Controller performance assessment of batch processes based on dynamic time warping

This paper reports on the application of dynamic time warping (DTW) to evaluate controller performance of batch processes. Batch processes often are characterized by unequal durations due to changing batch-to-batch operational conditions. To compare conveniently the performance different batch runs and different controllers, one requires equal batch duration. Therefore, batch data lengths are synchronized by compressing and expanding batch trajectories of various lengths in order to provide a unified controller performance index. Synchronization can be achieved by using DTW algorithm. The DTW algorithm is a nonlinear and flexible pattern matching technique. The applicability of the proposed method for controller performance assessment of batch processes is illustrated through an industrial example.