A New Method for Network Maintenance

This article proposes a novel knowledge injection method for leak detection and localization of leaks combining data collected from sensors located in the fluid delivery pipeline networks and thermographic images. The application of the knowledge injection method for the pipeline system is as follows. We separate and maturate stable expert knowledge from temporary volatile data. Next, these data are compiled for the detection of thermal anomalies. The method is applied to the detection of temperature, fluid flow, and pressure anomalies as indications for pipeline condition changes. We demonstrate this method in a typical application case: thermal leaks in a heating network. The data we may use come from drone camera and from network thermal parameters. The analysis quality stems from expert knowledge while pipeline conditions are constantly changing. This method is the basis for quick, effective, and efficient detection of problems.