The Decoupling Control Study And Stability Margin Evaluation Of Multivariable Uncertain Systems

Decoupling control is an old and living topic. Meanwhile, uncertainty is an ubiquitous troublesome in practice. In this dissertation, the linear multivariable system with uncertainty is regarded as the main research object. We lucubrate three representational decoupling theories, and carry out the in-depth analysis and comparison from two sides :the stability robustness and decoupling performance. These decoupling theories are: the decoupling control theory which was firstly put forward by Morgan; the decoupling control theory based on eigenstructure assignment and the decoupling theory based on H∞ control.The main works done in the dissertation are as follows:1. Retrospect the development and discern the status of decoupling control by referring to a lot of literatures.2. Under the condition that the number of input and output is equal, state feedback decoupling control method and output feedback decoupling control method based on Morgan's theory are presented. The former involves two schemes and the latter involves three schemes. The decoupling control system designed on this theory is not robust.3. The decoupling control theory based on eigenstructure assignment is studied, also includes two cases: full state feedback and output feedback. We discuss the nonrobust decoupling control method, the sufficient condition of stability robustness and sufficient condition of performance robustness successively. We also use the optimization index based on these two sufficient conditions to design the robust decoupling controller for control systems with structured state space uncertainty.4. We study how to translate the robust decoupling control into standard H∞ control by using the loop shaping technique of sensitivity and compensatesensitivity. Firstly the robust decoupling control method with single or double kinds of uncertainty is derived just by the loop shaping of those two sensitivities, this method can deal with both uncertainty and disturbance. Then we bring forward a newmethod , which has the advantages of both sensitivity loop shaping techinique and model matching technique, includes two schemes, one is for the system with unstructured uncertainty and disturbance, the other is for the system with mixed uncertainty and disturbance.5. A stability margin's evaluation method for the multivariable system which is applied to the engineering is studied. The margin is evaluated by the minimum singular value of return deference matrix of system. By analyzing and comparing those three decoupling methods ,we come into two conclusions: one is that decoupling performance and stability robustness are contradictory, the other is the nonrobust decoupling control method based on Morgan's theory and which based on eigenstructure assignment theory is equivalent.6. Utilizing the new robust decoupling control method which is put forward above, the design problem of decoupling controller for large angle of attack reentry missile's attitude control system is mainly investigated. By considering the presence of large aerodynamic uncertainties, resilience vibration, and disturbance, we discuss the necessary of decoupling control and design the robust decoupling control system, The simulation results and stability margin evaluation both show that control system can obtain strong robustness and good performance. It is proven that this new method provides an effective way for the decoupling of vehicle control system.The main innovations of this dissertation are as follows:1. An new method which have the advantages of both sensitivity loop shaping technique and model matching technique is bring forward. It includes two schemes, one is for the system with unstructured uncertainty and disturbance, the other is for the system with mixed uncertainty and disturbance.2. Decoupling performance and stability robustness are contradictory, we draw this conclusion by using stability margin's evaluation method which is based on the minimum singular value of return deference matrix for the multiva...