驾驶人分心状态判别支持向量机模型优化算法

驾驶人分心状态判别是分心驾驶预警系统的重要基础.建立以径向基为核函数的 驾驶人分心状态判别SVM模型，采用遗传算法(GA)优化SVM模型惩罚参数C和核函数参数 g，并利用模拟驾驶器实验平台采集的驾驶绩效数据对模型进行验证.结果表明，采用GASVM 模型能够准确识别自由流和拥挤流场景下驾驶人分心状态，判别精度分别为94.5%和 96.3%.与决策树C4.5 和交叉验证(CV)-SVM对比表明，GA-SVM在准确率、精准率、召回率和 F1值等模型性能方面均优于其他2 种方法.本文建立的模型能够有效地判别驾驶人分心状态， 可为驾驶人分心预警系统和分心控制策略提供依据.

Identification of Driver Distraction States with Optimized Support Vector Machine Method

The identification of driver distraction states is the important component of driver distraction warning system. This paper establishes Support Vector Machine (SVM) model based on Radial Basis Function to identify driver distraction states. The genetic algorithm (GA) is used to optimize the parameters of SVM model. The optimized SVM model is used in the identification of driver distraction states, and the effectiveness of the model is validated by experimental data that used the driving performance data. The results show that the recognition accuracies of driver distraction states in free flow condition and crowed flow condition are 94.5% and 96.3%, respectively. Compared to the C4.5 and Cross Validation (CV)-SVM, the performances of GA-SVM are superior to the other two methods. This model can effectively determine the driver distraction state, which can provide data support for driver distraction warning system and control strategy.