Holistic Principal Tunnel-Sewer Survey System (HS3) using Unmanned Aerial vehicle and Artificial Intelligence+Big Data

深埋主隧道下水道的调查过程非常繁琐，需要断开污水治理实体2个月的时间进行关键的通风和互通网络的搭建，3个班组的调查工作耗时0.5个月（假设隧道长4英里），总共耗时2.5个月，成本超过50万英镑。环境也特别不安全；它有啮齿动物和其他携带疾病的动物，即使在晾晒后也会携带有害的固体和化学物质。最近，它们被发现含有冠状病毒的痕迹（BBC 2020）。所有这些都使得每年对每个管道进行多次调查变得困难，导致间歇性的坍塌、堵塞，特别是不断的泄漏和相关的污染事件。由于有超过3500个深埋的主要隧道污水管道，超过40万英里的污水管道与之相连，污水从这些管道被运送到处理站，泄漏造成的污染事件是巨大而频繁的。泄漏导致英国50%以上的河流受到污染，而且还在上升（环境署，2018年）。此类泄漏的罚款通常数额巨大，对收入造成沉重打击，导致隧道业主迫切需要其他调查方法。一个受欢迎的案例是泰晤士水务公司，该公司在2017年因泰晤士河发生严重且可避免的污染事件而被罚款2000万英镑（环境署，2017年）。逃避主要是通过频繁的调查来产生快速的干预。因此，市场对高生产率（更快、更便宜和更安全）的调查系统的需求尚未得到满足，这将导致频繁的隧道下水道调查。因此，该项目将开发一个整体的隧道下水道调查系统（HS3），其中包括隧道调查专用无人机（T-suv）和人工智能分类模型（AI-CM），该模型将分析T-suv的视频，以进行故障分类和调查报告的制作。以每小时2英里的速度，T-suv将在2小时内拍摄一个大约4英里长的典型隧道下水道系统的调查视频。HS3的AI-CM将在大约30分钟内分析生成的调查视频并生成故障报告。

The survey process of deep principal tunnel-sewers is very tedious, needing sewerage regulation entity be disconnected for 2-months for crucial airing and intercommunication network setup, and survey done by 3-squads for 0.5months (assuming a 4-mile long tunnel), expending 2.5 months in total and over half a million pounds in costs. The environment is also particularly unsafe; it has rodents and other disease-carrying animals and carries harmful solids and chemicals even after airing. More recently, they have been found to contain traces of Corona virus (BBC 2020). All these make it difficult to conduct the required multiple survey annually of each, causing intermittent collapse, blockages, and particularly incessant leaks and associated pollution episodes.Pollution episodes from leaks are big and frequent because there are over 3,500 deep principal tunnel-sewers, to which the over 400,000 miles of sewers are connected to, and from which sewage is transported to treatment stations. The leaks have led to pollution of more than 50% of UK rivers and rising (Environment Agency, 2018). The penalty fee for such leaks are usually huge and hard hitting on revenue, causing tunnel owners to be desperate for alternative survey methods. A popular case is that of Thames Water that was given a £20 million penalty significant and avoidable pollution episodes on the River Thames in 2017 (Environment Agency, 2017). Avoidance comes mainly through frequent surveys that begets quick intervention.Thus, an unmet market need exists for a highly productive (quicker, cheaper and safer) survey system that will engender frequent tunnel-sewers surveys. This project will thus develop a holistic Tunnel-sewer survey system (HS3) that includes a tunnel survey-specific unmanned aerial vehicle (T-suv) and artificial intelligence classification models (AI-CM) that will analyse T-suv's videos for fault-classification and survey reports production. At 2-miles per hour T-suv will produce survey videos of a typical Tunnel-sewer system of circa 4-miles length in 2 hours. HS3's AI-CM will analyse the generated survey videos and produce fault reports in circa 30 minutes.