Model Identification In Thermal Processes And Optimizing Research In Coordinated Control

The control of power generation unit is a necessary measure to ensure the safety and economic operation of thermal equipment.Thermal system optimization consist of model building and controller design.Traditional model identification of the thermal process is mainly based on the test of dynamic characteristics of the process.Because of the characteristics of the power generation unit,the dynamic characteristic test in the field is often difficult to implement.Therefore,research on thermal process identification based on field data has been focused.Meanwhile,the research on improving the performance of neural networks is need for further research.Optimizing the parameters of PID controller is an effective method to improve control performance of coordinated system as PID controlling algorithm is most widely used in coal-fired power units at present.In spite of the predictive control algorithm has been applied to coordinated system in some text,the proper utilization of the coupling between boiler and turbine is worthy of researching.Therefore,the content in this paper consisted of thermal process model identification and PID/predictive control algorithm research has important theoretical significance and application value.The main research contents and results are as follows:1.An improved neural network training index composed of errors and error rate is proposed,as the traditional neural network modeling ignores the continuity between data.The second chapter details deducing the formula of weight matrix based on the advanced index.The results of simulating study indicated that compared to the traditional index,the neural network trained by the improved index has abilities to improve generalization and model quality at the same identification accuracy.2.An improved pruning strategy is proposed,it deletes input nodes based on the fundamental theory of sensitivity pruning,and avoid inadvertent deletion on account of avoiding hidden node sensitivity calculation.It deletes hidden node based on relevance pruning,and avoid different delete rules as nodes belong to different layers.The simulation results verify the effectiveness of the strategy.3.In consideration of the character of coordinated system,an index contains error rate is proposed in PID control system,according to the influence of different parameters and the requirement of different systems,an improved index is proposed.The results of simulating study indicated that improved indexes can make full use of heat storage and improve performance of PID control system.4.A new index contains error rate is proposed in predictive control system,the third chapter details deducing the formula of control increment,an adaptive index is proposed according to the influences of parameters.The results of simulating study indicated that the performance of new predictive control algorithm is superior to the traditional predictive control algorithm,the adaptive index can make full use of heat storage5.A thermal commissioning process is presented on MATLAB/SIMULINK.The process is concluded of data collection neural network modeling、transfer function extraction、controller design and nonlinear control system establishment.