Intellectual Nondestructive Testing System For Hardness Of Artillery's Parts

Based on analysis of the present situation of hardness detection,this paper improves hardness testing system, which is based on the FPGA devices and the period signal of standard frequency 100MHz,to measure resonated frequency by an equal precision frequency measuring technology.Aimed at nonlinear mapped relation between the hardness of artillery's parts and sonic parameters, this paper applies ANN network to sonic testing,and put to use modified particle swarm optimization (MPSO) to optimize the framework of BP Neural Networks and initial weight,MPSO algorithm translate from singleness into multitudinous different from PSO,so the precision of the researching is improved effectively and it balances part and whole search and can converge to optimal solution.Also it conquers the bug of BP Networks and come true the function of training and forecasting different parts'hardness.This paper proposed classificatory method based on sonic testing and probabilistic neural network(PNN),it realizes classification between different hardness. This paper adopts equal precision frequency measuring technology,which predigests design and enhance accuracy and reliability.Simultaneously optimized neural network algorithm make the system possess brainpower, which possesses gradual fitness as well as high convergence speed.This paper also exerts PNN to simulate hardness with sonic parameters together,to compare with examination instrument, whose precision is more better than instrument.Intellectual nondestructive testing system devised can advance overall performance,to a certain extent,which also has academic significance and practical value for national defence modern construction.