Supercritical Unit Coal Water Ratio Research On Inverse Model Control Of Neural Network

In recent years,China's energy supply method has developed in a diversified way.Compared with other power generation methods influenced by different constraints,the thermal power unit is more agile in depth peak load cycling on peak demand.What's more,in order to protect the environment,the development of large capacity and environmental generating unit admits no delay.In the case of the thermal power unit,many parameters such as temperature and pressure will change while the load varies widely,the fixed PID parameter cannot used as the high quality control objects,and it's difficult to meet the operating requirements of the generating unit by adjusting the PID parameter online.Therefore,the neural network inverse control method is proposed.First of all,this thesis analyzes the control characteristics of ultra-supercritical unit according to the features,control characteristics and coal-water ratio of ultra-supercritical once-through boiler.Secondly,it emphatically introduces the coal-water ratio control system of ultra-supercritical unit.The internal regulation and control of power plant fuel-water ratio is constituted by water supply control circuit,fuel-water ratio control command and fuel control circuit.At the same time,the inverse control method of neural network is proposed.Through modeling and identifying the control method can know,the neural network inverse model learns by offline learning,and the training error can fill the bill theoretically,and the nonlinear system input system has been tracking by the output.It is concluded that the neural network inverse model can successfully identify the nonlinear partial system which can meet the requirement.Finally,this thesis analyzes the influence of fuel-water ratio to the control characteristics of the neural network inverse model.Under three different conditions,the neural network inverse model and the conventional PID model are compared form static control,anti-jamming capability and robustness.The simulation results show that the neural network inverse model has good control effect,robustness,positional following and anti-jamming capability in fuel-water ratio of ultra-supercritical unit.