Industrial fault diagnosis (FD) often faces challenges due to the scarcity of labeled data and the inability of rule-based systems to handle high-dimensional nonlinearities. This study proposes expert system-assisted neural networks (ES-Nets), a novel hybrid framework featuring ES-guided decision-space sparsification to bridge symbolic reasoning with neural networks (NNs). Unlike traditional data-driven models, this knowledge-preconditioned architecture embeds domain-specific logic into a comprehensive knowledge base before training. Specifically, the optimization process constrains the gradient descent trajectory to a knowledge-consistent subspace, effectively regularizing the parameter updates based on symbolic expert logic rather than purely on statistical gradients. Advanced embedding techniques pre-configure the model, enabling early operational performance and significantly reducing training requirements. Validation on the Tennessee Eastman (TE) process and a real-world petrochemical plant case demonstrates the framework’s superiority in accuracy and operational efficiency. This study provides an efficient and interpretable solution, facilitating effective human–machine collaboration in complex industrial environments.

Accurate prediction of protein and peptide functions from amino acid sequences is essential for understanding biological processes and advancing biomolecular engineering. Due to the limitations of experimental methods, computational approaches, particularly machine learning, have gained significant attention. However, many existing tools are task-specific and lack adaptability. Here, we propose a BERT-BiLSTM-Attention-TCN Protein Function Prediction Framework (BBATProt), a versatile framework for predicting protein and peptide functions. BBATProt leverages transfer learning with a pretrained bidirectional encoder representations from transformer model to capture high-dimensional features. The custom network integrates bidirectional long short-term memory and temporal convolutional network to align with proteins' spatial characteristics, combining local and global feature extraction via attention mechanisms to achieve more precise predictions. Evaluations demonstrate that BBATProt consistently outperforms state-of-the-art models in tasks such as hydrolytic catalysis, peptide bioactivity, and post-translational modification (PTM) site prediction. Specifically, BBATProt improves accuracy by 2.96%-41.96% in antimicrobial peptide (AMP) prediction and by 0.64%-23.54% in PTM prediction tasks. In terms of area under the receiver operating characteristic curve, improvements range from 0.71% to 40.51% for AMP prediction and 0.62%-27.82% for PTM prediction. Visualizations of feature evolution and refinement via attention mechanisms validate the framework's interpretability, providing transparency into the feature-extraction process and offering deeper insights into the basis of property prediction.

Multiview subspace clustering (MSC) maximizes the utilization of complementary description information provided by multiview data and achieves impressive clustering performance. However, most of them are inefficient or even invalid among large-scale scenarios due to expensive computational complexity. Recently, anchor strategy has been developed to address this, which selects a few representative samples as anchor points for representation learning and anchor graph construction. However, most of them only explore single cross-view correlation, i.e., cross-view consistency from the global aspect or cross-view complementarity from the local aspect, which provides insufficient semantic correlation understanding and exploration for complex multiview data. To effectively address this issue, this study proposes a fast multiview subspace clustering (FMSC) with local-global anchor representation collaborative learning. FMSC integrates the discriminative anchor points learning and anchor graph construction with optimal structure into a joint framework. Furthermore, local (view-specific) and global (view-shared) anchor representations are learned collaboratively under two interaction strategies at different levels, providing beneficial guidance from global learning to local learning. Thus, the proposed FMSC can maximize the exploration of the complementarity-consistency among multiview data and capture a more comprehensive semantic correlation. More importantly, an effective algorithm with linear complexity is designed to solve the corresponding optimization problem of FMSC, making it more practical in large-scale clustering tasks. Extensive experimental results confirm the superiority of the proposed FMSC in both clustering performance and computational efficiency.

The emergence of new operating modes due to changes in manufacturing demands or production technologies leads to the occurrence of multimode industrial processes. To address the multimode problem, the dictionary learning method is widely utilized due to its excellent sparse representation technique capability. Despite the abundance of offline data enabling dictionary learning to model various modes, consolidating dictionaries from individual models considerably increases model complexity. Therefore, a method called adaptive importance coding dictionary learning is proposed in this study on the basis of dictionary learning. This approach utilizes dictionary learning techniques to determine the importance of dictionary atoms by encoding frequencies and selecting significant dictionary atoms to form a new dictionary, remarkably reducing the complexity of the dictionary model. In accordance with this idea, a comprehensive framework for monitoring multimode processes, including online fault detection, mode classification, fault isolation, and new mode updates, is introduced. Experimental validations are conducted through numerical examples, simulations of continuous stirred tank reactors, and application in an actual petrochemical process. The experimental results demonstrate the effectiveness and feasibility of the proposed method in monitoring multimode processes.

Nonlinear and non-Gaussian characteristics are common in industrial processes. Artificial neural correlation analysis (ANCA) is a good nonlinear process monitoring algorithm, which combines classical correlation analysis with artificial neural networks. However, its performance is not very satisfactory for industrial processes with non-Gaussian characteristics. To solve non-Gaussian problems, almost all the existing process monitoring algorithms only consider the effect of kurtosis. Nevertheless, both kurtosis and skewness affect the data distribution. To improve the limitations of existing algorithms, this study proposes a new process monitoring algorithm named Jarque–Bera-based ANCA. This new algorithm makes many improvements to ANCA scheme, and the designed loss function combines the influence of both kurtosis and skewness on the data distribution, which not only maintains the advantages of the ANCA algorithm in solving nonlinear problems, but also provides superior monitoring performance in non-Gaussian processes. Furthermore, the superior performance of the proposed new algorithm is verified through simulations using non-Gaussian and nonlinear numerical examples, the Tennessee Eastman process, and catalytic cracking units.

Fault detection has long been a hot research issue for industry. Many common algorithms such as principal component analysis, recursive transformed component statistical analysis and moments-based robust principal component analysis can deal with static processes only, whereas most industrial processes are dynamic. Therefore, dynamic principal component analysis and recursive dynamic transformed component statistical analysis have been proposed to deal with dynamic processes by expanding the dimensions. The computational complexity of these algorithms are greatly increased, and these algorithms cannot divide the data space accurately. In this paper, we propose a novel algorithm called recursive innovational component statistical analysis (RICSA), which estimates the dynamic structure of the data, accurately divides the data space into dynamic components and innovational components. In unsteady state process, the statistical characteristics of data will change, and RICSA can classify these characteristics into dynamic components by dividing the data space, instead of treating them as faults, thereby reducing the false alarm rate. Through a series of comparative experiments, especially on practical coal pulverizing system in the 1000-MW ultra-supercritical thermal power plant, Zhoushan Power Plant, we found the recursive innovational component statistical analysis to realize a higher accuracy rate and a lower false alarm rate and detection delay, which verifies its superiority. We also discuss the reduced computational complexity associated with the recursive innovational component statistical analysis. Note to Practitioners—Aiming at the dynamic processes, the recursive innovational component statistical analysis algorithm proposed in this paper can divide the data space into dynamic components and innovational components by estimating the dynamic structure of the data. In addition, in the monitoring process, computational complexity is also a key point. Compared with recursive dynamic transformed component statistical analysis, recursive innovational component statistical analysis has lower computational complexity and higher accuracy. After multiple sets of experiments, it can be verified that recursive innovational component statistical analysis has a great monitoring effect in the actual industrial process, and can provide early warning of faults.

Accurate process monitoring plays a crucial role in thermal power plants since it constitutes large-scale industrial equipment and its production safety is of great significance. Therefore, accurate process monitoring is very important for thermal power plants. The vigorous nature of the production process requires dynamic algorithms for monitoring. Since the common dynamic algorithm is mainly based on data expansion, the online computing complexity is too high because of data redundancy. Accordingly, this article proposes an innovative, dynamic process monitoring algorithm called autocorrelation feature analysis (AFA). AFA mines the dynamic information of continuous samples by calculating the correlation between the current time and past time features. While improving the monitoring effect, the AFA algorithm also has extremely low online computational complexity, even lower than common static algorithms, such as principal component analysis. Furthermore, this study exhibits the general form of dynamic additive faults for the first time and verifies the reliability of the algorithm through fault detectability analysis. Conclusively, the superiority of the AFA algorithm is verified on a numerical example, continuous stirred tank reactor (CSTR), and real data measured from a 1000-MW ultrasupercritical thermal power plant.

In modern industrial production, rotating machinery plays an important role. The gears in this machinery adjust the speed and transmission of torque. Therefore, when the gear fails, it is very important to be able to diagnose the fault quickly and accurately. Gear vibration signals are often used in gear fault diagnosis, but the fault signal is often overwhelmed by noises. To enable the scientific and efficient detection of faults, this study proposes a gear fault diagnosis method based on variational modal decomposition (VMD) and wide+narrow visual field neural networks (WNVNNs), namely VMD-WNVNN. VMD-WNVNN consists of two stages. In the feature extraction stage, VMD and Pearson correlation coefficients are used to decompose and reconstruct the original data to obtain the features of these data in the frequency domain. In the classification stage, WNVNN is used to classify the data based on features. The final results of the gear fault diagnosis experiments show that this method not only has higher classification accuracy but also has higher classification stability than other recently proposed methods. Note to Practitioners—The gearbox is composed of many mechanical parts, such as gears, shafts, and bearings. Therefore, the vibration signal collected by the vibration sensor on the gearbox housing will contain the vibration signal of each part and the noise caused by processing errors. Therefore, if some methods can be used to efficiently extract the characteristic signals required for diagnosis in the data processing stage, the efficiency of fault diagnosis will be greatly improved. This article takes the health of gears as the research object and proposes a method that combines adaptive signal decomposition and deep learning technology. Experimental results show that this method has higher classification accuracy and classification stability than other methods proposed recently.

In this article, a novel fault detection and process monitoring method referred to as artificial neural correlation analysis (ANCA) is proposed. Because nonlinear characteristics are common in complex industrial processes, the classic canonical correlation analysis (CCA) always perform poorly. Many scholars have noticed the nonlinear problem of the process and have also proposed some improved schemes, such as the kernel method. However, the selection of suitable parameters in the kernel method is extremely difficult, so most of the kernel learning methods are slightly unsatisfactory. Considering that the artificial neural network (ANN) can well extract the required feature components from the nonlinear data, we combined ANN and CCA from their respective principles, and proposed a new nonlinear monitoring method and the detailed gradient descent method derivation for the ANCA network is presented. In addition, we have designed two indices to monitor the changes of process variables and performance indicators. Finally, a numerical example, the Tennessee Eastman benchmark, and the Zhoushan thermal power plant process illustrate the superiority of the proposed method.