Inexact algorithms for correction of errors due to flexibility of dynamic structures

In the development of an intelligent controller to manage the flexibility of dynamic systems the combination of fuzzy logic and neural networks, inexact algorithms, can be applied in constructing an intelligent kernel. Using human operator´s experience and knowledge, correcting actions can be translated into groups of network connections and sets of control rules. In order to emulate the distortion of an end-effector caused by flexible mechanical structures, a spring mounted pen (SMP) is designed and engaged in the experiments. A machine vision station is constructed in the test rig to monitor the results of the SMP following process. By analysing the processing errors as well as feeding back on-line information to the intelligent controller, the problems of inaccuracy due to flexibility of dynamic systems can be overcome. The method developed is essentially trying to avoid using very complex sensors to monitor all the system rind environment factors. Through an on-line self-learning process-the intelligent kernel compares the difference between the desired pattern and the resultant pattern to make the appropriate compensation