Mechanical Fault Detection Research Articles

Genetic algorithm based production knowledge base for mechanical fault detection model

Mechanical fault detection has an important influence on production schedule and efficiency. With the development of intelligent technology, more and more intelligent detection technologies are applied to mechanical fault detection. In order to detect mechanical faults more efficiently and accurately, this experiment proposes a production knowledge base model based on genetic algorithm (GA algorithm). The model uses the unique biological genetics principle of genetic algorithm to evolve the interested population, and can conduct spatial search to find the global optimal solution. By comparing the performance of GA algorithm model with other similar detection models, it is found that the model proposed in the experiment has obvious advantages in mechanical fault detection performance. The experimental results show that the maximum accuracy of the GA algorithm is 0.935, 0.074 higher than the support vector machine (SVM) model, 0.118 higher than the linear discriminant analysis (LDA) model, 0.032 higher than the random forest (RF) model, and 0.166 higher than the K nearest neighbor (KNN) model. In addition, the error value of GA algorithm is the lowest among these models, which is 0.028. This proves that the genetic algorithm model has higher diagnostic accuracy and can play an important role in mechanical fault detection.

Deep learning based mechanical fault detection and diagnosis of electric motors using directional characteristics of acoustic signals

Early identification of rotating machinery faults is crucial to avoid catastrophic failures upon installation. Contact-based vibration acquisition approaches are traditionally used for the purpose of machine health monitoring and end-of-line quality control. In complex working conditions, it can be difficult to perform an accurate accelerometer based vibration test. Acoustic signals (sound pressure and particle velocity) also contain important information about the operating state of mechanical equipment and can be used to detect different faults. A deep learning approach, namely one-dimensional Convolution Neural Networks (1D-CNN) can directly process raw time signals thereby eliminating the human dependance on fault feature extraction. An experimental research study is conducted to test the proposed 1D-CNN methodology on three different electric motor faults. The results from the study indicate that the fault detection performance from the new acoustic-based method is very effective and thus can be a good replacement to the conventional accelerometer-based methods for detection and diagnosis of mechanical faults in electric motors.

Comparison of acoustic signal–based fault detection of mechanical faults in induction motors using image classification models

This study presents a novel deep neural network–based method for fault detection in induction motors. The focus was on identifying five types of mechanical cases: normal operation, shaft/load breakage, misalignment, mounting bolt looseness, and cooling fan problems. To increase the realism of the results, a laboratory-collected dataset of stereo microphone recordings was augmented with real factory noise. The audio data was transformed into image data using Mel-frequency cepstral coefficients as the feature extraction method and then processed with image-based classifiers. A comparison was made among 12 different networks in terms of accuracy and number of parameters, revealing that Mobilenet_v2, EfficientNetV2B0, and NASNetMobile had the best performance in terms of both network size and accuracy.

Artificial Intelligence and Machine Learning in Grid Connected Wind Turbine Control Systems: A Comprehensive Review

As grid-connected wind farms become more common in the modern power system, the question of how to maximize wind power generation while limiting downtime has been a common issue for researchers around the world. Due to the complexity of wind turbine systems and the difficulty to predict varying wind speeds, artificial intelligence (AI) and machine learning (ML) algorithms have become key components when developing controllers and control schemes. Although, in recent years, several review papers on these topics have been published, there are no comprehensive review papers that pertain to both AI and ML in wind turbine control systems available in the literature, especially with respect to the most recently published control techniques. To overcome the drawbacks of the existing literature, an in-depth overview of ML and AI in wind turbine systems is presented in this paper. This paper analyzes the following reviews: (i) why optimizing wind farm power generation is important; (ii) the challenges associated with designing an efficient control scheme for wind farms; (iii) a breakdown of the different types of AI and ML algorithms used in wind farm controllers and control schemes; (iv) AI and ML for wind speed prediction; (v) AI and ML for wind power prediction; (vi) AI and ML for mechanical component monitoring and fault detection; and (vii) AI and ML for electrical fault prevention and detection. This paper will offer researchers and engineers in the wind energy generation field a comprehensive review of the application of AI and ML in the control methodology of offshore and onshore wind farms so that more efficient and robust control schemes can be designed for future wind turbine controllers.

Vibration analysis for the detection of combined mechanical defects of a wind turbine

Wind energy is a source of renewable energy derived from the kinetic force of the wind. It has experienced the greatest expansion due, above all, to its low impact on the environment and the gradual reduction of costs, which produce clean, competitive and economically viable energy. In order to ensure the stability of the energy produced, many studies are working to develop a reliable maintenance process based on new online diagnostic techniques, fast and accurate, optimised for wind turbine systems. For this, an experimental study was carried out on a wind system based on electric generators with the aim of contributing to the diagnosis and predictive detection of combined mechanical faults, where the configuration is complex because one fault can hide another. For this, it is imperative to associate each symptom with the appropriate method or methods in order to highlight it unambiguously to detect each defect at the appropriate time. Hence the interest in using more elaborate techniques in order to improve the procedures for detecting and analysing combined faults.

Condition monitoring and fault diagnosis strategy of railway point machines using vibration signals

Abstract Condition monitoring of railway point machines is important for train operation safety and effectiveness. Referring to the fields of mechanical equipment fault detection, this paper proposes a fault detection and identification strategy of railway point machines via vibration signals. A comprehensive feature distilling approach by combining variational mode decomposition (VMD) energy entropy and time- and frequency-domain statistical features is presented, which is more effective than single type of feature. The optimal set of features was selected with ReliefF, which helps improve the diagnosis accuracy. Support vector machine (SVM), which is suitable for a small sample, is adopted to realize diagnosis. The diagnosis accuracy of the proposed method reaches 100%, and its effectiveness is verified by experiment comparisons. In this paper, vibration signals are creatively adopted for fault diagnosis of railway point machines. The presented method can help guide field maintenance staff and also provide reference for fault diagnosis of other equipment.

Equipment for Monitoring Synchronous Generators Condition through External Magnetic Field Waveforms

This paper presents an equipment for monitoring synchronous generators condition through characteristics of the time derivative of the external magnetic field. The developed monitoring methodology allows the identification of established or incipient faults, by detecting changes in the magnetic signature of the synchronous generator. In this methodology, the measurement of signals outside the machine gives this equipment a non-invasive characteristic, allowing its monitoring without interfering or disturbing its operation. The developed system includes the specification of magnetic field sensors, signal measurement and processing equipment, as well as software for analysis and monitoring. The validation of the methodology used in this system was carried out through the analysis of experimental data, presenting efficient results in the detection of electrical and mechanical faults in synchronous generators of an experimental test bench and a hydroelectric power plant. As a result, the commercial specification of this equipment was obtained and two units were implemented in a hydroelectric power plant to monitor 305 MVA synchronous generators.

Defective sewing stitch semantic segmentation using DeeplabV3+ and EfficientNet

Defective stitch inspection is an essential part of garment manufacturing quality assurance. Traditional mechanical defect detection systems are effective, but they are usually customized with handcrafted features that must be operated by a human. Deep learning approaches have recently demonstrated exceptional performance in a wide range of computer vision applications. The requirement for precise detail evaluation, combined with the small size of the patterns, undoubtedly increases the difficulty of identification. Therefore, image segmentation (semantic segmentation) was employed for this task. It is identified as a vital research topic in the field of computer vision, being indispensable in a wide range of real-world applications. Semantic segmentation is a method of labeling each pixel in an image. This is in direct contrast to classification, which assigns a single label to the entire image. And multiple objects of the same class are defined as a single entity. DeepLabV3+ architecture, with encoder-decoder architecture, is the proposed technique. EfficientNet models (B0-B2) were applied as encoders for experimental processes. The encoder is utilized to encode feature maps from the input image. The encoder's significant information is used by the decoder for upsampling and reconstruction of output. Finally, the best model is DeeplabV3+ with EfficientNetB1 which can classify segmented defective sewing stitches with superior performance (MeanIoU: 94.14%).

Construction cost optimization technology based on decision tree CHAID algorithm

With the rapid growth of the construction engineering industry, the construction industry has made great progress, but in the process of its development, there have been many irrational growth models such as extensive growth and real estate bubbles. In addition, with the continuous advancement of the development of construction projects, the types and modes of construction projects have become more and more complex, and the importance of construction project management has become more and more prominent. This article will discuss the cost of construction projects. Make a relative cost optimization technology research, this paper adopts the decision tree CHAID algorithm to solve the cost management problem of construction projects, and proposes corresponding cost optimization technical solutions, which is conducive to the integration of China's construction industry with the international industry and the promotion of my country's industrial structure transformation. From the perspective of decision tree structure, the earliest decision tree is a multi-forked tree structure, which is it brings some trouble to interpreting the rules of the decision tree, while the decision tree CHAID algorithm generates rules that are easy to understand, easy to implement, fast, able to process data-type and general-type attributes at the same time and calculate massive data good. This paper mainly studies the design of mechanical equipment fault detection robot, proposes a series of data model analysis, and analyzes the main theories and characteristics of the electronic tracing algorithm. 3.98s, the running time of single-machine multi-threading is 3.87s, while the running time of decision tree CHAID is 3.79s, the relative acceleration is slightly higher than that of single-machine single-threaded and single-machine multi-threaded.

Fuzzy Dispersion Entropy: A Nonlinear Measure for Signal Analysis

Entropy is a powerful tool for nonlinear analysis of time series. We have recently introduced dispersion entropy (DispEn), which has widely drawn attention from researchers in a variety of settings in data analysis. However, DispEn is sensitive to its parameters, especially number of classes (quantization level). Additionally, in some cases, a small change in the signal amplitude due to noise changes the quantized series and so entropy value. To address these limitations, we develop fuzzy dispersion entropy (FuzzDispEn) on the basis of fuzzy membership functions for the quantization of a signal and Shannon entropy (ShEn). In FuzzDispEn, the concept of fuzzification in ShEn-based methods is introduced for the first time. Several straightforward concepts in signal processing using a set of time series were utilized to show the advantages of FuzzDispEn over DispEn in terms of the detection of the dynamical variability of signals. The results showed that FuzzDispEn has lower sensitivity to noise and signal length compared to DispEn. FuzzDispEn was also applied to five real world applications: mechanical fault detection in gearbox, fault diagnosis of rolling element bearings, damage level detection of bearings, detection of blood pressure in salt-sensitive rats, and detection of seizure and seizure-free electroencephalographics (EEGs). The results suggest that FuzzDispEn outperforms DispEn as well as popular sample entropy and permutation entropy in distinguishing different kinds of dynamical patterns in synthetic and real data.

Experimental Assessment of a Broadband Vibration and Acoustic Emission Sensor for Rotorcraft Transmission Monitoring

Modern rotorcrafts rely on Health and Usage Monitoring Systems (HUMS) to enhance their availability, reliability, and safety. In those systems, data related to the health of key mechanical components is acquired, in addition to typical flight condition history data such as speed and torque. Commercial HUM systems usually rely on vibration measurements to assess the condition of shafts, gears, and bearings; using techniques such as spectral analysis, harmonic analysis, vibration trend and others. Recent research has shown that acoustic emissions (AE) can be advantageous in the detection of mechanical faults, in particular detecting very early small defects on bearings and gears, providing extra time for maintenance planning. However, the addition of extra sensors adds complexity and weight to the HUMS system, which is undesirable. This research is an experimental study to assess the monitoring capabilities of a broadband sensor, able to cover both low frequency vibration components as well as ultrasonic events, hence combining the benefits of both in a single compact sensing unit. The experimental results obtained from an instrumented rig using healthy components as well as seeded faults show the ability of the sensor to detect high frequency events, and compares the performance of the sensor in the low frequency range with a commercial accelerometer.

Vibration signal acquisition and computer simulation detection of mechanical equipment failure

Abstract The author in order to solve the problem of optimizing the accuracy of mechanical equipment failure detection proposes a vibration signal collection and computer simulation detection for mechanical equipment failure. Using wavelet domain Wiener filtering-based mechanical equipment fault detection method, the author first combined Wiener filtering and wavelet threshold filtering, established a vibration signal collection model for mechanical equipment, obtained the true signal and the filtered signal error and fusion of the principle of wavelet threshold filtering to perform orthogonal wavelet transform on noisy signals, and through the principle of fast independent component analysis to separate the vibration signals of mechanical equipment, build the initial separation matrix of the signal with unit variance, and found the estimated value of the source signal; the collection of vibration signals of mechanical equipment is completed according to the estimated value, realizing the optimization of the detection of mechanical faults and equipment failures. The simulation results prove that the signal-to-noise ratio of the vibration signal of mechanical equipment collected by this method is higher than 15.7% of the signal-to-noise ratio based on the FastICA method; this is mainly because when this method is used for anti-interference collection of vibration signals of mechanical equipment, combining the principle of fast independent component analysis to separate the vibration signals of mechanical equipment, construct the initial separation matrix of the signal with unit variance. Using the proposed method for signal acquisition can greatly reduce the error, and it can provide effective support for fault detection of mechanical equipment.

Retraction Note to: Detection and classification of mechanical faults of three phase induction motor via pixels analysis of thermal image and adaptive neuro-fuzzy inference system

Retraction Note to: Detection and classification of mechanical faults of three phase induction motor via pixels analysis of thermal image and adaptive neuro-fuzzy inference system

Effects of power transformer high-frequency equivalent circuit parameters non-uniformity on fault diagnosis using SFRA test

The electric power transformer is an essential component of an electrical power system because it regulates the voltage level to provide the best possible system operation. Thus, it is important to detect the transformer faults to maintain its reliability as high as possible. The sweep frequency response analysis (SFRA) test is one of the tests which, used to diagnose transformer faults. It determines the fault type by analyzing its response across a wide frequency range. The main advantage of the SFRA test is the detection of mechanical deformation faults since mechanical deformation faults cause the high-frequency equivalent circuit parameters to change which, could be detected easily using the SFRA test. Also, statistical features are defined for this SFRA response used in fault detection and classification. The transformer high-frequency equivalent circuit parameters are non-uniform distributed over discs due to normal leakages and aging. Thus, in the present paper, the non-uniform distribution of the transformer high-frequency equivalent circuit is studied in the SFRA test at different faults and various non-uniformity levels. Finally, when considering the non-uniform distribution of high-frequency equivalent circuit parameters, the results demonstrate that some features are greatly affected, and others are just minimally affected. Depending on how much each feature is affected by the uniformity of high-frequency equivalent circuit parameters, rules of fault diagnosis algorithms based on the SFRA test could be reweighted to improve their accuracy. Also, the obtained results could be compared to real measured data to estimate the actual non-uniformities of transformer parameters.

Permanent Magnet Synchronous Motor Driving Mechanical Transmission Fault Detection and Identification: A Model-Based Diagnosis Approach

This paper presents a model-based scheme for permanent magnet synchronous motor (PMSM) driving transmission fault detection and identification (FDI) in a steady-state condition. The proposed framework utilizes a PMSM state-space model and an approximated transmission model to construct the regression models for parameter estimation using the Recursive Least-Square (RLS) algorithm. The FDI are accomplished by the residual current spectrum thresholding method to assess the fault characteristic frequency magnitude and also by parameter clustering. Two types of mechanical transmission with three different fault conditions are tested in the experiments. As a preliminary effort in the condition monitoring of PMSM driving transmission, the study results demonstrate a promising approach by considering both residual current spectrum and parameter cluster, which achieved a satisfactory decision making in detecting and identifying the faulty condition.

Machine learning techniques applied to mechanical fault diagnosis and fault prognosis in the context of real industrial manufacturing use-cases: a systematic literature review.

When put into practice in the real world, predictive maintenance presents a set of challenges for fault detection and prognosis that are often overlooked in studies validated with data from controlled experiments, or numeric simulations. For this reason, this study aims to review the recent advancements in mechanical fault diagnosis and fault prognosis in the manufacturing industry using machine learning methods. For this systematic review, we searched Web of Science, ACM Digital Library, Science Direct, Wiley Online Library, and IEEE Xplore between January 2015 and October 2021. Full-length studies that employed machine learning algorithms to perform mechanical fault detection or fault prognosis in manufacturing equipment and presented empirical results obtained from industrial case-studies were included, except for studies not written in English or published in sources other than peer-reviewed journals with JCR Impact Factor, conference proceedings and book chapters/sections. Of 4549 records, 44 primary studies were selected. In 37 of those studies, fault diagnosis and prognosis were performed using artificial neural networks (n = 12), decision tree methods (n = 11), hybrid models (n = 8), or latent variable models (n = 6), with one of the studies employing two different types of techniques independently. The remaining studies employed a variety of machine learning techniques, ranging from rule-based models to partition-based algorithms, and only two studies approached the problem using online learning methods. The main advantages of these algorithms include high performance, the ability to uncover complex nonlinear relationships and computational efficiency, while the most important limitation is the reduction in model performance in the presence of concept drift. This review shows that, although the number of studies performed in the manufacturing industry has been increasing in recent years, additional research is necessary to address the challenges presented by real-world scenarios.

Unsupervised adversarial domain adaptive for fault detection based on minimum domain spacing

The deep learning model has gradually matured in the detection of mechanical faults. However, due to the changes in the mechanical operating environment and the application of new sensors in real work, the effect of the training model is not ideal in field applications. The key of this problem is the deviation of feature space mapping between training source domain and application target domain. This paper proposes an unsupervised adversarial domain adaptive fault diagnosis transfer learning model based on the minimum domain spacing to reduce the deviation. In adversarial network training, by training the weight parameters of the classifier, some features extracted by the composed classifier are added to the feature distribution of the target domain through weight changes, which reduces the feature distribution difference between the source domain and the target domain. It is reflected on the reduction of the maximum mean difference distance (MMD) between the two domains, and the fitting features of the data distribution are improved. Finally, through two experimental platforms of rolling bearing and planetary gearbox dataset, the results of six diagnostic tasks show that the new model reduces the amount of parameters by 33.66% and keeps accuracy more than 99% compared with the DANN model under the condition.

Mechanical fault detection based on machine learning for robotic RV reducer using electrical current signature analysis: a data-driven approach

Abstract Recently, prognostic and health management (PHM) has become a prominent field in modern industry. The rotate vector (RV) reducer is one of the widely used mechanical components in industrial systems, specifically in robots. The RV reducer is known for its unique characteristics of small size, efficient speed transmission, and high torsion. The RV reducer is prone to several kinds of faults, due to its continuous operation in an industrial robot. To keep the operation smooth and steady, timely PHM of the RV reducer has become essential. Previously, the RV reducer fault was diagnosed via various techniques, such as ferrography analysis, vibration analysis, and acoustic emission analysis. However, these conventional techniques have various issues. To resolve those issues, we introduce a novel approach to use the embedded electrical current system for the fault detection of the RV reducer. However, this is quite complicated to investigate mechanical fault using an electrical current signature, since the RV reducer is not an integral part of the electric motor, and finding a fault pattern in faulty components needs thorough examination. We therefore focus on the application of machine learning (ML) for fault classifications. We present an approach for feature extraction, feature selection, and feature reduction using the information obtained from the motor current signature analysis to create an ML-based fault classification system with distinguishable prominent features. Finally, the authenticity of the presented approach is justified via the improved values of evaluating parameters, such as accuracy, specificity, and sensitivity, for ML classifiers.

Classification of Fault Severity in Induction Machine Systems Based on Temporal Convolutions and Recurrent Networks

Detection and severity identification of mechanical and electrical faults by means of noninvasive methods such as electrical signatures of induction machine have attracted much attention in recent years. Since operating conditions of machines and severity of faults in incipient stages influence the amplitude of fault index in the fault detection process, diagnosing fault occurrence and severity can be more complicated. In this study, an efficient method for fault detection and classification in induction machine based on deep neural networks is introduced. The introduced method applies the long short-term memory (LSTM) and fully convolutional neural networks (FCNs) in a conjoined manner. The authors use the FCN architecture for feature extraction from the time-series signal and augment it with LSTM to improve classification performance. This structure has not been previously applied for fault severity detection in induction machine systems. The authors avoid manual feature engineering and, by eliminating the preprocessing phase, directly use time series of electrical signals for fault detection and classifications. The experimental results have been carried out in different fault severities and loads. The analysis of the results and comparison with other deep and classical methods show that the faulty cases can be separated based on severity and load levels with a high accuracy (98.92%), which shows that the adopted architecture is successful in automatically extracting discriminative features from the signal.

Mechanical faults detection for vehicle motors under nonstationary conditions based on Vold-Kalman order tracking method

The fault diagnosis and health management of motors are the key technology that needs to be resolved in the future with the development of electric vehicles. The drive motors usually working under nonstationary conditions. Moreover, the mechanical fault is the most common type among the motor fault types. Therefore, developing the mechanical fault detection method under variable speed conditions would be of great value to improve the diagnosis accuracy of vehicle motors. This paper deals with the mechanism of the bearing fault, the rotor eccentricity fault, and their compound fault of permanent magnet synchronous motors. The purpose of this paper is to propose a mechanical fault detection method based on Vold-Kalman for vehicle motors under variable working conditions. First, the vibration characteristics of the healthy and the faulty motors are analyzed. Next, for eliminating the influence of variable conditions on the mechanical-fault detection, the fault characteristics are extracted by applying the Vold-Kalman filter. Finally, experiments are carried out based on the selected part of the new European driving cycle test conditions. The experimental results show that the proposed approach can detect the mechanical fault of the motor under nonstationary conditions effectively.