Artificial Neural Network-based Evaluation on Soil Corrosion Situation along Pipelines

The remarkable nonlinear proximity is one of the characteristics of artificial neural network, which is utilized to evaluate soil corrosion situation of pipelines. This paper introduces the methodology and process in establishing the artificial neural network-based evaluation model on soil corrosion situation of pipelines. In a case, 4 factors, that is pH value of soil, soil's resistivity, oxidation-reduction potential of soil and electrolytic weightlessness, are taken into account in calculation. 9 studying samples selected are used to train the BP artificial neural network, and then, to evaluate the corrosion situation on 6 field test points of pipeline with trained BP artificial neural network and fuzzy comprehensive assessment method. Evaluation results show that the evaluation results obtained from the above two methods are entirely fitted each other, and suitable for the evaluation of soil corrosion situation of pipelines.