Research On Acoustic Recognition Algorithm For Construction Equipments Based On Hybrid Visible Feature

With the rapid development of society and the acceleration of urban modernization,noise pollution in cities is increasing.Noise pollution has attracted more and more attention from government departments and residents.It directly affects the reputation of the government and the living environment of residents.Among them,building noise is the most important issue in urban noise.During the urban construction process,the surrounding residents are often disturbed by the harsh noise emitted by some construction equipments,and people are seriously affected.Studies have shown that people living in a noisy environment for a long time will have a serious impacts on both physical and mental health.Therefore,in order to be able to monitor the noise of construction equipment,an algorithm system capable of accurately recognizing the noise of construction equipment is required.Based on the traditional acoustic recognition algorithms,this paper proposes to visualize the acoustic signals and extract the visible features of acoustic signals.The new features contain more information about the acoustic signals,and the recognition rate of construction equipment is much higher than the traditional methods.The accomplishments and innovations in this paper are as follows:(1)Based on the research of traditional acoustic recognition,this paper proposes to visualize the acoustic signals,and convert the acoustic signals into image through the visible technology,then,extracted the statistic visible features based on the visible acoustic signals;(2)PCNN neural network is introduced in this paper.An information entropy sequence feature algorithm for visible acoustics signal based on PCNN neural network is proposed.A series of binary images will be output after a visible acoustic image is iteratively processed by the PCNN neural network.Therefore,in this paper,we process the visible acoustic signals through the PCNN neural network model and calculate the information entropy of the output binary image.Then we will obtain the information entropy sequences of the visible acoustic signals after N iterations;(3)Based on the random forest classifier,the algorithm combined with the visible feature is studied.Besides,the common noise signals in construction equipment are studied.New hybrid features are acquired through the visible acoustic features combined with the traditional acoustic features.The performance of the comparative experiments was carried out with the visible acoustic feature,the new hybrid features,and the traditional acoustic features.The results of the experiment verify the advantages performance of the proposed visible acoustic features and the new hybrid features.