Copper electrolysis production quality closed-loop management and control method, system and equipment, medium and program product

Abstract

The invention relates to the field of electrolytic copper production, in particular to a copper electrolysis production quality closed-loop management and control method, which comprises the following steps: collecting cathode copper surface quality data and process parameters in real time, firstly carrying out data cleaning and normalization, and realizing automatic defect classification in combination with univariate probability density analysis; then, adopting Graphical Lasso multivariable association mining to lock core process parameters, and fusing a Bayesian network and a CART decision tree to construct an anomaly prediction model; and dynamically tracing an abnormal root based on a graph computing technology and generating a key path, and finally outputting a data-driven process parameter adjustment suggestion. By applying a single-factor and multi-factor analysis method and combining a decision tree model and a graph calculation technology, abnormal conditions are identified and early warned, factors with the strongest association are found out, reason tracing is carried out, and auxiliary guidance suggestions of process parameters are provided for teams and groups, so that the overall quality of cathode copper in the electrolysis production process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrolytic copper production, and specifically provides a quality closed-loop control method, system, equipment, medium and program product for copper electrolysis production. Background Art

[0002] Electrolytic copper is prepared by pre-making thick plates of crude copper as anodes, using the ISA method (permanent stainless steel cathode method) with stainless steel plates as cathodes, and a mixed solution of sulfuric acid and copper sulfate as the electrolyte. After power-on, copper dissolves from the anode into copper ions and moves towards the cathode. After reaching the cathode, it obtains electrons and pure copper is deposited on the cathode surface. The deposited pure copper can be peeled off from the stainless steel cathode plate through a peeling device, thus realizing continuous production.

[0003] The existing Chinese patent publication number is: CN114774990B, which provides a method for optimizing process parameters of copper electrolysis for energy conservation. This patent constructs a regression model of the production process parameters of copper electrolysis and electrolysis energy consumption through a random forest algorithm, and calculates the electrolysis energy consumption at different current densities, copper ion concentrations, sulfuric acid concentrations and electrolysis temperatures. Based on this model, an energy consumption optimization model for the copper electrolysis process is constructed, and a competitive swarm optimization algorithm introducing neighborhood control and adaptive mutation strategies is used for optimization and solution to obtain the optimal production process parameters. However, this method mainly focuses on the optimization of power consumption, and quality control in the copper electrolysis process is equally important. This invention does not involve the quality control of electrolytic copper.

[0004] The existing Chinese patent publication number is: CN103510106A, which provides a copper electrolysis additive and its usage method. This additive is used for copper electrolysis refining, making the cathode copper crystals dense, reducing dendritic crystals, canceling thiourea in the traditional additive, reducing the sulfur content in the cathode copper, enhancing the flocculation and sedimentation of anode mud, and ensuring product quality. However, this invention mainly relies on experimental methods to test the dosage of the additive, and has not established a data model to further optimize and predict the optimal dosage of the additive. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a quality closed-loop control method, system, equipment, medium and program product for copper electrolysis production, which solves the problems of long defect detection response time in traditional copper electrolysis production and fuzzy quality traceability path in complex production environments.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A quality closed-loop control method for copper electrolysis production includes the following steps:

[0007] S1. According to the growth situation of cathode copper during the electrolysis process, analyze the surface quality data of cathode copper collected by an industrial camera scanning the cathode plate surface, and classify defect abnormalities according to the performance indicators on the cathode copper surface.

[0008] S2. Normalize and clean the production condition data, production process data, and production control data to ensure the accuracy and consistency of the data;

[0009] S3. Based on the original data, conduct univariate analysis, fit the probability density distribution function of each type of defect respectively, as the basis for subsequent analysis; meanwhile, if parameters that significantly affect the generation of defects can be identified from the probability distribution diagram, add them to the parameter candidate set;

[0010] S4. Adopt the graphical lasso mining method to conduct multivariate joint analysis, analyze the influence of key process parameters on the conditional probability of abnormal types, and identify potential associated factors;

[0011] S5. For the associated factors of various abnormalities mined by graphical lasso, take the union of process parameters, add them to the parameter candidate set, and form a comprehensive list of associated factors;

[0012] S6. Based on the process parameters selected by graphical lasso, conduct multivariate Bayesian analysis to generate a conditional probability diagram, visually describing the relationship between various factors;

[0013] S7. Based on the process parameters selected by graphical lasso, conduct CART decision tree modeling to construct a decision tree model for abnormal classification and prediction;

[0014] S8. Conduct quality traceability of electrolysis abnormalities, construct a process parameter relationship network based on the Neo4j graph database, and on the basis of the path ranking algorithm, evaluate the importance of different quality event paths in the decision tree to determine the key paths and related nodes in the quality traceability process;

[0015] S9. Sort the paths according to the scores of the paths, and based on the results of path ranking, select the top three paths with the highest probability as the quality traceability results for display, providing a clear traceability direction for operators;

[0016] S10. According to the quality traceability results, dynamically suggest that the generation unit generate auxiliary suggestions for process parameter adjustment through the rule engine combined with case reasoning, providing specific parameter adjustment guidance for operators.

[0017] Preferably, the performance indicators in step S1 include, but are not limited to, nodule shape, area, height, color, and glossiness.

[0018] Preferably, in step S3, the probability density distribution function is as follows:

[0019]

[0020] Where μ is the mean and σ is the standard deviation. The distribution of each defect index is analyzed through this function.

[0021] Preferably, in the step S4, the graphical lasso formula is:

[0022]

[0023] Where S is the sample covariance matrix, Θ is the precision matrix, and λ is the regularization parameter. The potential associations between factors are identified through this method.

[0024] Preferably, in the step S6, the multivariate Bayesian analysis formula is as follows:

[0025]

[0026] Where P(X|Y) is the conditional probability, P(X|Y) is the likelihood probability, P(Y) is the prior probability, and P(X) is the marginal probability.

[0027] Preferably, in the step S7, the CART decision tree algorithm uses the Gini index to select the splitting attribute, and the purity of the dataset D can be represented by the Gini value as:

[0028]

[0029] Gini(D) reflects the probability that two randomly selected samples from the dataset D have inconsistent class labels; therefore, the smaller Gini(D) is, the higher the purity of the dataset D.

[0030] Preferably, a closed-loop quality control system for copper electrolysis production includes:

[0031] A data collection and storage module, which is used to collect production process data in real time, provide an analysis basis, and store the collected data;

[0032] A data processing and analysis module, which is used to clean the collected data and mine the correlations of process parameters;

[0033] A modeling and prediction module, which is used to build a model to achieve anomaly classification and prediction;

[0034] A quality traceability and decision-making module, which is used to locate the root cause of anomalies and provide control suggestions;

[0035] A visualization and interaction module, which is used to intuitively display the analysis results to guide on-site operations.

[0036] Preferably, a closed-loop quality control device for copper electrolysis production includes:

[0037] A process sensor array, including an industrial camera for scanning the surface of the cathode plate and other sensor devices, is used to collect the surface quality data of cathode copper and the key process parameters in the production process;

[0038] An industrial gateway is used to complete the calculations of data normalization and outlier filtering at the device end, reducing the data transmission load;

[0039] A multivariate analysis server is used to execute complex algorithm models to achieve anomaly correlation analysis and prediction.

[0040] The present invention provides a method, system, device, medium and program product for closed-loop quality control in copper electrolysis production. It has the following beneficial effects:

[0041] 1. By establishing a mathematical model and integrating expert experience, and applying advanced technologies such as artificial intelligence, machine learning, and big data analysis, the present invention realizes a closed-loop intelligent control method for the quality of the cathode copper production process that can provide scientific regulation decisions.

[0042] 2. By collecting the surface quality data and other process parameters in the cathode copper production process in real time, the present invention determines and conducts closed-loop control on the abnormal situations in the electrolysis process. Using single-factor and multi-factor analysis methods, combined with decision tree models and graph computing technologies, it identifies and warns of abnormal situations, finds the factors with the strongest correlation, traces the causes, and provides auxiliary guidance opinions on process parameters for the work teams. It provides process decisions and auxiliary operation guidance for the production site to improve the overall quality of cathode copper in the electrolysis production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic flow chart of the closed-loop quality control method for production of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0045] Embodiment 1:

[0046] As Figure 1 shown, the embodiment of the present invention provides a method for closed-loop quality control in copper electrolysis production, including the following steps:

[0047] Step 1: According to the growth of cathode copper during the electrolysis process, analyze the surface quality data of cathode copper collected by the industrial camera scanning the cathode plate surface, and classify the abnormalities of the nodule shape, area, height, color, and glossiness indicators on the cathode copper surface. Based on the experience of the operators, these abnormalities are divided into different categories and marked using one-hot encoding. For example, abnormality 1 is encoded as 0001, abnormality 2 is encoded as 0010, and so on.

[0048] Step 2: Normalize and clean the production condition data, production process data, and production regulation data. The normalization formula is:

[0049]

[0050] Through this formula, the data is normalized between 0 and 1, and at the same time, missing values and outliers are removed to ensure the accuracy and consistency of the data.

[0051] Step 3: Based on the original data, conduct univariate analysis and fit the probability density distribution function of various defects. The probability density function formula is:

[0052]

[0053] where μ is the mean and σ is the standard deviation. Through this function, analyze the distribution of each defect index; at the same time, if parameters that significantly affect the generation of defects can be identified from the probability distribution diagram, add them to the parameter candidate set.

[0054] Step 4: Adopt the graphical lasso mining method to conduct multivariate joint analysis and analyze the influence of key process parameters on the conditional probability of abnormal types. The graphical lasso optimization problem formula is:

[0055]

[0056] where S is the sample covariance matrix, Θ is the precision matrix, and λ is the regularization parameter. Through this method, identify the potential associations between various factors.

[0057] Step 5: For the associated factors of each abnormality mined by graphical lasso, take the union of the process parameters, add them to the parameter candidate set, and form a comprehensive list of associated factors for subsequent analysis and model construction.

[0058] Step 6: Based on the process parameters selected by graphical lasso, conduct multivariate Bayesian analysis to generate a conditional probability graph, intuitively display the relationships between various factors, and provide a basis for decision-making. The multivariate Bayesian analysis formula is:

[0059]

[0060] Among them, P(Y|X) is the conditional probability, P(X|Y) is the likelihood probability, P(Y) is the prior probability, and P(X) is the marginal probability. The conditional probabilities between various factors are calculated through this formula to generate a conditional probability graph.

[0061] Step 7: Based on the process parameters selected by graphical lasso, perform CART decision tree modeling to construct a decision tree model for anomaly classification and prediction, improving the interpretability and prediction accuracy of the model. The CART decision tree algorithm uses the Gini index to select the splitting attribute. The purity of the dataset D can be represented by the Gini value as:

[0062]

[0063] Gini(D) reflects the probability that two randomly selected samples from the dataset D have inconsistent class labels. Therefore, the smaller Gini(D) is, the higher the purity of the dataset D. The Gini index of the attribute a is defined as:

[0064]

[0065] Step 8: Conduct quality traceability of electrolysis anomalies. Based on the Neo4j graph database, construct a process parameter relationship network. Using the Neo4j graph database, take process parameters, quality events, and anomaly classification results as nodes, and the correlation between parameters as edges to construct a process parameter relationship network. Based on the path ranking algorithm, evaluate the importance of different quality event paths in the decision tree, determine the key paths and related nodes in the quality traceability process, and the calculation method is as follows:

[0066] Suppose the associated importance of the event in the defect path causing the defect to occur:

[0067]

[0068] For the defective electrolytic cathode copper I that needs to be traced, the event set N that satisfies the path events in the parameter data can be obtained. Therefore, the probability that the defective steel I is traced to this defect path is:

[0069]

[0070] [·] represents the Iverson bracket, which is 1 if the condition inside the square brackets is satisfied and 0 if not.

[0071] Step 9: Sort the paths according to the scores of the paths to determine the importance and relevance of the paths. The paths with higher probability scores are considered to be more relevant to the quality defects.

[0072] Step 10: Based on the results of path ranking, select the top three paths with the highest probabilities as the quality traceability results for retention.

[0073] Step 11: According to the quality traceability results, generate auxiliary suggestions for process parameter adjustment, providing specific parameter adjustment guidance for operators to ensure the stability of the production process and the improvement of product quality.

[0074] Step 12: Combining the experience of operators, the dynamic suggestion generation unit uses a rule engine combined with case-based reasoning to adjust the electrolysis process parameters, ensuring the practicality and effectiveness of the adjustment suggestions, and realizing the optimization of the production process and the continuous improvement of quality.

[0075] Example Two:

[0076] An embodiment of the present invention provides a closed-loop quality control system for copper electrolysis production, including:

[0077] A data acquisition and storage module for real-time acquisition of production full-process data, providing an analysis basis, and storing the acquired data. The data acquisition and storage module includes a production data acquisition unit and a data storage center. The production data acquisition unit is used to integrate sensors and databases to collect production conditions, process data, and control records, specifically including collecting cathode copper surface morphology data (such as surface flatness, defect distribution, etc.) through an industrial camera scanning the surface of the cathode plate, and collecting key process parameters (such as electrolyte temperature, current density, pH value, etc.) in the production process through other sensor devices; the data storage center uses a time-series database to store the original data and supports efficient query;

[0078] A data processing and analysis module for cleaning the acquired data and mining the correlation of process parameters; the data processing and analysis module includes a data preprocessing unit, a single-variable analysis unit, and a multi-variable correlation analysis unit; the data preprocessing unit is used to eliminate the dimension difference and process missing values and outliers; the single-variable analysis unit is used to fit the probability distribution of cathode copper defect indicators and identify the single-parameter abnormal threshold; the multi-variable correlation analysis unit constructs a sparse inverse covariance matrix based on the Graphical Lasso algorithm to identify key parameters; finally, according to the single-variable analysis results and multi-variable correlation analysis results, a list of associated factors is generated.

[0079] A modeling and prediction module for building a model to achieve anomaly classification and prediction; the modeling and prediction module includes a Bayesian network unit and a decision tree model unit. The Bayesian network unit generates a conditional probability graph (such as the probability chain of electrolyte temperature → additive concentration → dendritic defect) based on the Markov blanket theory; the decision tree model unit constructs a classification tree using the CART algorithm and selects the splitting node with information gain (such as "current density > 250A / m 2”As a splitting condition), output the decision rule for the abnormal type.

[0080] The quality traceability and decision-making module is used to locate the root cause of the abnormality and provide regulation suggestions; the quality traceability and decision-making module includes a graph calculation traceability unit and a dynamic suggestion generation unit. The graph calculation traceability unit constructs a process parameter relationship network based on the Neo4j graph database and uses the path ranking algorithm to evaluate the path weights; the dynamic suggestion generation unit combines case reasoning through a rule engine to match the historical optimal parameter combination (for example, when needle-like particles appear on the surface of cathode copper, it is recommended to reduce a specific amount of hydrochloric acid additive to reduce the chloride ion concentration).

[0081] The visualization and interaction module is used to intuitively display the analysis results to guide on-site operations.

[0082] Example 3:

[0083] The embodiment of the present invention provides a quality closed-loop control device for copper electrolysis production, including:

[0084] The process sensor array includes a temperature sensor, a current density detector, an electrolyte composition analyzer, and an industrial camera for scanning the surface of the cathode plate, and is used to collect the surface morphology data of the cathode copper and the production condition data;

[0085] The industrial gateway is used to complete the calculation of data normalization and outlier filtering at the device end to reduce the data transmission load;

[0086] The multi-variable analysis server is used to execute complex algorithm models to realize abnormal correlation analysis and prediction.

[0087] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirits of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for closed-loop quality control in copper electrolysis production, characterized in that: It includes the following steps: S1. Analyze the surface quality data of cathode copper collected by an industrial camera scanning the cathode plate surface according to the growth of cathode copper during the electrolysis process, and classify defect anomalies according to the performance indicators of the cathode copper surface; S2. Perform normalization processing and data cleaning on production condition data, production process data, and production regulation data to ensure the accuracy and consistency of the data; S3. Based on the original data, conduct univariate analysis, fit the probability density distribution function of each type of defect, which serves as the basis for subsequent analysis; meanwhile, if parameters that significantly affect the generation of defects can be identified from the probability distribution diagram, add them to the parameter candidate set; S4. Use the graphical lasso mining method to conduct multivariate joint analysis, analyze the influence of key process parameters on the conditional probability of abnormal types, and identify potential associated factors; S5. For the associated factors of various anomalies mined by graphical lasso, take the union of process parameters, add them to the parameter candidate set, and form a comprehensive list of associated factors; S6. Based on the process parameters selected by graphical lasso, conduct multivariate Bayesian analysis to generate a conditional probability diagram, visually describing the relationship between various factors; S7. Based on the process parameters selected by graphical lasso, conduct CART decision tree modeling to construct a decision tree model for anomaly classification and prediction; S8. Conduct quality traceability of electrolysis anomalies, construct a process parameter relationship network based on the Neo4j graph database, and evaluate the importance of different quality event paths in the decision tree based on the path ranking algorithm to determine the key paths and related nodes in the quality traceability process; S9. Sort the paths according to the scores of the paths, and based on the results of the path ranking, select the top three paths with the highest probability as the quality traceability results for display, providing a clear traceability direction for operators; S10. According to the quality traceability results, the dynamic suggestion generation unit combines case reasoning through a rule engine to generate auxiliary suggestions for process parameter adjustment, providing specific parameter adjustment guidance for operators.

2. The quality closed-loop control method for copper electrolysis production according to claim 1, characterized in that: The performance indicators in step S1 include but are not limited to nodule shape, area, height, color, and glossiness.

3. A quality closed-loop control method for copper electrolysis production according to claim 1, characterized in that: In step S3, the probability density distribution function is as follows: where μ is the mean and σ is the standard deviation, and the distribution of each defect index is analyzed through this function.

4. A quality closed-loop control method for copper electrolysis production according to claim 1, characterized in that: In step S4, the graphical lasso formula is: where S is the sample covariance matrix, Θ is the precision matrix, and λ is the regularization parameter, and potential associations between various factors are identified through this method.

5. A quality closed-loop control method for copper electrolysis production according to claim 1, characterized in that: In step S6, the multivariate Bayesian analysis formula is as follows: where P(Y|X) is the conditional probability, P(X|Y) is the likelihood probability, P(Y) is the prior probability, and P(Y) is the marginal probability.

6. A quality closed-loop control method for copper electrolysis production according to claim 1, characterized in that: In step S7, the CART decision tree algorithm uses the Gini index to select the splitting attribute, and the purity of the dataset D can be expressed by the Gini value as: Gini(D) reflects the probability that two randomly selected samples from the dataset D have inconsistent class labels; therefore, the smaller Gini(D) is, the higher the purity of the dataset D.

7. A quality closed-loop control system for copper electrolysis production, characterized in that: It includes: A data collection and storage module, which is used to collect production full-process data in real time, provide an analysis basis, and store the collected data. A data processing and analysis module, which is used to clean the collected data and mine the correlation of process parameters. A modeling and prediction module, which is used to build a model to achieve anomaly classification and prediction. A quality traceability and decision-making module, which is used to locate the root cause of anomalies and provide control suggestions. A visualization interaction module, which is used to intuitively display the analysis results to guide on-site operations.

8. A quality closed-loop control equipment for copper electrolysis production, characterized in that: It includes: A process sensor array, including an industrial camera for scanning the surface of the cathode plate and other sensor devices, which is used to collect the surface quality data of cathode copper and the key process parameters in the production process. An industrial gateway, which is used to complete the calculation of data normalization and outlier filtering at the device end to reduce the data transmission load. A multivariate analysis server, which is used to execute complex algorithm models to achieve anomaly correlation analysis and prediction.

9. An electronic device, characterized in that: It includes: One or more processors; A storage device, which is used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any one of claims 1-6.

10. A readable storage medium, characterized in that: A program or instruction is stored on the readable storage medium, and when the program or instruction is executed by a processor, the steps of the comment generation information processing method described in any one of claims 1-6 are implemented.

11. A computer program product, characterized in that: When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device executes the steps of the copper electrolysis production quality closed-loop control method described in any one of claims 1-6.

Images