重型叉车群智能体控制与自适应协同优化方法研究

Research on Multi-Agent Control and Adaptive Collaborative Optimization for Heavy-Duty Forklift Groups

摘要: 针对工业物流场景下重型叉车群在重载工况、变参数扰动及复杂动态环境下的作业效率低、协同安全性差等挑战，本文提出了一种基于多智能体系统(MAS)的自适应协同优化控制架构。首先，建立了计入负载偏移与阿克曼转向约束的高保真运动学与动力学模型，并引入Pacejka魔术公式描述轮胎非线性力学特性。其次，针对重载干扰设计了一种结合非线性扰动观测器(NDOB)与自适应滑模控制(ASMC)的数据驱动控制算法，有效抑制了负载波动带来的建模误差与控制抖振。在群体协同层面，提出一种混合式协同架构，利用负载感知粒子群算法(LA-PSO)实现多目标任务分配，并结合预测型人工势场法(P-APF)与分布式一致性协议解决了动态冲突消解与速度同步问题。仿真结果表明，该方法在保障重载工况稳定性的前提下，显著提升了群系统的路径跟踪精度与协同通量，为工业现场大规模叉车群自主作业提供了理论支撑。

Abstract: To address the challenges of low operational efficiency and poor collaborative safety for heavy-duty forklift groups in complex dynamic environments under heavy-load conditions and variable parameter disturbances, this paper proposes an adaptive collaborative optimization control architecture based on multi-agent Systems (MAS). Firstly, a high-fidelity kinematic and dynamic model accounting for load offset and Ackermann steering constraints is established, with the Pacejka Magic Formula integrated to describe nonlinear tire force characteristics. Secondly, for heavy-load interference, a data-driven control algorithm combining a Nonlinear Disturbance Observer (NDOB) and Adaptive Sliding Mode Control (ASMC) is designed to effectively suppress modeling errors and control chattering caused by load fluctuations. At the swarm coordination level, a hybrid collaborative architecture is proposed, utilizing a Load-Aware Particle Swarm Optimization (LA-PSO) for multi-objective task allocation, and combining a Predictive Artificial Potential Field (P-APF) method with distributed consensus protocols to solve dynamic conflict resolution and velocity synchronization. The simulation results demonstrate that the proposed method significantly improves path-tracking accuracy and collaborative throughput while ensuring stability under heavy-load conditions, providing a theoretical foundation for the autonomous operation of large-scale forklift groups in industrial settings.