基于物联网及神经网络的谐波诊断仪研制  

Development of Harmonic Diagnostic Instrument Based on Internet of Things and Neural Network

作　　者：郑思凡[1] 郭宝英[2] ZHENG Sifan GUO Baoying(College of Mechanical Engineering and Automation, Liming Vocational University, Quanzhou 362000, China College of Engineering and Technology, Yang＇ en University, Quanzhou 362000, China)

机构地区：[1]黎明职业大学机电工程与自动化学院,福建泉州362000 [2]仰恩大学工程技术学院,福建泉州362000

出　　处：《黎明职业大学学报》2017年第3期70-77,共8页Journal of LiMing Vocational University

基　　金：泉州市科技局科研项目(2014Z138)

摘　　要：随着深度学习等人工智能技术的发展,为了充分挖掘计算机数据处理潜力并应用在实际的电网故障检测上,在物联网及模式识别技术的基础上,实现一个Stm32为前端智能信号采集结点,后端以Matlab为数据处理引擎的基于"互联网+"架构的分布式电网暂态故障谐波诊断仪。实验证明:相比传统的谐波分析仪,该诊断仪具有在线实时提取故障谐波小波能量特征,并具有多层前馈神经网络的实时诊断识别故障类型的功能。同时,也表明仪器分布式的网络架构比传统仪器对新算法新技术具有更大的移植空间。With the development of artificial intelligence technology such as deep learning, in order to fully exploit the potential of computer data processing and apply it to the actual electricity power fault detection, this paper implements an internet-architecture distributed power quality transient fault diagnosis instrument which used Stm32 as a front-end intelligent signal acquisition node, and Matlab as the data processing engine based on the Interuet of things and the pattern recognition technology. Experiments show that, compared with the traditional harmonic analyzer, the diagnostic instrument developed in this paper has the performance of extracting harmonic wavelet energy characteristics and identifying the type of transient harmonics in real-time via a multi- layer feed forward neural network. At the same time, it also shows that the instrument, based on the distributed network architecture, has a broader space to update more powerful algorithm than the traditional instrument.

关 键 词：小渡能量谱 暂态谐波 Levenberg—Marquardt 机器学习 LWIP协议栈 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]