Аннотация
Pipelines are widely used for the transportation of water, oil, gas, and other liquid and gaseous media. The occurrence of leaks in pipeline systems leads to significant losses of natural resources and may pose a threat to the environment and public safety. In this regard, the problem of timely leak detection and localization remains one of the most important challenges in the oil and gas industry. This paper addresses the problem of determining the leak location in a pipeline based on pressure measurement data using artificial intelligence methods. An approach to leak localization is proposed based on approximating the relationship between pressure drop, leak flow rate, and the distance to the leak location using a radial basis function neural network implemented in the MATLAB environment. Experimental data were obtained during testing of a leak detection system under simulated leak conditions with different operating modes. The simulation results demonstrate that the developed model provides an effective approximation of the studied relationship and allows the leak location to be determined with a root-mean-square error of approximately 1.9 km for medium and large leaks, confirming the potential of neural network-based methods for pipeline monitoring applications.