Research On Intelligent Compensation Optimization For The Set Values Of Boiler Superheated Steam Temperature System

Due to their advantages of environmental protection and high efficiency,supercritical and ultra-supercritical coal-fired generating units have become the dominated units in the power grid of our country.At present,under the increasingly severe problem of new energy consumption and the load dispatching system centered on the power grid,the deep load-changing operation of large-capacity power units has become frequent situation,which puts forward higher requirements for the control level of the thermal power units.As an important parameter which affects the safety and economy of the boiler,superheated steam temperature being either too high or too low or fluctuant too much,will affect the normal operation of the unit.Water-Spray desuperheating is the most commonly used boiler steam temperature control method,cascade PID control scheme is widely adopted to improve its control quality.Because of the non-linearity,large-inertia and long-delay characteristics of the superheated steam system,the cascade PID control method with fixed parameters is often unable to achieve the required temperature control effect when the unit load changes greatly,and the online tuning of PID parameters is time-consuming,laborious and difficult to realize in actual operation.Therefore,based on the historical operation data of the unit and the neural network modeling method,a real-time intelligent dynamic compensation optimization scheme for the steam temperature set value is carried out on the top level of the PID control loop,to improve the steam temperature control effect of the water spray desuperheating system without changing the parameters of the PID controllers.In this thesis,a compensation optimization scheme of superheated steam temperature set values based on neural network prediction model was proposed,and detailed experimental tests were carried out with the aid of a full-scope simulator.Firstly,the working principle of neural network,BP algorithm and its realization in MATLAB were introduced,and the characteristics of superheated steam system and the influencing factors of superheated steam temperature were expounded.On this basis,the superheated steam temperature prediction models for both the first-stage and the second-stage desuperheaters were established with neural network method,and the models were verified with the simulator.Then,on the premise of not changing the original steam temperature control logic and PID parameters,a set values dynamic optimization compensation strategy which combines feedforward compensation based on prediction model with error-feedback compensation was designed on the top level of the steam temperature control loop.The real-time optimization compensation scheme of steam temperature set values was programmed based on MATLAB,and the control simulation experiments were carried out in detail with the aid of supercritical unit simulation system.The results showed that,compared with the original control,the control quality of the superheated steam temperature is obviously improved in terms of the overshoot and the adjustment time with the intelligent set value optimization compensation scheme.