Papers by Keyword: Fault Diagnosis

Abstract: When an early fault turns up in rotating machinery, the normal vibration signal will be modulated with periodic transient shock pulses. It’s significant to diagnose these periodic shock pulses for early fault prognostic and diagnostic tests. Usually demodulating is one of the most effective and common used method. Because of the strong background noise, it’s very difficult to select the parameters of band-pass filter. In this paper, we propose to use Ensemble Empirical Mode Decomposition (EEMD) coordinating with spectrum kurtosis theory to choose the Intrinsic Mode Functions (IMFs) to reduce the background noise and select the parameters of band-pass filter adaptively by fast-kurtogram. Energy operator demodulating method is used to demodulate the rebuilt signal to identify the faults frequencies. Energy operator demodulating displays better accuracy and little edge error. The achieved accuracy in the simulation indicates that this proposed transient faults diagnosis method is highly reliable and applicable in early transient faults diagnosis of industrial rotating machinery.

Abstract: A gear plays a crucial role in the performance of a gear box. The faults in a gear reduces the gear life and if problem arises in shaft it affects bearing. Gear box is finally affected due to these faults. Vibration signals carries information about condition of a gear box which are captured using piezoelectric accelerometer. In this paper, features are extracted and classified using K nearest neighbours (KNN) algorithms for both time and frequency domain. The effectiveness of KNN in classification of gear faults for both time and frequency domain is discussed and compared.

Abstract: Gearbox plays a vital role in various fields in the industries. Failure of any component in the gearbox will lead to machine downtime. Vibration monitoring is the technique used for condition based maintenance of gearbox. This paper discusses the use of machine learning techniques for automating the fault diagnosis of automobile gearbox. Our experimental study monitors the vibration signals of actual automobile gearbox with simulated fault conditions in the gear and bearing. Statistical features are extracted and classified for identifying the faults using decision tree and Naïve bayes technique. Comparison of the techniques for determining the classification accuracy is discussed.

Abstract: The proliferation of energy management systems leads to new potentials of data acquisition that can deliver improved machine information through intelligent linking. In addition to energy controlling, the newly gotten database creates further use cases for advanced purposes. This paper presents an exemplary application of a diagnostic scenario for industrial robots. For this objective, data fusion of energy data and operating logs is necessary to obtain detailed knowledge of the behavior of a production system. Hereby, an online measurement system will be described, which helps to uncover inefficiencies in production systems.

Abstract: An intelligent diagnostic method based on 3-D plot continuous wavelet transform (3-D plot CWT) and fuzzy inference system is presented to investigate the detectability and classification of rotor broken bars faults in induction machine (IM) and to overcome the limitation of classical Fourier Transform (FT). This approach is successfully used with Motor Current Signature Analysis (MCSA) and suitable developed model of IM in healthy and faulty mode using Matlab environment. As first step we performed new results using 3-D plot CWT to extract the discriminating features. The features extracted from the wavelet transformed signal are the second most predominant frequency, the time range at which it occurs and the corresponding wavelet coefficients .Then as second and last step a fuzzy Inference system is designed and implemented using Matlab software with these three features extracted from the wavelet transformed signal as inputs and generates an output that classifies the fault and no fault conditions. It is observed that the results are satisfactory.

Abstract: An intelligent diagnostic method based on 3-D plot continuous wavelet transform (3-D plot CWT) and fuzzy inference system is presented to investigate the detectability and classification of rotor broken bars faults in induction machine (IM) and to overcome the limitation of classical Fourier Transform (FT). This approach is successfully used with Motor Current Signature Analysis (MCSA) and suitable developed model of IM in healthy and faulty mode using Matlab environment. As first step we performed new results using 3-D plot CWT to extract the discriminating features. The features extracted from the wavelet transformed signal are the second most predominant frequency, the time range at which it occurs and the corresponding wavelet coefficients .Then as second and last step a fuzzy Inference system is designed and implemented using Matlab software with these three features extracted from the wavelet transformed signal as inputs and generates an output that classifies the fault and no fault conditions. It is observed that the results are satisfactory.

Abstract: Centrifugal pumps are widely used in industry and also domestically. It is commonly used for its robust design and its efficiency. Every machine has to be monitored periodically in order to maintain its efficiency and also to avoid unexpected failure which lead to loss of efficiency. So fault diagnosis is necessary to monitor the pump periodically for finding out the defects in pump and to replace it if necessary. Dismantling and assembling of pumps during fault diagnosis is a tedious process, vibration analysis can be helpful to monitor the performance of the pump system without dismantling. For the experimentation purpose mono-block centrifugal pumps have been used in this work. By using the Lab VIEW program and DAQ card as an interface, amplitude and frequency of vibration is obtained at different axes of the pump with the help of an accelerometer. Then the vibration spectrum is analyzed and defects are pointed out by identifying the frequency at which the amplitude of vibration is above the danger limit. The defects such as unbalance of impeller, bent shaft in pump, misalignment of shaft, hydraulic pulsation, cavitation and bearing defects are diagnosed using vibration data. The frequency at which different defects are occurring has been founded out by means of experimentation in the centrifugal pumps. Thus by diagnosing centrifugal pump using vibration data reduces cost and time for periodical maintenance. Shape memory alloy based ATDVA is used to control the amplitude of vibration due to hydraulic pulsation. Around 60% reduction in amplitude of vibration is evident for the varying excitation frequency between 336 Hz and 340 Hz due to hydraulic pulsation.

Abstract: In this paper, a new fault diagnosis method is proposed based on CFD technology in order to research the fault mechanism of mine axial fan. The method combined RANS equation and Realizable k-turbulence model realizes the three dimensional aerodynamic performance analysis of the fan impeller. Three-dimensional impeller model of axial fan was set up in three conditions which including normal, crack and blade loose, and comparing the turbulence characteristic of the three models. The research provides the theoretical basis for the fault mechanism of blade failure, and it has important reference value for fault diagnosis method based on CFD of mine axial fan.

Abstract: Bearings are one of the most important components in rotating machineries. Their failures will lead to great production loss and increase the maintenance cost. So, condition monitoring work of bearings can save and avoid the potential loss caused by bearing fault. Lucy-Richardson deconvolution (LRD) algorithm, as an image processing technique, started to be used in bearing fault diagnosis. However, only data of bearings working in electric motor are used to validate the method. In engineering cases, most bearings are working in gearbox. Therefore, the bearing fault signals are very weak compared to the gear vibration signal. It is usually difficult to detect the bearing fault in this case. LRD algorithm is used to enhance the bearing fault diagnosis and some characteristics in this case are discussed. Experiment data analysis demonstrates that LRD can enhance the periodic impulse signal effectively. Otherwise, if the desired fault signal is weak compared to non-desired signal, then, the desired fault signal will be continued weaken by LRD which is not benefit to bearing’s incipient fault detection.

Abstract: As the failure of a hydraulic pump is always instantaneous, the failure data are difficult to obtain. High-efficiency fault diagnosis under small-sample conditions for hydraulic pumps is urgently required in engineering applications. A fault diagnosis approach based on wavelet packet transform (WPT), singular value decomposition (SVD), and support vector machine (SVM) is proposed in this study. First, the nonlinear, non-stationary vibration signal of the hydraulic pump is decomposed into components by WPT. Second, singular value vectors are acquired as feature vectors by applying SVD to the components. Third, the health states of the hydraulic pumps are determined and classified with a SVM classifier. Furthermore, the SVM and Elman neural network classifiers are compared in terms of fault classification to demonstrate the superiority of SVM in dealing with small-sample problems. The results of the plunger pump rig test show that the proposed method can diagnose the faults of the hydraulic pump accurately even when the number of samples is small.