为量化评估高速公路连续下坡的货车行车风险，提出了基于货车制动毂温升的事故风险评价模型。首先，考虑货车下坡时制动性能持续降低诱发链式事故的风险，用制动毂温升表征货车下坡的关键风险致因，用货车事故率表征风险事故; 然后，以京昆高速雅安至西昌段的连续下坡为研究路段，通过制动毂温升数值模拟进行了风险致因和风险事故的量化，利用相关性分析揭示了制动毂温度-温度梯度-事故率的内在联系; 最后，采用多元非线性回归法拟合曲面函数，构建了事故风险度量方程，并结合风险分级标准建立了事故风险评价模型。结果表明: 累积 Logistic 函数能够较好的解释货车下坡事故风险的变化规律，模型评价结果与实际事故的平均误差为 0. 288，准确率达到80. 87%，模型精度较高，可为连续下坡路段的交通安全管控提供科学决策的依据。

An innovative model based on brake heating for crash risk evaluation was proposed to quantitatively evaluate the driving risk of trucks on continuous downhill． First，considering the risk of chain accidents caused by the continuous reduction of truck braking performance，the brake heating of the brake hub was used to represent the key risk cause of the crash accidents，and the truck accident rate was used to represent the risk accident.Then，taking the continuous downhill in Ya'an-Xichang section of Beijing-Kunming highway as the research section，the risk cause and risk accident were quantified by numerical simulation of the brake heating，and the inner connection between braking temperature, temperature gradient and accident rate was revealed by correlation analysis.Finally，the crash risk measurement equation was constructed by using multiple nonlinear regression method to fit the surface function，and the crash risk evaluation model was established in combination with the risk classification criteria． The results show that the Cumulative Logistic function can better explain the changing law of the truck crash risk on downhill． The average error between the final model evaluation result and the actual accident is 0. 288，and the accuracy rate is 80. 87%． The evaluation model shows high accuracy and can provide reference for scientific decision-making in traffic safety management and control of continuous downhill sections．