Fig21OH.

The multi-axle crane, a long vehicle with high inertia, has historically struggled with steering efficiency and path-tracking performance. Various control strategies, including Proportional-Integral-Derivative (PID), Linear Quadratic Regulator (LQR), and Model Predictive Control (MPC), have been employed to address these challenges. However, while improving steering efficiency, these strategies have often led to poor path-tracking performance. This work presents a significant advancement in the form of an optimized MPC for improved steering control of the multi-axle crane. A bicycle model of the multi-axle crane was adopted for the work. MPC was designed, and the smell agent optimization technique (SAO) was employed to optimize the steering input weighting factor, which determines the path-tracking performance. This provided an improved and accurate path-tracking performance for different driving speed conditions. Simulation and performance evaluation of the optimized MPC for the steering system were carried out on a curved road path for three different driving speed scenarios (25, 45, and 65 km/h). The results were compared with existing steering systems that utilized the MPC using steering efficiency, dynamic stability, and path-tracking performance. Results obtained showed improvements of 13.88%, 46.02%, and 18.35% in steering efficiency for the three scenarios over the benchmark scheme. Similarly, improvements of 2.29%, 1.03%, and 4.17%, respectively, were achieved in terms of dynamic stability for the three scenarios. For lateral error, improvements of 26.78%, 26.35%, and 27.52% were achieved, while 27.44%, 29.25%, and 28.93% were achieved for the yaw angle error in the three scenarios, respectively. A 3D simulation model for the multi-axle crane was developed in AnyLogic for visual interpretation and validation of the tracking results. These results showed that the developed MPC steering system achieved better steering performance than the existing scheme.多轴起重机（multi-axle crane）作为一种大惯量长车身车辆，长期以来始终面临转向效率低下与路径跟踪性能不佳的难题。此前已有多种控制策略被用于解决上述难题，包括比例-积分-微分（Proportional-Integral-Derivative，PID）控制器、线性二次调节器（Linear Quadratic Regulator，LQR）以及模型预测控制（Model Predictive Control，MPC）。然而，此类策略虽可提升转向效率，却往往会导致路径跟踪性能恶化。本研究提出一项突破性进展——面向多轴起重机转向控制优化的改进型模型预测控制方案。本研究采用自行车模型（bicycle model）作为多轴起重机的建模基础，设计了模型预测控制算法，并采用气味代理优化技术（Smell Agent Optimization Technique，SAO）对影响路径跟踪性能的转向输入加权因子进行优化。该方案可在不同行驶车速条件下实现更优异且精准的路径跟踪性能。针对该优化型模型预测控制转向系统，本研究在曲线路径上开展了仿真实验与性能评估，涵盖25、45、65 km/h三种不同行驶车速场景。研究以转向效率、动态稳定性及路径跟踪性能为评价指标，将所得结果与现有采用传统模型预测控制的转向系统进行了对比。实验结果表明，相较于基准方案，三种车速场景下的转向效率分别提升了13.88%、46.02%与18.35%；同样，三种场景下的动态稳定性分别实现了2.29%、1.03%与4.17%的提升。横向误差方面，三种场景下的改善幅度分别为26.78%、26.35%与27.52%；横摆角误差方面则分别实现了27.44%、29.25%与28.93%的性能提升。本研究基于AnyLogic平台搭建了多轴起重机三维仿真模型，用于可视化呈现并验证路径跟踪结果。实验结果证实，所提出的模型预测控制转向系统相较现有方案，可实现更优异的转向性能。