基于谱聚类的交通事故黑点鉴别及预测

交通事故黑点的鉴别及预测,可以避免更多的事故发生。选取事故率为评价指标,建立包含事故数据采集、事故数据筛选、基于谱聚类分析的事故黑点鉴别、事故黑点结果修正四部分的事故黑点鉴别体系;其次,基于交通事故的复杂性、多变性等原因,提出了基于BP神经网络的交通事故黑点位置预测模型。通过利用淄博市淄川区的实际交通事故数据进行验证,表明该文提出的交通事故黑点的鉴别与预测方法的可行性。

Select the accident rate as the evaluation index,establish accident black spot identification system including accident data collection,accident data screening,identification of accident black spots based on spectral clustering analysis and accident black point correction;Secondly,based on the complexity and variability of traffic accidents,the forecast model of traffic black point location based on BP neural network is proposed in this paper.By using the actual traffic accident data in Zibo City for verification,this paper proves the feasibility of the method for identifying and forecasting traffic black spots.