查看文章

Marine oil spill incidents are detrimental to both natural environment and human health. Water quality, marine ecosystems, and shoreline conditions can be deteriorated by the spilt oil. Swift and efficient response to an oil spill is crucial to minimize the adverse consequences. However, the oily waste generated from response operations may also become a challenge, requiring careful waste management strategies. Widely used oil spill response methods (OSRMs) include mechanical containment and recovery (MCR), in-situ burning, and the use of chemical dispersants. Choosing the most suitable method is a complex process depending on various factors. Among OSRMs, MCR is the most effective in removal of spilt oil from the marine environment. The management of oily wastewater generated during MCR requires careful attention, as it comprises a significant portion of overall oily waste. This study developed multiple tools to aid selecting OSRMs in harsh and remote offshore waters. These selection tools employ machine learning techniques and historical response data to predict appropriate OSRMs for new incidents. The tools were developed in MATLABTM using various artificial intelligence and soft computing techniques, such as fuzzy decision tree (FDT), Gaussian process regression (GPR), and artificial neural network, individually or in combination. FDT-based tools were also integrated with regression analysis techniques and an optimization algorithm to enhance their performance. Optimized FDTs integrated with regression analysis and GPR were found to be the most effective techniques based on the prediction power. Furthermore …