Research On Excavation Devices Localization Based On Decision Matrix And ELM Classifier

Underground pipeline networks suffered severe external breakage caused by earth surface excavation devices in a number of cities in many countries due to unordered excavations and noncanonical managements.The acoustic signal based excavation devices detection methods have been recently shown effective in the underground pipeline network surveillance system.However,merely relying on the recognition results may lead to a high false alarm rate in the real monitoring system due to that the urban underground pipelines are usually paved along fixed lines and operations out of the region may also trigger the surveillance system.Two acoustic signal based excavation device localization algorithms are developed in this paper as an aid to the surveillance system for urban underground pipeline network protection system.The novel recognition combining DOA estimation scheme can work efficiently in the real-world complex environment.The main contents are as follows:1.By four channel cross array,according to the principle of arrival of each microphone will have time delay,MUSIC single frequency algorithm is implemented.Acquisition and analysis of the voice signals of various mining equipment,to determine the appropriate reference frequency band for this project.According to the focus transformation,realize localization algorithm on real broadband signal,verifying the feasibility of the algorithm.2.Proposing pre-recognition operation in MUSIC algorithm,the interference caused by noise is eliminated through the signal recognition,which improve the accuracy of localization.Compared with the previous simple MUSIC algorithm,the location algorithm is more accurate,and the stability is improved.3.Based on the pre-recognition algorithm,the decision matrix(DM)is introduced,and the result of the recognition algorithm and the algorithm are summarized.Finally,the result of the decision algorithm is obtained.The improved degree of this algorithm is verified by experiments.Compared with the former two algorithms,it has better stability,and has a great improvement on the location accuracy of the individual devices.4.Based on decision matrix(DM),the localization of multiple sound source sound signals is achieved by using different sound source identification results classification.In the complex environment of the actual work of a variety of instruments at the same time,there is a good effect,to achieve the desired effect.