Rare earth metal electrolysis factory digital twinning system based on big data
Through the digital twin system of rare earth metal electrolytic factory, the electrolytic parameters are optimized using big data and neural network models, the problem of insufficient data utilization in the existing system is solved, and efficient and stable production processes and intelligent management are achieved.
Patent Information
- Application Number
- CN202510557339.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-29
- Publication Date
- 2025-08-19
AI Technical Summary
The existing rare earth metal electrolysis system fails to effectively manage and utilize massive data, resulting in insufficient optimization of production parameters and insufficient fault prediction, affecting production efficiency and stability.
The digital twin system of rare earth metal electrolytic factory based on big data is adopted, including information acquisition module, digital twin platform, monitoring and early warning module and production optimization module. Through real-time data analysis and neural network models, it provides a data-driven process optimization solution.
It significantly improves electrolytic efficiency, reduces human intervention and operational errors, improves the automation level and operation efficiency of the production line, ensures the stability of the production process, and reduces unplanned downtime.
Smart Images
Figure CN120505675A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of rare earth metal electrolysis, and in particular to a digital twin system of a rare earth metal electrolysis plant based on big data. Background Art
[0002] The extraction and production of rare earth metals is a typical process industry, with molten salt electrolysis being one of its core processes. The electrolysis process largely relies on manual operation, resulting in low efficiency and difficulty ensuring uniform and continuous operation. In the high-temperature, dusty, and highly corrosive environment of rare earth electrolysis, some equipment may malfunction over long periods of operation, leading to interruptions in feeding and reduced accuracy. Furthermore, the chemical reactions and physical changes during the electrolysis process exhibit nonlinear characteristics, making precise control difficult using simple linear models. The electrolysis reaction is long and influenced by multiple factors, resulting in significant lag in adjusting process parameters. Multiple parameters within the electrolytic cell, such as temperature, current density, and molten salt composition, influence each other. Changes in any one parameter affect the others, increasing the difficulty of process control.
[0003] Currently, 1. Some companies have developed automated feeding equipment, such as rare earth electrolysis powder feeders and automatic feeding control systems, to address the inefficiencies caused by manual operations, but automation remains in its early stages. 2. The rare earth molten salt electrolysis process generates a large amount of production data, including process parameters, equipment status, energy consumption data, and quality inspection data. This massive amount of data is not effectively managed and utilized. 3. Process parameter control is imprecise. Due to the complexity and nonlinearity of the electrolysis process, existing data analysis methods are mostly based on simple statistical models. As a result, the large amount of data generated is not effectively utilized and lacks in-depth analysis and mining. 4. Existing systems lack fault prediction capabilities, and repairs are often performed only after a fault occurs, resulting in increased unplanned downtime and impacting production efficiency. Summary of the Invention
[0004] The purpose of the present invention is to provide a digital twin system for rare earth metal electrolysis plants based on big data, aiming to solve the problems of insufficient production parameter optimization and insufficient fault prediction caused by the failure of existing systems to effectively manage and utilize massive data.
[0005] To achieve the above objectives, the present invention provides a digital twin system for a rare earth metal electrolysis plant based on big data, comprising an information acquisition module, a digital twin platform, a monitoring and early warning module, and a production optimization module. The information acquisition module is connected to the digital twin platform, and the monitoring and early warning module and the production optimization module are respectively connected to the digital twin platform.
[0006] The information acquisition module is used to collect operating data, environmental data and image information of equipment in the electrolysis plant in real time;
[0007] The digital twin platform is used to build a three-dimensional virtual model of the electrolysis plant, dynamically update the model based on real-time data, and generate early warning strategies and production optimization strategies through simulation analysis;
[0008] The monitoring and warning module is used to output an alarm signal to the management personnel according to the warning strategy;
[0009] The production optimization module is used to output process parameter optimization suggestions to management personnel based on the optimization strategy.
[0010] Wherein, the information acquisition module includes an operation data acquisition unit, an environment data acquisition unit and an image data acquisition unit;
[0011] The operation data collection unit is used to collect operation data of factory equipment;
[0012] The environmental data acquisition unit is used to collect environmental data of factory equipment;
[0013] The image data acquisition unit is used to acquire image data of factory equipment.
[0014] The digital twin platform includes a data processing unit and a neural network model and optimization algorithm unit;
[0015] The data processing unit is used to manage the entire process of the collected process parameters, equipment status and energy consumption data;
[0016] The neural network model and optimization algorithm unit trains the neural network model based on historical data to predict electrolysis efficiency and equipment failure, and combines the genetic algorithm to perform global optimization of electrolysis parameters.
[0017] Wherein, the data processing unit includes a data acquisition and access subunit, a data storage and management subunit, a data analysis and modeling subunit, and a visualization and interaction subunit;
[0018] The data acquisition and access subunit collects temperature, current, and voltage data in real time through the sensor network and transmits them to the system through the MQTT protocol;
[0019] The data storage and management subunit uses Hadoop HDFS distributed storage and InfluxDB time series database to store massive data, and uses Apache NiFi to clean and preprocess data;
[0020] The data analysis and modeling subunit uses Apache Flink to process real-time stream data and combines it with a neural network model to predict equipment failures and electrolysis efficiency.
[0021] The visualization and interaction subunit displays real-time analysis results through Unity3D and supports user perspective switching, data filtering and alarm prompts.
[0022] Among them, the neural network model includes an input layer, a hidden layer and an output layer. The input layer is used to receive current density, electrolysis temperature, molten salt composition, electrolytic cell voltage and electrolysis time parameters; the hidden layer is used to set 2-3 layers of hidden layers that dynamically adjust the number of neurons, using a ReLU activation function; the output layer is used to output the electrolysis efficiency prediction value and the equipment fault binary classification mark.
[0023] The digital twin system of a rare earth metal electrolysis plant based on big data of the present invention includes an information acquisition module, a digital twin platform, a monitoring and early warning module, and a production optimization module. The information acquisition module is connected to the digital twin platform, and the monitoring and early warning module and the production optimization module are respectively connected to the digital twin platform; the information acquisition module is used to collect operating data, environmental data and image information of equipment in the electrolysis plant in real time; the digital twin platform is used to construct a three-dimensional virtual model of the electrolysis plant, and dynamically update the model based on real-time data, and generate early warning strategies and production optimization strategies through simulation analysis; the monitoring and early warning module is used to output alarm signals to management personnel according to the early warning strategy; the production optimization module is used to output process parameter optimization suggestions to management personnel according to the optimization strategy.
[0024] The present invention is based on a digital twin system of a rare earth metal electrolysis plant based on big data. Based on the massive historical data and real-time data in the plant, it provides a data-driven process optimization solution through big data analysis, which significantly improves the electrolysis efficiency. The electrolysis efficiency and equipment failures are predicted by a neural network model to ensure the efficient reaction of electrolysis and the stability of the production process. Through real-time monitoring and early warning mechanisms, the system can predict equipment failures in advance and reduce unplanned downtime. Through the digital twin platform and big data analysis, the system can provide managers with real-time production optimization suggestions and decision support, and improve the intelligent level of production management. Driven by three-dimensional twin scenarios and real-time data, the system can dynamically update the model and monitor the production process in real time, provide an intuitive visual interface, and improve the flexibility and response speed of production. This solves the problem of insufficient optimization of production parameters and insufficient fault prediction caused by the failure of existing systems to effectively manage and utilize massive data, and improves the level of intelligent management of the factory. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0026] Figure 1 This is a schematic diagram of the digital twin system of a rare earth metal electrolysis plant based on big data provided by the present invention.
[0027] Figure 2 It is a schematic diagram of the information collection module.
[0028] Figure 3 It is a schematic diagram of the digital twin platform.
[0029] Figure 4 is a schematic diagram of a data processing unit.
[0030] Figure 5 This is a schematic diagram of the big data system.
[0031] Figure 6 and Figure 7 It is a flowchart for establishing a prediction model.
[0032] Figure 8 This is the principle diagram of BP neural network.
[0033] In the figure: 1-information acquisition module, 2-digital twin platform, 3-monitoring and early warning module, 4-production optimization module, 11-operation data acquisition unit, 12-environmental data acquisition unit, 13-image data acquisition unit, 21-data processing unit, 22-neural network model and optimization algorithm unit, 211-data acquisition and access subunit, 212-data storage and management subunit, 213-data analysis and modeling subunit, 214-visualization and interaction subunit. DETAILED DESCRIPTION
[0034] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.
[0035] See also Figures 1 to 8The present invention provides a digital twin system of a rare earth metal electrolysis plant based on big data. The present invention provides a digital twin system of a rare earth metal electrolysis plant based on big data, comprising an information acquisition module 1, a digital twin platform 2, a monitoring and early warning module 3, and a production optimization module 4. The information acquisition module 1 is connected to the digital twin platform 2, and the monitoring and early warning module 3 and the production optimization module 4 are respectively connected to the digital twin platform 2;
[0036] The information acquisition module 1 is used to collect the operating data, environmental data and image information of the equipment in the electrolysis plant in real time;
[0037] The digital twin platform 2 is used to build a three-dimensional virtual model of the electrolysis plant, dynamically update the model based on real-time data, and generate early warning strategies and production optimization strategies through simulation analysis;
[0038] The monitoring and warning module 3 is used to output an alarm signal to the management personnel according to the warning strategy;
[0039] The production optimization module 4 is used to output process parameter optimization suggestions to management personnel based on the optimization strategy.
[0040] In this embodiment, the information acquisition module 1 collects operational and image information corresponding to the target loading and unloading equipment. The digital twin platform 2 constructs a digital model of the target loading and unloading equipment and, based on the digital model, analyzes and processes the operational information and field-related parameters of the target equipment. Early warning strategies and production optimization strategies are formulated based on the analysis and processing results. The monitoring and early warning module 3 responds to the early warning strategy and outputs an alarm signal to management personnel. The production optimization module 4 responds to the optimization strategy and outputs an optimization signal to management personnel. Through big data analysis and optimization algorithms, the present invention enables the system to optimize process parameters based on historical and real-time data, providing data-driven process optimization solutions and significantly improving electrolysis efficiency. Real-time monitoring and optimization of the production process reduces human intervention and operational errors, improving the automation level and operational efficiency of the production line. Using a neural network model to predict electrolysis efficiency and equipment failures, the system can issue early warnings before anomalies occur, ensuring the stability of the production process. Through real-time monitoring and early warning mechanisms, the system can predict equipment failures in advance, reducing unplanned downtime. Through the digital twin platform 2 and big data analysis, the system can provide management personnel with real-time production optimization suggestions and decision support, enhancing the intelligent level of production management. Driven by 3D twin scenarios and real-time data, the system can dynamically update models and monitor the production process in real time, providing an intuitive visual interface that improves production flexibility and responsiveness. This addresses the existing system's inability to effectively manage and utilize massive amounts of data, leading to insufficient optimization of production parameters and inadequate fault prediction.
[0041] Furthermore, the information acquisition module 1 includes an operation data acquisition unit 11, an environment data acquisition unit 12 and an image data acquisition unit 13;
[0042] The operation data collection unit 11 is used to collect the operation data of the factory equipment;
[0043] The environmental data collection unit 12 is used to collect environmental data of factory equipment;
[0044] The image data acquisition unit 13 is used to acquire image data of factory equipment.
[0045] In this embodiment, the operation data acquisition unit 11 acquires operation data of factory equipment; the environment data acquisition unit 12 acquires environment data of factory equipment; and the image data acquisition unit 13 acquires image data of factory equipment.
[0046] Furthermore, the digital twin platform 2 includes a data processing unit 21 and a neural network model and optimization algorithm unit 22;
[0047] The data processing unit 21 is used to manage the entire process of the collected process parameters, equipment status and energy consumption data;
[0048] The neural network model and optimization algorithm unit 22 trains the neural network model based on historical data to predict electrolysis efficiency and equipment failure, and combines the genetic algorithm to perform global optimization on the electrolysis parameters.
[0049] In this embodiment, data collection and access collect temperature, current, voltage and other data in real time through the sensor network, and use MQTT to access the system to ensure the real-time and reliability of the data. Use distributed storage Hadoop HDFS and time series database InfluxDB to store massive data, and use the ETL tool Apache NiFi to clean and preprocess the data. Use the stream processing framework Apache Flink for real-time analysis and monitoring of anomalies. Train neural network models based on historical data to predict electrolysis efficiency and equipment failures; use optimization algorithms such as genetic algorithms to optimize electrolysis parameters. Display analysis results through the visualization tool Unity3D to support user interaction. Through big data technology, the rare earth molten salt electrolysis twin system can achieve efficient data management and intelligent analysis, providing support for process optimization and decision-making.
[0050] Furthermore, the data processing unit 21 includes a data acquisition and access subunit 211, a data storage and management subunit 212, a data analysis and modeling subunit 213, and a visualization and interaction subunit 214;
[0051] The data acquisition and access subunit 211 collects temperature, current, and voltage data in real time through the sensor network and transmits them to the system through the MQTT protocol;
[0052] The data storage and management subunit 212 uses Hadoop HDFS distributed storage and InfluxDB time series database to store massive data, and performs data cleaning and preprocessing through Apache NiFi;
[0053] The data analysis and modeling subunit 213 uses Apache Flink to process real-time stream data and combines it with a neural network model to predict equipment failures and electrolysis efficiency;
[0054] The visualization and interaction subunit 214 displays real-time analysis results through Unity3D, and supports user perspective switching, data filtering and alarm prompts.
[0055] In this embodiment, the neural network is trained using the training set data, and the Adam optimizer is selected as the optimization algorithm. During the training process, the model performance is verified by the test set to ensure that the model is not overfitting, thereby obtaining a neural network model with good generalization ability. The electrolysis parameters are optimized using a genetic algorithm. The optimization goal is to maximize the electrolysis efficiency (η) while minimizing the probability of equipment failure (F). The decision variables include current density (I), electrolysis temperature (T), molten salt composition (C), etc. The constraints are as follows:
[0056] Current density range: I min ≤I≤I max
[0057] Electrolysis temperature range: T min ≤T≤T max
[0058] Molten salt composition range: C min ≤C≤C max
[0059] Initialize the population: randomly generate a set of electrolysis parameter combinations (I, T, C).
[0060] Calculate fitness: Use the trained neural network to predict the electrolysis efficiency (η) and equipment failure probability (F) for each set of parameters. The fitness function is defined as:
[0061] Fitness = η - αF (where α is the fault penalty coefficient)
[0062] After selection, crossover, mutation, and iteration, the optimal electrolysis parameter combination (I*, T*, C*) and its corresponding electrolysis efficiency and failure probability are finally output.
[0063] Through experiments, we verify the actual effect of the optimal parameter combination, whether the electrolysis efficiency is close to the predicted value, and ensure that the failure probability is within an acceptable range.
[0064] Integrate the trained neural network model and optimization algorithm into the twin system to achieve real-time prediction and optimization.
[0065] Furthermore, the neural network model includes an input layer, a hidden layer and an output layer. The input layer is used to receive current density, electrolysis temperature, molten salt composition, electrolytic cell voltage and electrolysis time parameters; the hidden layer is used to set 2-3 layers of hidden layers that dynamically adjust the number of neurons, using a ReLU activation function; and the output layer is used to output the electrolysis efficiency prediction value and the equipment fault binary classification mark.
[0066] In this embodiment, input layer: input parameters include current density (I), electrolysis temperature (T), molten salt composition (such as rare earth oxide concentration C), electrolytic cell voltage (V) and electrolysis time (t). Hidden layer: set 2-3 hidden layers, the number of neurons in each layer is dynamically adjusted according to the complexity of the data, and the activation function uses ReLU to improve the nonlinear fitting ability of the model. Output layer: electrolysis efficiency prediction: 1 neuron, using a linear activation function. Equipment failure prediction: 1 neuron, represented by a binary classification (equipment failure flag F, 0 indicates normal, 1 indicates failure)
[0067] Beneficial effects:
[0068] 1. Improve electrolysis efficiency: Through big data analysis and optimization algorithms, the system can optimize process parameters based on historical data and real-time data, provide data-driven process optimization solutions, and significantly improve electrolysis efficiency.
[0069] 2. Improve production efficiency: Real-time monitoring and optimization of the production process reduces human intervention and operational errors, and improves the automation level and operating efficiency of the production line.
[0070] 3. Improve product quality stability: By using neural network models to predict electrolysis efficiency and equipment failures, the system can issue early warnings before abnormalities occur, ensuring the stability of the production process.
[0071] 4. Reduce the risk of equipment failure: Through real-time monitoring and early warning mechanisms, the system can predict equipment failures in advance and reduce unplanned downtime.
[0072] 5. Provide intelligent decision-making support: Through the digital twin platform and big data analysis, the system can provide managers with real-time production optimization suggestions and decision-making support, thereby improving the level of intelligent production management.
[0073] 6. Realize real-time monitoring and dynamic optimization: Driven by 3D twin scenarios and real-time data, the system can dynamically update the model and monitor the production process in real time, providing an intuitive visual interface and improving production flexibility and response speed.
[0074] The above disclosure is only a preferred embodiment of the digital twin system of the rare earth metal electrolysis plant based on big data of the present invention. Of course, this cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.
Claims
1. The digital twin system of rare earth metal electrolysis plant based on big data is characterized by: It includes an information acquisition module, a digital twin platform, a monitoring and early warning module, and a production optimization module. The information acquisition module is connected to the digital twin platform, and the monitoring and early warning module and the production optimization module are respectively connected to the digital twin platform. The information acquisition module is used to collect operating data, environmental data and image information of equipment in the electrolysis plant in real time; The digital twin platform is used to build a three-dimensional virtual model of the electrolysis plant, dynamically update the model based on real-time data, and generate early warning strategies and production optimization strategies through simulation analysis; The monitoring and warning module is used to output an alarm signal to the management personnel according to the warning strategy; The production optimization module is used to output process parameter optimization suggestions to management personnel based on the optimization strategy.
2. The digital twin system of a rare earth metal electrolysis plant based on big data according to claim 1, characterized in that: The information acquisition module includes an operation data acquisition unit, an environment data acquisition unit and an image data acquisition unit; The operation data collection unit is used to collect operation data of factory equipment; The environmental data acquisition unit is used to collect environmental data of factory equipment; The image data acquisition unit is used to acquire image data of factory equipment.
3. The digital twin system of a rare earth metal electrolysis plant based on big data according to claim 1, characterized in that: The digital twin platform includes a data processing unit and a neural network model and optimization algorithm unit; The data processing unit is used to manage the entire process of the collected process parameters, equipment status and energy consumption data; The neural network model and optimization algorithm unit trains the neural network model based on historical data to predict electrolysis efficiency and equipment failure, and combines the genetic algorithm to globally optimize the electrolysis parameters.
4. The digital twin system of a rare earth metal electrolysis plant based on big data according to claim 3, characterized in that: The data processing unit includes a data acquisition and access subunit, a data storage and management subunit, a data analysis and modeling subunit, and a visualization and interaction subunit; The data acquisition and access subunit collects temperature, current, and voltage data in real time through the sensor network and transmits them to the system through the MQTT protocol; The data storage and management subunit uses Hadoop HDFS distributed storage and InfluxDB time series database to store massive data, and uses Apache NiFi to clean and preprocess data; The data analysis and modeling subunit uses Apache Flink to process real-time stream data and combines it with a neural network model to predict equipment failures and electrolysis efficiency. The visualization and interaction subunit displays real-time analysis results through Unity3D and supports user perspective switching, data filtering and alarm prompts.
5. The digital twin system of a rare earth metal electrolysis plant based on big data according to claim 4, characterized in that: The neural network model includes an input layer, a hidden layer and an output layer. The input layer is used to receive current density, electrolysis temperature, molten salt composition, electrolytic cell voltage and electrolysis time parameters; the hidden layer is used to set 2-3 layers of hidden layers that dynamically adjust the number of neurons, using a ReLU activation function; and the output layer is used to output the electrolysis efficiency prediction value and the equipment fault binary classification mark.