Xudong Shi

Chemical processes involve complex operational mechanisms and high-dimensional variables with multiscale dependencies and temporal covariate shifts, posing significant challenges for soft sensing implementation. To address these issues, a self-constrained weighted garrote ordered memory network with adaptive distribution learning is proposed for industrial soft sensor modeling. Specifically, an improved long short-term ordered memory (LSTOM) architecture with a composite hierarchical information update mechanism is designed to capture multiscale dynamic features from industrial time-series data. Meanwhile, an adaptive distribution learning module is embedded into the LSTOM architecture to match high-order feature distributions across different temporal periods via maximally dissimilar segments. With the composite hierarchical update mechanism and embedded adaptive distribution learning, the proposed approach improves generalization in soft sensor modeling under multi-timescale dependencies and temporal covariate shifts. Finally, the shrinkage coefficients of the self-constrained weighted garrote are embedded into the LSTOM’s input weights to eliminate redundant variables and promote structural sparsity. The effectiveness of the proposed approach is validated on a public industrial-scale penicillin fermentation process and a real-world flue gas desulfurization case.

Owing to the complexity and uncertainty of industrial production, unknown faults inevitably occur. However, most existing methods struggle to effectively identify unknown fault categories in a new operating mode, and often either treat known and unknown categories equally or exhibit a bias toward known classes. To tackle these issues, this paper proposes a Multi-Classifier-Weighted Adversarial Network for Open-set fault diagnosis in industrial processes. During the adversarial training between the feature extractor and the extended classifier, ambiguous samples exhibit a higher probability of being biased toward the unknown category. Furthermore, an auxiliary classifier is designed to compute the membership of samples to the unknown class, thereby enabling adaptive and dynamic adjustment of the bias intensity toward the unknown category during adversarial training. Experiments on two chemical fault datasets verify that the proposed model can simultaneously identify known and unknown faults, outperforming other diagnostic models.

With the development of industrial intelligence, deep learning-based soft sensors are widely used in complex industrial systems to estimate key quality variables in real-time, with long short-term memory (LSTM) networks excelling in nonlinear dynamic modeling. However, conventional LSTM networks face the challenge of redundant memory weight allocation, which hinders efficient capture of critical temporal dependencies. Moreover, the probability distribution of industrial process variables often exhibits dynamic temporal shifts, significantly deteriorating model generalization performance. In this study, a novel long short-term complementary memory network based on temporal distribution matching (TDM-LSTCM) is proposed for product quality prediction in industrial processes. First, a novel memory update rate gate is designed based on LSTM to obtain adaptive update rate factors for different time steps, regulating the information flow rate. Second, the memory update rate gate is embedded into the input and forget gates, leading to the development of a complementary input-forget regulation gate structure to optimize the cell state update mechanism. Furthermore, the TDM algorithm is incorporated into the training process through a dual-stage optimization strategy. During the distribution characterization stage, the training dataset is partitioned into distinct periods with the most pronounced distributional discrepancies. In the distribution matching stage, cross-period discrepancies in hidden states are minimized to facilitate the extraction of common information. Finally, the proposed methodology is validated through a penicillin fermentation process and flue gas desulfurization system. Experimental results demonstrate that the developed approach outperforms other modeling methods, showcasing its promising application potential.

Gated recurrent units (GRUs) have been successfully applied to various industrial soft-sensor tasks. However, their linear coupling constraint in the hidden-state limits information propagation, while high-dimensional inputs and structural redundancies further challenge modeling. To address these issues, a speed-adaptive gated recurrent unit (SAGRU) network with weights dual-garrote regularization (WDG-SAGRU) is proposed for product quality prediction. Specifically, a speed-adaptive gating mechanism is integrated into the GRU update gate through a nonlinear exponential transformation. This mechanism is incorporated into the hidden-state update to break the linear coupling constraint, enhancing information flow and feature extraction. Meanwhile, the dual-garrote shrinkage coefficients are embedded into SAGRU’s input and hidden weight matrices, enabling synchronized optimization of input variable selection and structural sparsity. The proposed WDG-SAGRU is validated against state-of-the-art methods on a numerical example and an operational flue gas desulfurization system in a thermal power plant.

Data-driven soft sensors are extensively applied to predicting quality indicators. Recent studies have revealed that deep learning models can provide promising means for efficient predictive modeling. Nevertheless, most of them suffer from the intricate characteristics of spatiotemporal coupling relations, which hammers their widespread application in process industries. To handle those characteristics, this paper presents a spatiotemporal masked pre-training (SMPT) framework for soft sensing, which eases the masked modeling and guides the feature representation learning. Specifically, a graph-based memory unit with dynamic graph construction is devised to capture the connections among various process variables, accounting for the temporal and spatial correlations simultaneously. A multiple masked time series-based pre-training strategy is adopted to recover the randomly generated series to the original series via the weighted aggregation of similar neighbors, which assembles the ruined but complementary spatiotemporal information. In addition, a contrastive learning-based network training approach is designed for the effective high-level feature learning and series-wise similarity learning. Finally, SMPT is fine-tuned to obtain an output-feature mapping for predicting quality indicators. Compared with nine state-of-the-art (SOTA) methods, the proposed SMPT empirically presents superior results on two industrial datasets, yielding relative reductions in mean absolute error (MAE) of 18.0% and 9.0% over the second-best methods on the two datasets, respectively.