An improved neural network method for the prediction of comprehensive production indices in coking process
炼焦生产过程综合生产指标的改进神经网络预测方法

A prediction method based on the improved back propagation(BP) neural network is proposed to solve the problem of large time-delay in the detection of the comprehensive production indices (quality and quantity of coke, and energy consumption of coke oven) in the coking process. First, the input and output variables of the prediction models are determined by analyzing the process mechanism correlation between process parameters based on principal components analysis and grey relational analysis. Then, the BP neural network based on an improved differential evolution algorithm is applied to establish prediction models, which are compared with the basic BP neural network prediction models. Finally, the prediction models are verified. Simulation results show that the proposed prediction models provide a better convergence rate and higher prediction accuracy, and the prediction effect of the obtained models satisfy the technological requirements.