微型电脑应用

针对当前气体绝缘封闭开关(GIS)设备缺陷检测智能化水平较低的现状，提出一种基于无线传感网络(WSN)和超声波振动信号识别的GIS设备自动检测算法。该算法采用WSN进行超声波振动信号的采集与上送，通过人工鱼群算法(AFSA)进行变分模态分解(VMD)算法的固有模态分量数量及惩罚因子等参数的优化。同时利用优化后的VMD算法进行超声波振动信号分解，并将分解得到的多个固有模态分量作为长短期记忆(LSTM)网络的输入，以分析挖掘超声波振动特征与GIS设备缺陷之间的内在关系，从而实现缺陷的自动检测识别。仿真算例结果表明，相比于VMD-LSTM和AFSA-VMD-BPNN算法，所提出的AFSA-VMD-LSTM算法具有更为理想的缺陷检测效果，且平均缺陷识别率可达93.5%,能够有效提升变电站运维的智能化水平。

Aimed at the low intelligence level of gas insulated switchgear(GIS) equipment defect detection, this paper proposes an automatic detection algorithm for GIS equipment based on wireless sensor network(WSN) and ultrasonic vibration signal recognition. The algorithm uses WSN to collect and send ultrasonic vibration signal, and uses artificial fish swarm algorithm(AFSA) to optimize the number of intrinsic modal components and penalty factors of variational mode decomposition(VMD). The VMD algorithm after parameter optimization is used to decompose the ultrasonic vibration signal, and the decomposed intrinsic mode components are used as long short-term memory(LSTM) network inputs to analyze the internal relationship between the ultrasonic vibration characteristics and the defects of GIS equipment, so as to realize the automatic detection and identification of defects. The simulation results show that compared with VMD-LSTM and AFSA-VMD-BPNN algorithms, the proposed AFSA-VMD-LSTM algorithm has a more ideal defect detection effect, its average defect recognition rate is 93.5%, which effectively improves the intelligent level of substation operation and maintenance.