions occur simultaneously, it is determined to be a bubble accumulation stability imbalance and the bubble accumulation stability imbalance data is obtained: the average burst frequency fluctuates by more than ±35 times / second within 10 minutes, the temperature deviates from the set range by more than ±5°C, the average bubble diameter distribution deviation in the monitoring area exceeds 30%, and the center of gravity shift rate exceeds 0.5mm / s, where the temperature deviation from the set range is 35°C to 55°C; Integrate the data of abnormal local bubble accumulation, abnormal bubble shielding effect, and imbalanced bubble accumulation stability to calculate and obtain the abnormal bubble accumulation judgment results; The abnormal bubble accumulation judgment result is corrected based on the bubble shielding correction formula to generate a bubble shielding correction coefficient, where the bubble shielding correction formula is as follows: C bub =γ1·ΔI+γ2·Δf; Where C bub is the bubble shielding correction coefficient, ΔI is the sound intensity deviation, Δf is the frequency shielding amplitude, γ1 and γ2 are the sound intensity and frequency correction weight coefficients respectively.

5. The method for constructing an electrolysis energy consumption model according to claim 1, wherein: In step S3, the electrode corrosion pit density of the electrode surface grayscale image is identified, and surface roughness corrosion judgment is performed on the electrode surface grayscale image according to the electrode corrosion pit density, including: Calculate the local grayscale variance and texture gradient value of the grayscale image to form an initial corrosion feature response map; Detect local low-grayscale concave areas in the corrosion feature response image and mark potential corrosion pit candidate areas; Identify the boundary continuity and morphological closure of potential corrosion pit candidate areas and screen out effective corrosion pit areas; Extract the number of center points of the effective corrosion pit area and calculate the ratio of the center points to the unit area of ​​the image; obtain the corrosion pit density value data within the unit area; The surface roughness corrosion is judged on the grayscale image of the electrode surface according to the density of the electrode corrosion pits, and the surface roughness attenuation factor is generated.

6. The method for constructing an electrolysis energy consumption model according to claim 5, wherein: Surface roughness corrosion judgment of the grayscale image of the electrode surface based on the density of the electrode corrosion pit includes: The surface roughness corrosion is judged by the grayscale image of the electrode surface according to the density of the electrode corrosion pits. When the density of the corrosion pits in the grayscale image of the electrode surface exceeds 0.15 / mm 2 , and the area of ​​a single corrosion pit exceeds 0.02mm 2 When , it is judged as light corrosion, and light corrosion discrimination data is obtained; When the corrosion pit density in the grayscale image of the electrode surface is between 0.15 and 0.35 per mm 2 , and when the average depth of the corrosion pit is greater than 20 μm, it is judged to be moderate corrosion, and the moderate corrosion discrimination data is obtained; When the density of corrosion pits on the electrode surface exceeds 0.35 / mm 2 , and the depth of the corrosion pit exceeds 50μm, and the total area of ​​the corrosion pit accounts for more than 10% of the surface area, it is judged as severe corrosion and the severe corrosion discrimination data is obtained; Integrate the light corrosion discrimination data, moderate corrosion discrimination data and severe corrosion discrimination data into the corrosion grade discrimination result; The corrosion pit density and corrosion grade discrimination results are calculated based on the surface roughness attenuation quantification formula to generate the surface roughness attenuation factor. The surface roughness attenuation quantification formula is as follows: D r =a·r c +b·s g ; Where D r is the surface roughness attenuation factor, ρ c is the corrosion pit density, σ g is the grayscale standard deviation of the grayscale image on the electrode surface, α and β are the empirical weight coefficients of the corrosion pit density and the grayscale standard deviation, respectively.

7. The method for constructing an electrolysis energy consumption model according to claim 1, wherein: Step S4 includes the following steps: Step S41: normalizing the bubble shielding correction coefficient data and the surface roughness attenuation factor data to generate normalized correction factor data; inputting the normalized correction factor data into the electrolysis energy consumption initial model, dynamically adjusting the current efficiency parameter in the current-voltage efficiency curve, and generating an electrolysis energy consumption optimization model; Step S42: performing multi-operating condition simulation energy consumption verification based on the electrolysis energy consumption optimization model to generate simulation energy consumption data; Step S43: Calculate the deviation between the simulated energy consumption data and the actual unit product energy consumption value to obtain energy consumption deviation data; Step S44: Compare the energy consumption deviation data with the preset deviation threshold. If the energy consumption deviation data is lower than the preset deviation threshold, confirm that the electrolysis energy consumption optimization model is the final electrolysis energy consumption model; if the energy consumption deviation data is higher than or equal to the preset deviation threshold, cyclically adjust the current efficiency parameter in the current-voltage efficiency curve until the energy consumption deviation data is lower than the preset deviation threshold.

8. The method for constructing an electrolysis energy consumption model according to claim 7, wherein: Step S42 includes the following steps: Step S421: Combining and setting the key variable parameters input into the electrolysis energy consumption optimization model, including five parameters: electrolyte concentration, electrolysis temperature, electrode spacing, anode material conductivity, and electrolytic cell gas pressure intensity, setting three valid interval values ​​for each parameter, and using full factor permutation and combination to generate fixed-dimensional multi-operating condition parameter combination set data; Step S422: sequentially importing the multiple operating condition parameter combination set data into the electrolysis energy consumption optimization model, performing simulation calculations group by group, outputting the current-voltage operating response data under each group of operating conditions, and generating a simulation response data set; Step S423: performing energy consumption integration processing on each set of current-voltage response curves in the simulation response data set to generate unit product energy consumption data corresponding to each set of working conditions; Step S424: Perform a one-to-one mapping between the unit product energy consumption data and the multi-operating condition parameter combination set, and combine the parameter group and the corresponding energy consumption value into energy consumption mapping matrix data in a five-dimensional parameter space; perform a multi-dimensional offset rate analysis on the operating condition energy consumption mapping matrix data, identify abnormal operating condition parameter combinations whose unit product energy consumption value exceeds two standard deviations of the overall operating condition average deviation, and extract them as high-risk energy consumption operating condition data sets; Step S425: Classify and archive the unit product energy consumption data and the high-risk energy consumption condition data set to generate simulated energy consumption data.

9. A system for constructing an electrolysis energy consumption model, characterized in that: A method for constructing an electrolysis energy consumption model according to claim 1, wherein the system for constructing the electrolysis energy consumption model comprises: A model building module is used to use a multimodal sensor system to collect electrolysis process parameters, including current value, voltage value, electrolyte temperature and electrolysis time; construct an initial electrolysis energy consumption model based on the electrolysis process parameters, and input the electrolysis process parameters into the initial electrolysis energy consumption model to output unit product energy consumption value and current-voltage efficiency curve; The bubble analysis module is used to acoustically monitor the electrolytic bubbles on the side wall of the electrolytic cell to obtain the bubble burst sound wave signal; analyze the sound intensity and burst frequency of the bubble burst sound wave signal; and identify abnormal bubble accumulation based on the sound intensity and burst frequency of the bubble burst sound wave signal to generate a bubble shielding correction coefficient; The corrosion analysis module is used to obtain a microscopic image of the electrode surface using an industrial endoscope; perform grayscale gradient analysis on the microscopic image of the electrode surface to generate a grayscale image of the electrode surface; identify the electrode corrosion pit density in the grayscale image of the electrode surface, and perform surface roughness corrosion judgment on the grayscale image of the electrode surface based on the electrode corrosion pit density to generate a surface roughness attenuation factor; The model optimization module is used to input the bubble shielding correction coefficient and the surface roughness attenuation factor into the initial electrolysis energy consumption model to dynamically correct the current efficiency parameters of the current-voltage efficiency curve to generate an electrolysis energy consumption optimization model; the electrolysis energy consumption optimization model is simulated and verified for energy consumption. When the deviation between the simulated energy consumption and the unit product energy consumption value is lower than the preset value, the final electrolysis energy consumption model is generated.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for constructing an electrolysis energy consumption model as described in any one of claims 1 to 8 is implemented.

Images