Detection and material identification of loose particles inside the aerospace power supply via stochastic resonance and LVQ network

Abstract

The detection of loose particles inside an aerospace power supply is important to improve the reliability of the whole space system. This paper investigates the detection and material identification of loose particles within an aerospace power supply based on the particle impact noise detection (PIND) test. A stochastic resonance algorithm is employed to detect the presence of tiny particles. A learning vector quantization (LVQ)-based material identification method is proposed. Finally, experiments are conducted to demonstrate the effectiveness of the proposed technique. Experimental results show that the accuracies of particle detection and material identification are above 90% and 80%, respectively, which meets end-user requirements.

Pattern recognition approach to identify loose particle material based on modified MFCC and HMMs

“…[6,7]. Given that no standard database exists and the use of different test conditions and performance metrics, the results cannot be compared with the classifiers in primary study directly.…”

“…For example, Wang et al [5] employed the artificial neural network (ANN) classifier with features extracted from pulse duration time, spectrum shape and the linear prediction coefficients. Recently, a stochastic resonance algorithm is employed to detect the presence of tiny particles and a LVQ-based material identification method is proposed in Wang's study [6]. A similar investigation reported by Zhai [7] used support vector machine (SVM) and principal component analysis (PCA) to distinguish the different material particles.…”

Identification of loose particles using deep learning convolutional network

“…Table 2 shows the results compared with other methods. As there is no open dataset of loose particle signals, results of [2][3][4] in the table is given by corresponding papers, and others come from experiments on our datasets. The proposed method outperforms [2][3][4] with more classes and shorter signals, and outperforms all methods with higher accuracy.…”

“…Traditional methods usually use machine learning to solve the problem, the usual process is extracting features from acoustic signals firstly and training classifiers then. Shujuan Wang [2] proposed a LVQ (Learning Vector Quantization)-based method which uses energy distribution vectors in wavelet domain. Long Zhang [3] compared the wavelet Fisher discriminant with AR model and LVQ neural networks and proves that the wavelet Fisher discriminant is better than others.…”

Detecting loose particle signals in multichannel recordings with transductive confidence predictor

“…In another attempt to identify the types of material including tin dregs, glass scrap, wire and rubber particles, Wang et al (2007) employed an artificial neural network classifier with features extracted from pulse duration time, spectrum shape and the linear prediction coefficients of particle signals. Recently, a stochastic resonance algorithm has been employed to detect the presence of tiny particles and a learning vector quantization (LVQ)-based material identification method is proposed in Wang et al (2011). In order to improve the classification accuracy, Zhang et al (2013) employed a wavelet Fisher discriminant for loose particle classification instead of the conventional autoregressive model and LVQ neural networks.…”