模糊聚类方法在PIO探测中的应用  

Design of the membership function of fuzzy logic PIO detector

作　　者：颜世伟[1] 高正红[1] 刘艳[1] 田福礼[2] 

机构地区：[1]西北工业大学航空学院,陕西西安710072 [2]中国飞行试验研究院总师办,陕西西安710089

出　　处：《飞行力学》2011年第5期5-9,共5页Flight Dynamics

摘　　要：用于驾驶员诱发振荡(PIO)探测的特征参数可以有效地反映出PIO时飞机在频域和时域的特征,利用模糊逻辑方法对特征参数进行辨识可以有效地识别出是否发生了PIO。模糊逻辑方法中的隶属函数直接反映了特征参数的特点,其设计决定了是否能有效地对PIO进行识别。为了设计有效的隶属函数,根据特征数据的特点采用球壳型模糊聚类算法对数据进行分析与聚类,并对隶属函数进行修正。探测结果的对比验证了此方法对隶属函数的修正是有效的,对隶属函数的构建有指导意义。PIO detector is a kind of pattern-recognition method,which generally consists of data acquisition,data preprocessing,feature extraction and classifying decision.Using the short time Fourier transform（STFT） and wavelet transform to extract the features of PIO events.All these features show different between PIO and normal condition with quantum and quality.Then optimize the membership functions of fuzzy PIO detector with a fuzzy C-spherical shell clustering（FCSS） algorithm.This algorithm could cluster data set by its own distribution and features without any apriori knowledge.At last,this method is effective by comparing the original detect result with the result from improved detector with optimized membership functions by the FCSS algorithm.

关 键 词：短时傅里叶变换 小波变换 模糊逻辑 聚类分析 隶属函数 

分 类 号：V212.1[航空宇航科学与技术—航空宇航推进理论与工程]