Feedback Control by Online Learning an Inverse Model

• Published in: IEEE transaction on neural networks and learning systems Vol. 23; no. 10; pp. 1637 - 1648
• Main Authors: Waegeman, T., Wyffels, Francis, Schrauwen, B.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.10.2012; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive control; Algorithms; Applied sciences; Artificial intelligence; Artificial neural networks; Asymptotic stability; Computer science; control theory; systems; Computer Simulation; Connectionism. Neural networks; Distance learning; Exact sciences and technology; Feedback; Feedback control; Fundamental areas of phenomenology (including applications); heating tank; Inverse; inverted pendulum; Learning; Learning and adaptive systems; Mathematical models; Models, Statistical; neural network; Neural networks; Neural Networks (Computer); Neurons; Nonlinear dynamics; On-line systems; Online; Pattern Recognition, Automated - methods; Physics; pitch control; reservoir computing; Reservoirs; Solid dynamics (ballistics, collision, multibody system, stabilization...); Solid mechanics; Stability analysis; Studies; Training; Water heating; Inversed pendulum; Error estimation; Non linear control; Solid dynamic; Feedback regulation; Flight control; Adaptive control; Linear control; Modeling; Stirred tank reactor; reservoir computing; Active system; heating tank; Vibration control; pitch control; Learning algorithm; Storage tank; Intelligent control; Recurrent neural nets; Dynamic characteristic; Balancing; feedback control; Aerodynamics; Neural network; Experimental study; Closed feedback; Heat source; Heating; inverted pendulum; Non linear effect; Artificial intelligence
• ISSN: 2162-237X; 2162-2388; 2162-2388
• DOI: 10.1109/TNNLS.2012.2208655

A model, predictor, or error estimator is often used by a feedback controller to control a plant. Creating such a model is difficult when the plant exhibits nonlinear behavior. In this paper, a novel online learning control framework is proposed that does not require explicit knowledge about the plant. This framework uses two learning modules, one for creating an inverse model, and the other for actually controlling the plant. Except for their inputs, they are identical. The inverse model learns by the exploration performed by the not yet fully trained controller, while the actual controller is based on the currently learned model. The proposed framework allows fast online learning of an accurate controller. The controller can be applied on a broad range of tasks with different dynamic characteristics. We validate this claim by applying our control framework on several control tasks: 1) the heating tank problem (slow nonlinear dynamics); 2) flight pitch control (slow linear dynamics); and 3) the balancing problem of a double inverted pendulum (fast linear and nonlinear dynamics). The results of these experiments show that fast learning and accurate control can be achieved. Furthermore, a comparison is made with some classical control approaches, and observations concerning convergence and stability are made.