Predictive control for boiler-turbine unit using ANFIS

Coordinated Control System(CCS) in thermal power plant is a system with big inertia, large time delay and slow parameter variance as well as the property of fast parameter variance while the unit load changes. Conventional PID controller and Direct Energy Balance(DEB) Strategy which is tuned at typical operating point can hardly work well at different unit load. By using the resource of Automatic Generating Control(AGC), this paper proposes a novel predictive control of CCS based on Adaptive Neuro Fuzzy Inference System(ANFIS). Aiming at a 500 MW thermal unit, multiple linear models are obtained at several typical operating points. Corresponding multivariable predictive controllers are designed for the linear model group, which are used to train an ANFIS controller, thus ANFIS based multi-model preview predictive control is realized for CCS. Computer simulation results show that the new control strategy is very effective for CCS. It makes that the load has good tracking property, main steam pressure gets better dynamic and static performance, control input value is very smooth also. Very strong robust performance is obtained.