Index System And Evaluation Model Study Of Marine Engineering Pressure Container

Marine engineering pressure container is a key link of marine production platform,therefore, our country attaches great importance to its production safety. Although regulators intensify supervision, major accidents still happen now and then. In order to further enhance the security of pressure container, combining its complex conditions, it is the first time to take uncontrollable environmental factors as evaluation indexes, and establish index systems of security evaluation as well as early warning system, which lays a theoretic foundation for the evaluation work in the future.This paper selects BP neural network evaluation, because it can solve the complicated nonlinear problem, further optimize by genetic algorithm and overcome the drawback of itself. In the evaluation work, it combines with the Delphi method and analytic hierarchy process(AHP) to grade evaluation system and give weights, which greatly reduces the subjective evaluation of itself. it concludes that the next evaluation period is 3 to 6 years. To further verify the reliability of the evaluation results, the model of neural network is established. By comparing BP neural network and GA- BP neural network optimized by genetic algorithm, the following conclusion can be drawn under certain conditions: GA- BP neural network convergence step length is short,reflects the latter’s advantages of security evaluation work, and verifies accurate evaluation results of it security.Due to the outstanding performance of GA- BP neural network in security evaluation, this paper puts it in corrosion prediction of marine engineering spherical storage tank, and the corrosion measurement plays a decisive role in security evaluation work. It analyses one of spherical storage tank on the offshore platform and select the 14 points on the tank wall,taking the corrosion thickness detected for nearly three years as input data, and put it into GA- BP neural network to forecast. The result is close to the reality. In order to truly reflect the corrosion situation of the wall thickness of marine engineering spherical storage tank. It corresponds to wall thickness prediction and early warning system to distinguish different corrosion situations by the difference colors and take the worst point forecasts as heavy police and the staff should immediately conduct site inspection and find the reasons. The results show that GA-BP neural network is suitable for the security assessment of marine engineering pressure container and wall thickness prediction, and it is a scientific and high efficient evaluation model; the combination of wall thickness prediction and early warningsystem finish the testing work in the production scientifically and efficiently.