基于模拟加载系统油压信号的自动测试与识别技术研究  

Automatic measurement and recognition of oil pressure signals based on a simulated loading system

作　　者：王新晴[1] 王东[1] 赵洋[1] 朱会杰[1] 谢全民[2] 

机构地区：[1]中国人民解放军理工大学野战工程学院,南京210007 [2]武汉军械士官学校,武汉430075

出　　处：《振动与冲击》2013年第2期44-49,共6页Journal of Vibration and Shock

基　　金：国家科技重大专项子课题(2011ZX05056-003-0*)

摘　　要：在液压系统模拟加载与自动测试、识别过程中,工作装置油压波动信号是一种典型的非平稳信号。针对其影响因素多、不具备明显频域特征以及任何单一特征参量都无法对信号进行准确识别的难题,提出了对信号先进行状态分割,在分割基础上计算不同工作状态下的特征参量,并进行基于主成分分析(PCA)的特征提取方法,最后采用最小二乘支持向量机(LSSVM)构建多分类器,实现对工作装置6种不同工作状态的准确识别。实验结果验证了该方法的有效性,为同类液压系统的信号特征分析及模式识别提供了参考。In a process of automatic measurement and recognition, oil pressure signals of a working device based on a simulated loading system are typical non-stationary. Aiming at the signals with many influencing factors and without significant frequency domain features, and no single feature parameter used to recognize the signals exactly, a method was proposed here to exactly recognize six different working states of a working device. Firstly, the different working states of the signals were divided up. Then, the principal components from the original calculated feature parameters under different working states were extracted based on principal component analysis （PCA）. Finally, the multiple classifiers were constructed using a least square support vector machine （LSSVM）. The test results verified the effectiveness of the proposed method. The method provided a reference for characteristic analysis and pattern recognition of the same type hydraulic pressure signals.

关 键 词：油压信号 分割 特征提取 主成分分析 模式识别 最小二乘支持向量机 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]