Research On Fault Identification Of Drainage Pipe Blockage Based On Acoustic Active Detection

With the accelerated process of urbanization,the construction of drainage pipelines is also increasing.However,compared to the developed countries,China's pipeline failure rate is still higher.The blockage of drainage pipeline is partially blocked in the early stage.It will block up the area with the increase of accumulation.It will not only affect the urban drainage work but also pollute the environment and waste water resources.Based on this,it is very necessary to research the fault identification of the early partial blockage of the drainage pipeline.Among the numerous fault detection methods,acoustic detection is widely used because it is not affected by complex pipe network lines.At present,passive detection methods are mostly adopted for pipeline faults.Most of the feature extraction methods extract a single fault characteristic for fault identification,which can not meet the requirement of early identification of partial blockage.Therefore,this paper uses the active acoustic detection method to detect faults,and extract the characteristics of signals by multi-feature fusion to solve the identification problem of early blockage faults in drainage pipelines.The main research work of the dissertation is as follows:(1)Identification of drainage pipeline blockage based on the LMD features of acoustic signal fusion and SVM.Aiming at the detection problem of partial blockage within the urban drainage pipeline in the early stage,firstly,the LMD method is used to decompose the acoustic response signal to get the signal components of different scales,and the characteristic PF components are selected from the signal components by correlation analysis and energy distribution method.Then,the feature set is constructed by extracting its energy entropy,approximate entropy and average sound pressure by its characteristic PF components.Finally,the cross validation is used to optimize the parameters of the SVMs classifier to identify the blockage fault signal.The experiment shows that this method can effectively identify the normal pipeline and the blocked pipeline,and can preliminarily realize the purpose of identifying the blockage in pipeline.(2)Research on blockage identification of drainage pipeline based on wavelet packet features of acoustic signal fusion and SVM.In order to solve the problem of partial blockage detection of urban drainage pipeline and the difficulty in identifying the degree of blockage.Firstly,the acoustic response signal of the pipeline is decomposed by 3 layers of wavelet packet,and the high energy wavelet packet node is selected to reconstruct the signal to be feature components.Then the characteristics of energy entropy,approximate entropy and fractal box dimension of the feture components are extracted respectively,so the classification feature sets can be constructed.Finally,the particle swarm optimization algorithm is used to optimize the parameters of the SVM classifier to identify the blockage fault signal.This method can further identify the degree of blockage of the pipeline on the basis of completing the method 1 to identify the blockage,and is more in line with the actual needs of the project.(3)A multi-feature fusion drainage pipeline fault identification method based on distance separability criterion.Firstly,the acoustic response signals of the drainage pipeline are divided into frames as feature set.Then the characteristics of A weighted total sound pressure level,energy entropy and fractal box dimension of the feature components are extracted respectively,so the classification feature sets can be constructed.In addition,the distance separability criterion is introduced to calculate the distinguishing degree index of the feature vector so as to remove the low distinguishing feature vector.In addition,the identification weight of the single feature extracting mode is also calculated.Finally,random forest identification model are established on the classification feature sets based on the weighted fusion.Based on the functions of the above two methods,this method solves the problem of information redundancy and low discrimination in the feature set,and realizes the repeated blockage identification function,further improving the identification function.This paper mainly studies the problem of fault identification in urban drainage pipeline,and puts forward the above three methods.The three methods respectively solve the problem of identifying blockage faults,judging the degree of single blockage fault,identifying single blockage and repeated blocking and reducing feature set redundancy.The method of this paper can accurately realize the fault detection of the blockage of the drainage pipeline,which has certain practical significance in engineering.