CMAC Learning Controller for Servo Control of High Precision Machine Tools

There is an increasing number applications of high precision motion control systems in manufacturing, i.e. ultra-precision machining, assembly of small components and micro devices. Typical required positioning accuracies are in the range of 2.5¿m to 0.25¿m (1/10,000 in to 1/100,000 in). The trend is to require even higher accuracies. It is very difficult to assure such accuracies due to many factors affecting the precision of motion, such as friction and backlash in the drive system. The standard proportional-integral-derivative (PID) type servo control algorithms are not capable of delivering the desired precision under the influence of friction and backlash. A learning control algorithm based on Cerebellar Model Articulation Controller (CMAC) is studied for servo motion control under the presence of friction and backlash with the ultra-precision machine tool applications in mind. The CMAC learning controller is conceptually simpler, has faster learning convergence, and more practical than the well known feedforward backpropagation artificial neural network architectures. The CMAC control algorithm is implemented using C-language on an IBM-PC compatible.