矿热炉电磁场仿真与电极位置检测方法研究

为解决非接触式检测矿热炉电极位置的实际问题,使用COMSOL仿真软件,构建了矿热炉电磁场仿真模型,选取了外磁场信号的最优检测位置,并通过模拟退火(SA)优化的BP神经网络建立了矿热炉电极位置检测模型。首先根据矿热炉的几何构造和电路结构,建立了矿热炉仿真模型,对矿热炉电场和外磁场进行分析和计算。通过模型分析,选取外磁场信号的最优检测位置,并通过COMSOL后处理工具采集具有不同电极位置的矿热炉模型的外磁场信号,从而建立对应于最优测量位置的矿热炉外磁场信号样本集。在此基础上,通过SA优化的BP神经网络建立矿热炉电极位置检测模型,并且矿热炉电极位置检测准确率达到92.25%,平均绝对误差为0.0750。研究结果表明,通过COMSOL软件对矿热炉电磁场进行准确仿真,找到了外磁场信号的最优检测位置,实现了对矿热炉电极位置的非接触式精确检测。

Research on Electromagnetic Field Simulation and Electrode Position Detection Method of Submerged Arc Furnace

In order to solve the practical problems of non-contact detection of electrode positions in submerged arc furnace, the paper used COMSOL, a simulation software, to build the electromagnetic field simulation model of the submerged arc furnace, select the optimal detection location of the external magnetic field signal, and establish the electrode position detection model of the submerged arc furnace through simulated annealing(SA) optimized BP neural network. First of all, according to the geometric structure and circuit structure of the furnace, the paper established the submerged arc furnace simulation model, and the electric field and external magnetic field of the submerged arc furnace were analyzed and calculated. Through model analysis, the optimal detection positions of the external magnetic field signal were selected, and the external magnetic field signal of the model with different electrode positions was collected by COMSOL post-processing tool, so as to establish the sample set of the external magnetic field signal corresponding to the optimal detection positions. On this basis, through SA optimized BP neural network, the electrode position detection model of the submerged arc furnace was established. The detection accuracy of electrode positions can reach 92.25%, and the mean absolute error 0.0750. The results of research show that the paper accurately simulates the electromagnetic field of the furnace with COMSOL, finds the optimal detection positions of the external magnetic field signal, and realizes the accurate non-contact detection of the electrode positions in the submerged arc furnace.