基于函数链模糊Petri网的匝道控制方法

A Ramp Control Method Based on Functional Link Fuzzy Petri Nets

摘要: 函数链模糊Petri网(FLFPN)是一种具有自适应能力的拓展模糊Petri网(FPN)。本文构建了一个以FLFPN模型为核心的匝道控制系统，基于现有匝道控制研究去构建知识体系，搭建基于该知识库的FLFPN模型，根据采集到的真实交通数据进行训练与参数学习。除了能够直接预测匝道计量率外，模型构建中对当前拥堵状况的分析结果能够为用户提供后续匝道控制策略建议，从而实现完整的高速公路入口匝道控制决策过程。此外，本文对FLFPN模型的学习算法进行改进，将超参数对于实验结果的影响纳入考虑范围内，以此优化模型学习性能。

Abstract: Function-Link Fuzzy Petri Nets (FLFPNs) are an adaptive extension of Fuzzy Petri Nets (FPNs). This paper constructs a ramp control system centered on the FLFPN model, building a knowledge base based on existing research on ramp control. The FLFPN model is developed using this knowledge base and is trained and parameter-tuned with real traffic data. In addition to directly predicting the ramp metering rate, the analysis of current congestion within the model construction provides users with recommendations for subsequent ramp control strategies, thereby facilitating a complete decision-making process for highway ramp control. Furthermore, this paper improves the learning algorithm of the FLFPN model by considering the impact of hyper parameters on experimental outcomes, thereby optimizing model learning performance.