FRPFP模型滑坡监测预警关联规则挖掘分析——以三峡库区江津到奉节段为例

Association rules of monitoring and early warning by using landslides FRPFP model——Case study of Jiangjin-Fengjie reach in Three Gorges Reservoir area

采用传统的关联规则用于岩土工程监测预警领域的知识发现,在数据庞大情形下单机机器学习实时性差,无法获得多因素综合作用的规则。由于未对前后部项进行约束,得到的关联规则冗余度高,含有大量不符因果逻辑的规则。基于此,提出一种前后部项约束关联规则并行化FRPFP (fore-part and rear-part parallel FP-growth)算法,并在大数据分布式处理平台Spark下进行实现。通过对三峡库区奉节至江津库段滑坡的孕灾因子统计分类,采用7个滑坡发育基础因子和4个滑坡诱导因子作为前部集合,滑坡前缘、中部、后缘监测点位移参数为后部集合,采集研究区25个滑坡11年监测数据。以FRPFP算法为模型架构基于关联规则的滑坡监测预警大数据系统,设计区域滑坡危险性规则挖掘、典型滑坡危险性规则挖掘、滑坡发生原因分析挖掘3个功能,用于库岸滑坡稳定性预测和分析,为认清库岸滑坡的破坏机制和提升其预报水平提供新的思路。

When the traditional association rules are applied to the monitoring and early warning of geotechnical engineering,the machine learning has poor real-time performance and high redundancy of association rules. Aiming at the real-time and logic requirements of the association rules in the case of massive monitoring data of landslides, a fore-part and rear-part parallel FP-growth(FRPFP) algorithm is proposed. Through the statistical classification of landslide disaster factors from Fengjie to Jiangjin of Three Gorges reservoir, 7 basic factors and 4 induced factors are set as the front set, and the displacement parameters at the monitoring points at front, middle and rear edges of the landslide are set as the rear set. In addition, the monitoring data of 25 landslides in the study area for 11 years are collected. Based on the FRPFP algorithm, a large data system of intelligent landslide monitoring and warning based on the association rules is established. The three functions, hazard criterion mining of regional landslides, hazard criterion mining of typical landslides and occurrence analysis, and mining of landslides, are designed, and implemented under the large data-distributed processing platform Spark. The engineering verification shows that the proposed model has good real-time performance and logical rules. It is used to predict and analyze the stability of the landslide on the bank, which provides a new way of thinking for identifying the failure mechanism of the bank landslides and improving the forecast level.