Machine Health Monitoring Research Articles

Structural Health Monitoring (SHM) of Machine Using a Combination of Non-Destructive Testing for Heavy Manufacturing Industry

Markets are affected by assorted consumer requirements, which insist on superior quality, shorter delivery time, better customer support, and lower prices. Simultaneously, product life cycles are becoming shorter. Success relies on having either a cost-benefit or a value benefit, or, both in any competitive context. Therefore, non-destructive techniques (NDT) become vital but in the conventional system, the maintenance personnel has to visit the machine that consumes time and energy. In the present COVID-19 situation and to save energy and time, there is a necessity of making condition monitoring contactless as much as possible. Therefore, in this research work, a structural health monitoring analysis presented that covers: firstly, enlisting of the NDT infrastructure commonly available in heavy manufacturing industries; secondly, common causes and reasons of machine failures and finally, discusses need of embedded structural health monitoring (e-SHM) system with the combination of NDT in place of existing monitoring practice. The presented work suggested that a combination of NDT with e-SHM is better for timely fault detection to ensure effective condition monitoring.

Optimising Production through Intelligent Manufacturing

Intelligent manufacturing system (IMS) has been the focus of most industries since Industry 4.0 revolution. IMS is being implemented through the integration of Internet of Things, (IoT), Cyber-Physical Systems (CPS), digital twin and big data analytics to optimize production through smart manufacturing. This research presents a conceptual approach of an adaptive clustering algorithm (ACA) for advanced manufacturing decision-making for smart machining manufacturing. The work considers product monitoring and assessment, machine health and operating parameters monitoring, as an important factor for intelligent decision making on a machining production line through the developed cyber twin of the machine tool for production optimisation. Cyber twin of the machine tool is developed which runs on a realtime sequence with the physical asset fussed with smart sensors and controllers enabled with cloud computing, IoT and data analytics. The ACA enables resources monitoring, production monitoring, machine condition monitoring, cloud feedback notification, product monitoring, and assessment, for intelligent decision-making from a cluster of similar machines using ANN clustering tool for self-aware, self-predict and self-reconfiguration in a smart machining production line to detect a cutting tool chipping of less than 0.25mm size. The method is proposed to optimise production by increasing productivity through intelligent decision and prediction for tool change, tool failure, maintenance, adjustment of operating parameters.

Empirical Modelling of Frictional Noise and Two-Point Contact Using Ball-On-Disc Tribometer

Wear and its interrelated noise are inevitable occurrences when two materials are in contact with each other. The correlation of the two parameters leads to acoustic study for health monitoring of machines and components. Various empirical models have been proposed using ball-on-disc tribometer using a conventional one ball contact with the disc. In our investigation we used a two-point contact over a rotating disc. To imitate a machine environment where one component is in multiple contact at any instance, a ball holder is designed, which can be affixed on to a conventional tribometer and can engage two balls at the same time sliding over a disc specimen. For the noise parameter, non-contact technique is adapted instead of an acoustic analyzer where a microphone is placed at a fixed distance and in close vicinity to the noise source. The material of the balls was selected to be stainless steel while the material of the disc is mild steel. The balls slide over the disc for a fixed duration at a constant low speed rpm until the wear depth is significant for analysis. In-situ analysis of wear parameters including surface roughness, wear depth and wear profile of the disc was measured systematically. The results from the tribometer are in the form of penetration depth, while the sound signal obtained from microphone is filtered using microphone software and plotted as sound pressure (in dB). The results show a good correlation of wear to the frictional noise at both the contacts and helped us in our subsequent study of modelling wear dependencies. Comparison of wear parameters to noise parameters concludes that the frictional noise is directly related to wear generation at contact.

A Novel Fault Diagnosis Approach for Rolling Bearing Based on High-Order Synchrosqueezing Transform and Detrended Fluctuation Analysis

Time-frequency analysis always plays an important role in machine health monitoring owing to its advantage in extracting the fault information contained in non-stationary signal. In this paper, we present a novel technique to detect and diagnose the rolling bearing faults based on high-order synchrosqueezing transform (FSSTH) and detrended fluctuation analysis (DFA). With this method, the high-order synchrosqueezing transform is first utilized to decompose the vibration signal into an ensemble of oscillatory components termed as intrinsic mode functions (IMFs). Meanwhile, an empirical equation, which is based on the DFA, is introduced to adaptively determine the number of IMFs from FSSTH. Then, a time-frequency representation originated from the decomposed modes or corresponding envelopes is exhibited in order to identify the fault characteristic frequencies related to rolling bearing. Experiments are carried out using both simulated signal and real ones from Case Western Reserve University. Results show that the proposed method is more effective for the detection of fault characteristic frequencies compared with the traditional synchrosqueezing transform (SST) based fault diagnosis algorithm, which renders this technique is promising for machine fault diagnosis.

Deep Learning Algorithms for Bearing Fault Diagnostics—A Comprehensive Review

In this survey paper, we systematically summarize existing literature on bearing fault diagnostics with machine learning (ML) and data mining techniques. While conventional ML methods, including artificial neural network (ANN), principal component analysis (PCA), support vector machines (SVM), etc., have been successfully applied to the detection and categorization of bearing faults for decades, recent developments in deep learning (DL) algorithms in the last five years have sparked renewed interest in both industry and academia for intelligent machine health monitoring. In this paper, we first provide a brief review of conventional ML methods, before taking a deep dive into the state-of-the-art DL algorithms for bearing fault applications. Specifically, the superiority of DL based methods over conventional ML methods are analyzed in terms of fault feature extraction and classification performances; many new functionalities enabled by DL techniques are also summarized. In addition, to obtain a more intuitive insight, a comparative study is conducted on the classification accuracy of different algorithms utilizing the open-source Case Western Reserve University (CWRU) bearing dataset. Finally, to facilitate the transition on applying various DL algorithms to bearing fault diagnostics, detailed recommendations and suggestions are provided for specific application conditions such as the setup environment, the data size, and the number of sensors and sensor types. Future research directions to further enhance the performance of DL algorithms on health monitoring are also discussed.

Fault Diagnosis Method Based on Encoding Time Series and Convolutional Neural Network

In view of the shortcomings of traditional fault diagnosis methods based on time domain vibration analysis, which require complicated calculations of feature vectors, and are over-dependent on a prior diagnosis knowledge, effort for the design of the feature extraction algorithms, and have limited ability to extract the complex relationships in fault signals, in this paper we propose a convolutional neural network (CNN) framework for machine health monitoring based on the encoding of one-dimension (1-D) time series to two-dimension (2-D) images. This paper defines a new Gram matrix and through the Python programming environment, we emulate the new Gram matrix in 2-D images, thus feature extraction and recognition can be performed by CNNs. The proposed method which is tested on two datasets, including multi-stage centrifugal fan dataset for our lab, motor bearing dataset for Case Western Reserve University, has achieved prediction average accuracy of 94.07% and 96.28%, respectively. The results have been compared with other deep learning and traditional methods, including Recurrent neural network (RNN), Support Vector Machines (SVM), Multi-Genetic algorithm, shallow CNN and BP neural network. The results show that the method can improve fault diagnosis accuracy in an effective way and stability than other advanced techniques.

Development of a Digital Architecture for Distributed CNC Machine Health Monitoring

Abstract With the advancement of the Internet of Things, machines are expected to become more connected by transmitting operational data in ways that can readily be consumed by analytics engines. In recent years, a substantial amount of research has been conducted to create techniques of acquiring and utilizing machine data to inform supply chain decisions and monitor machine health. In modern manufacturing equipment, standard interfaces such as MTConnect and Open Platform Communications–Unified Architecture facilitate the data acquisition process by providing a means of easily capturing machine controller data. With data from these interfaces, monitoring quantities such as machine utilization are possible by monitoring the MTConnect interface and performing analysis on the machine state. In a distributed manufacturing environment, a means of remotely accessing this data must be developed. The infrastructure responsible for this data transmission must be secure, scalable, and standardized. If detailed machine health metrics are to be analyzed, the infrastructure should be able to accommodate data from sources beyond the machine controller itself. Supplementing the controller data with that from additional sensors, such as accelerometers, temperature, humidity, and current sensors, a more complete picture of machine health status can be created. This article will present a digital architecture that meets these criteria and demonstrate the use cases in machine utilization and health monitoring.

Support vector machines based non-contact fault diagnosis system for bearings

Bearing defects have been accepted as one of the major causes of failure in rotating machinery. It is important to identify and diagnose the failure behavior of bearings for the reliable operation of equipment. In this paper, a low-cost non-contact vibration sensor has been developed for detecting the faults in bearings. The supervised learning method, support vector machine (SVM), has been employed as a tool to validate the effectiveness of the developed sensor. Experimental vibration data collected for different bearing defects under various loading and running conditions have been analyzed to develop a system for diagnosing the faults for machine health monitoring. Fault diagnosis has been accomplished using discrete wavelet transform for denoising the signal. Mahalanobis distance criteria has been employed for selecting the strongest feature on the extracted relevant features. Finally, these selected features have been passed to the SVM classifier for identifying and classifying the various bearing defects. The results reveal that the vibration signatures obtained from developed non-contact sensor compare well with the accelerometer data obtained under the same conditions. A developed sensor is a promising tool for detecting the bearing damage and identifying its class. SVM results have established the effectiveness of the developed non-contact sensor as a vibration measuring instrument which makes the developed sensor a cost-effective tool for the condition monitoring of rotating machines.

A Silicon MEMS EM vibration energy harvester

This paper presents an optimized silicon-MEMS electromagnetic vibration energy harvester suitable for applications such as machine health monitoring. The harvester comprises a DRIE-etched silicon suspension, and pick-and-place N42 NdBFe magnets and copper coils, housed in a 3D-printed package. The harvester is designed to operate near 50 Hz with 0.5-1 g vibrations using a long-stroke suspension. Multi-domain harvester optimization results in an open-circuit voltage of 1.7 V, a matched-load power output of 2.2 mW, and a matched-load power-output density of 1.23 mW/cm3 at 1.1 g with a resonance frequency of 76.3 Hz.

Customized Vibration Generator for State of Health Monitoring of Prosthetic Implants and Pseudo-Bionic Machine–Human Feedbacks

Modern industrial, household and other equipment include sophisticated power mechanisms and complicated control solutions that require tighter human–machine–human interactions to form the structures known as cyber–physical–human systems. Their significant parts are human–machine command links and machine–human feedbacks. Such systems are found in medicine, e.g., in orthopedics, where they are important for the operation and functional abilities of orthopedic devices—wheelchair, prosthesis, rehabilitation units, etc. The mentioned feedbacks may be implemented based on the haptic perceptions that requires vibration actuators. In orthopedics, such actuators can be used also for diagnostic purposes. This research brings forward the idea of the use of 3D printing in conjunction with high quality permanent magnets. This allows for the achievement of better efficiency, smaller size, and the developing of actuators individually for particular circumstances. The obtained simulation, experimental data, and data about 3D manufacturing generally confirm the above hypothesis. In particular, the stiffness coefficient of the actuator’s membrane and attached mass, which can be changed easily during 3D printing, affects the frequency of maximal power output. Secondly, the 3D manufacturing process is quick, tunable and rather cheap. Finally, an elaboration of the design of the actuator that allows for the real-time modification of stiffness and mass in a program way is planned for future works.

Instantaneous Rotation Speed Measurement System Based on Variable Reluctance Sensors for Torsional Vibration Monitoring

In this paper, a system for the measurement of the instantaneous speed of rotating parts in turbines is proposed and modeled. The system is based on traditional and widely used variable reluctance sensors often mounted on-board for other purposes. By coupling this sensors with ad hoc rotating target (novel design) and front-end circuit, it is possible to extend the application of this sensor to accurate instantaneous angular speed monitoring, and to exploit it for torsional vibration monitoring or, in general, for the measurement of any other meaningful quantity related to the variation of the angular speed, useful for machine health monitoring and incipient fault diagnosis. The proposed rotating target produces a signal close to the pure sine wave, with the specific purpose of improving accuracy in torsional vibration monitoring. The complete system was modeled, also taking into account the effect of nonidealities in order to evaluate its performance. Experimental results confirm the satisfactory results obtained from simulations, and show the reduction of the magnitude of the higher order harmonics in the output signal spectrum ensuring improved performance for any torsional vibration monitoring assembly employing the proposed system.

A novel diagnostic and prognostic framework for incipient fault detection and remaining service life prediction with application to industrial rotating machines

Data-driven machine health monitoring systems (MHMS) have been widely investigated and applied in the field of machine diagnostics and prognostics with the aim of realizing predictive maintenance. It involves using data to identify early warnings that indicate potential system malfunctioning, predict when system failure might occur, and pre-emptively service equipment to avoid unscheduled downtime. One of the most critical aspects of data-driven MHMS is the provision of incipient fault diagnosis and prognosis regarding the system’s future working conditions. In this work, a novel diagnostic and prognostic framework is proposed to detect incipient faults and estimate remaining service life (RSL) of rotating machinery. In the proposed framework, a novel canonical variate analysis (CVA)-based monitoring index, which takes into account the distinctions between past and future canonical variables, is employed for carrying out incipient fault diagnosis. By incorporating the exponentially weighted moving average (EWMA) technique, a novel fault identification approach based on Pearson correlation analysis is presented and utilized to identify the influential variables that are most likely associated with the fault. Moreover, an enhanced metabolism grey forecasting model (MGFM) approach is developed for RSL prediction. Particle filter (PF) is employed to modify the traditional grey forecasting model for improving its prediction performance. The enhanced MGFM approach is designed to address two generic issues namely dealing with scarce data and quantifying the uncertainty of RSL in a probabilistic form, which are often encountered in the prognostics of safety-critical and complex assets. The proposed CVA-based index is validated on slowly evolving faults in a continuous stirred tank reactor (CSTR) system, and the effectiveness of the proposed integrated diagnostic and prognostic method for the monitoring of rotating machinery is demonstrated for slow involving faults in two case studies of an operational industrial centrifugal pump and one case study of an operational centrifugal compressor.

Stationary Wavelet-Fourier Entropy and Kernel Extreme Learning for Bearing Multi-Fault Diagnosis.

Bearing fault diagnosis methods play an important role in rotating machine health monitoring. In recent years, various intelligent fault diagnosis methods have been proposed, which are mainly based on the features extraction method combined with either shallow or deep learning methods. During the last few years, Shannon entropy features have been widely used in machine health monitoring, improving the accuracy of the bearing fault diagnosis process. Therefore, in this paper, we consider the combination of multi-scale stationary wavelet packet analysis with the Fourier amplitude spectrum to obtain a new discriminative Shannon entropy feature that we call stationary wavelet packet Fourier entropy (SWPFE). Features extracted by our SWPFE method are then passed onto a shallow kernel extreme learning machine (KELM) classifier to diagnose bearing failure types with different severities. The proposed method was applied on two experimental vibration signal databases of a rolling element bearing and compared to two recently proposed methods called stationary wavelet packet permutation entropy (SWPPE) and stationary wavelet packet dispersion entropy (SWPPE). Based on our results, we can say that the proposed method is able to achieve better accuracy levels than both the SWPPE and SWPDE methods using fewer failure features. Further, as our method does not require any hyperparameter calibration step, it is less dependent on user experience/expertise.

Fault Detection and Isolation in Electrical Machines using Deep Neural Networks

Condition and health monitoring of electrical machines during dynamic loading is a common, yet challenging problem in main battle tanks. Existing methods address this issue by extracting various features which are subsequently used in a classifier to isolate faults. However, this approach relies on the feature set being extracted and therefore most of the time does not provide expected accuracy in identification of faults. In this work, we have used convolution neural network that utilises the original time domain measurements for fault detection and isolation (FDI). Results from experimental studies indicate that the proposed approach can perform FDI with more than 95\% accuracy using commonly available current measurements.

Restoration of defocused ferrograph images using a large kernel convolutional neural network

Wear debris analysis is used for machine condition monitoring because it is capable of identifying wear related faults in machinery components. In particular, wear mechanisms can be explored based on debris shape and texture features. As current wear monitoring practices prefer rapid data interpretation, on-line ferrograph is favored where debris images are collected and processed on-site with an imaging apparatus. However, such images are often blurred due to lubricant contamination and limited optical field depth. This drawback makes the subsequent investigation of wear debris features a challenging task. To address this issue, a restoration method is developed to reduce the influence of out-of-focus blurring in on-line ferrograph images. Firstly, in order to model the out-of-focus blurring process, a degradation model is constructed with a convolutional neural network. Different from general deep learning structures, larger convolutional kernels are embedded in the network to more comprehensively characterize the blurring process. Next, training samples of different types of debris are collected in the laboratory. The neural network structure then is optimized to minimize the difference between restored images and their corresponding ground truths. The final restoration is conducted based on the trained network via a pixel-wise regression process. To evaluate the performance of the developed method, experiments are carried out with debris images collected from a gear-box test rig. Results have demonstrated that: (1) the proposed restoration strategy allows robust extractions of debris features in greater detail; (2) the developed method is more computationally efficient than traditional deblurring techniques, which enables it to be applicable in on-line machine health monitoring.

Machine Health Monitoring Using Adaptive Kernel Spectral Clustering and Deep Long Short-Term Memory Recurrent Neural Networks

Machine health monitoring is of great importance in industrial informatics field. Recently, deep learning methods applied to machine health monitoring have been proven effective. However, the existing methods face enormous difficulties in extracting heterogeneous features indicating the variation until failure and revealing the inherent high-dimensional features of massive signals, which affect the accuracy and efficiency of machine health monitoring. In this paper, a novel data-driven machine health monitoring method is proposed using adaptive kernel spectral clustering (AKSC) and deep long short-term memory recurrent neural networks (LSTM-RNN). This method include three steps: First, features in the time domain, frequency domain, and time-frequency domain are, respectively, extracted from massive measured signals. And, an Euclidean distance based algorithm is designed to select degradation features. Second, the AKSC algorithm is introduced to adaptively identify machine anomaly behaviors from multiple degradation features. Third, a new deep learning model (LSTM-RNN) is constructed to update and predict the failure time of the machine. The effectiveness of the proposed method is validated using a set of test-to-failure experimental data. The results show that the performance of the proposed method is competitive with other existing methods.

Combining translation-invariant wavelet frames and convolutional neural network for intelligent tool wear state identification

On-machine monitoring of tool wear in machining processes has found its importance to reduce equipment downtime and reduce tooling costs. As the tool wears out gradually, the contact state of the cutting edge and the workpiece changes, which has a significant influence on the vibration state of the spindle. The performance of traditional intelligent fault diagnosis methods depend on feature extraction of dynamic signals, which requires expert knowledge and human labor. Recently, deep learning algorithms have been applied widely in machine health monitoring. In this paper, we present a novel intelligent technique for tool wear state recognition using machine spindle vibration signals. The proposed technique combines derived wavelet frames (DWFs) and convolutional neural network (CNN). Constructed based on dual tree wavelets, DWF are equipped with merits of centralized multiresolution and nearly translation-invariance. In this method, DWFs are employed to decompose the original signal into frequency bands of different bandwidths and different center frequencies, which are more pronounced as the tool wears. Further, the reconstructed sub-signals are stacked into a 2-D signal matrix to match the structure of 2-D CNN while retaining more dynamic information. The 2-D convolutional neural network is utilized to automatically recognize features from the multiscale 2-D signal matrix. End-milling experiments were performed on a S45C steel workpiece at different machining parameters. The experiment results of the recognition for tool wear state show the feasibility and effectiveness of the proposed method.

Transient Signal Analysis Using Parallel Time-Frequency Manifold Filtering for Bearing Health Diagnosis

It is crucial to identify and extract the weak transient features embedded in the vibration signals for bearing health monitoring and fault diagnosis. However, due to the macro-structural disturbance and background noise interference, it is not easy to mine the transient features, especially at the apparent failure stage. Meanwhile, the actual mechanism of bearing fault detect can be not simply expressed by the formulated theory models without consideration of the actual physical collision process. To overcome these issues, motivated by the merits of time-frequency manifold (TFM), this paper proposes a new transient feature extraction method, called parallel time-frequency manifold (PTFM) filtering, by simultaneously using TFM-based reconstruction with TFM-based filter in parallel for transient feature extraction. First, to improve the computational efficiency of TFM, two-dimensional discrete wavelet transform is employed on the raw time-frequency distribution (TFD) with image compression. TFM learning is later used to mine the principle manifolds from those approximation sub-images. Then, the amplitudes of the raw time-frequency image can be reconstructed by TFM feature bases while the desired location of time-frequency feature can be captured by TFM morphology filter in a process of image morphology. With raw time-frequency phases in a series of inverse processes, the de-noised signal can be finally synthesized from these filtered images. The proposed method accomplishes a natural manifold feature denoising by combining the sparse theory with image morphology, and demonstrates attractive prospects in the following three aspects: signal de-noising with a self-learning mode in the view of image morphology processing combined with sparse theory, fault diagnosis with in-band noise/close interference removal, and machine health monitoring with capability in capturing sensitive failure information. Simulations and experiments confirmed the effectiveness of the proposed PTFM filtering method in noise suppression and feature enhancement, which is valuable for bearing health monitoring and diagnosis applications.

Security Health monitoring and Attestation of Virtual Machines in Cloud computing

Security Health monitoring and Attestation of Virtual Machines in Cloud computing

Development and Testing of a Combined Machine and Process Health Monitoring System

Process monitoring has been shown to be capable of observing the quality of a machining operation through sensor signals and analysis in both the literature and in commercially available systems. Some of these systems provide an additional benefit of monitoring the health of a machine tool. However, the commercially available systems tend to utilise relatively simple analysis techniques for both the process and machine health, limiting their application and robustness. Industrial interest in systems that can profit from the current advances in machine tool digitalisation and data analytics has grown considerably. This is especially true for the capability of early-detection of quality issues in components, whilst also ensuring machine tools are in a condition that can achieve high quality production. The present research includes the development and testing of a fingerprint routine which can be run at regular intervals to detect potential failure modes or machine tool degradation through signal analysis. Machining trials were carried out with the objective of detecting known defects in a workpiece through signal analysis. For both cases, a combined monitoring system was developed for data capture during testing, and a number of failure modes and defects were physically simulated to test the possibility of detection in the acquired signals. Time domain, frequency domain, and time-frequency domain signal processing techniques were applied to the sensor data with various levels of success. Continuous wavelet transforms (CWT) were of particular interest, as they successfully captured signal changes between tests for the physically simulated failure modes of the machine tool and the component. Therefore, a comparative CWT analysis was developed which successfully emphasised some of the machine tool failure modes and part defects when compared to baseline signals. The output of the comparative analysis may be well-suited to automation through machine learning techniques.