露天矿卡平行自主运输系统架构与方法

Architecture and Methodology for Parallel Autonomous Haulage System in Open-pit Mines

  • 摘要: 露天矿运输系统具有设备多样、作业点分布、作业流程时空耦合、环境动态变化等特性,导致其高效管理与控制存在挑战。此外,极端工况下系统测试困难以及特定场景的感知数据不足,使系统性能难以全面评估,导致决策与控制策略对场景的适应性差,泛化能力弱,运行稳定性与安全性难以保障,从而影响系统整体可靠性。为解决上述问题,本文提出一种基于平行理论与车路云融合控制理念的露天矿平行自主运输系统,旨在提升运输系统的安全性、高效性,降低运行成本的同时实现常态化运行。该系统由物理系统、工业互联网和人工系统三部分构成。其中,人工系统采用边缘云、矿山云和中心云三级云控架构,通过云控基础平台与云控应用平台实现多层级协同控制。云控基础平台是云计算中心,可保障各级云控应用对实时性与服务范围的多样化需求。云控应用平台包括资源库以及面向生成场景的平行学习与平行协同。资源库由模型库和算法库构成,实现建模方法选取和场景定义,为算法训练与协同控制提供方法支撑。通过虚实结合的平行训练与平行测试,平行学习构建了人工系统和实际运输系统之间的数据闭环,从而解决算法数据不足和极端场景测试困难问题,推动系统自演化。基于此,平行协同基于三级云控架构与多智能体技术,实现协同感知、态势预测和分层分布式协同控制,有效应对多设备协同管理复杂与实时控制难题。面向复杂多样的露天矿开采环境,所提露天矿平行自主运输系统具有更好的适应性、稳定性与安全性,为解决露天矿运输中的核心难题提供了有效解决方案。

     

    Abstract: Abstract: In the haulage system of open-pit mines, various equipment worked in different operating positions work have spatial and temporal coupling relationship, and the environment changes over time. These pose the challenge in efficient transportation management and control. Furthermore, testing under extreme working conditions is difficult, and scenario-specific sensing data is insufficient, which hinder comprehensive performance evaluation of this system. Consequently, decision-making and control strategies have poor adaptability to variousscenarios, and the operation stability and safety can not be guaranteed. That is, this system has weak reliability and generalization. To address these issues, parallel autonomous haulage system based on parallel theory and vehicle-road-cloud fusion control is proposed for open-pit mines, with the purpose of improving transportation safety and efficiency, reducing costs, and realizing normalized operation. It consists of physical and artificial systems, as well as industrial internet. The artificial system adopts a three-level cloud control architecture, including edge cloud, mine cloud, and center cloud, which realizes multi-level collaborative control through the cloud control foundation platform and cloud control application platform. The cloud control foundation platform, as the cloud computing center, meets diverse real-time and service demands on cloud control applications at all levels. The cloud control application platform includes a resource library, parallel learning and parallel collaboration for the generated scenarios. The resource library consists of model one and an algorithm one that select modeling method and define scenarios, respectively, with the purpose of providing support for algorithm training and cooperative control. Through parallel training and testing based on virtuality-reality fusion, parallel learning realizes the closed-loop interaction of data between the artificial system and actual transportation system, promoting self-evolution of this system. To handle the complex management and real-time control of multi-equipment collaboration, parallel collaboration realizes collaborative perception, situational prediction and hierarchical distributed collaborative control based on three-level cloud control architecture and multi-agent technology. For the complex and various environment of open-pit mines, the proposed parallel autonomous haulage system has better adaptability, stability and safety, providing an effective solution for their haulage systems.

     

/

返回文章
返回