Abstract:<p>Taking advantage of varying degrees of attention on specific features, attention-based long short-term memory (ALSTM) networks have made inroads into the industrial multivariate time series prediction sector recently. However, conventional ALSTM models usually employ static time-delays to constant select input/output pairings of multivariate data, which apparently ignores dynamics of transmission time between industrial process variables and degrades the prediction performance by incorrectly extracting process characteristics. In response to this problem, this paper proposes a novel approach to extracting dynamic time-delays to reconstruct (DTDR) multivariate data for an improved ALSTM prediction model. Therein, the temporal locations and spans of multivariate data are adaptively tailored to input/output pairings of the ALSTM network according to the dynamic time-delays. Specifically, the multivariate data can be accurately matched in temporal positions, and the data information in the original temporal spans with Status transfer time abnormal are replaced. Consequently, this prediction model not only appropriately utilizes dynamics between the predicting and correlated variables, but also makes better attentions on key features extracted from optimum data. Applied to industrial distillation and methanol production processes, the proposed method demonstrates the capability of significantly improving network training speeds as well as prediction accuracies in contrast to static time-delay based ALSTM and LSTM models, expecting even more applications.</p>

Multivariate Time Delay Estimation Based on Dynamic Characteristic Analytics

Investigation of Dynamic Multivariate Chemical Process Monitoring

Dynamic PCA-based Fault Detection and Diagnosis Analysis

Linear and Nonlinear Hierarchical Multivariate Time Delay Analytics for Dynamic Modeling and Process Monitoring

Cooperative Deep Dynamic Feature Extraction and Variable Time-Delay Estimation for Industrial Quality Prediction

Multi-mode process monitoring method based on PCA mixture model

A Fault Prognosis Strategy Based on Time-Delayed Digraph Model and Principal Component Analysis

Distance Between Vcs And Its Application To Step-By-Step Multi-Way Pca

Fault diagnosis in industrial processes using principal component analysis and hidden Markov model

Dynamic Processes Modeling and Monitoring Based on a Novel Dynamic Latent Variable Model.

DTDR–ALSTM: Extracting dynamic time-delays to reconstruct multivariate data for improving attention-based LSTM industrial time series prediction models

Autoregressive Dynamic Latent Variable Models for Process Monitoring

A Novel Dynamic Baysian Canonical Correlation Analysis Method for Fault Detection

A Generalized Multivariate Grey Prediction Model DGM(1, N, I) with Unknown Time Delays

A New Method of Dynamic Latent-Variable Modeling for Process Monitoring

Fault Detection Based on Multivariate Trajectory Analysis

New monitoring method for dynamic non-Gaussian process

On-line diagnosis of abnormal conditions of air separation process by dynamic PCA

Enhanced dynamic latent variable analysis for dynamic process monitoring

A model-free method for identifying time-delay characteristics of HVAC system based on multivariate transfer entropy

Evaluate Time Delay from Sensor's Data by Trend Similarity Search