Methods, apparatus, and electronic equipment for handling anolyte current during alumina electrolysis

Abstract

This application provides a method, apparatus, and electronic device for processing the anode current during alumina electrolysis. The method includes: acquiring multiple historical sub-anode currents for each sub-anode; inputting all historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model; acquiring the actual sub-anode current of the sub-anode and the predicted anode current output by the neural network model, and performing regression processing on the neural network model based on the actual and predicted anode currents to obtain an optimized neural network model; inputting the actual sub-anode current into the optimized neural network model, and adjusting the alumina feeding rate and frequency accordingly based on the output of the optimized neural network model. This solves the problem of low accuracy in controlling the anode feeding rate and frequency of alumina due to the low accuracy of anode current prediction in existing methods during alumina electrolysis.

Description

Technical Field

[0001] This application relates to the field of alumina electrolysis technology, and more specifically, to a method, apparatus, and electronic device for processing the anolyte current during alumina electrolysis. Background Technology

[0002] The application of artificial intelligence technology to electrolytic aluminum control systems still faces challenges such as limited hardware resources, small data samples, unclear mechanisms (black-box computation), and low reliability. Currently, the major technological bottlenecks that urgently need to be overcome in constructing an intelligent control system for acid-process alumina electrolysis are concentrated in the following aspects:

[0003] 1. There is a lack of information acquisition methods for complex and dynamic environments, which serve as the input and training basis for intelligent strategies;

[0004] 2. Lack of efficient reinforcement learning strategies for robot operation under dynamic, low-sample-rate tasks;

[0005] 3. There is a lack of effective methods to solve the entire process of object simulation, strategy generation, task generalization, and experience transfer in robot operation learning tasks.

[0006] In summary, the existing scheme has low accuracy in predicting the anode current of alumina during the electrolysis process, which leads to low accuracy in controlling the amount and frequency of alumina feed. Summary of the Invention

[0007] The main objective of this application is to provide a method, apparatus, and electronic device for processing the anode current during alumina electrolysis, so as to at least solve the problem that the accuracy of the control of the alumina feed rate and frequency is low due to the low accuracy of the prediction of the anode current of alumina during the electrolysis process in the existing scheme.

[0008] To achieve the above objectives, according to one aspect of this application, a method for processing the anode current during alumina electrolysis is provided. The method includes: acquiring multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple said sub-anodes; inputting all said historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model, wherein the neural network model is trained using multiple sets of training data, each set of training data including: a first historical sub-anode current and a corresponding second historical sub-anode current acquired within a historical time period, wherein the first historical sub-anode current is the current of the sub-anode at a first historical moment, and the second historical sub-anode current is... The sub-anode current is the current of the sub-anode at the second historical moment, where the first historical moment precedes the second historical moment. The actual sub-anode current and the predicted anode current output by the neural network model are obtained. Based on the actual sub-anode current and the predicted anode current, regression processing is performed on the neural network model to obtain an optimized neural network model, thereby optimizing the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment. The actual sub-anode current is input into the optimized neural network model, and the amount and frequency of alumina feeding are adjusted accordingly based on the output of the optimized neural network model.

[0009] Optionally, the neural network model is subjected to regression processing based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, including: determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current; if it is determined that regression processing should be performed on the neural network model, the neural network model is subjected to regression processing to obtain the optimized neural network model.

[0010] Optionally, determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current includes: obtaining an anode current difference, where the anode current difference is the difference between the actual sub-anode current and the corresponding predicted anode current; determining to perform regression processing on the neural network model if the absolute value of the anode current difference is greater than or equal to a first difference threshold and the absolute value of the anode current difference is less than or equal to a second difference threshold; and determining not to perform regression processing on the neural network model if the absolute value of the anode current difference is less than the first difference threshold or greater than the second difference threshold.

[0011] Optionally, after performing regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, the method further includes: obtaining the anode resistance of the sub-anode corresponding to the actual sub-anode current based on the actual sub-anode current corresponding to the predicted anode current; and determining the operating state of the corresponding sub-anode based on the anode resistance, wherein the operating state is one of the following: overheating state, undercooling state, or normal state.

[0012] Optionally, determining the operating state of the corresponding sub-anode based on the anode resistance includes: obtaining a first resistance threshold and a second resistance threshold; determining the operating state of the corresponding sub-anode as the normal state when the anode resistance is greater than or equal to the first resistance threshold and the anode resistance is less than or equal to the second resistance threshold; determining the operating state of the corresponding sub-anode as the subcooled state when the anode resistance is less than the first resistance threshold; and determining the operating state of the corresponding sub-anode as the overheated state when the anode resistance is greater than the second resistance threshold.

[0013] Optionally, before inputting all the historical sub-anode currents into the neural network model, the method further includes: acquiring first data and second data, wherein the first data is the actual sub-anode current and the second data is the estimated anode current, wherein the estimated anode current is an estimate of the current of the sub-anode at a future time; determining a first precision of the first data and a second precision of the second data, wherein both the first precision and the second precision are preset precisions; and inputting the first precision and the second precision into the neural network model to limit the output of the neural network model.

[0014] Optionally, before inputting all the historical sub-anode currents into the neural network model, the method further includes: extracting multiple target features from a database using histogram of oriented gradients (HARQ), the target features being used to characterize features related to the sub-anode other than the anode current of the sub-anode, the database storing multiple target features and the anode current of the sub-anode; classifying the target features using a support vector machine (SVM) algorithm to obtain multiple classification features; and inputting all classification features into the neural network model to limit the output of the neural network model.

[0015] Optionally, adjusting the amount and frequency of alumina feeding based on the output of the optimized neural network model includes: obtaining the current mapping relationship output by the optimized neural network model, wherein the current mapping relationship is the relationship between the predicted anode current and time within a predetermined future time period; and adjusting the amount and frequency of alumina feeding based on the current mapping relationship.

[0016] According to another aspect of this application, an apparatus for processing the anode current during alumina electrolysis is provided. The apparatus includes a first acquisition unit, a first processing unit, a second acquisition unit, and a second processing unit. The first acquisition unit is used to acquire multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple sub-anodes. The first processing unit is used to input all the historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model. The neural network model is trained using multiple sets of training data, each set of training data including: a first historical sub-anode current and a corresponding second historical sub-anode current acquired within a historical time period, wherein the first historical sub-anode current is the sub-anode current at a first historical moment. The first historical moment is earlier than the second historical moment. The second acquisition unit is used to acquire the actual sub-anode current and the predicted anode current output by the neural network model, and to perform regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, thereby optimizing the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment. The second processing unit is used to input the actual sub-anode current into the optimized neural network model, and to adjust the amount and frequency of alumina feeding according to the output of the optimized neural network model.

[0017] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including a method for processing the anolyte current during any of the aforementioned alumina electrolysis processes.

[0018] By applying the technical solution of this application, the sub-anode currents of each sub-anode of the alumina anode are used as input and iterated continuously to improve the accuracy of the neural network model, thereby improving the accuracy of subsequent anode current prediction and achieving the purpose of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feeding are adjusted accordingly, thus solving the problem that the accuracy of alumina feeding amount and frequency control is low due to the low accuracy of anode current prediction in the electrolysis process of existing solutions. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0020] Figure 1 A hardware structure block diagram of a mobile terminal for performing a method for processing anolyte current during alumina electrolysis, according to an embodiment of this application, is shown.

[0021] Figure 2 A schematic flowchart of a method for processing the anolyte current during alumina electrolysis according to an embodiment of this application is shown.

[0022] Figure 3 A structural block diagram of an apparatus for processing the anolyte current during alumina electrolysis, according to an embodiment of this application, is shown. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] As introduced in the background section, the major technical bottlenecks that urgently need to be overcome in constructing an intelligent control system for acid-process alumina electrolysis are concentrated in the following aspects: a lack of information acquisition methods for complex dynamic environments, serving as the input and training basis for intelligent strategies; a lack of efficient reinforcement learning strategies for robot operations under dynamic, low-sample task conditions; and a lack of effective methods to solve the entire process of object simulation, strategy generation, task generalization, and experience transfer in robot operation learning tasks. In summary, existing solutions have low accuracy in predicting the anode current of alumina during electrolysis, resulting in low accuracy in controlling the amount and frequency of alumina feed. To address the problem of low accuracy in predicting the anode current of alumina during electrolysis, which leads to low accuracy in controlling the amount and frequency of alumina feed, embodiments of this application provide a method, apparatus, and electronic device for processing the anode current during alumina electrolysis.

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0028] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a method of processing the anolyte current during alumina electrolysis according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0029] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the method for processing the anolyte current during alumina electrolysis in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0030] This embodiment provides a method for processing the anolyte current during alumina electrolysis running on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0031] Figure 2 This is a schematic flowchart illustrating a method for processing the anolyte current during alumina electrolysis according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0032] Step S201: Obtain multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple sub-anodes mentioned above;

[0033] Step S202: Input all the aforementioned historical sub-anode currents into the neural network model to process the aforementioned historical sub-anode currents using the aforementioned neural network model. The aforementioned neural network model is trained using multiple sets of training data. Each set of training data includes the following acquired within a historical time period: a first historical sub-anode current and a corresponding second historical sub-anode current. The aforementioned first historical sub-anode current is the current of the sub-anode at the first historical moment, and the aforementioned second historical sub-anode current is the current of the sub-anode at the second historical moment. The aforementioned first historical moment is preceding the aforementioned second historical moment.

[0034] Step S203: Obtain the actual sub-anode current and the predicted anode current output by the neural network model, and perform regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain the optimized neural network model, so as to optimize the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment.

[0035] In step S203, the neural network model is subjected to regression processing based on the actual sub-anode current and the predicted anode current to obtain the optimized neural network model. This includes: determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current; and performing regression processing on the neural network model if it is determined that regression processing should be performed to obtain the optimized neural network model.

[0036] Specifically, regression processing is an iterative process. Based on the actual sub-anode current and the corresponding predicted anode current, it is determined whether to optimize the input features of the prediction model and adjust the model parameters to improve the accuracy and reliability of the prediction. If regression processing is not required, step S204 is executed. After each iteration update of the sub-anode current, the real-time current data is compared with the predicted current data. The comparison can be performed using various indicators, such as root mean square error (RMSE), mean absolute error (MAE), or correlation coefficient. These indicators can quantify the degree of difference between the predicted and actual results. Through comparison and quantification, the performance of the prediction model can be evaluated, and adjustments and improvements can be made based on the evaluation results. If there is a large error between the predicted and actual results, the input features of the prediction model can be optimized and the model parameters adjusted to improve the accuracy and reliability of the prediction. This comparison and quantification method can ensure the validity of the prediction model and provide real-time feedback information during dynamic simulation. Through continuous iteration and improvement, the performance of the prediction model can be gradually improved to better adapt to the actual situation of the alumina mixing process in the electrolytic cell.

[0037] In one embodiment of this application, determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current includes: obtaining an anode current difference, wherein the anode current difference is the difference between the actual sub-anode current and the corresponding predicted anode current; determining to perform regression processing on the neural network model when the absolute value of the anode current difference is greater than or equal to a first difference threshold and less than or equal to a second difference threshold; and determining not to perform regression processing on the neural network model when the absolute value of the anode current difference is less than the first difference threshold or greater than the second difference threshold.

[0038] Specifically, the model continuously optimizes its prediction accuracy by comparing the difference between the predicted and actual values ​​with multiple difference thresholds.

[0039] In one embodiment of this application, after performing regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, the method further includes: obtaining the anode resistance of the sub-anode corresponding to the actual sub-anode current based on the actual sub-anode current corresponding to the predicted anode current; and determining the operating state of the corresponding sub-anode based on the anode resistance, wherein the operating state is one of the following: overheating state, undercooling state, or normal state.

[0040] Specifically, confirming the working status allows for timely adjustment of electrolysis parameters, making the electrolytic cell more stable and energy-efficient.

[0041] In one embodiment of this application, determining the operating state of the corresponding sub-anode based on the anode resistance includes: obtaining a first resistance threshold and a second resistance threshold; determining the operating state of the corresponding sub-anode as the normal state when the anode resistance is greater than or equal to the first resistance threshold and less than or equal to the second resistance threshold; determining the operating state of the corresponding sub-anode as the subcooled state when the anode resistance is less than the first resistance threshold; and determining the operating state of the corresponding sub-anode as the overheated state when the anode resistance is greater than the second resistance threshold.

[0042] Specifically, the anode resistance is compared with the first resistance threshold and the second resistance threshold to know the working state of the sub-anode and ensure that the working state of the sub-anode can be maintained in a normal state. In the case of overcooling or overheating, it is necessary to adjust the input power of the electrolytic cell or control the amount of alumina fed.

[0043] In one embodiment of this application, before inputting all the aforementioned historical sub-anode currents into the neural network model, the method further includes: acquiring first data and second data, wherein the first data is the actual sub-anode current and the second data is the estimated anode current, wherein the estimated anode current is an estimate of the current of the sub-anode at a future time; determining a first precision of the first data and a second precision of the second data, wherein both the first precision and the second precision are preset precisions; and inputting the first precision and the second precision into the neural network model to limit the output of the neural network model so that the output is closer to the actual value.

[0044] Specifically, based on two preset accuracy values, the learning rate of the neural network model on data from different sources is controlled, thereby controlling the importance of anode current, temperature, aluminum electrolyte composition, and alumina concentration, making the neural network model more accurate and the data sampling range wider.

[0045] In one embodiment of this application, before inputting all the aforementioned historical sub-anode currents into the neural network model, the method further includes: extracting multiple target features from a database using histogram of oriented gradients (HARQ) technology, wherein the target features are used to characterize other features related to the sub-anode besides the anode current of the sub-anode, and the database stores multiple target features and the anode current of the sub-anode; classifying the target features using a support vector machine algorithm to obtain multiple classification features; and inputting all classification features into the neural network model to limit the output of the neural network model.

[0046] Step S204: Input the actual sub-anode current into the optimized neural network model, and adjust the amount and frequency of alumina feeding according to the output of the optimized neural network model.

[0047] In one embodiment of this application, adjusting the amount and frequency of alumina feeding according to the output of the optimized neural network model includes: obtaining the current mapping relationship output by the optimized neural network model, wherein the current mapping relationship is the relationship between the predicted anode current and time within a predetermined future time period; and adjusting the amount and frequency of alumina feeding according to the current mapping relationship.

[0048] Specifically, by adjusting the amount and frequency of alumina feeding based on the current mapping relationship, the electrolysis efficiency of alumina can be improved. The current directly reflects the electrolyte resistance, and the electrolyte composition is closely related to the resistance. Therefore, by controlling the amount and frequency of alumina feeding, the electrolyte composition can be effectively controlled, and the current can be kept stable within a reasonable range.

[0049] Through the above embodiments, the accuracy of the neural network model is improved by using the sub-anode currents of each sub-anode of the alumina anode as input and iterating continuously. This improves the accuracy of subsequent anode current prediction, thereby achieving the goal of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feed are adjusted accordingly. This solves the problem that the accuracy of alumina feed amount and frequency control is low due to the low accuracy of anode current prediction during the electrolysis process in existing solutions.

[0050] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the anode current processing method during alumina electrolysis will be described in detail below with reference to specific embodiments.

[0051] This embodiment relates to a dynamic simulation method for anode current. The dynamic simulation method for anode current includes: setting the data precision of the first data used for dynamic simulation of anode current and the data precision of the second data used for dynamic simulation of anode current to a first precision and a second precision, respectively. The first data includes historical measured anode current and historical electrolytic cell temperature, and the second data includes at least one of real-time alumina concentration distribution, real-time fire eye identification data, real-time global cell voltage, and real-time operating parameters. The second precision is higher than the first precision. Based on the first data with the first precision and the second data with the second precision, dynamic simulation of anode current is performed.

[0052] In one specific embodiment of this application, the above-mentioned dynamic simulation method for anode current in acid-process alumina electrolysis further includes: receiving input data for dynamic simulation of anode current in acid-process alumina electrolysis, wherein the input data includes real-time current and operating time of different anodes; dividing the total anode current corresponding to the acid-process alumina electrolytic cell into multiple sub-anode currents processed by multiple processes in a distributed environment, wherein the step of performing dynamic simulation of anode current in acid-process alumina electrolysis includes: based on the input data, predicting, comparing and adjusting the real-time current, temperature and operating time of each anode in each sub-domain through the above-mentioned multiple processes.

[0053] In one specific embodiment of this application, the above-mentioned dynamic simulation method for anode current in acid-process alumina electrolysis includes the following steps for iteratively updating the multiple sub-anode currents: after each iteration update of the sub-anode current, comparing and quantifying the predicted current data and real-time current data; marking data points with differences between 0.02mV and 30mV; and processing the temperature and anode operating time of the data points and incorporating them into the learning dataset.

[0054] After each iteration update of the sub-anode current, the predicted current data is compared with the real-time current data and quantitatively evaluated. This comparison process verifies the accuracy of the prediction model and provides feedback to further improve its predictive performance. First, real-time current data needs to be acquired, which can be achieved through sensors or monitoring systems. Real-time current data represents the actual current situation in the electrolyzer. Simultaneously, a prediction model is needed to generate corresponding current prediction data. The prediction model can be built based on historical data and machine learning algorithms. This model can use past current data and other relevant variables as input to predict the current situation over a future period. After each iteration update of the sub-anode current, the real-time current data is compared with the predicted current data. Comparisons can be made using various metrics, such as root mean square error (RMSE), mean absolute error (MAE), or correlation coefficient. These metrics quantify the difference between predicted and actual results. Through comparison and quantification, the performance of the prediction model can be evaluated, and adjustments and improvements can be made based on the evaluation results. If there is a large error between the predicted and actual results, the input features of the prediction model can be optimized and the model parameters adjusted to improve the accuracy and reliability of the prediction. This method of comparison and quantification can ensure the validity of the prediction model and provide real-time feedback information during dynamic simulation. Through continuous iteration and improvement, the performance of the prediction model can be gradually improved to better adapt to the actual situation of the alumina mixing process in the electrolytic cell.

[0055] In one specific embodiment of this application, the above-described dynamic simulation method for anode current in acid-process alumina electrolysis includes the following steps for performing iterative updates: for each subdomain, in each iterative update, the following steps are performed: calculating the sub-anode current in the current iterative update based on the real-time data of the sub-anode current in the previous iterative update and the newly added dataset in the previous iterative update; and calculating the anode resistance and operating status in the current iterative update based on the anode current in the current iterative update.

[0056] In one specific embodiment of this application, the above-mentioned dynamic simulation system for anode current in acid-process alumina electrolysis includes: a precision configuration module, configured to: set the data precision of the first data used for dynamic simulation of anode current in acid-process alumina electrolysis and the data precision of the second data used for dynamic simulation of anode current in acid-process alumina electrolysis to a first precision and a second precision, respectively; based on HOG (Histogram of Oriented Gradients) features and SVM classification, through a learning phase and a detection phase, to achieve target identification and complete the acquisition of information on independent subdomains and the entire cell; and a mixed precision simulation module, configured to: perform dynamic simulation iteration of anode current in acid-process alumina electrolysis based on the first data with the first precision and the second data with the second precision.

[0057] In one specific embodiment of this application, the above-mentioned dynamic simulation system for anode current in acid-process alumina electrolysis includes the following steps in the learning phase: first, collecting sample data such as current, anode rod temperature, and fire-eye data; second, extracting the feature information of the samples into the feature vector space, representing the feature information of the image using a vector model, and obtaining the feature vector; third, inputting the feature vector into a classifier for training and learning, generating a target classifier; the detection phase includes the following steps: first, extracting features from the fire-eye image based on target detection; second, inputting the obtained target feature vector into the trained target classifier, classifying the target by scanning through the detection window, and marking it with a rectangular box; third, merging the rectangular boxes of the output results, merging overlapping small rectangles into a large rectangle to achieve target recognition; wherein, HOG (Histogram of Oriented Gradients) features are those where the detected local object contour can be characterized by the distribution of light intensity gradient or edge direction.

[0058] Fire-Eye imaging is an image acquisition method based on thermal infrared technology that can provide thermal radiation information of a target. In a Fire-Eye image, each pixel represents the temperature distribution of the target's surface. By processing and analyzing the image, the target's feature information can be extracted.

[0059] Feature extraction refers to extracting representative information from an image to describe the shape, texture, color, and other features of a target. Feature extraction in Huoyan images helps in target classification, recognition, and detection. Below are some commonly used Huoyan image feature extraction methods: Gray-level histogram: By statistically analyzing the distribution of pixel gray levels in an image, the overall brightness features of the image can be extracted. Gray-level histograms can be used to describe the brightness distribution of an image, thus characterizing the target. Texture features: Texture information in Huoyan images can be extracted using different texture descriptors. Commonly used methods include Gray-Level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP), and Gabor filters. These methods can capture the texture details of the target surface for texture analysis and recognition.

[0060] The current distribution signal of the anode is measured. The current in the acid-process alumina electrolytic cell is generally introduced by the column busbars connected to the cathode busbar of the previous cell, and enters the anode guide rods through the contact connection between the crossbeam busbar and the anode guide rods. After the current is introduced by the column busbar or the bridge busbar, it flows to the anode carbon blocks through the crossbeam. The current balance of each anode and the overall current can be expressed by the following formula:

[0061] The formulas are as follows: Formula 1: It = I3 + I4 = Ia + Ib + Ic + Id + I1 + I6;

[0062] Second formula: Ia = I2 - I1;

[0063] Third formula: Ib = I3 - I2;

[0064] Fourth formula: Ic = I4 - I5;

[0065] Fifth formula: Id = I5 - I6;

[0066] Where It is the total current entering the crossbeam section from the column or bridge busbar; I1, I2, I3, I4, I5, and I6 are the local cross-sectional currents of each crossbeam section; and Ia, Ib, Ic, and Id are the inflow currents of each anode conductor.

[0067] For the current flowing through the conductor rod, besides directly calculating it using Ohm's law by measuring the equidistant voltage drop across the rod, it can also be solved by summing the currents flowing through the two crossbeams that are in contact with the conductor rod. Therefore, based on the premise that the current on the crossbeams can be measured, for each anode conductor rod, the problem can be simplified to a problem of current inflow and outflow nodes, with the contact between the conductor rod and the crossbeams being the nodes.

[0068] Assuming the current flowing through the beam and guide rod can be considered as an ideal line current, then the current flowing through a certain section of the beam can be calculated using the potential difference along a specific length of the beam. The current I flowing through the guide rod can be calculated using the sixth formula:

[0069] The sixth formula is:

[0070] Where I is the sum of I1 and I2, R1 is the resistance value of the potential difference in the V1-V2 segment, and R2 is the resistance value of the potential difference in the V4-V3 segment. Therefore, if a certain segment of the potential difference V1-V2 and V4-V3 on the crossbeams on both sides of the guide rod can be measured, a physical quantity with the same meaning as the equidistant voltage drop can be obtained.

[0071] Based on the measured anode distribution current and combined with production data, the workflow of the dynamic simulation method for anode current in acid-process alumina electrolysis comprises a series of continuous iterations. Each iteration consists of three steps: exploration, labeling, and training. A crucial aspect of the entire training iteration process is the use of a pre-trained set of models based on data and deep learning. These models are trained from the same set of data, but differ in the initialization of their parameters. Iterations during training aim to minimize the loss function, which is a measure of the error between the measured and predicted anode current, voltage drop, superheat, or alumina concentration. With sufficient training data and appropriate configuration parameters, the prediction models under different production conditions should be highly accurate, thus producing predictions that are close to each other. Otherwise, predictions from different models will be dispersed with considerable variance. Therefore, the variance of the prediction set of a specific configuration's model set can be used as an error indicator and a criterion for selecting a configuration for labeling.

[0072] Because of the sheer volume of tasks delivered by the scheduler, manually managing script submissions and result collection would be both tedious and time-consuming. Using an intelligent task scheduler, however, is much more efficient. Typically, the scheduler automatically uploads input files, submits scripts for computation, maintains the job queue, and collects results upon job completion. Schedulers not only excel at handling various types of computational tasks but also adapt to different types of available computing resources.

[0073] A typical workflow of a scheduler consists of the following functions:

[0074] Checking for submissions: First, the scheduler will check if the task has been submitted; if not, the scheduler will upload the file and build the submission script from scratch; this will form a queue of tasks to be executed; otherwise, the scheduler will restore the existing queue and collect results from completed tasks, instead of submitting the script again.

[0075] File Upload and Download: When the predictor's main process runs on the HPC login node, it shares the file system with the compute nodes via the network file system; file upload and download can be easily implemented using the operating system and shutil (i.e., the regular shutil in Python modules); to improve I / O efficiency, the scheduler operates on symbolic links or moves files directly instead of copying or deleting them; when the predictor's main process runs on a machine other than the one performing the actual computation, files are transferred via SSH (Secure Shell protocol); the predictor provides a unified interface for both scenarios, so file transfer adapts to connection types and can be easily invoked; furthermore, new protocols for file transfer are easily implemented; Job Submission: After uploading the forward file and general files to the computer, the predictor generates a script for performing the required computational tasks; the job script is applicable to different types of computers, including workstations, etc. High-performance clusters and cloud computers with job scheduling systems: Workstations; this allows users to run predictors on a single computer (e.g., a personal laptop); in this case, the predictor prepares shell scripts that can be executed directly; HPCs (High-Performance Computing Clusters) with job scheduling systems, such as Slurm, PBS, and LSF (Distributed Cluster Management System Software); the resources for executing jobs need to be set in the submitted script, such as the number of CPUs / GPUs and the maximum execution idle time; the predictor provides an end-to-end interface to the user and translates the settings in the MACHINE file into a submitted script that the job scheduling system can accept; Cloud machines; the scripts for cloud computers are similar to those for workstations; the only difference is that the predictor scheduler runs additional commands, either before or after job execution, to start or terminate the computer; these commands are compatible with the application programming interface (API) provided by the cloud computing service;

[0076] Job monitoring: Monitoring the status of each submitted job is crucial; the predictive scheduler can identify job status and react accordingly by communicating with the various machines described above; if a job is running, the scheduler will not take any action; if a job terminates, the scheduler will attempt to resubmit the job up to three times; if the job still fails to execute successfully, it is likely that there is a problem with the job settings, for example, the parameters may be incorrect; in this case, user intervention is required; after a task is completed, the scheduler will download the back file and pass it to the scheduler; the task scheduler completes its task after all tasks are completed, and then the planner will move forward;

[0077] Finally, it should be emphasized that the prediction program is quite automated and industry-standard; once the input file is prepared and the prediction program is successfully run, no additional manual intervention is required. The dynamic simulation method for anode current in acid-process alumina electrolysis of this invention is based on the above description and assumptions, which combines the method of measuring beam voltage drop and indirectly calculating anode current with industry data, expert experience, and deep learning. Based on the measured data, it continuously learns and iterates to control the frequency and amount of alumina feeding, effectively reducing the anode effect by 40% to 60% and improving current efficiency by 3% to 5%.

[0078] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0079] This application also provides an apparatus for processing the anolyte current during alumina electrolysis. It should be noted that this apparatus can be used to execute the method for processing the anolyte current during alumina electrolysis provided in this application. This apparatus is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0080] The following describes the apparatus for processing the anolyte current during alumina electrolysis provided in the embodiments of this application.

[0081] Figure 3 This is a structural block diagram of an apparatus for processing the anolyte current during alumina electrolysis, according to an embodiment of this application. Figure 3As shown, the device includes a first acquisition unit 31, a first processing unit 32, a second acquisition unit 33, and a second processing unit 34. The first acquisition unit 31 is used to acquire multiple historical sub-anode currents of each sub-anode, wherein the anode of the alumina is composed of multiple such sub-anodes. The first processing unit 32 is used to input all the aforementioned historical sub-anode currents into a neural network model to process the aforementioned historical sub-anode currents using the neural network model. The neural network model is trained using multiple sets of training data. Each set of training data includes: a first historical sub-anode current and a corresponding second historical sub-anode current acquired within a historical time period. The first historical sub-anode current is the current of the sub-anode at the first historical moment, and the second historical sub-anode current is... The anode current is the current of the sub-anode at the second historical moment, and the first historical moment is before the second historical moment; the second acquisition unit 33 is used to acquire the actual sub-anode current and the predicted anode current output by the neural network model, and to perform regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, so as to optimize the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment; the second processing unit 34 is used to input the actual sub-anode current into the optimized neural network model, and to adjust the amount and frequency of alumina feeding according to the output of the optimized neural network model.

[0082] In the aforementioned device, the sub-anode currents of each sub-anode of the alumina anode are used as inputs and iterated continuously to improve the accuracy of the neural network model, thereby improving the accuracy of subsequent anode current predictions and achieving the goal of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feeding are adjusted accordingly, thus solving the problem that the accuracy of alumina feeding amount and frequency control is low due to the low accuracy of anode current prediction during the electrolysis process in existing solutions.

[0083] In one embodiment of this application, the second acquisition unit includes a first determining module and a first processing module. The first determining module is used to determine whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current. The first processing module is used to perform regression processing on the neural network model if it is determined that regression processing should be performed on the neural network model, so as to obtain the optimized neural network model.

[0084] In one embodiment of this application, the first determining module includes an acquisition submodule, a first determining submodule, and a second determining submodule. The acquisition submodule is used to acquire the anode current difference, which is the difference between the actual sub-anode current and the corresponding predicted anode current. The first determining submodule is used to determine whether to perform regression processing on the neural network model when the absolute value of the anode current difference is greater than or equal to a first difference threshold and less than or equal to a second difference threshold. The second determining submodule is used to determine whether to perform regression processing on the neural network model when the absolute value of the anode current difference is less than the first difference threshold or greater than the second difference threshold.

[0085] In one embodiment of this application, the device further includes a third processing unit and a first determining unit. After performing regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, the third processing unit is used to obtain the anode resistance of the sub-anode corresponding to the actual sub-anode current based on the actual sub-anode current corresponding to the predicted anode current. The first determining unit is used to determine the operating state of the corresponding sub-anode based on the anode resistance. The operating state is one of the following: overheating state, undercooling state, and normal state.

[0086] In one embodiment of this application, the third determining unit includes a first acquiring module, a second determining module, a third determining module, and a fourth determining module. The first acquiring module is used to acquire a first resistance threshold and a second resistance threshold. The second determining module is used to determine that the operating state of the corresponding sub-anode is the normal state when the anode resistance is greater than or equal to the first resistance threshold and the anode resistance is less than or equal to the second resistance threshold. The third determining module is used to determine that the operating state of the corresponding sub-anode is the subcooled state when the anode resistance is less than the first resistance threshold. The fourth determining module is used to determine that the operating state of the corresponding sub-anode is the overheated state when the anode resistance is greater than the second resistance threshold.

[0087] In one embodiment of this application, the device further includes a third acquisition unit, a second determination unit, and a fourth processing unit. Before inputting all the aforementioned historical sub-anode currents into the neural network model, the third acquisition unit is used to acquire first data and second data. The first data is the actual sub-anode current, and the second data is the estimated anode current, which is an estimate of the current of the sub-anode at a future time. The second determination unit is used to determine a first precision of the first data and a second precision of the second data, both of which are preset precisions. The fourth processing unit is used to input the first precision and the second precision into the neural network model to limit the output of the neural network model.

[0088] In one embodiment of this application, the device further includes a fifth processing unit, a sixth processing unit, and a seventh processing unit. Before inputting all the aforementioned historical sub-anode currents into the neural network model, the fifth processing unit is used to extract multiple target features from the database using directional gradient histogram technology. The target features are used to characterize other features related to the sub-anode besides the anode current of the sub-anode. The database stores multiple target features and the anode current of the sub-anode. The sixth processing unit is used to classify the target features using a support vector machine algorithm to obtain multiple classification features. The seventh processing unit is used to input all classification features into the neural network model to limit the output of the neural network model.

[0089] In one embodiment of this application, the second processing unit includes a second acquisition module and a fifth processing module. The second acquisition module is used to acquire the current mapping relationship output by the optimized neural network model, wherein the current mapping relationship is the relationship between the predicted anode current and time within a predetermined time period in the future. The fifth processing module is used to adjust the feeding amount and feeding frequency of the alumina according to the current mapping relationship.

[0090] The aforementioned device for processing the anolyte current during alumina electrolysis includes a processor and a memory. The first acquisition unit, first processing unit, second acquisition unit, and second processing unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve their respective functions. All of the above modules are located in the same processor; alternatively, the modules may be located in different processors in any combination.

[0091] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured. By adjusting the kernel parameters, the problem of low accuracy in controlling the amount and frequency of alumina feed caused by the low accuracy of predicting the anode current during the electrolysis process in existing solutions can be addressed.

[0092] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0093] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the anode current processing method during alumina electrolysis.

[0094] This invention provides a processor for running a program, wherein the program executes the method for processing the anolyte current during alumina electrolysis.

[0095] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps: acquiring multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple such sub-anodes; inputting all the aforementioned historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model, wherein the neural network model is trained using multiple sets of training data, and each set of training data includes: a first historical sub-anode current and a corresponding second historical sub-anode current acquired within a historical time period, wherein the first historical sub-anode current is the current of the sub-anode at the first historical moment. The current, where the second historical sub-anode current is the current of the sub-anode at the second historical moment, and the first historical moment precedes the second historical moment; the actual sub-anode current and the predicted anode current output by the neural network model are obtained, and regression processing is performed on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, thereby optimizing the output of the neural network model; the actual sub-anode current is the actual value of the current of the sub-anode at the current moment; the actual sub-anode current is input into the optimized neural network model, and the feeding amount and frequency of alumina are adjusted accordingly based on the output of the optimized neural network model. The devices mentioned in this article can be servers, PCs, PADs, mobile phones, etc.

[0096] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps: acquiring multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple such sub-anodes; inputting all such historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model, wherein the neural network model is trained using multiple sets of training data, each set of training data including: a first historical sub-anode current and a corresponding second historical sub-anode current acquired within a historical time period, wherein the first historical sub-anode current is the current of the sub-anode at a first historical moment, and the second historical sub-anode current is... The historical sub-anode current is the current of the sub-anode at the second historical moment, where the first historical moment precedes the second historical moment. The actual sub-anode current and the predicted anode current output by the neural network model are obtained. Based on the actual and predicted anode currents, regression processing is performed on the neural network model to obtain an optimized model, thereby optimizing its output. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment. The actual sub-anode current is input into the optimized neural network model, and the alumina feeding rate and frequency are adjusted accordingly based on the output of the optimized neural network model.

[0097] This application also provides an electronic device, characterized in that it includes: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include a method for processing the anode current during alumina electrolysis as described above. By using the sub-anode currents of each sub-anode of the alumina anode as input and iterating continuously, the accuracy of the neural network model is improved, thereby improving the accuracy of subsequent anode current prediction, thus achieving the purpose of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feed are adjusted accordingly, thereby solving the problem of low accuracy in controlling the amount and frequency of alumina feed due to the low accuracy of anode current prediction during alumina electrolysis in existing solutions.

[0098] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0099] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0104] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0105] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0106] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0107] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0108] 1) The anode current processing method during alumina electrolysis in this application uses the sub-anode current of each sub-anode of the alumina anode as input and iterates continuously to improve the accuracy of the neural network model, thereby improving the accuracy of subsequent anode current prediction and achieving the purpose of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feeding are adjusted accordingly, thus solving the problem that the accuracy of alumina feeding amount and frequency control is low due to the low accuracy of anode current prediction during alumina electrolysis in existing solutions.

[0109] 2) The anode current processing device for alumina electrolysis in this application uses the sub-anode current of each sub-anode of the alumina anode as input and iterates continuously to improve the accuracy of the neural network model, thereby improving the accuracy of subsequent anode current prediction and achieving the purpose of model optimization. Based on the output of the optimized model, the amount and frequency of alumina feeding are adjusted accordingly, thus solving the problem that the accuracy of alumina feeding amount and frequency control is low due to the low accuracy of anode current prediction in the electrolysis process of existing solutions.

[0110] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for processing the anolyte current during alumina electrolysis, characterized in that, include: Obtain multiple historical sub-anode currents for each sub-anode, wherein the anode of the alumina is composed of multiple said sub-anodes; All the historical sub-anode currents are input into a neural network model to process the historical sub-anode currents. The neural network model is trained using multiple sets of training data. Each set of training data includes: a first historical sub-anode current and a corresponding second historical sub-anode current obtained within a historical time period. The first historical sub-anode current is the current of the sub-anode at the first historical moment, and the second historical sub-anode current is the current of the sub-anode at the second historical moment. The first historical moment is before the second historical moment. The actual sub-anode current and the predicted anode current output by the neural network model are obtained. Based on the actual sub-anode current and the predicted anode current, regression processing is performed on the neural network model to obtain an optimized neural network model, thereby optimizing the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment. The actual sub-anode current is input into the optimized neural network model, and the amount and frequency of alumina feeding are adjusted according to the output of the optimized neural network model. The neural network model is subjected to regression processing based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model. This includes: determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current; and performing regression processing on the neural network model if it is determined that regression processing should be performed to obtain the optimized neural network model. Determining whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current includes: obtaining an anode current difference, where the anode current difference is the difference between the actual sub-anode current and the corresponding predicted anode current; determining to perform regression processing on the neural network model if the absolute value of the anode current difference is greater than or equal to a first difference threshold and the absolute value of the anode current difference is less than or equal to a second difference threshold; and determining not to perform regression processing on the neural network model if the absolute value of the anode current difference is less than the first difference threshold or greater than the second difference threshold.

2. The method according to claim 1, characterized in that, After performing regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, the method further includes: Based on the actual sub-anode current corresponding to the predicted anode current, the anode resistance of the sub-anode corresponding to the actual sub-anode current is obtained; Based on the anode resistance, the operating state of the corresponding sub-anode is determined, and the operating state is one of the following: overheating state, undercooling state, or normal state.

3. The method according to claim 2, characterized in that, Determining the operating state of the corresponding sub-anode based on the anode resistance includes: Obtain the first resistance threshold and the second resistance threshold; When the anode resistance is greater than or equal to the first resistance threshold and the anode resistance is less than or equal to the second resistance threshold, the working state of the corresponding sub-anode is determined to be the normal state. If the anode resistance is less than the first resistance threshold, the operating state of the corresponding sub-anode is determined to be the supercooled state. If the anode resistance is greater than the second resistance threshold, the operating state of the corresponding sub-anode is determined to be the overheating state.

4. The method according to claim 1, characterized in that, Before inputting all the historical sub-anode currents into the neural network model, the method further includes: Acquire first data and second data, wherein the first data is the actual sub-anode current and the second data is the estimated anode current, wherein the estimated anode current is an estimate of the current of the sub-anode at a future time; Determine a first precision of the first data and a second precision of the second data, wherein both the first precision and the second precision are preset precisions; The first precision and the second precision are input into the neural network model to limit the output of the neural network model.

5. The method according to claim 1, characterized in that, Before inputting all the historical sub-anode currents into the neural network model, the method further includes: Multiple target features are extracted from the database using directional gradient histogram technology. These target features are used to characterize other features related to the sub-anode besides the anode current of the sub-anode. The database stores multiple target features and the anode current of the sub-anode. The target features are classified using the support vector machine algorithm to obtain multiple classification features; All classification features are input into the neural network model to limit its output.

6. The method according to any one of claims 1 to 5, characterized in that, Based on the output of the optimized neural network model, the feeding amount and frequency of alumina are adjusted accordingly, including: Obtain the current mapping relationship output by the optimized neural network model, wherein the current mapping relationship is the relationship between the predicted anode current and time within a predetermined future time period; Based on the current mapping relationship, the amount and frequency of alumina feeding are adjusted accordingly.

7. A device for processing the anolyte current during alumina electrolysis, characterized in that, include: The first acquisition unit is used to acquire multiple historical sub-anode currents of each sub-anode, wherein the anode of the alumina is composed of multiple said sub-anodes; The first processing unit is used to input all the historical sub-anode currents into a neural network model to process the historical sub-anode currents using the neural network model. The neural network model is trained using multiple sets of training data. Each set of training data includes: a first historical sub-anode current and a corresponding second historical sub-anode current obtained within a historical time period. The first historical sub-anode current is the current of the sub-anode at a first historical moment, and the second historical sub-anode current is the current of the sub-anode at a second historical moment. The first historical moment is preceding the second historical moment. The second acquisition unit is used to acquire the actual sub-anode current of the sub-anode and the predicted anode current output by the neural network model, and to perform regression processing on the neural network model based on the actual sub-anode current and the predicted anode current to obtain an optimized neural network model, so as to optimize the output of the neural network model. The actual sub-anode current is the actual value of the current of the sub-anode at the current moment. The second processing unit is used to input the actual sub-anode current into the optimized neural network model, and adjust the amount and frequency of alumina feeding according to the output of the optimized neural network model. The second acquisition unit includes a first determining module and a first processing module. The first determining module is used to determine whether to perform regression processing on the neural network model based on the actual sub-anode current and the corresponding predicted anode current. The first processing module is used to perform regression processing on the neural network model if it is determined that regression processing should be performed on the neural network model, so as to obtain the optimized neural network model. The first determining module includes an acquisition submodule, a first determining submodule, and a second determining submodule. The acquisition submodule is used to acquire the anode current difference, which is the difference between the actual sub-anode current and the corresponding predicted anode current. The first determining submodule is used to determine whether to perform regression processing on the neural network model when the absolute value of the anode current difference is greater than or equal to a first difference threshold and the absolute value of the anode current difference is less than or equal to a second difference threshold. The second determining submodule is used to determine whether to perform regression processing on the neural network model when the absolute value of the anode current difference is less than the first difference threshold or greater than the second difference threshold.

8. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including a method for processing the anolyte current during alumina electrolysis as described in any one of claims 1 to 6.

Images