Font Size: a A A

The Research Of WNN Algorithms And Its Application In Flexible Redundant Robots' Vibration Curbing Control

Posted on:2002-06-08Degree:MasterType:Thesis
Country:ChinaCandidate:T ChengFull Text:PDF
GTID:2168360032455915Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The thesis focuses on the study of wavelet neural networks and the its application in flexible and redundant manipulators. Wavelet theory is a kind of new science. It has gotten success in applications of signal processing, image processing, pattern identification and target classification. But, the application potential of wavelet theory in controlling has not been exerted. This thesis uses artificial neural networks which is very popular in the controlling field, using its predominanee---non-linear function approaching ability together with its good local analyzing ability, and has probed and constructed wavelet self-adapt neural networks. After comparing with ordinary BP networks, least-square method and wavelet RBF networks, I have proved that the new wavelet self-adapt network has good studying performance and self-adapt ability which can been gotten by other neural network. And contraposing the defect of slow convergence and bad convergence impact of neural network?studying arithmetic, we have had optimized studying of studying arithmetic, and have deduced wavelet neural networks self-adapt algorithms. On the basis of constructive algorithms and destructive algorithms, the thesis puts forward a kind of method which is called dichotomy algorithm. In actual application, we use wavelet self-adapt neural networks in flexible and redundant manipulators?check-vibration. The thesis has designed a flexible and redundant manipulators?controlling system which is based on wavelet neural network, has given software and hardware implementation, has chose simple WYD direct current displacement sensor. It has used also the sensor to design terminal scaling devices of flexible and redundant manipulators, and has done simulation experiment on flexible and redundant manipulators in different conditions. The simulation result indicates that the algorithms have indeed excellent vibration curbing ability, manipulators?flexible manipulators?vibration has been well checked. This indicates that itl be to advantage if we put high-grade controlling strategies together. This thesis will be valuable in closely studying flexible and redundant manipulators.
Keywords/Search Tags:flexible and redundant manipulators, artificial neural networks, wavelet theory, self-adapt controlling, wavelet neural networks
PDF Full Text Request
Related items