基于神经网络的双球式空间电场仪测量数据在线处理

以神经网络为基础构建了球载双球式电场仪，测量电场数据的分类器根据空间电场变化的物理机理，将测量的空间电场强度数值分为3类：（1）时变空间电场强度；（2）随机脉冲干扰空间电场强度；（3）空间电荷放电场强度。利用分类结果对空间电场强度测量值进行不同处理，取得了满意的结果。该系统实现了在线学习、实时处理的目的。

A NEURAL NETWORK BASED ON-LINE PROCESS OF MEASURED DATA OF BALLOON-BORN TWO SPHERE ELECTRIC FIELD INSTRUMENT

Based on neural network a classifier for the space electric field data measured by balloon born two sphere electric field instrument is presented in this paper. According to the variation mechanism of space electric intensity of measured data, the space electric field is classified as three kinds. The first one is the normal electric field data, i.e., the field strength of time variation space. The second one is the data of electric field disturbed by random pulse noise. The third one is the field strength of space charge in the discharge field. By use of the result of classification the measured data of electric field is processed by different ways, and satisfactory results are obtained. In this system the on line study and real time processing are implemented. Therefore, it can well process the time variation data.