摘要
针对齿轮箱振动的非线性,利用非线性特征测度的方法提取齿轮箱振动信号的故障特征。并利用双子支持向量机(TWSVM)对齿轮箱故障类别的辨识性能进行研究。TWSVM努力构造两个非平行的超平面来实现分类,它比支持向量机(SVM)针对多分类问题具有更好的样本不均衡适应性,并且分类性能优势明显。对齿轮箱故障类别辨识的实验表明,与传统的SVM和BP神经网络算法相比较,TWSVM具有更高的分类准确率。
Aiming at gearbox vibration's nonlinearity,its vibration signals' fault features were extracted with the method of nonlinear characteristic measure. The twin support vector machine( TWSVM) technique was used to study gearbox 's fault category identification. With TWSVM, two nonparallel hyperplanes were constructed to realize classification,and it had a better adaptability than the support vector machine( SVM) technique did when samples were imbalance,and its classification performance had obvious advantages. Simulation tests for gearbox fault category identification showed that TWSVM has a higher classification accuracy rate than the BP neural network method and SVM do.
作者
曾柯
柏林
ZENG Ke, BO Lin(State Key Laboratory- of Mechanical Transmission, Chongqing University, Chongqing 400044, China)
出处
《振动与冲击》
EI
CSCD
北大核心
2018年第15期179-184,198,共7页
Journal of Vibration and Shock
基金
国家自然科学基金(51675064)
关键词
齿轮箱
故障诊断
非线性特征
TWSVM
gearbox
fault diagnosis
nonlinear characteristic
twin support vector machine (TWSVM)