A method and system for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven approaches

Abstract

This invention proposes a method and system for predicting anode deactivation based on electrochemical fingerprinting and knowledge co-driven approaches. It collects anode reaction potential, cell voltage, operating parameters, anode structural characteristics, and time-series electrochemical spectra. Peak current, volt-ampere integral charge, and redox response intensity are extracted from cyclic voltammetry curves. Anode potential, reaction overpotential, and Tafel slope at a fixed current density are extracted from linear sweep voltammetry curves. Cell voltage changes, change rates, and fluctuation levels are also extracted. Candidate fingerprints are screened and integrated with operating parameters, considering deactivation mechanisms such as side reaction competition. An initial prediction model for the anode reaction potential is constructed based on electrochemical kinetics and ohmic polarization constraints, and then corrected using a machine learning model. Based on the corrected prediction results, the time-series changes of the core fingerprint, and preset criteria, the anode deactivation state, deactivation rate, dominant deactivation type, and risk level are determined, and operational control recommendations are output.

Description

Technical Field

[0001] This invention proposes a method and system for predicting deactivation of electrowinning anodes based on electrochemical fingerprinting and knowledge co-driven approaches, which relates to the technical field of electrochemical recovery and intelligent process control. Background Technology

[0002] Electrowinning processes are widely used in non-ferrous metal smelting, hydrometallurgy, resource recovery of wastewater containing heavy metals, metal recycling from waste batteries, and targeted recovery of acidic metal ion solutions. During operation, the electrodepositing anode typically performs functions such as oxygen evolution, maintaining current flow, and stabilizing the electric field distribution. Its active state directly affects cell voltage, current efficiency, energy consumption, and the selectivity of target metal recovery.

[0003] In actual electrodeposition processes, the anode is exposed to a strong acid, high current density, and a coexistence of multiple metal ions and complex impurities for extended periods. This environment is prone to phenomena such as competition for side reactions on the anode surface, anode slime deposition, film densification, dissolution of catalytically active components, impurity ion adsorption poisoning, and limited mass transfer. These phenomena can lead to increased anode reaction potential, decreased oxygen evolution efficiency, and increased cell voltage fluctuations. In severe cases, they can even cause anode deactivation, production interruption, and reduced efficiency in the resource recovery process.

[0004] Current methods for determining anode deactivation largely rely on operator experience, abnormal cell voltage, periodic shutdown inspections, or single electrochemical test results. These methods have the following problems: First, cell voltage or single-point potential primarily reflects instantaneous changes in the anode state, which is easily affected by system operating conditions and other polarization factors, making it difficult to use alone to identify anode deactivation trends in advance. Second, cyclic voltammetry, Tafel slope, and impedance spectroscopy results are often used as independent characterization methods, reflecting electrode reactivity, kinetic characteristics, and interfacial mass / charge transfer behavior, respectively. However, this information is usually not systematically integrated into a quantifiable and comparable electrochemical fingerprint, making it difficult to directly use for identifying the deactivation state and predicting trends of electrodeposited anodes. Third, while traditional pure data-driven models can learn from historical data patterns, they are insufficient in expressing the anode deactivation mechanism and have limited generalization ability across operating conditions. Fourth, prediction models that rely solely on theoretical formulas are difficult to adapt to complex electrodeposited electrolyte compositions and dynamic operational disturbances.

[0005] Therefore, there is an urgent need for a predictive method that can simultaneously utilize operating condition data, electrochemical test data, and knowledge of anode deactivation mechanisms to transform visualized electrochemical responses into electrochemical fingerprints with structure-activity relationship significance. By combining electrochemical fingerprints with knowledge-driven approaches, it is possible to achieve early prediction, risk warning, and interpretable analysis of electrodeposition anode deactivation trends. Summary of the Invention

[0006] In view of this, in order to fill the gaps and deficiencies in the existing technology, this invention proposes a method and system for predicting the deactivation of electro-concentrated anodes based on electrochemical fingerprinting and knowledge co-driven approaches. Through the technical path of "spectrum acquisition - fingerprint extraction - core fingerprint screening - knowledge model - residual correction - deactivation prediction - early warning output", the anode operating condition parameters, structural features and core electrochemical fingerprints are fused together. Combined with anode deactivation mechanism constraints and machine learning residual correction, the accuracy, interpretability and cross-condition adaptability of electro-concentrated anode deactivation state identification and trend prediction are improved.

[0007] This invention proposes a method and system for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven approaches, comprising the following:

[0008] According to a first aspect of the present invention, the present invention proposes a method for predicting the deactivation of electrowinning anodes based on electrochemical fingerprinting and knowledge co-driven methods, characterized in that it includes the following:

[0009] Step S1: Collect the anodic operating parameters, structural characteristics, and time-series electrochemical scanning spectra of the electrodeposited anode;

[0010] Step S2: Preprocess the acquired data, including background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment.

[0011] Step S3: Extract candidate electrochemical fingerprints from the pre-processed time-series electrochemical scanning spectra, and screen them in conjunction with the anodic side reaction competition, film thickening and active site poisoning-induced inactivation mechanism to obtain the core electrochemical fingerprint;

[0012] Step S4: Perform feature fusion on the anode operating condition parameters, structural features, and core electrochemical fingerprint to construct an input feature vector;

[0013] Step S5: Construct a knowledge model based on electrochemical kinetics, ohmic polarization, mass transfer constraints, and anodic deactivation mechanism constraints to obtain the theoretical predicted value of the anodic reaction potential;

[0014] Step S6: Using the input feature vector as the model input, and the residual between the measured anode reaction potential and the theoretical prediction value as the learning target, the theoretical prediction value is corrected using a machine learning model to obtain the corrected anode reaction potential prediction value.

[0015] Step S7: Based on the corrected predicted anode reaction potential, the time-series change trend of the core electrochemical fingerprint, and the preset deactivation criteria, predict the change trend of the anode reaction potential within the preset time range in the future, determine the anode deactivation state, deactivation rate, dominant deactivation type, and deactivation risk level, and output corresponding operation and control suggestions.

[0016] Further, step S1 includes the following:

[0017] In step S1, the operating parameters, anode structural characteristics, and time-series electrochemical scanning spectra of the electrodeposition anode during the electrodeposition process are acquired at fixed intervals; the time-series electrochemical scanning spectra include cyclic voltammetry curves, polarization curves, and electrochemical impedance spectroscopy.

[0018] Further, step S2 includes the following:

[0019] In step S2, the time-series electrochemical scanning spectrum is preprocessed, and candidate electrochemical fingerprints are extracted from the preprocessed data. The preprocessing includes one or more of noise filtering, baseline correction, outlier removal, normalization, and time alignment. The candidate electrochemical fingerprints include reaction overpotential, Tafel slope, cyclic voltammetric integral charge, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance.

[0020] Further, step S3 includes the following:

[0021] In step S3, the candidate electrochemical fingerprints are screened based on the competition of anodic side reactions, film thickening, and active site inactivation mechanisms to obtain the core electrochemical fingerprint. The screening includes correlation analysis, redundancy analysis, and significance testing, including the following:

[0022] First, the correlation between candidate electrochemical fingerprints and anodic deactivation characterization parameters is calculated. These parameters include one or more of the following: measured anodic reaction potential, the increase in anodic reaction potential relative to the initial state, the rate of increase in anodic reaction potential per unit time, and anodic deactivation state labels determined according to a preset deactivation threshold. Candidate electrochemical fingerprints with a correlation to the anodic deactivation characterization parameters lower than the preset threshold are eliminated. Then, the correlation or redundancy between candidate electrochemical fingerprints is calculated, and redundant fingerprints that repeatedly express changes in anodic state are removed. Finally, based on the anodic deactivation mechanism, the core electrochemical fingerprints that can characterize changes in anodic reaction kinetics, film thickness, and reaction competition are retained.

[0023] Further, step S4 includes the following:

[0024] In step S4, the core electrochemical fingerprint, anode structure parameters, and operating condition parameters are time-aligned and feature-fused to form an input feature vector. The input feature vector includes one or more of the following parameters: reaction overpotential; Tafel slope; cyclic voltammetric integral charge; oxygen evolution current efficiency; charge transfer resistance; pore resistance; membrane capacitance; current density; electrolyte flow rate; electrolyte temperature; sulfuric acid concentration; and target recovered ion concentration.

[0025] Further, step S5 includes the following:

[0026] In step S5, the mechanism knowledge model includes an equilibrium potential term, a kinetic polarization term, and an ohmic polarization term, and further includes a mass transfer polarization term, a film impedance term, and a deactivation state term.

[0027] Further, step S6 includes the following:

[0028] In step S6, the machine learning model uses the difference between the measured anode reaction potential and the theoretical predicted anode reaction potential as the residual learning target, and superimposes the residual correction value output by the machine learning model with the theoretical predicted anode reaction potential to obtain the corrected anode reaction potential prediction value.

[0029] Further, step S7 includes the following:

[0030] The operational control recommendations include one or more of the following: anode cleaning, anode replacement, reducing current density, increasing electrolyte flow rate, adjusting electrolyte temperature, and adjusting electrolyte composition.

[0031] According to a second aspect of the present invention, the present invention proposes an electro-anode inactivation prediction system based on electrochemical fingerprint and knowledge co-driven, which is applied in the electro-anode inactivation prediction method based on electrochemical fingerprint and knowledge co-driven according to any one of the present invention, characterized in that it includes an operating condition acquisition module, an anode structure characterization module, a time-series electrochemical spectrum acquisition module, an electrochemical fingerprint extraction module, a fingerprint screening module, a mechanism knowledge model construction module, a machine learning residual correction module, an inactivation trend and type determination module, and an early warning output module;

[0032] The operating condition acquisition module is used to continuously or at preset time intervals acquire the operating parameters of the electrowinning reaction, including cell voltage, anode potential, current density, electrowinning liquid flow rate, electrowinning liquid temperature, sulfuric acid concentration, target recovered ion concentration, pH value, conductivity and cumulative operating time.

[0033] The anode structure characterization module is used to obtain anode film thickness, specific surface area, pore structure, and structural parameters characterizing changes in effective active area or accessibility of active sites, for analyzing anode deactivation mechanisms, including side reaction competition, film thickening, and active site poisoning.

[0034] The time-series electrochemical spectrum acquisition module is used to perform cyclic voltammetry, polarization scanning, and electrochemical impedance spectroscopy tests on the electro-anode at different sampling times during the operation of the electro-anode, and obtain time-stamped cyclic voltammetry curves, polarization curves, and electrochemical impedance spectra; the electrochemical fingerprint extraction module is used to extract features from the time-series electrochemical scan spectrum.

[0035] The fingerprint screening module is used to combine the anode deactivation mechanism and statistical methods, including correlation analysis, redundancy analysis and feature selection, to screen core electrochemical fingerprints that are related to the electrodeposition anode deactivation behavior and have low redundancy.

[0036] The mechanism knowledge model construction module is used to construct an anode reaction potential prediction model and generate mechanism prediction values ​​based on electrochemical kinetic relationships, ohmic polarization relationships, mass transfer limitation relationships and anode deactivation mechanism constraints.

[0037] The machine learning residual correction module takes the input feature vector as input and the residual between the theoretical prediction value and the measured potential as the learning target to correct the theoretical prediction value and obtain the corrected anode reaction potential prediction value.

[0038] The inactivation trend and type determination module is used to predict the trend of anode reaction potential change within a preset time range based on the corrected anode reaction potential prediction value, the time-series change trend of the core electrochemical fingerprint and the preset inactivation criteria, and output the anode inactivation rate, inactivation state, dominant inactivation type and inactivation risk level.

[0039] The early warning output module is used to generate the anode deactivation risk level based on the prediction results and provide control suggestions, including anode cleaning, replacement, current density adjustment, electrolyte flow rate adjustment, or electrolyte composition adjustment.

[0040] Further, the electrochemical fingerprint includes one or more of the following: reaction overpotential, anodic potential at a fixed current density or its change relative to a healthy state baseline, Tafel slope, cyclic voltammetric integral charge, oxidation peak current, reduction peak current, redox peak position, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance; the deactivation trend and type determination module is used to output a predicted anodic reaction potential curve, prediction confidence interval, and deactivation rate estimation results within a future preset time range, and to evaluate the deactivation state, deactivation risk, and dominant driving mechanism of the electro-deposited anode based on the prediction results; the preset time range is determined based on one or more of the following: electrochemical spectrum sampling frequency, available historical operating data length, electro-deposited anode operating cycle, and anode maintenance cycle.

[0041] The present invention has the following advantages:

[0042] 1. This invention unifies multi-source electrochemical spectra such as cyclic voltammetry, polarization curves, and impedance spectra into electrochemical fingerprints, which can more accurately reflect changes in anode activity than single cell voltage or single potential indicators;

[0043] 2. This invention incorporates inactivation mechanisms such as anodic side reaction competition, film thickening, and active site poisoning into the feature screening process, which can improve the physical meaning and interpretability of the model input features;

[0044] 3. This invention adopts a combination of electrochemical fingerprinting and knowledge co-driving, which retains the theoretical constraints of the electrode process while using historical operating data to correct nonlinear deviations under complex operating conditions;

[0045] 4. This invention can output the deactivation risk level and key contributing factors, providing an interpretable basis for anode maintenance, replacement decisions, electrolyte control, and current density optimization;

[0046] 5. This invention can be applied to the electrowinning process of metals such as copper, zinc, nickel, cobalt, and manganese, and can also be used in the resource-based electrowinning system for waste battery leachate and industrial wastewater containing heavy metals. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the steps of the present invention.

[0048] Figure 2 This is a schematic diagram of the process for extracting peak current characteristics based on cyclic voltammetric response and determining anodic activity decay according to the present invention.

[0049] Figure 3 This is a schematic diagram illustrating the process of fingerprint extraction, feature fusion, correlation analysis, state determination, and early warning output of multi-source anode state data according to the present invention.

[0050] Figure 4 This is a schematic diagram of the electrowinning anode deactivation prediction system described in this invention. Detailed Implementation

[0051] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0052] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0053] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0054] like Figures 1 to 4As shown, this invention proposes a method and system for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven approaches, comprising the following:

[0055] According to a first aspect of the present invention, the present invention proposes a method for predicting the deactivation of electro-anodes based on electrochemical fingerprinting and knowledge co-driven approaches, employing an electro-anode deactivation prediction system based on electrochemical fingerprinting and knowledge co-driven approaches as described in any one of the present invention, such as... Figure 1 As shown, it includes the following:

[0056] Step S1: Collect the anodic operating parameters, structural characteristics, and time-series electrochemical scanning spectra of the electrodeposited anode;

[0057] Step S2: Preprocess the acquired data, including background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment.

[0058] Step S3: Extract candidate electrochemical fingerprints from the pre-processed time-series electrochemical scanning spectra, and screen them in conjunction with the anodic side reaction competition, film thickening and active site poisoning-induced inactivation mechanism to obtain the core electrochemical fingerprint;

[0059] Step S4: Perform feature fusion on the anode operating condition parameters, structural features, and core electrochemical fingerprint to construct an input feature vector;

[0060] Step S5: Construct a knowledge model based on electrochemical kinetics, ohmic polarization, mass transfer constraints, and anodic deactivation mechanism constraints to obtain the theoretical predicted value of the anodic reaction potential;

[0061] Step S6: Using the input feature vector as the model input, and the residual between the measured anode reaction potential and the theoretical prediction value as the learning target, the theoretical prediction value is corrected using a machine learning model to obtain the corrected anode reaction potential prediction value.

[0062] Step S7: Based on the corrected predicted anode reaction potential, the time-series change trend of the core electrochemical fingerprint, and the preset deactivation criteria, predict the change trend of the anode reaction potential within the preset time range in the future, determine the anode deactivation state, deactivation rate, dominant deactivation type, and deactivation risk level, and output corresponding operation and control suggestions.

[0063] In this invention, step S1 includes the following:

[0064] Step S11: At fixed intervals, acquire the operating parameters, anode structural characteristics, and time-series electrochemical scanning spectra of the electrodeposition anode during the electrodeposition process. The time-series electrochemical scanning spectra include cyclic voltammetry curves, polarization curves, and electrochemical impedance spectroscopy.

[0065] At the t-th sampling time, the acquired anode state data can be expressed as:

[0066] ;

[0067] in, This represents the anode reaction potential at the t-th sampling time. Indicates operating condition parameters, Indicates the anode structure parameters, , , and The cyclic voltammetry curve, polarization curve, Tafel curve, and electrochemical impedance spectroscopy obtained at the t-th sampling time are respectively represented.

[0068] In this invention, step S2 includes the following:

[0069] Step S21: Preprocess the time-series electrochemical scan spectrum and extract candidate electrochemical fingerprints from the preprocessed data; the preprocessing includes one or more of noise filtering, baseline correction, outlier removal, normalization, and time alignment; the candidate electrochemical fingerprints include reaction overpotential, Tafel slope, cyclic voltammetric integral charge, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance;

[0070] The reaction overpotential can be expressed as:

[0071] ;

[0072] In the formula, Let be the reaction overpotential at the t-th sampling time. Let be the measured anode reaction potential at the t-th sampling time. The equilibrium potential of the target anode reaction. This is the anode current. It is an electrolytic electrolyte ohmic resistor.

[0073] The Tafel slope can be obtained through the following relationship:

[0074] ;

[0075] In the formula, Let J be the reaction overpotential at the t-th sampling time with current density j. Let be the Tafel intercept at the t-th sampling time. Let be the Tafel slope at the t-th sampling time.

[0076] j is the current density. This is the reference current density.

[0077] The cyclic voltammetric integral charge can be expressed as:

[0078] ;

[0079] In the formula, Let V be the cyclic voltammetric integrated charge at the t-th sampling time, and V be the cyclic voltammetric scan rate. and These are the lower and upper limits of the integral potential interval, respectively. Let be the response current at the scanning potential E at the t-th sampling time.

[0080] The oxygen evolution current efficiency can be expressed as:

[0081] ;

[0082] In the formula, Let be the oxygen evolution current efficiency at the t-th sampling time. The partial current density used for the oxygen evolution reaction, This represents the total current density.

[0083] Wherein, the total current density The ratio of the total anode current to the effective anode reaction area measured during the t-th sampling time period is obtained and expressed as:

[0084] ;

[0085] In the formula, Let A be the total anode current measured during the t-th sampling time period, and let A be the effective reaction area of ​​the anode.

[0086] The oxygen evolution reaction partial current density The amount of oxygen actually generated at the anode during the t-th sampling time period is calculated using Faraday's law, and is expressed as follows:

[0087] ;

[0088] In the formula, F is the Faraday constant. Sampling time period The amount of oxygen actually generated at the inner anode, with coefficient 4 representing the number of electrons transferred corresponding to the generation of 1 mol of oxygen; the amount of oxygen is obtained by measurement using one or more of the following: a gas collection device, a gas flow meter, an oxygen sensor, or a gas chromatograph.

[0089] In this invention, step S3 includes the following:

[0090] Step S31: Based on the competition of anodic side reactions, film thickening and active site deactivation mechanisms, the candidate electrochemical fingerprints are screened to obtain the core electrochemical fingerprints; the screening includes correlation analysis, redundancy analysis and significance test.

[0091] In a preferred embodiment, the correlation between candidate electrochemical fingerprints and anodic deactivation characterization parameters is first calculated. These anodic deactivation characterization parameters include one or more of the following: measured anodic reaction potential, the increase in anodic reaction potential relative to the initial state, the rate of increase in anodic reaction potential per unit time, and anodic deactivation state labels determined according to a preset deactivation threshold. Candidate electrochemical fingerprints with a correlation to the anodic deactivation characterization parameters lower than the preset threshold are then eliminated. Next, the correlation or redundancy between candidate electrochemical fingerprints is calculated, and redundant fingerprints that repeatedly express anodic state changes are removed. Finally, based on the anodic deactivation mechanism, core electrochemical fingerprints that can characterize anodic reaction kinetics, film thickness, and reaction competition changes are retained.

[0092] In this invention, step S4 includes the following:

[0093] Step S41: Time-align and feature-fuse the core electrochemical fingerprint, anode structure parameters, and operating condition parameters to form an input feature vector; the input feature vector can be expressed as:

[0094] ;

[0095] In the formula, Let be the input feature vector at the t-th sampling time. For charge transfer resistor, For hole resistance, For film capacitance, For current density, The flow rate of the electrolyte is [missing information]. The electrodeposition temperature,

[0096] This refers to the concentration of sulfuric acid. The target is to recover the ion concentration.

[0097] In this invention, step S5 includes the following:

[0098] Step S51: Based on electrochemical kinetics, ohmic polarization, mass transfer limitation, and anodic deactivation accumulation, an initial prediction model for the anodic reaction potential is constructed to obtain the initial predicted value of the anodic reaction potential within a preset time range. The initial prediction model for the anodic reaction potential can be expressed as:

[0099] ;

[0100] In the formula, Let h be the theoretical predicted value of the anode reaction potential at a future time (t+h), where h is the prediction step size. This represents a knowledge model consisting of electrochemical kinetic relationships, ohmic polarization relationships, mass transfer limitation relationships, and anodic deactivation accumulation relationships.

[0101] In a preferred embodiment, the anode deactivation comprehensive index can be expressed as:

[0102] ;

[0103] In the formula, Let be the comprehensive index of anode deactivation at the t-th sampling time. , , , and These are the normalized reaction overpotential, Tafel slope, cyclic voltammetric integral charge, charge transfer resistance, and film capacitance, respectively. to These are the corresponding weighting coefficients.

[0104] In this invention, step S6 includes the following:

[0105] Step S61: Using the input feature vector as the model input, and the residual between the measured anode reaction potential and the theoretical prediction value as the learning target, a machine learning model is used to correct the residual of the theoretical prediction value to obtain the corrected anode reaction potential prediction value.

[0106] The residual can be expressed as:

[0107] ;

[0108] In the formula, r(t+h) is the prediction residual at the future time (t+h). This represents the measured anode reaction potential at a future time (t+h).

[0109] The corrected predicted anodic reaction potential can be expressed as:

[0110] ;

[0111] In the formula This is the corrected predicted value for the anode reaction potential. The residual correction value is the output of the machine learning model. The machine learning model includes one or more of the following: Random Forest Regression, Support Vector Regression, XGBoost, CatBoost, LightGBM, Long Short-Term Memory Network, or a combination of the above models.

[0112] In this invention, step S7 includes the following:

[0113] Step S71: Based on the corrected predicted anode reaction potential, the trend of core electrochemical fingerprint change, and the preset deactivation criteria, output the anode deactivation status, the deactivation trend within the preset time range, the prediction confidence interval, the deactivation risk level, and operation control suggestions.

[0114] In a preferred embodiment, when the corrected predicted anode reaction potential continuously increases relative to the healthy state baseline, and one or more abnormal states such as increased charge transfer resistance, increased Tafel slope, decreased cyclic voltammetric integral charge, or decreased oxygen evolution current efficiency continue to reach a preset number of sampling cycles, it is determined that the electrolytic anode is at risk of deactivation; when the corrected predicted anode reaction potential reaches or exceeds a preset deactivation potential threshold, or when the abnormality of multiple core electrochemical fingerprints simultaneously reaches the corresponding fingerprint threshold, anode deactivation warning information is generated.

[0115] Furthermore, the method described in this invention is applicable to the electrowinning extraction and recovery processes of metals such as copper, zinc, nickel, cobalt, and manganese, as well as the prediction of anode potential and early warning of deactivation in the resource recovery of waste battery leachate and the resource treatment of industrial wastewater containing heavy metals. The machine learning model includes one or more of the following: random forest regression model, support vector regression model, XGBoost model, CatBoost model, LightGBM model, long short-term memory network model, and ensemble models composed of the above models.

[0116] According to a second aspect of the present invention, the present invention proposes an electro-anode deactivation prediction system based on electrochemical fingerprint and knowledge co-driven, characterized in that it includes an operating condition acquisition module, an anode structure characterization module, a time-series electrochemical spectrum acquisition module, an electrochemical fingerprint extraction module, a fingerprint screening module, a mechanism knowledge model construction module, a machine learning residual correction module, a deactivation trend and type determination module, and an early warning output module.

[0117] The operating condition acquisition module is used to continuously or at preset time intervals acquire one or more operating condition parameters during the electrowinning process, including anode reaction potential, current density, electrowinning fluid flow rate, electrowinning fluid temperature, sulfuric acid concentration, target recovered ion concentration, pH value, conductivity, and cumulative operating time, and form operating condition time series data corresponding to the sampling time.

[0118] The anode structure characterization module is used to obtain one or more anode structure parameters among the film thickness, specific surface area, pore structure and number of effective active sites of the electrodeposited anode, so as to characterize the changes in anode film thickness, pore blockage, deposit coverage and effective active sites during the electrodeposition process, and provide structural characteristic data for analyzing deactivation mechanisms such as anode side reaction competition, film thickness and active site poisoning.

[0119] The time-series electrochemical spectrum acquisition module is used to perform one or more of cyclic voltammetry, polarization scanning, and electrochemical impedance spectroscopy tests on the electro-anode at different sampling times during the operation of the electro-anode, so as to obtain time-series electrochemical scanning spectra such as cyclic voltammetry curves, polarization curves, and electrochemical impedance spectroscopy.

[0120] The electrochemical fingerprint extraction module is connected to the operating condition acquisition module, the anode structure characterization module, and the time-series electrochemical spectrum acquisition module, respectively. It is used to perform one or more preprocessing steps on the acquired operating condition parameters, anode structure parameters, and time-series electrochemical scan spectra, including background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment, and to extract candidate electrochemical fingerprints from the preprocessed data.

[0121] The candidate electrochemical fingerprints include one or more of the following: reaction overpotential, Tafel slope, volt-ampere integral charge, oxygen evolution current efficiency, anolyte bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance. The Tafel slope can be obtained by fitting the Tafel linear region of the polarization curve. The candidate electrochemical fingerprints are used to characterize reaction activity, interfacial reaction composition, mass transfer state, and film impedance characteristics.

[0122] The fingerprint screening module is used to combine the competition of anodic side reactions, film thickening and active site deactivation mechanisms to perform one or more of the following processes on candidate electrochemical fingerprints: correlation analysis, redundancy analysis, significance test and feature selection, to screen out core electrochemical fingerprints that are related to the deactivation behavior of electrodeposited anodes and have low redundancy.

[0123] The mechanism knowledge model construction module is used to construct an initial prediction model for the anode reaction potential based on electrochemical kinetic relationships, ohmic polarization relationships, mass transfer limitation relationships, and anode deactivation mechanism constraints. This initial prediction model decomposes the anode reaction potential into one or more of the following: equilibrium potential term, kinetic polarization term, ohmic polarization term, mass transfer polarization term, and film impedance term. The theoretical predicted value of the anode reaction potential is obtained based on each component term.

[0124] The machine learning residual correction module is used to fuse the core electrochemical fingerprint, anode structural parameters, and operating condition parameters to form an input feature vector; the input feature vector is used as the input of the machine learning model, and the difference between the measured anode reaction potential and the theoretical predicted anode reaction potential is used as the residual learning target to establish a residual prediction model; the residual correction value output by the residual prediction model is superimposed with the theoretical predicted anode reaction potential to obtain the corrected anode reaction potential prediction value.

[0125] The inactivation trend and type determination module is used to output the predicted anode reaction potential curve, prediction confidence interval, and inactivation rate estimation results within a preset time range, based on the corrected predicted anode reaction potential value, the time-series change trend of the core electrochemical fingerprint, and preset inactivation criteria. The preset time range is determined by the electrochemical spectrum sampling frequency, the length of available historical operating data, the electrowinning anode operating cycle, and the anode maintenance cycle.

[0126] The preset deactivation criteria are determined based on baseline data, historical operating data, maintenance and replacement records, or anode deactivation calibration tests during the healthy operation phase of the electrolytic anode. These criteria include: the increase in anode reaction potential relative to the healthy state baseline value reaching or exceeding a preset potential threshold; the rate of increase in anode reaction potential per unit time reaching or exceeding a preset rate threshold; the change in the core electrochemical fingerprint relative to the healthy state baseline value reaching or exceeding the corresponding fingerprint threshold; and the aforementioned abnormal state persisting for one or more of the preset sampling cycles. The core electrochemical fingerprint anomalies include one or more of the following: increased reaction overpotential or Tafel slope; increased charge transfer resistance or pore resistance; decreased cyclic voltammetric integrated charge or oxygen evolution current efficiency; and film capacitance deviating from the healthy state baseline range.

[0127] The inactivation trend and type determination module is also used to evaluate the inactivation type and dominant driving mechanism of the electrodeposited anode based on the abnormal combination of different core electrochemical fingerprints. The inactivation type or dominant driving mechanism includes one or more of the following: side reaction competition-dominated inactivation, film thickening or pore blockage-dominated inactivation, active site inactivation, and composite inactivation formed by the combined action of multiple inactivation mechanisms.

[0128] The early warning output module is used to generate early warning information for electrodeposition anode inactivation based on the anode reaction potential prediction curve, the anode reaction potential increase, the degree of core electrochemical fingerprint anomaly, the inactivation rate estimation result, and the inactivation type evaluation result. It classifies the inactivation risk of electrodeposition anode into low risk, medium risk, and high risk, and outputs operation and control suggestions corresponding to the inactivation risk level and the dominant driving mechanism.

[0129] The operational control recommendations include one or more of the following: anode cleaning, anode replacement, reducing current density, increasing or adjusting electrolyte flow rate, adjusting electrolyte temperature, and adjusting electrolyte composition.

[0130] Preferably, the anode structure characterization module obtains anode structure parameters by one or more of scanning electron microscopes, optical microscopes, three-dimensional profilometers, specific surface area and pore size analysis devices or surface element analysis devices, so as to characterize the morphology of the anode film layer, crack distribution, pore structure changes, deposit coverage and effective active site changes.

[0131] Preferably, the machine learning residual correction module uses one or more algorithms to construct a residual prediction model and outputs the residual correction value using a single machine learning model or a multi-model ensemble.

[0132] Preferably, the inactivation trend and type determination module determines the inactivation state and dominant inactivation type of the electrodeposited anode based on the abnormal changes in different core electrochemical fingerprints; wherein, a continuous increase in reaction overpotential or Tafel slope is used to characterize the deterioration of anode reaction kinetics, a decrease in voltammetric integral charge or anode bulk redox intensity is used to characterize the reduction of effective active sites, and abnormal changes in charge transfer resistance, pore resistance or film capacitance are used to characterize film thickening, pore blockage or deterioration of interfacial transport processes.

[0133] Preferably, the early warning output module comprehensively evaluates the increase in anode reaction potential, the degree of abnormality in the core electrochemical fingerprint, and the deactivation rate based on a pre-set risk threshold or an adaptive risk threshold established based on historical operating data, and generates the anode deactivation risk level, deactivation type, early warning reason, and corresponding operation control suggestions.

[0134] In addition to the above, the present invention also has related embodiments, including the following:

[0135] In one embodiment of the present invention, the acquisition of time-series electrochemical data of the electrodeposition anode includes the following:

[0136] This embodiment illustrates the acquisition methods for cyclic voltammetric data, linear scan voltammetric data, and cell voltage data during the operation of the electrowinning anode.

[0137] The oxygen-evolving anode in the acidic metal electrowinning system is selected as the monitoring target. The oxygen-evolving anode can be a titanium-based inert anode, an IrO2-Ta2O5 / Ti coated anode, a lead alloy anode, a mixed metal oxide coated anode, or other anodes suitable for acidic electrowinning systems.

[0138] The electrodeposition system can be an acidic electrodeposition system of copper, zinc, nickel, cobalt or manganese, or a resource-based electrodeposition system of waste battery leachate, hydrometallurgical leachate or industrial wastewater containing heavy metals.

[0139] During electrowinning operation, the cell voltage is continuously or at preset time intervals, and cyclic voltammetry and linear sweep voltammetry tests are performed on the electrowinning anode at different cumulative operating times.

[0140] The same potential range, scan rate, electrolyte composition, electrolyte temperature, effective electrode area, and testing device were used at different sampling times to reduce the impact of changes in testing conditions on the comparison results of electrochemical responses at different sampling times.

[0141] The sampling time is determined based on the operating cycle of the electrodeposited anode, the rate of change of the electrochemical response, and the anode maintenance cycle. In one embodiment, electrochemical tests can be performed during the initial operation phase, the intermediate operation phase, and the long-term operation phase of the anode to obtain time-stamped cyclic voltammetric responses and linear sweep voltammetric responses.

[0142] like Figure 2 As shown, the cyclic voltammetric response of the electrodeposited anode is collected at different operating times, and one or more features among oxidation peak current, reduction peak current, oxidation peak potential, reduction peak potential, cyclic voltammetric integrated charge, and overall redox response intensity are extracted from the cyclic voltammetric response.

[0143] Cyclic voltammetric characteristics obtained at different sampling times were compared with baseline characteristics during the healthy operation phase of the anode. When the oxidation peak current, reduction peak current, cyclic voltammetric integrated charge, or overall redox response intensity continuously decreases, it indicates that the effective active area or accessibility of active sites of the anode may be reduced, which can be used as a candidate electrochemical fingerprint of anode activity decay.

[0144] It should be noted that the decrease in cyclic voltammetric response is used to characterize changes in the effective active area or accessibility of active sites at the anode, and is not used alone to determine the specific reasons for the reduction in the number of active sites at the anode.

[0145] The changes in the cyclic voltammetric response can be further analyzed in conjunction with structural features and operational information such as anode surface deposit coverage, film densification, active component dissolution, impurity ion adsorption, and pore blockage.

[0146] After the data collection is completed, the cyclic voltammetric data, linear scan voltammetric data, and cell voltage data are uniformly labeled according to the sampling time, and correlated with operating condition parameters such as current density, electrolyte temperature, electrolyte flow rate, sulfuric acid concentration, target metal ion concentration, pH value, conductivity, and cumulative operating time to form an anode time-series state dataset.

[0147] When structural characterization conditions are available, one or more structural parameters, such as anodic film thickness, surface morphology, pore structure, deposit coverage, and surface elemental composition, can be acquired simultaneously or according to a preset cycle, and the structural parameters can be correlated with electrochemical data and operating condition parameters over time.

[0148] In one embodiment of the present invention, the extraction of anode dynamic features and the screening of core fingerprints include the following:

[0149] This embodiment illustrates a method for extracting anodic polarization and reaction kinetics characteristics using linear scanning voltammetric response, and for screening core electrochemical fingerprints from candidate electrochemical fingerprints.

[0150] The cyclic voltammetric characteristics, linear scan voltammetric characteristics, cell voltage data, anode structure parameters, and operating condition parameters obtained in Example 1 were aligned according to the sampling time.

[0151] The linear sweep volt-ampere response of the electrodeposited anode was collected at different operating times, and the anode polarization response at different sampling times was compared.

[0152] A preset current density is selected from the linear scan current-voltage response at different sampling times, and the anode potential corresponding to the preset current density is extracted.

[0153] When the anodic potential at a fixed current density continuously increases relative to the healthy state baseline, it indicates that the driving force required for the anodic to complete the same reaction rate has increased, and this can be used to characterize changes in the degree of anodic polarization and reactivity.

[0154] Furthermore, by selecting a fitting interval from the dynamic control region of the linear sweep volt-ampere response, the Tafel slope at different running times is obtained.

[0155] To ensure the comparability of Tafel slopes at different sampling times, the fitting intervals for each sampling time are selected according to the same rules, avoiding low-current regions where the background current has a significant impact and high-current regions where mass transfer is significantly limited.

[0156] When the anolyte potential at a constant current density continuously increases and the Tafel slope gradually increases, it indicates a deterioration trend in the anolyte reaction kinetics. Therefore, the anolyte potential at a constant current density, the degree of shift in the linear sweep voltammetric response, and the Tafel slope can be used as candidate electrochemical fingerprints.

[0157] The candidate electrochemical fingerprint may also include one or more of the following: cyclic voltammetric peak current, cyclic voltammetric integral charge, reaction overpotential, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance.

[0158] Considering the potential information duplication among different candidate electrochemical fingerprints, one or more methods, including correlation analysis, redundancy analysis, significance test, and feature importance evaluation, are used to screen candidate electrochemical fingerprints.

[0159] First, we analyze the correlation between candidate electrochemical fingerprints and anode reaction potential, cell voltage changes, maintenance and replacement records, or preset deactivation state tags, and remove features with low correlation to anode state changes.

[0160] Subsequently, the degree of information redundancy among different candidate electrochemical fingerprints was analyzed. When two or more candidate electrochemical fingerprints have similar expressive effects on the same anodic state change, representative features were retained based on feature stability, measurement repeatability, physical meaning, and model contribution.

[0161] During the fingerprint screening process, electrochemical fingerprints that can characterize the effective active area of ​​the anode, reaction kinetics, interfacial charge transfer, film transport state, and the degree of competition for side reactions are given priority.

[0162] Among them, the cyclic voltammetric peak current, the cyclic voltammetric integral charge, and the overall redox response intensity can be used to characterize changes in the effective active area or the accessibility of active sites; the anodic potential and Tafel slope at a fixed current density can be used to characterize changes in anodic reaction kinetics; the charge transfer resistance and pore resistance can be used to characterize the interfacial charge transfer and the transport state of the membrane pores; and the oxygen evolution current efficiency can be used to characterize the competition between the target oxygen evolution reaction and side reactions.

[0163] The core electrochemical fingerprints obtained after screening, together with the anode structure parameters and operating condition parameters, constitute the input features of the mechanism knowledge model and the machine learning residual correction model.

[0164] In one embodiment of the present invention, the tank voltage analysis and anode state determination include the following:

[0165] This embodiment illustrates the processing method of cell voltage timing data and the method of judging the anode operating status by combining multiple electrochemical fingerprints.

[0166] During constant current electrowinning, the cell voltage is continuously or collected at a preset sampling frequency to form time-stamped cell voltage timing data.

[0167] The change in cell voltage relative to the baseline value during the healthy operation phase, the rate of change per unit time, and the degree of fluctuation within a preset time window are extracted from the cell voltage time series data. The operational stability of the electrowinning process is evaluated based on the change, rate of change, and degree of fluctuation.

[0168] When the increase in cell voltage and the rate of increase in cell voltage continue to increase, it indicates that the total polarization of the electrodeposition system or its operating energy consumption may be increasing.

[0169] When the fluctuation of the cell voltage continues to increase, it indicates that the risk of decreased operational stability of the electrowinning process may be increasing. Such abnormal changes in cell voltage may be related to one or more of the following factors: bubble coverage, localized changes in mass transfer, uneven distribution of deposits on the anode surface, changes in connection status, or localized anode activity decay.

[0170] Since the cell voltage is affected by factors such as anodic polarization, cathodic polarization, ohmic voltage drop of the electrolyte, electrode spacing and contact resistance, this invention does not use abnormal cell voltage as the sole criterion for anodic deactivation, but rather analyzes it in conjunction with cyclic voltammetric response, linear sweep voltammetric response, anodic reaction potential and other operating condition parameters.

[0171] In the cell voltage timing data, the times of cyclic voltammetry testing, linear sweep voltammetry testing, sampling, electrolyte adjustment, shutdown operation, and equipment maintenance can be marked to distinguish between instantaneous voltage changes caused by test operations or operational adjustments and continuous changes caused by long-term anode operation.

[0172] like Figure 3 As shown, the cyclic voltammetric response, linear sweep voltammetric response, cell voltage timing data, anode reaction potential, anode structural parameters, and operating condition parameters are used as inputs for multi-source anode state data.

[0173] The multi-source anode state data is subjected to one or more preprocessing steps, including background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment, and candidate electrochemical fingerprints are extracted from the preprocessed data.

[0174] Electrochemical fingerprints from different sources are normalized and feature-fused to form a fused feature that can reflect the effective active area of ​​the anode, reaction kinetics, operational stability, interfacial transport state, and the degree of competition for side reactions.

[0175] Further analysis was conducted to determine the correlation between the cell voltage characteristics and the cyclic voltammetric characteristics, linear sweep voltammetric characteristics, anode reaction potential, and operating parameters.

[0176] When the peak current of cyclic voltammetry or the integral charge of cyclic voltammetry decreases, while the anode potential, Tafel slope and cell voltage at a fixed current density continue to increase, it indicates that the effective active area of ​​the anode, the accessibility of active sites and reaction kinetics may undergo abnormal changes simultaneously.

[0177] The correlation analysis is used to screen core electrochemical fingerprints related to anodic deactivation behavior, and is not used alone to determine the specific cause of anodic deactivation.

[0178] Based on the degree of deviation of the core electrochemical fingerprint from the health status benchmark and the duration of the abnormal state, the anode operating status is divided into healthy state, attention state, and high-risk state.

[0179] When the cell voltage and anode reaction potential change steadily, the cyclic voltammetric response does not show a significant decrease, and the anode potential and Tafel slope do not show a continuous abnormality under a fixed current density, the anode is judged to be in a healthy state.

[0180] When the cell voltage rises slowly, or when there are persistent abnormalities in some features of the cyclic voltammetric response, anode potential at a fixed current density, Tafel slope, and other core electrochemical fingerprints, the anode is determined to be in a state of interest.

[0181] When the rate of increase of the cell voltage or the anode reaction potential increases significantly, while the peak current of the cyclic voltammetry or the integral charge of the cyclic voltammetry continues to decrease, the anode potential under a fixed current density continues to increase, and the Tafel slope increases, it is determined that the anode has a high risk of deactivation.

[0182] For anodes under monitoring, the frequency of electrochemical testing and operational condition data acquisition can be increased, and the temperature, flow rate, acid concentration, target metal ion concentration, impurity ion concentration, and connection status of the electrolyte should be checked for any abnormalities.

[0183] For anodes in a high-risk state, one or more of the following measures can be taken, depending on the actual operating conditions: reducing current density, adjusting electrolyte flow rate, adjusting electrolyte temperature, optimizing electrolyte composition, cleaning the anode surface, or replacing the anode.

[0184] In one embodiment of the present invention, the mechanistic knowledge model, residual correction, and system application include the following:

[0185] This embodiment is used to illustrate the mechanism knowledge model, machine learning residual correction, and the system application of the present invention in the electrodeposition resource utilization process.

[0186] like Figure 4 As shown, the system of the present invention includes an operating condition acquisition module, an anode structure characterization module, a time-series electrochemical spectrum acquisition module, an electrochemical fingerprint extraction module, a fingerprint screening module, a mechanism knowledge model construction module, a machine learning residual correction module, an inactivation trend and type determination module, and an early warning output module.

[0187] During system application, the cell voltage is continuously or at preset time intervals by the voltage acquisition unit, and the anode reaction potential, cyclic voltammetric response, linear scan voltammetric response and other required electrochemical data are acquired by the reference electrode and electrochemical workstation.

[0188] Simultaneously, operating parameters such as current density, electrolyte temperature, electrolyte flow rate, sulfuric acid concentration, target metal ion concentration, pH value, conductivity, and cumulative operating time are collected. When structural characterization conditions are available, one or more structural parameters, including anolyte film thickness, surface deposit coverage, pore structure, crack distribution, and surface elemental composition, can also be acquired at preset intervals.

[0189] The collected cell voltage, anode reaction potential, electrochemical response, anode structural parameters, and operating condition parameters are transmitted to the electrochemical fingerprint extraction module after being uniformly time-stamped.

[0190] The electrochemical fingerprint extraction module performs one or more of the following processes on the input data: background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment, and extracts candidate features from data from different sources.

[0191] Specifically, the oxidation peak current, reduction peak current, peak potential, cyclic voltammetric integrated charge, and overall redox response intensity are extracted from the cyclic voltammetric response; the anode potential, reaction overpotential, and Tafel slope at a fixed current density are extracted from the linear sweep voltammetric response; and the change in cell voltage relative to the healthy state baseline, the rate of change per unit time, and the degree of fluctuation within a preset time window are extracted from the cell voltage time series data.

[0192] The fingerprint screening module combines deactivation mechanisms such as anode surface deposition coverage, film thickening, active component dissolution, active site poisoning, pore blockage, and mass transfer restriction to perform one or more of the following on candidate features: correlation analysis, redundancy analysis, significance test, and feature importance evaluation, to obtain core electrochemical fingerprints that are related to anode deactivation behavior and have low redundancy.

[0193] The core electrochemical fingerprint is time-aligned and feature-fused with anode structural parameters and operating condition parameters to form the input features of the mechanism knowledge model and machine learning residual correction model.

[0194] The mechanistic knowledge model makes an initial prediction of the trend of anodic reaction potential based on electrochemical kinetics, ohmic polarization, mass transfer limitation, film impedance, and anodic deactivation accumulation.

[0195] The mechanistic knowledge model establishes a correspondence between the change in anodic reaction potential and one or more parameters among current density, electrolyte conductivity, electrode spacing, electrolyte flow rate, electrolyte temperature, target metal ion concentration, anodic potential at a fixed current density, reaction overpotential, Tafel slope, and membrane transport state.

[0196] Due to the presence of complex factors in actual electrodeposition systems, such as impurity ion adsorption, competition among multiple metal ions, bubble shielding, flow field disturbance, cathode state changes, and random distribution of deposits on the anode surface, there may be deviations between the initial prediction results of the mechanistic knowledge model and the measured anode reaction potential.

[0197] The machine learning residual correction module takes the core electrochemical fingerprint, anode structural parameters, and operating condition parameters as inputs, and uses the deviation between the measured anode reaction potential and the initial prediction result of the mechanism knowledge model as the learning target to correct the prediction result of the mechanism knowledge model and obtain the corrected anode reaction potential prediction result.

[0198] The machine learning model may be one or more of the following: random forest regression, support vector regression, XGBoost, CatBoost, LightGBM, long short-term memory network, or an ensemble model composed of the above models.

[0199] The deactivation trend and type determination module, in accordance with the state determination method described in Example 3, combines the corrected anode reaction potential prediction results, the time-series change trend of the core electrochemical fingerprint, the cell voltage change trend, and the preset deactivation criteria to determine the anode deactivation state, deactivation rate, dominant deactivation type, and deactivation risk level.

[0200] The dominant inactivation types include one or more of the following: inactivation dominated by side reaction competition, inactivation dominated by film thickening or pore blockage, inactivation dominated by active site poisoning, and composite inactivation formed by the combined action of multiple inactivation mechanisms.

[0201] When the corrected anodic reaction potential prediction result or multiple core electrochemical fingerprints reach the corresponding preset inactivation criteria within multiple consecutive sampling cycles, the early warning output module generates anodic inactivation early warning information.

[0202] The early warning output module generates corresponding operation and control suggestions based on the anode deactivation risk level, abnormal core electrochemical fingerprint, and possible dominant deactivation type.

[0203] The operational control recommendations include increasing the detection frequency, reducing the current density, adjusting the electrolyte flow rate, adjusting the electrolyte temperature, adjusting the sulfuric acid concentration, optimizing the electrolyte composition, cleaning the anode surface, or replacing the anode, or one or more of these measures.

[0204] This embodiment transforms multi-source anode state data into a core electrochemical fingerprint with physical meaning by jointly processing cyclic voltammetric response, linear scan voltammetric response, anode reaction potential, cell voltage timing data, anode structural parameters, and operating condition parameters. Combined with mechanistic knowledge models and machine learning residual correction, it enables anode reaction potential prediction, deactivation trend analysis, determination of dominant deactivation type, and risk warning.

[0205] Compared to relying solely on abnormal cell voltage, single electrochemical test results, or manual shutdown for inspection, this invention can identify anode deactivation risks from multiple aspects, including the effective active area of ​​the anode, reaction kinetics, interfacial transport status, and electrodeposition operation stability, providing a basis for anode maintenance and electrodeposition process control.

[0206] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven approaches, characterized in that, Includes the following: Step S1: Collect the anodic operating parameters, structural characteristics, and time-series electrochemical scanning spectra of the electrodeposited anode; Step S2: Preprocess the acquired data, including background subtraction, noise filtering, baseline correction, outlier removal, normalization, and time alignment. Step S3: Extract candidate electrochemical fingerprints from the pre-processed time-series electrochemical scanning spectra, and screen them in conjunction with the anodic side reaction competition, film thickening and active site poisoning-induced inactivation mechanism to obtain the core electrochemical fingerprint; Step S4: Perform feature fusion on the anode operating condition parameters, structural features, and core electrochemical fingerprint to construct an input feature vector; Step S5: Construct a knowledge model based on electrochemical kinetics, ohmic polarization, mass transfer constraints, and anodic deactivation mechanism constraints to obtain the theoretical predicted value of the anodic reaction potential; Step S6: Using the input feature vector as the model input, and the residual between the measured anode reaction potential and the theoretical prediction value as the learning target, the theoretical prediction value is corrected using a machine learning model to obtain the corrected anode reaction potential prediction value. Step S7: Based on the corrected predicted anode reaction potential, the time-series change trend of the core electrochemical fingerprint, and the preset deactivation criteria, predict the change trend of the anode reaction potential within the preset time range in the future, determine the anode deactivation state, deactivation rate, dominant deactivation type, and deactivation risk level, and output corresponding operation and control suggestions.

2. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S1 includes the following: In step S1, the operating parameters, anode structural characteristics, and time-series electrochemical scanning spectra of the electrodeposition anode during the electrodeposition process are acquired at fixed intervals; the time-series electrochemical scanning spectra include cyclic voltammetry curves, polarization curves, and electrochemical impedance spectroscopy.

3. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S2 includes the following: In step S2, the time-series electrochemical scanning spectrum is preprocessed, and candidate electrochemical fingerprints are extracted from the preprocessed data. The preprocessing includes one or more of noise filtering, baseline correction, outlier removal, normalization, and time alignment. The candidate electrochemical fingerprints include reaction overpotential, Tafel slope, cyclic voltammetric integral charge, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance.

4. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S3 includes the following: In step S3, the candidate electrochemical fingerprints are screened based on the competition of anodic side reactions, film thickening, and active site inactivation mechanisms to obtain the core electrochemical fingerprint. The screening includes correlation analysis, redundancy analysis, and significance testing, including the following: First, calculate the correlation between candidate electrochemical fingerprints and anodic deactivation characterization parameters; The anodic deactivation characterization parameters include one or more of the following: measured anodic reaction potential, the increase in anodic reaction potential relative to the initial state, the rate of increase in anodic reaction potential per unit time, and anodic deactivation state labels determined according to a preset deactivation threshold. Candidate electrochemical fingerprints with a correlation lower than the preset threshold to the anodic deactivation characterization parameters are eliminated. The correlation or redundancy between candidate electrochemical fingerprints is then calculated to remove redundant fingerprints that repeatedly express changes in anodic state. Finally, in conjunction with the anodic deactivation mechanism, the core electrochemical fingerprints that can characterize changes in anodic reaction kinetics, film thickness, and reaction competition are retained.

5. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S4 includes the following: In step S4, the core electrochemical fingerprint, anode structure parameters, and operating condition parameters are time-aligned and feature-fused to form an input feature vector. The input feature vector includes one or more of the following parameters: reaction overpotential; Tafel slope; cyclic voltammetric integral charge; oxygen evolution current efficiency; charge transfer resistance; pore resistance; membrane capacitance; current density; electrolyte flow rate; electrolyte temperature; sulfuric acid concentration; and target recovered ion concentration.

6. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S5 includes the following: In step S5, the mechanism knowledge model includes an equilibrium potential term, a kinetic polarization term, and an ohmic polarization term, and further includes a mass transfer polarization term, a film impedance term, and a deactivation state term.

7. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S6 includes the following: In step S6, the machine learning model uses the difference between the measured anode reaction potential and the theoretical predicted anode reaction potential as the residual learning target, and superimposes the residual correction value output by the machine learning model with the theoretical predicted anode reaction potential to obtain the corrected anode reaction potential prediction value.

8. The method for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods according to claim 1, characterized in that, Step S7 includes the following: The operational control recommendations include one or more of the following: anode cleaning, anode replacement, reducing current density, increasing electrolyte flow rate, adjusting electrolyte temperature, and adjusting electrolyte composition.

9. A system for predicting electrodeposition anode deactivation based on electrochemical fingerprinting and knowledge co-driven methods, applied in any one of claims 1 to 8, characterized in that, It includes a working condition acquisition module, an anode structure characterization module, a time-series electrochemical spectrum acquisition module, an electrochemical fingerprint extraction module, a fingerprint screening module, a mechanism knowledge model construction module, a machine learning residual correction module, an inactivation trend and type determination module, and an early warning output module; The operating condition acquisition module is used to continuously or at preset time intervals acquire the operating parameters of the electrowinning reaction, including cell voltage, anode potential, current density, electrowinning liquid flow rate, electrowinning liquid temperature, sulfuric acid concentration, target recovered ion concentration, pH value, conductivity and cumulative operating time. The anode structure characterization module is used to obtain anode film thickness, specific surface area, pore structure, and structural parameters characterizing changes in effective active area or accessibility of active sites, for analyzing anode deactivation mechanisms, including side reaction competition, film thickening, and active site poisoning. The time-series electrochemical spectrum acquisition module is used to perform cyclic voltammetry, polarization scanning and electrochemical impedance spectroscopy tests on the electro-anode at different sampling times during the operation of the electro-anode, and obtain time-stamped cyclic voltammetry curves, polarization curves and electrochemical impedance spectra. The electrochemical fingerprint extraction module is used to extract features from time-series electrochemical scanning spectra; The fingerprint screening module is used to combine the anode deactivation mechanism and statistical methods, including correlation analysis, redundancy analysis and feature selection, to screen core electrochemical fingerprints that are related to the electrodeposition anode deactivation behavior and have low redundancy. The mechanism knowledge model construction module is used to construct an anode reaction potential prediction model and generate mechanism prediction values ​​based on electrochemical kinetic relationships, ohmic polarization relationships, mass transfer limitation relationships and anode deactivation mechanism constraints. The machine learning residual correction module takes the input feature vector as input and the residual between the theoretical prediction value and the measured potential as the learning target to correct the theoretical prediction value and obtain the corrected anode reaction potential prediction value. The inactivation trend and type determination module is used to predict the trend of anode reaction potential change within a preset time range based on the corrected anode reaction potential prediction value, the time-series change trend of the core electrochemical fingerprint and the preset inactivation criteria, and output the anode inactivation rate, inactivation state, dominant inactivation type and inactivation risk level. The early warning output module is used to generate the anode deactivation risk level based on the prediction results and provide control suggestions, including anode cleaning, replacement, current density adjustment, electrolyte flow rate adjustment, or electrolyte composition adjustment.

10. The electro-anode deactivation prediction system based on electrochemical fingerprinting and knowledge co-driven as described in claim 1, characterized in that, The electrochemical fingerprint includes one or more of the following: reaction overpotential, anodic potential at a fixed current density or its change relative to a healthy state baseline, Tafel slope, cyclic voltammetric integral charge, oxidation peak current, reduction peak current, redox peak position, oxygen evolution current efficiency, anodic bulk redox intensity, charge transfer resistance, pore resistance, and film capacitance. The deactivation trend and type determination module is used to output a predicted anodic reaction potential curve, a predicted confidence interval, and an estimated deactivation rate within a preset time range, and to evaluate the deactivation state, deactivation risk, and dominant driving mechanism of the electro-deposited anode based on the prediction results. The preset time range is determined based on one or more of the following: electrochemical spectrum sampling frequency, available historical operating data length, electro-deposited anode operating cycle, and anode maintenance cycle.

Images