Natural Gas Quality Analysis by Recurrent Neural Networks

Abstract

In this paper we propose the analysis of recurrent neural networks that can be used for natural gas quality analysis. The choice of a statistical model for the gas quality analysis problem is mostly made by heuristic methods due to the lack of a general algorithm for choosing a model and a variety of both statistical models and architectures of specific models. This paper provides an analysis of the main recurrent neural networks that can be used to solve the task under discussion. The accuracy characteristics of simple recurrent neural network, recurrent neural network with long short-term memory, and neural network with gated recurrent unit are shown. Based on the conducted analysis, the conclusions are made about a neural network model that is most appropriate in the discussed problem.