A Phase Division Strategy For Multiphase Batch Process Monitoring Based On Particle Swarm Optimizer (Pso)

Batch process generally consists of multiple stages of operation. In view of such characteristic, multi-phase modeling and analysis help to improve monitoring performance and enhance understanding of the process. Although many partition algorithms have been put forward, these is no quantification index to evaluate the performance of phase model. Besides, the results may be affected by human factors since they have many tunable parameters. In order to overcome the above problems, an optimization based phase division algorithm is presented in this work. In the algorithm, a quantization index (the cumulative control limit error, CCLE) is first defined to evaluate the performance of the phase model. Then, the PSO-based optimization algorithm is used to search the optimal phase partition results for a given target phase number. Through the percentage of performance improvement threshold, the proposed algorithm achieves the compromise between modeling complexity and monitoring performance, and ultimately gets the number of phases. A chip packaging process are conducted to demonstrate the feasibility and effectiveness of the proposed algorithm.