Explicit Coordinated Signal Timing Optimization Using Soft Actor–Critic and Model Predictive Control Along an Arterial
4 affiliations
- 1School of Traffic and Transportation Engineering, Central South University, Changsha 410075, China
- 2Smart Transport Operation Laboratory, Changsha Communications Investment Holding Group Co., Ltd., Changsha 410014, China
- 3Neusoft Research, Neusoft (China), Shenyang 110179, China
- 4School of Infrastructure Engineering, Dalian University of Technology, Dalian 116024, China
Abstract
Arterial signal coordination requires coupled decisions on cycle length, phase green times, and offsets in an explicit timing plan. This study presents a modular explicit coordinated signal timing optimization (ECSTO) method for a three-intersection arterial. At each cycle-level decision, a soft actor–critic (SAC) agent selects a common background cycle length and offset corrections for the upcoming cycle. A time–space relationship provides base offsets. Modified model predictive control (MMPC) then allocates phase green times using non-negative, intersection-specific cycle adjustments derived from changes between consecutive offset corrections. It does so without a separately scheduled explicit transition cycle. SAC is trained with a phase-weighted discharged-vehicle reward. System vehicle delay and coordinated-direction space-mean speed are used to evaluate and select policies. Across 1000 common demand scenarios in the simulated arterial, ECSTO had 7.16–11.73% lower aggregate mean system vehicle delay and 5.30–10.19% lower total vehicle travel time than five selected explicit timing-control baselines. These reductions remained statistically significant after Holm adjustment. For coordinated-direction space-mean speed, its mean was significantly below four baselines, illustrating a trade-off between system-wide efficiency and coordinated-direction speed. Ablation results describe differences among four tested timing configurations. The findings are limited to this three-intersection simulation and do not establish field performance or effects in larger networks.
Related
- Coordinated Lane-Level Variable Speed Limits and Ramp Metering for Successive Weaving Segments Considering Merging/Diverging Risks: A Hybrid Model Predictive Control and Multi-Agent Reinforcement Learning Approach
- Health-aware optimal charging of lithium-ion batteries using deep-neural networks-based explicit constrained model predictive control
- Towards Sustainable Hydrogen Systems: Supply Chain Optimization with Model Predictive Control and Reinforcement Learning
- Distributed electricity–carbon co-optimization of hydrogen energy storage microgrids with dynamic laplace perturbation and model predictive control
- Coordinated Optimization of Train Timetables and Maintenance Windows for Plateau Railways
BibTeX
@article{zhang2026explicit,
title = {{Explicit Coordinated Signal Timing Optimization Using Soft Actor–Critic and Model Predictive Control Along an Arterial}},
author = {Kun Zhang and Helai Huang and Hanchu Zhou and Junwu Liu and Qiming Zheng and Hongfeng Xu},
journal = {Mathematics},
year = {2026},
month = nov,
doi = {10.3390/math14214060},
url = {https://www.mdpi.com/2227-7390/14/21/4060},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Explicit Coordinated Signal Timing Optimization Using Soft Actor–Critic and Model Predictive Control Along an Arterial
AU - Kun Zhang
AU - Helai Huang
AU - Hanchu Zhou
AU - Junwu Liu
AU - Qiming Zheng
AU - Hongfeng Xu
PY - 2026
DA - 2026/11/05
JO - Mathematics
DO - 10.3390/math14214060
UR - https://www.mdpi.com/2227-7390/14/21/4060
AB - Arterial signal coordination requires coupled decisions on cycle length, phase green times, and offsets in an explicit timing plan. This study presents a modular explicit coordinated signal timing optimization (ECSTO) method for a three-intersection arterial. At each cycle-level decision, a soft actor–critic (SAC) agent selects a common background cycle length and offset corrections for the upcoming cycle. A time–space relationship provides base offsets. Modified model predictive control (MMPC) then allocates phase green times using non-negative, intersection-specific cycle adjustments derived from changes between consecutive offset corrections. It does so without a separately scheduled explicit transition cycle. SAC is trained with a phase-weighted discharged-vehicle reward. System vehicle delay and coordinated-direction space-mean speed are used to evaluate and select policies. Across 1000 common demand scenarios in the simulated arterial, ECSTO had 7.16–11.73% lower aggregate mean system vehicle delay and 5.30–10.19% lower total vehicle travel time than five selected explicit timing-control baselines. These reductions remained statistically significant after Holm adjustment. For coordinated-direction space-mean speed, its mean was significantly below four baselines, illustrating a trade-off between system-wide efficiency and coordinated-direction speed. Ablation results describe differences among four tested timing configurations. The findings are limited to this three-intersection simulation and do not establish field performance or effects in larger networks.
SN - 2227-7390
ER -