Study On The Bolt Nondestructive Testing Based On Probabilistic Neural Network

With the advantages of simple construction and low cost,Rock Bolting technology has been more and more widely used in tunnels,mines,mountain reinforcement and other construction projects.The quality of the support’s performance is very important for the safety and reliability of the supported project.The research on the non-destructive detection of the quality and performance of bolt anchoring system has been one of the key research directions of scholars at home and abroad.In this paper,the frequency response function of different anchor types is studied,using Principal Component Analysis(PCA)to reduce the dimension of the measured frequency response function set,the improved Particle Swarm Optimization is used to optimize the smoothing factor of the Probabilistic Neural Networks(PNN),and finally the Probabilistic Neural Networks is used to identify the damage of different anchor types in this test.The specific contents of this paper are as follows.(1)According to the characteristics of bolt anchor system,a non-destructive testing system was designed and built.The corresponding frequency response function of different damage type anchor system was obtained by collecting excitation and corresponding signals.(2)Studying the principle of the stress wave reflection method in the bolt anchoring system is introduced,introducing the principle of the frequency response function and the feasibility as the characteristic parameter of the bolt anchoring system.The basic principles and the dimension reduction methods of Principal Component Analysis are introduced,and using PCA to realize the measured frequency response function set.(3)Describing the probabilistic neural network,and the principal components are used as the input of probabilistic neural network,finding the damage of the measured bolt anchoring system,and analyzing the accuracy of damage identification.Discussing the factors which affect the recognition accuracy of probabilistic neural network from the aspects of the setting of smoothing factors and the selection of training data.The results show that the probabilistic neural network can meet the requirements of the classification of the damage of the anchorage system.(4)Optimize the standard PSO algorithm from two aspects.On the one hand,it improves the problem of the out of bounds,which will reducing the probability of PSO falling into local optimum at the boundary.Using the classical function to verify the feasibility of the feasibility of the improved particle swarm optimization method.Using the improved PSO algorithm to optimize the smoothing factor of PNN,and then it is compared with the traditional particle swarm optimization algorithm and applied to the bolt anchoring system to confirm the feasibility in practical.The results show that the improved PSO algorithm has the characteristics of fast convergence and high precision,which improves the recognition rate of the damage classification of the anchor system by the probabilistic neural network.