论文部分内容阅读
航天继电器长期处于贮存环境,为保证其各阶段始终保持在备用激活状态,必须对继电器的贮存寿命进行预测。本文将因子分析法和回归分析法引入到表征触点电接触可靠性的重要参数——接触电阻的转换中,将25台继电器样品的200对触点在125℃下的接触压降和释放电压双参数数据交叉分为4组进行处理,分析两者与接触电阻的关系,建立函数链神经网络,对接触电阻进行动态预测,进而得到继电器的贮存寿命。分析神经网络预测的整体误差,用92℃的数据对该方法进行检验,得出神经网络的预测误差低于3.5%,证实了统计方法和函数链神经网络的适用性。
Aerospace relays long-term storage environment, in order to ensure that all stages of its standby has been activated, you must predict the shelf life of the relay. In this paper, factor analysis and regression analysis are introduced to characterize the reliability of contact electrical contact important parameters - contact resistance conversion, the 25 relay samples 200 pairs of contacts at 125 ℃ contact pressure drop and release voltage The two-parameter data crossover is divided into four groups for processing. The relationship between the two and the contact resistance is analyzed. Functional chain neural network is established, and the contact resistance is dynamically predicted, and the storage life of the relay is obtained. The whole error of neural network prediction is analyzed. The method is tested with the data of 92 ℃. The prediction error of neural network is less than 3.5%, which confirms the applicability of statistical method and function chain neural network.