Reliability Evaluation Of Submarine Pipeline Based On Corrosion Data

With the continuous development of China's Marine energy industry,the laying of submarine oil and gas pipelines is increasing greatly.Oil and gas pipelines have been under the sea for a long time,the environment is complex and changeable,and the transmission medium is diversified,so they are vulnerable to corrosion during operation.In order to ensure the safe operation of pipelines,it is very important to analyze the reliability of submarine oil and gas pipelines.Because of the traditional reliability assessment methods rely too much on expert experience and cannot meet the requirements of pipeline health management,more scientific methods are needed to evaluate the reliability of pipelines.Based on the reliability assessment of submarine oil and gas pipelines,this paper uses machine learning algorithm to analyze submarine oil and gas pipelines depend on the original data of submarine oil and gas pipelines detected by magnetic leakage detection technology and geometric detection technology.The main research contents are as follows:Firstly,the corrosion classification is considered for the submarine oil and gas pipelines based on machine learning algorithm.Because corrosion is an important factor that causes pipeline damage,only by knowing the extent of pipeline corrosion damage,we can understand the health of the pipeline.In order to solve the problem of complicated calculation and low accuracy in the classification of pipeline corrosion grades,this paper used Dropout and momentum coefficient to improve the combined algorithm of SVM and DBN.Secondly,the residual life of submarine oil and gas pipeline is predicted by WS-LSTM method.In this paper,WS-LSTM prediction method is proposed,that is,a long-term memory network(LSTM)prediction model based on wavelet transform(WT)and stack-compiled(SAE).Then the basic idea of this prediction method is to use wavelet transform denoising to process the data collected in the field.Moreover use stack self-coding for feature selection,and use LSTM model to predict the processed data set.Furthermore,the experimental results show that the accuracy of the WS-LSTM prediction model constructed in this paper is 98%.Finally,the integrated learning algorithm is used to predict the maintenance coefficient of offshore oil and gas pipelines.This paper compares GBM algorithm with Adaboost algorithm,Adaboost+KNN algorithm and RF algorithm,and finds that GBM algorithm has the best prediction effect...