一种基于模糊Kohonen网络的飞行数据智能处理方法

An Intelligent Approach to Flight Data Processing Basedon a Fuzzy Kohonen Neural Network

为了充分利用飞行参数记录仪 (黑匣子 )记录的数据 ,对飞行质量进行评估 ,需要把数据按动作进行分段。以是否含有标准飞行动作为分段标准 ,设计了一种飞行数据的分段模型———模糊Kohonen网络。首先对每帧飞行数据进行特征检测 ,将根据专家经验模糊处理得到的隶属度矢量作为模糊Kohonen的输入 ,经过该网络决策 ,最终得到以时间划分的标准飞行动作数据段和非标准飞行动作数据段。经实际飞行数据检验 ,该方法准确、有效。

In order to make good use of the data in the flight data recorder and to evaluate the pilot's wingmanship, it is necessary to divide the data into some pieces by action. This paper is concerned with fuzzy Kohonon neural network, a dividing model for flight data, whose design is based on the presence of standard flight actions. First the degree of dependence vector obtained by applying the experts' experience is input as fuzzy vector, and then the pattern of data frame is classified automatically through this fuzzy Kohonen. Finally, the flight data classified according to time is divided into two segments-the data of standard flight and the data of nonstandard flight, which will be used as the basis of automatic flight data dentification. Proved by a large number of actual flight data, this processing method is accurate and effective.