论文部分内容阅读
利用神经网络方法开发了ＬＦ—ＶＤ—ＣＣ钢液温度预报模型。生产现场实时预报检验结果表明，预报值与实测值的误差在０～７ ℃之间的炉次为８５ ％ ，最大误差不超过１０ ℃。程序中的数据更新功能，可使预报精度不受渐变因数的影响。
The temperature prediction model of LF-VD-CC molten steel was developed by using neural network method. The results of real-time forecasting on the production site show that the error between the predicted value and the measured value is 85% between 0 and 7 ° C, and the maximum error does not exceed 10 ° C. The data update function in the program can make the prediction accuracy not affected by the gradual change factor.