22 research outputs found
Health assessment and fault diagnosis for centrifugal pumps using Softmax regression
No full textReal-time health monitoring of industrial components and systems that can detect, classify, and predict impending faults is critical to reduce operating and maintenance costs. This paper presents a softmax regression-based prognostic method for on-line health assessment and fault diagnosis. System conditions are evaluated by processing the information gathered from access controllers or sensors mounted at different points in the system, and maintenance is performed only when the failure or malfunction prognosis is indicated. Wavelet packet decomposition and fast Fourier transform techniques are used to extract features from non-stationary vibration signals. Wavelet packet energies and fundamental frequency amplitude are used as features, and principal component analysis is used for feature reduction. Reduced features are input into softmax regression models to assess machine health and identify possible failure modes. The gradient descent method is used to determine the parameters of softmax regression models. The effectiveness and feasibility of the proposed method are illustrated by applying to a real application
Bearing performance degradation assessment and prediction based on EMD and PCA-SOM
No full textBearings are used in a wide variety of rotating machineries. Bearing vibration signals are non-stationary signals. According to the non-stationary characteristics of bearing vibration signals, a bearing performance degradation assessment/prediction and fault diagnosis method based on empirical mode decomposition (EMD) and principal component analysis (PCA)-self organizing map (SOM) is proposed in this paper. First, vibration signals are decomposed into a finite number of intrinsic mode functions, after which the EMD energy feature vector, which is composed of all the IMF energy, is obtained. PCA is then introduced to reduce the dimension of feature vectors. After that, the reduced feature vectors are selected as input vectors of the SOM neural network for performance degradation and fault diagnosis. Finally, the degradation trend of bearing is predicted by Elman neural network. The analysis results from bearings with different fault degrees or degradation trend and fault patterns show that the proposed method can assess and predict the degradation of bearing suitably and achieve a fault recognition rate of over 95 % for various bearing fault patterns
Bearing performance degradation assessment and prediction based on EMD and PCA-SOM
No full textBearings are used in a wide variety of rotating machineries. Bearing vibration signals are non-stationary signals. According to the non-stationary characteristics of bearing vibration signals, a bearing performance degradation assessment/prediction and fault diagnosis method based on empirical mode decomposition (EMD) and principal component analysis (PCA)-self organizing map (SOM) is proposed in this paper. First, vibration signals are decomposed into a finite number of intrinsic mode functions, after which the EMD energy feature vector, which is composed of all the IMF energy, is obtained. PCA is then introduced to reduce the dimension of feature vectors. After that, the reduced feature vectors are selected as input vectors of the SOM neural network for performance degradation and fault diagnosis. Finally, the degradation trend of bearing is predicted by Elman neural network. The analysis results from bearings with different fault degrees or degradation trend and fault patterns show that the proposed method can assess and predict the degradation of bearing suitably and achieve a fault recognition rate of over 95 % for various bearing fault patterns
Bearing performance degradation assessment and prediction based on EMD and PCA-SOM
No full textBearings are used in a wide variety of rotating machineries. Bearing vibration signals are non-stationary signals. According to the non-stationary characteristics of bearing vibration signals, a bearing performance degradation assessment/prediction and fault diagnosis method based on empirical mode decomposition (EMD) and principal component analysis (PCA)-self organizing map (SOM) is proposed in this paper. First, vibration signals are decomposed into a finite number of intrinsic mode functions, after which the EMD energy feature vector, which is composed of all the IMF energy, is obtained. PCA is then introduced to reduce the dimension of feature vectors. After that, the reduced feature vectors are selected as input vectors of the SOM neural network for performance degradation and fault diagnosis. Finally, the degradation trend of bearing is predicted by Elman neural network. The analysis results from bearings with different fault degrees or degradation trend and fault patterns show that the proposed method can assess and predict the degradation of bearing suitably and achieve a fault recognition rate of over 95 % for various bearing fault patterns
Evolution of Gas-Liquid Two-Phase Flow in an M-Shaped Jumper and the Resultant Flow-Induced Vibration Response
No full textThe vibration excited by gas-liquid multiphase flow endangers the structural instability and fatigue life of subsea jumpers due to the cyclic behavior. In this paper, the multiphase flow-induced vibration (MFIV) of an M-shaped jumper is numerically investigated using a two-way fluid-structure interaction (FSI) approach. The effect of gas-liquid ratios (β) ranging from 1:1 to 1:5 is examined with a fixed flow velocity of 3 m/s, and the influence of mixture velocity (vm) in the range 2–6 m/s is evaluated with a gas-liquid ratio of 1:1. The numerical results reveal the detailed flow evolution of the gas-liquid mixture along the jumper. With inflow of slugs, the pattern successively experiences the slug flow, wavy flow, imperfect annular flow, stratified flow, churn flow, wavy flow and imperfect annular flow in the pipe segments when β = 1:1 and vm = 3 m/s. This development of mixture flow is significantly altered by changing either the gas-liquid ratio or the mixture velocity. In comparison with the flow evolution in a stationary jumper, the pattern in each pipe segment is not been substantially changed due to the limited response amplitude of order of 10−3D (D is the outer diameter of the jumper). Due to the complex flow evolution, the pressure acting on the six bends of the jumper fluctuate in multiple frequencies. Nevertheless, the dominant fluctuation frequency is approximately equal to the inflow slug frequency. Moreover, the inflow slug frequency also dominates the in-plane response of the jumper. Both the in-plane and out-of-plane responses of the jumper exhibit spatial-temporal variation characteristics. The most vigorous oscillation occurs at the midspan of the jumper. As β is reduced, the out-of-plane response of the jumper midspan is suppressed while the in-plane response is enhanced. In contrast, both the in-plane and out-of-plane oscillations of the jumper midspan are amplified with the increase of vm
Bearing performance degradation assessment and prediction based on EMD and PCA-SOM
No full textBearings are used in a wide variety of rotating machineries. Bearing vibration signals are non-stationary signals. According to the non-stationary characteristics of bearing vibration signals, a bearing performance degradation assessment/prediction and fault diagnosis method based on empirical mode decomposition (EMD) and principal component analysis (PCA)-self organizing map (SOM) is proposed in this paper. First, vibration signals are decomposed into a finite number of intrinsic mode functions, after which the EMD energy feature vector, which is composed of all the IMF energy, is obtained. PCA is then introduced to reduce the dimension of feature vectors. After that, the reduced feature vectors are selected as input vectors of the SOM neural network for performance degradation and fault diagnosis. Finally, the degradation trend of bearing is predicted by Elman neural network. The analysis results from bearings with different fault degrees or degradation trend and fault patterns show that the proposed method can assess and predict the degradation of bearing suitably and achieve a fault recognition rate of over 95 % for various bearing fault patterns
Health assessment and fault diagnosis for centrifugal pumps using Softmax regression
No full textReal-time health monitoring of industrial components and systems that can detect, classify, and predict impending faults is critical to reduce operating and maintenance costs. This paper presents a softmax regression-based prognostic method for on-line health assessment and fault diagnosis. System conditions are evaluated by processing the information gathered from access controllers or sensors mounted at different points in the system, and maintenance is performed only when the failure or malfunction prognosis is indicated. Wavelet packet decomposition and fast Fourier transform techniques are used to extract features from non-stationary vibration signals. Wavelet packet energies and fundamental frequency amplitude are used as features, and principal component analysis is used for feature reduction. Reduced features are input into softmax regression models to assess machine health and identify possible failure modes. The gradient descent method is used to determine the parameters of softmax regression models. The effectiveness and feasibility of the proposed method are illustrated by applying to a real application
Signal processing and health assessment techniques in structural health monitoring
No full textConsidering the practical necessity of realizing the aircraft structure health monitoring (SHM), not only some advanced sensors must be adopted, an appropriate signal processing method is also necessary, in order to achieve the goals of signal characteristic extraction and health assessment. Based on the fact that piezoelectric (PZT) sensor signals extracted from the monitored structure carry a lot of useful health related information, this paper proposes to utilize the power characteristics of PZT signals for health assessment of the aircraft structure. The Hilbert-Huang Transform (HHT) algorithm is employed to calculate the power characteristic vector, after that, the self-organizing map (SOM) is used to map the corresponding characteristics into a confidence value (CV) which represents the health state of the monitored structure. The experimental results demonstrate that the proposed method is reliable
Health assessment and fault diagnosis for centrifugal pumps using Softmax regression
No full textReal-time health monitoring of industrial components and systems that can detect, classify, and predict impending faults is critical to reduce operating and maintenance costs. This paper presents a softmax regression-based prognostic method for on-line health assessment and fault diagnosis. System conditions are evaluated by processing the information gathered from access controllers or sensors mounted at different points in the system, and maintenance is performed only when the failure or malfunction prognosis is indicated. Wavelet packet decomposition and fast Fourier transform techniques are used to extract features from non-stationary vibration signals. Wavelet packet energies and fundamental frequency amplitude are used as features, and principal component analysis is used for feature reduction. Reduced features are input into softmax regression models to assess machine health and identify possible failure modes. The gradient descent method is used to determine the parameters of softmax regression models. The effectiveness and feasibility of the proposed method are illustrated by applying to a real application
Health assessment and fault diagnosis for centrifugal pumps using Softmax regression
No full textReal-time health monitoring of industrial components and systems that can detect, classify, and predict impending faults is critical to reduce operating and maintenance costs. This paper presents a softmax regression-based prognostic method for on-line health assessment and fault diagnosis. System conditions are evaluated by processing the information gathered from access controllers or sensors mounted at different points in the system, and maintenance is performed only when the failure or malfunction prognosis is indicated. Wavelet packet decomposition and fast Fourier transform techniques are used to extract features from non-stationary vibration signals. Wavelet packet energies and fundamental frequency amplitude are used as features, and principal component analysis is used for feature reduction. Reduced features are input into softmax regression models to assess machine health and identify possible failure modes. The gradient descent method is used to determine the parameters of softmax regression models. The effectiveness and feasibility of the proposed method are illustrated by applying to a real application
