摘要
提出了基于机器学习的埋地燃气管线防腐层绝缘电阻值预测方法,以北京燃气集团的626条燃气管线防腐检测结果为基础,并收集相应的特征数据,分别采用了线性模型、决策树以及随机森林等方法进行预测,其中随机森林的效果最好,R^(2)为0.73,相关系数为0.87。特征重要度显示管线的投运年限是最主要的影响因素,该预测方法提供了一种燃气管线健康状况预测方法。
Based on the 626 anti-corrosion test results of Beijing Gas Group Co,Ltd,and collected the corresponding characteristic data,respectively used linear model,decision tree and random forest to predict,among which random forest is the best,R^(2) is 0.73,correlation coefficient is 0.87.The characteristic importance shows that the service life of the pipeline is the most important factor.This prediction method provides a prediction method for the health status of the gas pipeline.
出处
《城市燃气》
2021年第7期4-7,共4页
Urban Gas
关键词
绝缘电阻值
机器学习
随机森林模型
影响因素
insulation resistance
machine learning
random forest model
influence factor
