关键词: 炼焦生产过程, 多目标遗传算法, 优化控制, 主成分分析, 灰色关联分析, 神经网络

Abstract: A multi-objective and layered optimization control strategy was proposed to deal with the optimization control problem of coking plant production process. Firstly, the correlation models between the comprehensive production targets and the local optimization targets were established with the principal component analysis, the gray correlation analysis and the improved Back Propagation (BP) neural network methods. The comprehensive production targets were mapped into local optimization targets. Secondly, a multi-objective optimization model was constructed with the maximum of coke yield and minimum of energy consumption as the optimization objective, the coke quality and production boundary as constraints, and the local optimization targets as decision variables. Then the multi-objective genetic algorithm was applied to solve this multi-objective optimization problem to obtain local optimization targets. Finally, the local optimization targets were served as the set values in the process control of subsystems and decision variables in the optimal scheduling system in order to realize the expectant comprehensive production targets by making dynamic adjustment of operation parameters. The practical operation results showed that the optimization control strategy proposed had good application effects.

Key words: coking plant production process, multi-objective genetic algorithm, optimization control, principal component analysis, gray correlative analysis, neural network