Problem Research And System Development Of Tubing Corrosion Prediction And Early Warning Based On Data Mining

Oil tubing corrosion is an important factor affecting timing and material cost of oiling. Tubing corrosion can reduce the working life of tubes, reducing the efficiency of oil extraction. It is important to predict the rate of oil tubing corrosion for reducing cost of time and materials with changing and fixing oil tubes. If corrosion rules could be found according a big data set of oil tubing corrosion tests, cost of human and material resource could be reduced dramatically. Most of methods about analyzing and researching oil tubing corrosion are traditional, with low dimensions and insufficiency on researching of oil tubing corrosion test data. This dissertation extracts useful information with data mining technology.This dissertation applies Naive Bayes Algorithm, Bayes Net Algorithm, C4.5 Algorithm, Random Forest Algorithm and Multi Layer Perceptorn(MLP) Algorithm to the discovery of corrosion rules, prediction, and building of prediction model, analyses the results of them. Results show that MLP Algorithm is suitable to this issue. This dissertation studies MLP Algorithm, comes to a conclusion that it has a good ability to simulate any function, and can be applied in tubing corrosion prediction with parameters adjusting.This dissertation applies MPL to tubing corrosion prediction data mining program., which builds corrosion prediction model and predicts value with tube working parameters through data mining technology. This dissertation realizes tubing corrosion prediction and pre warning system based on corrosion prediction program with Weka and MLP, and development with Tomcat, Maven and Nexus, proofs that solving tube corrosion prediction problem with data mining is effective and feasible.