摘要
通过以某口径高射机枪自动机为研究对象,提出一种运用固有时间尺度分解(ITD)分形模糊熵与RBF神经网络进行故障诊断的方法。由于自动机振动信号短时、非平稳、高冲击的特性,将ITD引入自动机的故障诊断中,通过对ITD分解得到不同频段的合理旋转(proper rotation简称PR)分量,然后分别提取分形维数和模糊熵组成组合特征向量。由于RBF神经网络结构简单、收敛速度快具有很高的分类准确率,所以采用RBF神经网络分类识别。最后得到理想识别效果的同时验证了ITD分形模糊熵与RBF的自动机早期故障诊断方法的优越性。
By taking a caliber machine gun automaton as research object,a fault diagnosis method based on the intrinsic time scale decomposition(ITD)Fractal fuzzy entropyand the RBF neural network is proposed. Because of the short time,non-stationary and high impact characteristic of the automatic vibration signal,the ITD is introduced into the fault diagnosis of the automatic machine.The rational rotation(proper rotation PR)of different Frequency bands component is through ITD decomposition,and then extracting the fractal dimension and Fuzzy entropy respectively combine feature vector. The RBF neural network has the advantages of simple structure,fast convergence speed and has high classification accuracy. So the paper uses RBF neural network classification.Finally the paper could obtain ideal recognition results and verify the superiority of automaton early fault diagnosis method of ITD Fractal fuzzy entropy and RBF neural network.
作者
赵雄鹏
潘宏侠
刘广璞
安邦
ZHAO Xiong-peng;PAN Hong-xia;LIU Guang-pu;AN Bang(School of Mechanical and Power Engineering,North University of China,Shanxi Taiyuan 030051,China;System Identification and Diagnosis Technology Research Institute,North University of China,Shanxi Taiyuan 030051)
出处
《机械设计与制造》
北大核心
2019年第1期134-137,共4页
Machinery Design & Manufacture
基金
国家自然科学基金项目-基于运动形态分解和信息熵融合的高速自动机早期故障诊断研究(51175480)
基于广义形态学与多场信息融合的复杂供输弹系统早期故障预示方法研究(51675491)
关键词
自动机
固有时间尺度分解
分形维数
模糊熵
RBF神经网络
故障诊断
Automaton
Intrinsic Time-Scale Decomposition
Fractal Dimension
Fuzzy Entropy
RBF Neural Netw-ork
Fault Diagnosis
