Performance Assessment Of Industrial Cascade Control Systems

Process control system is the basis of industrial production automation and plays a crucial role in the modern manufacturing. At the beginning, control systems are always running in good conditions in practical industrial process. However, the performance of the system will degrade due to the lack of maintenance, equipment wear, system failure and many other reasons which not only reduce the economic benefits but also endanger the safety of the enterprise. Therefore, the performance assessment of control systems has attracted a majority of scholars in the past two decades and it is to be a hot issue about how to evaluate the system by closed-loop operating data.This thesis researched on performance assessment of industrial single-loop control systems and cascade control systems. The main contents of this thesis include as follows:(1) A brief review of the subspace identification and a detailed summary of the orthogonal and oblique projection which laid a foundation for the application of the subspace identification technique to the performance assessment.(2) The method of joint input and output subspace identification was applied into the single-loop control systems to get the process model, noise transfer function and process delays. The MV benchmark and LQG benchmark were used to evaluate the single-loop control systems. The MV benchmark had the disadvantages of strong control action, poor robustness and conservative evaluation compared to the LQG benchmark according to the simulation results. What’s more, the LQG benchmark not only takes the output variance into consideration, but also takes account of control costs. Thus, the LQG benchmark is more practical.(3) The performance assessment of cascade control systems based on the LQG benchmark was proposed. The joint input and output subspace identification algorithm was used to estimate the process model, noise transfer function, the inner loop controller and the outer loop controller. And three possible scenarios could be considered to evaluate cascade control systems. First, none of the inner and outer controllers were known and the goal was to design both controllers to get the global optimum. Second, the outer controller was considered to be fixed and one was looking for the best achievable performance by designing the inner controller. Third, the inner controller was considered to be fixed and one was looking for the best achievable performance by designing the outer controller. These three scenarios were used to evaluate the performance of the global, the inner controller performance and the outer controller. According to the simulation results, performance assessment of cascade control systems based on the LQG benchmark not only can get the global performance of the system, but also can show how much improvement may be achieved by designing inner and outer controllers.(4) A set of software for performance assessment was developed based on MATLAB/GUI. The effectiveness of the software was illustrated by applying it to the single frame strip gauge control system.