Hybrid Model/data-based Robust Design for Unknown Dynamic Systems

In order to reduce product cost for the increasingly competition, we were requested by ASM - a local industry to design the snap cure oven without temperature controller during the cure process. This poses a great challenge to the fundamental research on system design because the off-line design has to play some roles of online control functions.One of difficulties in system design is to minimize the influence from uncontrollable variations and the model uncertainty, and another is to design the dynamic system for both stability and robustness. The goal of this project is to develop an effective approach for robust design for the unknown dynamic system with application to the selected industrial process – snap cure oven. The nonlinear complexity of unknown systems will be separated into the approximate linear structure. Then, all the model uncertainty and design variations can be expressed using the perturbation sensitivity matrix, and estimated from the process data. With help of the approximate model, a relation between eigenvalues of the matrix and design variables can be established for system design. For the dynamic system, the traditional system design method has to be modified to consider the complex egienvalues. The designed process should be able to have a satisfactory dynamic perfomance without on-line control.