作者：李鹏, 彭嘉潮, 黄培炜, 刘根柱, 杜艺博

作者单位：华东交通大学机电与车辆工程学院，江西 南昌 330013

关键词：结构状态监测;传感器分布优化;非劣分层多目标遗传算法;支持向量数据描述

摘要：

研究提出一种基于传感器分布优化的转向架构架状态监测方法，采用小波变换和主成分分析对构架正常状态下的多工况振动信号进行特征提取和降维，以支持向量数据描述（SVDD）超球体半径定义双目标优化函数：传感器数量和超球体聚类指标，并基于改进的非劣分层多目标遗传算法（NSGA-II）对传感器分布进行优化。在此基础上，搭建转向架构架状态监测实验平台进行构架异常状态识别研究，结果表明：1） 经优化后的传感器分布方案能以较少的数量保证很好的构架异常状态监测效果，当传感器分布优化方案中传感器数量分别为1，2和3时，识别率分别为79.24%，94.80%和99.81%；2） 以转向架构架正常状态样本集构建的SVDD模型对异常状态具有很好的识别效果。

Condition monitoring for bogie frame based on two-objective sensor distribution optimization
LI Peng, PENG Jiachao, HUANG Peiwei, LIU Genzhu, DU Yibo
School of Mechatronics & Vehicle Engineering, East China Jiaotong University, Nanchang 330013, China
Abstract: This paper proposed a framework condition monitoring method based on sensor distribution optimization. Wavelet transform and principal component analysis was used to extract and reduce the dimensions of vibration signals under multiple operating conditions of normal conditions of the framework. A two-objective optimization function was defined by the support vector data description (SVDD) hypersphere radius: the number of sensors and the hypersphere clustering index, and the sensor distribution was optimized based on the non-dominated sorting genetic algorithm II(NSGA-II). On this basis, a bogie condition monitoring experimental platform was built to identify the abnormal state of the framework. The results show: 1) the optimized sensor distribution scheme can ensure a good monitoring effect of abnormal state of the framework with a small number of sensors. When the number of sensors in the optimized sensor distribution scheme is 1, 2 and 3, the recognition rate is 79.24%, 94.80% and 99.81%, respectively. 2) the SVDD model constructed from the normal state sample set of bogie frame has a good recognition effect on the abnormal state.
Keywords: structural condition monitoring;sensor distribution optimization;NSGA-II;SVDD
2020, 46(8):131-135,147  收稿日期: 2020-01-24;收到修改稿日期: 2020-03-13
基金项目: 国家自然科学基金资助项目（51365012）
作者简介: 李鹏（1976-），男，江西南昌市人，副教授，博士，研究方向为智能结构与材料、模式识别
参考文献
[1] 王兴华, 于卫东, 张弘. 客车转向架构架裂损故障分析[J]. 铁道车辆, 2015, 53(2): 38-41
[2] 蔡俊鹏, 吴炳福, 陈德旺. 基于机器学习的高速列车转向架振动信号监测[J]. 计算机技术与发展, 2019, 29(8): 130-135
[3] 张楠. 高速动车组构架的传感器优化布置方法研究[D]. 兰州: 兰州交通大学, 2018.
[4] LI P, LIU Y, ZOU T, et al. Optimal design of microvascular networks based on non-dominated sorting genetic algorithm II and fluid simulation[J]. Advances in Mechanical Engineering, 2017, 9(7): 1-9
[5] 丁晨阳, 彭军. 基于改进粒子群优化算法的传感器部署机制[J]. 仪表技术与传感器, 2016(12): 176-180
[6] 彭珍瑞, 张楠, 殷红, 等. 基于频响函数的动车组构架传感器优化布置[J]. 西南交通大学学报, 2019, 54(2): 402-407,414
[7] LUO H, WANG Y, CUI J. A SVDD approach of fuzzy classification for analog circuit fault diagnosis with FWT as preprocessor[J]. Expert Systems with Applications, 2011, 38(8): 10554-10561
[8] 同晓雅, 陈春俊, 杨劼立, 等. 基于小波阈值的动车组模型脉动压力提取方法研究[J]. 仪表技术与传感器, 2019(2): 90-93,97
[9] 郭宇庭, 肖树岩. 基于小波变换的滚动轴承故障信号分析[J]. 中国测试, 2017, 43(S1): 142-147
[10] HAN S, ZHANG W, LI P, et al. Characterization of aromatic liquor by gas chromatography and principal component analysis[J]. Analytical Letters, 2016, 50(5): 777-786
[11] 金星, 邵珠超, 王盛慧. 基于PCA和IGWO-SVM的水泥回转窑故障诊断研究[J]. 中国测试, 2017, 43(10): 92-96
[12] TAX D M J, DUIN R P W. Support vector data description[J]. Machine Learning, 2004, 54(1): 45-66
[13] 肖剑, 王雨时, 张志彪. 基于Pareto遗传算法的无后坐炮内弹道多目标优化设计[J]. 弹道学报, 2019, 31(3): 53-57
[14] 李鹏, 刘远, 王裕华. 基于NSGA-II的通信电缆绞合节距优化[J]. 北京邮电大学学报, 2017, 40(2): 88-92
[15] 张鲲鹏, 陈哲武. 结构试验相似理论研究及实例分析[J]. 低温建筑技术, 2015, 37(7): 48-51