Multi-Subsystem-Entropy Based Operating-States Identification For Combustion Process Of Coke Ovens

Coke oven temperature is an important parameter influencing coke quality and energy consumption. In combustion process many uncertain factors change heat level and operating states of coke oven, which affects stability of coke oven temperature. Identifying operating states online and taking control strategy based on operating states is meaningfull for improving performance of control system.After making a deep analysis of mechanisms in combustion process of coke oven this paper defines operating states in combustion process and heating subsystem. Aiming at periodicity, dynamic, and uncertainty in coking process a method based on qualitative trend analysis and fuzzy partition for coking stage judgment is proposed. First the trend of raw gas temperature in a single coking process is analyzed offline and maps between trends and coking stages is established. Then considering transition between coking stages, fuzzy membership degree is taken to divide coking stage and expert rules for coking stage judgment is established. Finally according to the trend obtained online stage of coking chamber is judged by reasoning.Based on judging stage of coking chamber and analyzing its effects on heat state of subsystem a method based on fuzzy reasoning for judging heat state of subsystem is proposed and uncertainty of subsystem is depict correctly. In combustion process each subsystem has different effects on operating states and all subsystems codetermine operating states. Therefore, a method based on information entropy of multiple subsystems is proposed to identify operating states. Subsystem’s objective weight is determined by information entropy and used to evaluate the importance in operating states identification. Finally operating state in combustion process is judged comprehensively by contribution and objective weight of all subsystems.In this paper operating-states identification algorithms for combustion process of coke oven is simulated and the feasibility is verified. Then coke oven temperature control system based on multiple operating states is established and actual running results show that performance of the new system based on operating-states identification is superior. Therefore, effectiveness and practical value of the classification and identification method proposed in this paper is verified.