Research On Active Vibration Control Of Beams By Neural Network

The range of Flexibility structure vibration is great and during long time,which affect precision of instrument and life of structure. So the technology of vibration control of this kind structure is abroad attention by people in the field of vibration engineer. This dissertation deals with identification and vibration control of composite cantilever beam.Neural control method appling to vibration control of cantilever beam and building a real-time vibration active-control system based on Windowsare the target , Firstly,an composite beam was fabricated ,and its experimental modal frequencies were measured by LMS.Secondly,a vibration test system based on Matlab/XPC was builded.it was used to excite the beam by the white noise and the response was collected.Thirdly,a C++ class of BP neural network algorithm was programed,on this base an off-line neural network NARMA system model and a indirect model reference adaptive neural controller were bulided by the feedforward neural network. fourthly, A data acquisition program by VC was finished in the circumstance of Windows based on PCL1800.Lastly,The neural controller and the data acquisition program make up of the neural network real-time vibration active-control system.The experiment result of vibration control of beam testified not only the validity of the neural controller but also the practicability of the active vibration control system. The reseach described in this paper is financially supported by National Natural Science Foundation(50135030) and Jiangsu Natural Science Foundation(BK2002090).