Mechanical Fault Detection Research Articles

Discrimination Between Radial Deformation and Axial Displacement in Power Transformers Using Analysis of Electromagnetic Waves

In recent years, the electromagnetic wave application has been proposed for the detection of mechanical defects in power transformer windings. After detection of a defect, the determination of defect type is the next stage of monitoring. In this paper, a comparative method is presented to discriminate between radial deformation and axial displacement in power transformer high-voltage winding using electromagnetic waves. In this paper, the magnitude Euclidean distance (MED) index is used to compare the received signals in sound and defective conditions. In the proposed method, two receiving antennas and one transmitter antenna are used to emit ultra wideband signals to the winding and receive their reflections. Two receiving antennas receive signals at two different positions, as a result, in the case of a mechanical defect occurrence, the amplitude of the received signals and the MED index will be different. Hence, taking the advantage of this feature, the type of defect can also be determined after its detection. The proposed method is evaluated by a simulated model, and then, experimental tests are conducted on an actual transformer winding to show its validity.

Fracture mechanics and mechanical fault detection by artificial intelligence methods: A review

Abstract Artificial intelligence (AI) researchers created new techniques, developed and applied them to solve engineering problems since two decades. Although lots of AI techniques and approaches are available in mechanical engineering, there isn’t any survey aiming to review the existing works, systems and applications in the field of fracture mechanics. In this paper, the state of the art of five AI methods which are used in the field of fracture mechanics, is surveyed. This review is performed from the technical point of view on particular applications of artificial neural networks, Bayesian networks, genetic algorithms, fuzzy logic and case-based reasoning. After an overview of AI methods, sub-domains of engineering fracture mechanics with respect to the fault and failure analysis are described. The existing works from 1990 to 2016 are analysed and discussed in four categories as sub-domains of fracture mechanics: (a) failure mode and failure mechanism identification, (b) damage and failure detection and diagnosis, (c) fault and error detection, diagnosis and (d) mechanical fracture and fracture parameters. We analyse literature based on a classification of these five AI methods in order to highlight their main concepts and explain how they are applied in these sub-domains of fracture mechanics. Our analysis and discussion in this paper shows the advantages, limitations and research gaps in this field. Finally, perspectives and future research directions are outlined.

Simulation-driven machine learning: Bearing fault classification

Abstract Increasing the accuracy of mechanical fault detection has the potential to improve system safety and economic performance by minimizing scheduled maintenance and the probability of unexpected system failure. Advances in computational performance have enabled the application of machine learning algorithms across numerous applications including condition monitoring and failure detection. Past applications of machine learning to physical failure have relied explicitly on historical data, which limits the feasibility of this approach to in-service components with extended service histories. Furthermore, recorded failure data is often only valid for the specific circumstances and components for which it was collected. This work directly addresses these challenges for roller bearings with race faults by generating training data using information gained from high resolution simulations of roller bearing dynamics, which is used to train machine learning algorithms that are then validated against four experimental datasets. Several different machine learning methodologies are compared starting from well-established statistical feature-based methods to convolutional neural networks, and a novel application of dynamic time warping (DTW) to bearing fault classification is proposed as a robust, parameter free method for race fault detection.

A Wireless Sensor System for Real-Time Monitoring and Fault Detection of Motor Arrays.

This paper presents a wireless fault detection system for industrial motors that combines vibration, motor current and temperature analysis, thus improving the detection of mechanical faults. The design also considers the time of detection and further possible actions, which are also important for the early detection of possible malfunctions, and thus for avoiding irreversible damage to the motor. The remote motor condition monitoring is implemented through a wireless sensor network (WSN) based on the IEEE 802.15.4 standard. The deployed network uses the beacon-enabled mode to synchronize several sensor nodes with the coordinator node, and the guaranteed time slot mechanism provides data monitoring with a predetermined latency. A graphic user interface offers remote access to motor conditions and real-time monitoring of several parameters. The developed wireless sensor node exhibits very low power consumption since it has been optimized both in terms of hardware and software. The result is a low cost, highly reliable and compact design, achieving a high degree of autonomy of more than two years with just one 3.3 V/2600 mAh battery. Laboratory and field tests confirm the feasibility of the wireless system.

A Novel Parallel Modelling-Wavelet Based Mechanical Fault Detection Using Stator Current Signature of Induction Machine under Variable Load Conditions

Induction Machine (IM) fault detection techniques such as Fast Fourier Transform (FFT) which is a popular steady-state analysis method is recognized to be highly dependent on the IM loading and speed conditions. Nonetheless, implementing an FFT or even Short Time Fourier Transform (STFT) will result in low resolution frequency characteristics especially under a variable speed and loading conditions. Consequently, fault detection and classification under variable loading and speed conditions is quite inconvenient. Since, mechanical faults are one of the major breakdowns, which occur in IMs, it needs to be addressed to prevent breakage andfault extension. This paper investigates and detects faults under variable loading and speed conditions by studying the Motor Current Signature Analysis (MCSA) using a novel developed parallel technique based on the discrete wavelet transform (DWT).The proposed model input would be MCSA with the similar drive, loading condition and constraints for both healthy and faulty electric motors and the Discrete Wavelet Transform (DWT) is used as a detection technique. The proposed technique uses the DWT to the IM’s stator current to extract the desired features including Min, Max, Standard deviation, and Energy of the signal as a specific vector for each electric motor. In addition, different mechanical faults including Rotor broken bar(s), eccentricity and bearing have been detected using the proposed technique. Then, the accuracy of proposed parallel model using DWT is verified using experimental set-up of the parallel machines under variable loading and speed conditions for different mechanical faults. Finally the results of the proposed technique for all mechanical faults is tabulated and presented.

Amplification of Single Mechanical Fault Signatures Using Full Adaptive PMSM Observer

This paper presents a localized mechanical fault detection method using a full adaptive permanent-magnet synchronous machine (PMSM) observer in a d / q reference frame. It appears that the fault sensitivity of PMSM stator currents and rotor speed measurements is very low with regard to single mechanical faults. That is why an amplification of single mechanical fault signatures in frequency domain is proposed by adjusting the adaptive PMSM observer gains. To study the influence of the adaptive PMSM observer gains on fault signatures, the observer model is first of all linearized at an operating point. Then, the static gain, the resonance, and the quality factor Q of different transfer functions are analyzed. A dedicated experimental setup based on an original mechanical fault emulator and a 7.8 kW PMSM drive is designed to validate the simulation results. The simulation and the experimental results show that the proposed method is effective to amplify single mechanical fault signatures and the calculated fault indicator.

Fast–ICA for Mechanical Fault Detection and Identification in Electromechanical Systems for Wind Turbine Applications

Recently, the approaches based on source separation are increasingly adopted for the fault diagnosis in several industrial applications. In particular, Independent Component Analysis (ICA) method is attractive, thanks to its simplicity of implementation. In the context of electrical rotating machinery with a variable speed, namely the wind turbine type, the interaction between the electrical and mechanical parts along with the fault is complex. Therefore, the essential system variables are affected and it thereby requires to be analyzed in order to detect the presence of certain faults. In this paper, the target system is the classical association of a doubly-fed induction motor to a two stage gearbox for wind energy application system. The investigated mechanical fault is a uniform wear of two gear wheels for the same stage. The idea behind the proposed technique is to consider the fault detection and identification as a source separation problem. Based on the analysis into independent components, Fast–ICA algorithm is adopted to separate and identify the sources of the gear faults. Afterwards, a spectral analysis is applied on the signals resulting from the separation in order to identify the fault components related to the damaged wheels. The efficiency of the proposed technique for the separation and identification of the fault components is evaluated by numerical simulations.

A novel intelligent method for bearing fault diagnosis based on affinity propagation clustering and adaptive feature selection

Bearings faults are one of the main causes of breakdown of rotating machines. Thus detection and diagnosis of mechanical faults in bearings is very crucial for the reliable operation. A novel intelligent fault diagnosis method for roller bearings based on affinity propagation (AP) clustering algorithm and adaptive feature selection technique is proposed to better equip with a non-expert to carry out diagnosis operations. Ensemble empirical mode decomposition (EEMD) and wavelet packet transform (WPT) are utilized to accurately extract the fault characteristic information buried in the vibration signals. Moreover, in order to improve the efficiency of clustering algorithm and avoid the curse of dimensionality, a new adaptive features selection technique is developed in this work, whose effectiveness is verified in comparison with other methods. The proposed intelligent method is then applied to the bearing fault diagnosis. Results demonstrate that the proposed method is able to reliably and accurately identify different fault categories and severities of bearings.

Online condition monitoring of wind turbines through three-phase electrical signature analysis

In the context of the KAStrion European project, a complete solution was proposed in order to monitor wind turbines. The developed solution comprises both hardware and software parts of the condition monitoring system. In terms of software, modules for vibration analysis and electrical signal analysis have been developed. The current paper presents the electrical analysis solution proposed in the context of this project. The electrical module is able to detect both mechanical and electrical faults in a wind turbine system. The goal of the mechanical fault detection using electrical signals is to confirm the faults also detected by vibration analysis, while the main focus of the module remains the detection of electrical faults. Results showing the performance of mechanical fault detection are presented using electrical signals acquired on the test-bench developed for testing the KAStrion system. Moreover, results regarding the electrical unbalance are presented using signals acquired on a three-phase transformer. The final solution has been implemented on two onshore wind turbines since the end of 2014, and online condition monitoring results are presented at the end of the paper.

Soft sensor design for mechanical fault detection in PMSM at variable speed

Abstract This paper presents a soft sensor (SS) used to localize mechanical faults in a Permanent Magnet Synchronous Machine (PMSM) at variable speed. The speed contains crucial information about the fault and the characteristic frequencies sought can be amplified by a proper observer design to improve the detection procedure. On the other hand, all the measurements are expressed as a function of estimated position in order to obtain a speed independent signature. An original experimental setup based on a mechanical fault emulator is proposed to validate the simulation results. Simulations and experimental results show that the proposed method is effective to amplify the single fault signatures and calculate fault indicator at variable speed.

Wavelet transform and least square support vector machine for mechanical fault detection of an alternator using vibration signal

This article deals with fault detection of an alternator based on vibration signals using wavelet transform and least square support vector machine. Firstly, the noise in the vibration signal is removed using wavelet denoising. The denoised signals are then analysed using discrete wavelet transform with Daubechies mother wavelet. Several statistical features are then extracted from discrete wavelet transform coefficients of the signals. Finally, least square support vector machine is employed to detect and classify the different alternator conditions. The results show that the detection accuracy reached 90.48%. Hence, the proposed procedure is capable of detecting the alternator faults, and thus can be used for practical applications.

Determination of Transformer Winding Radial Deformation Using UWB System and Hyperboloid Method

Online monitoring of the transformer winding using electromagnetic waves has recently been proposed for the detection of mechanical defects of transformer winding. In this paper, a new method, which uses the ultrawideband radar measurement results and is based on locus of objects in the space, is proposed for the 3-D positioning of transformer winding radial deformation location. The proposed experimental setup for this method named hyperboloid method is modeled using Computer Simulation Technology software. In this paper, Vivaldi antennas suitable for measurements in environments with multipath, are used and the analysis is performed in the time domain. The simulation results show that the exact 3-D location of the radial deformation can be detected with a good accuracy.

Combination of process and vibration data for improved condition monitoring of industrial systems working under variable operating conditions

The detection and diagnosis of faults in industrial processes is a very active field of research due to the reduction in maintenance costs achieved by the implementation of process monitoring algorithms such as Principal Component Analysis, Partial Least Squares or more recently Canonical Variate Analysis (CVA). Typically the condition of rotating machinery is monitored separately using vibration analysis or other specific techniques. Conventional vibration-based condition monitoring techniques are based on the tracking of key features observed in the measured signal. Typically steady-state loading conditions are required to ensure consistency between measurements.In this paper, a technique based on merging process and vibration data is proposed with the objective of improving the detection of mechanical faults in industrial systems working under variable operating conditions. The capabilities of CVA for detection and diagnosis of faults were tested using experimental data acquired from a compressor test rig where different process faults were introduced. Results suggest that the combination of process and vibration data can effectively improve the detectability of mechanical faults in systems working under variable operating conditions.

Current-Based Mechanical Fault Detection for Direct-Drive Wind Turbines via Synchronous Sampling and Impulse Detection

Online fault detection is an effective means to improve wind turbine reliability and performance and reduce wind turbine downtime and operating and maintenance costs. Current-based wind turbine fault detection techniques have received more and more attention in academia and industry due to their nonintrusive character and economic advantages. This paper presents a novel computationally efficient high-resolution wideband synchronous sampling algorithm for the mechanical fault detection of variable-speed direct-drive wind turbines (i.e., no gearbox) only using nonstationary generator stator current measurements. The proposed algorithm synchronously resamples the current signals such that the varying characteristic frequencies of the excitations generated by wind turbine faults in the current signals become constant values. An impulse detection algorithm is then proposed to detect the faults by identifying the excitations from the frequency spectra of the synchronously sampled stator current signals. Experimental studies are carried out to demonstrate the effectiveness of the proposed algorithms for the detection of rotor eccentricity and bearing faults of a direct-drive wind turbine operating in variable-speed conditions.

Mechanical fault detection in permanent magnet synchronous motors using equal width discretization-based probability distribution and a neural network model

This paper focuses on detecting the static eccentricity and bearing faults of a permanent magnet synchronous motor (PMSM) using probability distributions based on equal width discretization (EWD) and a multilayer perceptron neural network (MLPNN) model. In order to achieve this, the PMSM stator current values were measured in the cases of healthy, static eccentricity, and bearing faults for the conditions of three speeds and five loads. The data was discretized into several ranges through the EWD method, the probability distributions were computed according to the number of current values belonging to each range, and these distributions were then used as inputs to the MLPNN model. We conducted eighteen experiments to evaluate the performance of the proposed model in the detection of faults. The proposed method was very successful in full load and high speed for some experiments. As a result, we showed that the proposed model resulted in a satisfactory classification of accuracy rates.

A Comparative Study on Multiwavelet Construction Methods and Customized Multiwavelet Library for Mechanical Fault Detection

Inner product transform principle reveals that the basis functions most relevant or similar to the fault features are pivotal to the meaningful fault detection. Customized multiwavelet methods and practices have continued to improve over the recent years, focused on two-scale similarity transform (TST), lifting transform (LT), and lifting scheme (LS). Due to the respective advantages and disadvantages, a comparative study on the multiwavelet construction methods by TST, symmetric and dissymmetric LT, and LS is discussed in the paper, covering the differences of construction theories, the synthetic analyses of construction strategies, and the comparison of waveform characteristics along with their applicable occasions. Comprehensively utilizing the capabilities of the construction methods, a novel customized multiwavelet library is established for the accurate fault detection. The proposed method is applied to incipient fault detection of rolling bearing for electric locomotive to verify the effectiveness and feasibility.

An observer-based approach for sensors (position/speed, currents and DC-link voltage) fault detection and diagnosis in PMSM drive

The paper describes a strategy for the mechanical, current and DC bus voltage sensor Fault Detection and Diagnosis (FDD) of a Permanent Magnet Synchronous Machine (PMSM) drive. The method is based on three observers. The Extended Kalman Filter (EKF) is designed to estimate the position in case of mechanical sensor fault. A Model Reference Adaptive System (MRAS) estimates the phase currents using the actual position and speed and a nonlinear observer computes the DC link voltage from the phase currents and the duty cycles. The computation and sort of the residuals (difference between measured and estimated values) allows the fault isolation. The method is evaluated on a test bench with mechanical, phase current and DC voltage sensor faults. The experimental results show the effectiveness and the capability of fault detection and diagnosis of the proposed strategy.

A Review Study on Mathematical Methods for Fault Detection Problems in Induction Motors

—Induction motors are frequently used in industrial processes. Failure of these machines may cause economic, quality and safety losses. In this paper, the mathematical methods used in detection of mechanical and electrical faults of these motors are reviewed together with theory and application examples on the current and vibration data which is acquired during performance tests of the motors followed by accelerated aging

Induction machine faults detection using stator current parametric spectral estimation

Current spectrum analysis is a proven technique for fault diagnosis in electrical machines. Current spectral estimation is usually performed using classical techniques such as periodogram (FFT) or its extensions. However, these techniques have several drawbacks since their frequency resolution is limited and additional post-processing algorithms are required to extract a relevant fault detection criterion. Therefore, this paper proposes a new parametric spectral estimator that fully exploits the faults sensitive frequencies. The proposed technique is based on the maximum likelihood estimator (MLE) and offers high-resolution capabilities. Based on this approach, a fault criterion is derived for detecting several fault types.The proposed technique is assessed using simulation signals, issued from a coupled electromagnetic circuits approach-based simulation tool for mechanical unbalance and electrical asymmetry faults detection. It is afterward validated using experiments on a 0.75-kW induction machine test bed for the particular case of bearing faults.

Signal Measurement Error Analysis of the Bayesian Network of Mechanical Fault Detection

Based on measurement error of observation nodes is commom in mechanical system fault detection, but the traditional denoising method has many shortcomings. This paper introduce the Gibbs sampling method, which can be used to denoise and eliminate measurement error for node discreted information. We discuss it, and expect some promotion in practical application.