Machine learning-based active control for lightweight antenna with force density method and nested genetic algorithm

Abstract

This work proposes the integration of a tensegrity structure composed of struts and cables into an antenna, thus achieving a lightweight and optimized design for the large-scale system. Active control is introduced to ensure the deformation accuracy of the cable-strut antenna, while a shape sensing method based on force density information is adopted to obtain the deformation field of actual wind cases from monitoring the cable forces. Numerical simulations are performed to verify the accuracy of the method, and a multi-objective optimization framework is employed to determine the optimal positions and control of actuators for active control. A machine learning-based framework is proposed to achieve rapid response of the active control system under actual wind cases. The framework is composed of a two-stage control system that combines the shape sensing method with the multi-objective optimization approach. The machine learning framework establishes a quick response system that monitors changes in the cable forces, thus ensuring structural accuracy control under wind load without structural analysis.