Single Sensor Signal Research Articles

Detection of Hot Cracking for 6000 Series Aluminum Alloys using a Multi-Sensor Based Deep Learning Model

This study developed a monitoring technology using a multi-sensor based deep learning model to diagnose hot cracking in aluminum alloy laser welding. Hot cracking that occurs during the laser welding process of aluminum alloys is difficult to diagnose accurately with a single sensor signal, necessitating multi-sensor based process monitoring technology. To monitor these hot cracks, laser-induced plasma, acoustic, and elastic wave signals were simultaneously measured using a spectrometer, non-contact acoustic sensor, and contact acoustic sensor during the overlap laser welding process of 6000 series aluminum alloys. The welded specimens were classified into normal and cracked specimens through bead analysis, and features related to hot cracking were extracted from each sensor signal to utilize the measured multi-sensor signals for monitoring. The extracted features from each signal were used as inputs for a Deep Neural Network (DNN) model capable of learning complex nonlinear relationships, and the hyperparameters of the DNN model were optimized using a genetic algorithm. The DNN model trained with multi-sensor data diagnosed hot cracking with an accuracy of 93.75%.

MRCFN: A multi-sensor residual convolutional fusion network for intelligent fault diagnosis of bearings in noisy and small sample scenarios

Bearing fault diagnosis is of great importance to ensure the safe and stable operation of mechanical equipment. The actual collected bearing fault signals are susceptible to strong noise interference and bearing samples for each fault state may be insufficient, which increases the difficulty of capturing effective features. Most of the existing diagnostic methods extract features from a single sensor signal for pattern recognition and fault diagnosis. The fault information provided by a single sensor is limited and incomplete, which is usually very difficult to meet the demand for accurate and reliable fault diagnosis in complex scenarios. To solve these problems, this paper proposes a multi-sensor residual convolutional fusion network (MRCFN) for intelligent fault diagnosis of bearings. Firstly, a convolutional pooling module (CPM) is coupled with the designed double ring residual module (DRRM) to rough feature extraction and deep feature mining, which not only captures the discriminative fault features from multi-sensor signal, but also avoids the performance degradation of network. Secondly, a spatial channel reconstruction module (SCRM) is further introduced to eliminate redundant information in the features and improve the network training efficiency. Finally, the presented global interactive perception fusion module (GIPFM) is connected with a classification block (CB) to globally fuse the features extracted from the acoustic and vibration signals and conduct automatic fault identification, which both can realize the complementarity and calibration of multi-sensor feature information and high precision diagnosis. The experiments and a series of comparisons are implemented on two datasets to efficaciously verify the superiority of the proposed method over five existing representative multi-sensor fusion diagnosis methods (i.e., FAC-CNN, MB-CNN, MsACNN, MRSDF, and IMSFDFL) under strong noise and small samples.

Multiscale cyclic frequency demodulation-based feature fusion framework for multi-sensor driven gearbox intelligent fault detection

Accurate fault detection is extremely important to ensure stable gearbox operation. Data-driven schemes using cyclic spectral have received significant attention due to their robust demodulation performance. However, these schemes are mainly applied to process single sensor signals, and they are unable to accurately obtain precise fault features. This paper proposed a novel multiscale cyclic frequency demodulation (MCFD)-based feature fusion framework for multi-sensor driven gearbox intelligent fault diagnosis. Firstly, the MCFD is proposed to analyze the vibration signals from multi-sensor driven gearbox, which acquires the multi-sensor mode information without setting parameters in advance. Thereafter, the grey relational degree between the multi-sensor mode information and original signal is calculated, and its results are normalized to obtain the relationship coefficients. Finally, the acquired coefficients are performed for multi-sensor information fusion to form the covariance matrix for gearbox fault diagnosis. The effectiveness of the proposed feature fusion framework is validated using the gearbox case. The comparative experiments indicate that this framework outperforms comparative algorithms for multi-sensor driven gearbox fault diagnosis.

Blood pressure estimation and classification using a reference signal-less photoplethysmography signal: a deep learning framework.

The markers that help to predict th function of a cardiovascular system are hemodynamic parameters like blood pressure (BP), stroke volume, heart rate, and cardiac output. Continuous analysis of hemodynamic parameters such as BP can detect abnormalities earlier, preventing cardiovascular diseases (CVDs). However, sometimes due to motion artifacts, it becomes difficult to monitor the BP accurately and classify it. This work presents an optimized deep learning model having the capability to estimate the systolic blood pressure (SBP) and diastolic blood pressure (DBP) and classify the BP stages simultaneously from the same network using only a single channel photoplethysmography (PPG) signal. The proposed model is designed by exploiting the deep learning framework of a convolutional neural network (CNN), exhibiting the inherent ability to extract features automatically. Moreover, the proposed framework utilizes the superlet transform method to transform a 1-D PPG signal into a 2-D super-resolution time-frequency (TF) spectrogram. A superlet transform separates the peaks related to true PPG signal components and motion artifacts components. Thus, the superlet provides a robust realtime approach to accurately estimating and classifying BP using a single PPG sensor signal and does not require additional ECG and PPG sensor signals for reference. Using a super-resolution spectrogram and CNN model makes the method profitable in motion artifact removal, feature selection, and extraction. Hence the proposed framework becomes less complex for deployment on wearable devices having limited battery resources. The performance of the proposed framework is demonstrated on the publicly available larger dataset MIMIC-III. This work obtained a mean absolute error (MAE) of 2.71mmHg and 2.42mmHg for SBP and DBP, respectively. The classification accuracy for the SBP prediction is about 96.79%, whereas it is 98.94% for DBP. From a motion artifact-affected PPG signal, SBP and DBP are estimated. Then the estimated BP is classified into three categories: normotension, prehypertension, and hypertension, and is compared with the state of art methods to show the effectiveness of the proposed optimized framework.

Research on Performance Degradation Assessment and Abnormal Health Status Detection of Quayside Crane Lifting Gearbox Based on Information Fusion

In order to solve the problems of subjectivity in the extraction of traditional degradation features and incomplete degradation information contained in a single sensor signal, a performance degradation assessment and abnormal health status detection method based on information fusion for the quayside crane lifting gearbox is proposed. Firstly, the correlation between the vibration and temperature of the gearbox is analyzed; secondly, the Convolutional Neural Network (CNN) and entropy degradation features from the full fault cycle vibration and temperature data of the lifting gearbox are extracted respectively; then, the final degradation indicators of the vibration and temperature data are obtained, respectively, through feature optimization, and the fusion degradation indicator is obtained by combining the two indicators; finally, the performance degradation assessment and abnormal health status detection of the gearbox are carried out. The effectiveness and superiority of the proposed method in the performance degradation evaluation of the gearbox are verified by comparison, and the proposed method can identify the initial degradation point of the gearbox earlier than the method based on the single vibration degradation index and the method based on the fusion of the traditional vibration degradation feature and the temperature entropy degradation feature.

Graph features dynamic fusion learning driven by multi-head attention for large rotating machinery fault diagnosis with multi-sensor data

Recently, rotating machinery fault diagnosis studies based on graph neural networks (GNN) have received some satisfactory achievements. But most of them are based on the analysis of the single sensor signals, which cannot capture the comprehensive fault information, especially aiming at large rotating machineries. A few research using GNN for multi-sensor fault diagnosis only fuse multi-source features in the construction of the input graph, and the fusion effect largely depends on the manual feature selection. Graph attention network (GAT), as an emerging GNN, can give trainable weights to vertices based on the self-attention mechanism to improve the effectiveness of feature learning. And it has not yet been used in the field of multi-sensor fault diagnosis. To fill this gap and utilize GAT’s advantages, this paper presents a multi-sensor multi-head GAT (MMHGAT) model for large rotating machinery fault diagnosis. With the input of several subgraphs, the designed MMHGAT model consisting of two graph attention layers (GAL), a feature fusion process and a Softmax classifier, can dynamically fuse and mine the high-level fault characteristics during the training process. By employing the experiment on the axial flow pump, the effectiveness and superiority of the proposed method are validated.

Engine remaining useful life prediction model based on R-Vine copula with multi-sensor data

Aeroengine is a highly complex and precise mechanical system. As the heart of an aircraft, it has a crucial impact on the overall life of the aircraft. Engine degradation process is caused by multiple factors, so multi-sensor signals are used for condition monitoring and prognostics of engine performance degradation. Compared with the single sensor signal, the multi-sensor signals can more comprehensively contain the degradation information of the engine and achieve higher prediction accuracy of the remaining useful life (RUL). Therefore, a new method for predicting the RUL of an engine based on R-Vine Copula under multi-sensor data is proposed. Firstly, aiming at the phenomenon that the engine performance parameters change over time, and the performance degradation presents nonlinear characteristics, the nonlinear Wiener process is used to model the degradation process of a single degradation signal. Secondly, the model parameters are estimated in the offline stage to integrate the historical data to obtain the offline parameters of the model. In the online stage, when the real-time data is obtained, the Bayesian method is used to update the model parameters. Then, the R-Vine Copula is used to model the correlation between multi-sensor degradation signals to realize online prediction of the remaining useful life of the engine. Finally, the C-MAPSS dataset is selected to verify the effectiveness of the proposed method. The experimental results show that the proposed method can effectively improve prediction accuracy.

On Sensor Data Clustering for Machine Status Monitoring and Its Application to Predictive Maintenance

Predictive maintenance is one of the main approaches which Industry 4.0 is based on, since it aims at reducing unplanned downtime and maintenance costs of industrial machines. In this work, a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">time-aware clustering-based approach to the analysis of sensor data</i> is presented for the purpose of monitoring the time evolution of the health status of an industrial machine. A possible application of the proposed framework to predictive maintenance is then proposed. As a relevant representative application scenario, the focus is on one of the key machines in a pharmaceutical plant: a freeze-dryer. The illustrated procedure allows to carry out a time segmentation of the properly sensed data. More precisely, the corresponding operational points (associated with features of the sensed data) are clustered using various algorithms, among which Density-Based Spatial Clustering of Applications with Noise (DBSCAN) turns out to be the best. The benefits of the proposed approach are (i) its general nature and (ii) the limited amount of needed features that have to be extracted from a single sensor signal. The proposed procedure is attractive when the collected data (e.g., from a single sensor) are not sufficient to build an accurate physical model of the monitored component.

A new convolutional neural network-based framework and data construction method for structural damage identification considering sensor placement

In the application of data driven structural damage identification (SDI) based on supervised deep learning technology, valid data demarcation is the foundation; a convolutional neural network model with learning ability and capability of processing rich signal information is the core. Based on this understanding, this work makes three contributions: Firstly, the structural damage location and severity are jointly demarcated, and the SDI problem is transformed into a multi-classification task. Secondly, a 3D signal processing convolutional neural networks (3DS-CNN) is designed with an attempt to identify the complex and slight damages using the most basic network structure. Thirdly, a ‘major and subsidiary’ data construction (MSDC) method integrating the key intrinsic mode function is proposed to construct 3D data. Then the proposed schemes are verified by two different structures. The results show that the 3DS-CNN has excellent damage identification ability for small-size data with noise pollution. MSDC method can enrich the feature information of the damage signals and help the network with deep feature excavation, even if the vibration signals are heavily polluted. Going one step further, the impact of sensor placement is discussed, and it is found that when external excitation is obvious, better SDI accuracy can be achieved even using a single sensor signal with slight noise. When the noise interference is obvious, the generalization ability and noise robustness of the network can be enhanced by optimizing sensor placement. In this case, the sensor placement criteria and the sensitive nodes of the structure should be comprehensively and carefully considered to avoid mutual ‘coupling’ interference of data between sensors.

Fault diagnosis of marine electric thruster bearing based on fusing multi-sensor deep learning models

The data-driven fault diagnosis of bearings is important in the marine electric thruster. For avoiding information loss when manually extracting features and unreliable diagnosis by a single sensor, a novel method of fusing multi-sensor deep learning models is proposed. The improved one-dimensional convolution network (1DCNN) can adaptively extract features from single-sensor signals and use them for preliminary fault diagnosis at first. Then, the diagnosis results of different sensors are fused by the evidential reasoning (ER) rule. We found that the diagnosis accuracy at the training set size of 40 % is 99.4 %, which is better than four machine learning methods and ten state-of-the-art deep learning methods. Furthermore, at different noise levels (0–10 dB), the diagnostic accuracy is higher than 89 %, showing more robustness than single-sensor deep learning methods. Meanwhile, its suitability is further validated under different torque conditions.

A new multichannel deep adaptive adversarial network for cross-domain fault diagnosis

Currently, most fault diagnosis methods can achieve desired results from a single signal source. However, a single sensor signal has limited features and adaptability to the working environment, which will greatly affect the diagnosis results. To overcome this weakness, a multichannel deep adaptive adversarial network (MCDAAN) based on fusing acoustic and vibration signals is proposed in this paper. The training process of MCDAAN primarily includes the following aspects. First, the acoustic and vibration signals extracted by the neural network feature extraction are fused after being adjusted by the convolutional block attention module in channel and spatial dimensions. Next, the fusion features of the source and target domains are measured by the Wasserstein distance. Finally, the fused features are classified by the label and domain classifiers. The proposed MCDAAN is tested using acoustic and vibration signals collected at ten transfer tasks. The results demonstrate that the diagnostic accuracy of the proposed MCDAAN can reach more than 99% in both groups of experiments. MCDAAN can accurately classify all kinds of fault samples, and the classification accuracy is superior to other comparison methods.

Voltage Difference Envelope to the Control Simulating of Connection Frequency of Seamless Shoring

Through the intelligent control of single chip microcomputer and Hall sensor detection technology and signal transmission technology, the frequency, voltage and voltage phase of the shore power and the ship power are detected to form the lowest point of voltage difference envelope, and the seamless connection of the shore power and the ship power is controlled. The load transfer control between the shore power and the ship power realizes the ideal synchronous control of grid-connected closing and load transfer by establishing the ratio control of frequency difference between the shore power and the ship power and power change during load transfer.

Fault diagnosis of wind turbine based on multi-signal CNN-GRU model

Deep Learning has been widely used in the monitoring and diagnosis of wind turbines. However, most of the current fault diagnosis methods only use single sensor signal as the input of DL model, which leads to the limitation of the model performance. Therefore, this paper proposes a multi-signal CNN-GRU model. Firstly, the acquired multiple sensor signals are converted to time–frequency images by Multi-Synchrosqueezing S-Transform, the frequency domain features of multiple sensors are extracted by Convolutional Neural Network and fused by Attention Mechanism, then the multi-source time-frequency features are extracted by Gated Recurrent Unit and finally classified by SoftMax. Experiments are conducted on the CWRU dataset and the field gearbox dataset. The results show that the proposed method achieves an average accuracy of 99.69% and 100% on the two datasets, which are both higher than existing DL-based fault diagnosis methods. The proposed method can effectively fuse signals from multiple sensors, thus improving the classification accuracy and stability of the model, which has high practicality and reliability for fault diagnosis of wind turbines.

Intelligent recognition of tool wear in milling based on a single sensor signal

Intelligent recognition of tool wear in milling based on a single sensor signal

Fault Diagnosis Algorithm of Gearbox Based on NAPSO-VMD Self-Adaptive Noise Reduction and Dual-Sensor Feature Fusion

Condition monitoring plays a very important role in equipment fault diagnosis technology. However, existing monitoring methods often collect equipment fault signals from a single dimension, resulting in a major lack of fault information. To improve the problem, we built a gearbox preset fault test bench and constructed a dual-sensor acquisition system to realize the multiple dimensions of vibration signal acquisition in the horizontal and vertical directions of the gearbox. At the same time, given the poor adaptability of most current signal preprocessing methods, the improved nonlinear adaptive inertial weight particle swarm optimization algorithm (NAPSO) and variational modal decomposition (VMD) are combined to optimize the key parameters in VMD with the maximum correlation kurtosis convolution (MCKD) as the fitness function. Further, after extracting fault features from the intrinsic mode functions (IMFS) decomposed by VMD, the single-layer sparse autoencoder network (SAE) and the double-layer stacked sparse autoencoder network (SSAE) with different structures are used to realize an effective fusion of multidimensional information and deep feature extraction. Finally, the hybrid fault diagnosis of gearboxes is realized by using the random forest algorithm (RF) as the classifier. The experimental results show that the accuracy of the method proposed in this paper can reach 96.0%, and the accuracy can be improved by 3.0% and 4.0%, respectively, when compared with a single horizontal or vertical sensor signal input.

Multisource Data Fusion Diagnosis Method of Rolling Bearings Based on Improved Multiscale CNN

Intelligent diagnosis applies deep learning algorithms to mechanical fault diagnosis, which can classify the fault forms of machines or parts efficiently. At present, the intelligent diagnosis of rolling bearings mostly adopts a single‐sensor signal, and multisensor information can provide more comprehensive fault features for the deep learning model to improve the generalization ability. In order to apply multisensor information more effectively, this paper proposes a multiscale convolutional neural network model based on global average pooling. The diagnostic model introduces a multiscale convolution kernel in the feature extraction process, which improves the robustness of the model. Meanwhile, its parallel structure also makes up for the shortcomings of the multichannel input fusion method. In the multiscale fusion process, the global average pooling method is used to replace the way to reshape the feature maps into a one‐dimensional feature vector in the traditional convolutional neural network, which effectively retains the spatial structure of the feature maps. The model proposed in this paper has been verified by the bearing fault data collected by the experimental platform. The experimental results show that the algorithm proposed in this paper can fuse multisensor data effectively. Compared with other data fusion algorithms, the multiscale convolutional neural network model based on global average pooling has shorter training epochs and better fault diagnosis results.

Application of Instantaneous Parameter Characteristic in Active Lamb Wave Based Monitoring of Plate Structural Health

In the active Lamb wave based monitoring of plate structural health, it is difficult to extract damage information from active Lamb waves based on single sensor collection. Based on the Hilbert–Huang transform (HHT) instantaneous processing method, this paper proposes to study the damage information carried by a single sensor monitoring signal from the instantaneous parameter characteristics of the signal. The instantaneous phase change caused by the phase difference between the damage scattering wave and the direct wave is studied. The change of the marginal spectrum amplitude in the effective range caused by the damage scattering wave is studied in continuous multiple frequency bands. Finally, the damage information extraction based on a single sensor monitoring signal is realized. From the model analysis and experimental results, it is reliable and feasible to realize the active Lamb wave based monitoring of plate structural health according to the instantaneous parameter change characteristics from a single sensor signal.

Signal frequency domain analysis and sensor fault diagnosis based on artificial intelligence

Measurement bias or drift often occurs after long-term use of the sensor. Measuring faults will inevitably mislead the control system, making the goal of advanced control strategies impossible. To this end, this paper proposes an artificial intelligence diagnosis method based on wavelet neural network for the diagnosis of sensor faults. Wavelet analysis is used to extract the frequency domain features of the data, and then the neural network is used to diagnose the frequency domain characteristic data of the signal. On this basis, the data of a single sensor signal is used for self-diagnosis. The advantages of wavelet analysis in extracting signal characteristics and the advantages of neural network in feature learning and discrimination are comprehensively utilized. Based on the correlation of sensor signals, this paper proposes a joint information diagnosis method. To determine the correlation of sensor signals, models based on energy balance and flow-pressure balance were established. The simulation results show that the method can effectively diagnose the fault of the sensor in the multi-sensor system.

A generic framework for multisensor degradation modeling based on supervised classification and failure surface

In condition monitoring, multiple sensors are widely used to simultaneously collect measurements from the same unit to estimate the degradation status and predict the remaining useful life. In this article, we propose a generic framework for multisensor degradation modeling, which can be viewed as an extension of the degradation models from one-dimensional space to multi-dimensional space. Specifically, we model each sensor signal based on random-effect models and characterize failure events by a multi-dimensional failure surface, which is an extension of the conventional definition of the failure threshold for a single sensor signal. To overcome the challenges in estimating the failure surface, we transform the degradation modeling problem into a supervised classification problem, where a variety of classifiers can be incorporated to estimate the degradation status of the unit based on the underlying signal paths, i.e., the collected sensor signals after removing the noise. As a result, the proposed method gains great flexibility. It can also be used for sensor selection, can handle asynchronous sensor signals, and is easy to implement in practice. Simulation studies and a case study on the degradation of aircraft engines are conducted to evaluate the performance of the proposed framework in parameter estimation and prognosis.

Fault Separation and Detection for Compound Bearing-Gear fault Condition Based on Decomposition of Marginal Hilbert Spectrum

For the rotating machinery system, it is a challenge to explore fault detection and diagnosis for multiple-faults condition, which simultaneously contains faulty bearing components and faulty gear components. In the study, a fault feature separation and extraction approach is proposed for the bearing-gear fault condition through combining empirical mode decomposition (EMD), Hilbert transform (HT), principal component analysis (PCA), independent component analysis (ICA) techniques. Firstly, EMD is implemented to decompose the single sensor signal to obtain multiple sub-band signals termed as the intrinsic mode functions (IMFs). Secondly, the most relevant IMFs to bearing and gear fault features are selected to construct multiple-channels model with the help of the simulated bearing and gear fault signals. Thirdly, HT is utilized to compute marginal Hilbert spectrum (MHS) for each IMF in multiple-channels model, to construct an MHS matrix. Finally, some statistically independent components are obtained by decomposing the MHS matrix with PCA and ICA, and multiple fault features are identified from these components. The experimental application of the proposed method is put into a bearing-belt-gearbox union machinery system to evaluate its validity. The experimental analysis results indicate that the proposed method is effective to separate and extract a bearing fault and a gear fault for two types of compound bearing-gear fault conditions.