Application Of Ensemble Learning Model In Marine Oil Pollution Detection

With the continuous development of marine mining technology and ship technology,the ocean has become a hard-hit area for oil spills or oily wastewater discharge,which has a great impact on the natural environment and the human economy.After the oil spill accident at sea,satellite remote sensing is used to achieve first-time oil spill detection,while the oil spill pollution detection in the potentially polluted sea area is realized by the combination of sample collection device and laboratory analysis.Subject to the limitations of data collection and analysis methods,marine water sample detection equipment often cannot achieve efficient and rapid oil spill detection tasks.This paper uses the monitoring data of the "Deepwater Horizon" oil pollution leakage accident as training data,trains the ensemble learning model,and discusses the marine oil spill detection technology.The specific research content has the following two points1)Propose ensemble learning model,mainly XGBoost algorithm(eXtreme Gradient Boosting),applied to the pre-test of marine oil pollution,assisting and supporting traditional laboratory analysis methods.And compare the performance of random forest and GBDT models on the same data set,and adjust the XGBoost model structure to make the model get better classification results.The experimental results show that the method adopted in this paper can achieve timely and efficient marine oil spill detection.2)Based on the training results of the XGBoost model for data mining,more information that is not expressed by the usual data analysis methods is obtained,which is consistent with the related research on marine environment and ecological pollution.At the same time,the model is applied to other water sample datasets of the Deepwater Horizon accident.The obtained analysis results prove that the model trained in this paper can achieve accurate marine oil pollution detection.