Modeling Of Boiler Steam Turbine Based On Adaptive Neuro-fuzzy Inference System

By the end of 2018,thermal power still occupies a dominant position in the current operation of generating units in China,but the delay of the thermal object itself is large,and there is a strong coupling between the physical objects,so it is difficult to establish a more accurate model.In this paper,the boiler steam turbine which is a common thermal control system is modeled.The model of Bell-A boiler steam turbine is widely used in the study of the effectiveness of advanced control of thermal objects,and its model is a nonlinear model.A 300 MW boiler steam turbine unit in China is a linear model.The modeling data used in this paper are generated by these two existing mathematical models,and then modeled respectively.The boiler steam turbine system is a complex nonlinear system,and it is often very difficult to establish an accurate model for the nonlinear system,so the working state of the nonlinear system is often divided into several regions in engineering.establish a local model corresponding to each region.Adaptive neuro-fuzzy system(adaptive neuro-fuzzy inference system,ANFIS)was proposed by Jang in 1993,its own principle is similar to doing many local approximations,and has good ability of nonlinear function approximation,so this paper uses ANFIS to model the Bell-A boiler steam turbine of160 MW and a domestic boiler turbine unit of 300 MW.In addition,the basic ANFIS is improved by Fletcher-Reeves update method,and the improved effect is verified by modeling.The 160 MW model is a nonlinear model and 300 MW is a linear model.Basic ANFIS and improved ANFIS are used to model 160 MW and 300 MW boiler steam turbine units respectively.It is found that the modeling accuracy reaches the level of practical engineering application,which verifies the effectiveness of the modeling.Moreover,the modeling effect of the improved ANFIS is better than that of the basic ANFIS,so the effect of the improved algorithm is also verified.