Analog Signal Detection And Application Of A Kind Of Mechanical Rotor System

Stochastic resonance is a phenomenon that the cooperative action of noise,signal and nonlinear system is generated under the joint excitation of noise,signal and nonlinear system,which makes the Brownian particle oscillate and leads to the enhancement of the output of nonlinear system.Stochastic resonance widely exists in nonlinear systems and is applied in various scientific fields.With the improvement of productivity and the sustainable development mode of green and low carbon,the faults of mechanical rotor systems such as wind turbines in the process of operation can not be underestimated,which puts forward high standards and strict requirements for the early weak fault signal identification of the system.Therefore,the extraction and detection of weak fault signals based on stochastic resonance method has practical application value.The main research work of this paper is as follows:1.The classical stochastic resonance system is theoretically derived,and the image is drawn according to the potential function equation.The influence of each parameter in the potential function on the performance can be preliminarily determined.Under the adiabatic condition,the two evaluation indicators,mean first passage time(MFPT)and signal to noise ratio(SNR),are derived to determine how to adjust the parameters to optimize the system.Because the signal amplification function of asymmetric system is better than that of symmetric system,the theoretical derivation and optimization of classical asymmetric system are carried out.2.Based on the classical asymmetric model and the segmented method,the general asymmetric stochastic resonance system is segmented,and a segmented potential function model is established.First,under the adiabatic condition,the expressions of the mean first passage time and the signal-to-noise ratio of the general asymmetric stochastic resonance system and the piecewise asymmetric stochastic resonance system are derived respectively.The influence of the analytical expression of the signal-tonoise ratio varying with the noise intensity under different parameters is studied.In addition,in the application of digital signal processing,the fourth-order Runge-Kutta algorithm is adopted.Based on the random weight particle swarm optimization algorithm,the signal-to-noise ratio is used as the performance evaluation index of the test system to compare the two systems.The simulation and experimental results show the superiority of the segmented asymmetric stochastic resonance system.3.Based on the physical model of horizontal platform system(HPS),a new potential function model is established.The evaluation indexes of the system are derived,and the effects of platform inertia moment,damping factor,accelerometer scale coefficient and input periodic signal amplitude on the system performance are studied respectively.In order to verify the actual availability of the system,simulation and experimental verification are carried out.4.In order to efficiently identify a large number of fault signals,based on deep learning,a model combining wavelet scattering network and bidirectional long and short time memory artificial neural network is used to classify a large number of unknown weak fault signals.The experimental results show that the accuracy and stability of the signal classification results using the wavelet scattering network in the bidirectional long and short time memory artificial neural network is better than that using the feature classification toolbox of matlab in the bidirectional long and short time memory artificial neural network,which proves the superiority of the classification method.