A method and system for load collaborative control of a smelting furnace cluster based on hybrid optimization

Abstract

The application discloses a kind of based on mixed optimization's ore-heating furnace cluster load collaborative control method and system, the method includes: real-time acquisition in each ore-heating furnace of cluster multiple-source heterogeneous time series data, dynamic construction is fused state characteristic index by attention fusion model, and optimization characteristic parameter is calculated based on fused state characteristic index;The total regulating power of ore-heating furnace cluster is obtained, and the optimization characteristic parameter and working condition self-adaptive dynamic constraint condition of each ore-heating furnace are combined, and dynamic power distribution model is constructed to generate individual power regulation instruction of each ore-heating furnace;Each ore-heating furnace is regulated by executing corresponding individual power regulation instruction;According to the actual power value of each ore-heating furnace after power regulation, the actual total power is calculated, the deviation of actual total power and total regulating power is obtained and feedback correction is carried out according to the deviation.The application realizes the balanced regulation of ore-heating furnace cluster power by constructing the optimization characteristic parameter and working condition self-adaptive dynamic constraint condition of ore-heating furnace.

Description

Technical Field

[0001] This invention belongs to the field of centralized power scheduling and optimization control technology for submerged arc furnaces, specifically relating to a load coordination control method and system for submerged arc furnace clusters. Background Technology As a core piece of equipment in high-energy-consuming smelting, the load control of submerged arc furnaces faces challenges of strong nonlinearity, time-varying characteristics, and multiple constraints. Existing single-furnace control technologies focus on local process stability, while multi-furnace cluster operations often employ simple power allocation or independent control modes, resulting in the following deep-seated problems: First, the state perception is superficial. Cluster optimization relies heavily on directly collected raw operating data (such as power and current) or simple calculated indicators (such as adjustable margin), lacking a holistic representation of the adjustable potential and economic efficiency of individual furnaces, leading to insufficient optimization basis. Second, constraint handling is rigid. Existing schemes often use static, isolated upper and lower power limits as constraints, failing to adapt to the dynamic changes in furnace conditions and process stages during submerged arc furnace operation, resulting in conservative adjustments or potential risks. Third, the allocation strategy is simplistic, failing to consider the different smelting stages of each furnace, such as start-up, reduction, and pre-tapping, and lacking deep integration with multi-timescale control requirements, resulting in insufficient balance between response speed, overall economy, and equipment operational stability.

[0002] The aforementioned issues make it difficult to coordinate and optimize the accuracy, economy, and security of the overall response to dispatch instructions of the electric arc furnace cluster, thus restricting the ability of the electric arc furnace cluster to participate in grid interaction as a high-quality adjustable load. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a method and system for coordinated load control of electric arc furnace clusters. Through substantial improvements such as adaptive dynamic safety constraints, multi-source heterogeneous fusion state vectors, and hybrid optimized power allocation algorithms, it solves the problems of power allocation, response speed, economy, and safety in multi-furnace coordinated control. It supports grid frequency regulation and peak shaving, and can also meet the requirements of integrated power generation, grid, load, and storage projects and green electricity direct connection projects for flexible load control, thereby improving the overall energy efficiency and operational stability of the system.

[0004] The present invention adopts the following technical solution.

[0005] A method for coordinated load control of a cluster of submerged arc furnaces based on hybrid optimization, the method comprising the following steps: S1: Real-time acquisition of multi-source heterogeneous time-series data from each submerged arc furnace in the cluster, dynamic construction of fusion state feature indicators through attention fusion model, and calculation and optimization of feature parameters based on fusion state feature indicators; S2: Obtain the total regulating power of the electric arc furnace cluster, combine the optimized characteristic parameters of each electric arc furnace with the adaptive dynamic constraints of the operating conditions, construct a dynamic power allocation model to generate individual power regulation commands for each electric arc furnace; S3: Each submerged arc furnace adjusts its power by executing the corresponding individual power adjustment command; S4: Calculate the actual total power based on the actual power value of each submerged arc furnace after power adjustment, obtain the deviation between the actual total power and the total adjusted power, and make feedback corrections based on the deviation.

[0006] Preferably, S1 specifically includes: S101: Collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster to be controlled, and preprocess the collected multi-source heterogeneous time-series data to obtain the input feature matrix. S102, Construct an attention weight fusion mechanism driven by the smelting stage to obtain the attention weight matrix; S103, Generate deep state feature vectors based on attention weight matrix and input feature matrix; S104 dynamically constructs optimized feature parameters based on deep state feature vectors. The optimized feature parameters include: dynamic adjustment cost per unit power, dynamic safety adjustable margin, and comprehensive adjustability index.

[0007] Preferably, the multi-source heterogeneous time-series data specifically includes: three-phase current, voltage, active / reactive power, power factor, electrode position, electrode current, furnace temperature, material level, unit power consumption, electricity price, smelting stage indicator, safety status indicator, cumulative iron output, and cumulative electricity consumption. The smelting stages are marked as follows: shutdown stage, start-up stage, reduction stage, iron tapping preparation stage, and iron tapping stage. Safety status indicators include normal, alarm, and fault indicators. When the safety status indicator is normal, it means that the electric arc furnace can participate in regulation. When the safety status indicator is alarm or fault, it means that the electric arc furnace cannot be regulated.

[0008] Preferably, in S103, the deep state feature vector includes: multi-dimensional cross-correlation feature variables, furnace condition dynamic stability feature variables, regulation potential and boundary feature variables, regulation response capability feature variables, and regulation economy feature variables; Among them, the multi-dimensional cross-correlation characteristic variables of the i-th submerged arc furnace at time t specifically include: power-furnace temperature regulation coupling coefficient. Power-electrode current coupling coefficient Coupling coefficient of unit power consumption during smelting stage Smelting stage - adjustment cost coupling coefficient Furnace condition stability - adjustment cost coupling coefficient Operating power-equipment safety status coupling coefficient ; The dynamic stability characteristics of the i-th submerged arc furnace at time t include: the furnace temperature fluctuation rate over the N sampling periods prior to the current time. Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Power steady-state sustainability index Electrode adjustment frequency index ; The regulation potential and boundary characteristic variables of the i-th submerged arc furnace at time t include: the margin of electrode action times per unit time. The maximum rate of change of single furnace power is limited by the current smelting stage. Adjustable range limited by transformer capacity Adjustable range limited by electrode current Adjustable range limited by electrode position Adjustable range limited by the current smelting stage ; The characteristic variables of the regulation response capability of the i-th submerged arc furnace at time t include: regulation response reachability index. Power regulation tracking accuracy ; The economic characteristics of the i-th submerged arc furnace at time t include: unit regulation power consumption cost. Unit adjustment process loss cost Unit adjustment equipment wear cost .

[0009] Preferably, in step S104, the adjustment cost includes the adjustment costs of power consumption, process losses, and equipment wear. The adjustment cost is then corrected based on the smelting stage and furnace condition stability to obtain the dynamic adjustment cost per unit power. The calculation formula is as follows:

[0010] in, The cost of dynamic adjustment per unit power.

[0011] Preferably, in step S104, a comprehensive adjustability index is calculated based on dimensions such as power adjustability potential, furnace condition stability, adjustment response capability, equipment safety, and process adaptability. The calculation formula is as follows:

[0012] in, Let be the sub-index of the power tunability potential at time t. Let be the furnace condition stability sub-index at time t. Let be the sub-index of the regulatory response capability at time t. The equipment safety and process adaptability index at time t. , , , These represent the weighting coefficients for power adjustable potential, furnace condition stability, adjustment response capability, and equipment safety and process adaptability, respectively. Based on the electrode action margin per unit time at time t Power steady-state sustainability index Calculation of adjustable power range The calculation formula is as follows:

[0013] In the formula,

[0014]

[0015] Based on the power steady-state duration index at time t Furnace temperature fluctuation rate Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Electrode adjustment frequency index calculate The calculation formula is as follows:

[0016] Based on the reachability index of the adjustment response at time t Power regulation tracking accuracy calculate The calculation formula is as follows:

[0017] Based on the operating power-equipment safety state coupling coefficient at time t Power-furnace temperature regulation coupling coefficient Power-electrode current coupling coefficient calculate The calculation formula is as follows:

[0018] in, and This is the baseline value for the coupling coefficient used in process calibration.

[0019] Preferably, in step 104, the theoretical margins for power adjustment upward and downward are calculated based on four power constraints: transformer capacity, electrode current, electrode position, and smelting stage. The margins are then dynamically corrected based on equipment safety and furnace stability to obtain a safe adjustable margin, which includes a safe downward adjustment margin and a safe upward adjustment margin. No.i Safety margin for each submerged arc furnace at time t The calculation formula is as follows:

[0020] in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. The adaptive correction coefficient for power reduction is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate The calculation is as follows:

[0021] No. i Safety margin adjustment for each submerged arc furnace at time t The calculation formula is as follows:

[0022] in, The adaptive correction coefficient for power adjustment is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate Electrode current fluctuation rate The calculation is as follows: .

[0023] Preferably, S2 specifically includes: S201: Receive the total power command issued by the dispatching system, and obtain the total regulated power by subtracting the real-time total power from the total power command. ; S202, define the decision variable as the power regulation command value allocated to each boiler. Combined with total regulating power Construct a multi-objective optimization model that comprehensively considers the total operating cost of cluster power regulation and the total regulation loss of cluster equipment; S203, based on the comprehensive adjustability index, performs safety screening of submerged arc furnaces to obtain the range of furnace groups participating in the optimized allocation; S204: Use the combined state characteristic index of the current regulation capacity and regulation economy of each submerged arc furnace obtained in S1 as the input of the optimization model, and solve the optimization model to obtain the individual power regulation command of each submerged arc furnace participating in the regulation.

[0024] Preferably, the multi-objective optimization model includes an objective function and model constraints, wherein the objective function... J The minimization formula is expressed as:

[0025] in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. This represents the normalized value of the single-furnace power command at time t for the i-th submerged arc furnace; Let be the unit power dynamic adjustment cost of the i-th submerged arc furnace at time t. , , These are the weighting coefficients for cluster instruction tracking accuracy, power fluctuation, and power regulation cost, respectively.

[0026] The constraints of the multi-objective optimization model include: power regulation range constraints, power change rate constraints, power balance constraints, and equipment safety constraints. The power adjustment range is constrained as follows:

[0027] in, This is a dynamically adjustable margin constraint value; when the power is increased, When the power is reduced, ; The power change rate constraint is as follows:

[0028] in, This represents the maximum rate of change in power output of a single furnace during the current smelting stage. This represents the normalized value of the actual power of the i-th submerged arc furnace at time t-1; The power balance constraints are as follows:

[0029] Where M is the total number of ferroelectric furnaces in the cluster; The specific safety constraints for the equipment are as follows: ,

[0030] in, This refers to the minimum allowable value for equipment safety compatibility specified in the process calibration. This is the maximum permissible number of electrode movements per unit time as specified in the process calibration.

[0031] Preferably, S203 specifically includes: using a comprehensive adjustability index As a screening criterion, the safety constraint verification module screens all submerged arc furnaces in the cluster to determine the range of furnace groups to participate in the optimized allocation; The specific selection criteria are as follows: Non-adjustable furnace: <0.2 indicates no moderating ability and does not participate in modifiable allocation; Low adjustable furnace: 0.2≤ <0.5, for low-adjustability furnaces, in the subsequent optimization allocation process, the dynamic adjustability margin constraint value of the i-th submerged arc furnace at time t. satisfy: When the power is increased ; When the power is reduced ; in, , These are the safety margin for the i-th submerged arc furnace at time t, respectively; Normal adjustable furnace: ≥0.5, no adjustment range limit, participates in subsequent optimization allocation.

[0032] This invention also proposes a hybrid optimization-based load coordination control system for a cluster of submerged arc furnaces, used to implement the load coordination control method for the submerged arc furnace cluster. The system includes: a cluster monitoring center, a single furnace controller, and a real-time communication network. The cluster monitoring center is used to collect real-time operating status data of each submerged arc furnace and process it to obtain individual power commands for each submerged arc furnace; the cluster monitoring center includes a data acquisition module, a parameter calculation module, a power allocation module, and a correction module; Specifically, the data acquisition module is used to collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster in real time, and to obtain the total regulating power of the submerged arc furnace cluster. The parameter calculation module is used to dynamically construct fusion state feature indicators through a multi-source heterogeneous time-series data attention fusion model, and calculate and optimize feature parameters based on the fusion state feature indicators; The power allocation module is used to combine the optimized characteristic parameters of each submerged arc furnace and the adaptive dynamic constraints of the operating conditions to build a dynamic power allocation model and generate individual power adjustment commands for each submerged arc furnace. The correction module is used to calculate the actual total power based on the actual power value of each submerged arc furnace, as well as the deviation between the actual total power and the total regulating power command, and to perform feedback correction based on the deviation.

[0033] Each furnace controller corresponds to a single submerged arc furnace and is used to receive the individual power adjustment command and control each submerged arc furnace to adjust its power by executing the corresponding individual power adjustment command. The real-time communication network connects the cluster monitoring center with all the individual furnace controllers and transmits instructions and data in real time.

[0034] The present invention also proposes a terminal, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the hybrid optimization-based ferroelectric furnace cluster load coordination control method.

[0035] The present invention also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the hybrid optimization-based load collaborative control method for ferroelectric furnace clusters.

[0036] The beneficial effects of the present invention are that, compared with the prior art, the present invention has at least the following beneficial effects: 1. This invention introduces an attention fusion model driven by the smelting stage, which aggregates the multi-dimensional time-series state data (such as electrical quantities, process quantities, state quantities, and time-series quantities) of each submerged arc furnace into a deep state feature vector, decodes and generates the core parameters of adjustment cost and adjustment capability required for optimization decision-making, breaks through the limitations of shallow perception in traditional schemes, realizes dimensionality reduction and data aggregation, and improves the accuracy and speed of allocation decision-making. 2. This invention replaces the traditional static power constraint with a dynamic safety adjustable margin. Based on four power constraints—transformer capacity, electrode current, electrode position, and smelting stage—the adjustable range is calculated and dynamically corrected in real-time operating conditions. The furnace group is graded and screened based on the comprehensive adjustability index, which solves the problem of poor adaptability of traditional static constraints and greatly improves the system safety and adaptability. 3. This invention adopts a hierarchical collaborative architecture and a hybrid optimization algorithm to construct a multi-objective optimization model that comprehensively considers cluster command tracking accuracy, power fluctuation range and total power adjustment cost. It takes into account both the power grid peak shaving and enterprise production needs, and achieves global optimization and real-time dynamic adjustment of power allocation, effectively supporting the power grid frequency regulation and peak shaving needs. Attached Figure Description

[0037] Figure 1 This is a flowchart of the load coordination control method for electric arc furnace clusters in this invention; Figure 2 This is a schematic diagram of the intelligent state feature fusion process of the submerged arc furnace cluster in this invention; Figure 3 This is a schematic diagram of the method for generating individual power adjustment commands for each submerged arc furnace by constructing a dynamic power allocation model in this invention. Figure 4 This is a diagram of the collaborative control architecture for a single submerged arc furnace in this invention; Figure 5This is a structural diagram of the load coordination control system for the electric arc furnace cluster in this invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, other embodiments obtained by those skilled in the art without creative effort are all within the protection scope of this invention.

[0039] Figure 1 A flowchart illustrating a hybrid optimization-based load coordination control method for a cluster of submerged arc furnaces, as provided in this embodiment of the invention, is shown below. Figure 1 As shown, the proposed method for coordinated load control of submerged arc furnace clusters based on hybrid optimization includes the following steps: S1: The cluster monitoring center collects multi-source heterogeneous time-series data of each submerged arc furnace in the cluster in real time. Combining the multi-source heterogeneous time-series data and the multi-source heterogeneous time-series data attention fusion model, it dynamically constructs optimized characteristic parameters that reflect the current regulation capacity and regulation economy of each submerged arc furnace. The cluster monitoring center's communication and data acquisition module continuously collects multi-dimensional real-time operating status data of each furnace through a real-time communication network at a 1-second cycle. This includes three-phase current, voltage, active / reactive power, power factor, electrode position, electrode current, furnace temperature, material level, unit power consumption, electricity price, smelting stage indicator, safety status indicator, cumulative iron output, and cumulative electricity consumption.

[0040] In this example, the unit power consumption is obtained by dividing the power consumption over a set time by the weight of molten iron tapped, reflecting the efficiency of power utilization. The smelting stage indicators are divided into shutdown stage, start-up stage, reduction stage, molten iron tapping preparation stage, and molten iron tapping stage. Safety status indicators include three types: normal, alarm, and fault. A normal safety status indicator means the submerged arc furnace can participate in regulation, while an alarm or fault safety status indicator means the submerged arc furnace cannot be regulated.

[0041] The optimized characteristic parameters calculated in this step include: dynamic adjustment cost per unit power. Dynamic safety adjustable margin , and comprehensive adjustability index .

[0042] like Figure 2 As shown, S1 of the present invention specifically includes the following steps: S101: Collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster to be controlled, and preprocess the collected multi-source heterogeneous time-series data to obtain the input feature matrix; Specifically, the three-phase current, voltage, active / reactive power, power factor, electrode position, electrode current, furnace temperature, material level, unit power consumption, electricity price, smelting stage indicator, safety status indicator, cumulative iron output, and cumulative electricity consumption data of the i-th submerged arc furnace over the past N sampling periods are processed for differential normalization.

[0043] Preferably, N=30 is set, that is, time series data of the past 30 seconds are collected.

[0044] The normalization preprocessing of the collected multi-source heterogeneous data specifically includes: For dimensional continuous time series data (such as current, voltage, power, temperature, position, height, and iron output), interval normalization is applied and mapped to the [0, 1] interval; Dimensionless discrete state data (such as smelting stage identifiers and safety status identifiers) are converted into numerical features using one-hot encoding. Linear normalization is applied to ordered categorical data (such as equipment grades) to preserve their sequential characteristics.

[0045] The normalized multi-source heterogeneous time-series data are concatenated to obtain an initial time-series feature matrix. A positional encoding layer is then used to add temporal positional information to the initial time-series feature matrix to eliminate positional ambiguity in the time-series data, ultimately generating the input feature matrix. , where d is the original feature dimension, which is the number of types of collected data.

[0046] S102: Construct an attention weight fusion mechanism driven by the smelting stage to obtain the attention weight matrix. ; Taking the current smelting stage identifier (start-up, reduction, iron tapping preparation, iron tapping) of the i-th submerged arc furnace as the core, a smelting stage-driven attention mechanism is constructed to calculate the attention weight of the operating status characteristics at each historical moment on the current operating condition evaluation, thereby realizing the integration of process knowledge and operating data.

[0047] First, the current smelting stage identifier is transformed into a low-dimensional smelting stage embedding vector through an embedding layer, and then a query vector is generated through a linear transformation layer. , input feature matrix The key vectors are generated through two independent linear transformation layers. Sum value vector ,satisfy , , ,in The query / key feature dimension of the attention mechanism after linear transformation. In this embodiment, the value feature dimension of the attention mechanism after linear transformation is... .

[0048] In this embodiment, the scaling dot product attention mechanism is preferably used to calculate the attention weights, and its core formula is as follows: ,in The scaled attention score. The function is used to normalize the attention score to the [0, 1] interval and generate the attention weight matrix. Attention weight matrix Each element corresponds to a weight value for a historical sampling period, reflecting the importance distribution of the operating status characteristics at that historical moment to assess the current regulation capacity of the submerged arc furnace under the process requirements of the current smelting stage.

[0049] S103: Based on the attention weight matrix With the input feature matrix Generate deep state feature vectors; attention weight matrix With the input feature matrix Weighted summation is performed to highlight the feature contributions of high-weight historical moments and suppress noise interference from low-weight moments, resulting in a weighted temporal feature matrix. This matrix is ​​then fed into a multi-head attention layer for cross-fusion of multiple feature dimensions, followed by feature compression and nonlinear mapping via a feedforward neural network (FFN), ultimately generating the deep state feature vector of the i-th submerged arc furnace. , where m is the total depth feature dimension.

[0050] This depth state feature vector It is a high-dimensional abstract representation of the multi-dimensional cross-correlation, furnace condition stability, regulation potential, response capability, and regulation economy of electric arc furnaces, comprehensively integrating the core information of multi-source operating data, specifically including: (1) Multi-dimensional cross-correlated feature variables: This set of variables is used to quantify the linkage patterns, coupling degree, and synergistic change characteristics among the five independent categories of electrical, technological, equipment, economic, and safety features of electric arc furnaces, solving the problem of multi-parameter synergistic operation mechanisms that cannot be characterized by single-dimensional features. Specifically, it includes the power-furnace temperature regulation coupling coefficient of the i-th electric arc furnace at time t. Power-electrode current coupling coefficient Coupling coefficient of unit power consumption during smelting stage Smelting stage - adjustment cost coupling coefficient Furnace condition stability - adjustment cost coupling coefficient Operating power-equipment safety status coupling coefficient .

[0051] (2) Dynamic stability characteristics of furnace conditions: These include the rate of change, fluctuation amplitude, and steady-state characteristics of core process and electrical parameters over N sampling periods prior to the current time t, reflecting the stability and trend of the smelting conditions. Specifically, this includes the furnace temperature fluctuation rate of the i-th submerged arc furnace at time t. Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Power steady-state sustainability index Electrode adjustment frequency index .

[0052] (3) Regulation potential and boundary characteristics variables: This set of variables quantifies the regulation potential and boundary characteristics of the submerged arc furnace, specifically including the margin of electrode action times per unit time for the i-th submerged arc furnace at time t. The maximum rate of change of single furnace power is limited by the current smelting stage. Adjustable range limited by transformer capacity Adjustable range limited by electrode current Adjustable range limited by electrode position Adjustable range limited by the current smelting stage .

[0053] in, These are the lower and upper limits of the adjustable power of the i-th submerged arc furnace at time t, which are limited by the transformer capacity. These are the lower and upper limits of the adjustable power of the i-th submerged arc furnace at time t, which are limited by the electrode current, respectively. These are the lower and upper limits of the adjustable power of the i-th submerged arc furnace at time t, which are limited by the electrode position, respectively. These are the lower and upper limits of the adjustable power of the i-th submerged arc furnace at time t, which are limited by the current smelting stage.

[0054] (4) Regulation response capability characteristic variables: This group of variables quantifies the ability of the submerged arc furnace to execute power regulation commands, and determines the reachability of regulation commands and control accuracy. Specifically, it includes the regulation response reachability index of the i-th submerged arc furnace at time t. Power regulation tracking accuracy .

[0055] (5) Economic characteristics variables: This group of variables quantifies the full-cycle costs and benefits of power regulation, specifically including the unit regulation power consumption cost of the i-th submerged arc furnace at time t. Unit adjustment process loss cost Unit adjustment equipment wear cost .

[0056] S104: Based on deep state feature vectors Dynamically construct and optimize feature parameters; Based on depth state feature vector Based on the operating characteristics and safety requirements of the electric arc furnace, four core optimization feature parameters are dynamically constructed: dynamic adjustment cost per unit power. Comprehensive Adjustability Index Safe adjustable margin , This provides a quantitative decision-making basis for adaptive operation of cluster load coordination optimization. The specific implementation is as follows: (1) Unit power dynamic adjustment cost The adjustment cost covers three dimensions: power consumption, process loss, and equipment wear. It incorporates the smelting stage and furnace condition stability to correct the adjustment cost, and calculates the dynamic adjustment cost per unit power. The specific calculation formula is as follows:

[0057] in, Used to correct power consumption during the smelting stage, reflecting the dynamic impact of power consumption efficiency on adjustment costs; Used to correct adjustment costs during the smelting stage. If the coefficient is greater than 1 during a process-sensitive stage (such as the tapping stage), the adjustment cost will be increased and the adjustment priority of the furnace will be reduced. Used to correct the adjustment cost of furnace condition stability. When the furnace condition fluctuation exceeds the time limit, the coefficient is greater than 1, which increases the adjustment cost of unstable furnaces and avoids process risks. (2) Comprehensive adjustability index The comprehensive adjustability index is calculated by integrating four core dimensions: adjustable power potential, furnace condition stability, adjustment response capability, equipment safety, and process adaptability. The specific calculation formula is as follows:

[0058] in, The sub-index of power adjustable potential is the margin of electrode action times per unit time. Power steady-state sustainability index The power adjustable range is calculated using the following formula:

[0059] In the formula, and The first and second intermediate parameters are respectively, and their calculation formulas are as follows:

[0060]

[0061] in, The sub-index for furnace condition stability is determined by the power steady-state sustainability index. Furnace temperature fluctuation rate Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Electrode adjustment frequency index Calculated. The closer the value is to 1, the more stable the furnace operation of the submersible arc furnace. The calculation formula is as follows:

[0062] in, To adjust the response capability sub-index, the response accessibility index is adjusted. Power regulation tracking accuracy Calculated. The closer the value is to 1, the stronger the regulation response capability and the higher the command tracking accuracy of the submerged arc furnace. The calculation formula is as follows:

[0063] in, The sub-index for equipment safety and process adaptability is determined by the coupling coefficient between operating power and equipment safety status. Power-furnace temperature regulation coupling coefficient Power-electrode current coupling coefficient The calculation is as follows:

[0064] In the formula, and This is the baseline value for the coupling coefficient used in process calibration.

[0065] in, , , , These represent the weighting coefficients for power adjustable potential, furnace condition stability, adjustment response capability, equipment safety, and process adaptability, respectively. In this example, the values ​​are 0.5, 0.2, 0.2, and 01, respectively.

[0066] (3) Adjustable safety margin , Based on four power constraints—transformer capacity, electrode current, electrode position, and smelting stage—the theoretical margins for power adjustment (both upward and downward) are calculated. Then, through dynamic correction based on equipment safety and furnace stability characteristics, the adaptive safety margin for power adjustment of the i-th submerged arc furnace at time t is obtained. and safety margin .

[0067] No. i Safety margin for each submerged arc furnace at time t The calculation formula is as follows:

[0068] in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. The adaptive correction coefficient for power reduction is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate The calculated adaptive correction coefficient for power downgrade is... The calculation formula is as follows:

[0069] No. i Safety margin adjustment for each submerged arc furnace at time t The calculation formula is as follows:

[0070] in, The adaptive correction coefficient for power adjustment is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate Electrode current fluctuation rate The calculation is as follows:

[0071] S2: Obtain the total regulating power of the electric arc furnace cluster, combine the optimized characteristic parameters of each electric arc furnace with the adaptive dynamic constraints of the operating conditions, and generate individual power commands for each electric arc furnace through a dynamic power allocation algorithm. like Figure 3 As shown, S2 includes the following steps: S201, The cluster monitoring center receives the total power command issued by the scheduling system. According to the total power command The total regulating power is calculated by subtracting the real-time total power. ; S202, Establish a multi-objective optimization model: Define the decision variable as the power regulation command value allocated to each boiler. Construct a multi-objective optimization model that comprehensively considers the total operating cost of cluster power adjustment and the total adjustment loss of cluster equipment; The power allocation problem of an electric arc furnace cluster is formalized into a mathematical optimization model, defining the decision variable as the power adjustment command value allocated to the i-th electric arc furnace. , where i = 1 ~ M, and M is the total number of electric arc furnaces.

[0072] Design a multi-objective optimization model that comprehensively considers cluster command tracking accuracy, power fluctuation, and power regulation cost. The multi-objective optimization model includes an objective function and model constraints. objective function J The minimization formula is expressed as:

[0073] in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. This represents the normalized value of the single-furnace power command for the i-th submerged arc furnace at time t. , , These are the weighting coefficients for cluster instruction tracking accuracy, power fluctuation, and power regulation cost, respectively.

[0074] With cluster power command tracking accuracy as the primary objective, and power fluctuation and power regulation costs as secondary objectives, in this example... , , The values ​​are 0.6, 0.2, and 0.2 respectively.

[0075] The constraints of the multi-objective optimization model include: power regulation range constraints, power change rate constraints, power balance constraints, and equipment safety constraints; The power adjustment range constraints are as follows:

[0076] in, This is a dynamically adjustable margin constraint value; when the power is increased, When the power is reduced, .

[0077] The power change rate constraint is as follows:

[0078] in, This represents the maximum rate of change in power output of a single furnace during the current smelting stage. This represents the normalized value of the actual power of the i-th submerged arc furnace at time t-1.

[0079] The power balance constraints are as follows:

[0080] The specific safety constraints for the equipment are as follows: ,

[0081] in, This refers to the minimum allowable value for equipment safety compatibility specified in the process calibration. This is the maximum permissible number of electrode movements per unit time as specified in the process calibration.

[0082] S203: Based on the comprehensive adjustability index, a safety screening of submerged arc furnaces is performed to obtain the range of furnace groups that can participate in the optimized allocation; Before solving the optimization model, the algorithm first calls the safety constraint verification module to screen all submerged arc furnaces. The overall adjustability index decoded by S1 is used as the basis for this process. Using the security constraint verification module as the sole screening criterion, all submerged arc furnaces within the cluster are screened to determine the range of furnace groups participating in the optimized allocation.

[0083] The specific selection criteria are as follows: ① Non-adjustable furnace: <0.2 indicates no moderating ability and does not participate in modifiable allocation; ② Low adjustable furnace: 0.2≤ <0.5, only small adjustments are allowed. For low-adjustable furnaces, during subsequent optimization and allocation processes, when the power is increased, ,contrast With safety margin The smaller of the two values ​​is used to correct the dynamic adjustability margin constraint value of the low-adjustability furnace. When the power is reduced, ,contrast With safety margin reduction The smaller of the two values ​​is used to correct the dynamic adjustability margin constraint value of the low-adjustability furnace. ; ③ Normal adjustable furnace: ≥0.5, no adjustment range limit, participates in subsequent optimization allocation.

[0084] S204: Use the combined state characteristic index of the current regulation capacity and regulation economy of each submerged arc furnace obtained in S1 as the input of the optimization model, and solve the optimization model to obtain the individual power command of each submerged arc furnace participating in regulation.

[0085] The combined state characteristic index of the current regulating capacity and regulating economy of each submerged arc furnace obtained in step 1 is used as the input to the optimization model. For the submerged arc furnaces that have been screened and can participate in regulation, the above optimization problem is transformed into a standard linear programming or quadratic programming problem. Given the requirements of real-time control, a solver with high computational efficiency and good convergence is preferred. The global optimizer module sends the objective function, constraints, and real-time operating status data to the solver, and the solver outputs the optimal power allocation scheme, that is, the individual power command for each furnace participating in regulation.

[0086] S205: Exception Handling and Feasibility Verification. If the solver returns "no feasible solution," it means that the total power command requirement cannot be met under the current safety constraints. In this case, the algorithm reports the command as infeasible to the scheduling system and suggests an achievable power range. Before the command is issued, the safety constraint verification module will perform a final verification of the allocation results.

[0087] S3: The cluster monitoring center sends individual power commands to the individual furnace controllers corresponding to each submerged arc furnace through the communication and data acquisition module. Each individual furnace controller coordinates the control of the electrode regulation system (APC), on-load tap changer and reactive power compensation device to execute the individual power commands.

[0088] like Figure 4 As shown, the cluster monitoring center sends individual power commands to the corresponding single-furnace controllers. Each single-furnace controller collaboratively controls the Automatic Electrode Regulation (APC), On-Load Tap Changer (OLTC), and Static Var Generator (SVG) to execute the commands. The APC system finely adjusts the electrode insertion depth and changes the loop resistance through second-level response, thereby precisely adjusting the input power to quickly approach the commanded value. When the power deviation of the submerged arc furnace continuously exceeds the APC's adjustment capability, its single-furnace controller determines that coarse adjustment is needed and triggers the OLTC to perform a minute-level response range switching, shifting the entire power operating range. The APC then performs fine adjustment again on the new voltage platform. Throughout the entire control process, dynamic reactive power compensation with millisecond-level response is achieved through the SVG to maintain voltage stability and ensure the power factor meets the target.

[0089] S4: Calculate the actual total power based on the actual power value of each submerged arc furnace, calculate the deviation between the actual total power and the total regulated power, and make feedback corrections based on the deviation.

[0090] Preferably, closed-loop control is performed using a PID algorithm to ensure that the real-time total power of the cluster accurately matches the instructions issued by the scheduling system.

[0091] Figure 5 An architecture diagram of a load-coordinated control system for a cluster of submerged arc furnaces is provided as an embodiment of the present invention, such as... Figure 5As shown, the hybrid optimization-based load collaborative control system for ferroelectric furnace clusters proposed in this invention is used to implement the above-mentioned hybrid optimization-based load collaborative control method for ferroelectric furnace clusters. The system adopts a centralized-distributed collaborative control architecture, including: a cluster monitoring center, a single furnace controller, and a real-time communication network. The cluster monitoring center is deployed in the factory's central control room and is equipped with a high-performance industrial server. It is used to collect real-time operating status data of each submerged arc furnace, receive total power instructions from the scheduling system, run the dynamic power allocation algorithm, and issue individual power instructions. The cluster monitoring center includes: a data acquisition module, a parameter calculation module, a power allocation module, and a correction module. Specifically, the data acquisition module is used to collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster in real time, and to obtain the total regulating power of the submerged arc furnace cluster. The parameter calculation module is used to dynamically construct fusion state feature indicators through a multi-source heterogeneous time-series data attention fusion model, and calculate and optimize feature parameters based on the fusion state feature indicators; The power allocation module is used to combine the optimized characteristic parameters of each submerged arc furnace and the adaptive dynamic constraints of the operating conditions to build a dynamic power allocation model and generate individual power adjustment commands for each submerged arc furnace. The correction module is used to calculate the actual total power based on the actual power value of each submerged arc furnace, as well as the deviation between the actual total power and the total regulating power command, and to perform feedback correction based on the deviation.

[0092] Each furnace controller corresponds to a single submerged arc furnace and is used to receive individual power adjustment commands and control each submerged arc furnace to adjust its power by executing the corresponding individual power adjustment commands. Each furnace controller corresponds to a single submerged arc furnace (SAF) and is used to receive individual power commands. It then controls the local electrode regulation system, active power regulating transformer, and reactive power compensation device of the SAF to execute the corresponding individual power commands for power adjustment. The reactive power compensation device is preferably a static var generator (SVG). Specifically, each SAF is configured one-to-one with a SAF, employing a highly reliable PLC or dedicated controller. It communicates and controls the Automatic Electrode Control System (APC), On-Load Tap Changer (OLTC), and Static Var Generator (SVG) via hardwiring or a real-time communication network.

[0093] The preferred real-time communication network is an industrial Ethernet ring network, which connects the cluster monitoring center with all individual furnace controllers and transmits commands and data in real time, ensuring that the transmission delay of commands and data is less than 100ms to meet real-time control requirements.

[0094] The control system supports communication with the dispatching system and receives automatic generation control or demand response commands. The dispatching system is preferably an integrated source-grid-load-storage control system, a power grid load management system, or a virtual power plant platform.

[0095] The beneficial effects of this invention are that, compared with the prior art, the control method and system proposed in this invention, through multi-source heterogeneous feature-level fusion of state vectors, dynamic threshold constraints with adaptive operating conditions, and multi-objective hybrid optimization improvement of dynamic power allocation algorithms, effectively solve the problems of slow response, poor economy, high safety risks, and difficulty in guaranteeing product quality in the traditional load control of submerged arc furnace clusters. It can achieve accurate, fast, safe and coordinated control of cluster loads, significantly improve the efficiency of multi-furnace coordinated operation, reduce product power consumption and equipment wear, effectively support grid frequency regulation and peak shaving, and is suitable for high-energy-consuming smelting scenarios such as ferrosilicon alloys and industrial silicon.

[0096] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0097] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0098] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0099] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for coordinated load control of a cluster of submerged arc furnaces based on hybrid optimization, characterized in that, The method includes the following steps: S1: Real-time acquisition of multi-source heterogeneous time-series data from each submerged arc furnace in the cluster, dynamic construction of fusion state feature indicators through attention fusion model, and calculation and optimization of feature parameters based on fusion state feature indicators; S2: Obtain the total regulating power of the electric arc furnace cluster, combine the optimized characteristic parameters of each electric arc furnace with the adaptive dynamic constraints of the operating conditions, construct a dynamic power allocation model to generate individual power regulation commands for each electric arc furnace; S3: Each submerged arc furnace adjusts its power by executing the corresponding individual power adjustment command; S4: Calculate the actual total power based on the actual power value of each submerged arc furnace after power adjustment, obtain the deviation between the actual total power and the total adjusted power, and make feedback corrections based on the deviation.

2. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 1, characterized in that, S1 specifically includes: S101: Collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster to be controlled, and preprocess the collected multi-source heterogeneous time-series data to obtain the input feature matrix. S102, Construct an attention weight fusion mechanism driven by the smelting stage to obtain the attention weight matrix; S103, Generate deep state feature vectors based on attention weight matrix and input feature matrix; S104 dynamically constructs optimized feature parameters based on deep state feature vectors. The optimized feature parameters include: dynamic adjustment cost per unit power, dynamic safety adjustable margin, and comprehensive adjustability index.

3. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 2, characterized in that, The multi-source heterogeneous time-series data specifically includes: three-phase current, voltage, active / reactive power, power factor, electrode position, electrode current, furnace temperature, material level, unit power consumption, electricity price, smelting stage indicator, safety status indicator, cumulative iron output, and cumulative electricity consumption. The smelting stages are marked as follows: shutdown stage, start-up stage, reduction stage, iron tapping preparation stage, and iron tapping stage. Safety status indicators include normal, alarm, and fault indicators. When the safety status indicator is normal, it means that the electric arc furnace can participate in regulation. When the safety status indicator is alarm or fault, it means that the electric arc furnace cannot be regulated.

4. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 2, characterized in that, In S103, the deep state feature vector includes: multi-dimensional cross-correlation feature variables, furnace condition dynamic stability feature variables, regulation potential and boundary feature variables, regulation response capability feature variables, and regulation economy feature variables. Among them, the multi-dimensional cross-correlation characteristic variables of the i-th submerged arc furnace at time t specifically include: power-furnace temperature regulation coupling coefficient. Power-electrode current coupling coefficient Coupling coefficient of unit power consumption during smelting stage Smelting stage - adjustment cost coupling coefficient Furnace condition stability - adjustment cost coupling coefficient Operating power-equipment safety status coupling coefficient ; The dynamic stability characteristics of the i-th submerged arc furnace at time t include: the furnace temperature fluctuation rate over the N sampling periods prior to the current time. Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Power steady-state sustainability index Electrode adjustment frequency index ; The regulation potential and boundary characteristic variables of the i-th submerged arc furnace at time t include: the margin of electrode action times per unit time. The maximum rate of change of single furnace power is limited by the current smelting stage. Adjustable range limited by transformer capacity Adjustable range limited by electrode current Adjustable range limited by electrode position Adjustable range limited by the current smelting stage ; The characteristic variables of the regulation response capability of the i-th submerged arc furnace at time t include: regulation response reachability index. Power regulation tracking accuracy ; The economic characteristics of regulation at time t include: unit regulation power consumption cost. Unit adjustment process loss cost Unit adjustment equipment wear cost .

5. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 4, characterized in that, In step S104, the adjustment cost includes the adjustment costs of power consumption, process losses, and equipment wear. The adjustment cost is then adjusted based on the smelting stage and furnace condition stability to obtain the dynamic adjustment cost per unit power. The calculation formula is as follows: in, The cost of dynamic adjustment per unit power.

6. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 4, characterized in that, In step S104, a comprehensive adjustability index is calculated based on dimensions such as power adjustability potential, furnace condition stability, adjustment response capability, equipment safety, and process adaptability. The calculation formula is as follows: in, Let be the sub-index of the power tunability potential at time t. Let be the furnace condition stability sub-index at time t. Let be the sub-index of the regulatory response capability at time t. The equipment safety and process adaptability index at time t. , , , These represent the weighting coefficients for power adjustable potential, furnace condition stability, adjustment response capability, and equipment safety and process adaptability, respectively. Based on the electrode action margin per unit time at time t Power steady-state sustainability index Calculation of adjustable power range The calculation formula is as follows: In the formula, Let i be the rated power of the i-th submerged arc furnace. and The calculation formulas are as follows: Based on the power steady-state duration index at time t Furnace temperature fluctuation rate Electrode current fluctuation rate Total active power fluctuation rate Furnace temperature change rate exceeding the limit Electrode adjustment frequency index calculate The calculation formula is as follows: Based on the reachability index of the adjustment response at time t Power regulation tracking accuracy calculate The calculation formula is as follows: Based on the operating power-equipment safety state coupling coefficient at time t Power-furnace temperature regulation coupling coefficient Power-electrode current coupling coefficient calculate The calculation formula is as follows: in, and This is the baseline value for the coupling coefficient used in process calibration.

7. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 4, characterized in that, In step 104, the theoretical margins for power adjustment and reduction are calculated based on four power constraints: transformer capacity, electrode current, electrode position, and smelting stage. The margins are then dynamically corrected based on equipment safety and furnace stability to obtain a safe adjustable margin, which includes a safe downward adjustment margin and a safe upward adjustment margin. No. i Safety margin for each submerged arc furnace at time t The calculation formula is as follows: in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. The adaptive correction coefficient for power reduction is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate The calculation is as follows: No. i Safety margin adjustment for each submerged arc furnace at time t The calculation formula is as follows: in, The adaptive correction coefficient for power adjustment is determined by the margin of electrode action times per unit time. Power steady-state sustainability index Operating power-equipment safety status coupling coefficient Furnace temperature fluctuation rate Electrode current fluctuation rate The calculation is as follows: 。 8. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 1, characterized in that, S2 specifically includes: S201: Receive the total power command issued by the dispatching system, and obtain the total regulated power by subtracting the real-time total power from the total power command. ; S202, define the decision variable as the power regulation command value allocated to each boiler. Combined with total regulating power Construct a multi-objective optimization model that comprehensively considers the total operating cost of cluster power regulation and the total regulation loss of cluster equipment; S203, based on the comprehensive adjustability index, performs safety screening of submerged arc furnaces to obtain the range of furnace groups participating in the optimized allocation; S204: Use the combined state characteristic index of the current regulation capacity and regulation economy of each submerged arc furnace obtained in S1 as the input of the optimization model, and solve the optimization model to obtain the individual power regulation command of each submerged arc furnace participating in the regulation.

9. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 6, characterized in that, The multi-objective optimization model includes an objective function and model constraints. The objective function... J The minimization formula is expressed as: in, This represents the normalized value of the actual power of the i-th submerged arc furnace at time t. This represents the normalized value of the single-furnace power command at time t for the i-th submerged arc furnace; Let be the unit power dynamic adjustment cost of the i-th submerged arc furnace at time t. , , These are the weighting coefficients for cluster instruction tracking accuracy, power fluctuation, and power regulation cost, respectively. The constraints of the multi-objective optimization model include: power regulation range constraints, power change rate constraints, power balance constraints, and equipment safety constraints. The power adjustment range is constrained as follows: in, This is a dynamically adjustable margin constraint value; when the power is increased, When the power is reduced, ; The power change rate constraint is as follows: in, This represents the maximum rate of change in power output of a single furnace during the current smelting stage. This represents the normalized value of the actual power of the i-th submerged arc furnace at time t-1; The power balance constraints are as follows: Where M is the total number of ferroelectric furnaces in the cluster; The specific safety constraints for the equipment are as follows: , in, This refers to the minimum allowable value for equipment safety compatibility specified in the process calibration. This is the maximum permissible number of electrode movements per unit time as specified in the process calibration.

10. The method for coordinated load control of submerged arc furnace clusters based on hybrid optimization according to claim 6, characterized in that, S203 specifically includes: using a comprehensive adjustability index As a screening criterion, the safety constraint verification module screens all submerged arc furnaces in the cluster to determine the range of furnace groups to participate in the optimized allocation; The specific selection criteria are as follows: Non-adjustable furnace: <0.2 indicates no moderating ability and does not participate in modifiable allocation; Low adjustable furnace: 0.2≤ <0.5, for low-adjustability furnaces, in the subsequent optimization allocation process, the dynamic adjustability margin constraint value of the i-th submerged arc furnace at time t. satisfy: When the power is increased ; When the power is reduced ; in, , These represent the safety margin for the i-th submerged arc furnace at time t, with the margins being the upper and lower limits respectively. Let be the rated power of the i-th submerged arc furnace; Normal adjustable furnace: ≥0.5, no adjustment range limit, participates in subsequent optimization allocation.

11. A hybrid optimization-based load coordination control system for a cluster of submerged arc furnaces, used to implement the load coordination control method for a cluster of submerged arc furnaces as described in any one of claims 1-10, characterized in that, The system includes: a cluster monitoring center, a single furnace controller, and a real-time communication network; The cluster monitoring center is used to collect real-time operating status data of each submerged arc furnace and process it to obtain individual power commands for each submerged arc furnace; the cluster monitoring center includes a data acquisition module, a parameter calculation module, a power allocation module, and a correction module; Specifically, the data acquisition module is used to collect multi-source heterogeneous time-series data of each submerged arc furnace in the cluster in real time, and to obtain the total regulating power of the submerged arc furnace cluster. The parameter calculation module is used to dynamically construct fusion state feature indicators through a multi-source heterogeneous time-series data attention fusion model, and calculate and optimize feature parameters based on the fusion state feature indicators; The power allocation module is used to combine the optimized characteristic parameters of each submerged arc furnace and the adaptive dynamic constraints of the operating conditions to build a dynamic power allocation model and generate individual power adjustment commands for each submerged arc furnace. The correction module is used to calculate the actual total power based on the actual power value of each submerged arc furnace, as well as the deviation between the actual total power and the total regulating power command, and to perform feedback correction based on the deviation. Each furnace controller corresponds to a single submerged arc furnace and is used to receive the individual power adjustment command and control each submerged arc furnace to adjust its power by executing the corresponding individual power adjustment command. The real-time communication network connects the cluster monitoring center with all the individual furnace controllers and transmits instructions and data in real time.

12. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-10.

Images