论文部分内容阅读
为实现多影响因素作用下的道路交通事故预测,将灰色系统理论和神经网络理论相结合,发挥灰色理论提高可用信息利用率、弱化数据序列波动性的优点及神经网络特有的非线性适应性信息处理能力,提出道路交通事故灰色-径向基函数神经网络多元预测模型,并以某算例进行了不同预测方法结果对比。分析结果表明:与灰色系统预测和径向基函数神经网络预测相比,多元预测模型平均绝对误差、平均绝对百分比误差分别降低50.0%和12.5%,不等系数降低54.5%和16.6%,有效度提高2.7%和0.3%,说明该组合预测能够有效提高系统建模效率与模型精度。
In order to realize the prediction of road traffic accident under the influence of many factors, the combination of gray system theory and neural network theory can make full use of gray theory to improve the availability of information, weaken the advantages of data sequence fluctuation and neural adaptive nonlinear adaptive information Processing ability, a multi-forecasting model of gray-radial basis function neural network for road traffic accidents is proposed, and the results of different prediction methods are contrasted by a certain example. The results show that compared with gray system prediction and radial basis function neural network prediction, the average absolute error and average absolute percentage error of multivariate prediction model are reduced by 50.0% and 12.5% respectively, and the inequalities are reduced by 54.5% and 16.6% respectively. The validity Increased by 2.7% and 0.3% respectively, indicating that the combination forecast can effectively improve the system modeling efficiency and model accuracy.