Font Size: a A A

Research On Lightning Signals Recognition Based On Fractal Features

Posted on:2021-03-22Degree:MasterType:Thesis
Country:ChinaCandidate:Y FangFull Text:PDF
GTID:2370330623482077Subject:Intelligent information processing
Abstract/Summary:PDF Full Text Request
Lightning is a complex discharge phenomenon in nature,and it is also a kind of disastrous weather,which will cause irreparable loss to human life and even human death.Therefore,the study of lightning is very important,and it is of great significance to find a newer and more effective method to study lightning early warning.Lightning is a kind of instantaneous and nonlinear complex signal.In this paper,fractal theory,a nonlinear analysis method,is used to study the lightning signal.The purpose is to reveal the deeper characteristics of lightning signal,mainly analyze the fractal characteristics of lightning signal,and further use the fractal characteristics to classify and identify the lightning signal.The main contents of this paper are as follows:1.Empirical mode decomposition(EMD)and fractal box dimension method are combined.EMD can reflect the detail characteristics of each layer of lightning signal well.Combined with fractal box dimension method,it can reflect the fractal characteristics of each layer of lightning signal more carefully.Then,the dominant component of lightning signal is selected by the correlation number,and the box dimension of its dominant component is analyzed and compared.It is found that different types of lightning signal correspond to different box dimension of dominant component.At the same time,the Hurst index of lightning signal is analyzed to grasp the development trend of lightning signal.2.Only using a single fractal can not completely show the deeper laws and characteristics of lightning signal.Therefore,the multifractal method is used to analyze the multifractal spectrum parameters of lightning signal.At the same time,the multifractal analysis of special lightning events can further enrich the recognition of lightning signal types.3.The box dimension and multifractal spectrum parameters of the dominant component of lightning signal are taken as the feature variables,and support vector machine is used to identify cloud lightning,ground lightning,return stroke and bipolar narrow pulse events.The experimental results show that the single fractal and multifractal are complementary,and the combination of them can describe the fractal characteristics of lightning signal comprehensively from the whole to the local,so as toget a higher recognition rate.
Keywords/Search Tags:Lightning Signals, EMD decomposition, box counting, multifractal, classification and recognition
PDF Full Text Request
Related items