Abstract:Deep learning is an appropriate methodology for modeling complex industrial data in the field of soft sensors, owing to its powerful feature representation capability. Given the nonlinear and dynamic nature of the process industry, the key challenge for soft sensor technology is to effectively mine dynamic information from long sequences and accurately extract features of relevance to quality. A dual temporal attention mechanism-based convolutional long short-term memory network (DTA-ConvLSTM) under an encoder-decoder framework is proposed as a soft sensor model to acquire quality-relevant dynamic features from serial data. Considering different influences of process variables for prediction at multiple time steps and various locations, ConvLSTM and temporal self-attention mechanism are utilized as the encoder to adaptively fuse spatiotemporal features and capture long-term dynamic properties of process in order to capture the trends of industrial variables. Furthermore, a quality-driven temporal attention mechanism is employed throughout the decoding process to dynamically select relevant features to more accurately track quality changes. The encoder-decoder model meticulously analyses the interactions between process and quality variables by incorporating dual-sequence dynamic information to improve the prediction performance. The validity and superiority of the DTA-ConvLSTM model was validated on two industrial case studies of the debutanizer column and sulfur recovery unit. Compared to the traditional LSTM model, the proposed model demonstrated a substantial improvement with the accuracy R2 up to 97.3% and 94.9% and the root mean square error reducing to 0.122 and 0.022.

Residual-aware deep attention graph convolutional network via unveiling data latent interactions for product quality prediction in industrial processes

Self-adaptive Graph Convolution Networks with Application to Industrial Soft Sensor Modeling

Layer-Wise Residual-Guided Feature Learning with Deep Learning Networks for Industrial Quality Prediction

Multivariate Deep Reconstruction Neural Network for Multi-step-ahead Prediction of Industrial Process Quality Variables

Path Enhanced Bidirectional Graph Attention Network for Quality Prediction in Multistage Manufacturing Process

Deep Learning for Industrial KPI Prediction: when Ensemble Learning Meets Semi-Supervised Data

Causal Variable Selection for Industrial Process Quality Prediction Via Attention-Based GRU Network

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

Concurrent Analytics of Temporal Information and Local Correlation for Meticulous Quality Prediction of Industrial Processes

Deep Latent Variable Predictive Modeling with Online Bayesian Soft Attention Mechanism

Gated Stacked Target-Related Autoencoder: A Novel Deep Feature Extraction and Layerwise Ensemble Method for Industrial Soft Sensor Application

Product Quality Time Series Prediction with Attention-Based Convolutional Recurrent Neural Network

Improving Data-Driven Inferential Sensor Modeling by Industrial Knowledge: A Bayesian Perspective

A Multiphase Dual Attention-Based LSTM Neural Network for Industrial Product Quality Prediction

Deep learning for quality prediction of nonlinear dynamic processes with variable attention-based long short-term memory network

Variational Inference over Graph: Knowledge Representation for Deep Process Data Analytics

Dual Temporal Attention Mechanism-based Convolutional LSTM Model for Industrial Dynamic Soft Sensor

Graph convolutional network soft sensor for process quality prediction

A new approach for quality prediction and control of multistage production and manufacturing process based on Big Data analysis and Neural Networks

Quality-Driven Regularization for Deep Learning Networks and Its Application to Industrial Soft Sensors

Deep Nonlinear Dynamic Feature Extraction for Quality Prediction Based on Spatiotemporal Neighborhood Preserving SAE