Optimal Sensors Placement Based On Effective Independence-Entropy Energy Fusion Method

The core content of sensor optimization layout is to arrange a limited number of sensors in the optimal position of the structure,so as to achieve the purpose of more comprehensive information data collected by testing and more accurate identification of modal parameters.In this paper,based on the Effective Independence Method of optimal placement of the sensor methods only consider the layout and point to the target modal vector between linear contribution,ignoring the point of Energy contribution,leading to the problem of shortage of the Method of noise resistance,this paper proposes a Fusion algorithm based on Entropy can effectively independent-(Effective Independence-Entropy Energy Fusion Method)Method for optimal placement of the sensor.On the premise of maintaining linear independence among the modes,the proposed method can not only make the measuring points located at larger vibration energy,but also obtain more modal information.The main research contents of this paper are as follows:1.To solve the problem of insufficient anti-noise of the effective independent method,an optimal sensor placement method based on the effective independent-entropy energy fusion algorithm is proposed.In this method,the information entropy of nodes is fused with the modal strain energy coefficient of nodes,and the effective independent components are corrected twice.On the one hand,the proposed method was compared with the effective independent method and the effective independent-improved modal strain energy method.Five evaluation criteria were used to evaluate the optimal placement results of the three methods.The results show that the effective independent-entropy energy fusion algorithm is superior to the other two algorithms.On the other hand,with the increase of the added noise intensity,the five evaluation indexes of the effective independent-entropy fusion algorithm are stable,and the number of modal conditions representing the anti-noise ability also remains basically unchanged,indicating that the proposed method has a good anti-noise performance.2.Mode number and the number of sensors to determine the influence the result of the optimal placement of the sensor,this paper puts forward a covariance matrix of information increment matrix to determine the target number of modal method,and based on the Fisher information matrix trace,and 2-norm,comparing the two methods and combined with MAC guidelines to further determine the number of sensors.The results show that when the maximum value of diagonal element is taken as the evaluation index by MAC,the optimal sensor layout result is obtained by the number of modes determined by the information increment matrix based on covariance matrix,which indicates that the method proposed in this paper is feasible.3.This paper combined with the depth study and finite element model updating is put forward based on the finite element model of the depth of the residual network correction method,this method will acceleration data node of Fourier transform(FFT)spectrum is generated after further spectrum data can be converted to grayscale figure as input features,using the depth of the residual network based on MATLAB environment(Res Net)for training.Compared with the finite element model correction method based on genetic algorithm,the results show that the error of finite element model correction using deep residual network is less than that of genetic algorithm,which verifies the feasibility of this method.4.Experimental verification and analysis.The acceleration signals of the nodes of the frame structure are picked up and the modal analysis of the frame structure is carried out by the enhanced frequency domain decomposition method(EFDD).The experimental scheme based on different methods,different number of modes and different number of sensors is set up.The method proposed in this paper is comprehensively verified by five evaluation criteria.The results show that when the optimal position of the frame structure is determined by different sensor optimal placement methods,the first 40 modes and 12 sensors' effective independent-entropy fusion algorithm have the best identification effect on the modal parameters.At the same time,the feasibility of the method to establish the number of modes and the improved effective independent method is also verified.