استخدام الشبكات العصبونية الاصطناعية في آلية التنبؤ بالتسرب وتحديد موقعه في خطوط أنابيب المياه الرئيسية. "دراسة حالة محطة ضخ الجكرة"

Abstract

Leaking water pipes in drinking water pumping stations are a common and significant global problem, causing a waste of water resources. Therefore, rapid detection, location, and repair of pipe leaks are crucial to ensuring their safety. In this study, a water leak prediction system was designed in the main transmission line. The studied case was the pumping station located in the Tartous countryside. A leak was observed in the main transmission line connecting the first station (collection tank) and the second station (main tank). The design of this system was based on the metrics used in the prediction process (such as flow rate in the transmission line, water pressure inside the transmission line, and line vibration), in addition to the environmental metrics (temperature and humidity surrounding the transmission line). The Odoo software application was used, which provides the ability to monitor and follow everything that happens in the pumping station through a web browser via the Internet. Through it, the Odoo module was created to predict water leakage using artificial neural networks that were previously trained on the metrics used in the prediction process. The designed system was tested, and as a result, good accuracy in prediction was achieved by evaluating its performance by calculating the accuracy measure, where the training accuracy reached (89%) and the test accuracy (90%). The results of the training and verification loss graph and the accuracy of the training and verification set showed that the improvements that occurred in the model for the training set appear The results are fairly consistent with the improvements on the validation set, indicating that overfitting is not a significant problem with the model