A lwpt-dtw-based method for unequal long-term synchronization of intermittent processes

Abstract

The invention discloses a batch process unequal-length time period synchronization method based on LWPT-DTW (lifting wavelet package transform-dynamic time warping), and belongs to the technical field of batch process modeling and monitoring. According to the method, the LWPT is used for performing high-low-frequency multi-stage decomposition on batch process unequal-length time periods obtained after the use of a K-mean clustering algorithm; a coefficient matrix at different frequency bands in different sub time periods is obtained; then, the DTW is used for synchronizing the frequency band coefficient matrix; the synthesis is performed through ILWPT (inverse lifting wavelet package transform); the equal-length time period after the synchronization is obtained. All information in the batch process original time period is sufficiently extracted; the influence of a Gibbs phenomenon on the signal synthesis is reduced; the synchronization of the batch process unequal-length time period is realized; the synchronization precision is improved; the operation speed is high; the operation is stable; reliable data support is provided for the batch process unequal-length time sub period modeling and monitoring.

Description

technical field [0001] The invention relates to a method for synchronizing an unequal time period of an intermittent process, belonging to the technical field of intermittent process modeling and monitoring, and in particular to a method for synchronizing an unequal time period of an intermittent process based on LWPT-DTW. Background technique [0002] Batch process is an important production method in modern industry, and has a wide range of applications in fine chemicals, pharmaceuticals and food. Batch processes have complex process characteristics, and multi-period is an inherent feature of batch processes. If a single batch of multi-period batch processes is regarded as a whole for process modeling and monitoring, it will be difficult for the established process model to accurately describe the batch process. And cause a high rate of false alarms and missed alarms in the monitoring of the batch process; at the same time, the operation time of the corresponding sub-perio...

AI Technical Summary

Problems solved by technology

The batch process has complex process characteristics, and multi-period is an inherent feature of the batch process. If the single-batch multi-period batch process is regarded as a whole for process modeling and monitoring, it will make it difficult for the built process model to accurately describe the batch process. And resulting in a high false alarm rate and false alarm rate of batch process monitoring; at the same time, the operation time of different batches corresponding to sub-periods of the batch process is not synchronized due to the difference in the quality of raw materials and the difference in initial condition settings, and it is difficult to meet the requirements based on MPCA's batch process monitoring requires the consistency of process data length, so solving the problem of synchronization of batch processes with different long periods can provide reliable process data for MPCA-based batch process monitoring

[0003] Dynamic Time Warping (DTW) can dynamically compress and expand the trajectory according to the local characteristics between the two trajectories, and then make the lengths of the two trajectories consistent. However, the DTW algorithm has a large amount of calculation and low efficiency, which restricts its practical application. ; Although the MSDTW (Multi-scale DTW) constructed by DTW and discrete wavelet transform can improve the running speed of the DTW algorithm, the MSDTW algorithm only performs multi-level decomposition on the low-frequency part of the signal, and does not mine the useful information of the high-frequency part. Discrete wavelet transform will appear Gibbs phenomenon when processing the signal with abrupt point, which will increase the synthesis error of the signal, so that the MSDTW algorithm has certain limitations; Synchronization over long periods of time

Images