Acoustic imaging in the spherical harmonic domain (SHD) reduces computational complexity by representing the 3D sound field using spherical harmonic coefficients, thereby avoiding direct calculations on dense spatial grids. The spatial resolution is limited by the truncation number of spherical harmonics, especially at low-to-mid frequencies. Non-synchronous measurements (NSM) provide a practical way to increase the effective spatial sampling by moving the array to multiple positions. To further enhance acoustic imaging resolution, this paper considers a spherical microphones array and introduces NSM into SHD. Based on this framework, two imaging methods are proposed: spherical harmonic beamforming of NSM (NSM-SHB) and sparse Bayesian learning of NSM in the SHD (NSM-SBL-SHD). Simulations and laboratory experiments with a 64-channel spherical microphone array across 8 scanning positions are conducted for validation. The results demonstrate that the proposed methods improve both accuracy and efficiency. At 200 Hz, the Direction of Arrival (DoA) error is reduced from 2.99∘ (NSM-CBF) to 0.26∘ (NSM-SBL-SHD), while NSM-SHB achieves a 3.09 times reduction in computation time (10.6 s vs. 3.4 s). At 2000 Hz, the DoA error decreases from 3.12∘ to 0.09∘, while NSM-SHB achieves a 3.18 times reduction in computation time (11.9 s vs. 3.7 s). The effectiveness of the proposed methods is further verified in industrial blower applications, where NSM-SBL-SHD provides accurate localization with robustness to moderate reverberation.

Infrared thermography faces persistent challenges in temperature accuracy due to material emissivity variations, where existing methods often neglect the joint optimization of radiometric calibration and image degradation. This study introduces a physically guided neural framework that unifies temperature correction and image enhancement through a symmetric skip-CNN architecture and an emissivity-aware attention module. The pre-processing stage segments the ROIs of the image and and initially corrected the firing rate. A novel dual-constrained loss function strengthens the statistical consistency between the target and reference regions through mean-variance alignment and histogram matching based on Kullback-Leibler dispersion. The method works by dynamically fusing thermal radiation features and spatial context, and the model suppresses emissivity artifacts while recovering structural details. After validating the industrial blower system under different conditions, the improved network realizes the dynamic fusion of thermal radiation characteristics and spatial background, with accurate calibration results in various industrial conditions.

In industry, sound source localization technology is used for detecting and diagnosing equipment faults by precisely locating noise sources to detect normal or abnormal functioning, and improve the maintenance efficiency. However, the localization of multiple sound sources with various motion patterns is often hindered by the different motion characteristics and spatial distributions of these sources, making their distinction difficult for traditional methods. This complex problem of localizing sound sources with different motion patterns is addressed in this study, with a specific focus on the identification and separation of static, linearly moving, and rotating sound sources. An innovative method is proposed that integrates Modal Composition Beamforming (MCB) with the equivalent source approach. A Multi-Motion Mode Sound Source Power Propagation(M3-S2-PP) model is introduced and hybrid deconvolution approach considered in relation to sources of the static, linearly moving and rotating types. Cross-PSF matrix equation is effectively solved using Least Absolute Shrinkage and Selection Operator (LASSO) within Alternating Direction Method of Multipliers (ADMM). LASSO's regularization capability enhances predictive accuracy and promotes solution sparsity, enabling the effective identification of active sound sources. In a first step, extensive simulations are conducted rigorously test the effectiveness of this method and to explore how the simulation scenario parameters influence the localization results. In scenarios where the MCB method is applicable, the proposed method achieves precise separation and localization results. In scenarios where the application of the MCB method may produce side lobes, the localization results are observed to exhibit varying degrees of deviation as the experimental scenario parameters change. Then experiments are carried out to rigorously examine the effectiveness of this technique and confirm the applicability conditions. Accuracy, reliability, and robustness of this method consistently validated across various scenarios. Even under challenging conditions, accurate performance in noise immunity and sound source localization accuracy is demonstrated this method. Potential applications in fields such as industrial monitoring, environmental noise assessment, and acoustic imaging are indicated by its performance.

Inverse problems arise across scientific and engineering domains, where the goal is to infer hidden parameters or physical fields from indirect and noisy observations. Classical approaches, such as variational regularization and Bayesian inference, provide well established theoretical foundations for handling ill posedness. However, these methods often become computationally restrictive in high dimensional settings or when the forward model is governed by complex physics. Physics Informed Neural Networks (PINNs) have recently emerged as a promising framework for solving inverse problems by embedding physical laws directly into the training process of neural networks. In this paper, we introduce a new perspective on the Bayesian Physics Informed Neural Network (BPINN) framework, extending classical PINNs by explicitly incorporating training data generation, modeling and measurement uncertainties through Bayesian prior modeling and doing inference with the posterior laws. Also, as we focus on the inverse problems, we call this method BPINN-IP, and we show that the standard PINN formulation naturally appears as its special case corresponding to the Maximum A Posteriori (MAP) estimate. This unified formulation allows simultaneous exploitation of physical constraints, prior knowledge, and data-driven inference, while enabling uncertainty quantification through posterior distributions. To demonstrate the effectiveness of the proposed framework, we consider inverse problems arising in infrared image processing, including deconvolution and super-resolution, and present results on both simulated and real industrial data.

A thermal fault detection method for high-speed direct-driven blower components is proposed, using thermal and visible image fusion along with semantic segmentation. The proposed method follows three steps: multimodal image fusion, component semantic segmentation of the fused image, and temperature level segmentation. First, an end-to-end image fusion network based on an improved denoising diffusion model is used, a perceptually prioritized weighted loss is introduced for training, and an alternate training strategy is used to improve the quality of the fused images. In the second step, a lightweight segmentation network is proposed to reduce the model size and inference time while improving the segmentation accuracy. Finally, thermal images are processed by clustering methods. Experiments on real industrial objects show that the proposed method composed of infrared and optical image fusion, semantic segmentation, and temperature clustering networks improves significantly the fault temperature detection on different blower components.

Infrared and visible image fusion aims to generate fused images that maintain the advantages of each source such as temperature information and detailed textures. This paper presents Bayesian Model-based Fusion-Net, a novel approach for infrared and visible image fusion. By formulating image fusion as an inverse problem within a hierarchical Bayesian framework, our method leverages physical priors and data-driven techniques to enhance model interpretability and transferability. Compared to traditional and deep learning-based fusion methods, the proposed Bayesian Model-based Fusion-Net achieves promising performance with significantly reduced computational complexity (0.07G FLOPs). Extensive experiments on multiple datasets, including industrial public dataset, demonstrate the effectiveness of the proposed method in preserving texture details, maintaining structural integrity, and enhancing feature clarity. Furthermore, our approach exhibits robustness when trained with limited data, maintaining consistent performance even when using only 10% of the training dataset. These characteristics make the proposed Bayesian Fusion-Net particularly suitable for industrial monitoring applications where computational resources and the amount of training dataset are limited.

Industrial blowers are energy-efficient and widely used, but fault monitoring is challenging. Infrared and visible light sensors monitor their status, capturing temperature distribution and time evolution in images and videos. Due to the complexity of the industrial scene and the large difference between the characteristics of infrared and visible images, the existing registration methods are unable to accurately align the infrared and visible images. This paper proposes a Neural Network (NN) based registration method of infrared and visible images for industrial blowers using contours and Weight Global Shape Context Descriptor (W-GSCD), by considering the distance between different feature points as weights in our feature registration method which improves the accuracy of the registration of IR and visible images compared to classical image registration methods. Experiments on real data show that the proposed method effectively addresses registration challenges from complex shapes and heterogeneous backgrounds, achieving promising results on the blower dataset.

In industrial scenarios, direct-drive blowers are essential for operations such as exhaust emission control, air ventilation, compression, and conditioning. Their continuous operation often results in substantial noise pollution. The unusual level of noise is usually associated with mechanical failures or design flaws. This paper focuses on simulated non-synchronous measurements (NSM) for high-resolution acoustic imaging and noise localisation of direct-drive blowers at characteristic fre-quencies. The acoustic imaging results of single measurement and Joint Maximum A Posterior (JMAP) of NSM are compared in the paper. The results indicate that the JMAP of NSM method provides sharper imaging quality with fewer side lobes and higher resolution. The results will be used to guide the application of NSM in industrial scenarios to monitor the direct-drive blowers' operational conditions and optimize noise reduction designs.

Infrared thermography is widely used to detect abnormal body temperature because of its noncontact and large scale. Attenuation during infrared propagation causes the measured temperature to always be less than the actual value. This article proposes a novel method of temperature calibration based on Bayesian inference. In the first step, we propose an improved infrared radiation model (IIRM), which accounts for emissivity and measures the distance between the radiation source and the infrared imager. This study leverages naive Bayesian inference (NBI) to derive surface emissivity. Then, using the improved model, the parameters of the model and the temperature distribution are reconstructed by joint maximum a posterior (JMAP). The IIRM and JMAP method (IIRM-JMAP) improved the accuracy of temperature measurement. The improved infrared thermal radiation model is suitable for measuring scenarios with different measuring distances, different humidity factors, and different emissivities. The proposed method has been validated to have small errors through various experiments on a blackbody and high-speed direct-drive blower.

The Modal Composition Beamforming (MCB) method can quickly achieve the localization of fast-rotating sound sources; however, its low resolution and unclear applied conditions severely limit its industrial application. This study aims to investigate the MCB-based high-resolution localization method for multiple rotating sound sources and apply it to the localization and identification of axial-fan blade-noise. In this paper, the Variational Bayesian Approximation (VBA) and the Subspace Variational Bayesian Approximation (SVB) methods are used to solve the MCB-based rotating sound power propagation (RSP) model, denoted as RSP-VBA and RSP-SVB, respectively. The effectiveness of the proposed RSP-VBA and RSP-SVB are experimentally validated for the first time. Compared to the conventional MCB method, the resolution is significantly improved, and the proposed high-resolution methods are even faster than the classical Rotating Source Identifier (ROSI) method in most conditions. More importantly, the applied condition of the MCB, RSP-VBA, and RSP-SVB methods is given and verified by using three evaluation indicators. Then a schematic of the applied condition with examples is provided. With the guide of the applied conditions, the RSP-VBA and RSP-SVB methods are applied to the blade-noise localization of various multiblade high-speed axial fans.

Traditional sound source localization methods encounter significant challenges in simultaneously locating rotating and static sources. These challenges arise from the differing motion patterns of these two types of sound sources, and they are typically not situated on the same plane. To address this issue, a method based on Modal Composition Beamforming (MCB) and the equivalent source method is proposed for separating rotating and static sound sources that can fully utilize priori knowledge of the spatiotemporal properties of the sources. The proposed approach involves establishing a Rotating-Static Sources Power Propagation (R-S2P) model, utilizing the relationship between the beamforming output of equivalent sources and the actual beamforming output. Solving this model allows for matching the contributions of rotating and static equivalent sources. Simulations for three cases with different source strengths are presented, and the R-S2P model is solved using the Least Absolute Shrinkage and Selection Operator (LASSO). This method enables accurate separation and localization of rotating and static sources on different planes with varying relative intensities, even if the background noise is strong

Neural Networks (NN) has been used in many areas with great success. When a NN's structure (Model) is given, during the training steps, the parameters of the model are determined using an appropriate criterion and an optimization algorithm (Training). Then, the trained model can be used for the prediction or inference step (Testing). As there are also many hyperparameters, related to the optimization criteria and optimization algorithms, a validation step is necessary before its final use. One of the great difficulties is the choice of the NN's structure. Even if there are many "on the shelf" networks, selecting or proposing a new appropriate network for a given data, signal or image processing, is still an open problem. In this work, we consider this problem using model based signal and image processing and inverse problems methods. We classify the methods in five classes, based on: i) Explicit analytical solutions, ii) Transform domain decomposition, iii) Operator Decomposition, iv) Optimization algorithms unfolding, and v) Physics Informed NN methods (PINN). Few examples in each category are explained.

Traditional sound source localization methods encounter significant challenges in simultaneously locating rotating and static sources. These challenges arise from the different motion patterns of these two types of sound sources, and they are typically not situated on the same plane. To address this issue, a method based on Modal Composition Beamforming (MCB) and the equivalent source method is proposed for separating rotating and static sound sources, which fully utilizes the prior knowledge of the spatiotemporal properties of these sources. The proposed approach involves establishing a Rotating-Static Sources Power Propagation (R-S2P) model, utilizing the relationship between the equivalent source strength and the actual beamforming output. By employing this forward model and applying an appropriate inversion method, it is possible to separate the components of rotating and static sources. Simulations for three cases with different source strengths are presented, and the R-S2P inversion problem is resolved using the Least Absolute Shrinkage and Selection Operator (LASSO) method. We showed that this method enables accurate separation and localization of rotating and static sources on different planes with varying relative intensities, even if the background noise is strong.

Industry prioritizes using infrared (IR) sensors to obtain temperature information and detect equipment failures early. Existing methods have accuracy, segmentation, and real-time detection issues. This article uses infrared video data to detect temperature faults in blowers. To improve equipment part segmentation accuracy, a robust segmentation algorithm is adopted based on improved point rendering infrared blower images. A hierarchical multiscene anomaly detection method for analyzing temperature profiles in the blower region is proposed. ${K}$ -means++ clustering is used for coarse classification, followed by k-nearest neighbor (KNN) analysis to identify anomalous data. Anomalous clustering center data is then scored using sliding windows. These scores are then used as the new anomaly indicators, and the time interval of the anomaly is estimated. The main contribution of this work is to propose a method to improve the accuracy and robustness of temperature anomaly detection in industrial objects that are in complex environments and background scenarios.

Infrared images have been widely used in many research areas, such as target detection and scene monitoring. Therefore, the copyright protection of infrared images is very important. In order to accomplish the goal of image-copyright protection, a large number of image-steganography algorithms have been studied in the last two decades. Most of the existing image-steganography algorithms hide information based on the prediction error of pixels. Consequently, reducing the prediction error of pixels is very important for steganography algorithms. In this paper, we propose a novel framework SSCNNP: a Convolutional Neural-Network Predictor (CNNP) based on Smooth-Wavelet Transform (SWT) and Squeeze-Excitation (SE) attention for infrared image prediction, which combines Convolutional Neural Network (CNN) with SWT. Firstly, the Super-Resolution Convolutional Neural Network (SRCNN) and SWT are used for preprocessing half of the input infrared image. Then, CNNP is applied to predict the other half of the infrared image. To improve the prediction accuracy of CNNP, an attention mechanism is added to the proposed model. The experimental results demonstrate that the proposed algorithm reduces the prediction error of the pixels due to full utilization of the features around the pixel in both the spatial and the frequency domain. Moreover, the proposed model does not require either expensive equipment or a large amount of storage space during the training process. Experimental results show that the proposed algorithm had good performances in terms of imperceptibility and watermarking capacity compared with advanced steganography algorithms. The proposed algorithm improved the PSNR by 0.17 on average with the same watermark capacity.

In 3D acoustic imaging, the spherical array is used to collect signals. The spherical harmonic function has good orthogonality and is suitable for solving the signals collected by the spherical array. In the spherical harmonic domain (SHD), the main lobe and sidelobe widths of acoustic imaging are influenced by the spherical array radius and the spherical harmonic expansion order of the spherical harmonic function. It has to do with the number of microphones. In this paper, the non-synchronous measurements (NSM) method is introduced into the spherical harmonic domain. The NSM is a measurement technique that uses a prototype array to move continuously to a fixed position to scan the sound source and can approximate the measurement results of large apertures and increased microphone density. In this paper, the most important change is the approximate expansion of Green's function in the NSM method by spherical harmonics. Secondly, the sparse Bayesian learning (SBL) method of NSM is applied to the SHD. Finally, the effectiveness of NSM in the SHD is verified by simulation and experiment using some point sound sources. In the SHD, the NSM method can not only increase the number of microphones but also improve the resolution of polar coordinate acoustic imaging and save time.

Rotating source beamforming techniques have been effective means of noise localization on rotary machines. In this letter, we derive an alternative expression for modal composition beamforming (MCB) and subsequently consider the equivalent source assumption and cyclostationarity of the constant angular-speed rotating sound source so that a rotating sound source power (RSP) propagation model is derived. By estimating a suitable solution for the RSP model using the subspace variational Bayesian (SVB) technique with sparsity and total variation (TV) priors, the validity of the RSP model was established. According to the simulation results, the proposed RSP-SVB method leads to a significantly higher resolution than the MCB method. It can localize multiple fast-rotating sound sources accurately, rapidly, and effectively in environments with strong background noise interference. Therefore, our proposed RSP-SVB can offer a reliable solution for identifying fast-rotating blade noise.

This paper proposes an online prediction method for the roundness of grinding workpieces based on vibration signals. Vibration sensors are used to collect vibration signals during grinding, and wavelet packet denoising is used to preprocess original signals to obtain effective vibration signals. Then use time domain analysis and frequency domain analysis to extract features and normalize them to form feature vectors. The roundness of the finished workpiece is measured using a shape-measuring instrument and integrated with the feature vectors to generate a usable data set. The support vector machine (SVM) algorithm is implemented using A Library for Support Vector Machines (LIBSVM), and a prediction model is constructed. Use the data set to train the model and evaluate the accuracy of the model to verify the effectiveness of the model. The results show that the prediction accuracy of the prediction method can reach 92.86%, and it can better predict whether the roundness is qualified.

The width of the mainlobe in the spherical harmonic domain (SHD) acoustic imaging is mainly influenced by the microphone aperture and number. A research method proposed in this paper is to introduce the non-synchronous measurements (NSM) into the SHD in order to further improve the resolution of acoustic imaging in the spherical harmonic domain. The NSM is a sequential measurement technique that uses a prototype array to scan the sound source in some fixed positions, which can approximate the measurement results of large apertures and increased microphone density. The main contributions of this paper are: (1) the NSM technology is successfully merged into the SHD; (2) The spherical harmonic beamforming of NSM (NSM-SHB) and the sparse Bayesian learning of NSM in the SHD (NSM-SBL-SHD) methods are proposed. The effectiveness of the proposed methods is verified by simulation and loudspeaker experiments. The results demonstrate that the resolution of acoustic imaging is both improved by the NSM-SHB and NSM-SBL-SHD in the middle and low frequencies compared with the single measurement. Meanwhile, the computational speed of these two methods is further improved compared to conventional 3D acoustic imaging.