土石坝巡检信息智能感知与安全动态诊断方法

Numerous examples show that routing inspection is important to ensure dam safe operation, especially for the large number of small and medium earth rock-fill dams without enough monitoring instruments. However, many problems, such as multi-source heterogeneity, poor timeliness and low utilization rate, exist for current routing inspection, which have stunted dam safety diagnose. In this study the Internet of things technology based inspection is studied with intelligent perception model. Multiple disaster factors affecting dam safety are excavated, and the safety information fusion model is proposed. According to recorded failure cases, dam slope stability experiments under various conditions are carried out to reveal main failure modes and corresponding induced basic events. Combing the event tree method and the Bayesian network method, the inspection index system and the safety diagnosis model for earth rock-fill dam are constructed. According to the established disaster factors database, the prior and transformation probability of inspection indicator and its evaluation standard status are determined using the reliability theory and the information fusion method. Meanwhile considering uncertainty of human factors, the improving method of the state probability of inspection indicator is also improved. Furthermore, the fusion method of system failure probability, considering the posterior probability of additional information and the series-parallel system, is proposed. Finally the intelligent dynamic diagnosis model of dam safety, based on intelligent patrol information, is built. This intelligent model can implement dynamic diagnosis for small and medium earth rock-fill dams without enough monitoring instruments.

诸多实例表明巡检对保障大坝安全运行至关重要，尤其是对监测设施不完备的众多中小型土石坝安全运行更为关键，然而当前巡检信息存在多源异构、时效性差和利用率低等不足，不能实时掌握大坝安全性态。本项目基于物联网技术，研究巡检信息智能感知模式，挖掘影响土石坝安全的多重相关要素，构建土石坝安全信息融合模型。开展多工况下坝坡失稳室内模型实验和案例验证，揭示土石坝主要灾变模式和诱发事件；借助事件树和贝叶斯网络方法，构建土石坝巡检指标体系和安全诊断模型。依据已建土石坝病害要素数据库，采用可靠度理论和信息融合方法，制定巡检指标状态先验及其转化概率和评判标准，分析人因环境等对状态取值影响，改进指标状态概率量化方法；提出新增信息指标状态后验概率和串并联系统灾变概率融合方法，建立基于指标状态的土石坝安全智能动态诊断模型，实时掌握土石坝安全性态，确保安全运行，本项目可望实现大坝智能巡检与安全动态诊断方法的突破和创新。

诸多实例表明巡检对保障大坝安全运行至关重要，尤其是对监测设施不完备的众多中小型土石坝安全运行更为关键，然而当前巡检信息存在多源异构、时效性差和利用率低等不足，不能实时掌握大坝安全性态。为此，本项目基于物联网技术，研究巡检信息智能感知模式，挖掘影响土石坝安全的多重相关要素，构建土石坝安全信息融合模型，实现大坝智能巡检与安全动态诊断。主要研究内容和取得结论如下：.（1）针对土石坝实际工程特点，基于物联网、云计算及大数据挖掘技术，开展了巡检信息感知方法研究；通过MVC框架体系搭建云应用平台，实现了智能巡检信息的实时在线监测；基于运行管理需求及大数据挖掘技术，借用云计算，通过数学分析模型，实现了基于巡检信息的分析计算和智能诊断。.（2）基于土石坝失事案例，辨识了土石坝坝坡失稳、渗透变形等破坏的致因、诱发条件。开展了土石坝坝坡失稳及裂缝缺陷影响下的渗流模型试验，发现裂缝在降雨后会出现一定程度的愈合，但内部土体连贯性无法恢复至裂缝前，严重影响坝坡稳定，同时旱涝急转工况具备了水力劈裂的发生必要条件，土石坝安全影响巨大。揭示了局部缺陷病害对土石坝系统整体失效的驱动作用。.（3）通过针对土石坝破坏溃决致因及病害机理分析，考虑了溃坝中间过程中可能会出现的发展趋势，分析了土石坝溃坝模式。从土石坝溃决模式过程中根据识别的安全隐患，挖掘了巡检指标，根据土石坝溃坝诱发事件的外在宏观表现，确定了土石坝各种溃决方式对应的巡检指标，构建了土石坝巡检指标体系。.（4）提出了基于温度阈值和图像像素两种不同的异常部位面积计算方法；采用了数字图像识别技术获取裂缝扩展过程中各参数变化。通过基质吸力变化反映黏土体受旱程度，基于裂缝扩展过程分析建立了裂缝宽度与裂缝深度关系，实现了从裂缝宽度反演裂缝深度，构建了基于表观宽度的裂缝深度预测模型，并提出了基于裂缝表观参数的大坝渗流预警指标，研发了基于巡检监控的动态诊断及预警系统。