Intelligent Motion Control for Electromechanical Servos Using Evolutionary Learning and Adaptation Mechanisms

Electromechanical servos integrate servomotors, power converters, sensors, integrated circuits, and controllers. In conventional applications, analog and digital proportional-integral-derivative controllers are widely used, and electric servomotors can be straightforwardly controlled by making use of the electromagnetic and energy conversion phenomena. High-performance servos must be designed to achieve specified criteria and standards in expanded operating envelopes. These requirements can be guaranteed implementing advanced control algorithms designed applying novel control methods. This paper stresses the need to design intelligent systems to solve the intelligent motion control problem. It must be emphasized that in general, motion control cannot be viewed merely as robust tracking control because in addition to accuracy, stability and disturbance attenuation, other criteria (e.g., efficiency, reliability, noise, vibration, and electromagnetic interference) and tasks (e.g., decision making, intelligence, diagnostics, and health monitoring) must be performed. Intelligent motion control can be achieved by implementing learning and adaptation mechanisms (through evolutionary learning, reconfiguration, rescheduling, tuning, and optimization) with the ultimate objective being to attain the optimal overall performance. The performance (objective) functional is evaluated using measured control, reference, and output vectors that are the system performance variables. It is demonstrated that using the evolutionary learning and adaptation, the intelligent motion control problem can be solved without linguistic or mathematical models of electromechanical servos. In particular, unfalsified and premium control laws are designed, and experimental results are documented.