Pipeline data analysis utilizing machine learning method is present in this paper. Three machine learning models using Artificial Neuro Network methods are constructed to use nominal flow rate and head loss as input and pipeline roughness change, internal diameter change, a potential leak (location and leakage rate) as output. Half of the data created from the hydraulics model were used to train the Neuro Network. The rest of the data were used to validate the results of the network. The Neuro Network model was trained with the assistance of hydraulics equations. The advantage of the new method is that it provides an artificial intelligence to reveal the flow condition such as the ID, roughness, and leakage directly from the pressure and flow rate measurement data through machine learning without the necessity to compare with hydraulics model.
The study presented in the paper try to demonstrate the potential way of using machine learning and artificial intelligence in pipeline data analysis and operation analysis.
Number of Pages
Looking for more?
Some of the OnePetro partner societies have developed subject- specific wikis that may help.