基于机器学习方法的长输管线地质灾害预测建模

Prediction Modeling of the Long-distance Pipeline Geological Hazard Based on Machine Learning Method

地质灾害是影响长输管线安全运行的重要因素,为减少地质灾害对长输管线的影响,有效预防、规避长输管线的安全风险,本文基于西南油气田仁寿作业区的地质灾害数据,以研究区地质灾害为预测变量,高程、坡度、坡向、地表覆盖为响应变量,采用人工神经网络、决策树、支持向量机3种机器学习方法分析研究区域地质灾害的发生规律,并进一步预测研究区内长输管线发生地质灾害的概率。通过模型对比,相较于人工神经网络和支持向量机,决策树的方法具有更高的预测精度,其预测正确率达0.86。本文的分析结果对研究区管线管理部门具有重要参考价值,相关部门可根据模型预测出的管线点发生地质灾害的概率,提前做好地质灾害预防工作。

Geological hazard is an important factor affecting the safe operation of long distance pipeline.In order to reduce the impact of geological disasters on the long-distance pipeline and effectively prevent and avoid the safety risk of long transmission pipeline,this paper is based on the geological hazard data of Renshou area in Southwest Oil and Gas field Company,taking the geological disaster in the study area as the prediction variable,the elevation,slope,slope and surface coverage as the response variable,using Artificial neural network,Decision tree and Support vector machine three machine learning methods to analysis the rule of geological disaster and use the model parameters to predict the probability of pipeline point geological hazards in the study area.Compared with Artificial neural network and Support vector machine,the Decision tree method has higher prediction accuracy and its prediction accuracy is 0.86.The research results of this paper has important reference significance for the pipeline management department in the study area,according to the probability of occurrence of geological hazards predicted by the model,the relevant departments can take measures for geological hazards are done in advance.