A novel intelligent fault diagnosis method based on deep learning and particle swarm optimization support vectors machine (PSO-SVM) is proposed. The method uses deep learning neural network (DNN) to extract fault features automatically, and then uses support vector machine to classify diagnose faults based on extracted features. DNN consists of a stack of denoising autoencoders. Through pre-training and fine-tuning of DNN, features of input parameters can be extracted automatically. This paper uses particle swarm optimization algorithm to select the best parameters for SVM. The extracted features from multiple hidden layers of DNN are used as the input of PSO-SVM. Experimental data is derived from the data of rolling bearing test platform of West University. The results demonstrate that deep learning can automatically extract fault feature, which removes the need for manual feature selection, various signal processing technologies and diagnosis experience, and improves the efficiency of fault feature extraction. Under the condition of small sample size, combining the features of the multiple hidden layers as the input into the PSO-SVM can significantly increase the accuracy of fault diagnosis.
The theoretical analysis of corrective characteristics of three kinds of polishing methods for mid-frequency errors was studied, which was aimed to confirm the possibility that computer control optical surfacing and computer control activelap can be replaced by bonnet polishing in the machining process. The first step was to calculate the removal functions of three kinds of polishing technologies and use fast Fourier transform to figure out the frequency spectrum of each method. After that, according to the frequency spectra, curves of cut-off frequencies related to the working ranges of spatial frequencies errors were obtained. It revealed that the affected scope of spatial frequencies is determined by the polishing method, diameter size of polishing tool and shape of removal function. Moreover, only low-frequency errors could be modified and mid-frequency errors could not be corrected or created by computer control active-lap, and computer control optical surfacing can correct part of the mid-frequency errors and low-frequency errors in the polishing process, but at the same time can produce some new mid-frequency errors; as for bonnet polishing, it can be computer control active-lap-like in smoothing which only modified and created the low-frequency errors or computer control optical surfacing-like which corrected and created the mid-frequency errors in local polishing. Otherwise, the efficiency of bonnet polishing is higher than the other two methods. As a result, seen from the point of correction ability of mid-frequency or polishing efficiency, bonnet polishing could replace computer control active-lap and computer control optical surfacing for finishing two polishing stages by only one tool, which is significant to extending the application of bonnet polishing in optical manufacturing.
In the absence of a priori knowledge, manual feature selection is too blind to find the sensitive features which can effectively classify the different fault features. And it is difficult to obtain a large number of typical fault samples in practice to train the intelligent classifier. A novel intelligent fault diagnosis method based on feature selection and deep learning is proposed for rotating machine mechanical in the paper. In this method, the deep neural network is not only used for feature extraction but also for fault diagnosis. First, the deep neural network 1 is used to extract feature from the spectral signal of the original signal. In addition, the original vibration signal is decomposed to a series of intrinsic mode function components by empirical mode decomposition, and the statistical features of each intrinsic mode function component are extracted by the deep neural network 2 in time domain and frequency domain. Second, the extraction features of the original signal spectrum and the extraction features of each intrinsic mode function component are evaluated, respectively. After features evaluation, the selected sensitive features are combined together to construct a joint feature. Finally, the joint feature is put into the deep neural network 3 to realize the automatic recognition of different fault states of rotating machinery. The experimental results show that the method proposed in this paper which integrated time-domain, frequency-domain statistical characteristics, empirical mode decomposition, feature selection, and deep learning methods can obtain the fault information in detail and can select sensitive features from a large number of fault features. The method can reduce the network size, improve the mechanical fault diagnosis classification accuracy, and has strong robustness.
Some procedures or functions had be added to an ESTA (Expert System Shell for Text Animation) so that the ESTA and MATLAB can communicate via some data files.On this basis,a deep learning-DBN(Deep Belief Network) and two BP(back propagation) artificial neural network based on the MATLAB programming were researched by using directly DGA (Dissolved Gas Analysis) and characteristic gas method in transformer oil chromatographic analysis.The transformer fault diagnosis expert system based on a three ratio and characteristic gas method of DGA and ESTA including the DBN and two BP artificial neural network programmed in MATLAB had be created.The basic application shows the effectiveness of the expert system.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.