Minimization of Cable-Net Reflector Shape Error by Target-Approaching and Procedural-Learning Method

Abstract

Cable length adjustment is a critical step to minimize the shape error of cable-net reflectors. The adjustment requires tedious repetitive work, but the surface accuracy is often unsatisfactory because of inevitable operating error and modeling inaccuracy. To solve this problem, this work proposes a target-approaching and procedural-learning method. At the initial stage of the adjustment process, the target-approaching method is proposed to ensure the shape error is improved and that adjustment data are generated. Then an online prediction model is built by sensitivity analysis and least-squares support vector machine. On this basis, a multidimensional forward-backward algorithm is applied to calculate the optimal adjustment amounts in order to achieve fast convergence of adjustment accuracy. Finally, the proposed method is verified by both numerical simulation and prototype experimentation. The results show that the proposed method can well reflect the adjustment characteristics of cable-net reflectors, effectively improve shape accuracy, and greatly shorten adjustment time.