Developing a predictive model using multivariate analysis and Long Short-Term Memory (LSTM) to assess corrosion degradation in mining pipeline thickness

Abstract

Pipeline corrosion has significant impacts on the human, economic, and natural environment. To help better detect and prevent it over time, in this paper, we propose a multivariate approach using machine learning. More precisely, we propose to study the evolution of the thickness of the mining pipeline using a multivariate approach and to implement a predictive model using the Long Short-Term Memory (LSTM) artificial neural network. Indeed, LSTM is a specific recurrent neural network (RNN) architecture designed to model temporal sequences. The proposed predictive model achieved an accuracy of 80% and a loss of 0.01 and was able to predict variations in eight thickness measurements over one hundred days.