基于人在回路的大型星载平面天线形面精度调控

As for large-scale spaceborne planar antennas, surface accuracy determines their electrical performance, and how to efficiently and accurately conduct surface accuracy adjustment is a crucial challenge in manufacturing. To address this problem, a novel method based on human in the loop for large-scale spaceborne planar antennas surface accuracy optimization and control is proposed. The immediate feedback of experts and prior engineering experience are integrated into the optimization and control of antennas surface accuracy to enhance efficiency while achieving optimal adjustment. Firstly, a mathematical formulation of human knowledge based on user belief reinforcement and a preference learning model is constructed. Based on traditional Bayesian optimization, expert knowledge is organically incorporated into the Bayesian optimization process. Subsequently, the human in the loop Bayesian optimization model is constructed, and the immediate feedback of experts is seamlessly incorporated into the optimization process. Then, based on the human in the loop Bayesian optimization algorithm, with minimization of surface accuracy as the objective function and adjustable rod lengths as optimization variables, the craftsmen's expertise is integrated and the surface accuracy control of large-scale spaceborne planar antennas is constructed. Finally, numerical case analysis and practical assembly experiments on a scaled antenna prototype are conducted to validate the proposed method. The results indicated that the proposed approach significantly accelerated optimization convergence speed and greatly enhanced surface accuracy and control efficiency.