Compressed sensing (CS) can substantially enhance the transmission efficiency of wireless sensor networks (WSNs). To tackle the difficulties of high transmission delay and reconstruction failure caused by compressible measurements loss, this article proposes a measurements–residuals collaboration sparse reconstruction (MCSR). First, the acquisition node performs embedded compressed sampling (ECS) to improve transmission efficiency. The measurements obtained from the ECS are transmitted wirelessly to the gateway node, and can be random lost owing to unstable communication link. In addition, sensing matrix adaptive matching is proposed to address the mismatch between the dimensions of the missing measurements and the dimensions of the sensing matrix resulting in a failure of the reconstruction, providing a basis for subsequent effective reconstruction. Moreover, learning dictionary-based split Bregman iteration (SBI-LD) sparse reconstruction algorithm is adopted to realize the initial signal reconstruction based on the effective measurements obtained from wireless transmission. Furthermore, based on the initial reconstruction signal, the learning dictionary-based residuals reconstruction algorithm is proposed to obtain the reconstructed signal of the measurement residuals. Finally, the experimental results demonstrate that the proposed algorithm achieves higher reconstruction accuracy, compared with other popular methods. This provides a solution of great significance for efficient and reliable mechanical vibration monitoring in WSN.

The edge recognition of mechanical faults using deep learning models requires the deployment and operation of neural networks, which consume a large amount of memory. However, edge devices have limited memory. To address the problem, a neural network micromemory control strategy is proposed. This strategy addresses the memory resource constraints through process quantization and byte indexing while maintaining the model structure and accuracy. First, the neural network parameters with Gaussian and uniform distributions are tested and corrected to achieve process quantization. Then, a parameter memory control is implemented to achieve low-memory deployment and runtime. Subsequently, a bytes indexing is proposed for parameter location reading, enabling low-memory process inversed quantization. Finally, temporary variables is computed through micro memory overlay. This memory control method using process quantization and bytes indexing not only allows for ultra-low memory deployment and operation of neural networks but also improves test accuracy. It achieves over 97% test accuracy and an edge computation speed of over 200 KB/s with a maximum of 3.89 KB of runtime memory.

Operating condition monitoring of wind turbine (WT) key components is of significant importance to preventative maintenance and the improvement of WT reliability. To realize this industrial target, health indicator (HI) construction is a crucial and indispensable step. While most of the recently reported HIs are emphasized effective in stationary cases, they are insufficiently applicable to variable speed conditions. To address this issue, a novel HI through operating speed slicing and discrepancy compensation is proposed in this article for WT generator bearing condition monitoring. First, signal probability distributions of the collected degradation data are appropriately characterized by an optimized multiparameter regression method. Then, benchmark distributions established at the normal state are identified through operating speed slicing, and the discrepancies induced by the time-varying operating condition are subsequently calibrated with a compensation strategy. On this basis, a globally comparable metric, by quantitatively evaluating the degree to which the currently established distribution deviates from the corresponding slice-related benchmark, is accordingly constructed. Experimental tests demonstrate that the proposed HI can make a more effective health state assessment for WT generator bearing under variable speed conditions when compared with the conventional indicators.

Limitations pertaining to the storage capacity and bandwidth of wireless sensor networks (WSNs) for mechanical vibration monitoring (MVM) pose difficulties in the realization of reliable and energy-efficient communication of big data. To prolong the lifetime of battery-limited WSNs, this article proposes a multilayer joint optimization of packet size and adaptive transmission scheduling (ATS) to jointly optimize the energy consumption of acquisition nodes from the storage and transmission layers. First, owing to the frequent storage and transmission of massive mechanical vibration data in WSNs, the packet size significantly affects the energy consumption during storage and transmission. The proposed packet size optimization based on the lowest energy consumption (PSO-LEC) model optimizes the energy consumption of acquisition nodes from three aspects: 1) writing; 2) reading; and 3) transmission of data. Second, when transmitting massive vibration data frequently, the direct beacon frame conflict and indirect beacon frame conflict markedly induce data conflict and delay problems. Therefore, ATS based on the time division multiple access (TDMA/ATS) method is proposed to optimize the errors in data collision and transmission delay. Thus, the packet size and ATS are considered jointly to improve the network lifetime for MVM. Comprehensive experimental results indicate that the transmission reliability and energy efficiency of the proposed scheme are significantly improved, and the network life is effectively prolonged for MVM.

In wireless sensor networks (WSNs) for mechanical vibration monitoring systems, the network topology is initially in equilibrium. However, it progressively unbalances over time owing to the high sample frequency and large amount of data transmitted by each node. This can result in different remaining energy levels of each node and shorten the system lifetime. Therefore, dynamic adjustment of the topology is crucial to balance the networks in a lifetime cycle. This article proposes a time–space incentives control algorithm to dynamically improve the topology equilibrium for WSNs used in mechanical vibration monitoring systems. In the proposed algorithm, multiple relevant parameters that influence the network topology are designed and adopted. Two calculation methods involving expert experience and the data driver approach are devised to determine the parameter weights. A static evaluation model is constructed to measure the performance of the network topology in the space dimension. Based on this model, the dynamic topological equilibrium control algorithm and strategy are proposed considering time–space incentives. Finally, the transmission power of each node is changed by considering the comprehensive evaluation value, and the network topology is dynamically adjusted. The proposed algorithm constitutes an advanced approach because it comprehensively integrates the various parameters affecting the equilibrium of the network topology in the time and space dimensions. The experimental results indicate that the time–space incentives model contributes significantly to the dynamic improvement of the network topology equilibrium.

This article proposes a novel edge collaborative compressed sensing for mechanical vibration monitoring (MVM) to address the severe shortage of storage and computational resources and long delay in transmitting massive vibration data in wireless sensor networks (WSN) for MVM. It first combines compressed sensing and edge computing (EC) into a WSN to effectively solve the above-mentioned problems. Based on the self-developed acquisition node (AN) and EC node (ECN), the proposed approach is implemented in the AN and ECN, respectively. This enhances the acquisition efficiency and computing capacity of the WSN. Meanwhile, the sparse pattern of a mechanical fault signal is analyzed. In addition, the adaptive parameter adjustment mechanism for the alternating direction of multiplier method based on Anderson acceleration is proposed to generate a convex optimization algorithm with fast convergence to realize data reconstruction in the ECN. The results of comprehensive experiments demonstrate that the proposed method can reduce the transmission data by 70%, while maintaining high-precision bearing fault feature detection, compared with the Shannon sampling theory. Furthermore, the acquisition and transmission efficiencies are improved significantly. This provides a potential solution for practical engineering applications.

This article proposes a new multisensor fusion fault diagnosis method for gearbox, namely residual gated dynamic sparse network, to improve the multisensor feature learning and fusion ability. Considering that the fault sensitivity of the sensor varies with mounted location and complex transfer path modulation causes information from multisensor redundant, the lightweight channel attention unit is designed to strengthen the feature extraction ability of the network. The developed gated dynamic sparse unit is inserted into the deep architecture to eliminate ineffective components caused by high noise interference. Besides, the loss function is improved with multiple activation criteria to enhance convergence ability. The results of experiments and the engineering application show that the proposed method is more effective than other methods under varying degrees of noise interference.

To detect the incipient faults of rotating parts used in electromechanical systems widely, a novel transient feature extraction method based on the improved orthogonal matching pursuit (OMP) and one-dimensional K-SVD algorithm is explored in this paper. First, the stopping criterion of adaptive spark is developed, and then the corresponding OMP algorithm is used to remove the modulated and harmonic signals adaptively. Second, the residual signal is reformulated as a signal matrix by period segmentation and circulating shift, and the initial transient dictionary is constructed via the time-domain average technique. Subsequently, a novel K-SVD algorithm is proposed to get the optimized transient dictionary for the one-dimensional signal. Finally, the repetitive transient signal is recovered by the optimized dictionary. The simulated and experimental results show that the proposed method can not only much faster extract the fault characteristics than the traditional K-SVD method, but also more accurately detect the repetitive transients than the infogram method and the traditional K-SVD method.

Diagnosis of a civil aircraft engine, which is operating under speed variation conditions, is a representative problem encountered in aeronautic industry. It is still very challenging to estimate the instantaneous angular speed (IAS) through the aircraft engine vibration signal when no encoder or tachometer is available due to cost or technological reasons. However, for the currently available tacholess order tracking algorithms, many vital parameters must be initialized manually in advance, which lead to user-friendliness, even false diagnosis. To address this issue, a novel method is proposed and the merits of nonlinear mode decomposition are inherited, so the IAS can be estimated adaptively without prior knowledge. The vibration signal collected from a civil aircraft engine is used for validation; the experimental results exhibit that the proposed method is more accurate and flexible when compared with the conventional methods.

This article develops new deep learning methods, namely, deep residual shrinkage networks, to improve the feature learning ability from highly noised vibration signals and achieve a high fault diagnosing accuracy. Soft thresholding is inserted as nonlinear transformation layers into the deep architectures to eliminate unimportant features. Moreover, considering that it is generally challenging to set proper values for the thresholds, the developed deep residual shrinkage networks integrate a few specialized neural networks as trainable modules to automatically determine the thresholds, so that professional expertise on signal processing is not required. The efficacy of the developed methods is validated through experiments with various types of noise.