摘要
目的:利用滑坡过程中的多源异构数据,减少滑坡带来的伤害。方法:搭建滑坡综合实验系统,实时监测滑坡过程中降雨量、土壤含水率、土壤应力、地表位移的数据,对整个滑坡过程的数据进行分析。然后通过TOPSIS法,利用滑坡过程中的多源数据划分滑坡危险度。利用海鸥算法(SOA)优化BP神经网络,改进DS证据理论中证据冲突的问题。利用优化改进后的BP-DS模型,判断滑坡危险状态。结果:利用优化改进后的BP-DS模型能够将预测滑坡危险状态的精度提高到94.5%以上。结论:优化改进后的BP-DS模型能够精确地判断滑坡危险状态,实现滑坡预警。
Aims:The multi-source heterogeneous data were used in the process of landslides to reduce the damage caused by landslides.Methods:A comprehensive landslide experimental system was built to monitor the data of rainfall,soil moisture content,soil stress and surface displacement in the process of landslides in real time.To analyze the data of the whole landslide process,the TOPSIS method was used to divide the landslide risk degree through the multi-source data in the landslide process.The seagull algorithm(SOA)was used to optimize the BP neural network to improve evidence conflict in the DS evidence theory.The optimized BP-DS model was used to judge the dangerous state of landslide.Results:The prediction accuracy of landslide danger state was improved by more than 94.5%by using the improved BP-DS model.Conclusions:The optimized BP-DS model can accurately judge the dangerous state of landslides and realize early landslide warning.
作者
李枫林
陈华民
童仁园
李青
LI Fenglin;CHEN Huamin;TONG Renyuan;LI Qing(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China;Jinhua Geological Environment Monitoring Station,Jinhua 321000,China)
出处
《中国计量大学学报》
2021年第4期489-496,共8页
Journal of China University of Metrology
基金
国家重点研发计划课题(No.2017YFC0804604)
浙江省重点研发计划项目(No.2018C03040)。