The present paper deals with the novel approach for clustering using the image feature of stabilization diagram for automated operational modal analysis in parametric model which is stochastic subspace identification (SSI)-COV. The evolution of automated operational modal analysis (OMA) is not an easy task, since traditional methods of modal analysis require a large amount of intervention by an expert user. The stabilization diagram and clustering tools are introduced to autonomously distinguish physical poles from noise (spurious) poles which can neglect any user interaction. However, the existing clustering algorithms require at least one user-defined parameter, the maximum within-cluster distance between representations of the same physical mode from different system orders and the supplementary adaptive approaches have to be employed to optimize the selection of cluster validation criteria which will lead to high demanding computational effort. The developed image clustering process is based on the input image of the stabilization diagram that has been generated and displayed separately into a certain interval frequency. and standardized image features in MATLAB was applied to extract the image features of each generated image of stabilisation diagrams. Then, the generated image feature extraction of stabilization diagrams was used to plot image clustering diagram and fixed defined threshold was set for the physical modes classification. The application of image clustering has proven to provide a reliable output results which can effectively identify physical modes in stabilization diagrams using image feature extraction even for closely spaced modes without the need of any calibration or user-defined parameter at start up and any supplementary adaptive approach for cluster validation criteria.
Previous studies have indicated that the pipe-surface-mounted helical strakes effectively reduce vortex-induced vibration (VIV) under a uniform flow application, particularly during the lock-in region. Since VIV experiments are time-consuming, observation is generated with an interval helical strakes parameter in pitch and height to lessen tedious procedures and repetitive post-processing analyses. The aforementioned result subset is insufficient for helical strakes design optimisation because the trade-off between the helical strakes dimension, lock-in region and flow velocity are non-trivial. Thus, a parametric model based on an improved recursive least squares (RLS) parameter estimation technique is proposed to define the statistical relationship between input, or strakes and pipe dimension, and output, or VIV amplitude ratio. As results suggested, revised RLS estimated VIV model demonstrated an optimal prediction with the highest coefficient of determination and lowest Integral Absolute Error. The feasibility of VIV parametric model was validated by embed into Genetic Algorithm (GA) as the fitness function to acquire a desirable helical strakes dimension with minimum VIV amplitude. The rapid generation of optimal helical strakes dimension which returned the highest VIV suppression implied a superior simulation method compared to the experimental outcome.
This paper presents parameters analysis for the estimated modal damping ratio using a new version of the automated enhanced frequency domain decomposition (AEFDD). The purpose of this study is to provide a better choice of a maximum number of points of time segments and modal assurance criterion (MAC) index number regarding to the variable level of system damping (low and high damped structure) and degree of freedom of the system. According current literature, frequency domain (FD) methods seem to have the problem with providing a correct identification of the modal damping ratio, since the correct estimate of modal damping is still an open problem and often leads to biased estimates. This technique is capable of providing consistent modal parameters estimation, particularly for modal frequencies and mode shapes. As a necessary fundamental condition, the algorithm has been assessed first from computed numerical responses according to random white noise, acting on different shear-type frame structures and corrupted with noise. Results indicate that reducing the value of natural frequencies and modal damping ratios of the modes under analysis demands longer time segments and a high value of the maximum number of points for adequate information on the decaying correlation functions when computing a modal damping ratio. In addition, the results also prove that the MAC index does not significantly affect the results for the low damped system. However, the use of a high MAC index value for the high damped system significantly introduces large error bound and it becomes worse, particularly for the higher modes, as the standard deviation of percentage error increases gradually. Furthermore, the use of a MAC index for a high number of points of time segments significantly increases the standard deviation of the percentage error.
Rotating machinery is one of the major components of industries that suffer from various faults due to the constant workload. Therefore, a fast and reliable fault diagnosis method is essential for machine condition monitoring. In this study, noise eliminated ensemble empirical mode decomposition (NEEEMD) was used for fault feature extraction. A convolution neural network (CNN) classifier was applied for classification because of its feature learning ability. A generalized CNN architecture was proposed to reduce the model training time. A sample size of 64×64×3 pixels RGB scalograms are used as the classifier input. However, CNN requires a large number of training data to achieve high accuracy and robustness. Deep convolution generative adversarial network (DCGAN) was applied for data augmentation during the training phase. To evaluate the effectiveness of the proposed feature extraction method, scalograms from related feature extraction methods such as ensemble empirical mode decomposition (EEMD), complementary EEMD (CEEMD), and continuous wavelet transform (CWT) are classified. The effectiveness of scalograms is also validated by comparing the classifier performance using grayscale samples from the raw vibration signals. All the outputs from bearing and blade fault classifiers showed that scalogram samples from the proposed NEEEMD method obtained the highest accuracy, sensitivity, and robustness using CNN. DCGAN was applied with the proposed NEEEMD scalograms to further increase the CNN classifier's performance and identify the optimal number of training data. After training the classifier using augmented samples, the results showed that the classifier obtained even higher validation and test accuracy with greater robustness. The proposed method can be used as a more generalized and robust method for rotating machinery fault diagnosis.
A major issue of machinery fault diagnosis using vibration signals is that it is over-reliant on personnel knowledge and experience in interpreting the signal. Thus, machine learning has been adapted for machinery fault diagnosis. The quantity and quality of the input features, however, influence the fault classification performance. Feature selection plays a vital role in selecting the most representative feature subset for the machine learning algorithm. In contrast, the trade-off relationship between capability when selecting the best feature subset and computational effort is inevitable in the wrapper-based feature selection (WFS) method. This paper proposes an improved WFS technique before integration with a support vector machine (SVM) model classifier as a complete fault diagnosis system for a rolling element bearing case study. The bearing vibration dataset made available by the Case Western Reserve University Bearing Data Centre was executed using the proposed WFS and its performance has been analysed and discussed. The results reveal that the proposed WFS secures the best feature subset with a lower computational effort by eliminating the redundancy of re-evaluation. The proposed WFS has therefore been found to be capable and efficient to carry out feature selection tasks.