The invention relates to a multi-sensor traction machine excitation loss monitoring method based on one-dimensional multi-scale convolution. The method mainly solves the technical problem that a multivariable and nonlinear relation contained in a traditional convolutional network is difficult to accurately describe. According to the technical scheme, the excitation loss monitoring method comprises the specific steps that S1, excitation loss sensitive characteristic parameters are collected from existing data, and the types and arrangement forms of multiple paths of needed sensors are determined; s2, acquiring field loss characteristic signal data of the permanent magnet synchronous traction machine under typical working conditions through a multi-path sensor, and constructing a representative field loss characteristic database; s3, preprocessing the characteristic signal data in the field loss characteristic signal database according to the physical attribute and the sensitive expression form of each field loss sensitive characteristic, and updating the field loss characteristic database; s4, building a lightweight network model; s5, determining the output type of the network according to the field loss degree interval division; and performing network configuration and network training.
本发明涉及一种基于一维多尺度卷积的多传感器曳引机失磁监测方法。主要解决传统的卷积网络存在的难以准确拟刻画出其中蕴含的多变量、非线性的关系的技术问题。本发明采用的技术方案是:该失磁监测方法的具体步骤为:S1:从已有资料中收集失磁敏感特征参数,确定所需多路传感器类型及传感器布置形式;S2:通过多路传感器采集永磁同步曳引机典型工况下的失磁特征信号数据,构建代表性的失磁特征数据库;S3:依据各失磁敏感特征物理属性及敏感表现形式,对失磁特征信号数据库中的特征信号数据进行预处理,更新失磁特征数据库;S4:搭建轻量化的网络模型;S5:依据失磁程度区间划分,确定网络的输出类型;进行网络配置及网络训练。
Multi-sensor traction machine field loss monitoring method based on one-dimensional multi-scale convolution
一种基于一维多尺度卷积的多传感器曳引机失磁监测方法
2023-02-03
Patent
Elektronische Ressource
Chinesisch
Portable multi-machine multi-point traction turnout synchronism detection system
Europäisches Patentamt | 2021
|Multi-stator traction machine with parallel structure
Europäisches Patentamt | 2022
|Europäisches Patentamt | 2021
|Traction Motor In-Service Multi-Technology Condition Monitoring
British Library Conference Proceedings | 2008
|