Spatio-temporal Multi-models Based Process Monitoring and Control

Complex distributed parameter systems (DPS) widely exist in production industry. Except the complex spatio-temporal nature, the stronger nonlinearity will make the process dynamics change more significant between operating conditions than the traditional lumped parameter system. Since all the previously developed models work only at vicinity of the operating condition, tracking control between operating conditions will be more difficult. Furthermore, the final performance of the process, which is actually a complex function of process variables or other unknown factors, can not be measured directly. All these difficulties make the process monitoring nearly impossible.We aim to develop a novel spatio-temporal multi-models based process control and monitoring for the DPS, with application to the cure process. The model developed should be able to work in a wider range of operating conditions, so that the cure process can be globally simulated. First, the whole operating region will be divided into several small zones, upon which a spatio-temporal Wiener model is developed for the local modelling, and then integrated by the fuzzy system for a global model. Based on each spatio-temporal Wiener model, a low-order observer will be designed to estimate dominant states; then the observer-based local controller will be developed to stabilize the spatio-temporal process at each operating condition. These local controllers will be integrated by a T-S fuzzy system to maintain a satisfactory tracking performance in the whole operating region. Finally, a SPC-based expert system will be developed to work with the global fuzzy multi-models to on-line monitor performance variance and capture any quality drift caused by the fault or aging components during the production.