摘要
对于具有强耦合性和不确定性的水电机组故障,基于统一特征向量的故障表征方法容易淹没有效特征的表征效果,直接限制了故障诊断的准确性和时效性。鉴于此,提出一种仿生故障诊断方法,该方法模仿人脑思维,提出关联特征向量概念,并用特征选择树描述其逻辑结构,以特征提取树求取。以其嵌套结构深层次挖掘故障之间的错综复杂关系,实现对水电机组故障之间量差异和质差异的综合挖掘,深层次揭示不同故障间以及故障与特征之间的固有联系,有效提取不同故障的本质区别,提高故障表征的有效性。进而结合概率神经网络实现智能分类。实验结果表明,该方法能够完成水电机组的故障诊断,同时取得了优秀的诊断有效性和时效性。
For the faults of hydropower unit with strong coupling and uncertainty,the traditional fault expression method based on the uniform feature vector is easy to submerge the effect of fault expression and limit the fault diagnosis accuracy and timeliness directly.Given this,we presented a bionic fault diagnosis method,which imitates the human brain thinking and proposes the concept of dependent feature vector(DFV),which logical structure is described by feature selection tree and extracted by feature extraction tree.DFV deeply excavates the intricate relationship between faults by using its nested structure,and realizes the comprehensive mining of the quantity and quality differences between faults of hydroelectric units.It deeply reveals the intrinsic relationship between different faults and faults and features,effectively extracts the essential differences of different faults,and improves the validity of fault expression.Furthermore,the intelligent classification method is realized by combining DFV and probabilistic neural network,the experimental results show that the method can complete the fault diagnosis of hydropower units and achieve excellent diagnostic validity and timeliness.
作者
陈晓玥
葛荡
黄江平
刘雄
张晓燕
CHEN XiaoYue;GE Dang;HUANG JiangPing;LIU Xiong;ZHANG XiaoYan(School of Electrical and Automation Engineering,East China Jiaotong University,Nanchang 330013,China;School of Hydropower and Information Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Ningbo Kingband Automobile Electronics Co.,Lad.,Ningbo 315040,China)
出处
《机械强度》
CAS
CSCD
北大核心
2020年第6期1271-1276,共6页
Journal of Mechanical Strength
基金
国家自然科学基金青年项目(51609088)
第63批中国博士后科学基金面上项目(2018M632864)
江西省教育厅青年基金项目(GJJ170408)
国家自然科学基金地区项目(61663011)资助。
关键词
水电机组
关联特征向量
综合差异挖掘
仿生诊断
特征选择
Hydroelectric generating set
Dependent feature vector
Comprehensive differences mining
Bionic fault diagnosis
Feature selection
