This work focuses on the application of artificial intelligence methods to enhance pipeline monitoring systems, specifically Third-Party Interference (TPI) and leak detection. A critical aspect of pipeline monitoring revolves around determining the operational state of the pipeline. This is paramount because the processing algorithms are intricately linked to this information.
Traditionally, the pipeline's state has been determined through ad-hoc systems known for their robustness and reliability, despite occasional downtime and delays. However, these limitations may occasionally lead the monitoring systems to resort to less reliable algorithms, resulting in false alarms.
This innovative approach incorporates machine learning and deep learning techniques to create a data-driven system, significantly improving overall system performance in terms of both reliability and robustness. This approach enables us to extract valuable features from the data, constructing a data-driven model capable of accurately detecting the true state of the pipeline with minimal error rates and zero delays.