Modeling And Control Of Intermediate Point Temperature For An Ultra Supercritical Unit

Ultra supercritical(USC)power generation technology has developed rapidly during the past decades,and will inevitably take the leading position in the future with high combustion efficiency and low pollution emissions.The intermediate point temperature is one of the core parameters of power units,which is a key point in the energy conversion process.It affects the operation and efficiency of the energy conversion,and therefore,is of great significance to safety and stability of the operation,as well as efficiency and economy of the development of the thermal power plant.However,with the improvement of the requirement of model accuracy and control performance,traditional modeling methods and control schemes have become hardly to achieve satisfactory performance for the industrial system.New theories and methods need be introduced to solve this dilemma.Therefore,a Fuzzy Recursive Least Squares(FRLS)algorithm is proposed to realize model identification.In order to achieve higher identification accuracy,a Fuzzy Particle Swarm Optimization(FPSO)algorithm is proposed.Furthermore,based on the obtained model,a Fuzzy Generalized Predictive Control(FGPC)algorithm is used as the control scheme.The main research contents and contributions of this dissertation are given as follows:(1)A Fuzzy Recursive Least Square(FRLS)algorithm with uniform partition is proposed based on the principle of traditional Recursive Least Square(RLS)algorithm.In order to cope with the weakness of the traditional RLS,the global region is uniformly divided to several local regions to reduce the nonlinearity.Then the local models are identified through traditional RLS.And the global model is obtained by fuzzy fusion according to the fuzzy rules and the weight calculation results.The simulation results show that the proposed FRLS with uniform partition performs better than traditional RLS with greater accuracy.(2)Aiming at breaking through the shortcoming of traditional RLS,and exploring the characteristics of operation region of the intermediate point temperature,a Fuzzy Particle Swarm Optimization(FPSO)with the structure of the fuzzy K-Means network is first proposed and applied as the model identification method.Considering the strong nonlinear dynamic operation process of the intermediate point temperature,the global operation region is divided into several local regions by the K-Means clustering algorithm,by which the nonlinearity is reduced to an acceptable range.Then,in order to identify the accurate local models,an Improved Particle Swarm Optimization(IPSO)algorithm is introduced,which is optimized by linearly decreasing inertia weight and the reinitialization strategy to avoid local extremum.After that,the global model of the intermediate point temperature is obtained by the fusion of the local models according to the fuzzy inference rules.Finally,detailed analyses on the model identification results through simulations are addressed to demonstrate the effectiveness of the proposed FPSO.(3)A Fuzzy Generalized Predictive Control algorithm based on the Fuzzy Particle Swarm Optimization(FPSO-FGPC)is proposed as the control scheme.In order to overcome the shortcomings of the standard Generalized Predictive Control(GPC)algorithm,such as the single method to obtain the controlled object parameters and the algorithm parameters cannot be changed according to the actual operation,FPSO-FGPC gives the following improvements: 1)The current operation region is judged by the antecedent variables,which is the last intermediate point temperature value.Then,the local model parameters are obtained and fused online to improve the accuracy of the controlled object parameters.2)The Particle Swarm Optimization(PSO)algorithm is used to search the optimal coefficients of the Generalized Predictive Control(GPC)in each local region,which improves the adaptability of parameters to different working conditions.Finally,the controlled object parameters and the optimal coefficients are used for the control law derivation.The simulation results show that the quality of control for intermediate point temperature is significantly improved and the robustness of the control system is enhanced by the Fuzzy Generalized Predictive Control algorithm based on the Fuzzy Particle Swarm Optimization(FPSO-FGPC).