Basic evaluation for simulation-based crash-risk-models - multiscale modelling regarding dynamic traffic flow states

碰撞风险评估是基础设施规划领域的一项重要工作，目前主要通过统计建模来实现。然而，到目前为止，这组模型并不能根据交通流量的动态状态做出可靠的预测。实际研究提案的目的是，为即将到来的模拟辅助碰撞风险估计建立数学模型和分析考虑的基础。在方法论上，这将通过扩展现有的和验证的交通流模型来实现，通过交通流的非均匀动态来实现。因此，将负责任的驾驶员行为分析为交通的非均匀状态，并将其整合到模型中。将实际交通中识别的动态交通状态下的驾驶行为和实验室驾驶模拟器实验中的驾驶行为纳入微观交通模型。从这个小尺度系统出发，严格地数学推导出一个大尺度模型。然后在大尺度宏观交通流模型的基础上，对可能的碰撞风险进行了基于模型的预测。基于所获得的数据，对所导出的模型进行了校准和验证。通过这项研究项目，期望对数学交通建模有实质性的投入。根据这一发现，未来可以开发基于风险的个人车辆引导或集体交通影响的决策模型。

Crash risk estimation is an important task in the area of infrastructure planning, which nowadays is realized by statistical modeling. The group of models however until now does not result in reliable predictions depending on dynamic states of traffic flow. The aim of the actual research proposal is, to create the basis for a mathematical model and analytic consideration for a coming simulation-assisted estimate of crash risk. Methodically, this will be achieved by an extension of the existing and validated traffic flow models by inhomogeneous dynamic state of traffic flow. Therefore the responsible drivers behavior is analyzed as an inhomogeneous state of traffic and so integrated into the model. The driving behavior in dynamic traffic states identified in real traffic and in laboratory driving simulator experiments is incorporated into a microscopic traffic model. From this small-scale system rigorously mathematically a large scale model is derived. Model-based predictions of possible crash risks are then derived on the basis of large-scale macroscopic traffic flow model. The derived models are calibrated and validated based on the obtained data. By the research project a substantial input to mathematical traffic modeling is expected. From this findings risk-based decision models for individual vehicle guidance or collective traffic influencing can be developed in future.