| Main steam temperature is of great significance for the safety, stability and economy of power unit. Therefore, the requirements of it are very strict, and generally require deviation from setpoint remains within ?? 5C. However, the object has the properties of large time delay, large inertia, time-varying and many disturbance factors. Especially in recent years, the unit becoming larger, the object gets more delay and more complex. Based on the feedback, traditional PID can not regulate large delay object, for it often causes big overshoot. And using fixed parameters in the changing production environment, the control effect is not ideal.This paper studies the application of predictive control in main steam temperature system of 300 MW power unit. For the characteristics of main steam temperature, generalized predictive control is used. The basic principle is: forecasting model, rolling optimization and feedback correction. Firstly, the methods and processes of modeling are summarized. According to the data getting from step disturbance in 300 MW simulation machine, transfer function of reducing water and main steam temperature is established, using different model structure and identifying parameters with particle swarm optimization. Secondly, transfer function is discreted to differential equation to meet the demand of GPC. Taking into account the practical limitations of the valve, the calculated input is constrainted. The impacts of parameters in control are also discussed. Finally, GPC needs to solve Diophantine equations, resulting in computational complexity, so that this paper presents an improved algorithm. Form of forecasting model is changed from CARIMA to a more common CARMA and derivation and calculation are simplified. Simulations prove the feasibility of improved GPC in noise and external disturbances.The results show that: predictive control can effectively control the large delay object and has strong robustness. |
PDF Full Text Request