论文部分内容阅读
根据航天器系统级电性能测试工作中数据量大、任务繁重的特点,设计了基于人工神经网络的智能分类系统,对原始测试数据进行智能化分类,将非线性的调试经验以数据的形式储备,可在减少测试工作中对人为经验依赖的同时为航天器信号识别快速提供专家知识。考虑到经典的人工神经网络系统有训练时间长和对网络初始权值的依赖程度高等不足,利用主成分分析对数据进行压缩和自动编码技术对网络权值进行初始化。实验数据测试表明:与传统方法相比,本文提出的改进学习系统的分类准确率、稳定性和响应速度均得到显著提高。
According to the large amount of data and heavy workload in the system-level electrical performance testing of spacecraft, an intelligent classification system based on artificial neural network is designed to classify the original test data intelligently and store the nonlinear debugging experience in the form of data , Which can rapidly provide expert knowledge for spacecraft signal recognition while reducing the dependence on human experience in testing efforts. Taking into account the classic artificial neural network system has a long training time and the initial weight of the network is higher than the initial dependence, the use of principal component analysis of data compression and automatic coding technology to initialize the network weights. Experimental results show that compared with the traditional methods, the classification accuracy, stability and response speed of the improved learning system proposed in this paper have been significantly improved.