A control system of multi-variable PID neuron network has been given by merging the decoupling control theory with the neuron network principle

and the limitations of PID decoupling control in multi-variable control system and the feasibility of intro- ducing the neuron network control system being analysed.A simulation test of temperature control for reheated steam and for outlet steam from the low-temperature superheater has been carried out by using the given control algorithm.Results of simulation show that the said control algorithm has better decoupling performance and self-learning control property for the reheated steam tempera- ture control

and the system has stronger robust in the event of parameters variation concerning the controlled object.