Research On Active Vibration Control By Neural Network

Vibration control is an important part of vibration engineering. Its essence is the use of certain technical means to make the vibration level is accorded with the design requirements. Vibration control includes the use of vibration and vibration suppression. The purpose of active vibration control is to control adverse vibration. The process of active vibration control can be divided into two steps. First, the control law was designed according to the collected vibration signals. And then, the actuator output was controlled to decrease or eliminate the vibration of system.Based on unique nonlinear approximation ability, neural network has a huge potential for identification. In this paper, a cantilever beam vibration system was identified by neural network. The neural network controller was designed for the active vibration control of the cantilever beam vibration system.In this paper, a cantilever beam was be used as the object of study. Firstly, the cantilever beam model was established and the control law was designed. And then, the active vibration control strategy was established in dSPACE real-time system which based on neural network. Finally, the cantilever beam was controlled by the use of neural network. This thesis started from the purpose of active vibration control and the new methods of domestic and foreign. The contents of this paper included the following sections:mathematical model of the cantilever beam was established. The cantilever beam was simplified to three degree of freedom system, and calculated the differential equations of motion. The natural frequencies were calculated respectively by the numerical calculation and analysis software of ANSYS for the purposes of neural network reference and active vibration control. The cantilever beam was identified by NNSYSID toolbox with the signal which was obtained from experiment platform. At the same time, the cantilever beam was simulated by neural network in SIMULINK. The experimental block diagram neural network control algorithm was established and downloaded in dSPACE real-time system. The vibration control of cantilever beam was validated by experiment, and the results were analyzed.Through the cantilever beam vibration system was identified by NNSYSID toolbox, indicated that the application NNSYSID toolbox can easily select the reference input, sampling frequency, the estimated model and the validate model, set initialize variables and parameters to achieve the identification of the vibration system. The simulation results showed that the neural network controller can control the actual output of the vibration system, made tracking the output of the reference model. The on-line control the cantilever beam vibration system which was based on the dSPACE real-time system showed that:: cantilever incentives frequency from the inherent frequency more far, the anti-vibration effects the better; the better control effect can be obtained through the high sampling frequency.In this paper, neural network was applied to the active vibration control of a cantilever beam. Achieved the expected control requirements, at the same time, also confirmed that the neural network in the active vibration control is effective. By use of dSPACE real-time control systematic platform, researchers can easily develop control algorithm and test it in this platform, even correct and design it many times.