测绘与空间地理信息
測繪與空間地理信息
측회여공간지리신식
GEOMATICS & SPATIAL INFORMATION TECHNOLOGY
2015年
6期
10-12
,共3页
元胞自动机( CA)%溢油模拟%逻辑回归%DeepSpill
元胞自動機( CA)%溢油模擬%邏輯迴歸%DeepSpill
원포자동궤( CA)%일유모의%라집회귀%DeepSpill
cellular automata ( CA)%simulation of oil spill%logistic regression%DeepSpill
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DeepSpill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。
結閤邏輯迴歸方法和元胞自動機模型構建瞭邏輯迴歸CA模型，模型中的邏輯迴歸方法能夠很方便地穫取影響因子的權重，再設置起止條件等，便可以模擬齣溢油的動態變化情況。把模型應用到DeepSpill項目的溢油模擬實驗中，結果錶明，從模型結果的形態看，模擬結果與檢驗結果吻閤程度較高，能夠很好地模擬齣溢油擴散與漂移等重要的溢油特性。從統計分析中可知，模擬結果的總精度可達96.8％，Kappa繫數達到0.834。
결합라집회귀방법화원포자동궤모형구건료라집회귀CA모형，모형중적라집회귀방법능구흔방편지획취영향인자적권중，재설치기지조건등，편가이모의출일유적동태변화정황。파모형응용도DeepSpill항목적일유모의실험중，결과표명，종모형결과적형태간，모의결과여검험결과문합정도교고，능구흔호지모의출일유확산여표이등중요적일유특성。종통계분석중가지，모의결과적총정도가체96.8％，Kappa계수체도0.834。
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.
