Study On System Identification For The Magnetic Suspension Active Vibration Isolation System

Mute capability of the submarine is an important indicator of submarine combat capability. Isolation technology can effectively reduce the mechanical vibration, improve Mute capability of the submarine. Floating raft system has the features of "tuning effect","Mass Effect","mixed neutralized effect", widely used in the submarine, which reducing the radiated noise and vibration and having a weakening affect for the high-frequency signals. However, as a passive vibration isolation system, floating raft system plays the role of amplification for vibration frequency that is less than or equal to the natural frequency of the vibration isolation system. Active vibration isolation system has the ability of adjusting the stiffness and damping and vibration isolation parameters dynamically, overcoming the defects of the passive vibration isolation. Active and passive combined technology, achieving full frequency vibration isolation, is the trend and direction of the isolation field. Magnetic Bearing technology has some characteristics of no contact, no lubrication and long life. So, the magnetic suspension vibration isolation applied to the floating raft system is of great significance. Magnetic active vibration isolation system is a strong coupling, nonlinear complex systems, and the traditional modeling approach is difficult to solve problem of system dynamic modeling. This article has modeled magnetic active vibration isolation system dynamic modeling by system identification methods.First, compared with several main forms of least squares method in model accuracy, convergence properties, the amount of storage and calculation time, the recursive least squares method is determined to system identification for magnetic single-layer vibration isolation system in the time domain and frequency domain. Acceleration sensor signals acquisition to the corresponding location of the magnetic vibration isolation is the objective, this paper comparing the results of the time domain with frequency domain in terms of fitting degree, pole and step response. The comparison results show that:compared to the frequency domain identification, time domain identification has better accuracy of the identification model, slow the step response, long convergence time.Secondly, the paper makes a brief introduction to the technology of magnetic levitation bearing, and analysises the characteristics of the magnetic levitation vibration isolators as well as structure. The theory electromagnetic force formula of differential magnetic levitation vibration isolation is deduced in the case of ignoring flux, leakage, and magnetic saturation. On this basis, the theory electromagnetic force is amended by fitting actual static electromagnetic force. In order to verify the invariance of the system, three kinds of different voltage signal were entered into system. The results show that system identification models consistent high and magnetic active isolation vibration is time-invariant system.Finally, the active isolation vibration for floating raft system is motivated by entering the random voltage signal, collected acceleration signals from floating raft. According to input and output signals, the paper take recursive least squares method to identify the system model. In order to ensure system accuracy, stability, transient response characteristics, it is necessary to comprehensively consider of the fitting degree, the poles and zeros, the step response. Applied the transfer function of the identification model to the feed-forward control system, we take advantage of simulation and experiment to verify the validity of the model identified by recursive least squares method. Simulation and experimental results show that:feed-forward control algorithms on magnetic active vibration isolation platform has better vibration isolation effect in the low frequency, especially near the resonance region of the isolation effect is more obvious, improved15dB-25dB, which verify the recursive least squares method for model identification of the effectiveness of the magnetic levitation system.