Font Size: a A A

The Study Of Shaking Table's Servo Hydraulic System Based On Artificial Neutral Network

Posted on:2010-05-06Degree:MasterType:Thesis
Country:ChinaCandidate:L Q XiaFull Text:PDF
GTID:2198330338482350Subject:Structural engineering
Abstract/Summary:PDF Full Text Request
The structural seismic test is a necessary step in research of quakeproof and disaster preventing. With the development of the control theory and computer technology, the tend of seismic test methods is to be large-scaled, digitized, accurated and intelligent. Application of new testing methods on Quasi-static testing, Pseudo-dynamic testing and the Shaking table test is the hotspot of the research of the seismic testing of structures, especially the shaking table test based on intelligent control methods, which even became the leading-edge of the structural seismic test. The idiographic work is reconstructing the servo hydraulic control system of the shaking table, which is important to the research of the seismic simulated shaking table test method.In this paper, the basic control methods of electro-hydraulic shaking table and the trend of use of intelligent control technology are analysed, and it is focusing on application of iterative learning control algorithm for neural networks in the shaking table motion acceleration waveform reproduction technology. For the need of Structure Laboratory of Hunan University shaking table acceleration waveform reproduction, based on the PID control loop, the application of neural networks on-line iterative learning control and multi-parameter control theory build an external acceleration loop to form the double-closed-loop control. One use of MTS control system which is PID control method of the existing. The outer loop by configuring the acceleration sensors, velocity sensors, displacement sensors and the preparation of the waveform conversion and loading process, using neural network iterative learning control algorithm to achieve an external closed-loop control to modify the PID parameters.This outer loop through the RBF network to recognize the information of the sensitivity on the variate of control input and feedback, and online train the RBF network. A certain algorithm was improved by adding or deleting nodes in hidden layer neural network to accelerate the speed of online learning algorithm. This article also attempts to genetic algorithm to optimize the neural network initial parameters, increasing the network's ability of global optimization. The final adoption of genetic algorithm optimization neural network initial weights and parameters, and the RBF neural network and BP neural network together constitute the parameters of the outer-conditioning systems in closed-loop PID algorithm, explorations tests show that the algorithm has good adaptability.
Keywords/Search Tags:Shaking table, Electro-hydraulic servo system, Neural network, On-line iterative
PDF Full Text Request
Related items