Bottleneck prediction optimization method, device and equipment for battery positive electrode material production line

Abstract

The invention belongs to the technical field of industrial control, and discloses a bottleneck prediction optimization method, device and equipment for a battery positive electrode material production line, and the method comprises the steps: collecting the spatial data and time series data of the production line in real time; synchronizing the spatial data and the time sequence data to a digital twinborn model, wherein the digital twinborn model is pre-constructed based on the production line; converting the spatial data and the time sequence data into spatio-temporal data; inputting the spatio-temporal data into a pre-trained spatio-temporal diagram convolutional network to obtain a prediction result of the bottleneck of the production line; when it is judged that a production line bottleneck exists according to the prediction result, based on the prediction result, an optimization strategy is generated according to a priority rule, the optimization strategy is operated on the digital twin model in a simulation mode, a simulation result is obtained, and if it is judged that the simulation result meets the expectation, the production line is controlled according to the optimization strategy. Efficient and stable operation of a production line can be ensured, and production efficiency and stability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial control, and in particular relates to a bottleneck prediction and optimization method, device and equipment for a battery positive electrode material production line. Background Art

[0002] Existing battery positive electrode materials (such as bismuth oxide , cobalt oxide CoO2, lithium iron phosphate LiFePO4, etc.) production lines mainly rely on traditional automated control systems (such as PLC and SCADA) to monitor equipment and perform basic operations. Although these systems have achieved automation of the production process to a certain extent, they have obvious deficiencies in intelligent decision-making, optimization capabilities, and coping with complex environmental changes, and cannot meet the needs of efficient and stable production.

[0003] For example: Traditional automated control systems mainly rely on preset rules and fixed operating parameters to perform tasks, lack adaptive capabilities, and cannot flexibly adjust operating strategies when the production environment changes. They lack intelligent decision-making and prediction capabilities based on real-time data and cannot effectively identify potential bottlenecks or unstable factors in production. Summary of the invention

[0004] The purpose of the present invention is to provide a bottleneck prediction optimization method, device, equipment and storage medium for a battery positive electrode material production line, which can ensure efficient and stable operation of the production line and improve production efficiency and stability.

[0005] The first aspect of the present invention discloses a bottleneck prediction and optimization method for a battery positive electrode material production line, comprising: Real-time collection of spatial data and time series data of production lines; Synchronizing the spatial data and the time series data to a digital twin model, wherein the digital twin model is pre-built based on the production line; Converting the spatial data and the time series data into spatiotemporal data; Inputting the spatiotemporal data into a pre-trained spatiotemporal graph convolutional network to obtain a prediction result of the production line bottleneck; When it is determined that there is a production line bottleneck according to the prediction results: Based on the prediction results, an optimization strategy is generated according to priority rules, and the optimization strategy is simulated and run on the digital twin model to obtain simulation results. If it is determined that the simulation results meet expectations, the production line is controlled according to the optimization strategy.

[0006] In some embodiments, after collecting the spatial data and time series data of the production line in real time, the method further includes: Transmitting the spatial data and the time series data to an edge computing device; On the edge computing device, the spatial data and the time series data are preprocessed.

[0007] In some embodiments, a machine learning model is deployed on the edge computing device. On the edge computing device, the spatial data and the time series data are also input into the machine learning model to obtain a prediction result, and the production line is controlled according to the prediction result.

[0008] In some embodiments, converting the spatial data and the time series data into spatiotemporal data comprises: Setting nodes of a space-time graph according to the spatial data, wherein the nodes in the space-time graph at least include equipment and production links on the production line; Setting feature vectors of nodes of the space-time graph according to the time series data; Setting edges between nodes of the space-time graph according to the spatial data and the historical spatial data; The data of the space-time graph are combined into the space-time data.

[0009] In some embodiments, setting edges between nodes of the space-time graph according to the spatial data and the historical spatial data includes: Find whether there is a material flow or mutual dependence relationship between different devices according to the spatial data, obtain a first object, set a first edge between nodes of the space-time graph corresponding to the first object, and set the attribute of the first edge to spatial correlation; According to the spatial data and the historical spatial data, searching whether there is state transfer of the same device in adjacent time steps, obtaining a second object, setting a second edge of the node of the space-time graph corresponding to the second object between adjacent time steps, and setting an attribute of the second edge to time correlation; According to the spatial data and the historical spatial data, find out whether there is an interactive relationship between different devices or between devices and materials in adjacent time steps, obtain a third object, set the third edge of the node of the space-time graph corresponding to the third object between adjacent time steps, and set the attribute of the third edge to space-time correlation.

[0010] In some embodiments, the priority rule is to set the flow bottleneck and / or pressure bottleneck as a high priority bottleneck, set the temperature bottleneck as a medium priority bottleneck, and set the vibration bottleneck as a low priority bottleneck. Based on the prediction result, an optimization strategy is generated according to the priority rule, including: When it is determined according to the prediction result that a high priority bottleneck exists, the optimization strategy is to adjust the parameters of the first target device according to the prediction result, the first target device including a delivery pipeline, a pump and a valve; When it is determined according to the prediction result that a medium priority bottleneck exists, the optimization strategy is to adjust parameters of a second target device according to the prediction result, the second target device including a temperature control device and / or a cooling device; When it is determined according to the prediction result that a low priority bottleneck exists, the optimization strategy is to adjust the parameters of a third target device according to the prediction result, and the third target device is a device whose vibration value exceeds a preset threshold.

[0011] In some embodiments, after synchronizing the spatial data and the time series data to the digital twin model, the method further includes: The spatial data and the time series data, the historical spatial data and the historical time series data are analyzed in the digital twin model, and when it is determined that equipment performance has declined or the operating status is abnormal, the parameters of the production line are adjusted.

[0012] The second aspect of the present invention discloses a bottleneck prediction and optimization device for a battery positive electrode material production line, comprising: The acquisition module is used to collect spatial data and time series data of the production line in real time; A digital twin module, for synchronizing the spatial data and the time series data to a digital twin model, wherein the digital twin model is pre-built based on the production line; A conversion module, used for converting the spatial data and the time series data into spatiotemporal data; A prediction module, used for inputting the spatiotemporal data into a pre-trained spatiotemporal graph convolutional network to obtain a prediction result of the production line bottleneck; A control module is used for, when it is determined that a production line bottleneck exists according to the prediction result: based on the prediction result, generating an optimization strategy according to a priority rule, simulating and running the optimization strategy on the digital twin model to obtain a simulation result, and if it is determined that the simulation result meets expectations, controlling the production line according to the optimization strategy.

[0013] The third aspect of the present invention discloses an electronic device, comprising a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the bottleneck prediction and optimization method for the battery positive electrode material production line disclosed in the first aspect.

[0014] The fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the bottleneck prediction and optimization method for a battery positive electrode material production line disclosed in the first aspect.

[0015] The beneficial effects of the present invention are that the spatiotemporal data are modeled and analyzed using a spatiotemporal graph convolutional network to identify and predict bottlenecks in the production process and production optimization strategies; a digital twin model is used to perform dynamic simulation based on real-time sensor data and optimization strategies, and by combining the output results of the digital twin model and the spatiotemporal graph convolutional network, the production parameters in the production process are dynamically adjusted to reduce bottlenecks and unstable factors, thereby ensuring efficient and stable operation of the production line and improving production efficiency and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings herein show specific examples of the technical solutions described in the present invention, and together with the specific implementation methods, constitute a part of the specification, and are used to explain the technical solutions, principles and effects of the present invention.

[0017] Unless otherwise specified or defined, the same reference numerals in different drawings represent the same or similar technical features, and the same or similar technical features may also be represented by different reference numerals.

[0018] Figure 1 It is a flow chart of a bottleneck prediction and optimization method for a battery positive electrode material production line disclosed in an embodiment of the present invention; Figure 2 is a flow chart of data preprocessing using an edge computing device according to an embodiment of the present invention; Figure 3 is a flow chart of converting spatial data and time series data into spatiotemporal data according to an embodiment of the present invention; Figure 4 yes Figure 3 A flowchart of setting edges between nodes of a spatiotemporal graph in an embodiment; Figure 5 is a schematic diagram of a time-space diagram of an embodiment of the present invention; Figure 6 is a structural schematic diagram of a bottleneck prediction and optimization device for a battery positive electrode material production line according to an embodiment of the present invention; Figure 7 It is a structural schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] Unless otherwise specified or defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art. In combination with the technical solution of the present invention in a realistic scenario, all technical and scientific terms used herein may also have meanings corresponding to the purpose of implementing the technical solution of the present invention. The "first, second..." used herein is only used to distinguish the names and does not represent a specific quantity or order. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0020] It should be noted that when a component is considered to be "fixed to" another component, it can be directly fixed to the other component or there can be a central component; when an component is considered to be "connected to" another component, it can be directly connected to the other component or there can be a central component at the same time; when an component is considered to be "installed on" another component, it can be directly installed on the other component or there can be a central component at the same time. When an component is considered to be "set on" another component, it can be directly set on the other component or there can be a central component at the same time.

[0021] Unless otherwise specified or defined, the "said" and "the" used in this document refer to the technical features or technical contents mentioned or described before the corresponding position, and the technical features or technical contents may be the same as or similar to the technical features or technical contents mentioned therein. In addition, the terms "including" and "having" and any variations thereof used in this document are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0022] The existing automated control systems for battery positive electrode materials usually rely on human intervention or fixed rules. Although they can adjust some operating parameters, they are unable to automatically generate adjustment plans based on changes in the production environment. Bottlenecks are prone to occur in the production process, resulting in low equipment utilization and high energy consumption.

[0023] To address the above problems, the present invention creates a digital twin model that is highly consistent with the actual production line. It models and analyzes spatiotemporal data through the spatiotemporal graph convolutional network (STGCN), identifies and predicts bottlenecks and anomalies in the production process, determines the optimization strategy, and uses the digital twin model to simulate and run the optimization strategy. After determining that the simulation effect meets the expectations, it automatically adjusts the production parameters to optimize the production process, making the production process efficient and stable.

[0024] The embodiment of the present invention takes the bismuth compound production line as an example to disclose a production line control method, but it is not limited to the bismuth compound production line, and can also be used in other battery positive electrode material production lines. The execution subject of the method can be an electronic device such as a computer, a laptop, a tablet computer, or a control chip embedded in an electronic device, and the present invention does not limit this.

[0025] The process flow of the bismuth compound production line of this embodiment is as follows: after the customer material (bismuth compound) is fully mixed in the reactor, it is conveyed to the centrifuge; the centrifugal material is received by a receiving vehicle and connected to a negative pressure gas vacuum conveying system; the material is pumped to a conical screw belt vacuum dryer for drying; after drying, the material falls into the suction hopper of the vacuum loader below; the material is pumped to the buffer bin through the vacuum conveying system; the buffer bin is connected to a hammer crusher below; after the material is crushed, it is unloaded to the suction hopper of the vacuum loader below; the material in the suction hopper of the vacuum loader is conveyed to the homogenizer through the vacuum conveying system; after homogenization, the material falls into the hopper below; the hopper is connected to the lifting docking device; the material in the hopper is conveyed to the automatic baler for packaging through the material conveying system.

[0026] In order to facilitate the understanding of the present invention, specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings.

[0027] like Figure 1 As shown, the method comprises the following steps: Step S100: collecting spatial data and time series data of the production line in real time; Temperature, pressure, flow, humidity, vibration, weight and other sensors are deployed on the bismuth compound production line to collect real-time data of each link and obtain time series data. The sensors also collect the equipment location, current production link and the interaction between different links to obtain spatial data. By collecting spatial data and time series data at the same time, the spatiotemporal correlation of the production process can be obtained. Among them, the production links of the bismuth compound production line include centrifugation, drying, crushing, homogenization, etc., and the interaction between the links is such as the next link of the centrifugation link is drying, etc.

[0028] Specifically, this embodiment deploys sensors on key equipment and conveying systems. Sensors are deployed on the reactor to monitor the temperature, pressure, and flow rate during the chemical reaction process to ensure stable reaction conditions; sensors are deployed at the centrifuge to monitor material flow and vibration to ensure centrifugal separation effect and equipment health; sensors are deployed at the dryer to measure temperature, humidity, and flow rate to ensure the effectiveness of the drying process; sensors are deployed at the crusher to monitor equipment vibration and pressure to prevent uneven crushing or equipment damage; sensors are deployed at the homogenizer to monitor flow rate to ensure uniform mixing; sensors are deployed at the baler to monitor weight and material flow rate during the baling process; flow sensors and pressure sensors are installed on the material conveying system to monitor the flow state of the material and ensure smooth transfer of materials between different links.

[0029] The deployed sensors can obtain spatial data and time series data. Spatial data includes the physical location of each production link (reactor, centrifuge, dryer, crusher, etc.), the relationship between each production link, equipment load status, etc. Time series data includes the flow, speed, output, etc. of each link in the production process. Time series data is dynamic and can reflect the status of each link of the production line at different times.

[0030] like Figure 2 As shown, in this embodiment, after collecting the spatial data and time series data of the production line in real time, the following steps are also included: Step S110: Transmitting spatial data and time series data to an edge computing device; Step S120: preprocessing the spatial data and time series data on the edge computing device; Traditional automation control systems rely on centralized data processing and remote monitoring, which results in information transmission delays. This makes it impossible for the control system to reflect the status of the production line in real time, especially when dealing with emergencies, and it is slow to respond and unable to achieve timely adjustments and optimizations.

[0031] In this embodiment, the spatial data and time series data collected by the sensor are first transmitted to the edge computing device, and the edge computing device is used to pre-process the real-time data of these sensors. Through local processing, data transmission delay can be reduced. The processing results are then fed back to the control system in real time to accelerate the decision-making process.

[0032] In some embodiments, a machine learning model can also be deployed on the edge computing device. After the spatial data and time series data are transmitted to the edge computing device, the edge computing device first pre-processes the sensor data (such as noise filtering, data normalization, etc.) and extracts useful features. These features are input into the machine learning model, which can support rapid evaluation and prediction of equipment status, material flow, production efficiency, etc. After obtaining the prediction results, the production line is controlled according to the prediction results.

[0033] Using edge computing devices to process and feedback sensor data in real time can quickly respond to changes in the production process and automatically optimize various parameters of the production process (such as temperature, pressure, flow rate, etc.) to ensure efficient and stable operation of the production line.

[0034] Step S200: synchronizing the spatial data and the time series data to the digital twin model, where the digital twin model is pre-built based on the production line; By building a virtual model that is highly consistent with the actual production line and combining it with physical modeling such as fluid mechanics and equipment status, a digital twin model can be formed to accurately simulate the production process. After the spatial data and time series data collected in real time are integrated into the digital twin model, various behaviors in the production process can be dynamically simulated to ensure that the digital twin model can accurately reflect the actual production status and provide data support for subsequent optimization and decision-making.

[0035] Specifically, the digital twin model construction process mainly includes physical modeling and sensor data integration. During physical modeling, physical models of various production links including reactors, centrifuges, dryers, crushers, etc. are established. These models describe the structure, working principle and behavioral characteristics of each device; during sensor data integration, the data collected by actual sensors (such as temperature, flow, pressure, vibration, etc.) in real time are synchronized to the digital twin model to ensure that the digital twin model can accurately reflect the dynamic state of the actual production line.

[0036] Once the digital twin model is built, real-time data synchronization and simulation can be performed. Specifically, real-time data is collected from production line sensors and input into the digital twin model through a data interface. The digital twin model simulates the production line status based on this real-time data and accurately presents the current status of the production line (such as material flow, equipment load, etc.).

[0037] In this embodiment, the digital twin model is also used to judge the abnormal conditions of the production line, so as to quickly respond to the abnormal conditions on the production line. Specifically, after synchronizing the spatial data and time series data to the digital twin model, the saved historical spatial data and historical time series data are combined in the digital twin model for analysis, for example: the current collected data and historical data of the same sensor are compared and analyzed to determine whether the performance of the current equipment has declined or whether the operating status is abnormal. When an abnormal situation occurs, the parameters of the production line are directly adjusted.

[0038] In this embodiment, the data collected by sensors is combined with the digital twin model to monitor the operating status (such as temperature, pressure, vibration, flow and other key parameters of the equipment) of key equipment in the production line (such as reactors, centrifuges, dryers, etc.) in real time, analyze the real-time data, and combine the historical operating status of the equipment to identify potential failure risks. When the performance of the equipment decreases or the operating status is abnormal, an early warning signal is issued to prompt the operator or trigger the automatic adjustment mechanism. It can respond quickly to abnormal conditions to ensure the stability of the production process.

[0039] Step S300: converting spatial data and time series data into spatiotemporal data; Combine spatial data and time series data according to the preset data structure to form spatiotemporal data. For example, the dimension of spatiotemporal data at each time step is , where T is the time step, P is the number of devices, and Q is the total dimension of all sensor data.

[0040] In order to better represent the spatiotemporal characteristics in spatial data and time series data, this embodiment performs spatiotemporal graph modeling based on spatial data and time series data. After the spatiotemporal graph modeling, the spatiotemporal graph convolutional network can better capture the spatiotemporal dependency between devices.

[0041] like Figure 3 As shown in the figure, the specific steps for converting spatial data and time series data into spatiotemporal data include: Step S310: setting nodes of the space-time graph according to the spatial data, where the nodes in the space-time graph at least include equipment and production links on the production line; Step S320: setting feature vectors of nodes of the space-time graph according to the time series data; Step S330: setting the edges between nodes of the space-time graph according to the spatial data and the historical spatial data; Spatiotemporal graph modeling includes two steps: constructing the spatiotemporal graph of the production line and constructing the time dependency. Among them, when constructing the spatiotemporal graph of the production line, each device in the spatial data (such as reactor, centrifuge, dryer, etc.) and the related production links are used as nodes of the spatiotemporal graph. The feature vector of each node contains the real-time status of the device obtained from the time series data, such as temperature, pressure, flow, vibration, etc. For example, the node feature vector of the reactor may contain: temperature, pressure, flow, flow rate, etc. The relationship between nodes (such as material flow, temperature transfer, equipment collaboration, etc. between devices) is defined as an edge. The edge between nodes can be determined based on spatial data and historical spatial data, and the weight of the edge can be set based on information such as physical connection, transmission speed, workload, etc. between devices. For example, if there is material flow between the reactor and the centrifuge, they can be connected by an edge, and the weight of the edge can be set based on the material transmission rate. The edge between nodes also includes the time dependency at different time steps. The time series data determines the state of each node at different times. The characteristics of each node in the graph will change over time. For example, the state of each device (such as flow, pressure) changes over time, forming a time series.

[0042] This embodiment also sets the edges between nodes to be spatially related, temporally related, and spatiotemporally related, which can better reflect the temporal dependency. Specifically, the edges between the nodes of the spatiotemporal graph are set according to the spatial data and the historical spatial data. The specific steps are as follows: Figure 4 As shown, including: Step S331: according to the spatial data, find out whether there is material flow or mutual dependence between different devices, obtain the first object, set the first edge between the nodes of the space-time graph corresponding to the first object, and set the attribute of the first edge to spatial correlation; Spatial correlation reflects the correlation between different devices at the same time step. Equipment with material flow or mutual dependence is selected from the spatial data and set as the first object. The node corresponding to the first object is determined in the space-time graph. Then, the first edges between these nodes are set in the space-time graph according to the correlation between the first objects, and the attributes of these first edges are set as spatial correlation.

[0043] Step S332: according to the spatial data and the historical spatial data, find out whether there is state transfer of the same device in adjacent time steps, obtain the second object, set the second edge of the node of the space-time graph corresponding to the second object between adjacent time steps, and set the attribute of the second edge to time correlation; Time correlation reflects the correlation between the same devices at different time steps. Based on spatial data and historical spatial data, devices with state transfer are screened out, and these devices are set as second objects. The nodes corresponding to the second objects are determined in the space-time graph. Then, based on the correlation between the second objects, the second edges of these nodes between adjacent time steps are set in the space-time graph, and the attributes of these second edges are set as time correlation.

[0044] Step S333: According to the spatial data and historical spatial data, find out whether there is an interactive relationship between different devices or between devices and materials in adjacent time steps, obtain the third object, set the third edge of the node of the space-time graph corresponding to the third object between adjacent time steps, and set the attribute of the third edge to space-time correlation.

[0045] Spatiotemporal correlation reflects the correlation between different devices or between devices and materials at different time steps. Based on spatial data and combined with historical spatial data, equipment and materials with interactive relationships are screened out, and these equipment and materials are set as third objects. The nodes corresponding to the third objects are determined in the spatiotemporal graph, and then the third edges of these nodes between adjacent time steps are set in the spatiotemporal graph according to the correlation between the third objects, and the attributes of these third edges are set as spatiotemporal correlation.

[0046] The time-space diagram of this embodiment is as follows Figure 5As shown in the figure, the time-space diagram at different times t1 and t2 is shown. Among them, the green dots are nodes, and each node represents the key equipment or material status in the production line, such as centrifuges, reactors, materials, etc. In each time step t1 and t2, the node vectors represent the state characteristics of the corresponding equipment respectively; the orange arrows are spatial correlations, indicating the material flow or interdependence between equipment in the same time step. For example: at time step t1, the centrifuge receives materials from the reactor, and the state of the reactor affects the operation of the centrifuge. At the same time, the operating state of the reactor directly determines the quality and flow of the material output; the blue arrows are time correlations, indicating the state transfer of the same equipment at different times. For example: the state of the reactor at time t1 (such as temperature, pressure) affects its operating parameters at time t2; the purple arrows are time-space correlations, indicating the interactive relationship between nodes across time steps. For example: the operating state of the centrifuge at time t1 may have a direct impact on the reactor and material state at t2.

[0047] Step S340: Combine the data of the space-time graph into space-time data.

[0048] After the spatiotemporal graph modeling is completed, the spatiotemporal graph data can be combined and output according to a predetermined data structure to obtain spatiotemporal data. Combining and outputting the spatiotemporal graph data is a conventional technique in the art and will not be described in detail here.

[0049] Step S400: input the spatiotemporal data into the pre-trained spatiotemporal graph convolutional network to obtain a prediction result of the production line bottleneck; Spatial-Temporal Graph Convolutional Networks (STGCN) is a deep learning model for processing spatiotemporal data, especially suitable for the combination of graph structured data (such as social networks, traffic networks, etc.) and time series data. STGCN combines the advantages of graph convolutional networks (GCN) and temporal convolutional networks (TCN), and can effectively capture spatial and temporal dependencies.

[0050] The main structure of the STGCN model consists of two parts: a graph convolutional layer (GCN) and a temporal convolutional layer (CNN). The graph convolutional layer (GCN) is used to process the spatial dependencies between nodes and extract the relational features between devices, while the temporal convolutional layer (CNN) is used to process time series information and extract the behavior patterns of each node that change over time.

[0051] The spatiotemporal data is input into the STGCN network. The spatiotemporal data combines time series data and spatial data. The spatial relationship between devices is learned through the graph convolution layer, and the time series features are extracted through the time series convolution layer. Therefore, the STGCN network can learn and capture the complex spatiotemporal relationship between various devices and links in the production line. The output result of the STGCN network is whether a bottleneck will occur for each device within a given time window, usually using a binary classification result (0 means no bottleneck, 1 means bottleneck).

[0052] When training the spatiotemporal graph convolutional network, the training data set is first preprocessed, such as moving average denoising and normalization steps. Then the preprocessed time series data is divided into fixed time windows (for example, 10 time steps). The data in each window will be used as the input of the spatiotemporal graph convolutional network. The data set is divided into a training set and a validation set in an 8:2 ratio to train the spatiotemporal graph convolutional network.

[0053] The loss function can adopt the cross entropy loss function, the goal is to minimize the difference between the predicted value and the true label. The specific expression is: , Among them, y is the true label, is the predicted output of the model.

[0054] Use Adam or other optimization algorithms to update the parameters of the spatiotemporal graph convolutional network until the loss function converges. Use the training set data to train the spatiotemporal graph convolutional network and evaluate the network performance through the validation set. Adjust hyperparameters (such as learning rate, batch size, etc.) and select the optimal model. After training, the STGCN network can predict in real time whether bottlenecks occur in various devices in the production process based on real-time sensor data (such as flow, temperature, pressure, etc.) to dynamically adjust the operating parameters in the production process.

[0055] Bottleneck prediction directly affects equipment load, fluid flow behavior, and material transfer efficiency. For example, when equipment load is too high, it may cause fluid transfer obstruction or material backlog, which in turn affects overall production efficiency. Through accurate bottleneck prediction, potential problems can be identified in advance and the impact of these changes on other links can be predicted. When bottlenecks exist, optimization strategies can be generated based on real-time data and prediction results to adjust parameters in the production process (such as temperature, pressure, flow rate, etc.) to reduce waste and improve production efficiency and equipment utilization.

[0056] Step S500: When it is determined that there is a production line bottleneck based on the prediction results: based on the prediction results, an optimization strategy is generated according to the priority rules, and the optimization strategy is simulated and run on the digital twin model to obtain simulation results. If it is determined that the simulation results meet expectations, the production line is controlled according to the optimization strategy.

[0057] Since there are many devices on the production line and they are related to each other, there are many optimization strategies when a production line bottleneck occurs. When multiple adjustment requirements occur at the same time, the processing order is determined according to the priority rule. High priority first, then medium priority, and finally low priority.

[0058] Specifically, based on the prediction results, domain knowledge and expert experience, a set of priority rules combined with specific equipment status and process flow is designed. When it is determined that there is a bottleneck in the production line based on the prediction results, the rule system is used to generate an optimization strategy based on the operating status of different equipment and the real-time working conditions of the production line. Among them, the optimization strategy is used to dynamically adjust relevant parameters (such as the operating rate of the equipment, temperature setting value, material flow rate, etc.). The optimization strategy is then fed back to the digital twin model, and the simulation function of the digital twin model is used to simulate the adjustment plan (such as adjusting the temperature, pressure or production speed), and verify the feasibility and effect of the improvement measures. If it is determined that the simulation results meet expectations, such as the bottleneck is eliminated, the production line is controlled according to the optimization strategy. Thereby optimizing the production process and improving overall efficiency.

[0059] In this embodiment, when STGCN predicts a potential bottleneck, the virtual test environment in the digital twin model is started (wherein the virtual environment is a simulation environment isolated from actual production in the digital twin model, which is created through real-time or historical data, physical modeling and process modeling to test and verify the effects of different adjustment schemes). At this time, the virtual test environment runs in parallel with real-time monitoring. The virtual test environment is used to test the feasibility of different optimization strategies. Real-time monitoring is synchronized with actual production to monitor the operating status and key parameters of the equipment in real time. In the virtual test environment, based on the adjustment scheme corresponding to the simulation optimization strategy, new status data is obtained, and the new status data is input into STGCN. The production line operation status is analyzed in time and space again to verify whether the potential bottleneck point is eliminated or alleviated; if STGCN predicts that there is still a bottleneck, the optimization strategy is adjusted to continue testing until the corresponding bottleneck is eliminated. Then, the corresponding parameters are adjusted in the actual production implementation to avoid potential bottleneck problems. For example, if a flow bottleneck is detected, a simulation test is performed by adjusting the parameter combination of the material flow rate, and finally a flow adjustment scheme verified by STGCN is selected and implemented in actual production. After the test is completed, the virtual test environment is exited and the real-time synchronization status of the digital twin model and actual production is restored.

[0060] In the production process, low flow means that the material cannot be effectively transported, which may be caused by pipeline blockage, equipment failure (such as pump damage) or other flow obstacles. When the flow rate is lower than a certain threshold, the equipment may have a bottleneck, and priority should be given to problems such as blockage of the conveying pipeline, reduced pump efficiency or increased material flow resistance; high temperature may cause equipment overload, especially for equipment that requires precise temperature control, such as reactors. If the temperature exceeds the preset range, it may be a temperature control system failure or cooling system failure. When the temperature exceeds the set safety upper limit, overheating problems may occur, resulting in equipment overload or low reaction efficiency; when the pressure rises or falls abnormally, it may indicate problems with the equipment or pipeline system, such as equipment blockage, valve failure or pipeline leakage. Low pressure may indicate pipeline leakage or pump failure. Excessive pressure may indicate equipment overload or valve failure. When the vibration value exceeds the normal range, it may be a sign of mechanical failure, causing the equipment to fail to work properly. High vibration may indicate mechanical damage, bearing failure or imbalance problems, which may cause the equipment to fail to reach normal production capacity.

[0061] In this embodiment, it is considered that flow problems may cause stagnation of material transportation, directly affecting the continuity of the production line; pressure problems may cause equipment damage, seriously threatening production safety and equipment stability; temperature abnormalities may have a greater impact on product quality and the stable operation of equipment; vibration abnormalities are usually a manifestation of mechanical failure. Although they may cause equipment damage, their immediate impact on production is relatively small. Therefore, the priority rules established are to set flow bottlenecks and / or pressure bottlenecks as high-priority bottlenecks, temperature bottlenecks as medium-priority bottlenecks, and vibration bottlenecks as low-priority bottlenecks. For example: high-priority issues such as insufficient flow and abnormal pressure are handled first, because these problems may directly affect the continuity of the production line; then deal with medium-priority issues such as abnormal temperature, which may affect product quality and equipment stability; finally, deal with low-priority issues such as abnormal vibration, which may be early manifestations of mechanical failure and are usually arranged for subsequent maintenance.

[0062] In the above description, illustratively, the specific process of generating an optimization strategy based on the prediction result and the priority rule is as follows: When a high-priority bottleneck is determined to exist based on the prediction results, the optimization strategy is to adjust the parameters of the first target equipment such as the delivery pipeline, pumps and valves based on the prediction results; when a medium-priority bottleneck is determined to exist based on the prediction results, the optimization strategy is to adjust the parameters of the second target equipment such as the temperature control equipment and / or cooling equipment based on the prediction results; when a low-priority bottleneck is determined to exist based on the prediction results, the optimization strategy is to adjust the parameters of the third target equipment whose vibration value exceeds the preset threshold based on the prediction results.

[0063] By real-time monitoring of the operating status of each link in the production line, using real-time collected sensor data, model prediction results and priority rules, combined with the output results of the digital twin model and the STGCN model, it is possible to dynamically adjust the control strategy in the production process, including optimizing equipment parameters and adjusting material flow, thereby reducing manual intervention, improving production efficiency and equipment utilization, and ensuring the stability and efficiency of the production process.

[0064] After each production cycle, the actual data in the production process can be compared with the prediction results of the spatiotemporal graph convolutional network and the digital twin model, and the digital twin model and STGCN can be iteratively optimized based on the differences. Through continuous feedback and optimization, the prediction accuracy and optimization capabilities of the production process can be gradually improved. For example: compare the actual data in the production process with the predicted data, identify the source of errors, and adjust the parameters of the digital twin model and STGCN in a timely manner; through continuous feedback and correction of data, the digital twin model and STGCN are continuously optimized to improve the prediction accuracy of production line operation and the effectiveness of control strategies.

[0065] In summary, this embodiment controls the bismuth compound production line based on digital twins, edge computing and STGCN. By creating a virtual digital twin model that is highly consistent with the actual production line, sensor data is integrated in real time to monitor and predict the behavior of each link in the production process, including material transportation, equipment status, and process efficiency. Use edge computing devices to process field sensor data in real time to ensure instant data feedback and accelerate the decision-making process. Use STGCN to process spatiotemporal data, optimize bottlenecks and unstable factors in the production process in real time, automatically adjust production parameters, ensure efficient and stable operation of the production line, improve the automation level of the production line, improve equipment utilization, and reduce downtime.

[0066] like Figure 6 As shown, based on the bottleneck prediction and optimization method of the battery positive electrode material production line, the embodiment of the present invention discloses a bottleneck prediction and optimization device of the battery positive electrode material production line, including: The acquisition module 600 is used to acquire spatial data and time series data of the production line in real time; A digital twin module 610, for synchronizing the spatial data and the time series data to a digital twin model, wherein the digital twin model is pre-built based on the production line; A conversion module 620, configured to convert the spatial data and the time series data into spatiotemporal data; A prediction module 630, configured to input the spatiotemporal data into a pre-trained spatiotemporal graph convolutional network to obtain a prediction result of a production line bottleneck; The control module 640 is used to generate an optimization strategy based on the prediction result according to the priority rule when it is determined that there is a production line bottleneck based on the prediction result, simulate and run the optimization strategy on the digital twin model to obtain a simulation result, and if it is determined that the simulation result meets expectations, control the production line according to the optimization strategy.

[0067] like Figure 7 As shown, an embodiment of the present invention discloses an electronic device, including a memory 401 storing executable program codes and a processor 402 coupled to the memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute the bottleneck prediction and optimization method for the battery positive electrode material production line described in the above embodiments.

[0068] An embodiment of the present invention further discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the bottleneck prediction and optimization method for a battery positive electrode material production line described in the above embodiments.

[0069] The purpose of the above embodiments is to exemplarily reproduce and deduce the technical solution of the present invention, and to fully describe the technical solution, purpose and effect of the present invention. Its purpose is to make the public understand the disclosed content of the present invention more thoroughly and comprehensively, and it does not limit the scope of protection of the present invention.

[0070] The above embodiments are not exhaustive enumerations of the present invention, and there may be multiple other implementations not listed. Any replacement and improvement made without violating the concept of the present invention shall fall within the protection scope of the present invention.

Claims

1. A bottleneck prediction and optimization method for a battery positive electrode material production line, characterized in that: include: Real-time collection of spatial data and time series data of production lines; Synchronizing the spatial data and the time series data to a digital twin model, wherein the digital twin model is pre-built based on the production line; Converting the spatial data and the time series data into spatiotemporal data; Inputting the spatiotemporal data into a pre-trained spatiotemporal graph convolutional network to obtain a prediction result of the production line bottleneck; When it is determined that there is a production line bottleneck according to the prediction results: Based on the prediction results, an optimization strategy is generated according to priority rules, and the optimization strategy is simulated and run on the digital twin model to obtain simulation results. If it is determined that the simulation results meet expectations, the production line is controlled according to the optimization strategy.

2. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 1, characterized in that: After real-time collection of spatial data and time series data of the production line, it also includes: Transmitting the spatial data and the time series data to an edge computing device; On the edge computing device, the spatial data and the time series data are preprocessed.

3. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 2, characterized in that: A machine learning model is deployed on the edge computing device. On the edge computing device, the spatial data and the time series data are also input into the machine learning model to obtain a prediction result, and the production line is controlled according to the prediction result.

4. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 1, characterized in that: Converting the spatial data and the time series data into spatiotemporal data comprises: Setting nodes of a space-time graph according to the spatial data, wherein the nodes in the space-time graph at least include equipment and production links on the production line; Setting feature vectors of nodes of the space-time graph according to the time series data; Setting edges between nodes of the space-time graph according to the spatial data and the historical spatial data; The data of the space-time graph are combined into the space-time data.

5. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 4, characterized in that: Setting edges between nodes of the space-time graph according to the spatial data and the historical spatial data includes: Find whether there is a material flow or mutual dependence relationship between different devices according to the spatial data, obtain a first object, set a first edge between nodes of the space-time graph corresponding to the first object, and set the attribute of the first edge to spatial correlation; According to the spatial data and the historical spatial data, searching whether there is state transfer of the same device in adjacent time steps, obtaining a second object, setting a second edge of the node of the space-time graph corresponding to the second object between adjacent time steps, and setting an attribute of the second edge to time correlation; According to the spatial data and the historical spatial data, find out whether there is an interactive relationship between different devices or between devices and materials in adjacent time steps, obtain a third object, set the third edge of the node of the space-time graph corresponding to the third object between adjacent time steps, and set the attribute of the third edge to space-time correlation.

6. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 1, characterized in that: The priority rule is to set the flow bottleneck and / or pressure bottleneck as a high priority bottleneck, set the temperature bottleneck as a medium priority bottleneck, and set the vibration bottleneck as a low priority bottleneck. Based on the prediction result, an optimization strategy is generated according to the priority rule, including: When it is determined according to the prediction result that a high priority bottleneck exists, the optimization strategy is to adjust the parameters of the first target device according to the prediction result, the first target device including a delivery pipeline, a pump and a valve; When it is determined according to the prediction result that a medium priority bottleneck exists, the optimization strategy is to adjust parameters of a second target device according to the prediction result, the second target device including a temperature control device and / or a cooling device; When it is determined according to the prediction result that a low priority bottleneck exists, the optimization strategy is to adjust the parameters of a third target device according to the prediction result, and the third target device is a device whose vibration value exceeds a preset threshold.

7. The bottleneck prediction and optimization method for a battery positive electrode material production line according to claim 1, characterized in that: After synchronizing the spatial data and the time series data to the digital twin model, the method further includes: The spatial data and the time series data, the historical spatial data and the historical time series data are analyzed in the digital twin model, and when it is determined that equipment performance has declined or the operating status is abnormal, the parameters of the production line are adjusted.

8. A bottleneck prediction and optimization device for a battery positive electrode material production line, characterized in that: include: The acquisition module is used to collect spatial data and time series data of the production line in real time; A digital twin module, for synchronizing the spatial data and the time series data to a digital twin model, wherein the digital twin model is pre-built based on the production line; A conversion module, used for converting the spatial data and the time series data into spatiotemporal data; A prediction module, used for inputting the spatiotemporal data into a pre-trained spatiotemporal graph convolutional network to obtain a prediction result of the production line bottleneck; A control module is used for, when it is determined that a production line bottleneck exists according to the prediction result: based on the prediction result, generating an optimization strategy according to a priority rule, simulating and running the optimization strategy on the digital twin model to obtain a simulation result, and if it is determined that the simulation result meets expectations, controlling the production line according to the optimization strategy.

9. An electronic device, characterized in that: It comprises a memory storing executable program code and a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the bottleneck prediction and optimization method for the battery positive electrode material production line according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program enables a computer to execute the bottleneck prediction and optimization method for a battery positive electrode material production line according to any one of claims 1 to 7.

Images