ALGORITHMS FOR THE SYNTHESIS OF ADAPTIVE CONTROL SYSTEMS BASED ON THE SPEED-GRADIENT METHOD

This paper discusses the synthesis algorithms for adaptive control systems based on the speed-gradient method. Adaptive control systems with implicit reference and adjustable models are synthesized using speed-gradient techniques, which reduce the requirements for the main control loop structure and the completeness of measurement data. Stable adaptive decentralized control algorithms are developed for a class of interconnected systems with nonlinear local dynamics and uncertainties, ensuring the stability of individual subsystems and the overall system while accounting for their interactions. To incorporate inter-subsystem interactions into the overall control law, an adaptation algorithm based on the speed-gradient method is introduced. A synthesis procedure is proposed for control systems using multilayer neural networks based on the speed-gradient method, which guarantees system stabilization under bounded external disturbances that are not directly measurable. Including the state vector of the controlled object in the neural network training functional ensures consistency between network parameter tuning and control of the dynamic system. Stable synthesis algorithms for adaptive neural network control systems are developed, ensuring the fulfillment of a limiting condition that guarantees achievement of the target control objective using the analytical design method for aggregated regulators. Based on the proposed algorithms, an adaptive control system for the tape-drawing process in conveyors is introduced. This adaptive system, grounded in the speed-gradient method, stabilizes process operation modes and improves overall efficiency.