Abstract: The reaction time of drivers reading traffic signs on highways is a key factor for determining the number and information capacity of traffic signs installed on highways. Based on the queueing network-adaptive control of thought-rational (QN-ACTR) theory, a cognitive computational model was developed. By quantifying and predicting the reading response time of single/multiple traffic signs, the adjustment mechanism of driving experience on response performance was elucidated. A production system incorporating traffic sign screening rules was established, and an integrated experimental platform combining the QN-ACTR model with TORCS driving simulator was constructed. Three hypotheses for driver visual search and response strategies were proposed and validated, and the effectiveness of the proposed model was verified using parallel dual task experimental data. The research results show that the proposed model can accurately predict reaction time variations under different numbers of traffic signs and road name information, with a mean absolute percentage error (EMAP) of 2.47% and a root mean square error (ERMS) of 0.06 s.

Key words: transportation engineering; driver behavior modeling; cognitive system architecture; queuing network; human factors

摘要： 驾驶人在高速公路阅读交通标牌的反应时间是决定高速公路设置交通标牌数量及信息容量的关键因素。基于排队网络-理性思维自适应控制理论(QN-ACTR)，构建了认知计算模型，通过量化预测单/多交通标牌的阅读反应时间，解析了驾驶经验对响应性能的调节机制；建立融合交通标牌筛选规则的产生式系统，搭建QN-ACTR模型与TORCS驾驶模拟器联合实验平台，提出并验证了3种驾驶人视觉搜索与响应策略假设，采用并行双任务实验数据验证了该模型的有效性。研究结果表明：该模型能精确预测不同交通标牌数量及路名信息量下的反应时间变化，平均绝对百分比误差(EMAP)为2.47%，均方根误差(ERMS)为0.06 s。

关键词: 交通工程；驾驶行为建模；认知体系架构；排队网络；人为因素