基于改进人工神经网络的航天器电信号分类方法

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。

Spacecraft electrical signal classification method based on improved artificial neural network

To solve the problem of multiple data and arduous task in the aircraft test and intellectualize the management of the testing work, an intelligent classification system based on artificial neural networks was designed. The system can classify the original test data intelligently, reduce the workload and reliance on testing experience and store the nonlinear debugging experience in the form of expert database. This system has many deficiencies, such as, long training time and high dependence on the initial threshold. To this end, the principal component analysis was used to compress the raw data and auto-encoder in deep learning was applied to initialize the network weights. Experimental data indicates that compared with traditional methods, the accuracy, stability and response speed of the improved learning system are significantly increased.