高彩霞, 冯凌云, 党建军.基于双级PNN的直驱永磁同步电机匝间短路故障诊断方法[J].河南理工大学学报（自然科学版）,2025,44(2):116-127.

GAO C X, FENG L Y, DANG J J.Interturn short circuit fault diagnosis method of direct-drive permanent magnet synchronous motor based on two-stage PNN[J].Journal of Henan Polytechnic University(Natural Science) ,2025,44(2):116-127.

基于双级PNN的直驱永磁同步电机匝间短路故障诊断方法

高彩霞, 冯凌云, 党建军

河南理工大学 电气工程与自动化学院，河南 焦作 454000

摘要: 目的 早期故障的精准定位对于预防电机系统故障恶化和制定故障后运行维护策略至关重要。针对多支路并联直驱永磁同步电机（MB-DDPMSM）匝间短路故障诊断问题，   方法 提出一种基于双级PNN的匝间短路故障检测与短路线圈自动定位的新方法。首先，根据MB-DDPMSM特殊的结构特点，选择支路差值电流（BDC）作为故障支路定位特征信号，支路电流残差（BRC）作为短路线圈定位特征信号；其次，建立MB-DDPMSM匝间短路故障状态的有限元模型，利用MB-DDPMSM有限元模型分析短路线圈位置对特征信号的影响规律，并建立BDC与故障支路位置的映射关系以及BRC与短路线圈位置的映射关系，为了消除故障程度对特征信号的影响并增强故障特征，对提取的BDC幅值进行L₂范数归一化处理，得到故障支路定位特征向量，对提取的BRC幅值进行Z-Score标准化处理，得到短路线圈定位特征向量；最后，建立基于双级PNN的故障诊断模型，利用第一级PNN实现匝间短路故障检测与故障支路定位，利用第二级PNN实现短路线圈自动定位。   结果 仿真和实验结果表明，所提出的方法能及时检测出电机早期匝间短路故障并实现短路线圈精准定位，识别正确率达100%。   结论 基于双级PNN的匝间短路故障诊断模型具有建模简单、收敛速度快、故障诊断精准度高等优势。 

关键词:直驱永磁同步电机;匝间短路故障;支路差值电流;支路电流残差;双级PNN;短路线圈定位

doi:10.16186/j.cnki.1673-9787.2023030061

基金项目:国家自然科学基金资助项目（52177039）；河南省科技攻关项目（222102220017）；河南理工大学博士基金资助项目（B2021-22，B2018-48）；河南理工大学创新团队项目（T2023-2）；河南省高校基本科研业务费专项项目（NSFRF240306）

收稿日期:2023/03/25

修回日期:2023/09/25

出版日期:2025-03-05

Interturn short circuit fault diagnosis method of direct-drive permanent magnet synchronous motor based on two-stage PNN

GAO Caixia, FENG Lingyun, DANG Jianjun

School of Electrical Engineering and Automation， Henan Polytechnic University， Jiaozuo 454000， Henan， China

Abstract: Objectives Accurate positioning of early faults is crucial for preventing the deterioration of motor system faults and formulating operation and maintenance strategies after faults. Methods Aiming at the problem that the inter-turn short-circuit fault diagnosis of multi-branch parallel direct-drive permanent magnet synchronous motor （MB-DDPMSM）， a novel method of inter-turn short-circuit fault detection and short-circuit coil automatic positioning based on two-stage PNN was proposed. Firstly， according to the special structural characteristics of MB-DDPMSM， the branch difference current （BDC） was selected as the fault branch location characteristic signal， the branch current residual （BRC） was selected as the short-circuit coil location characteristic signal. Secondly， the finite element model of MB-DDPMSM inter-turn short circuit fault state was established， the MB-DDPMSM finite element model was used to analyze the influence of short-circuit coil position on the characteristic signal， and the mapping relationship between BDC and fault branch position， BRC and short-circuit coil position was established. Then， in order to eliminate the influence of fault degree on the characteristic signals and enhance the fault characteristics， the extracted BDC amplitude was normalized by L2 norm to obtain the fault branch location feature vector. The extracted BRC amplitude was normalized by Z-Score to obtain the short-circuit coil positioning feature vector. Finally， a fault diagnosis model based on two-stage PNN was established. The first-stage PNN was used to realize inter-turn short-circuit fault detection and fault branch location， and the second-stage PNN was used to realize short-circuit coil automatic location.  Results The simulation and experimental results showed that the proposed method could detect the early inter-turn short-circuit fault in time and realize the accurate positioning of the short-circuit coil， and the recognition accuracy was as high as 100%.  Conclusions The inter-turn short circuit fault diagnosis model based on two-stage PNN had the advantages of simple modeling， fast convergence speed and high fault diagnosis accuracy. 

Key words:direct-drive permanent magnet synchronous motor;interturn short-circuit fault;branch differential current;branch residual current;two-stage PNN;fault coil location