Font Size: a A A

Analog Circuit Fault Diagnosis Based On Wavelet Analysis And Particle Swarm Optimization Neural Network

Posted on:2014-09-24Degree:MasterType:Thesis
Country:ChinaCandidate:F J XuFull Text:PDF
GTID:2268330425961145Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
Since the1990s, neural network using as an adaptive pattern recognitiontechnology has drawn academia’s attention because it provides an effective approachto solve problems of fault diagnosis of analog circuits. In recent years, the emergenceof multiple bionics algorithms expands the application of the neural network invarious fields. Taking advantage of dominant species, the particle swarm algorithm, akind of swarm intelligence algorithm, provides a new thought of optimization ofsolving the complex problems.Firstly,this article introduced a specific step and method of general analogcircuit fault diagnosis based on neural network. Then, this paper expounds the basicstructure, basic principle of BP neural network and its advantages and disadvantagesand improvement method in fault diagnosis. Secondly, this article described thetheory of Particle Swarm Optimization (PSO) in detail,we can study its socialbehavior and convergence performance by analyzing evolvable equation of particle’svelocity.Thirdly, by combining the strong global searching characteristic of particleswarm optimization algorithm with the fast local search ability of BP algorithm, wecan optimize neural network effectively. At the same time, wavelet analysis candeeply reflect the nature of the circuit operation state, which can reduce the neuralnetwork input and simply the network architectures. Thus, a new method of analogcircuit fault diagnosis based on PSO algorithms and wavelet neural network isproposed.Throughout this paper, the main line is neural network and the research object isamplifier circuit, combining wavelet transform and PSO algorithms algorithm toapply to the analog circuit fault diagnosis, then simulating the fault diagnosis processwith ORCAD and MATLAB software. In diagnosis on the circuit, after beingnormalized and principal component analysis using wavelet decomposition as anpreprocessing tool, it effectively reduce the input dimension of neural networks; Andthen use PSO algorithm to optimize the BP neural network’s weights and thresholds,to overcome the shortcomings of the BP network and improve its structure. Theresults show that wavelet analysis can effectively deal with the input data and extractthe fault signal feature vector, shorten the time of the fault diagnosis, by joining thegenetic algorithm, the BP neural network structure is greatly improved, the number of training steps is reduced, diagnostic accuracy is greatly increased.
Keywords/Search Tags:Analog circuit, fault diagnosis, neural network, particle swarm, wavelet analysis
PDF Full Text Request
Related items