电力基坑有害气体风险评估方法设计  

Design of Hazardous Gas Risk Assessment Method for Power Pits

作　　者：文宗山 景国明 樊彦国 WEN Zongshan;JING Guoming;FAN Yanguo(Henan Power Transmission and Transformation Construction Co.,Ltd.,Zhengzhou 450000,China;Safety Supervision Department,State Grid Henan Electric Power Company,Zhengzhou 450000,China)

机构地区：[1]河南送变电建设有限公司,河南郑州450000 [2]国网河南省电力公司安全监察部,河南郑州450000

出　　处：《自动化仪表》2024年第2期122-126,共5页Process Automation Instrumentation

基　　金：国网河南省电力公司科技基金资助项目(KJXM(2022)74)。

摘　　要：为了提升风险评估准确率,设计了一种电力基坑有害气体风险评估方法。创新性地利用密度和加权方法改进聚类算法,实施电力基坑有害气体传感数据的挖掘。使用改进阈值函数去噪方法对传感器数据实施去噪处理。基于静态响应和动态响应构建传感器响应灵敏度控制模型,提高数据质量。基于机器学习中的图神经网络构建电力基坑有害气体风险评估模型,实现有害气体风险等级评估。测试结果表明,该方法能够实现风险等级评估,并将风险评估结果的准确率均值提升到95%以上。评估结果显示:工程中基坑4、基坑5、基坑6的风险评估等级为低等级;基坑2为中低等级;基坑1、基坑3、基坑8为中等级;基坑7为中高等级。该方法的评估结果可用于对电力基坑施工安全进行控制,具有实际工程应用价值。To improve the accuracy of risk assessment,a hazardous gas risk assessment method for power pits is designed.The density and weighting methods are innovatively utilized to improve the clustering algorithm and implement the mining of power pit hazardous gas sensing data.The denoising of sensor data is implemented using the improved threshold function denoising method.Constructing sensor response sensitivity control model based on static response and dynamic response to improve data quality.Based on the graph neural network in machine learning,the hazardous gas risk assessment model of power pits is constructed to realize the hazardous gas risk level assessment.The test results show that the method can realize the risk level assessment and increase the average accuracy of the risk assessment results to more than 95%.The assessment results show that the risk assessment levels of pit 4,pit 5 and pit 6 in the project are low;pit 2 is medium-low;pit 1,pit 3 and pit 8 are medium;pit 7 is medium-high.The assessment results of this method can be used to control the construction safety of power pits,which has the value of practical application in engineering.

关 键 词：机器学习 电力基坑 改进聚类算法 图神经网络 有害气体风险评估 传感器控制 气体浓度 密度可达性 

分 类 号：TH-39[机械工程]