高速公路雾天能见度预测方法

Method of Visibility Prediction on Highway in Foggy Weather

以多要素气象检测器采集的样本数据为基础,将温度、风速及湿度作为输入变量,雾天能见度作为输出变量,分别采用三层结构BP神经网络和支持向量机非线性回归预测方法,建立了雾天能见度的预测模型;将预测结果与实际数据进行对比分析,结果表明:BP神经网络和支持向量机均能较好地预测雾天能见度,其中BP神经网络和支持向量机模型预测值与实际值的相关性分别为0.895和0.978,支持向量机预测结果的误差更稳定,表明支持向量机更适于处理小样本、非线性、维数灾难和局部极小等问题。

Based on sample data collected by the multi factor meteorological detector,temperature,wind speed and humidity are used as input variables,visibility is regarded as output variable and two nonlinear regression method that BP neural network and support vector machine are respectively used to establish prediction models of visibility,the forecast results are compared with measured data.The results indicate that the BP neural network and support vector machine models can predict the visibility significantly,the correlation value of BP neural network and support vector machine model predictive value and the actual value are 0.914and 0.895,which belong to a high degree of correlation,the error of support vector machine is more stable.It is proved that the support vector machines are more suitable for problems such as processing small sample,nonlinear,high dimension and local minima.