Control of boiler-turbine unit based on adaptive neuro-fuzzy inference system

Boiler-turbine unit is a complex multivariable system with strong time-variation, uncertainty and coupling. The improvement of its control quality is a key problem for automation improvement in thermal engineering. Adaptive Neuro-Fuzzy Inference System (ANFIS) is the neural network realization for Takagi and Sugeno fuzzy inference system. ANFIS can not only approach any linear and nonlinear function with any precision, but also quicken convergence speed, decrease errors, and lessen training data that are needed. The paper put forward to use ANFIS as a controller to control boiler-turbine unit. Under several typical operating points, the Linear Quadratic Regulators (LQR) are designed and generate the training data for the ANFIS. Therefore, LQR control is realized based on ANFIS, on the coordinated control system of a thermal power unit, multi-model LQR control is approached. The simulation results show that the control system has better control performance and stronger robustness.