Performance Assessment Of Multivariable Control Systems

There are hundreds of control loops in a typical industrial process.The performance of the system will degrade due to the lack of maintenance,equipment wear and many other reasons,while market competitiveness strives the process of control systems to achieve optimal state.So the performance assessment of control systems is expected to a kind of effective tool to monitor performance for normal production.In this thesis,the performance assessment of multivariable control systems is researched so that effective ideas and methods of assessment are provided.The main contents of this thesis include as follows:The background,the purpose and the significance of the control system performance assessment is introduced;the criterion and the index of performance is introduced;the general situation of development about performance assessment is introduced.The subspace identification method and other relevant definition is introduced.The orthogonal and oblique projection are important for the subspace identification arithmetic.In order to get the subspace matrix,the regression analysis method is also needed.The subspace identification arithmetic based on single closed-loop data is deduced,getting the subspace matrix and the covariance matrix of the noise.The LQG benchmark is used to evaluate the single-loop control system.The simulation results prove the effectiveness of the arithmetic.Due to the difficulty of application of MVC benchmark in multivariable control systems,the subspace identification method is applied to this kind of system.The impulse response series is separated from the obtained subspace matrix.The condition where the process model has already obtained is researched and the method prove to be effective according to the simulation result.Based on the software MATLAB/GUI and the theory preparation about control performance assessment,a set of software for performance assessment is designed.The software can be applied to no matter SISO or MIMO.