Feng Dong

Electrical resistance tomography (ERT) is a non-intrusive method used to determine internal conductivity distributions using measurement voltage acquired from electrodes positioned along a boundary. Although, ERT delivers distinct strengths, reliably generating high-resolution conductivity distributions continues to be demanding owing to unstable and complex non-linear structure of inverse problem. In response, a deep neural network architecture, termed SURE-Net (spatially up-sampled residual encoder network), is proposed to enhance image clarity and shape integrity in ERT image reconstruction. The framework integrates hierarchical feature extraction with electrode-adjacency alignment, residual-based refinement, and a non-learnable feature expansion stage to promote spatially consistent reconstruction. The SURE-Net further maintains stable reconstruction behavior under limited input conditions and preserves fine structural boundaries that are typically blurred in traditional approaches. Through extensive simulation and experimental evaluations, including robustness tests under varying signal-to-noise conditions, SURE-Net demonstrates consistent conductivity reconstruction with improved feature clarity. The model reflects high accuracy under different conductivity distributions and challenging structures. Moreover, effective model adaptation from restricted inputs is supported by its design, ideally suited in practical ERT applications.

Electrical impedance tomography (EIT) is a non-invasive imaging technique that is promising for multiphase flow detection and biomedical monitoring; however, its nonlinear inverse problem is highly ill-posed. Deep learning-based image reconstruction methods have gained considerable attention; however, many suffer from excessive computational costs or blurred boundary recovery, limiting their real-time performance. This study proposes a physics-guided pyramidal refinement and coarse correction network (PIPRC-Net) to balance geometric robustness, reconstruction fidelity, and efficiency. Embedding classical EIT regularization principles, it decouples global localization and detail refinement, using a bias-free learned linear inverse module for coarse physical priors and a ResUNet with a squeeze-and-excitation attention for nonlinear residual refinement. Evaluated on 145 627 EIDORS-generated samples, PIPRC-Net achieved an root mean squared error of 0.1076, Structural Similarity of 0.9685, and Dice coefficient of 0.8925, while maintaining clear boundaries under 20 dB noise. The results were further supported by expanded comparisons, ablation studies, and preliminary physical phantom validation after supervised fine-tuning of the measured EIT voltage-mask pairs. With 2.934 million parameters and 3.17 ms inference latency, PIPRC-Net provides a compact real-time-capable framework under the evaluation protocol.

Electrical impedance tomography (EIT) technology has garnered extensive attention in the field of tactile perception for electronic skin owing to its non-invasiveness, low wiring complexity, and excellent flexible adaptability. However, in practical applications, flexible sensing materials are often subjected to complex and uncontrollable mechanical conditions such as stretching, compression, and local damage, which induce drastic fluctuations in load impedance. Meanwhile, the contact state at the electrode-material interface is prone to change, further increasing the uncertainty in the measurement process. Under such dynamic working conditions, traditional current excitation methods are susceptible to load modulation, thereby reducing the system’s measurement stability and introducing imaging errors. To address the aforementioned problems, this paper designs a constant-current excitation circuit based on closed-loop feedback. By monitoring and regulating the output state, the circuit achieves stable control of the injected current under complex variable impedance conditions. Compared with schemes that solely rely on high-performance devices or complex compensation networks, this method not only ensures basic measurement accuracy, but also strikes a balance between circuit implementation complexity and system stability. Experimental results demonstrate that the proposed excitation circuit can effectively improve the consistency of current injection, thereby enhancing the stability and reliability of EIT imaging results.

Electrical Impedance Tomography (EIT) has emerged as a promising technique across various fields; however, its imaging quality is highly contingent upon the contact quality between electrodes and the subject. To address the challenges of artifact interference and measurement-space error coupling induced by electrode failure in practical applications, this study presents a sparse-decoupled diagnostic scheme for EIT faulty electrode monitoring, incorporating an adaptive strategy to rectify erroneous data. Specifically, a measurement-space error vector is constructed by leveraging reciprocity errors and waveform distortion features. By establishing an electrode-measurement coupling matrix and employing a sparse inversion algorithm, channel errors are mapped into the physical electrode space, thereby achieving the decoupling of faulty data. Furthermore, an adaptive thresholding strategy based on difference comparison is introduced to alleviate the decision ambiguity caused by interference from adjacent faulty electrodes. Experimental results demonstrate that the proposed framework consistently achieves accurate faulty electrode identification under various scenarios, including single-fault, multi-fault, and adjacent-coupled-fault conditions. When integrated with image reconstruction algorithms, the results indicate that the compensated data significantly mitigate noise artifacts, thereby enhancing the robustness and imaging quality of the EIT system in complex operating environments.

Transfer learning has been widely applied to intelligent fault diagnosis to address the challenge of insufficient labeled data. However, the efficacy of existing semi-supervised domain adaptation (SSDA) methods is often constrained by their critical dependence on pseudo-label quality and the adoption of indiscriminate global alignment strategies. To address the negative transfer induced by these limitations, a deep gated network (DGN) for targeted transfer learning is proposed in this article. First, an anchor-guided manifold clustering (AGMC) method is developed to generate high-quality pseudo-labels by exploiting both the local manifold structure and anchor supervision in the target domain. Subsequently, a feature extractor is constructed to learn discriminative representations from the source domain, integrate high-quality supervisory information from the target domain, and impose constraints on the feature space. Furthermore, a gated domain alignment strategy is designed to achieve precise class-level transfer. This strategy incorporates an integrated gating mechanism to selectively filter out domain-specific features while employing the local maximum mean discrepancy (LMMD) to align the conditional distributions across domains. Finally, the effectiveness and superiority of the proposed method are validated through transfer experiments on both cross-machine bearing and cross-condition two-phase flow datasets.

Stored pulverized coal is prone to oxidation, leading to internal temperature rise and increased risks of spontaneous combustion. Therefore, accurate monitoring of the internal temperature is of critical importance. Acoustic thermometry is regarded as a promising technique for tracking internal temperature changes within stored pulverized coal, owing to its high penetration ability, non-invasive nature, rapid response and low equipment cost. Nonetheless, the effectiveness of its penetration hinges on employing low-frequency sound waves, which may suffer from measurement inaccuracy due to reverberation interference and ambient noise, especially for temperature monitoring in enclosed spaces. To address this issue, this study developed a deep learning dereverberation and parameter extraction algorithm using CNN-Transformer. Experimental results demonstrate that the method effectively suppresses reverberation in enclosed spaces and enables high-precision recovery of temperature-related acoustic parameters. These improvements are expected to support subsequent acoustic tomography, in which accurate parameter estimation is essential for reconstructing the temperature distribution of stored pulverized coal.

Electrical resistance tomography (ERT) uses a network of boundary electrodes to visualize and analyze processes involving multiphase flow effectively. Nonlinearity and ill-posed problems are challenging for ERT reconstruction. However, reconstruction image accuracy is in demand. Due to a lack of suitable training, the existing deep learning network for ERT image reconstruction depends on sparse information flow and gradient flow. It is challenges to represent the structural topological link among internal conductivity characteristics using the usual reconstruction approach. By propagating feature representation across topological connections, the non-Euclidean structure of the ERT measurement encodes the problems. Capturing spatial and topological dependencies requires adaptive learning to handle irregular prominent feature positions and conductivity domains. The adaptive learning is modulated by edge priority for different graph states in a dynamic graph where edge features embed spatial constraints and graph geometry. Therefore, a multi-head attention mechanism is incorporated into the graph convolutional network to maintain the local smoothness while preserving sharp transitions. The dynamic graph is learned through edge weights jointly with node features using the RMSprop optimization method. Adaptive edge learning can identify object geometries and real-time flow patterns. Using extensive numerical simulation and experiment results, the proposed algorithm improves the imaging accuracy compared to the traditional methods without dense connections. It is compared with existing deep learning methods.

Oil–gas–water three-phase flow process exhibits randomness and transience. Under varying flow conditions and environmental factors, the process presents a dual challenge in state monitoring: data distribution discrepancies between source and target domains, along with the presence of unknown states in the target domain. Therefore, a cross-domain open-set state monitoring method based on multigrained adversarial network with hierarchical attribute causality (MANHAC) is presented in this work. A domain adversarial architecture is designed to distinguish unknown from known target flow states, in which the weighted thresholding method based on information entropy can adjust the decision boundary adaptively. Besides, to mitigate conditional distribution mismatch and negative transfer caused by forced global domain alignment, MANHAC introduces hierarchical causal attributes for describing different flow states, where attribute features are influenced by upstream cause attribute and optimized through competition with multiple discriminators. The proposed MANHAC employs a dual-alignment mechanism to achieve both global and fine-grained domain adaptation, which can effectively carry out cross-domain open set state identification. More importantly, it can describe unknown states by attribute vectors, providing more meaningful monitoring information. Dynamic experiments of three-phase flow demonstrate its effectiveness and superiority.