基于遥感智能解译技术的围堰溃决洪水淹没分析--以旭龙水电站为例

Inundation analysis of cofferdam break flood based on remote sensing intelligent interpretation technolog:a case of Xulong Hydropower Station

水电站围堰溃决洪水突发性强且非常规,应急调查时效性要求高。为分析水电站围堰溃决洪水风险,以旭龙水电站为例,提出了一套基于遥感智能解译技术的溃堰洪水淹没分析方法用于实物指标应急调查,并选取与居民财产安全密切相关的建筑物为典型对象,通过构建U-Net卷积神经网络模型,对旭龙水电站下游区域进行建筑物提取。结果表明:该方法可有效识别出建筑物分布情况,F-score指标精度在93%以上,在算法效率上也明显优于人工解译。

Cofferdam break flood of hydropower station is sudden and unconventional,which requires high efficiency of emergency investigation.A remote sensing intelligent interpretation framework analysis of cofferdam break flood was proposed for emergency investigation of physical indicators,and buildings closely related to residential property safety were selected as typical objects for analysis.The U-Net convolutional neural network model was constructed to extract the building in the lower reaches of Xulong Hydropower Station.The results demonstrated that the proposed method could effectively identify the distribution of buildings.The accuracy of F-score quantitative evaluation index was above 93%,and the algorithm efficiency was also significantly better than manual interpretation.