Adaptive prediction and cooperative control of mixed vehicle groups at intersections under time-varying delay and false data injection attacks

The inevitable presence of non-ideal communication conditions in open wireless communication networks leads to insufficient coordination and optimization among mixed traffic flows during cloud-based centralized cooperative scheduling and control in intersection systems, thereby compromising traffic safety and operational efficiency. This situation imposes heightened requirements for coordinated operation between mixed vehicle platoons composed of conventional Human-Driven Vehicles (HDVs) and Intelligent Connected Vehicles (ICVs) under imperfect communication conditions. To address these challenges, this paper proposes a multi-vehicle model-free adaptive prediction and cooperative control method for intersection scenarios that considers time-varying delays and false data injection attacks. By employing a compact-form dynamic linearization prediction scheme and an enhanced model-free receding horizon predictive control technique, we construct a networked adaptive prediction and cooperative controller for mixed vehicle clusters. This framework enables the estimation and compensation of multi-vehicle state information within mixed vehicle groups, followed by the distribution of calculated adjustment speeds to target controlled ICVs. The proposed approach guides mixed vehicle platoons to achieve dynamic formation adjustments in response to environmental variations while enhancing coordinated control performance under time-varying delays and false data attacks through regulation of vehicle queue speeds. Simulation results verify that this methodology effectively improves the traffic efficiency at intersections based on safety.