基于LDA的潜艇机械噪声识别算法研究  

Submarine Mechanical Noise Recognition Algorithm Based on Linear Discriminant Analysis

作　　者：刘杰[1] 张福生[1] 冯达[1] 

机构地区：[1]91388部队92分队

出　　处：《舰船电子工程》2013年第4期140-142,共3页Ship Electronic Engineering

摘　　要：通过选取一定数量的潜艇机械设备噪声信号作为训练样本,利用线性判别分析算法(LDA)对噪声信号建立识别分类模型,然后将待测信号与已知信号库(机械设备正常工况下的噪声信号和所有已知故障情况下的噪声信号)中所有信号分别投影到识别分类模型上一一进行比对,确定潜艇机械噪声的类别。论文使用噪声模拟信号进行了相关实验,结果表明该方法能够比较准确的达到识别分类的效果。A certain quantity of submarine mechanical noise signals has been chosen as training samples and of linear discriminant analysis algorithm is used to build a recognition and classification model ofnoise signals, this thesis corapares all the signals from the untreated sig- nal bank with the known signal bank when the signals are proiected onto the recognition and classifieation model to confirm the type of submarine mechanical noise. The known signal bank consists of all the noise signals when the machines work well or not. This thesis also makes related experiments by using the analog signals, the result of which shows that this method can recognize and classify signals comparatively accurately.

关 键 词：线性判别分析 LDA 主成分分析 PCA 潜艇机械噪声 噪声识别 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]