Multi-model Identification And Control Application Research On Complex Indurstrial Process

Multi-variable,strong nonlinearity and environment disturbance are problems existed in complex indurstrial process.The traditional modeling methods usually can not reflect the whole process in this circumstance beacuase its simple structrue,which lead the control result can meet the need.So,the research on the multi-model identification and its correspoding control strategy for complex indurstrial process becomes hotspot and difficult thing in related fields.Multi-model identification and control application problems are studied in this paper based on the existed work.Agent multi-model and online clustering multi-model identification methods which procide new idea for multi-model identification is presented to solve the drawbacks of the traditional methods.In addition,the indirect adaptive fuzzy control and generalized predictive control based on the corresponding identification multi-model are designed.At the same time,the numerical example or application simulations are given following each algrithm.At last,the algrithm is applied to 660 MW power plant swirl burner for its identification multi-model and control system.The mainly research content and are as follows:Firstly,combine agent with fuzzy T-S model,an agent fuzzy multi-model system is investigated,and it is proved that the agent based multivariable fuzzy T-S multi-model can approximate any linear or nonlinear process at arbitrary accuracy.Other than the traditional multi-model methods,each agent in this system represents a work condition,which can be described with a dynamic equation.The agent can execute a task alone or collaborate with other agents,which brings a considerable flexibility to describe the complex system.The viability and the efficiency of the algorithm is demonstrated in an electrical water heater system.Secondly,for the multi variables nonlinear system under disturb noise,an online identification algorithm for multiple model based on minimum entropy clustering is investigated.The number of local models and corresponding weights is calculated by the entropy based fuzzy subtractive clustering approach,and the regularity degree of the local system is considered along with the clustering process.Parameters of local models could be estimated online by the weighted recursive least square method,it realize the online identification of the local model,and the validity is tested in the numerical example.In addation,the initial algorithm for model parameter identification is studied,and the identification results of electrical valve circuit in the cooling water system show the self-adaptability for uncertain external disturbances.Thirdly,an adaptive fuzzy controller is designed based on the Lyapunov function.The optimal fuzzy adaptive control law that can make sure stability is given to the multi-model object which constructed by combining agent with fuzzy logical.A simulation experiment is presented for waste heat utilizing process,and the comparition with traditional PID controller is conducted to test the performance of the algorithm.Fourthly,the identification and control system is studied for 660 MW power plant swirl burner based on the entropy clustering method and generalized predictive controller.The structure and running principle of the swirl burner are analyzed,and the main variables are determined,then a fuzzy multi-model system whose parameters can be adjusted along with the turning process is presented,the parameter identification approach is listed.The MIMO generalized predictive control strategy for the swirl burner is complished.At last,to prove the validity of the method the strategy is used to the power plant swirl burner.