Abstract
Pipeline integrity management is widely used as an effective means for pipeline safety management, in which integrity evaluation is an important part. To some extent, pipeline integrity can be interpreted as the safety condition of the pipeline, while safety is an eternal topic for pipeline operators. In numerous recent studies, the evaluation of pipeline integrity generally focuses on the evaluation of remaining strength and/or residual life, which is based on the defect size such as corrosion, dents, etc., obtained during inspection. However, pipeline integrity is not only related to the pipe body, all factors that may threaten the operation safety of the pipe should be considered, including the pipe body, ancillary facilities, the pipe security system, and the surrounding environment, etc.. Although some comprehensive models have been established recently to assess pipeline condition, there still exist limitations for practical application, such as quantification of integrity and complexity of analysis. Therefore this paper presents the development of a comprehensive integrity evaluation method based on multi-factor analysis. The method is developed by an integrated application of fuzzy mathematics, grey correlation analysis theory, and the artificial neural network technique. After establishing integrity evaluating indexes, fuzzy analysis is used to quantify and classify pipeline integrity, and grey correlation analysis to screen key influence indicators. Then a comprehensive predictive evaluation model can be generated using large amount of relevant sample data based on the artificial neural network technique. In the end of the paper, a simple case is applied to validate feasibility of this comprehensive integrity evaluation method. The comprehensive evaluation method is expected to be applied to determine the condition of pipeline integrity, and to grade and rank the integrity condition of pipes, so as to assist and optimize pipeline maintenance decision for pipeline operators.
Comprehensive Evaluation Method of Transformer Area State Based on Hopfield Neural Network