G-BP算法在烧结矿FeO指标预测中的应用

The Application of G-BP in FeO Content Prediction During Sintering

实现烧结过程工艺参数的优化,首先要进行烧结矿质量预测·采用遗传算法与BP神经网络相结合的方法,建立了烧结矿FeO含量预测模型,并改进BP学习算法·仿真表明,该方法可以优化神经网络结构,缩短学习时间·与传统的BP神经网络模型相比,预测值与实际值间的相对误差由6 534%降低至1 400%,其精度高于传统BP网络模型·该方法为实现在线预测奠定基础·

?It is necessary to predict sintering quality in order to realize optimization of technology parameters in sinter process. BP neural network combined with Genetic Algorithms was used to build up a prediction model for FeO content during sinter process. Learning method of BP neural network was specially examined to get better results. The simulation shows that the method can optimize the structure of the network and shorten the learning time. Compared with traditional BP network, the relative error between the prediction by this model and the actual FeO value is reduced from 6534% to 1400%. Therefore, the accuracy of the new method is much higher than that by traditional BP network. The method can also be used on line.