| Pressure system in coke oven gas-collecting pipe is a multi-coupling, nonlinear, strong disturbed, multi-variable and time-varying system. It is hard to control it in usual methods. Along with the development of control theory and the mature of computer technology, some advanced control strategies are adopted which combined fuzzy control and expert system. But fuzzy rules are generally created, which are in the principia of designing, depending on operators'knowledge and manipulators'experiences, based on experience conclusions and expert knowledge.In the multitudinous factors, which affect the coke oven gas-collecting pipe pressure, the change of some factors values has a small influence on the undulation of the pressure of the gas-collecting pipe, but the change of others has a great influence. Therefore, with the core identity of the Rough Set—attribute reduce, it can obtain the more influencing factors to produce the mechanism model of the pressure system about the gas-Collecting Pipe in the paper at first. After getting the mechanism model, this paper proposed a new way to acquire fuzzy rules with clustering analysis in data mining. Clustering can categorize to data by comparing variable, so the data having the similar characters are categorized together. So that lots of real time data is converted to class set by clustering.Through treating the original data which are collected from TangGang coking and gassing factory, we can gain the states of actuating equipment in different production conditions, called fuzzy control rules. Simuted in matlab6.5 and compared control effect with the constructing fuzzy control rules method given in this paper with control effect with fuzzy control rules constructed from experience conclusions. It is proved the method given in this paper is superior in rising time, super-regulating value and steady-state error, though it is comparable in peak value time.And it showed that the method given in this paper has itselt rationality and feasibility. |
PDF Full Text Request