Inner-phase-evolution-traced statistical modeling and online monitoring for uneven batch processes

Most batch processes have multiple phases with different characteristics. Within each phase, processes usually evolve following certain underlying rules, called inner-phase evolution here. For normal processes, these evolution rules must be obeyed, and any violation of the inner-phase evolutions is deemed to be abnormal. In this paper, a new statistical modeling and online monitoring method is proposed by combining principal component analysis (PCA) and qualitative trend analysis (QTA) to trace inner-phase evolutions of batch processes. By this method, inner-phase evolutions are traced, offering more information about whether the process is operating under normal status. Meanwhile, the problem of uneven-duration batches can be handled. Transitions are also modeled and monitored effectively. A chart showing the evolutions of variable contributions to a fault is designed for fault diagnosis. This method is applied to an injection molding process, revealing satisfactory monitoring and fault detection results.