Font Size: a A A

Vibration Signal Analysis And Feature Identification Of Forest Tree Pests

Posted on:2020-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:S H LiuFull Text:PDF
GTID:2370330575995024Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
The larvae of the forest pests live inside the trunk and the species is difficult to identify.In order to quickly and accurately identify the species of the larvae of the longhornedbeetles and carry on biological control,a method for testing and analyzing the vibration signals of the larvae of the longhornedbeetles eating the trees,extracting the behavioral characteristics of the larvae of the longhornedbeetles and judging the species of the longhornedbeetles is proposed.The following results are obtained:Pests in the trunk will have eating trunks,clearing the passages,defending themselves and crawling.These behaviors will cause a waveform similar to the vibration waveform,while the outside strong winds and other disturbances will also cause vibration.The vibration signal caused by pests that eat trees has a rising edge which is a steep pulse and then the pulse is attenuated by an approximate exponential envelope,which is clearly different from the vibration waveform caused by external disturbances.The maximum relative amplitude of vibration caused by pest eating trees is above 60%,the maximum relative amplitude of the clearing passages vibration is 40%-60%,the maximum relative amplitude of defending themselves vibration is 20%-40%,and the maximum relative amplitude of crawling vibration is less than 20%,they can be directly discriminated.It has been verified in future research.Using variational mode decomposition to reduce noise of the vibration signal caused by pests eating trees.The number of modes is determined by the central frequency discrimination method and the operation time and accuracy of different penalty factors are compared to determine the value of the penalty factor.The principle of VMD noise reduction and the method of noise component identification by waveform and time entropy are introduced.The reconstructed signal is decomposed by three-layer wavelet packet,perform Fourier transform on the decomposed components and calculate the energy ratio to extract the characteristics of the vibration caused by pests eating trees.This paper collected a large amount of data of the Batocera horsfieldi(Fraxinus chinensis),the Anoplophora glabripennis(Platanus acerifolia),the Aromia bungii(Prunus cerasifera and Malus micromalus).Time domain characteristics are extracted for the vibration signal after noise reduction.It mainly includes the vibration waveform and the difference in vibration duration.For frequency domain characteristics,it mainly includes frequency composition and node energy ratio.The difference in the vibration of different larvae is very large,which is suitable as a characteristic parameter to judge the species of the longhornedbeetles.In the same time,the characteristics of the vibration signal of the same host are clearly different Prove that the method based on vibration signal analysis can effectively identify pests.Through the comprehensive discrimination of time domain and frequency domain,the identification of pests is completed.For the same kind of trees,there may be a variety of longhornedbeetles.This paper randomly collect Fraxinus chinensis,Platanus acerifolia,Prunus cerasifera and Malus micromalus and other trees,collected the corresponding signals for analysis,the correct rate reached 90%.No corresponding pests are found in the remaining trees,but there are other kinds of longhornedbeetles.This paper proves the feasibility of using vibration signals to identify the species of the longhornedbeetles.It is of great practical value to formulate scientific prevention and control programs for early identification of pest species.It has laid a solid foundation for future biological control.
Keywords/Search Tags:Variational Modal Decomposition, Wavelet Packet Decomposition, Dry Vibration Signal, Longhorn Control
PDF Full Text Request
Related items