مقاله انگلیسی رایگان در مورد استفاده از یک الگوریتم ژنتیک برای بهبود پیش بینی نشت نفت – الزویر ۲۰۱۸

The performance of oil spill models is strongly influenced by multiple parameters. In this study, we explored the ability of a genetic algorithm (GA) to determine optimal parameters without the need for time-consuming manual attempts. An evaluation function integrating the percentage of coincidence between the predicted polluted area and the observed spill area was proposed for measuring the performance of a Lagrangian oil particle model. To maximise the objective function, the oil spill was run numerous times with continuously optimised parameters. After many generations, the GA effectively reduced discrepancies between model results and observations of a real oil spill. Subsequent validation indicated that the oil spill model predicted oil slick patterns with reasonable accuracy when equipped with optimal parameters. Furthermore, multiple objective optimisation for observations at different times contributed to better model performance.

Introduction

Oil spills are a major environmental concern and regarded as one of worst types of marine pollution, some of which may have disastrous consequences for open oceans and coastal seas. Considerable research has been conducted on the transport of spilled oil using field and laboratory investigations. Numerical oil spill models, which predict the transport and behaviour of oil spills, are an essential instrument for risk assessment and clean-up during an actual accident. However, it is still not possible to predict the actual trajectories of oil spills with any degree of certainty. Over the last three decades, numerous detailed oil spill models have been presented with the goal of improving oil spill forecasting (ASCE, 1996; Reed et al., 1999; Spaulding, 2017). These models have been developed from two-dimensional horizontal models to three-dimensional multiphase models, from considering only oil on the surface to oil distributed in multiple interacting phases, and from including a single environmental factor to atmosphere–wave–current coupled effects. Although these theories and data are valid, oil behaviour is complex, and many aspects of this behaviour are far from being clarified satisfactorily. Currently, oil spill models incorporate a range of parameters, partly due to a lack of knowledge of the underlying mechanisms behind oil transport and reaction processes. Hodges et al. (2015) argued that empirical parameters are one of four major contributors to uncertainty in an oil spill model. Complicated environmental conditions and the complex mixture of hundreds of chemicals make every spill different, and determining a unique set of appropriate parameters for each event is impractical and difficult. For example, the 3% wind drift factor for oil movement considers average conditions, and the implication just represents average conditions and the actual factor ranges from 1 to 6%. Once submerged, oil particles driven only by water currents have a net lower drift speed than that assumed by the 3% rule. Moreover, oil converging in windrows accelerates, and the transport velocity becomes higher than the average 3%, and some variables, such as wind deflection angle, are disputed. Because an oil layer is too thin to experience the full Ekman spiral, the wind deflection angle has previously been set to zero (Coppini et al., 2011; Huntley et al., 2011). However, Samuels et al. (1982) argued that the veering angle is related to wind speed; when wind speeds are low, the average deflection angle can be as high as 20°. Understanding the model structure and underlying principles is a key requirement for increasing model reliability.