A leakage detection system on the Water Pipe Network through Support Vector Machine method

Clean Water is the primary and basic need for human. Thus, the provision of it has to be assured by preserving the quality, quantity, and also the pressure. However, in fact, there are many leakage cases in the distribution process which become the world´s problem. One of the causes of the leakage, technically, is the leakage on the pipes. The leaky pipe changed the pressure on each junction / node in water pipe network. The pattern of the pressure changes, then, is able to be analyzed computationally to detect the location and the size of the leakage. In this research, there is a detector of the leakage that will be used computerized through Support Vector Machine (SVM) method. As the research material, there is Water Pipe Network System in Taman Khayangan Resident Makassar which was made by using EPANET 2.0 software. The data of the Pipe Network System is obtained from the Data of Drinking Water Local Company (PDAM) of Makassar. The output of this Leakage Detection System is in the form of model which detects the size and the location of the leakage on the pipe. The results of the simulation showed that the SVM method is relatively accurate in predicting by resulting the RMSE average of 0.06785 for the leakage size and RMSE = 0.1382 for the leakage location.