Electrolysis system and method of charge control

Abstract

This application provides an electrolysis system and a feeding control method. The electrolysis system includes: a homogeneous group, comprising multiple homogeneous groups, each homogeneous group including multiple electrolytic cells whose similarity between implemented electrolysis processes meets a predetermined process threshold, wherein the feeding interval between multiple homogeneous groups is less than a start-up waiting window period, the start-up waiting window period being determined based on the historical feeding time sequence of each homogeneous group; a controller configured to: in response to receiving an instruction to feed a first feeding request to an electrolytic cell in a target homogeneous group, time itself, and generate a feeding command in response to reaching a start-up trigger condition, the start-up trigger condition including: receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail feeding request and the first feeding request is less than the start-up waiting window period, or not receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, but the start-up waiting window period has been reached; and a feeding component configured to feed the electrolytic cells of each homogeneous group in the homogeneous group in response to the feeding command.

Description

Technical Field

[0001] This application relates to the field of aluminum electrolysis production technology, and more specifically, to an electrolysis system and a feeding control method. Background Technology

[0002] With the rapid development of automation and intelligent control technologies in the non-ferrous metal smelting industry, the intelligent upgrading of aluminum electrolysis production has become a development direction for the industry to improve quality and efficiency and stabilize production processes. Among them, precise control of the feeding process of aluminum electrolysis cells is an important link to ensure the material balance of electrolysis cells and the stability of process parameters.

[0003] However, the current feeding process in aluminum electrolysis cells still largely relies on manual operation. Operators must determine the fullness of the material by tapping the hoppers to detect the echo, which inherently suffers from high labor intensity, low inspection efficiency, and accuracy heavily influenced by human factors. While it's possible to equip each hopper with a level gauge for automated feeding, the high procurement and maintenance costs of this hardware result in poor economic efficiency and hinder large-scale implementation. Furthermore, timed feeding or simple low-level trigger feeding schemes cannot match the actual material consumption differences between different hoppers, easily leading to uneven feeding, disrupting the material balance of the electrolysis cell, and thus affecting the long-term stability of the electrolysis process. Summary of the Invention

[0004] In view of this, this application provides an electrolysis system and a feeding control method.

[0005] One aspect of this application provides an electrolysis system, comprising:

[0006] Homogeneous groups include multiple homogeneous groups, which include multiple electrolytic cells whose similarity between implemented electrolysis processes meets a predetermined threshold. The feeding interval between multiple homogeneous groups is less than the start-up waiting window period, which is determined based on the historical feeding time series of each homogeneous group in the homogeneous group.

[0007] The controller is configured to: in response to receiving an initial feed request instructing the addition of material to an electrolyzer in a target homogeneous group, time itself, and in response to the achievement of a start-up trigger condition, generate a feed command, wherein the start-up trigger condition includes: receiving a tail feed request from the last homogeneous group in the homogeneous group, and the time interval between the tail feed request and the initial feed request is less than a start-up waiting window period; or not receiving a tail feed request from the last homogeneous group in the homogeneous group, but the start-up waiting window period has been reached; and

[0008] The feeding component is configured to feed the electrolyzers in each homogeneous group in response to a feeding command.

[0009] According to embodiments of this application, the electrolysis system further includes:

[0010] At least one hopper, configured to hold material added to the electrolytic cell via a feeding assembly, and to feed the electrolytic cell at a predetermined frequency; and

[0011] The material level gauge is installed in the hopper of the target electrolytic cell of the same group. It is configured to trigger the generation of a feeding request and send the feeding request to the controller when the content of the material in the hopper is detected to be lower than a first predetermined material threshold.

[0012] According to an embodiment of this application, the controller is further configured to: in response to receiving a first stop request indicating to stop feeding, time the process, and in response to reaching a shutdown trigger condition, generate a stop command so that the feeding component stops feeding the electrolytic cells in each homogeneous group in response to the stop command, wherein the shutdown trigger condition includes: receiving a tail stop request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail stop request and the first stop request is less than the shutdown waiting window period, or not receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, but the shutdown waiting window period has been reached.

[0013] According to an embodiment of this application, the electrolysis system further includes: a material level gauge, which is installed in the material bin of the target electrolytic cell of the same group, and is configured to trigger the generation of a stop request and send the stop request to the controller when it is detected that the content of the material in the material bin is higher than a second predetermined threshold.

[0014] According to an embodiment of this application, the electrolysis system further includes: a processor configured to:

[0015] Based on the electrolysis process parameters of each electrolytic cell, the electrolysis process characteristics of each electrolytic cell are determined.

[0016] Based on the electrolysis process characteristics of each electrolytic cell, multiple electrolytic cells are clustered to identify multiple electrolytic cells belonging to the same homogeneous group.

[0017] And based on the number of electrolytic cells in the homogeneous group, determine the target number of target electrolytic cells from the homogeneous group for sending feeding requests to the controller.

[0018] According to embodiments of this application, the processor is further configured to:

[0019] For each homogeneous group within a homogeneous group, the historical feeding sequence information of the homogeneous group is input into the autoregressive model, and the predicted feeding time series information for future periods is output.

[0020] Based on the predicted feeding time series information of multiple homogeneous groups, the initial start-up waiting window period is determined;

[0021] Furthermore, based on the startup drift trend term, the initial startup waiting window period is corrected to determine the startup waiting window period, wherein the startup drift trend term is used to characterize the difference between the predicted initial startup waiting window period and the actual startup waiting window period.

[0022] According to an embodiment of this application, the processor is further configured to: determine start-up deviation time series information based on real historical feeding time series information and predicted historical feeding time series information for a target historical period, wherein the predicted historical feeding time series information is determined by inputting real historical feeding time series information for historical periods prior to the target historical period into an autoregressive model; and determine a start-up drift trend term based on the start-up deviation time series information.

[0023] According to an embodiment of this application, the processor is further configured to: input historical shutdown time series information of each homogeneous group into an autoregressive model for each homogeneous group, and output predicted shutdown time series information for future periods; determine an initial shutdown waiting window period based on the predicted shutdown time series information of each of the multiple homogeneous groups; and correct the initial shutdown waiting window period based on the shutdown drift trend term to determine the shutdown waiting window period, wherein the shutdown drift trend term is used to characterize the difference between the predicted initial shutdown waiting window period and the actual shutdown waiting window period.

[0024] According to an embodiment of this application, the feeding assembly includes: a pipeline valve disposed on a pipeline for connecting to the electrolytic cell and the material bin, configured to open the pipeline in response to a feeding command; and a blower connected to the pipelines configured in each electrolytic cell, configured to start operation in response to the opening of the pipeline, so that the material in the material bin is respectively transported to the electrolytic cells in each homogeneous group through the pipeline.

[0025] Another aspect of this application provides a feeding control method applied to an electrolysis system, the electrolysis system including a homogeneous group, a controller, and a feeding component. The method includes: the controller responding to receiving a first feeding request instructing the feeding of a cell in a target homogeneous group within the homogeneous group, timing the process, wherein the similarity between the electrolysis processes implemented by each cell in the target homogeneous group satisfies a predetermined process threshold, the homogeneous group including multiple homogeneous groups, the feeding time interval between the multiple homogeneous groups being less than a start-up waiting window period, the start-up waiting window period being determined based on the historical feeding time sequence of each homogeneous group in the homogeneous group; the controller generating a feeding instruction responding to the achievement of a start-up triggering condition, wherein the start-up triggering condition includes: receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail feeding request and the first feeding request being less than the start-up waiting window period, or not receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, but the start-up waiting window period has been reached; and the feeding component feeding the cells in each homogeneous group within the homogeneous group in response to the feeding instruction sent by the controller. Attached Figure Description

[0026] The above and other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0027] Figure 1 A schematic diagram of an electrolysis system according to an embodiment of this application is shown.

[0028] Figure 2 A schematic diagram of an electrolysis system according to another embodiment of this application is shown;

[0029] Figure 3 A schematic diagram of an electrolysis system according to another embodiment of this application is shown;

[0030] Figure 4 A flowchart illustrating the feeding control method provided in an embodiment of this application is shown. Detailed Implementation

[0031] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0032] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0033] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0034] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0035] Figure 1 A schematic diagram of an electrolysis system according to an embodiment of this application is shown.

[0036] Embodiments of this application provide an electrolysis system, such as Figure 1 As shown, it includes: homogeneous groups, including multiple homogeneous groups, each homogeneous group comprising multiple electrolytic cells whose similarity between implemented electrolysis processes meets a predetermined threshold for the process, the feeding interval between multiple homogeneous groups being less than the start-up waiting window period, and the start-up waiting window period being determined based on the historical feeding time series of each homogeneous group in the homogeneous group.

[0037] The controller is configured to: in response to receiving an initial feed request instructing the feeding of an electrolyzer in a target homogeneous group, time itself, and in response to the achievement of a start-up triggering condition, generate a feed instruction, wherein the start-up triggering condition includes: receiving a tail feed request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail feed request and the initial feed request is less than the start-up waiting window period, or not receiving a tail feed request sent by the last homogeneous group in the homogeneous group, but the start-up waiting window period has been reached.

[0038] The feeding component is configured to feed the electrolyzers in each homogeneous group in response to a feeding command.

[0039] In some specific embodiments, the electrolysis process may include aluminum electrolysis, but is not limited to this; it may also include wet electrolysis processes for non-ferrous metals such as copper electrolysis, zinc electrolysis, nickel electrolysis, and lead electrolysis. The materials can be adaptively adjusted according to the type of electrolysis process; for example, the materials may be alumina, alumina, copper sulfate, zinc oxide, nickel sulfate, etc.

[0040] The process parameters of the electrolysis process can include the cell operating voltage, molecular ratio, alumina concentration, and feeding interval of each electrolytic cell.

[0041] The similarity between electrolysis processes meets a predetermined threshold, which can be understood as similar process parameters. For example, this could mean multiple electrolytic cells with operating voltages close to 5V or multiple electrolytic cells with alumina concentrations close to 10%.

[0042] According to the electrolysis system of this application, the homogeneous group includes multiple homogeneous clusters, and each homogeneous cluster contains multiple electrolytic cells. The similarity of the electrolysis processes of these electrolytic cells all meets a preset process threshold, thereby ensuring a high degree of convergence in material consumption and feeding patterns among the electrolytic cells within the same cluster. The system also sets a start-up waiting window T, which is determined based on the historical feeding time series of each homogeneous cluster within the same cluster. Simultaneously, it pre-constrains the feeding interval between multiple homogeneous clusters to be less than this start-up waiting window T, providing a basis for subsequent batch aggregation feeding requests.

[0043] For example, homogeneous groups include homogeneous group 1, homogeneous group 2, and homogeneous group 3. Homogeneous group 1 includes electrolytic cell 1, electrolytic cell 2, electrolytic cell 3, and electrolytic cell 4. Homogeneous group 2 includes electrolytic cell 5, electrolytic cell 6, and electrolytic cell 7. Homogeneous group 3 includes electrolytic cell 8 and electrolytic cell 9. The waiting window period is determined by the historical feeding time sequence of each homogeneous group within the homogeneous group, for example, it can be 10 minutes. The feeding interval between multiple homogeneous groups is less than this start-up waiting window period T, for example, it can be 4 minutes. However, it is not limited to this. It can be adjusted according to the actual situation.

[0044] When the controller receives the first feeding request to add material to the electrolytic cells within the target homogeneous group, it starts timing. During the timing process, it continuously receives feeding requests from other homogeneous groups within the same group and determines in real time whether the triggering conditions are met. Once the triggering conditions are met, a feeding command is generated. There are two types of triggering conditions: one is that a feeding request from the last homogeneous group is received within the timing period, and the time interval between the last feeding request and the first feeding request is less than the start-up waiting window; the other is that not all feeding requests from homogeneous groups have been received, but the timing period has reached the start-up waiting window. After receiving the feeding command from the controller, the system's feeding component can perform feeding operations on the electrolytic cells of each homogeneous group within the same group.

[0045] For example, one implementation could be as follows: In the first minute, homogeneous group 1 sends a first feeding request to the controller. Upon receiving the first feeding request, the controller immediately starts timing, with the current start-up waiting window T = 10 minutes. In the third minute, the controller receives a feeding request from homogeneous group 2. In the fifth minute, the controller receives a feeding request from homogeneous group 3, which is the last feeding request from the last homogeneous group that has not yet sent a request. The time interval between the last feeding request and the first feeding request is 4 minutes, which is less than the preset start-up waiting window of 10 minutes, thus satisfying the first start-up triggering condition. The controller stops timing and generates a batch feeding command. After receiving the feeding command, the feeding component synchronously completes the feeding operation for all electrolytic cells within the homogeneous group.

[0046] For example, another implementation could be: In the first minute, homogeneous group 1 sends a first feeding request to the controller. Upon receiving the request, the controller immediately starts timing, with the current start-up waiting window T = 10 minutes. In the fifth minute, the controller receives a valid feeding request from a typical electrolytic cell within homogeneous group 2. Throughout the timing period, homogeneous group 3 does not trigger a material empty signal within the start-up waiting window T, and the controller never receives a tail feeding request from the last homogeneous group (homogeneous group 3) within the homogeneous group. The controller's timing duration has reached the 10-minute waiting window, and it has not received feeding requests from all homogeneous groups within the homogeneous group, thus satisfying the second start-up triggering condition. The controller stops timing and generates a batch feeding command. Upon receiving the feeding command, the feeding component synchronously completes the feeding operation for all electrolytic cells within the homogeneous group.

[0047] According to the embodiments of this application, the electrolysis system utilizes the design of homogeneous groups to accurately infer the feeding pattern of the entire homogeneous group through the historical feeding time sequence of each homogeneous group, and determine the start-up waiting window period; in addition, by aggregating feeding requests and batch centralized feeding during the start-up waiting window period, the frequent start-up and shutdown of the feeding equipment is reduced, effectively reducing system energy consumption.

[0048] According to an embodiment of this application, the feeding assembly may include: a pipeline valve disposed on a pipeline for connecting to the electrolytic cell and the material bin, configured to open the pipeline in response to a feeding command; and a blower connected to the pipelines configured for each electrolytic cell, configured to start operation in response to the opening of the pipeline, so that the material is conveyed through the pipeline to the material bins of the electrolytic cells in each homogeneous group.

[0049] According to embodiments of this application, the feeding component is configured in a coordinated manner with pipeline valves and a blower. The pipeline valves are installed on the pipelines connecting the electrolytic cell and the material bin, and can accurately open the target pipelines corresponding to the same homogeneous group in response to batch feeding commands. This flexibly adapts to the feeding needs of different batches, ensuring accurate delivery of materials to the electrolytic cell and avoiding uneven feeding that could affect the stability of the electrolysis process. The blower is connected to each pipeline and can start synchronously in response to pipeline opening. It determines the feeding and stopping points based on the material bin filling time pattern inferred from the group feeding pattern, operating only during feeding and shutting down after filling, significantly reducing the number of blower start-ups and shutdowns and substantially reducing system energy consumption.

[0050] Figure 2 A schematic diagram of an electrolysis system according to another embodiment of this application is shown.

[0051] like Figure 2 As shown, the electrolysis system may also include a processor.

[0052] According to an embodiment of this application, the processor is configured to: determine the electrolysis process characteristics of each electrolytic cell based on the electrolysis process parameters of each electrolytic cell; cluster multiple electrolytic cells based on the electrolysis process characteristics of each electrolytic cell to determine multiple electrolytic cells belonging to the same homogeneous group; and determine a target number of target electrolytic cells from the homogeneous group for sending a feeding request to the controller based on the number of electrolytic cells in the homogeneous group.

[0053] In some specific embodiments, clustering multiple electrolytic cells can be performed using clustering algorithms, with the specific steps as follows.

[0054] (1) Data acquisition: Obtain the core process parameters (such as cell working voltage, molecular ratio, alumina concentration, feeding interval, etc.) of each electrolytic cell in the target production area within the preset statistical period. For example, the preset statistical period is 1 month. The arithmetic mean of each process parameter in each shift is calculated using a single production shift as the statistical unit to eliminate the interference of instantaneous fluctuations of process parameters on the clustering results. For example, 3 production shifts are set per day, and the monthly statistical period is 30 days. 90 sets of average process parameter data are obtained for each electrolytic cell.

[0055] (2) Vector transformation: The collected process parameters are vectorized. For the i-th electrolytic cell in the target production area (1≤i≤M, where M is the total number of electrolytic cells in the target production area, for example, M=37), the mean of each core process parameter in a single shift is converted into a p-dimensional feature vector, where p is the number of dimensions of the process parameter. In a single shift, the feature vectors of M electrolytic cells together constitute an M×p-dimensional process status feature matrix. A total of 90 sets of M×p-dimensional process status feature matrices are formed in a single monthly statistical period.

[0056] (3) Data standardization: The Z-score standardization method is used to standardize all the transformed feature vectors. After the standardization, the mean of the parameters of each feature dimension is 0 and the standard deviation is 1.

[0057] (4) Clustering: Based on the total number M of electrolytic cells in the target production area, the number of cluster centers is selected in combination with the experience value of the production site. For example, for a target production area with 37 electrolytic cells, 6 cluster centers are selected. The Euclidean distance is used as the measure of process similarity. The Euclidean distance between the feature vectors of each electrolytic cell after standardization is calculated. The Euclidean distance is negatively correlated with the process similarity. That is, the smaller the Euclidean distance between the feature vectors, the higher the electrolysis process similarity between the two electrolytic cells. Finally, the clustering calculation is completed by the clustering algorithm to form multiple clusters corresponding to the number of cluster centers. Each cluster is a homogeneous group. The Euclidean distance between the feature vector of any electrolytic cell in the same homogeneous group and the corresponding cluster center of the group satisfies the preset process similarity threshold.

[0058] In some specific embodiments, multiple electrolytic cells can be clustered, for example, by using time series clustering algorithms. Based on the time series data of empty and full materials confirmed by manual tapping during the historical feeding process of the electrolytic cells, the period, amplitude, and stability characteristics of the feeding time series of each electrolytic cell are quantitatively analyzed, and electrolytic cells with similar feeding behavior patterns are divided into the same homogeneous group. The specific steps are as follows.

[0059] For each electrolytic cell within the target production area, the empty time and full time of each cell were checked and recorded by manually tapping the material bins, forming time-series data of electrolytic cell feeding events. Based on this time-series data, the feeding behavior was quantitatively characterized as follows.

[0060] Periodic characteristics: Calculate the time interval between two consecutive "material empty" signals of a single electrolytic cell, and record it as the material empty feeding cycle T1 (i.e., time interval 1); calculate the time interval between two consecutive "material full" signals, and record it as the material full feeding cycle T2 (i.e., time interval 2); statistically analyze the average value of all T1 and the average value of T2 of all electrolytic cells within a preset historical period (30 days), and use it as the periodic characteristics of the electrolytic cell.

[0061] Amplitude characteristics: For the empty feeding cycle T1 sequence of a single electrolytic cell, the difference between the maximum and minimum values ​​in the sequence is calculated, i.e., the T1 fluctuation range R1, which is used as the amplitude characteristic corresponding to T1; for the full feeding cycle T2 sequence, the difference between the maximum and minimum values ​​in the sequence is calculated, i.e., the T2 fluctuation range R2, which is used as the amplitude characteristic corresponding to T2.

[0062] Stability characteristics: For the T1 and T2 sequences of a single electrolytic cell, the standard deviations σ1 and σ2 of the sequences are calculated respectively. The standard deviation is used as the stability characteristic. The smaller the standard deviation, the smaller the fluctuation of the feeding time interval and the stronger the regularity, that is, the higher the stability of the feeding behavior.

[0063] Electrolytes that exhibit similar periodicity, amplitude, and stability characteristics are classified into the same homogeneous group.

[0064] According to embodiments of this application, the processor configuration provides support for the homogeneous grouping and target electrolytic cell selection of the electrolysis system, serving as a prerequisite for achieving low-cost and precise material supply. It obtains process characteristics by quantifying electrolysis process parameters, and then uses this as a basis for clustering and grouping homogeneous cells. This allows for the precise selection of electrolytic cells with highly similar processes and feeding patterns, ensuring consistent material consumption characteristics within the same group. This lays a data foundation for subsequently inferring the overall feeding demand within the group based on the target electrolytic cells. Simultaneously, determining the number of target electrolytic cells based on the number of electrolytic cells in the homogeneous group and completing the selection process enables rational sampling of target electrolytic cells, replacing subjective judgment in manual grouping. This improves the accuracy and efficiency of homogeneous grouping, providing a reliable basis for subsequent group feeding pattern modeling, material supply request aggregation and batch feeding, and stoppage control.

[0065] According to some specific embodiments, for a target production area containing M electrolytic cells, L target electrolytic cells can be selected, where L is much smaller than M (for example, when M=37, L=6).

[0066] For example, if there are 6 electrolytic cells in a single homogeneous group, 1 to 2 cells can be randomly selected from the group as the target typical cells.

[0067] According to an embodiment of this application, the electrolysis system further includes: at least one hopper configured to hold materials added to the electrolytic cell via a feeding assembly, and to feed the electrolytic cell at a predetermined frequency.

[0068] In the electrolysis system of this application, the material bin can serve as a material buffer unit to hold the material conveyed by the feeding component. It can stably feed the material into the electrolytic cell at a predetermined frequency required by the electrolysis process, effectively smoothing the supply fluctuations during the material conveying process, ensuring a continuous and stable supply of materials, and avoiding abnormal electrolysis process parameters caused by material supply fluctuations.

[0069] In an optional example, multiple feed boxes can be set in each electrolytic cell, and the multiple feed boxes can be set side by side to ensure that the material fed into the electrolytic cell is uniformly distributed within the electrolytic cell.

[0070] For example, 6-10 material bins can be arranged in an array inside the electrolytic cell. The number of material bins can be adjusted according to the actual size of the electrolytic cell and the material bins.

[0071] According to an embodiment of this application, the electrolysis system further includes: a material level gauge, which is installed in the material tank of the target electrolytic cell of the same group, and is configured to trigger the generation of a feeding request and send the feeding request to the controller when it is detected that the content of the material in the material tank is lower than a first predetermined material threshold.

[0072] In a preferred embodiment of this application, the material level gauge in the electrolysis system is only installed in the material bin of the target electrolytic cell of the same group, eliminating the need to configure it for all material bins of all electrolytic cells, thus significantly reducing hardware procurement, installation and maintenance costs from the source; it can monitor the material level in the bin in real time, and automatically triggers the generation of a feeding request when the material content is lower than the first predetermined material threshold, completely replacing the material level inspection mode of manually tapping the bin, eliminating the safety risks of manual operation, and avoiding the errors and lags of manual judgment.

[0073] According to an embodiment of this application, the electrolysis system may further include: a material level gauge, disposed in the hopper of the target electrolytic cell of the same group, configured to trigger the generation of a stop-material request and send the stop-material request to the controller when the content of the material contained in the hopper is detected to be higher than a second predetermined material threshold.

[0074] The full-level gauge and empty-level gauge of the electrolysis system in this application form a two-way control of the material level, further improving the coordinated optimization of system feeding and shutdown. This full-level gauge is only installed in the hopper of a typical target electrolytic cell within the same group. It can monitor the remaining material level in the hopper in real time. When the material content exceeds a second predetermined threshold, it automatically triggers a shutdown request and sends it to the controller. It can accurately terminate the corresponding feeding process, fundamentally avoiding material overflow and waste caused by overfeeding. It also provides the controller with accurate data on the full-level trigger moment and can simultaneously coordinate with the controller to stop feeding in a timely manner, reducing the equipment's ineffective operating time, further reducing system energy consumption, and ensuring the stable operation of the electrolysis process.

[0075] According to embodiments of this application, the method further includes dynamically modeling the material feeding pattern based on signals from the empty material level gauge and the full material level gauge in the target electrolytic cell, wherein the steps are as follows:

[0076] (1) In each tank of each target electrolytic cell, two rotary paddle level gauges, one high and one low, are installed and set as "full" and "empty" trigger points, respectively (the one set at the high point is the full level gauge, and the one set at the low point is the empty level gauge). This converts the level information into a readable binary (full / empty) status signal, where true indicates that the corresponding level gauge is covered by material and is in the triggered state, and false indicates that the corresponding level gauge is not covered by material and is in the untriggered state. Based on the above definition, when the output of the low-level rotary paddle level gauge is false, it indicates that the material level is below the low-level detection position, and is judged as empty; when the output of the high-level rotary paddle level gauge is true, it indicates that the material level has reached or exceeded the high-level detection position, and is judged as full. The system collects "empty" and "full" signals from all material bins within the target electrolytic cell in real time, recording the timing of each signal. Each target electrolytic cell contains six material bins. An "empty" signal is defined as when only one of the low-position rotary paddle level gauges shows a "false" signal, while a "full" signal is defined as when all high-position rotary paddle level gauges show a "true" signal. The system accumulates and records the level signal data, and based on this data, establishes and continuously updates a model of the group's material feeding pattern.

[0077] (2) Calculation of individual feeding cycle: First, record the box number corresponding to the “empty” and “full” signals of each electrolytic cell. Record the time interval between two consecutive “empty” signals of each box as a feeding cycle. Record the time interval between each “empty” and “full” signal. Calculate the average feeding cycle and feeding interval and its variance of the target electrolytic cell within the sliding time window.

[0078] (3) Group feeding modeling: Align and aggregate the time series of "empty" and "full" events of all target electrolytic cells. Using time series analysis, through time correlation analysis and cluster analysis between events, determine whether the electrolytic cells in the area are in a highly synchronous feeding state or whether there are multiple staggered "feeding batches". Give the time pattern of the first empty tank of the target electrolytic cell and the time pattern of all tanks being filled after emptying. Establish an autoregressive model to describe the slow drift trend of the group feeding rhythm over time and establish the feeding pattern of the entire target production area.

[0079] For example, taking a target electrolytic cell containing 6 material bins as an example, an empty event is triggered when only one of the empty level gauges in the 6 material bins is false, and a full event is triggered when all 6 material bins are true. The feeding cycle is calculated based on the time interval between two consecutive empty events, and the feeding interval is calculated based on the time interval from empty to full. The average feeding cycle and variance are calculated within a sliding time window to update the feeding rhythm of a single material bin. Furthermore, the time series of empty and full events of multiple target electrolytic cells in the region are aligned and aggregated. The feeding batches of the group are identified through time correlation analysis and cluster analysis, and an autoregressive model is established to represent the group. The drift trend of the material feeding rhythm; wherein, the first material predetermined threshold can be a threshold for the number of hoppers that trigger the empty material event. For example, the first material predetermined threshold can be 1, 2 or 3 hoppers or any range between them. When the number of hoppers with the empty material level indicator being false is greater than or equal to the first material predetermined threshold, feeding is triggered; the second material predetermined threshold can be a full material determination threshold. For example, the second material predetermined threshold can be all hoppers full, the full material ratio not less than 90% or the number of full hoppers not less than 5 hoppers or any range between them. When the second material predetermined threshold is met, feeding stops, thereby realizing dynamic modeling of the material feeding rhythm and control of the feeding process based on the material level event.

[0080] Figure 3 A schematic diagram of an electrolysis system according to another embodiment of this application is shown.

[0081] like Figure 3 As shown, the electrolysis system also includes a fluorine-loaded silo, air supply valves, and a dense phase transport link. The fluorine-loaded silo centrally stores raw materials (such as fluorine-loaded alumina) for the entire system. The air supply valves are linked with the feed fan, responding to control commands to open and close the air path, providing air pressure power for the dense phase transport. The electrolysis production unit includes multiple (37) electrolytic cells connected in parallel to the main transport pipeline, which can receive materials synchronously. Through clustering algorithms, they are divided into multiple homogeneous groups with similar processes, providing a basis for batch centralized feeding. Each electrolytic cell is equipped with multiple material bins (6), which serve as material buffer containers to quantitatively feed materials to the electrolytic cells. The material bins of the target electrolytic cell (equivalent to a typical electrolytic cell) in each homogeneous group are equipped with high and low resistance rotary level gauges, which trigger material full stop and material empty feed requests, respectively. By setting only a target number of level gauges, the feeding pattern of the entire area can be inferred, significantly reducing hardware and maintenance costs.

[0082] According to an embodiment of this application, the processor is further configured to: input historical feeding sequence information of each homogeneous group into an autoregressive model for each homogeneous group, and output predicted feeding time series information for future periods; determine an initial start-up waiting window based on the predicted feeding time series information of each of the multiple homogeneous groups; and correct the initial start-up waiting window based on the start-up drift trend term to determine the start-up waiting window, wherein the start-up drift trend term is used to characterize the difference between the predicted initial start-up waiting window and the actual start-up waiting window.

[0083] An autoregressive model (AR) can be used to predict future data using historical data. It can employ a large language model network structure, but is not limited to this. It can also use a transformer network structure, or a Long Short-Term Memory (LSTM) network. Any network structure that can combine historical feeding sequence information with temporal relationship information to predict future feeding time series information is acceptable.

[0084] According to embodiments of this application, the processor learns from the historical feeding sequences of homogeneous groups using an autoregressive model, and can output the predicted feeding time sequence for future periods of each homogeneous group. This effectively captures the inherent patterns of the group's feeding rhythm, replacing the traditional experience-based fixed window setting. Based on the initial start-up waiting window determined by the predicted feeding sequences of multiple homogeneous groups, the accuracy of the start-up waiting window is improved, allowing for better matching of the feeding time distribution of each homogeneous group. This ensures that more feeding requests can be aggregated within the window, reducing the number of start-ups and stop-ups of the feeding components. Simultaneously, by dynamically correcting the initial start-up waiting window using the start-up drift trend term, it effectively adapts to the slow drift of the group's feeding rhythm with the operating conditions. This avoids both excessively long windows leading to material shortages in the electrolyzer and affecting process stability, and excessively short windows preventing insufficient aggregation of requests and inadequate energy-saving effects.

[0085] For example, for six homogeneous groups, the real historical feeding time series information of the past 30 days of the homogeneous groups (such as the material shortage trigger time, feeding cycle, and time interval between adjacent feeding events) is input into the autoregressive model. The model outputs the predicted feeding time series information of each homogeneous group in the next 24 hours (such as the predicted material shortage trigger time and the time distribution pattern of feeding events for each homogeneous group).

[0086] Based on the predicted feeding time series of 6 homogeneous groups, the initial start-up waiting window was determined to be 120s.

[0087] To calculate the start-up drift trend term, the actual historical feeding time series of the target historical period over the past 7 days is compared with the predicted historical feeding time series output by the model for the corresponding period to generate start-up deviation time series information. The average value of the deviation series is calculated to obtain the start-up drift trend term as +15s (indicating that the predicted initial start-up waiting window is 15s shorter than the actual initial start-up waiting window corresponding to feeding). Based on this drift trend term, the initial start-up waiting window is corrected, and the final start-up waiting window is determined to be 135s.

[0088] In some comparative examples, the processor did not set up a startup drift trend term correction step, and directly determined the startup waiting window based on the predicted feeding time series information of multiple homogeneous groups.

[0089] For example, the specific steps are as follows: For the six homogeneous groups, extract the real historical feeding time series information of each homogeneous group over the past 30 days, input it into the autoregressive model, and output the predicted feeding time series information of each homogeneous group for the next 24 hours; directly based on the predicted feeding time series, determine a fixed start-up waiting window of 120s, without subsequent drift correction and dynamic adjustment steps.

[0090] The controller performs feeding control based on the 120s start-up waiting window determined in this comparative example. Since no drift trend term is introduced to adaptively and dynamically correct the start-up waiting window, this start-up waiting window can only aggregate some homogeneous feeding requests compared to this application, resulting in an increase in the number of start-ups and shutdowns of the feeding system and a significant decrease in energy-saving effect compared to this application. At the same time, due to fluctuations in the electrolytic cell operating conditions and changes in material consumption characteristics, the feeding rhythm slowly drifts, and the fixed start-up waiting window will continuously deviate from the actual feeding rhythm, resulting in the electrolysis process stability in the comparative example being inferior to that in this application.

[0091] According to an embodiment of this application, the method for calculating the start-up drift trend term includes: determining start-up deviation time series information based on the real historical feeding time series information and the predicted historical feeding time series information for the target historical period, wherein the predicted historical feeding time series information is determined by inputting the real historical feeding time series information of the historical period before the target historical period into an autoregressive model; and determining the start-up drift trend term based on the start-up deviation time series information.

[0092] According to embodiments of this application, the method for calculating the startup drift trend term in the processor obtains the startup deviation time series by comparing the actual feeding time series of the target historical period with the predicted feeding time series output by the model. Based on this, the startup drift trend term is calculated, which can accurately quantify the actual deviation between the autoregressive model's predicted value and the actual feeding rhythm. This effectively captures the slow drift in the feeding rhythm caused by fluctuations in the electrolytic cell's operating conditions and changes in material consumption characteristics, solving the problem that fixed window periods or purely predicted window periods cannot adapt to dynamic changes in operating conditions. The drift trend term obtained by this method corrects the initial startup waiting window period, effectively compensating for model prediction errors. Furthermore, this method can be iteratively updated with the accumulation of historical data, continuously optimizing the model's prediction accuracy, ensuring that the window period always matches the actual feeding pattern, and continuously balancing the system's timely feeding, process stability, and energy-saving effects.

[0093] For example, the calculation method for initiating the drift trend term can be as follows:

[0094] (1) Taking one day as a time period, the complete feeding process from "empty" to "full" of a single typical electrolytic cell is recorded as one valid feeding event; the number of events occurring on a single day is counted, and data is collected continuously for 30 days to form a feeding event sequence.

[0095] An electrolytic cell being empty or full within a day is considered a small event. The number of times this event occurs within a day is recorded as X_1. Data collected over a sufficiently long period (e.g., 30 days) is used to form a sequence.

[0096] (2) Based on the standardized sequence training autoregressive model, the 30-day historical data is used as the model input to predict the number of daily material feeding events in the next 7 days. The actual number of material feeding events during the same period is collected simultaneously to form the predicted value sequence and the actual value sequence respectively.

[0097] (3) Calculate the daily deviation between the predicted value and the actual value, and accumulate them to obtain the cumulative deviation sequence; perform linear fitting on the cumulative deviation sequence, and the slope of the fitted line is the starting drift trend term. The positive and negative values ​​and absolute values ​​of the slope correspond to the drift direction and amplitude, respectively.

[0098] According to an embodiment of this application, the controller in the electrolysis system is further configured to: in response to receiving a first stop request indicating to stop feeding, time the process, and in response to reaching a shutdown trigger condition, generate a stop command so that the feeding component stops feeding the electrolytic cells in each homogeneous group in response to the stop command, wherein the shutdown trigger condition includes: receiving a tail stop request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail stop request and the first stop request is less than the shutdown waiting window period, or not receiving a tail feed request sent by the last homogeneous group in the homogeneous group, but the shutdown waiting window period has been reached.

[0099] According to an embodiment of this application, the controller uses the stop request triggered by the full material level gauge as the signal basis, aggregates stop requests from multiple homogeneous groups through a closing waiting window, and determines in real time whether the closing trigger condition is met. Once the closing trigger condition is met, a stop command is generated. The closing trigger condition has two implementations: one is that within the timing period, a tail stop request sent by the last homogeneous group is received, and the time interval between the tail stop request and the first stop request is less than the closing waiting window; the other is that not all stop requests from homogeneous groups are received, but the timing duration has reached the start waiting window. Upon receiving the stop command from the controller, the system's feeding component can stop feeding the electrolytic cells in each homogeneous group. Based on the above-mentioned rules for stopping feeding, the automation and intelligence of stopping feeding can be improved.

[0100] For example, one implementation could be: at the 2nd minute, a typical electrolytic cell in homogeneous group 1 detects a full material state and sends a first stop request to the controller; at the 10th minute, a typical electrolytic cell in homogeneous group 3 detects a full material state and sends a last stop request to the controller; the current closing waiting window period is 20 minutes; the time interval between the last stop request and the first stop request is 8 minutes, which is less than the closing waiting window period; the controller generates a batch stop command, and the feeding component stops feeding materials to the electrolytic cells in each homogeneous group.

[0101] For example, another implementation could be that the timing period has reached the start-up waiting window (20 minutes), and no stop requests have been received from all homogeneous groups within the homogeneous group, thus satisfying the second start-up trigger condition; the controller generates a batch stop command, and the feeding component stops feeding materials to the electrolytic cells in each homogeneous group.

[0102] In some specific embodiments, determining the shutdown waiting window period of the homogeneous group to which the target homogeneous group belongs includes: for each homogeneous group in the homogeneous group, inputting the historical shutdown time series information of the homogeneous group into an autoregressive model, and outputting the predicted shutdown time series information for future periods; determining the initial shutdown waiting window period based on the predicted shutdown time series information of each of the multiple homogeneous groups; and correcting the initial shutdown waiting window period based on the shutdown drift trend term to determine the shutdown waiting window period, wherein the shutdown drift trend term is used to characterize the difference between the predicted initial shutdown waiting window period and the actual shutdown waiting window period.

[0103] According to embodiments of this application, by learning the historical shutdown time series of homogeneous groups through an autoregressive model, the predicted shutdown time series for future periods of each homogeneous group can be accurately output. This effectively captures the inherent patterns of the group's shutdown rhythm, replacing the traditional experience-based fixed window period setting, and providing scientific data for determining the initial shutdown waiting window period. Based on the initial window period determined by the predicted shutdown sequences of multiple homogeneous groups, the shutdown time distribution of each homogeneous group can be maximized, a reasonable boundary can be reserved for the aggregation of shutdown requests, and multi-cell batch coordinated shutdown can be achieved, reducing the frequent start-up and shutdown of the feeding equipment and the invalid running time. At the same time, by dynamically correcting the initial window period through the shutdown drift trend term, the deviation between the model prediction value and the actual shutdown rhythm can be accurately quantified, adapting to the shutdown rhythm drift caused by operating condition fluctuations. This avoids insufficient feeding of some electrolytic cells due to an excessively short window period, and also prevents material overflow caused by an excessively long window period, continuously balancing the feeding accuracy, process stability, and system energy-saving effect.

[0104] In some specific embodiments, the calculation method for the shutdown drift trend term includes: determining shutdown deviation sequence information based on the real historical shutdown time series information and the predicted historical shutdown time series information for the target historical period, wherein the predicted historical shutdown time series information is determined by inputting the real historical shutdown time series information of the historical period before the target historical period into the autoregressive model; and determining the shutdown drift trend term based on the shutdown deviation sequence information.

[0105] According to embodiments of this application, this method compares the actual shutdown time series of a target historical period with the predicted historical shutdown time series output by the autoregressive model to obtain shutdown deviation sequence information and calculate the shutdown drift trend term accordingly. This accurately quantifies the true deviation between the model's predicted value and the actual shutdown rhythm, effectively capturing the shutdown rhythm drift caused by changes in operating conditions and material consumption in the electrolytic cell, thus solving the problem that a fixed shutdown window period cannot adapt to dynamic changes in operating conditions. Correcting the initial shutdown window period based on the drift trend term obtained by this method can effectively compensate for prediction errors and optimize prediction accuracy.

[0106] Figure 4 A flowchart illustrating the feeding control method provided in an embodiment of this application is shown.

[0107] like Figure 4 As shown, a feeding control method is provided, which is applied to an electrolysis system. The electrolysis system includes a homogeneous group, a controller, and a feeding component. The feeding control method includes operations S410 to S430.

[0108] In operation S410, the controller responds to receiving a first feed request instructing the addition of material to the electrolyzer in the target homogeneous group within the homogeneous group, and begins timing.

[0109] The similarity between the electrolysis processes implemented in each electrolyzer in the target homogeneous group meets the predetermined threshold of the process. The homogeneous group includes multiple homogeneous groups, and the feeding time interval between multiple homogeneous groups is less than the start-up waiting window period. The start-up waiting window period is determined based on the historical feeding time series of each homogeneous group in the homogeneous group.

[0110] When operating S420, the controller generates a feeding command in response to the achievement of the start trigger condition.

[0111] The trigger conditions for activation include: receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail feeding request and the first feeding request is less than the activation waiting window period; or not receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, but the activation waiting window period has been reached.

[0112] In operation S430, the feeding component responds to the feeding command sent by the controller and feeds the electrolyzers in each homogeneous group.

[0113] According to the feeding control method of this application, relying on the convergent characteristics of electrolysis processes within homogeneous groups, the first feeding request is used as the timing starting point. Two start-up triggering conditions are set in conjunction with a preset start-up waiting window. This achieves efficient aggregation of feeding requests within homogeneous groups while ensuring timely feeding through the start-up waiting window, preventing process parameter fluctuations caused by material shortages in the electrolytic cell. Simultaneously, this method coordinates the feeding components to perform batch centralized feeding, significantly reducing the start-up and shutdown frequency of the feeding components, lowering equipment idle and operating energy consumption, and improving the overall energy efficiency of the feeding system. Furthermore, this feeding control method determines the start-up waiting window based on historical feeding time series, ensuring that feeding control conforms to actual feeding patterns and achieving a balance between feeding request aggregation and process stability.

[0114] According to embodiments of this application, the feeding control method further includes:

[0115] Based on the electrolysis process parameters of each of the multiple electrolyzers, the electrolysis process characteristics of each electrolyzer are determined.

[0116] Based on the electrolysis process characteristics of each electrolyzer, multiple electrolyzers are clustered to identify multiple electrolyzers belonging to the same homogeneous group.

[0117] Based on the number of electrolytic cells in the homogeneous group, the target number of target electrolytic cells for sending feeding requests to the controller is determined from the homogeneous group.

[0118] Optionally, the initial feeding request can be obtained based on the signal sent by the material level gauge installed in any target electrolytic cell.

[0119] According to embodiments of this application, the feeding control method further includes:

[0120] In response to receiving the first stop request indicating to stop feeding, a timer is set, and in response to the achievement of the shutdown trigger condition, a stop command is generated so that the feeding component responds to the stop command and stops feeding the electrolyzers in each homogeneous group.

[0121] The closing trigger conditions include: receiving a tail stop request from the last homogeneous group in the homogeneous group, and the time interval between the tail stop request and the first stop request is less than the closing waiting window period; or not receiving a tail feed request from the last homogeneous group in the homogeneous group, but the closing waiting window period has been reached.

[0122] According to embodiments of this application, the feeding control method for determining the start-up waiting window for homogeneous groups includes:

[0123] For each homogeneous group within a homogeneous group, the historical feeding sequence information of the homogeneous group is input into the autoregressive model, and the predicted feeding time series information for future periods is output.

[0124] Based on the predicted feeding time series information of multiple homogeneous groups, the initial start-up waiting window period is determined; and

[0125] Based on the startup drift trend term, the initial startup waiting window period is corrected to determine the startup waiting window period. The startup drift trend term is used to characterize the difference between the predicted initial startup waiting window period and the actual startup waiting window period.

[0126] According to embodiments of this application, the method for controlling the feeding of homogeneous groups includes determining the closing waiting window period, which includes:

[0127] For each homogeneous group within a homogeneous group, the historical material stoppage time series information of the homogeneous group is input into the autoregressive model, and the predicted material stoppage time series information for future periods is output.

[0128] Based on the predicted shutdown time series information of multiple homogeneous groups, the initial shutdown waiting window period is determined; and

[0129] Based on the closure drift trend term, the initial closure waiting window period is corrected to determine the closure waiting window period. The closure drift trend term is used to characterize the difference between the predicted initial closure waiting window period and the actual closure waiting window period.

[0130] According to embodiments of this application, the feeding control method for controlling the feeding of materials into electrolytic cells within each homogeneous group includes:

[0131] Open the pipeline valves on the pipelines connected to the electrolytic cells in each homogeneous group, wherein the other end of the pipeline is connected to the material box;

[0132] Control the operation of the fans connected to each pipeline so that the materials are transported through the pipelines to the material bins of the electrolytic cells in each homogeneous group.

[0133] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Without departing from the scope of this application, those skilled in the art can make various substitutions and modifications, all of which should fall within the scope of this application.

Claims

1. An electrolysis system, characterized in that, The electrolysis system includes: Homogeneous groups, including multiple homogeneous groups, wherein the similarity between the implemented electrolysis processes meets a predetermined threshold, and the feeding interval between the multiple homogeneous groups is less than the start-up waiting window period, wherein the start-up waiting window period is determined based on the historical feeding time series of each homogeneous group in the homogeneous group; The controller is configured to: in response to receiving an initial feed request instructing the addition of material to an electrolyzer in a target homogeneous group, time itself, and in response to reaching a start-up triggering condition, generate a feed command, wherein the start-up triggering condition includes: receiving a tail feed request from the last homogeneous group in the target homogeneous group, and the time interval between the tail feed request and the initial feed request is less than the start-up waiting window period; or not receiving a tail feed request from the last homogeneous group in the target homogeneous group, but the start-up waiting window period has been reached; and The feeding assembly is configured to feed the electrolyzers in each of the homogeneous groups in response to the feeding command.

2. The electrolysis system according to claim 1, characterized in that, The electrolysis system also includes: At least one hopper configured to hold material added to the electrolytic cell via the feeding assembly, and to feed the electrolytic cell at a predetermined frequency; and The material level gauge is installed in the hopper of the target electrolytic cell of the homogeneous group and is configured to trigger the generation of the feeding request and send the feeding request to the controller when the content of the material in the hopper is detected to be lower than a first predetermined material threshold.

3. The electrolysis system according to claim 2, characterized in that, The controller is also configured to: In response to receiving an initial stop request instructing the feeding to cease, a timer is established. In response to the achievement of a shutdown trigger condition, a stop command is generated so that the feeding component, in response to the stop command, stops feeding the electrolytic cells in each of the homogeneous groups. The shutdown trigger condition includes: receiving a tail stop request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail stop request and the initial stop request is less than the shutdown waiting window period; or not receiving a tail feed request sent by the last homogeneous group in the homogeneous group, but the shutdown waiting window period has been reached.

4. The electrolysis system according to claim 3, characterized in that, The electrolysis system also includes: A material level gauge is installed in the hopper of the target electrolytic cell of the homogeneous group. It is configured to trigger the generation of the material stop request and send the material stop request to the controller when the content of the material in the hopper is detected to be higher than a second predetermined material threshold.

5. The electrolysis system according to claim 3, characterized in that, The electrolysis system also includes: The processor is configured as follows: Based on the electrolysis process parameters of each electrolytic cell, the electrolysis process characteristics of each electrolytic cell are determined; Based on the electrolysis process characteristics of each electrolytic cell, the multiple electrolytic cells are clustered to identify multiple electrolytic cells belonging to the same homogeneous group; and Based on the number of electrolytic cells in the homogeneous group, a target number of target electrolytic cells for sending the feeding request to the controller is determined from the homogeneous group.

6. The electrolysis system according to claim 5, characterized in that, The processor is also configured to: For each homogeneous group in the homogeneous group, the historical feeding sequence information of the homogeneous group is input into the autoregressive model, and the predicted feeding time series information for future periods is output. Based on the predicted feeding time series information of each of the multiple homogeneous groups, the initial start-up waiting window period is determined; as well as Based on the startup drift trend term, the initial startup waiting window period is corrected to determine the startup waiting window period, wherein the startup drift trend term is used to characterize the difference between the predicted initial startup waiting window period and the actual startup waiting window period.

7. The electrolysis system according to claim 6, characterized in that, The processor is also configured to: Based on the actual historical feeding time series information and the predicted historical feeding time series information for the target historical period, the start-up deviation time series information is determined. The predicted historical feeding time series information is determined by inputting the actual historical feeding time series information for historical periods prior to the target historical period into the autoregressive model. Based on the startup deviation time series information, the startup drift trend term is determined.

8. The electrolysis system according to claim 5, characterized in that, The processor is also configured to: For each homogeneous group in the homogeneous group, the historical material stoppage time series information of the homogeneous group is input into the autoregressive model, and the predicted material stoppage time series information for future periods is output. Based on the predicted shutdown time series information of each of the multiple homogeneous groups, the initial shutdown waiting window period is determined; as well as Based on the closure drift trend term, the initial closure waiting window period is corrected to determine the closure waiting window period, wherein the closure drift trend term is used to characterize the difference between the predicted initial closure waiting window period and the actual closure waiting window period.

9. The electrolysis system according to any one of claims 2 to 8, characterized in that, The feeding assembly includes: A pipeline valve, disposed on a pipeline connecting the electrolytic cell and the feed hopper, is configured to open the pipeline in response to the feeding command; and A blower, connected to the pipelines configured for each of the electrolytic cells, is configured to start operation in response to the opening of the pipelines, so that materials are conveyed through the pipelines to the hoppers of the electrolytic cells in each of the homogeneous groups.

10. A feeding control method, characterized in that, Applied to an electrolysis system, the electrolysis system including a homogeneous group, a controller, and a feeding assembly, the method includes: The controller, in response to receiving a first feeding request instructing the feeding of a cell in a target homogeneous group within the homogeneous group, begins timing, wherein the similarity between the electrolysis processes implemented by each cell in the target homogeneous group meets a predetermined process threshold, the homogeneous group includes multiple homogeneous groups, the feeding time interval between the multiple homogeneous groups is less than a start-up waiting window period, and the start-up waiting window period is determined based on the historical feeding time series of each homogeneous group in the homogeneous group; The controller generates a feeding command in response to the achievement of a start-up trigger condition, wherein the start-up trigger condition includes: receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, and the time interval between the tail feeding request and the first feeding request is less than the start-up waiting window period; or not receiving a tail feeding request sent by the last homogeneous group in the homogeneous group, but the start-up waiting window period has been reached; and The feeding component, in response to the feeding command sent by the controller, feeds the electrolyzers in each of the homogeneous groups.

Images