基于子空间辨识的火电锅炉风烟系统预测控制研究-控制理论与控制工程专业论文.docxVIP

AbstractABSTRACT Abstract ABSTRACT With the increasing competition from the power market，the shortage of global resources，and the severe requirement on environment protection，it is quite urgent for the thermal power units to improve the operation level of the boiler combustion system，and to meet the need of energy saving and emission reduction．This dissertation mainly focuses on the advanced control strategy design of air system of thermal plant，based-on the theoretical research and industrial process data analysis． On the basis of Subspace Identification Method(SIM)，a novel method of Model Predictive Controller design is proposed for the air system of thermal power station． First of all，the basic framework and characteristics of SIM is analyzed systematically， including the fundamental theory and algorithms．It can be concluded that the classic SIM can be summarized into three key steps．Step l：The projection of the row space of specific data Hankel matrices is calculated．Step 2：with the tool of singular value decomposition，the extended observability matrix and a Kalman filter estimation of the state sequences can be obtained from the projection．Step 3：the system matrices are extracted from the observability matrices and／or the estimated state sequences． Then，the performance of SIM is improved by using Principal Component Analysis (PCA)．Meanwhile，the subject on how to determine the optimal order of state—space model i S also discussed． Model Predictive Control(MPC)technology has gained considerable application in industrial process for its good control performance．In this dissertation，the system architecture and algorithm theory of MPC is introduced，followed by MPC design method based-on state space model obtained by SIM．Then，by comparison between SIM and Model Algorithm Control(MAC)，a data—driven MPC design method is proposed by using input and output Hankel matrices to obtain impulse response sequences of the system directly． Air system is one of the most important