Fault Diagnosis Of Mechanical Systems Research Articles

Enhancing Fault Diagnosis in Mechanical Systems with Graph Neural Networks Addressing Class Imbalance

Recent advancements in intelligent diagnosis rely heavily on data-driven methods. However, these methods often encounter challenges in adequately addressing class imbalances in the context of the fault diagnosis of mechanical systems. This paper proposes the MeanRadius-SMOTE graph neural network (MRS-GNN), a novel framework designed to synthesize node representations in GNNs to effectively mitigate this issue. Through integrating the MeanRadius-SMOTE oversampling technique into the GNN architecture, the MRS-GNN demonstrates an enhanced capability to learn from under-represented classes while preserving the intrinsic connectivity patterns of the graph data. Comprehensive testing on various datasets demonstrates the superiority of the MRS-GNN over traditional methods in terms of classification accuracy and handling class imbalances. The experimental results on three publicly available fault diagnosis datasets show that the MRS-GNN improves the classification accuracy by 18 percentage points compared to some popular methods. Furthermore, the MRS-GNN exhibits a higher robustness in extreme imbalance scenarios, achieving an AUC-ROC value of 0.904 when the imbalance rate is 0.4. This framework not only enhances the fault diagnosis accuracy but also offers a scalable solution applicable to diverse mechanical and complex systems, demonstrating its utility and adaptability in various operating environments and fault conditions.

WBUN: an interpretable convolutional neural network with wavelet basis unit embedded for fault diagnosis

Convolutional Neural Network (CNN) is extensively applied in mechanical system fault diagnosis. However, the absence of transparent decision mechanisms in CNNs hinders credibility. To address these challenges, this paper proposes an interpretable wavelet basis unit convolutional network (WBUN). This network incorporates meticulously designed wavelet basis unit (WBU) functions into convolutional layer, creating the interpretable wavelet basis unit convolutional (WBUConv) layer. Convolutional kernels with clear physical significance enable the WBUConv layer to extract fault-related features in both time and frequency domains, enhancing diagnostic performance, and interpreting the CNN’s attention frequency along with the convolutional kernel’s training outcomes. In this paper, three WBU functions are designed to construct the corresponding WBUNs, and their effectiveness and interpretability are verified through three sets of mechanical fault diagnosis experiments. Meanwhile, experimental results demonstrate the WBUConv layer’s remarkable advantages in noise robustness, convergence speed, and strong generalization ability.

DECVAE: Data augmentation via conditional variational auto-encoder with distribution enhancement for few-shot fault diagnosis of mechanical system

Conditional variational autoencoder (CVAE) has the potential for few-sample fault diagnosis of mechanical systems. Nevertheless, the scarcity of faulty samples leads the augmented samples generated using CVAE suffer from limited diversity. To address the issue, a novel CVAE variant namely CVAE with distribution augmentation (DECVAE) is developed, to generate a set of high-quality augmented samples that are different but share very similar characteristics and categories with the corresponding real samples. First, DECVAE add a new sample distribution distance loss into the optimization objective of traditional CVAE. Amplifying this loss in training process can make the augmented samples cover a larger space, thereby improving diversity. Second, DECVAE introduces an auxiliary classifier into traditional CVAE to enhance the sensitivity to category information, keeping the augmented samples class invariance. Furthermore, to ensure that the information of edge-distributed samples can be fully learned and make augmented samples representative and authentic, a novel multi-model independent fine-tuning strategy is designed to train the DECVAE, which utilizes multiple independent models to fairly focus on all samples of the minority class during DECVAE training. Finally, the effectiveness of the developed DECVAE in few-shot fault diagnosis of mechanical systems is verified on a series of comparative experiments.

TFN: An interpretable neural network with time-frequency transform embedded for intelligent fault diagnosis

Convolutional neural network (CNN) is widely used in fault diagnosis of mechanical systems due to its powerful feature extraction and classification capabilities. However, the CNN is a typical black-box model, and the mechanism of CNN’s decision-making is not clear, which limits its application in high-reliability-required fault diagnosis scenarios. To tackle this issue, we propose a novel interpretable neural network termed as time-frequency network (TFN), where the physically meaningful time-frequency transform (TFT) method is embedded into the traditional convolutional layer as a trainable preprocessing layer. This preprocessing layer named as time-frequency convolutional (TFconv) layer, is constrained by a well-designed kernel function to extract fault-related time-frequency information. It not only improves the diagnostic performance but also reveals the logical foundation of the CNN prediction in a frequency domain view. Different TFT methods correspond to different kernel functions of the TFconv layer. In this study, three typical TFT methods are considered to formulate the TFNs and their diagnostic effectiveness and interpretability are proved through three mechanical fault diagnosis experiments. Experimental results also show that the proposed TFconv layer has outstanding advantages in convergence speed and few-shot scenarios, and can be easily generalized to other CNNs with different depths to improve their diagnostic performances. The code of TFN is available on https://github.com/ChenQian0618/TFN.

A Highly Sensitive Wear Debris Sensor Based on Differential Detection

Wear debris in the oil contains a wealth of information about the friction pairs of the mechanical equipment. By analyzing the size and type of wear debris through oil detection technology, condition monitoring and fault diagnosis of mechanical systems can be realized. This paper presents an inductive sensor based on differential detection and its signal conditioning circuit, which can detect metal wear debris in the oil. The sensor adopts the structure of two induction coils embedded in one excitation coil. The differential signal is obtained by reverse connecting two induction coils with the same parameters, which can suppress the common mode interference and eliminate the influence of ambient noise so that the sensor has extremely low noise. Through the designed signal conditioning circuit, the detection signal is phase-sensitive detected, and the information of wear debris is extracted by amplification and filtering. In this paper, the sensing principle of the sensor is derived, the spacing between the two induction coils is optimized using finite element simulation, and the optimal excitation frequency, detection limit, and detection error of the sensor are investigated through experiments. The experiment results show that the sensor can detect 20 μm iron particles and 130 μm copper particles in a 2 mm flow channel, and the detection error of the sensor is less than 22%. The sensor has the advantages of simple structure and high sensitivity and can be applied to detect metal wear debris in hydraulic oil.

Analysis of key techniques of condition monitoring and fault diagnosis of mechanical system

Analysis of key techniques of condition monitoring and fault diagnosis of mechanical system

Mechanical varying non-stationary signal separation method based on instantaneous frequency estimation

Mechanical equipment often works on variable speed condition, its corresponding vibration signal presents multi-component, modulation coupling with fast time-varying instantaneous frequency (IF), how to effectively compute IF and realize fasting varying non-stationary signal decoupling separation plays an important role in mechanical system fault diagnosis. In this paper, a sparse representation method called multi-scale chirp sparse representation (MSCSR) is introduced to identify, extract, and trend IF for achieving a highly concentrated time-frequency energy. Simulation demonstrates that the proposed method performs better than traditional IF estimation method. Furthermore, an adaptive time-varying filter is constructed using the extracted instantaneous frequency to decouple non-stationary fast signal. Ultimately, by rapid instantaneous frequency fluctuation experiment, the effectiveness of proposed method for fast strong time-varying signal is validated, it can effectively extract rapid oscillation instantaneous frequency, and the error is less than 10%.

Application of Fault Diagnosis Method Combining Finite Element Method and Transfer Learning for Insufficient Turbine Rotor Fault Samples.

Deep learning has led to significant progress in the fault diagnosis of mechanical systems. These intelligent models often require large amounts of training data to ensure their generalization capabilities. However, the difficulty of obtaining turbine rotor fault data poses a new challenge for intelligent fault diagnosis. In this study, a turbine rotor fault diagnosis method based on the finite element method and transfer learning (FEMATL) is proposed, ensuring that the intelligent model can maintain high diagnostic accuracy in the case of insufficient samples. This method fully exploits the finite element method (FEM) and transfer learning (TL) for small-sample problems. First, FEM is used to generate data samples with fault information, and then the one-dimensional vibration displacement signal is transformed into a two-dimensional time-frequency diagram (TFD) by taking advantage of the deep learning model to recognize the image. Finally, a pre-trained ResNet18 network was used as the input to carry out transfer learning. The feature extraction layer of the network was trained on the ImageNet dataset and a fully connected layer was used to match the specific classification problems. The experimental results show that the method requires only a small amount of training data to achieve high diagnostic accuracy and significantly reduces the training time.

A Novel Approach for Fault Diagnosis in Mechanical Systems Using Time-Frequency Analysis and Unsupervised Learning

A Novel Approach for Fault Diagnosis in Mechanical Systems Using Time-Frequency Analysis and Unsupervised Learning

Design of a Three-Coil Microfluidic Lubricating Oil Detection Sensor With Multiple Signal Characteristics

In this article, a three-coil microfluidic lubricating oil detection sensor is designed by combining microfluidic principle with three-coil inductive detection method. The sensor adopts a microchannel design to reduce the position error, improve the detection accuracy of a single particle, and reduce the occurrence of detection errors caused by multiple particles passing through at the same time. The inductive electric potential of the induction coil was obtained by a simulation analysis and experimental verification methods. It is found that the influence of particles on the internal magnetic field of the micro coil is relatively enhanced when the micro coil is used as excitation and induction coil. This results in two peaks and valleys shown from the generated inductive electric potential, increasing of the signal’s characteristic point position, and improvement in signal recognition. Therefore, it will provide more information for the subsequent material identification and multiparticle mixing signal differentiation. Meanwhile, a signal processing circuit with the function of low-pass filtering and amplification is designed to suppress the noise value of the output signal to around 1.25 mV, which could improve the stability of the system. A series of lubricating oil abrasive particle detection tests are conducted, and the results show that the new detection sensor is able to detect iron particles below 64 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu \text{m}$ </tex-math></inline-formula> and copper particles below 125 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mu \text{m}$ </tex-math></inline-formula> . The abrasive particle information collected by this sensor can provide more precise wear status information for the health assessment and fault diagnosis of mechanical systems, which is important for the health monitoring of mechanical systems.

An Intelligent Diagnostic Method for Multisource Coupling Faults of Complex Mechanical Systems

In the actual environment, there are difficult points such as complicated mechanical system fault types, random fault locations, and inconspicuous minor fault signals, which make it difficult to accurately diagnose faults. This paper proposes a new method for fault diagnosis of an adaptive multisensor bearing-gear system based on GAF/MTF (Gramian angular fields and Markov transition fields) and ResNet (deep residual network). First, we establish a multisensor signal acquisition system to monitor the running signals of the bearing-gearbox composite test bench in real time. Faulty parts include multiple types of composite faults of different sizes, different fault types, and different transmission stages. Second, based on GAF/MTFs, the multichannel timing signal collected by using the acquisition system is converted into multichannel pictures, and pictures are fused and compressed into three-channel pictures. Finally, we input these pictures into ResNet for fault diagnosis. The experimental results show that the GAF/MTF-ResNet model has a recognition accuracy of 72.14% for a total of 520 classification label test sets under different motor speeds, different sampling times, and different types of faults. Among them, the accuracy of the motor speed and sampling time is close to 100%, and the accuracy of gearbox failure and bearing failure is 75.25% and 88.97%, respectively. This shows that the method provides new ideas for the composite fault diagnosis of mechanical systems under different working conditions and different types of faults and has theoretical guiding significance.

Ball-Bearing Fault Classification Using Comparative Analysis of Wavelet Coefficient based on Entropy Measurement

Ball bearing is used to provide free rotation around a fixed axis. Various kinds of faults exist in the bearing such as inner race fault, outer race fault and ball race fault. One very effective method to diagnose the bearing fault is vibration signal analysis. Empirical mode decomposition (EMD) has been used for ball-bearing fault diagnosis in mechanical systems using vibration signal analysis. Classification of the ball-bearing fault has always been a challenging task. Various classification schemes such as Extreme learning machine (ELM), K-means Clustering, and Support vector machine (SVM), have been reported in the literature for ball-bearing fault classification. SVM is restricted by multiclass classification efficiency, and ELM is restricted by the longer training. In this paper, the entropy analysis of the wavelet coefficient obtained from the third level decomposition of the residue signal (obtained after subtracting the highest frequency component from the raw signal) has been done for ball-bearing fault classification. A comparative analysis of the wavelet coefficient based on entropy measurement has been presented here. High fault classification accuracy has been achieved using the proposed method for the detection of ball-bearing fault. Shannon entropy, Average Shannon entropy, and Renyi’s entropy are parameters for the justification of the proposed approach. The result shows the best wavelet to be chosen among the available discrete wavelet based on various entropy measurements.

A deep learning framework for adaptive compressive sensing of high‐speed train vibration responses

Onboard monitoring plays an important role in real-time condition assessment of rail systems. However, the data amount is typically tremendous due to the high sampling rate needed and long traveling distance, especially for vibration data collected from high-speed trains (HSTs). As for fault diagnosis of mechanical systems, compressive sensing (CS) has been increasingly adopted to reduce the data amount. In comparison to rotary bearings and bolted joints in machinery that operate in relatively steady working environments, HSTs run in an open and varying environment throughout the traveling mileage, and the data amount is normally much larger, making it tricky to directly apply the classical CS methods. This study aims to bridge the gap by investigating the sparsity of HST vibration signals and CS approaches. Considering the lack of sparsity and long reconstruction time, we propose an efficient adaptive CS approach for dynamic responses of HSTs. More specifically, we unroll the iterative soft thresholding algorithm (ISTA) in a deep learning (DL) framework and configure it into a data reconstruction machine. Compared to the conventional CS methods, our approach exhibits two advantages: (i) The dictionary learning and signal reconstruction are integrated into one neural network and can be conducted in an end-to-end manner; (ii) the process is highly efficient since encapsulating ISTA in a DL framework can naturally leverage the capability of GPU. The proposed approach is validated using data collected from an in-service HST, and results show that our approach achieves superior reconstruction performance over fixed bases and redundant dictionaries.

Acoustic Signal Classification Using Symmetrized Dot Pattern and Convolutional Neural Network

The classification of sound signals can be applied to the fault diagnosis of mechanical systems, such as vehicles. The traditional sound classification technology mainly uses the time-frequency domain characteristics of signals as the basis for identification. This study proposes a technique for visualizing sound signals, and uses artificial neural networks as the basis for signal classification. This feature extraction method mainly uses a principle to convert a time domain signal into a coordinate symmetrized dot pattern, and presents it in the form of snowflakes through signal conversion. To verify the feasibility of this method to classify different noise characteristic signals, the experimental work is divided into two parts, which are the identification of traditional engine vehicle noise and electric motor noise. In sound measurement, we first use the microphone and data acquisition system to measure the noise of different vehicles under the same operating conditions or the operating noise of different electric motors. We then convert the signal in the time domain into a symmetrized dot pattern and establish an acoustic symmetrized dot pattern database, and use a convolutional neural network to identify vehicle types. To achieve a better identification effect, in the process of data analysis, the effect of the time delay coefficient and weighting coefficient on the image identification effect is discussed. The experimental results show that the method can be effectively applied to the identification of traditional engine and electric vehicle classification, and can effectively achieve the purpose of sound signal classification.

Highly imbalanced fault diagnosis of mechanical systems based on wavelet packet distortion and convolutional neural networks

The healthy operations of mechanical systems are crucially important for ensuring human safety and economic benefits, so that there is a high demand on the automatic fault diagnosis techniques. However, the number of available faulty samples of mechanical systems is often far less than healthy samples, and thereby the traditional data-driven methods often suffer a high rate of misdiagnosis. In this paper, a new fault diagnosis method is developed on the basis of wavelet packet distortion and convolutional neural networks. First, wavelet packet distortion means that wavelet packet coefficients are distorted to augment fault samples, in order to achieve the equilibrium between healthy and faulty classes. Second, a convolutional neural network-based classification model is trained using the balanced training dataset. Third, the trained model is applied to classify the testing samples. Finally, the efficacy of this developed method in imbalanced fault diagnosis of mechanical systems is demonstrated through a number of experiments.

Mechanical fault diagnosis by using dynamic transfer adversarial learning

Different machine learning approaches have been developed for the fault diagnosis of mechanical systems. To achieve desired diagnosis performance, lots of labeled one-dimensional (1D) signals are required for training machine learning models. however, those signals collected under various working conditions are difficult to be used for both diagnosis model training and testing. For real applications, moreover, the collection of labeled data is more difficult than that of unlabeled ones. To tackle the above challenging points, a dynamic transfer adversarial learning (DTAL) network is proposed for dealing with unsupervised fault diagnosis missions. To this end, an improved feature extractor is developed to deal with 1D mechanical vibration signals. A dynamic adversarial factor is presented to automatically adapt the marginal distribution of the global domain. The conditional distribution of the local domain is employed to make the model independent of training multiple classifiers, so as to reduce the computational burden of the proposed method. The addressed DTAL was evaluated using fault diagnosis experiments for a wind turbine gearbox and benchmark bearings. Compared with other state-of-the-art methods, it has better accuracy and robustness as highlighted by experimental results. The developed model can improve the diagnosis performance under various workloads for mechanical systems.

Fault diagnosis for mechanical system using dynamic Bayesian network

The present study focuses on the fault diagnosis of mechanical systems. Mechanical systems are considered with interconnected components that work together to achieve a common function or purpose. On the one hand, the fault diagnosis result is affected by strong dependence between each component. One the other hand, diagnostic results may be different at different time slices because of the performance degradation of components when the same fault symptoms are given. To deal with these problems in diagnosis, a dynamic Bayesian network (DBN) model is proposed. First, series and parallel systems are converted to a Bayesian network. And the relationship between components and reliability of the system is expressed by the Bayesian network. Then, the dynamic Bayesian network is established to model the dynamic degradation of components in a system under additional information by using the wear data. The parameters of the model are estimated by historical data. Finally, a case is investigated to verify the proposed model in this study. Fault diagnosis is conducted through a backward analysis of the DBN model proposed, and the weakest component is identified. The dynamic probabilities of the mechanical system are obtained through forwarding analysis of the DBN model.

A Comparative Analysis of Artificial Intelligence-Based Methods for Fault Diagnosis of Mechanical Systems

Abstract The present research studied fault diagnosis of composite sheets using vibration signal processing and artificial intelligence (AI)-based methods. To this end, vibration signals were collected from sound and faulty composite plates. Using different time-frequency signal analysis and processing methods, a number of features were extracted from these signals and the most effective features containing further information on these composite plates were provided as input to different classification systems. The output of these classification systems reveals the faults in composite plates. The different types of classification systems used in this research were the support vector machine (SVM), adaptive neuro-fuzzy inference system (ANFIS), k-nearest neighbor (k-NN), artificial neural networks (ANNs), Extended Classifier System (XCS) algorithm, and the proposed improved XCS algorithm. The research results were reflective of the superiority of ANFIS in terms of precision, while this method had the highest process duration with an equal number of iterations. The precision of the proposed improved XCS method was lower than that of ANFIS, but the duration of the process was shorter than the ANFIS method with an equal number of iterations.

Composite-Graph-Based Sparse Subspace Clustering for Machine Fault Diagnosis

Sparse subspace clustering (SSC) is an effective method to cluster sensing signal for fault diagnosis in mechanical systems. SSC is based on a global expression strategy describing each data point by other data points from all the potential clusters. A drawback of this strategy is that it generates nonzero elements in the nondiagonal blocks of the similarity matrix, thereby reducing the performance of the matrix for data discrimination. To improve SSC's performance, a composite-graph-based SSC (CG-SSC) method is developed by introducing distance among the data points into SSC, where the L1-norm of the sparse coefficients is replaced by the product of the data distance and the sparse coefficients. The distance serves as a constraint to the amplitude of the sparse coefficients. A sparse coefficient with small amplitude is assigned if the distance between the expressed data and the expressing data is large, and vice versa. This method supports the construction of a compact cluster. The effectiveness of the presented method is experimentally verified by data measured on several bearings and gearboxes with different types of faults. The method is also compared with classical clustering methods, and the results indicate its advantage for data sorting and clustering.

An experimental platform for statistical fault diagnosis in propelled mechanical systems: an academic exercise

Statistical data analysis for fault diagnosis in mechanical systems is a fundamental tool, for instance, in applied mechanical engineering. In order to capture a feasible data set, a well designed electronic instrumentation and excitation system signal stages are mandatory. Hence, one objective of this paper is to develop a low cost vibration sensor based on an inductive LC-tank oscillator (a resonant inductive-capacitive electronic circuit carefully designed to produce an harmonic electrical signal), and then to tune an effective excitation system signal to our experimental platform. This platform uses a propelled drone motor mounted on a beam structure to emulate a propelled rotating machine. Essentially, two data set were acquired. One for the healthy behaviour of the developed system, and the other for a programmed faulty scenario. This defective case was realized by introducing a small mechanical fault in one blade extreme of the mechanical propelled system. To note, this faulty scenario is almost impossible to deduce by just seen the raw data. The other objective of this paper is to analyze the obtained data sets by utilizing a statistical data analysis tool. Then, by employing box-plot diagrams, the healthy and faulty cases become evidenced. Finally, and due to we are proposing a low-cost academic experimental platform for fault diagnosis based on data analysis, our platform’s toll was around 120 euros. Hence, this platform results applicable to teach data analysis from dynamical systems.