Showing papers on "Condition monitoring published in 2018"

Intelligent Bearing Fault Diagnosis Method Combining Compressed Data Acquisition and Deep Learning

Abstract: Effective intelligent fault diagnosis has long been a research focus on the condition monitoring of rotary machinery systems. Traditionally, time-domain vibration-based fault diagnosis has some deficiencies, such as complex computation of feature vectors, excessive dependence on prior knowledge and diagnostic expertise, and limited capacity for learning complex relationships in fault signals. Furthermore, following the increase in condition data, how to promptly process the massive fault data and automatically provide accurate diagnosis has become an urgent need to solve. Inspired by the idea of compressed sensing and deep learning, a novel intelligent diagnosis method is proposed for fault identification of rotating machines. In this paper, a nonlinear projection is applied to achieve the compressed acquisition, which not only reduces the amount of measured data that contained all the information of faults but also realizes the automatic feature extraction in transform domain. For exploring the discrimination hidden in the acquired data, a stacked sparse autoencoders-based deep neural network is established and performed with an unsupervised learning procedure followed by a supervised fine-tuning process. We studied the significance of compressed acquisition and provided the effects of key factors and comparison with traditional methods. The effectiveness of the proposed method is validated using data sets from rolling element bearings and the analysis shows that it is able to obtain high diagnotic accuracies and is superior to the existing methods. The proposed method reduces the need of human labor and expertise and provides new strategy to handle the massive data more easily.

Structural health monitoring: Closing the gap between research and industrial deployment:

Abstract: There has been a large volume of research on structural health monitoring since the 1970s but this research effort has yielded relatively few routine industrial applications. Structural health monitoring can include applications on very different structures with very different requirements; this article splits the subject into four broad categories: rotating machine condition monitoring, global monitoring of large structures (structural identification), large area monitoring where the area covered is part of a larger structure, and local monitoring. The capabilities and potential applications of techniques in each category are discussed. Condition monitoring of rotating machine components is very different to the other categories since it is not strictly concerned with structural health. However, it is often linked with structural health monitoring and is a relatively mature field with many routine applications, so useful lessons can be read across to mainstream structural health monitoring where there ar...

Nonconvex Sparse Regularization and Convex Optimization for Bearing Fault Diagnosis

Abstract: Vibration monitoring is one of the most effective ways for bearing fault diagnosis, and a challenge is how to accurately estimate bearing fault signals from noisy vibration signals. In this paper, a nonconvex sparse regularization method for bearing fault diagnosis is proposed based on the generalized minimax-concave (GMC) penalty, which maintains the convexity of the sparsity-regularized least squares cost function, and thus the global minimum can be solved by convex optimization algorithms. Furthermore, we introduce a k-sparsity strategy for the adaptive selection of the regularization parameter. The main advantage over conventional filtering methods is that GMC can better preserve the bearing fault signal while reducing the interference of noise and other components; thus, it can significantly improve the estimation accuracy of the bearing fault signal. A simulation study and two run-to-failure experiments verify the effectiveness of GMC in the diagnosis of localized faults in rolling bearings, and the comparison studies show that GMC provides more accurate estimation results than L1-norm regularization and spectral kurtosis.

Prognosis of Bearing Acoustic Emission Signals Using Supervised Machine Learning

Abstract: Acoustic emission (AE) technique can be successfully utilized for condition monitoring of various machining and industrial processes. To keep machines function at optimal levels, fault prognosis model to predict the remaining useful life (RUL) of machine components is required. This model is used to analyze the output signals of a machine whilst in operation and accordingly helps to set an early alarm tool that reduces the untimely replacement of components and the wasteful machine downtime. Recent improvements indicate the drive on the way towards incorporation of prognosis and diagnosis machine learning techniques in future machine health management systems. With this in mind, this work employs three supervised machine learning techniques; support vector machine regression, multilayer artificial neural network model and gaussian process regression, to correlate AE features with corresponding natural wear of slow speed bearings throughout series of laboratory experiments. Analysis of signal parameters such as signal intensity estimator and root mean square was undertaken to discriminate individual types of early damage. It was concluded that neural networks model with back propagation learning algorithm has an advantage over the other models in estimating the RUL for slow speed bearings if the proper network structure is chosen and sufficient data is provided.

Deep Learning for Infrared Thermal Image Based Machine Health Monitoring

Abstract: The condition of a machine can automatically be identified by creating and classifying features that summarize characteristics of measured signals. Currently, experts, in their respective fields, devise these features based on their knowledge. Hence, the performance and usefulness depends on the expert's knowledge of the underlying physics or statistics. Furthermore, if new and additional conditions should be detectable, experts have to implement new feature extraction methods. To mitigate the drawbacks of feature engineering, a method from the subfield of feature learning, i.e., deep learning (DL), more specifically convolutional neural networks (NNs), is researched in this paper. The objective of this paper is to investigate if and how DL can be applied to infrared thermal (IRT) video to automatically determine the condition of the machine. By applying this method on IRT data in two use cases, i.e., machine-fault detection and oil-level prediction, we show that the proposed system is able to detect many conditions in rotating machinery very accurately (i.e., 95 and 91.67% accuracy for the respective use cases), without requiring any detailed knowledge about the underlying physics, and thus having the potential to significantly simplify condition monitoring using complex sensor data. Furthermore, we show that by using the trained NNs, important regions in the IRT images can be identified related to specific conditions, which can potentially lead to new physical insights.

Condition Monitoring in Advanced Battery Management Systems: Moving Horizon Estimation Using a Reduced Electrochemical Model

Abstract: Efficient battery condition monitoring is of particular importance in large-scale, high-performance, and safety-critical mechatronic systems, e.g., electrified vehicles and smart grid. This paper pursues a detailed assessment of optimization-driven moving horizon estimation (MHE) framework by means of a reduced electrochemical model. For state-of-charge estimation, the standard MHE and two variants in the framework are examined by a comprehensive consideration of accuracy, computational intensity, effect of horizon size, and fault tolerance. A comparison with common extended Kalman filtering and unscented Kalman filtering is also carried out. Then, the feasibility and performance are demonstrated for accessing internal battery states unavailable in equivalent circuit models, such as solid-phase surface concentration and electrolyte concentration. Ultimately, a multiscale MHE-type scheme is created for State-of-Health estimation. This study is the first known systematic investigation of MHE-type estimators applied to battery management.

Energy Harvesting Technologies for Achieving Self-Powered Wireless Sensor Networks in Machine Condition Monitoring: A Review.

Abstract: Condition monitoring can reduce machine breakdown losses, increase productivity and operation safety, and therefore deliver significant benefits to many industries. The emergence of wireless sensor networks (WSNs) with smart processing ability play an ever-growing role in online condition monitoring of machines. WSNs are cost-effective networking systems for machine condition monitoring. It avoids cable usage and eases system deployment in industry, which leads to significant savings. Powering the nodes is one of the major challenges for a true WSN system, especially when positioned at inaccessible or dangerous locations and in harsh environments. Promising energy harvesting technologies have attracted the attention of engineers because they convert microwatt or milliwatt level power from the environment to implement maintenance-free machine condition monitoring systems with WSNs. The motivation of this review is to investigate the energy sources, stimulate the application of energy harvesting based WSNs, and evaluate the improvement of energy harvesting systems for mechanical condition monitoring. This paper overviews the principles of a number of energy harvesting technologies applicable to industrial machines by investigating the power consumption of WSNs and the potential energy sources in mechanical systems. Many models or prototypes with different features are reviewed, especially in the mechanical field. Energy harvesting technologies are evaluated for further development according to the comparison of their advantages and disadvantages. Finally, a discussion of the challenges and potential future research of energy harvesting systems powering WSNs for machine condition monitoring is made.

Development and trend of condition monitoring and fault diagnosis of multi-sensors information fusion for rolling bearings: a review

Abstract: A rolling bearing is an essential component of a rotating mechanical transmission system Its performance and quality directly affects the life and reliability of machinery Bearings’ performance and reliability need high requirements because of a more complex and poor working conditions of bearings A bearing with high reliability reduces equipment operation accidents and equipment maintenance costs and achieves condition-based maintenance First in this paper, the development of technology of the main individual physical condition monitoring and fault diagnosis of rolling bearings are introduced, then the fault diagnosis technology of multi-sensors information fusion is introduced, and finally, the advantages, disadvantages, and trends developed in the future of the detection main individual physics technology and multi-sensors information fusion technology are summarized This paper is expected to provide the necessary basis for the follow-up study of the fault diagnosis of rolling bearings and a foundational knowledge for researchers about rolling bearings

Machine Learning approach for Predictive Maintenance in Industry 4.0

Abstract: Condition monitoring together with predictive maintenance of electric motors and other equipment used by the industry avoids severe economic losses resulting from unexpected motor failures and greatly improves the system reliability. This paper describes a Machine Learning architecture for Predictive Maintenance, based on Random Forest approach. The system was tested on a real industry example, by developing the data collection and data system analysis, applying the Machine Learning approach and comparing it to the simulation tool analysis. Data has been collected by various sensors, machine PLCs and communication protocols and made available to Data Analysis Tool on the Azure Cloud architecture. Preliminary results show a proper behavior of the approach on predicting different machine states with high accuracy.

Fault Diagnosis Techniques for Permanent Magnet AC Machine and Drives—A Review of Current State of the Art

Abstract: This paper reviews the current state of the art of condition monitoring and fault diagnosis techniques for permanent magnet (PM) machines. It also takes into account the past research on this subject as appropriate. The discussion in this paper includes issues related to common abnormalities and specific faults in PM machines and drives, such as magnet damage and demagnetization, rotor eccentricity, unbalanced magnetic pull, open- and short-circuit windings, and switch faults. In this paper, a detailed review is present on the sources of these faults, their analytical model and fault detection tools. Finally, an integrated methodology for condition monitoring of PM machines is discussed.

Wind Turbine Blade Breakage Monitoring With Deep Autoencoders

Abstract: Monitoring wind turbine blade breakages based on supervisory control and data acquisition (SCADA) data is investigated in this research. A preliminary data analysis is performed to demonstrate that existing SCADA features are unable to present irregular patterns prior to occurrences of blade breakages. A deep autoencoder (DA) model is introduced to derive an indicator of impending blade breakages, the reconstruction error (RE), from SCADA data. The DA model is a neural network of multiple hidden layers organized symmetrically. In training DA models, the restricted Boltzmann machine is applied to initialize weights and biases. The back-propagation method is subsequently employed to further optimize the network structure. Through examining SCADA data, we observe that the trend of RE will shift by the blade breakage. To effectively detect RE shifts through online monitoring, the exponentially weighted moving average control chart is deployed. The effectiveness of the proposed monitoring approach is validated by blade breakage cases collected from wind farms located in China. The computational results prove the capability of the proposed monitoring approach in identifying impending blade breakages.

Wind Turbine Fault Detection and Identification Through PCA-Based Optimal Variable Selection

Abstract: An effective condition monitoring system of wind turbines generally requires installation of a high number of sensors and use of a high sampling frequency in particular for monitoring of the electrical components within a turbine, resulting in a large amount of data This can become a burden for condition monitoring and fault detection systems This paper aims to develop algorithms that will allow a reduced dataset to be used in wind turbine fault detection This paper first proposes a variable selection algorithm based on principal component analysis with multiple selection criteria in order to select a set of variables to target fault signals while still preserving the variation of data in the original dataset With the selected variables, this paper then describes fault detection and identification algorithms, which can identify faults, determine the corresponding time and location where the fault occurs, and estimate its severity The proposed algorithms are evaluated with simulation data from PSCAD/EMTDC, Supervisory control and data acquisition data from an operational wind farm, and experimental data from a wind turbine test rig Results show that the proposed methods can select a reduced set of variables with minimal information lost whilst detecting faults efficiently and effectively

Bearing faults diagnosis using fuzzy expert system relying on an Improved Range Overlaps and Similarity method

Abstract: Bearing fault diagnosis represents the core of induction machines condition monitoring. This paper presents an application of fuzzy expert system (FES) to bearing faults diagnosis. Here, fuzzy rules are automatically induced from numerical data using the Similarity partition method. Data of faulty bearings presents high noise level. Thus, an Improved Range Overlaps method (IRO) is proposed to select input feature vectors by giving them validity degrees. The Similarity method partition was found confused with features presenting range overlap. Consequently, the new proposed Improved Range Overlaps method is found quite suitable for improving the classifier accuracy. The model validity and efficiency were proved using experimental bearing faults data from Case Western Reserve University database and the NSF I/UCR Center on Intelligent Maintenance Systems (IMS) database.

Methods for Advanced Wind Turbine Condition Monitoring and Early Diagnosis: A Literature Review

Abstract: Condition monitoring and early fault diagnosis for wind turbines have become essential industry practice as they help improve wind farm reliability, overall performance and productivity. If not detected and rectified at early stages, some faults can be catastrophic with significant loss or revenue along with interruption to the business relying mainly on wind energy. The failure of Wind turbine results in system downtime and repairing or replacement expenses that significantly reduce the annual income. Such failures call for more systematized operation and maintenance schemes to ensure the reliability of wind energy conversion systems. Condition monitoring and fault diagnosis systems of wind turbine play an important role in reducing maintenance and operational costs and increase system reliability. This paper is aimed at providing the reader with the overall feature for wind turbine condition monitoring and fault diagnosis which includes various potential fault types and locations along with the signals to be analyzed with different signal processing methods.

SCADA-based wind turbine anomaly detection using Gaussian process models for wind turbine condition monitoring purposes

Abstract: The penetration of wind energy into power systems is steadily increasing; this highlights the importance of operations and maintenance, and specifically the role of condition monitoring. Wind turbine power curves based on supervisory control and data acquisition data provide a cost-effective approach to wind turbine health monitoring. This study proposes a Gaussian process (a non-parametric machine learning approach) based algorithm for condition monitoring. The standard IEC binned power curve together with individual bin probability distributions can be used to identify operational anomalies. The IEC approach can also be modified to create a form of real-time power curve. Both of these approaches will be compared with a Gaussian process model to assess both speed and accuracy of anomaly detection. Significant yaw misalignment, reflecting a yaw control error or fault, results in a loss of power. Such a fault is quite common and early detection is important to prevent loss of power generation. Yaw control error provides a useful case study to demonstrate the effectiveness of the proposed algorithms and allows the advantages and limitations of the proposed methods to be determined.

A review of condition monitoring techniques and diagnostic tests for lifetime estimation of power transformers

Abstract: Power transformers are a key component of electrical networks, and they are both expensive and difficult to upgrade in a live network. Many utilities monitor the condition of the components that make up a power transformer and use this information to minimize the outage and extend the service life. Routine and diagnostic tests are currently used for condition monitoring and appraising the ageing and defects of the core, windings, bushings and tap changers of power transformers. To accurately assess the remaining life and failure probability, methods have been developed to correlate results from different routine and diagnostic tests. This paper reviews established tests such as dissolved gas analysis, oil characteristic tests, dielectric response, frequency response analysis, partial discharge, infrared thermograph test, turns ratio, power factor, transformer contact resistance, and insulation resistance measurements. It also considers the methods widely used for health index, lifetime estimation, and probability of failure. The authors also highlight the strengths and limitations of currently available methods. This paper summarizes a wide range of techniques drawn from industry and academic sources and contrasts them in a unified frame work.

Performance-Based Gas Turbine Health Monitoring, Diagnostics, and Prognostics: A Survey

Abstract: Health monitoring is an essential part of condition-based maintenance and prognostics and health management for gas turbines. Various health monitoring systems have been developed based on the measurement and observation of the fault symptoms including turbine performance parameters such as heat rate, and nonperformance symptoms such as structural vibration. This paper focuses on surveying state-of-the-art condition monitoring, diagnostic and prognostic techniques using performance parameters acquired from gas-path data that are mostly available from the operating systems of gas turbines. Performance parameters and the corresponding effective factors are presented in the beginning. Structure of performance monitoring and diagnostic systems are systematically laid out next, and the recent developments in each section are surveyed and discussed. Observing the importance of the prognostics in the recent trend of health monitoring research, an emphasis is given on the prognostic frameworks and their implementation for the remaining useful life prediction. A conclusion along with a brief discussion on the current state and potential future directions is provided at the end.

Rotor-Current-Based Fault Diagnosis for DFIG Wind Turbine Drivetrain Gearboxes Using Frequency Analysis and a Deep Classifier

Abstract: Fault diagnosis of drivetrain gearboxes is a prominent challenge in wind turbine condition monitoring. Many machine learning algorithms have been applied to gearbox fault diagnosis. However, many of the current machine learning algorithms did not provide satisfactory fault diagnosis results due to their shallow architectures. Recently, a class of machine learning models with deep architectures called deep learning has received more attention, because it can learn high-level features of inputs. This paper proposes a new fault diagnosis method for the drivetrain gearboxes of the wind turbines equipped with doubly-fed induction generators (DFIGs) using DFIG rotor current signal analysis. In the proposed method, the instantaneous fundamental frequency of the rotor current signal is first estimated to obtain the instantaneous shaft rotating frequency. Then, the Hilbert transform is used to demodulate the rotor current signal to obtain its envelope, and the resultant envelope signal contains fault characteristic frequencies that are in proportion to the varying DFIG shaft rotating frequency. Next, an angular resampling algorithm is designed to resample the nonstationary envelope signal to be stationary based on the estimated instantaneous shaft rotating frequency. After that, the power spectral density analysis is performed on the resampled envelope signal for the gearbox fault detection. Finally, a classifier with a deep architecture that consists of a stacked autoencoder and a support vector machine is proposed for gearbox fault classification using extracted fault features. Experimental results obtained from a DFIG wind turbine drivetrain test rig are provided to verify the effectiveness of the proposed method.