基于神经网络的多参数在线检测系统的设计

Design of Multi-Parameter Online Detection System Based on Neural Network

针对井下变电站传统检测方式工人劳动强度大、检测结果不准确、成本较高等问题,对变电站智能远程在线检测系统进行研究,分析了变电站常见故障类型以及异常特征,利用神经网络算法搭建多参数远程检测系统模型,并完成了硬件系统选型,利用模块化思想编写了软件控制程序。经现场应用,基于神经网络的变电站在线检测系统检测效率大大提高,异常响应时间仅为1.89 s,故障诊断率为98.4%,减少工人4~5名。

For underground substation traditional detection methods of workers labor intensity,inaccurate test results,higher cost,study the remote substation intelligent online detection system,analyzes the common fault type and abnormal characteristics,using neural network algorithm build multiple parameter remote detection system model,and completed the hardware system selection,using modular ideas to write the software control program.After field application,it shows that the detection efficiency of the substation online detection system based on neural network is greatly improved,the abnormal response time is only 1.89 s,the fault diagnosis rate is 98.4%,reducing 4~5 workers.