Rare earth molten salt electrolysis process parameter optimization method and system based on big data analysis

Abstract

The invention relates to the technical field of rare earth metallurgy, in particular to a rare earth molten salt electrolysis process parameter optimization method and system based on big data analysis, a sensing-analysis-decision-execution full-process closed-loop control system is constructed, and the system comprises a data acquisition layer, an edge calculation layer, a cloud platform layer, an algorithm layer and an execution layer. Wherein the data acquisition layer acquires data in a rare earth molten salt electrolytic bath, the data is transmitted through the cloud platform layer after being preprocessed in the edge calculation layer, prediction optimization application is executed in the algorithm layer, and the execution layer converts an optimization instruction into equipment action and feeds back a result; according to the method, by collecting multi-dimensional data of the electrolysis process in real time, combining edge calculation and cloud collaborative analysis, dynamically optimizing key process parameters, and particularly through multi-source data fusion and intelligent algorithm optimization, the problems that multi-parameter dynamic collaborative regulation lags behind, the abnormal working condition response is slow (more than 5 minutes) and global energy efficiency optimization is insufficient are solved; and self-adaptive accurate control of process parameters is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of rare earth metallurgy, and in particular to a method and system for optimizing rare earth molten salt electrolysis process parameters based on big data analysis. Background Art

[0002] The molten salt electrolysis preparation of rare earth metals is a core process in the field of rare earth metallurgy, which obtains high-purity metals by electrolyzing a molten salt system. In the current process, key parameters such as electrolysis temperature, current density, and electrolyte composition mainly rely on manual experience to set, and frequent adjustments are required during the production process to maintain stability. However, due to the complex dynamic relationship between multiple parameters in the electrolytic cell (such as the thermal-electric coupling effect of temperature and current density), traditional single-variable control methods are difficult to achieve precise coordinated regulation, which often leads to large fluctuations in cell voltage, high energy consumption, and significant impact of process fluctuations on metal purity.

[0003] Although some existing research attempts to optimize parameters through numerical simulation, such methods rely on offline calculations and lack real-time data feedback, making them unable to cope with dynamic operating conditions such as changes in molten salt composition and electrode loss during the production process. Furthermore, the value of process data is not fully exploited, and massive operating data is only used for post-analysis, failing to form real-time decision support. These issues directly restrict the improvement of current efficiency (generally below 85%) and the further reduction of energy consumption (power consumption per ton of product >12,000kWh). Summary of the Invention

[0004] The purpose of the present invention is to provide a rare earth molten salt electrolysis process parameter optimization method and system based on big data analysis, aiming to solve the technical problems of traditional rare earth molten salt electrolysis process, which relies on manual experience to set parameters and is difficult to coordinate the coupling relationship of multiple parameters such as current density, temperature and electrolyte composition in real time, resulting in high energy consumption, low current efficiency and large fluctuations in metal purity.

[0005] To achieve the above objectives, the present invention provides a rare earth molten salt electrolysis process parameter optimization system based on big data analysis, including a data acquisition layer, an edge computing layer, a cloud platform layer, an algorithm layer, and an execution layer. The five layers are divided into three layers and work together. The data acquisition layer collects data from the rare earth molten salt electrolysis cell, and transmits it through the cloud platform layer after preprocessing at the edge computing layer. The algorithm layer executes the prediction optimization application. The execution layer converts the optimization instructions into equipment actions and feeds back the results, completing the closed-loop optimization from data acquisition to execution control.

[0006] The data acquisition layer deploys a sensor network at key locations of the electrolytic cell, including temperature monitoring, electrical parameter monitoring, and gas analysis. Temperature monitoring is achieved by arranging a K-type thermocouple at the bottom of the cell, an infrared thermometer in the middle of the molten salt, and a thermocouple array on the surface of the electrolytic cell; electrical parameter monitoring is achieved by installing a Hall current sensor on the anode busbar and a reference electrode voltage acquisition module on the cathode side wall; gas analysis is completed by installing an online mass spectrometer at the exhaust port on the top of the cell.

[0007] Among them, the edge computing layer is deployed with an industrial computer and a built-in real-time data processing engine, which is responsible for data cleaning, feature extraction and anomaly detection.

[0008] Among them, the cloud platform layer includes a three-dimensional simulation model of the electrolytic cell, a dynamic database and a process knowledge graph. The three-dimensional simulation model of the electrolytic cell is used to couple the electromagnetic field, thermal field and material transfer equations to simulate the molten salt flow and metal deposition process under different parameters; the dynamic database is responsible for storing real-time process data, optimization records and equipment status information, and supports time series data retrieval and comparative analysis; the process knowledge graph is responsible for integrating historical production data, literature rules and expert experience.

[0009] The algorithm layer adopts a hybrid intelligent optimization model, including an LSTM efficiency prediction model and an adaptive particle swarm optimization algorithm;

[0010] The LSTM efficiency prediction model includes an input layer, an LSTM layer, and a fully connected layer, wherein the input layer receives 8-dimensional process parameters; the number of hidden units in the LSTM layer is 64, and the time step is consistent with the data acquisition window; the fully connected layer includes two linear transformation layers and a ReLU activation function;

[0011] The adaptive particle swarm optimization algorithm introduces a dynamic adjustment mechanism of inertia weight based on the standard particle swarm optimization, that is, the initial value is 0.9, which decreases linearly to 0.4 with the number of iterations. The objective function is:

[0012]

[0013] where η Predicted The current efficiency predicted by the LSTM model, E current is the current energy consumption, E base is the benchmark value.

[0014] Among them, the execution layer is responsible for optimizing instruction conversion, specifically adjusting current density through thyristor rectifiers, controlling feeding through feeding valves, and implementing temperature closed-loop control using fuzzy PID algorithm.

[0015] The present invention also proposes a rare earth molten salt electrolysis process parameter optimization method based on big data analysis, using the rare earth molten salt electrolysis process parameter optimization system based on big data analysis, comprising the following steps:

[0016] Step 1: Real-time data collection;

[0017] Step 2: Edge computing feature extraction;

[0018] Step 3: LSTM efficiency prediction;

[0019] Step 4: Adaptive particle swarm optimization algorithm for multi-objective optimization;

[0020] Step 5: Closed-loop execution.

[0021] The present invention provides a method and system for optimizing the process parameters of rare earth molten salt electrolysis based on big data analysis, and constructs a full-process closed-loop control system of "perception-analysis-decision-execution". The system includes a data acquisition layer, an edge computing layer, a cloud platform layer, an algorithm layer and an execution layer. The data acquisition layer collects data in the rare earth molten salt electrolysis cell, transmits it through the cloud platform layer after preprocessing in the edge computing layer, executes predictive optimization applications in the algorithm layer, and the execution layer converts the optimization instructions into equipment actions and feeds back the results. The present invention dynamically optimizes key process parameters by real-time collection of multi-dimensional data of the electrolysis process, combining edge computing with cloud-based collaborative analysis. Specifically, through multi-source data fusion and intelligent algorithm optimization, it solves the problems of delayed dynamic collaborative regulation of multiple parameters, slow response to abnormal working conditions (>5 minutes) and insufficient global energy efficiency optimization, and realizes adaptive and accurate control of process parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] 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.

[0023] Figure 1 This is a schematic diagram of the architectural principle of a rare earth molten salt electrolysis process parameter optimization system based on big data analysis of the present invention.

[0024] Figure 2 This is a schematic diagram of the framework data flow in a rare earth molten salt electrolysis process parameter optimization system based on big data analysis of the present invention.

[0025] Figure 3 It is a schematic diagram of the electrolytic cell sensor layout of the present invention.

[0026] Figure 4It is a specific flow chart of a method for optimizing rare earth molten salt electrolysis process parameters based on big data analysis of the present invention.

[0027] Figure 5 It is a schematic diagram of the optimized process of the NdFeB alloy electrolytic cell in the specific embodiment 1 of the present invention.

[0028] Figure 6 This is a schematic diagram of abnormal temperature response to a thermocouple failure in specific embodiment 2 of the present invention.

[0029] Figure 7 It is a schematic diagram comparing the performance improvement after the implementation of the present invention.

[0030] Figure 8 It is a schematic diagram of the abnormal response process of the present invention. DETAILED DESCRIPTION

[0031] 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.

[0032] The present invention provides a rare earth molten salt electrolysis process parameter optimization system based on big data analysis, including a data acquisition layer, an edge computing layer, a cloud platform layer, an algorithm layer and an execution layer. The five-layer structure is divided into a data acquisition layer, an edge computing layer, a cloud platform layer, an algorithm layer and an execution layer. The five-layer structure performs respective tasks and works in collaboration. The data acquisition layer collects data in the rare earth molten salt electrolysis cell, transmits it through the cloud platform layer after pre-processing at the edge computing layer, executes the predictive optimization application at the algorithm layer, and the execution layer converts the optimization instructions into equipment actions and feeds back the results, completing the closed-loop optimization from data acquisition to execution control.

[0033] See also Figure 1 and Figure 2 , Figure 1 The principle architecture of the rare earth molten salt electrolysis process parameter optimization system based on big data analysis described in the present invention is shown in FIG. Figure 2 The following is a diagram of the execution process of data flow at each level. The following is a further explanation of the content and execution process of each part:

[0034] 1. Data Collection Layer

[0035] Deploy high-precision sensor networks at key locations in the electrolyzer (such as Figure 3 shown):

[0036] Temperature monitoring: T1: tank bottom (K-type thermocouple, range 800-1200°C, ±1.5°C), T2: middle of molten salt (infrared thermometer, ±2°C), T3: upper surface (thermocouple).

[0037] Electrical parameter monitoring: A Hall current sensor (0-10kA, 0.5 level accuracy) is installed on the anode, and a reference electrode voltage acquisition module is configured on the cathode.

[0038] Gas analysis: An online mass spectrometer (detecting CO2 and CO concentrations with a resolution of 1 ppm) is installed at the tank's top exhaust port, sampling every 30 seconds. Sensor data is transmitted to the edge layer via Industrial Ethernet, using Modbus TCP / IP as the communication protocol, ensuring a 1-second data update frequency.

[0039] 2. Edge computing layer

[0040] Deploy high-performance industrial computers with built-in real-time data processing engines:

[0041] Data cleaning: Wavelet threshold denoising is used (wavelet threshold denoising is a noise reduction method based on signal decomposition that retains valid signals by filtering out high-frequency noise) to eliminate electromagnetic interference and random noise.

[0042] Feature extraction: Synchronously calculate time domain statistics (mean, variance, kurtosis), frequency domain features (FFT dominant frequency, harmonic distortion rate) and time-frequency domain indicators (wavelet packet energy entropy).

[0043] Anomaly detection: Based on the 3σ criterion and sliding window algorithm (window length 60 seconds), it can identify abnormal operating conditions such as sudden temperature rise and current fluctuation in real time and trigger early warning signals.

[0044] 3. Cloud platform layer

[0045] Building a digital twin system for process optimization:

[0046] Multi-physics model: A three-dimensional simulation model of the electrolytic cell is established based on COMSOL Multiphysics. The electromagnetic field (Maxwell's equations), thermal field (Fourier's law), and material transport equations are coupled to simulate the molten salt flow and metal deposition processes under different parameters.

[0047] Process knowledge graph: Integrates historical production data (more than 100,000 sets), literature rules (such as "for every 1% decrease in LiF concentration, current efficiency decreases by 0.8%), and expert experience to form a decision-making network with more than 2,000 nodes.

[0048] Dynamic database: stores real-time process data, optimization records, and equipment status information, and supports time series data retrieval and comparative analysis.

[0049] 4. Algorithm layer

[0050] Adopting a hybrid intelligent optimization model, including:

[0051] LSTM efficiency prediction model: The input layer receives 8-dimensional parameters (temperature gradient, current density, LiF concentration, CO2 concentration, etc.), extracts time series features through a three-layer LSTM network (64 neurons per layer, dropout rate 0.2), and outputs the current efficiency prediction value (η_predicted). The model is trained using the Adam optimizer (learning rate 0.001), with a prediction error of ≤1.5%.

[0052] The LSTM efficiency prediction model includes an input layer, an LSTM layer, and a fully connected layer, wherein:

[0053] The input layer receives 8-dimensional process parameters;

[0054] The number of hidden units in the LSTM layer is 64, and the time step is consistent with the data acquisition window;

[0055] The fully connected layer contains two linear transformation layers and a ReLU activation function, and finally outputs the current efficiency prediction value.

[0056] By capturing the temporal correlation of process parameters through the LSTM layer and combining it with a fully connected layer to implement nonlinear mapping, the prediction accuracy is improved by 15% to 20% compared to traditional BP neural networks. The LSTM model is deployed on the edge computing layer (Advantech UNO-2484G), with inference latency ≤ 50ms, meeting real-time optimization requirements.

[0057] Adaptive Particle Swarm Optimization (APSO): Based on the standard PSO, a dynamic adjustment mechanism of inertia weight is introduced (the initial value is 0.9, which decreases linearly to 0.4 with the number of iterations). The objective function is:

[0058]

[0059] where η Predicted The current efficiency predicted by the LSTM model, E current is the current energy consumption, E base is the benchmark value.

[0060] 5. Execution Layer

[0061] Convert optimization instructions into device actions:

[0062] Current density regulation: DC power output is adjusted by thyristor rectifier (response time ≤ 10ms), with a control accuracy of ±0.5A / dm 2 .

[0063] Feeding control: A stepper motor drives a screw feeder (accuracy ±0.1g) to dynamically adjust the feed ratio of rare earth oxide to fluoride salt according to the predicted LiF concentration.

[0064] Temperature closed-loop control: Fuzzy PID algorithm is used to maintain the tank temperature within the set value ±10℃ by adjusting the heater power and cooling fan speed.

[0065] The present invention also proposes a rare earth molten salt electrolysis process parameter optimization method based on big data analysis, using the rare earth molten salt electrolysis process parameter optimization system based on big data analysis, comprising the following steps:

[0066] Step 1: Real-time data collection;

[0067] Step 2: Edge computing feature extraction;

[0068] Step 3: LSTM efficiency prediction;

[0069] Step 4: Adaptive particle swarm optimization algorithm for multi-objective optimization;

[0070] Step 5: Closed-loop execution.

[0071] Among them, the implementation process of the adaptive particle swarm optimization algorithm (APSO) includes the following steps:

[0072] (1) Algorithm initialization:

[0073] Particle swarm size: 50 particles

[0074] Parameter search range:

[0075] Current density: [0.5, 1.2] A / dm2

[0076] Temperature setting value: [950,1100]℃

[0077] LiF feeding frequency: [1,5] times / minute

[0078] (2) The fitness value of each particle is calculated by the following formula:

[0079]

[0080] in:

[0081] η Predicted Current efficiency predicted for the LSTM model

[0082] E current is the energy consumption under the current parameters

[0083] E base Baseline energy consumption (historical average)

[0084] Dynamic inertia weight adjustment

[0085] (3) The inertia weight ω decreases linearly with the number of iterations:

[0086]

[0087] Where iter is the current number of iterations, and max-iter is the total number of iterations (the default is 100).

[0088] (4) Algorithm termination condition:

[0089] Reached the maximum number of iterations (100 times)

[0090] The fitness value change rate for 10 consecutive iterations is less than 1%

[0091] Specifically, such as Figure 4 As shown in the figure, the system operation process includes the following key links:

[0092] 1. Real-time data collection and preprocessing:

[0093] The sensor network collects raw data at a period of 1 second. After denoising at the edge layer, 15 key features (such as temperature mean, current harmonic distortion rate, wavelet packet energy entropy, etc.) are extracted.

[0094] When data is abnormal (such as a sudden temperature change exceeding ±20°C), the redundant sensor switching mechanism is immediately activated to ensure data continuity.

[0095] 2. Efficiency prediction and parameter optimization:

[0096] The LSTM model receives feature data every 5 minutes and predicts the current efficiency trend for the next 15 minutes. If the predicted value falls below a threshold (e.g., η_predicted < 88%), the optimization process is triggered.

[0097] APSO algorithm is used under the constraint condition (current density 0.5~1.2A / dm 2 , temperature 950~1100℃) to search for the optimal solution, and the calculation time is controlled within 30 seconds.

[0098] 3. Exception handling and security assurance

[0099] Fault-tolerance mechanism: Key sensors (such as tank bottom thermocouples) utilize a dual-redundancy design. If the primary sensor fails, it automatically switches to the backup device and generates an alarm. In the event of a communication interruption, the edge layer activates local caching mode, storing the last 10 minutes of data and automatically retransmitting it when the network is restored.

[0100] Emergency control strategy: When the CO2 concentration is detected to be above 5% (indicating anode effect), immediately reduce the current density to a safe value (0.6A / dm 2 ) and activate the emergency exhaust system. When the molten salt level exceeds the warning line (85% of the tank height), the automatic discharge program is triggered to prevent overflow accidents.

[0101] Furthermore, the present invention also proposes specific embodiments for auxiliary explanation:

[0102] The following are the environment setting parameters of the embodiment:

[0103] 1. Hardware configuration:

[0104] Data acquisition module (such as Figure 3 K-type thermocouple (accuracy ±1.5°C), Hall effect current sensor (accuracy 0.5 level)

[0105] Edge computing devices: Industrial computers

[0106] Communication network: Industrial Ethernet + 5G redundant transmission

[0107] 2. Software implementation:

[0108] Data preprocessing: using wavelet transform denoising algorithm

[0109] Feature engineering: extract time domain (mean, variance), frequency domain (FFT main frequency), and time-frequency domain (wavelet energy) features.

[0110] Model training: Using the PyTorch framework, the training data contains 100,000 sets of historical process data.

[0111] 3. Control logic:

[0112] When the tank temperature is detected to deviate from the set value by ±15°C, the emergency control program is started.

[0113] Adjust the current density according to the current efficiency calculated in real time, and adjust the formula:

[0114] ΔI=k×(η_target-η_current)×I_base

[0115] (where ΔI is the current density adjustment amount, k is the adjustment coefficient, η_target is the target current efficiency, η_current is the real-time current efficiency, and I_base is the base current value).

[0116] Example 1: Optimization of NdFeB Alloy Electrolyzer

[0117] Initial parameters: current density 0.8A / cm 2 , temperature 1050℃, LiF concentration 25%.

[0118] Optimization process, such as Figure 5 As shown:

[0119] 1. The system detects that the CO2 concentration has risen to 4.2% (normal range ≤ 3.5%).

[0120] 2. LSTM predicted current efficiency dropped to 88.5%.

[0121] 3. PSO algorithm calculates the optimal solution: current density is adjusted to 0.76A / cm 2 , feeding frequency increased by 15%.

[0122] result:

[0123] The current efficiency recovered to 91.2% and the power consumption decreased by 18%.

[0124] Metal composition analysis: Nd content 99.96%, C impurity ≤50ppm.

[0125] Example 2: Abnormal operating condition handling

[0126] Simulation scenario: A thermocouple failure causes abnormal temperature data.

[0127] After applying this system, the system successfully coped with thermocouple failures and electrolyte composition deviations, and the recovery time from abnormal working conditions was shortened from five minutes with traditional methods to within 30 seconds, avoiding interruptions such as Figure 6 shown.

[0128] System Response:

[0129] 1. The edge computing layer quickly filters noise to avoid false triggers, executes commands directly to the device, and reduces latency.

[0130] 2. Dynamic decision-making: Minor anomalies are directly triggered by the edge layer to adjust the local PID, with a response time of ≤10 seconds. Severe anomalies are affected by simulated faults in the cloud, and emergency plans are generated in combination with knowledge graphs, with a decision-making time of ≤20 seconds.

[0131] In summary, the present invention achieves the following specific effects through multi-source data fusion and intelligent optimization algorithm: Figure 7 As shown:

[0132] 1. Comprehensive performance improvement of rare earth molten salt electrolysis process:

[0133] Improved current efficiency:

[0134] After optimization, the current efficiency is increased from 85% to 87% of the traditional process to 90% to 93%, the electrolysis reaction is more complete, and the metal deposition rate is increased.

[0135] Typical case: After the application of a neodymium electrolytic cell, the current efficiency increased from 87.2% to 92.5%.

[0136] Significantly reduced energy consumption:

[0137] The DC power consumption per ton of product has been reduced by 15% to 20%, from an average of 12,500 kWh to 10,000 to 10,500 kWh.

[0138] Main reasons for energy saving: optimizing current density distribution and reducing ineffective heat loss.

[0139] Enhanced quality stability:

[0140] The standard deviation of rare earth metal purity has been reduced from ±0.15% to ±0.03%, and the C and O impurity content has been reduced by more than 50%.

[0141] The uniformity of metal composition is improved (e.g., the fluctuation of Nd content in NdPr alloy is ≤0.1%).

[0142] 2. Enhanced abnormal response capabilities:

[0143] The recognition time of abnormal working conditions such as anode effect and temperature runaway is ≤30 seconds (e.g. Figure 8 The system can identify and respond to events (such as electrolyte separation and sudden temperature changes) within 30 seconds, preventing accidents.

[0144] The system completes the entire process of "detection-prediction-optimization-execution" within 30 seconds, avoiding an efficiency drop of more than 5%.

[0145] The above disclosure is merely one or more preferred embodiments of the present invention, and certainly cannot be used to limit the scope of the present invention. A person skilled in the art 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 still fall within the scope of the invention.

Claims

1. A rare earth molten salt electrolysis process parameter optimization system based on big data analysis, characterized in that: The five layers consist of a data acquisition layer, an edge computing layer, a cloud platform layer, an algorithm layer, and an execution layer. Each layer performs its own tasks and works collaboratively. The data acquisition layer collects data from the rare earth molten salt electrolyzer, which is then pre-processed by the edge computing layer and transmitted via the cloud platform layer. The algorithm layer then executes predictive optimization applications. The execution layer converts optimization instructions into device actions and provides feedback, completing a closed-loop optimization from data acquisition to execution control. The data acquisition layer deploys a sensor network at key locations of the electrolytic cell, including temperature monitoring, electrical parameter monitoring, and gas analysis. Temperature monitoring is achieved by arranging a K-type thermocouple at the bottom of the cell, an infrared thermometer in the middle of the molten salt, and a thermocouple array on the surface of the electrolytic cell; electrical parameter monitoring is achieved by installing a Hall current sensor on the anode busbar and a reference electrode voltage acquisition module on the cathode side wall; gas analysis is completed by installing an online mass spectrometer at the exhaust port on the top of the cell.

2. The rare earth molten salt electrolysis process parameter optimization system based on big data analysis according to claim 1, characterized in that: The edge computing layer is deployed with industrial computers and a built-in real-time data processing engine, which is responsible for data cleaning, feature extraction and anomaly detection.

3. The rare earth molten salt electrolysis process parameter optimization system based on big data analysis according to claim 2, characterized in that: The cloud platform layer includes a three-dimensional simulation model of the electrolytic cell, a dynamic database, and a process knowledge graph. The three-dimensional simulation model of the electrolytic cell is used to couple electromagnetic fields, thermal fields, and material transport equations to simulate the molten salt flow and metal deposition process under different parameters. The dynamic database is responsible for storing real-time process data, optimization records and equipment status information, and supports time series data retrieval and comparative analysis; the process knowledge graph is responsible for integrating historical production data, literature rules and expert experience.

4. The rare earth molten salt electrolysis process parameter optimization system based on big data analysis according to claim 3, characterized in that: The algorithm layer adopts a hybrid intelligent optimization model, including an LSTM efficiency prediction model and an adaptive particle swarm optimization algorithm; The LSTM efficiency prediction model includes an input layer, an LSTM layer, and a fully connected layer, wherein the input layer receives 8-dimensional process parameters; the number of hidden units in the LSTM layer is 64, and the time step is consistent with the data acquisition window; the fully connected layer includes two linear transformation layers and a ReLU activation function; The adaptive particle swarm optimization algorithm introduces a dynamic adjustment mechanism of inertia weight based on the standard particle swarm optimization, that is, the initial value is 0.9, which decreases linearly to 0.4 with the number of iterations. The objective function is: where η Predicted The current efficiency predicted by the LSTM model, E current is the current energy consumption, E base is the benchmark value.

5. The rare earth molten salt electrolysis process parameter optimization system based on big data analysis according to claim 4, characterized in that: The execution layer is responsible for optimizing instruction conversion, specifically by adjusting current density through thyristor rectifiers, controlling feeding through feeding valves, and implementing temperature closed-loop control using fuzzy PID algorithm.

6. A method for optimizing rare earth molten salt electrolysis process parameters based on big data analysis, using the rare earth molten salt electrolysis process parameter optimization system based on big data analysis according to any one of claims 1 to 5, characterized in that: The following steps are involved: Step 1: Real-time data collection; Step 2: Edge computing feature extraction; Step 3: LSTM efficiency prediction; Step 4: Adaptive particle swarm optimization algorithm for multi-objective optimization; Step 5: Closed-loop execution.

Images