D. J. Bordoloi

Optimization of support vector machine based multi-fault classification with evolutionary algorithms from time domain vibration data of gears

In the present work, a multi-fault classification of gears has been attempted by the support vector machine learning technique using the vibration data in time domain. A proper utilization of the support vector machine is based on the selection of support vector machine parameters. The main focus of this article is to examine the performance of the multiclass ability of support vector machine techniques by optimizing its parameters using the grid-search method, genetic algorithm and artificial bee colony algorithm. Four fault conditions were considered. A group of statistical features were extracted from time domain data. The prediction of fault classification is attempted at the same angular speed as the measured data as well as innovatively at the intermediate and extrapolated angular speed conditions. This is due to the fact that it is not feasible to have measurement of vibration data at all continuous speeds of interest. The classification ability is noted and it shows an excellent prediction performance.

Monitoring of Induction Motor Mechanical and Electrical Faults by Optimum Multiclass-Support Vector Machine Algorithms Using Genetic Algorithm

The induction motor (IM) may lose their normal efficiency and finally fail due to chronic mechanical or electrical faults or both. For the prevention of failure, the early detection of these faults is necessary. The vibration and current signals are measured and collected for varying speeds and load conditions of IMs from an experimental laboratory test rig. Experiments are conducted for four different mechanical fault conditions and five electrical fault conditions including one intact condition. The identification of fault predictions is studied by considering of all mechanical faults, electrical faults and no fault condition. The one-against-one Multiclass-Support Vector Machine Algorithms (MSVM) with radial basis function (RBF) kernel has been trained at various operating conditions of IMs and predictions performance is presented. Two MSVM algorithms, C-SVM and nu-SVM, are used for the investigation. The RBF kernel parameter (gama) and MSVM parameter (C and nu) are optimally selected by the Genetic Algorithm (GA) for better performance for each case. Prediction performances are presented for different speeds and load conditions.

Optimisation of SVM Methodology for Multiple Fault Taxonomy of Rolling Bearings from Acceleration Records

Health monitoring of bearings is very critical for satisfactory working of complex machinery. Thus, the ability to detect bearing faults and classify them based on their nature becomes very important aspect of health monitoring of machines. In the machine learning methodology for the fault taxonomy, the support vector machine (SVM) is well recognized for its generalization capabilities. In this work, the taxonomy of rolling element bearing faults has been discussed. Acceleration signatures are classified by the support vector machine (SVM) learning algorithm. The tuning of the SVM and kernel parameters is necessary for better taxonomy. The novelty of the paper is in comparing the ability to classify a set of faults by the tuned SVM and kernel parameters with the help of grid-search method (GSM), genetic algorithm (GA) and artificial bee colony algorithm (ABCA). Four fault settings along with no-fault condition were considered. Three statistical features were obtained from acceleration signatures. The fault taxonomy was performed at the identical rotational speed at which signals were captured. The taxonomy capability is observed and it depicted a very good prediction performance especially at higher speeds.

Health Monitoring of Gear Elements Based on Time-Frequency Vibration by Support Vector Machine Algorithms

Health monitoring of a gear box has been attempted by the support vector machine (SVM) learning technique with the help of time-frequency (wavelet) vibration data. Multi-fault classification capability of the SVM is suitably demonstrated that is based on the selection of SVM parameters. Different optimization methods (i.e., the grid-search method (GSM), the genetic algorithm (GA) and the artificial bee colony algorithm (ABCA)) have been performed for optimizing the SVM parameters. Four fault conditions have been considered including the no defect case. Time domain vibration signals were obtained from the gearbox casing operated in a suitable speed range. The continuous wavelet transform (CWT) and wavelet packet transform (WPT) are extracted from time domain signals. A set of statistical features are extracted from the wavelet transform. The classification ability is noted and compared against predictions when purely time domain data is used, and it shows an excellent prediction performance.Copyright

Health Monitoring of Gears Based on Vibrations by Support Vector Machine Algorithms

Health monitoring of gears is very critical for satisfactorily overall working of the complex machinery. Thus, the ability to detect gear faults and classify them based on their nature becomes very important aspect of health monitoring of machines. In this paper, SVM algorithms have been used for the multiclass prediction of faults with the help of time domain vibration signals obtained from the gearbox casing operated in a suitable speed range. Moreover, it tries to examine the performance of the SVM technique by optimizing its parameters on utilization of time domain data from multi-fault gear box. The SVM software was fed with the training data and testing data at similar operating speeds for three types of defects and no defect case, and classification ability of SVM was noted and found to be excellent. The sensitivity analysis of optimized parameters is studied and conclusions are drawn.Copyright