Modal Parameter Identification Of Vibration Signal Based On Convolutional Neural Network

The modal parameter identification of vibration signal is the core of vibration detection technology in large-scale structural health monitoring and structural damage detection field,and it becomes an effective and practical method in fault diagnosis and safety detection.The accurate extraction of modal parameters directly affects the health monitoring of engineering structures.At present,the commonly modal parameter extraction algorithms are divided into time domain method,frequency domain method,time domain method and modern signal processing technology.However,these methods are sensitive to noise and poor robustness.Therefore,this paper is on the basis of the modern signal processing technology,modal parameter identification of vibration signal based on convolutional neural network was proposed.Firstly,The forward propagation process and the back propagation process of the existing convolution neural network are deduced theoretically.The relationship between the input layer,convolution layer,sampling layer and output layer of convolutional neural network is studied;On the basis of the existing convolution neural network,the CNN applied to two-dimensional image processing was changed into the CNN to deal with one-dimensional signal.The input layer was changed into the vibration signal set of modal parameters to be extracted,and the intermediate layer was changed into several one-dimensional convolution layers,sampled layers,and output layer was the set of N-order modal parameters corresponding to the signal.Then,in the error evaluation,the network calculation result(N-order modal parameter set)was reconstructed by the vibration signals.Finally the squared sum of the difference between the reconstructed signal and the input signal was taken as the network learning error,which makes the network become an unsupervised learning network,and avoids the ordering problem of modal parameters extraction.Secondly,Aiming at the problem that the BP algorithm of the improved convolutional neural network is easy to fall into the minimum and generalization,particle swarm optimization algorithm instead of reverse propagation of the BP algorithm for network weight and offset update was proposed.The vibration signal is divided into test set and training set,the test set was taken into particle swarm optimization to improve the volume and offset of the network in the convolutional neural network,After completing the parameter update,the training set was taken to the network to extract the modal parameters of the vibration signal.Finally,the simulation experiment of vibration signal modal parameter extraction is carried out by BP learning algorithm with improved convolution neural network to study the anti-noise of the modified convolution neural network.The simulation results show that the improved convolution neural network improves the antinoise performance,then,the existing modal recognition method,the BP algorithm of the improved convolutional neural network and the particle swarm optimization improved convolutional neural network are simulated and compared.the particle swarm optimization improved convolutional neural network is improved to improve the generalization performance of the convolution neural network and has strong noise resistance.Above all,on the basis of the existing convolution neural network,this paper studies the convolutional neural network method for vibration signal modal parameter extraction and applies this algorithm in actual bridge modal parameters extraction,the algorithm is verified by the practical application of noise resistance.