The Modeling Method Study Of Coke Production Process Quality Model

The coke production process is a complex industrial process with characteristics of serious nonlinear, time-varying, multi-parameters and uncertainty. Coke is the raw materials and fuel of metallurgical, machinery and chemical industry. Its mass is directly related to the stability of the follow-up of industrial production. The coke quality modeling is necessary to realize quality control of the coke production process. As for the coke production process, due to its complexity coupled with industrial noise pollution, the quality of their model is very difficult. In this paper, coke production process quality modeling is conducted in-depth research, and made some meaningful results. The main work and contents are as follows:Based on the analysis of coke production process, the quality modeling problem of the coke production process is proposed. The main factors affecting the coke quality and coke quality indexes are analyzed and discussed in depth to determine the input and the output of the quality model. And the coke production process data pretreatment is done.Based on the analysis of the adaptive genetic algorithm and BP neural network algorithm, the basic adaptive genetic algorithm is improved and the modeling method of coke quality model based on BP neural network is given. The simulation results show that the proposed method converges quickly, and its prediction accuracy and prediction hit rate is high.Based on the analysis of particle swarm optimization algorithm and RBF neural network algorithm, and reference on cross-ideological of genetic algorithms, particle swarm optimization algorithm is improved. Combined the advantage of PSO with that of RBF network, the idea of particle swarm optimization RBF network is introduced to the modeling. The coke quality model of particle swarm optimization RBF network is established. The simulation results show that the prediction accuracy is higher and the prediction effect is better.Finally, the comparative analysis was done for two modeling methods and quality prediction results. On this basis, the coke quality model output optimization problem is proposed, and the analysis and discussion is carried out. The solid foundation is provided to further do the coke quality optimization control.