Neurofuzzy mathematical model for monitoring flow parameters of natural gas

摘要

Natural gas exists as associated gas (AG) or non-associated gas (NAG). Each individual industrial process engineering design is based on a particular gas quality specification. This paper presents a new approach, based on a three-layer neurofuzzy network theory, to detect the occurrence of off-specification gas in a supply/distribution network. We also present a new technique for the treatment of overlaps among adjoining fuzzy sets. The model combines the learning capabilities of neural networks with fuzzy control. This neurofuzzy model has been simulated on a digital computer, using C++ programming language and training data from three local process plants. It is shown that the model can successfully be implemented on microcomputers in a feedback loop, to individually control temperature, heat capacity and gas pressure.