The presence of incorrect data leads to the decrease of condition-monitoring big data quality. As a result, unreliable or misleading results are probably obtained by analyzing these poor-quality ...data. In this paper, to improve the data quality, an incorrect data detection method based on an improved local outlier factor (LOF) is proposed for data cleaning. First, a sliding window technique is used to divide data into different segments. These segments are considered as different objects and their attributes consist of time-domain statistical features extracted from each segment, such as mean, maximum and peak-to-peak value. Second, a kernel-based LOF (KLOF) is calculated using these attributes to evaluate the degree of each segment being incorrect data. Third, according to these KLOF values and a threshold value, incorrect data are detected. Finally, a simulation of vibration data generated by a defective rolling element bearing and three real cases concerning a fixed-axle gearbox, a wind turbine, and a planetary gearbox are used to verify the effectiveness of the proposed method, respectively. The results demonstrate that the proposed method is able to detect both missing segments and abnormal segments, which are two typical incorrect data, effectively, and thus is helpful for big data cleaning of machinery condition monitoring.
The key function of rotating machine condition monitoring (CM) is to detect structural changes during machine operations. This paper presents a novel statistical time-frequency analysis method for ...this purpose. In particular, frequency spectrum is extracted from the machine condition signals based on periodogram estimation. Undirected weighted graph is then constructed from the resulting periodograms, where the so-called median graph is introduced and adopted to describe the normal machine status. Statistical analysis is performed to investigate newly observed data with respect to the median graph for change decision making. The proposed method has been applied to three different engineering applications to evaluate its effectiveness: load CM; early bearing failure detection; and speed CM. The results were compared with some benchmark methods reported in the literature, where significant improvements of the proposed method were demonstrated, indicating its good potentials in engineering applications.
•Uses compressive sensing and sparse over-complete feature learning.•Uses the unsupervised sparse autoencoder for learning feature representations.•Achieved high classification accuracy even from ...highly compressed measurements.•Our method needs significantly less computation time compared to other methods.•Our method improves the classification accuracy in machine fault diagnosis.
Condition classification of rolling element bearings in rotating machines is important to prevent the breakdown of industrial machinery. A considerable amount of literature has been published on bearing faults classification. These studies aim to determine automatically the current status of a roller element bearing. Of these studies, methods based on compressed sensing (CS) have received some attention recently due to their ability to allow one to sample below the Nyquist sampling rate. This technology has many possible uses in machine condition monitoring and has been investigated as a possible approach for fault detection and classification in the compressed domain, i.e., without reconstructing the original signal. However, previous CS based methods have been found to be too weak for highly compressed data. The present paper explores computationally, for the first time, the effects of sparse autoencoder based over-complete sparse representations on the classification performance of highly compressed measurements of bearing vibration signals. For this study, the CS method was used to produce highly compressed measurements of the original bearing dataset. Then, an effective deep neural network (DNN) with unsupervised feature learning algorithm based on sparse autoencoder is used for learning over-complete sparse representations of these compressed datasets. Finally, the fault classification is achieved using two stages, namely, pre-training classification based on stacked autoencoder and softmax regression layer form the deep net stage (the first stage), and re-training classification based on backpropagation (BP) algorithm forms the fine-tuning stage (the second stage). The experimental results show that the proposed method is able to achieve high levels of accuracy even with extremely compressed measurements compared with the existing techniques.
Process monitoring is necessary in machining operation to increase productivity, improve surface quality, and reduce unscheduled downtime. Tool wear and breakage are important and common source of ...machining problems due to high temperatures and forces of the machining process. Therefore, it is highly beneficial to develop an online tool condition monitoring (TCM) system. This paper investigates a robust tool wear monitoring system for milling operation. Recent developments in machine learning, in particular deep learning methods, result in significant improvement in automation of different industries. Therefore, in this research, we employed convolutional neural network (CNN) as a well-established and powerful deep learning algorithm for tool wear estimation. Wavelet packet-based features are extracted for tool wear monitoring as a powerful time-frequency fault indicator. Moreover, a hybrid feature extraction method is proposed using wavelet time-frequency transformation and spectral subtraction algorithms to intensify the effect of tool wear in the signal and reduce the effect of other cutting parameters. CNN-based monitoring systems are compared with three other machine learning methods (support vector machine, Bayesian rigid network, and K nearest neighbor method) as the baseline. The research is validated using different datasets. The algorithms are implemented and compared using experimental force and vibration signals from LIPPS lab of ETS university as well as using current signals as the fault indicator from Nasa_Ames dataset.
The ever increasing size of wind turbines and the move to build them offshore have accelerated the need for optimised maintenance strategies in order to reduce operating costs. Predictive maintenance ...requires detailed information on the condition of turbines. Due to the high costs of dedicated condition monitoring systems based on mainly vibration measurements, the use of data from the turbine supervisory control and data acquisition (SCADA) system is appealing. This review discusses recent research using SCADA data for failure detection and condition monitoring (CM), focussing on approaches which have already proved their ability to detect anomalies in data from real turbines. Approaches are categorised as (i) trending, (ii) clustering, (iii) normal behaviour modelling, (iv) damage modelling and (v) assessment of alarms and expert systems. Potential for future research on the use of SCADA data for advanced turbine CM is discussed.
Capacitors are widely used in dc links of power electronic converters to balance power, suppress voltage ripple, and store short-term energy. Condition monitoring (CM) of dc-link capacitors has great ...significance in enhancing the reliability of power converter systems. Over the past few years, many efforts have been made to realize CM of dc-link capacitors. This article gives an overview and a comprehensive comparative evaluation of them with emphasis on the application objectives, implementation methods, and monitoring accuracy when being used. First, the design procedure for the CM of capacitors is introduced. Second, the main capacitor parameters estimation principles are summarized. According to these principles, various possible CM methods are derived in a step-by-step manner. On this basis, a comprehensive review and comparison of CM schemes for different types of dc-link applications are provided. Finally, application recommendations and future research trends are presented.