Industrial silicon intelligent power distribution and control system and method

Abstract

The invention discloses an industrial silicon intelligent power distribution and control system and method. The system comprises a data acquisition and processing module, a shared feature extraction module, an intelligent control module and a power distribution module which are connected in sequence. Multi-mode operation data is collected in real time, a shared feature extraction module is adopted to perform feature extraction on the operation data, data input is provided for subsystems of the intelligent control module, and information isolated island is avoided; through cooperative work of all the modules, global optimization control over the industrial silicon production process is achieved. The intelligent control module comprises an intelligent industrial silicon submerged arc furnace baking system, an automatic electrode lifting system, an automatic submerged arc furnace transformer gear adjusting system and an automatic electric energy quality adjusting system, and each system performs task prediction based on corresponding characteristics by adopting a pre-trained model and outputs optimized parameters; and the power distribution module generates a power distribution instruction based on the optimization parameters, so that intelligent distribution of power resources is realized, and dynamic adjustment and optimal distribution of energy consumption are realized.

Description

Technical Field

[0001] The present invention relates to the field of industrial production process control, and in particular to an industrial silicon intelligent power distribution and control system and method. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Industrial silicon is an important basic industrial material, widely used in fields such as semiconductors, solar photovoltaics, power electronics, and alloy materials. The production of industrial silicon is an energy-intensive process that requires a stable and substantial power supply. Submerged arc furnaces, as key equipment, require significant power resources to operate. Operating efficiency and energy consumption are directly related to production costs and economic benefits.

[0004] With the continuous development of industrial silicon production technology, the requirements for power distribution and control systems are also increasing. Traditional power distribution and control methods mainly rely on manual operation and empirical judgment. They cannot accurately adjust in real time according to production needs and the status of the submerged arc furnace. They lack real-time response capabilities and have adjustment lags, resulting in waste of power resources and low energy efficiency. They cannot meet the needs of modern industrial production for high precision, high efficiency, and high safety. Although some improved technologies have gradually been applied to industrial silicon production, such as power distribution systems, their data utilization rate is low and they do not fully utilize real-time operating data from the production process for dynamic optimization. Most of them use preset rules and fixed parameters based on experience, have poor adaptability to dynamic load changes, and cannot adapt to complex working conditions. In addition, existing methods usually only focus on a single link and fail to achieve global optimization of the entire production process. Summary of the Invention

[0005] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides an industrial silicon intelligent power distribution and control system and method, and designs an integrated industrial silicon intelligent power distribution and control system, which integrates data acquisition, processing, intelligent control, safety protection and human-computer interaction. Based on real-time operation data and intelligent optimization models, it realizes intelligent control of power distribution of industrial silicon ore-fired furnaces, ensures the rational allocation of power resources, and improves energy utilization efficiency.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0007] In a first aspect, the present invention provides an industrial silicon intelligent power distribution and control system, comprising a data acquisition and processing module, a shared feature extraction module, an intelligent control module, and a power distribution module connected in sequence;

[0008] The data acquisition and processing module is used to collect and preprocess the operating data of the industrial silicon ore furnace in real time, and transmit the preprocessed operating data to the shared feature extraction module;

[0009] The shared feature extraction module is used to extract features from the operating data, obtain corresponding features and transmit them to the intelligent control module;

[0010] The intelligent control module includes an intelligent baking system for an industrial silicon submerged arc furnace, an automatic electrode lifting system, an automatic gear adjustment system for the submerged arc furnace transformer, and an automatic power quality adjustment system. Each system uses a pre-trained model to perform task prediction based on corresponding features, outputs optimized parameters, and sends the optimized parameters to the power distribution module.

[0011] The power distribution module is used to generate power distribution instructions based on the optimization parameters to achieve intelligent distribution of power resources.

[0012] According to a further technical solution, the operating data include the temperature, pressure, current, voltage, power and power factor of the submerged arc furnace.

[0013] A further technical solution is to perform a stationarity test on the operating data, including graphical analysis and statistical testing. Specifically, by observing the graph of the operating data, a preliminary judgment is made as to whether it is stationary, and then a unit root test is performed on the operating data to determine whether it is stationary data. If it is non-stationary data, differential processing is performed.

[0014] In a further technical solution, the shared feature extraction network includes a feature extraction layer and a feature dimensionality reduction layer connected in sequence, and the feature extraction layer includes a long short-term memory network, a one-dimensional convolution layer and an embedding layer.

[0015] As a further technical solution, the automatic electrode lifting system adopts a differential autoregressive moving average model to predict the current value, takes the time series characteristics as input features, outputs the predicted value of the future set time, and obtains the real-time height adjustment amount of the electrode according to the predicted value.

[0016] A further technical solution is to use the differential autoregressive moving average model to predict the current value as follows:

[0017] Draw autocorrelation and partial autocorrelation plots of time series characteristics, and determine the number of autoregressive terms, difference orders, and moving average terms of the model;

[0018] Use statistical methods to estimate the parameters of the model, perform residual tests on the fitted model, and optimize the model based on the test results to obtain the optimized differential autoregressive moving average model;

[0019] The optimized differential autoregressive moving average model is used for prediction.

[0020] According to a further technical solution, the intelligent baking system of the industrial silicon ore-fired furnace and the automatic power quality adjustment system adopt a reinforcement learning model for dynamic adjustment.

[0021] In a second aspect, the present invention provides an industrial silicon intelligent power distribution and control method, comprising:

[0022] Real-time collection of industrial silicon submerged arc furnace operating data and pre-processing to obtain pre-processed operating data;

[0023] Performing feature extraction on the operating data to obtain corresponding features;

[0024] Input the corresponding features into the pre-trained model for task prediction to obtain the optimized parameters;

[0025] A power allocation instruction is generated based on the optimization parameters to achieve intelligent allocation of power resources.

[0026] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the industrial silicon intelligent power distribution and control method as described in the second aspect.

[0027] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the industrial silicon intelligent power distribution and control method as described in the second aspect are implemented.

[0028] One or more of the above technical solutions have the following beneficial effects:

[0029] The present invention collects multi-modal operating data in real time, uses a shared feature extraction module to extract features from the operating data, and provides data input for each subsystem of the intelligent control module to avoid information islands. Through the collaborative work of each module, global optimization control of the industrial silicon production process is achieved, and in the intelligent control module, each subsystem independently completes a specific task and outputs corresponding optimization parameters to achieve dynamic optimization of furnace power, electrode height, transformer gear and power quality, realize dynamic adjustment and optimal distribution of energy consumption, significantly reduce energy waste, and achieve efficient and accurate power distribution and equipment control.

[0030] On the one hand, the present invention achieves energy saving and consumption reduction. In the intelligent baking system of the industrial silicon ore-burning furnace, the baking curve is dynamically adjusted through reinforcement learning to minimize heat energy waste; the current is predicted in the automatic electrode lifting system to improve the efficiency of power utilization; the genetic algorithm is used in the automatic adjustment system of the transformer gear of the ore-burning furnace to match the load with the optimal gear and reduce reactive power consumption; on the other hand, the operating stability of the industrial silicon ore-burning furnace is improved. The intelligent baking system of the industrial silicon ore-burning furnace dynamically adjusts the heating power through reinforcement learning strategy to stably control the temperature; the time series prediction model (such as ARIMA) of the automatic electrode lifting system adjusts the electrode height in advance to reduce the impact of current fluctuations on the equipment; the automatic transformer gear adjustment system ensures stable power output to avoid grid shock or equipment overload due to gear mismatch.

[0031] The present invention also uses a deep Q learning model through the automatic power quality adjustment system to dynamically adjust the compensation equipment to ensure that the harmonic level meets the standard; by dynamically adjusting the filter, the impact of voltage fluctuations on equipment operation is avoided, thereby improving the overall power quality of the system.

[0032] The intelligent control module of the present invention is based on technologies such as reinforcement learning and deep Q learning. The system can respond to changes in operating status in real time and make optimal adjustments.

[0033] This invention incorporates multiple safety protection measures to ensure safe and stable production. It also offers an intuitive and user-friendly interface for ease of use and management. It has broad application prospects and will be able to meet the high-precision, high-efficiency, and high-security demands of modern industrial production. Furthermore, the system can be customized and expanded to meet the needs of enterprises of varying sizes and production requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0035] Figure 1 This is a flow chart of the industrial silicon intelligent power distribution and control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0037] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0038] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0039] Example 1

[0040] like Figure 1 As shown, this embodiment discloses an industrial silicon intelligent power distribution and control system, comprising a data acquisition and processing module, a shared feature extraction module, an intelligent control module and a power distribution module connected in sequence;

[0041] The data acquisition and processing module is used to collect and preprocess the operating data of the industrial silicon ore furnace in real time, and transmit the preprocessed operating data to the shared feature extraction module;

[0042] The shared feature extraction module is used to extract features from the operating data, obtain corresponding features and transmit them to the intelligent control module;

[0043] The intelligent control module includes an intelligent baking system for an industrial silicon submerged arc furnace, an automatic electrode lifting system, an automatic gear adjustment system for the submerged arc furnace transformer, and an automatic power quality adjustment system. Each system uses a pre-trained model to perform task prediction based on corresponding features, outputs optimized parameters, and sends the optimized parameters to the power distribution module.

[0044] The power distribution module is used to generate power distribution instructions based on the optimization parameters to achieve intelligent distribution of power resources.

[0045] In this embodiment, the data acquisition and processing module includes various high-precision sensors deployed around the industrial silicon submerged arc furnace to collect real-time operating data (key parameters) such as the furnace's temperature, pressure, current, voltage, power, and power factor. Specifically, this module collects: the furnace transformer's primary current and voltage; the furnace transformer's secondary output phase and line voltages; the furnace transformer's primary active and reactive power; the furnace transformer's primary power factor; the furnace transformer's primary active and reactive power; various furnace transformer fault signals, gear position display, oil temperature, furnace bottom temperature, cooling water status, three-phase electrode voltage, current, and power; three-phase electrode position and pressure release; and main flue gas temperature.

[0046] After preprocessing, the collected operational data is transferred to the shared feature extraction module for feature extraction. Preprocessing includes missing value processing, outlier processing, and data standardization. Specifically, missing data can be filled using methods such as interpolation and mean substitution to ensure data continuity. Outliers can be detected and processed using statistical methods to remove or correct outliers caused by sensor failures, data transmission errors, and other reasons to improve data accuracy. Based on the data's characteristics, the collected operational data is standardized to eliminate the impact of different dimensions and units and make the data comparable.

[0047] Perform stationarity tests on the time series data (voltage, current, power, power factor, furnace bottom temperature, etc.) in the operating data, including graphical analysis and statistical tests.

[0048] Further, graphical analysis: By observing the graph of the time series data in the running data, a preliminary judgment is made as to whether it is stationary. The graph of a stationary time series usually fluctuates around a constant level without obvious trends or periodicity. Statistical test: Use statistical test methods such as ADF (Augmented Dickey-Fuller) to perform a unit root test on the time series data to determine whether it is stationary data. If the data is non-stationary, differential processing is required. For non-stationary time series data, the differential method is used for stabilization. Differentiation can be understood as a subtraction operation on the time series to eliminate irregular changes such as trends and seasonality. The data after differential processing should meet the stationarity requirements.

[0049] In this embodiment, the shared feature extraction module extracts key features from multimodal data (collected operational data) and establishes a unified feature space to provide input for the subsequent intelligent control module. The shared feature extraction module includes a feature extraction layer and a feature dimensionality reduction layer connected in sequence. The feature extraction layer includes a long short-term memory network (LSTM), a one-dimensional convolutional layer (1D-CNN), and an embedding layer.

[0050] The specific steps to obtain the corresponding features are:

[0051] (1) Receiving input data: The received operating parameters include electrical parameters (continuous variables such as current, voltage, power factor, etc.), equipment status (discrete variables such as gear position, cooling water status, fault signal, etc.) and environmental data (continuous variables such as furnace bottom temperature and flue temperature).

[0052] Furthermore, discrete variables (such as fault signals) are one-hot encoded, and the received time series data is segmented according to fixed time windows.

[0053] (2) Feature extraction layer: Long short-term memory (LSTM) networks are used to extract temporal trend features from time series data (voltage, current, power factor, furnace bottom temperature, etc.), obtain historical context influences, and output the hidden state representation of the time series (i.e., time series features). Time series features include the mean, rate of change, slope, standard deviation, and difference between the maximum and minimum values ​​of parameters such as voltage, current, power factor, and furnace bottom temperature.

[0054] A one-dimensional convolutional neural network (1D-CNN) is used to extract local correlations from spatially correlated data (such as three-phase electrode voltage, current, and power), capture the correlations between the three phases, and output the convolved feature vector (i.e., spatial features). Spatial features include the symmetry of the three-phase voltage and current and the ratio of phase voltage to line voltage.

[0055] The embedding layer is used to extract features from equipment status data (such as cooling water status, gear position, and fault signals), and its discrete variables are mapped into continuous features (i.e., discrete variable embedding features).

[0056] (3) Feature dimensionality reduction layer: Use the autoencoder to learn the low-dimensional representation of the features and output the corresponding feature vector of fixed dimension to represent the current operating status of the submerged arc furnace.

[0057] In this embodiment, the intelligent control module includes an intelligent baking system for an industrial silicon submerged arc furnace, an automatic electrode lifting system, an automatic transformer gear adjustment system for the submerged arc furnace, and an automatic power quality control system. This module is the core of the industrial silicon power distribution and control system. By leveraging the collaborative operation of multiple subsystems, it performs predictions and optimizations based on their respective characteristics, providing real-time decision-making for each operational link.

[0058] The intelligent control module receives feature vectors from the shared feature extraction module, which contain the real-time operating status information of the submerged arc furnace. Each subsystem independently runs the prediction task based on the pre-trained model and outputs the optimization parameters of each subsystem, including the furnace curve, electrode height, gear adjustment parameters, reactive compensation amount, etc.; it also updates the models within each subsystem in real time and adjusts the control model parameters according to the collected data to achieve closed-loop optimization.

[0059] (1) Intelligent baking system for industrial silicon ore-fired furnace: Dynamically adjusts the baking curve to adapt to different raw materials and environmental conditions (such as seasonal temperature fluctuations) to avoid thermal stress damage and energy waste.

[0060] The oven process is modeled as a reinforcement learning problem. The goal is to dynamically adjust the oven heating power to keep the oven temperature within the target range and optimize the energy consumption. Specifically, state S t , including current furnace temperature, target temperature, heating rate, and historical furnace data; Action A t , heating power adjustment value; reward R t, measure the effect of the furnace curve. Environment to simulate the thermodynamic characteristics of industrial silicon submerged arc furnace.

[0061] Reward function R t Expressed as:

[0062] R t =-α(ΔT) 2 -βE t +γP m

[0063] Where ΔT is the deviation between the actual temperature and the target temperature, E t is the energy consumption per unit time, P m is the material protection coefficient (set according to the material stability), and α, β, and γ are weight parameters.

[0064] The Deep Deterministic Policy Gradient (DDPG) algorithm is used for continuous action optimization. When training the model, the simulation environment generates baking data, trains the intelligent agent, and optimizes the policy network to maximize the cumulative reward.

[0065] The intelligent baking system of the industrial silicon submerged arc furnace automatically adjusts various parameters in the baking process (adjusting baking conditions such as power and current, etc.) according to the dynamically adjusted baking curve and process parameters (such as electrode parameters, temperature parameters, pressure parameters and electrical parameters) to ensure the stability and controllability of the baking process. The intelligent baking system can realize the automation of the baking process, reduce manual intervention, improve production efficiency, and achieve seamless connection between the baking process and the smelting process by integrating with the control system of the industrial silicon furnace.

[0066] (2) Automatic electrode lifting system: Based on the prediction of the current in the furnace, the electrode height is adjusted in advance to reduce the risk of short circuit and electrode consumption, and improve the utilization rate of electricity.

[0067] Taking the time series features as input features, the ARIMA model is used to predict the current value and output the real-time height adjustment of the electrode. Specifically:

[0068] Autocorrelation and Partial Autocorrelation Analysis: Plotting the autocorrelation plot (ACF) and partial autocorrelation plot (PACF) of time series data allows you to determine the ARIMA model's orders p and q by analyzing their characteristics. The autocorrelation plot reflects the correlation between the time series data and its past values, while the partial autocorrelation plot reflects the correlation between the time series data and its past values ​​after excluding the influence of intermediate variables.

[0069] Model Construction: Based on the analysis of the autocorrelation and partial autocorrelation plots, as well as comparisons of statistical indicators such as the AIC / BIC criteria, the orders p, d, and q of the ARIMA model are determined. Here, p is the number of autoregressive terms, d is the differencing order (determined in the differencing process), and q is the number of moving average terms.

[0070] Parameter estimation: Use statistical methods such as least squares, maximum likelihood, or Bayesian estimation to estimate the parameters of the ARIMA model. These parameters include the autoregressive coefficient, moving average coefficient, and intercept term. The goal of parameter estimation is to minimize the model's forecast error.

[0071] Model diagnostics: Perform a residual test on the fitted ARIMA model to check whether the residuals are white noise sequences. A white noise sequence is one in which the residuals are independent and normally distributed. If the residuals do not meet these conditions, there may be a problem with the model, and you may need to readjust the model parameters or choose a different model.

[0072] Model optimization: Based on the results of model diagnostics, namely the residual test results, the ARIMA model is optimized. Model fitting and forecasting performance can be improved by adjusting model parameters and increasing or decreasing the number of model terms.

[0073] Model Prediction and Evaluation: The optimized ARIMA model is used to generate forecasts, outputting predicted values ​​for a set future time. The forecast results are evaluated, and statistical metrics such as prediction error are calculated to measure the model's predictive performance. These evaluation results provide strong support for optimized operation and decision-making in the industrial silicon intelligent power distribution system.

[0074] The electrode lifting amount is calculated based on the prediction results, and the real-time height adjustment amount of the electrode is output.

[0075] The automatic electrode lifting system predicts current and calculates the electrode lifting amount by integrating the coupling relationship between the three-phase electrode operating parameters. A controller is then developed to automatically control the three-phase electrode lifting and lowering in real time based on the received electrode lifting amount, optimizing and balancing the three-phase electrode power. The coupling relationship is the mutual influence between current and voltage and electrode lifting and lowering. Specifically, electrode lifting and lowering affect current and voltage: electrode lifting and lowering directly affect arc length, which in turn changes arc resistance. According to Ohm's law, changes in resistance lead to corresponding changes in current and voltage. When the electrode rises, the arc length increases, arc resistance increases, current decreases, and voltage increases. Conversely, when the electrode descends, arc length decreases, arc resistance decreases, current increases, and voltage decreases. Current and voltage feedback on electrode lifting and lowering: Changes in current and voltage are also fed back to the automatic electrode lifting and lowering system. For example, when the current or voltage is excessively high, the system may automatically adjust the electrode lifting and lowering position to maintain the current and voltage within an appropriate range to protect the electrodes and the furnace equipment.

[0076] In an industrial silicon submerged arc furnace, electrodes act as a carrier for converting electrical energy into thermal energy, with discharges at their lower ends melting the raw materials. As smelting progresses, the lower end of the electrodes wears out, shortening the length of the electrodes inserted beneath the material layer, which in turn affects the conductivity of the molten pool and smelting efficiency. Therefore, timely adjustment of the electrodes' actual position to maintain optimal conditions is crucial for maintaining a stable smelting environment within the furnace and improving product quality. The automatic electrode lifting system uses an optimized ARIMA model to predict current and controls electrode lifting based on this predicted current, preventing excessive or insufficient current from adversely affecting the smelting process.

[0077] (3) Automatic adjustment system for transformer gear of electric arc furnace: Genetic algorithm is used to dynamically adjust the gear of transformer according to current load and predicted demand to ensure stable power output and minimum energy consumption.

[0078] The specific steps of optimization using genetic algorithm are as follows:

[0079] 1) Define the optimization goal, specifically: dynamically select the optimal transformer gear according to the current load and predicted demand to ensure stable power output and minimum energy consumption. The optimization objective function is:

[0080] f(L)=-λ·|v|-θ·u-τ·w

[0081] Where f(L) is the fitness value of gear L, v is the load deviation (actual load minus the load supported by gear L), u is the power loss (energy consumption of transformer gear L), w is the switching cost (operating time and mechanical wear required for gear change), and λ, θ, and τ are weight parameters.

[0082] 2) Initialize the population: Each individual is a gear combination, L = [L1, L2, ..., L n ] represents the gear settings of multiple transformers. A certain number of gear combinations are randomly generated and coded individually.

[0083] 3) Calculate the fitness value f(L) for each individual.

[0084] 4) Roulette wheel is used to select individuals. Individuals are randomly selected according to the ratio of fitness value f(L). Individuals with higher fitness value have greater probability of being selected.

[0085] 5) Perform single-point crossover and mutation operations.

[0086] 6) Update the population: Select the top N best individuals according to the fitness value to enter the next generation; if the maximum number of iterations is reached or the fitness value converges, the algorithm is terminated.

[0087] Through genetic algorithms, the transformer gear automatic adjustment system can efficiently search for the global optimal gear combination, reduce power loss and switching cost while meeting load demand, and ultimately achieve the goal of stable and efficient power output.

[0088] The automatic transformer gear adjustment system works based on the transformer's voltage regulation principle. The transformer's primary side is connected to the power grid. Due to fluctuations in grid voltage or changes in load, the transformer's secondary voltage will also fluctuate accordingly. To maintain stable voltage within the furnace, the system monitors electrical parameters (such as load) in real time. Based on preset setpoints and actual production experience, it uses a genetic algorithm to automatically adjust the transformer's gear position, thereby changing the secondary voltage.

[0089] (4) Automatic power quality regulation system: achieves harmonic suppression, reactive power compensation and voltage fluctuation balance, and uses the deep Q learning (DQN) model to dynamically adjust compensation equipment and filters.

[0090] Modeling: The power quality regulation problem is modeled as a reinforcement learning problem, specifically: state S t ', including the current total harmonic distortion rate THD, power factor, voltage fluctuation, current configuration of compensation equipment and filter; Action A t ′, adjust the output of the compensation device and start / stop the filter (the action set includes specific device parameter adjustments, such as reactive power injection value or filter switch status); reward R t ′, rewards are given based on the improvement of power quality indicators. The formula is as follows:

[0091]

[0092] Among them, H t is the current total harmonic distortion rate, ΔP f is the power factor deviation, ΔV is the voltage fluctuation, 1. is the reward weight of power quality, is the harmonic suppression reward weight, l is the power factor reward weight, Reward weight for voltage fluctuation.

[0093] Deep Q-learning model: DQN uses a neural network to approximate the Q-value function and adjust action strategies in real time. It includes a main network that updates the Q-value, while the target network generates a stable target value.

[0094] Output: Based on the input characteristics, the output action corresponds to the specific adjustment value of the compensation device and the switching state of the filter.

[0095] Through the above steps, the system can dynamically adjust compensation equipment and filters in real time to improve power quality, suppress harmonics, optimize power factor, and balance voltage fluctuations.

[0096] The automatic power quality control system monitors the power quality of the power system in real time and takes appropriate measures, such as adjusting the input of filters and reactive power compensation equipment, to improve power quality and smelting efficiency. Power quality monitoring primarily covers voltage fluctuations and flicker, harmonics, and power factor. If the harmonics of the submerged arc furnace exceed the standard, the filter is adjusted to control the harmonics generated by the submerged arc furnace, thereby reducing harmonic pollution to the power grid. If the power factor of the submerged arc furnace is too low, the low-voltage reactive power compensation device automatically compensates the secondary side reactive power of the submerged arc furnace transformer, thereby improving power quality and smelting efficiency.

[0097] The intelligent control module is the core of this system. Based on advanced artificial intelligence technologies such as machine learning, reinforcement learning, and optimization algorithms, it analyzes and predicts features extracted from operational data, monitors the operating status of the submerged arc furnace in real time, and automatically adjusts and optimizes according to the requirements of the smelting process. Based on production needs and equipment status, the intelligent control module automatically adjusts power distribution to ensure optimal operation of the industrial silicon submerged arc furnace. Furthermore, through the coordinated collaboration of various subsystems, the stability and efficiency of the smelting process are ensured, energy consumption is reduced, product quality is improved, and the stable operation of the power grid is maintained.

[0098] In addition, the intelligent control module also has an alarm function, which will promptly issue an alarm and take measures when parameters are abnormal. Potential problems can be discovered in advance and appropriate measures can be taken to avoid accidents. Specifically, temperature detection is set in the submerged arc furnace to monitor the surface temperature, furnace bottom temperature, etc., and the temperature range under normal conditions is set. If the temperature exceeds or falls below the normal range, an alarm will be issued and appropriate measures will be taken. The voltage and current signals of the submerged arc furnace are monitored. If abnormal parameters such as voltage imbalance, large voltage fluctuation, and current instability occur (such as voltage too high, voltage too low, current imbalance, current fluctuation, etc.), an alarm will be issued and measures such as adjusting the transformer gear or electrode lifting will be taken.

[0099] In this embodiment, the power distribution module receives optimized parameters from the intelligent control module's subsystems, such as those for furnace optimization, electrode adjustment, transformer gear adjustment, and power quality control. Based on these optimized parameters, it generates specific control instructions. Based on these control instructions, the system intelligently allocates power resources. By optimizing power distribution, the system can reduce energy consumption and improve energy efficiency.

[0100] The intelligent power distribution and control system also includes a safety protection module, including intelligent circuit breakers and intelligent protection devices. These devices feature comprehensive safety protection mechanisms and can monitor equipment operating status in real time. Upon detecting an abnormality such as overcurrent, overvoltage, or overheating, the system immediately activates safety protection programs, shutting off power to the affected equipment and preventing accidents. Furthermore, intelligent circuit breakers and intelligent protection devices enable precise control and protection of circuits, further enhancing production safety.

[0101] A user-friendly human-machine interface is also provided, allowing operators to monitor the equipment's operating status and power distribution in real time. The interface is intuitive and easy to use, supporting multiple operation methods, such as touch screen and keyboard. Operators can manually intervene and make adjustments as needed, enhancing the system's flexibility and adaptability.

[0102] Example 2

[0103] This embodiment discloses an industrial silicon intelligent power distribution and control method, including:

[0104] Real-time collection of industrial silicon submerged arc furnace operating data and pre-processing to obtain pre-processed operating data;

[0105] Performing feature extraction on the operating data to obtain corresponding features;

[0106] Input the corresponding features into the pre-trained model for task prediction to obtain the optimized parameters;

[0107] A power allocation instruction is generated based on the optimization parameters to achieve intelligent allocation of power resources.

[0108] Example 3

[0109] The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of the second embodiment when executing the program.

[0110] Example 4

[0111] The purpose of this embodiment is to provide a computer-readable storage medium, a computer-readable storage medium having a computer program stored thereon, which executes the steps of the method of embodiment 2 when the program is executed by a processor.

[0112] The steps involved in the apparatuses of Examples 3 and 4 above correspond to those of Method Example 2. For detailed implementation, please refer to the relevant description of Example 2. The term "computer-readable storage medium" should be understood to mean a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and causing the processor to perform any method of the present invention.

[0113] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computer device. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0114] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0115] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. An industrial silicon intelligent power distribution and control system, characterized in that: It includes a data acquisition and processing module, a shared feature extraction module, an intelligent control module and a power distribution module which are connected in sequence; The data acquisition and processing module is used to collect the operating data of the industrial silicon ore furnace in real time and perform preprocessing, and transmit the preprocessed operating data to the shared feature extraction module; The shared feature extraction module is used to extract features from the operating data, obtain corresponding features and transmit them to the intelligent control module; The intelligent control module includes an industrial silicon ore-fired furnace intelligent baking system, an electrode automatic lifting system, an ore-fired furnace transformer gear automatic adjustment system, and an electric energy quality automatic adjustment system. Each system uses a pre-trained model to perform task prediction based on corresponding features, outputs optimization parameters, and sends the optimization parameters to the power distribution module; The power distribution module is used to generate power distribution instructions based on the optimization parameters to achieve intelligent distribution of power resources.

2. An industrial silicon intelligent power distribution and control system as claimed in claim 1, characterized in that: The operating data include the temperature, pressure, current, voltage, power and power factor of the ore-fired furnace.

3. An industrial silicon intelligent power distribution and control system as claimed in claim 2, characterized in that: The operation data is subjected to a stationarity test, including graphical analysis and statistical test, specifically: by observing the graph of the operation data, a preliminary judgment is made as to whether it is stationary, and then a unit root test is performed on the operation data to determine whether it is stationary data, and if it is non-stationary data, differential processing is performed.

4. The industrial silicon intelligent power distribution and control system according to claim 1, characterized in that: The shared feature extraction network includes a feature extraction layer and a feature dimension reduction layer connected in sequence, and the feature extraction layer includes a long short-term memory network, a one-dimensional convolution layer and an embedding layer.

5. The industrial silicon intelligent power distribution and control system according to claim 1, characterized in that: The automatic electrode lifting system adopts a differential autoregressive moving average model to predict the current value, takes the time series characteristics as input characteristics, outputs the predicted value of the future set time, and obtains the real-time height adjustment amount of the electrode according to the predicted value.

6. An industrial silicon intelligent power distribution and control system as claimed in claim 5, characterized in that: The current value predicted by the differential autoregressive moving average model is specifically: Draw autocorrelation and partial autocorrelation plots of time series characteristics, and determine the number of autoregressive terms, difference orders, and moving average terms of the model; Use statistical methods to estimate the parameters of the model, perform residual tests on the fitted model, and optimize the model based on the test results to obtain the optimized differential autoregressive moving average model; The optimized difference autoregressive moving average model is used for prediction.

7. The industrial silicon intelligent power distribution and control system according to claim 1, characterized in that: The industrial silicon ore-fired furnace intelligent baking system and the power quality automatic adjustment system adopt a reinforcement learning model for dynamic adjustment.

8. An industrial silicon intelligent power distribution and control method, characterized in that: include: Collect the operation data of industrial silicon ore-fired furnace in real time and pre-process it to obtain the pre-processed operation data; Performing feature extraction on the operation data to obtain corresponding features; Input the corresponding features into the pre-trained model for task prediction to obtain the optimized parameters; A power allocation instruction is generated based on the optimization parameters to achieve intelligent allocation of power resources.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the industrial silicon intelligent power distribution and control method as described in claim 8 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the industrial silicon intelligent power distribution and control method as described in claim 8 are implemented.

Images