Online monitoring and fault diagnosis for uneven length batch process based on multi‐way orthogonal enhanced neighborhood preserving embedding

Abstract

In the practical batch process, the duration of each batch is probably different, and the key event happened may also vary from batch to batch. This paper proposes an effective fault diagnosis algorithm to synchronize the batch runs with uneven duration and keep the key features that reflect the running state of batch process. To address the batch out of sync, the relaxed greedy time warping is used to accurate on-line synchronization of ongoing batch and avoid evaluating the optimal path every time a new sample is available. For online monitoring and fault diagnosis, an important step is how to extract key features that reflect the running state of batch process, on which basis a weight matrix considers the distance between the neighbor data is embedded to construct the enhanced objective function, and then the mutually orthogonal basis functions which represent the local geometrical structure are computed to extract key features of batch process. After a fault is detected, the contribution plot method is used to diagnose the variables that cause this fault. The effectiveness of the proposed algorithm is verified by the penicillin fermentation process.

A Fault-Tolerant Soft Sensor Algorithm Based on Long Short-Term Memory Network for Uneven Batch Process

“…In addition, various modifications of DTW have been utilized for aligning uneven batches of data [35,36]. Nevertheless, they run the risk of distorting the relationship between process variables [37]. Another common approach is to resample process variables using methods such as interpolation, which has now gained applications in areas such as fault diagnosis [38].…”

“…Another common approach is to resample process variables using methods such as interpolation, which has now gained applications in areas such as fault diagnosis [38]. The RGTW-MOENPE algorithm is introduced to align batch processes, having varied durations while preserving the essential features of critical events [37]. A dynamic multi-stage and multi-mode modeling approach is proposed [39].…”

Batch process monitoring based on sequential phase division multiway sparse weighted neighborhood preserving embedding

“…Different batches with uneven lengths are collected under normal operation condition. Then, the RGTW method is used to synchronize the ongoing batches [27,28]. For batches of data with uneven-lengths, a batch close to the middle length of the duration is selected as the reference batch F r .…”