基于逻辑回归CA模型的溢油模拟与预测

Simulation and Prediction of Oil Spill Based on Logistic Regression CA Model

结合逻辑回归方法和元胞自动机模型构建了逻辑回归CA模型,模型中的逻辑回归方法能够很方便地获取影响因子的权重,再设置起止条件等,便可以模拟出溢油的动态变化情况。把模型应用到Deep Spill项目的溢油模拟实验中,结果表明,从模型结果的形态看,模拟结果与检验结果吻合程度较高,能够很好地模拟出溢油扩散与漂移等重要的溢油特性。从统计分析中可知,模拟结果的总精度可达96.8%,Kappa系数达到0.834。

In this paper , a new logistic-regression CA model was proposed by combining cellular automata and logistic regression for simulating oil spill.The logistic regression in the model can easily obtain the weights .Then the weights, initial image and end condi-tions are input into the model to simulate the dynamic changes .The model is applied to simulate oil spill in DeepSpill project .From visual comparison , the simulation results are consistent with real situation .Statistic analysis shows that the overall accuracy is up to 96.8%and Kappa coefficient is 0.834.Can simulate the diffusion and drift of oil spills and other important characteristics of the oil spill.