Adaptive Predictive Control And Its Application To DCS In Power Plants

In recent years DCS has been introducing into power plants in China widely, and an environment has been provided for the application of advanced control strategies. However, due to the high complexity, the unique requirement for safety and reliability, the relative blockade of DCS in the large-scale power plant, the application of advanced control strategies in DCS is somewhat stunted. Under the support of State 863 project, he Application of Advanced Control Strategies to DCS in Large-scale Power Plant and the Development of Control Software Package? the research of adaptive predictive control theory and its application to DCS in large power plant have been carried out in this paper. The dissertation is dedicated to solve the practical control problems in power plant on the basis of theoretical research. Under the existence of the dynamics characterized by non-linear, significant delay and time varying, the work strives to deal with the problems of significant delay objects control in large power plants, which have puzzled the power plants for long time. Smith estimator is a classical method to deal with control problem of significant delay. The main disadvantages are poor robustness and no systematic way to determine controller parameters. The model reference adaptive predictive control method (MRAPC), which was been put forward first by Prof.Zhi-yuan Yang, combines the model reference adaptive method with Smith estimator ingeniously and great successes in theory research and practical applicant in power plants have been achieved. Based on which mentioned above, the works in this thesis are to continually study the theory problem in MRAPC and put forward the improved algorithm. In the meanwhile, spreading the successful experience of MRA PC's practical application in power plants to DCS is another important content. The organization of the dissertation is as follows. Firstly, based on the analysis of the tuning process of MRAPC for systems with time-delays, the conception that time-delays exist in MRAPC tuning loops has been described. Then an improved MRAPC algorithm with tunable parameters has been put forward. The new modified approach, which moves all or part tunable parameters forward to reference route in adaptive structure and improves the means of time-delay implement in computers, is proposed to solve this problem. Secondly, in view of the identification problem after the change of MRAPC structure, the research has been extended to a kind of continuous system with time delay, and the robust convergence analysis for this model reference adaptive identification algorithm has been given. Using the intermediate value theorem in this analysis, the approximate problem in original analysis has been solved partly. Moreover, the robust convergence process for this model reference adaptive identification or control problem with uncertain parameters has been discussed by using frequency analysis methods. Based on the internal model control and modem robust control, a new tuning algorithm of one-dimension robust PJD controller for the special structure of improved MRAPC systems is presented in this paper. Moreover, an easy PJD on-line tuning method, which has a PID parameter-list only based on inertia time of controlled process, is been put forward for MRAPC self-tuning structure, too. Based on the analysis of the actual curves for power plant controlled processes (such as steam temperature control process) and interfe...