Chaotic Identification Of Maglev Vibration Measurement System Based On Neural Network

In the field of vibration measurement are the most frequently used relative measurement.In this paper,a vibration measuring system with magnetic levitation ball as measuring element is designed.It belongs to the absolute vibration measurement.It has the advantages of no mechanical connection,no mechanical friction loss and high accuracy compared with the traditional measurement.Because the magnetic levitation system is a nonlinear system,the change of initial conditions will lead to chaotic motion,which is not conducive to the accuracy of the measurement results and the stability of the system.Therefore,it is particularly important to identify the chaotic state of the magnetic levitation vibration measurement system.In this paper,the chaotic state of the magnetic levitation vibration measurement system is identified by the BP neural network and the improved neural network based on genetic algorithm.The components and working principles of the magnetic levitation vibration measurement system are introduced.The magnetic suspension vibration measurement system model and control circuit are given,which theoretically proves the feasibility and superiority of the magnetic suspension absolute measurement.According to the output characteristics of photoelectric displacement sensor,the control circuit and the motion equation of magnetic levitation ball,a Simulink model of magnetic suspension vibration measuring system is established.The other parameters remain unchanged,and the output waveform is simulated by changing the values of zeros and poles.The experiment shows that under certain conditions,there are four states of the system: double attractor,single attractor,periodic motion and stability.The chaotic characteristics of vibration process are systematically analyzed,and chaos model is established.The chaotic characteristics are verified by simulation,and the boundary condition of chaotic motion is summarized.The BP neural network algorithm is derived,and the shortcomings of the BP neural network are summarized,and the model of the GA-BP neural network is established.Selecting the eigenvalues,The output waveform is obtained by simulation.The eigenvalues are extracted as network training and test samples.BP neural network is used to train data.Through the process of decreasing error,a good network is obtained.Finally,the measured data is substituted for the identification of chaotic state.The experimental results show that the network can accurately recognize the chaotic motion state of the maglev system.Comparing the recognition results of the two algorithms,it is found that the GA-BP neural network has faster convergence rate and higher accuracy than the BP neural network.