Search

CN-121995819-A - Alloy addition control method and system in steelmaking process

CN121995819ACN 121995819 ACN121995819 ACN 121995819ACN-121995819-A

Abstract

The invention belongs to the technical field of metallurgical process control, and discloses a method and a system for controlling alloy addition in a steelmaking process. The method comprises the steps of collecting state information such as molten steel components, temperature and weight in real time, obtaining alloy components and setting molten steel target components, utilizing an alloy element yield prediction model constructed based on a neural network to predict the yield of the current heat in real time, and then calculating an optimal alloy addition amount scheme meeting all component constraints by using an optimization algorithm with the lowest total alloy addition cost as an optimization target. The corresponding system comprises a data acquisition module, an alloy composition management module, a target setting module, a yield prediction module, an optimization calculation module and an output module. The invention overcomes the defect of the traditional dependence on fixed yield experience value by combining intelligent prediction with cost optimization, realizes accurate control and cost minimization of alloy addition, and is suitable for various steelmaking processes such as converters, electric furnaces, refining and the like.

Inventors

  • CUI MENG
  • ZHANG LIFENG
  • LI DONGHUA
  • WANG WEIJIAN
  • MA GUIFEN
  • DONG LITONG
  • Dong Zhanjian

Assignees

  • 天津市新天钢联合特钢有限公司
  • 北方工业大学

Dates

Publication Date
20260508
Application Date
20251224

Claims (10)

  1. 1. The method for controlling the alloy addition amount in the steelmaking process is characterized by comprising the following steps of: Step 1, acquiring molten steel state information of smelting heat in real time, wherein the molten steel state information comprises molten steel components, temperature and weight; Step 2, obtaining component information of various currently used alloys; Step 3, setting target contents of all alloy elements in the molten steel; step 4, based on the molten steel state information, predicting the yield of each alloy element of the current heat in real time through a pre-trained neural network model; And 5, calculating an alloy addition amount scheme meeting all component constraints by adopting an optimization algorithm by taking the lowest total alloy addition cost as an optimization target based on the molten steel state information, the alloy component information, the target content and the predicted yield.
  2. 2. The method according to claim 1, wherein in step 4, the training process of the neural network model comprises: collecting historical production data to construct a training data set, wherein the historical production data comprises molten steel state information and corresponding actual alloy element yield; normalizing the training data set; Training and testing the neural network by using the processed data set until the model prediction precision reaches the preset requirement.
  3. 3. The method of claim 2, wherein the normalization of the data is performed using the formula: (2) Wherein y is a normalized value, y min is a minimum value of y, 1 is taken, y max is a maximum value of y, 1 is taken, X is an original input variable, and X min and X max are minimum and maximum values of the input variables which are statistically determined according to historical data.
  4. 4. The method of claim 2, wherein the transfer function between neurons in the neural network model uses a combination of an S-shaped function for the middle layer and a linear function for the output layer.
  5. 5. The method of claim 4, wherein the number of middle tier nodes num of the neural network model is determined according to the number of input tier nodes and the number of output tier nodes by the following formula: (3); wherein m is the number of nodes of an input layer, n is the number of nodes of an output layer, and a ranges from 1 to 15.
  6. 6. The method of claim 4, wherein the transfer function between neurons in the neural network model is formulated as follows: (4) (5) In the formulas (4) and (5), w represents the weight value connected between the neuron nodes, i represents the ith node of the current layer of the neural network, j represents the jth node of the previous layer of the neural network, x represents the normalized variable, and θ represents the threshold value of the neuron node.
  7. 7. The method according to claim 1, wherein in step 5, the optimization algorithm is any one of an interior point method, a quadratic programming method, or a penalty function method.
  8. 8. The method according to claim 1, wherein the calculation formula of the addition amount of the single alloying element i is as follows: (7); Wherein m i represents the added mass of the alloy element i, kg, w (i) g represents the target content of the alloy element i in the steel,%, "w (i) d represents the content of the alloy element i in the steel under the current condition,%," m steel represents the mass of the molten steel, kg; the content of the alloy element i in the ferroalloy is shown in percent, and Y (i) shows the yield of the ferroalloy containing the element i when the ferroalloy is added into molten steel.
  9. 9. The method of claim 1, wherein the objective function of the optimization objective is: (8); wherein, cost is the Cost of the total alloy addition, cost i is the unit price of the alloy element i per kg, and m i is the addition mass of the alloy element i per kg.
  10. 10. A steelmaking process alloy addition control system for implementing the method of any one of claims 1-9, comprising: The data acquisition module is used for acquiring molten steel state information and alloy material information in real time; The alloy component management module is used for storing and managing component data of various alloys; the target setting module is used for setting the target content of each alloy element in the molten steel; The yield prediction module is internally provided with the neural network model as claimed in claim 1 and is used for predicting the yield of each alloy element in real time; The optimization calculation module is used for calculating an optimal alloy addition amount scheme by adopting an optimization algorithm with the aim of minimizing the total alloy cost; and the output module is used for outputting the optimal alloy addition amount scheme.

Description

Alloy addition control method and system in steelmaking process Technical Field The invention belongs to the technical field of metallurgical process control, and particularly relates to a method and a system for controlling alloy addition in a steelmaking process. Background Along with the continuous improvement of the requirements of the modern steel industry on the product quality and the production efficiency, the accurate control of alloy elements in the steelmaking process has become a key link for guaranteeing the mechanical property, corrosion resistance and process stability of steel. In actual smelting operation, in order to meet the component design requirements of different steel grades, various alloying elements such as carbon, silicon, manganese, chromium, niobium and the like are added into molten steel. However, the traditional alloy adding mode is highly dependent on experience judgment of operators, and lacks systematic quantitative analysis on the current state of molten steel, alloy yield and cost constraint, so that the fluctuation of component control is large, the primary hit rate is low, the condition that the alloy needs to be added for the second time frequently occurs, the smelting period is prolonged, and the consumption of raw materials and the energy waste are obviously increased. The accurate calculation of the alloy addition amount depends on comprehensive consideration of multidimensional parameters such as initial components, target components, molten steel amount, alloy grade, element yield and the like of molten steel. In the prior art, although an attempt is made to introduce a basic material balance formula for estimation, the alloy yield is generally regarded as a fixed empirical value, and the dynamic change characteristics of the alloy under different temperatures, oxygen potentials, molten steel components and operating conditions are ignored, so that the theoretical calculation value is seriously deviated from the actual effect. Meanwhile, a component coupling relation exists among various alloys, the adjustment of a single element can influence the final content of other elements, and the traditional method lacks the capability of multi-variable cooperative optimization, so that the alloy cost is difficult to minimize on the premise of meeting all component constraints. Therefore, there is a need for an intelligent control method and system for alloy addition in a steelmaking process, which combines real-time data sensing, dynamic yield prediction and multi-objective optimization decision, so as to break through experience dependence bottleneck and improve component control precision and production economy. Disclosure of Invention The invention aims to provide a method and a system for controlling the alloy addition in the steelmaking process, which can effectively solve the problems in the background technology. In order to achieve the purpose, the technical scheme adopted by the invention is a method for controlling the alloy addition in the steelmaking process, which comprises the following steps: Step 1, collecting molten steel state information of smelting heat in real time, wherein the molten steel state information comprises molten steel components, temperature and weight; step 2, obtaining component information of each type of alloy at present; Setting target content of each alloy element in molten steel, wherein the target content can be automatically acquired from the system or manually input, when an automatic acquisition mode is adopted, the system calls component requirements of corresponding steel types from a preset steel type standard component database, and when a manual input mode is adopted, an operator inputs a specific component control target value through a human-computer interface. And 4, predicting the yield of each alloy element of the current heat in real time through a pre-trained neural network model based on the molten steel state information, wherein the alloy element yield is the ratio of the mass of the alloy in the molten steel in the production process to the mass of the alloy actually added into the molten steel. (1) Wherein m represents the added mass of the alloy element i, kg, m g represents the mass of the alloy in the molten steel after the addition of the alloy, and m d represents the mass of the alloy element i in the molten steel before the addition of the alloy; the content of the alloy element i in the ferroalloy is shown in percent, and Y is the yield of the ferroalloy containing the element i when the ferroalloy is added into molten steel. The alloy element yield can be calculated based on a neural network model, and can be manually input according to actual production experience; And 5, calculating an alloy addition amount scheme meeting all component constraints by adopting an optimization algorithm based on the molten steel state information, the alloy component information, the target content and the predicted al