Nondestructive Testing Method Of Bolt Anchorage Quality Based On PSO-SVM

Bolt supporting technology plays an increasingly important role in many engineering fileds,such as mines,bridges,water conservancy and so on.The performance of bolt support is closely related to the safety and reliability of the whole engineering system.Therefore,plenty of researchers work hard to improve the anchor bolt quality of the study.In this thesis,the characteristic matrix is composed of time frequency characteristic of acceleration signals of different anchor types.Principle component analysis(PCA)is used to reduce the feature matrix,and the support vector machine(SVM)is introduced to distinguish the recognition of different types of bolts.The nuclear parameters of SVM and classification accuracy are inextricably linked,determined by particle swarm optimization(PSO).The specific contents of this thesis are as follows:(1)The basic principle of the stress wave reflection method is introduced in detail,and the feasibility of the method used for classification and identification of bolt is illustrated.(2)DH5923 dynamic signal testing and analysis system is used to collect the acceleration signals for different types of bolts,which providing data support for classification and identification of bolts.Noise reduction processing is used for the acceleration signals collected and the initial feature matrix is composed of the time domain and frequency domain characteristic extracted from the denoised signal.(3)The basic principle of PCA is described,and the feature matrix is reduced by PCA.The contribution rate of each principle component is calculated and arranged in a descending order.The results show that PCA can save the original information as much as possible while achieving the compression of the original data and simplifying the processing problem.(4)After the dimension reduction of PCA,the first principal components with larger cumulative contribution rate are extracted as the input of SVM.The classification order of BT-SVM method is determined,and PSO is used to find the optimal parameter value of SVM quickly and accurately.The results show that the method can be used to classify and identify different types of anchor bolt system effectively.(5)The inertia weight and learning factor of PSO are improved,and signal peak function Sphere and multi peak function Ackley are used in testing,The simulation results show that the accuracy precision and convergence speed are improved.The improved PSO is applied to the classification and identification of bolts,and the results show that the convergence speed is improved obviously while the accuracy is guaranteed.