中国空间科学技术
中國空間科學技術
중국공간과학기술
CHINESE SPACE SCIENCE AND TECHNOLOGY
2014年
6期
51-56,64
,共7页
人工神经网络%格栅反射器%型面精度%优化设计%天线%卫星通信
人工神經網絡%格柵反射器%型麵精度%優化設計%天線%衛星通信
인공신경망락%격책반사기%형면정도%우화설계%천선%위성통신
Artificial neural network%Grating reflector%Surface accuracy%Optimization design%Antenna%Satellite communications
基于人工神经网络的预测方法，利用数量适当的有限元计算结果，建立人工神经网络模型，对反射器型面进行精度预测分析，得到在最小型面精度结果下的结构设计参数。计算结果显示训练好的神经网络模型能够较精确地预测格栅反射器的型面精度，节省计算时间，并且以型面精度最小为准则进行参数分析，能够指导反射器的结构设计。
基于人工神經網絡的預測方法，利用數量適噹的有限元計算結果，建立人工神經網絡模型，對反射器型麵進行精度預測分析，得到在最小型麵精度結果下的結構設計參數。計算結果顯示訓練好的神經網絡模型能夠較精確地預測格柵反射器的型麵精度，節省計算時間，併且以型麵精度最小為準則進行參數分析，能夠指導反射器的結構設計。
기우인공신경망락적예측방법，이용수량괄당적유한원계산결과，건립인공신경망락모형，대반사기형면진행정도예측분석，득도재최소형면정도결과하적결구설계삼수。계산결과현시훈련호적신경망락모형능구교정학지예측격책반사기적형면정도，절성계산시간，병차이형면정도최소위준칙진행삼수분석，능구지도반사기적결구설계。
The precision of reflector antenna on orbit is an important design specification, and the reasonable structure design can improve the reflector precision. By taking full advantages of the high strength, high modulus, low coefficient of thermal expansion of the composite material,and the surface precision of the antenna could be improved significantly. By using the artificial neural networks'highly nonlinear mapping capability,and based on the principle of the minimum RMS,the design parameters were optimized.The calculation results indicate that the computing time can be effectively saved. The results show that the optimization method is reasonable, and the structure design meets the design requirements. The design scheme and analytical methods provide a direction for the structure design of the reflector.
