目前，在国家“双碳”战略目标的驱动下，煤炭行业持续推进数字化转型和智能化升级。悬臂式掘进机作为掘进工作面的核心设备，其定向掘进主要依靠操作人员通过手动方式进行纠偏作业，这种方式不仅安全性与可靠性较差，而且极易造成超挖、欠挖现象，难以保证巷道成形质量。因此，实现掘进机的自主纠偏和智能控制，不仅是提高掘进效率的关键，更是贯彻“少人则安”生产理念的必然要求。

论文围绕目前掘进机智能化程度不高、人员操作依赖性强等问题，提出了一种数字孪生驱动的掘进机自主纠偏控制总体方案，通过研究轨迹跟踪控制与阀控马达调速控制技术，实现了掘进机的纠偏控制。在此基础上，将循迹控制与数字孪生技术进行有机融合，开发了数字孪生驱动的掘进机自主纠偏控制系统，在虚拟空间中实现远程监测与纠偏控制指令的生成与下发，有效提升了巷道掘进的智能化水平。

根据掘进机的结构特点以及运动约束，对掘进机行走部进行运动分析与数学建模。基于掘进机差速转向原理，采用瞬时转动中心的方法研究了机身转向机制，建立了表征掘进机机身运动规律的运动学和动力学模型，同时对掘进机液压驱动系统进行了原理分析与数学建模，为后续轨迹跟踪、阀控马达调速等控制问题的研究奠定了基础。

针对人工控制掘进机纠偏对中存在可靠性差、安全性低等问题，提出了基于双层协同控制策略的掘进机轨迹纠偏控制方法。在上层位置控制层面，采用改进滑模控制的轨迹跟踪控制方法，通过设计一种新型趋近律，并结合边界层法实现了系统偏差的快速收敛以及抖振现象的有效抑制。由理论分析验证了该趋近律的存在性、可达性以及稳定性，同时推导了干扰稳态误差界。在下层速度控制层面，考虑到常规PID难以满足掘进机液压驱动系统的控制需求，采用改进RBF-PID的阀控马达调速控制方法作为下层速度控制。通过动量优化算法对梯度更新进行优化，并在此基础上提出了学习速率和动量因子自适应调整算法用于改进神经网络学习过程，解决了网络收敛速度慢以及局部极值的问题，同时为确保被控系统稳定，对控制参数的调节范围进行了限制。最后通过仿真分析，验证了所提方法的有效性。

针对视频监控与人工跟机相结合的远程监测与控制方式存在可视化差、效率低等问题，开发了掘进机孪生驱动控制系统平台。构建了掘进机自主纠偏控制数字孪生模型，通过“虚拟空间建立、物理空间状态感知、孪生数据交互”等步骤，将掘进工作面和掘进机的实时状态映射至数字化的虚拟空间中，实现环境与设备状态的可视化监测。并在系统平台中嵌入轨迹跟踪控制算法，由虚拟端实现纠偏控制指令的生成和下发，同时利用物理空间传感器数据驱动虚拟空间同步变化，实现智能化远程控制。

最后，搭建实验平台，分别对系统平台主要功能、掘进机纠偏控制性能以及掘进机行走速度控制性能进行测试与验证。结果表明，系统平台能够实现指令下发、数据上传等远程控制功能，在此基础上验证了改进滑模控制方法对于掘进机纠偏控制的可行性，且系统在运行过程中能够保证虚实空间的同步性与一致性；改进RBF-PID控制方法能够使掘进机行走速度较快响应给定速度值，满足速度控制需求。该研究为掘进机智能化纠偏控制提供了新的思路。

At present, driven by the national "dual carbon" strategic goal, the coal industry continues to promote digital transformation and intelligent upgrading. As the core equipment of the tunneling face, the boom-type roadheader relies mainly on manual correction by operators for directional tunneling. This method not only has poor safety and reliability, but also easily causes over-excavation and under-excavation, making it difficult to ensure the quality of roadway formation. Therefore, achieving autonomous correction and intelligent control of roadheader is not only the key to improving tunneling efficiency, but also an inevitable requirement for implementing the production concept of "less people, more safety".

The paper addresses the current issues of low intelligence and strong dependence on human operation in roadheader tunneling machines, and proposes a digital twin-driven autonomous correction control system for roadheader tunneling machines. By studying trajectory tracking control and valve-controlled motor speed control technology, the correction control of roadheader tunneling machines is achieved. On this basis, the tracking control and digital twin technology are organically integrated to develop a digital twin-driven autonomous correction control system for roadheader tunneling machines, which enables remote monitoring and generation and issuance of correction control instructions in virtual space, effectively improving the intelligence level of roadway tunneling.

 Based on the structural characteristics and motion constraints of the roadheader, this paper conducts a motion analysis and mathematical modeling of the roadheader's traveling unit. Utilizing the principle of differential speed steering, the mechanism of the machine's body steering is studied through the method of instantaneous centers of rotation. This establishes kinematic and dynamic models that characterize the movement laws of the roadheader's body. Additionally, a principled analysis and mathematical modeling of the roadheader's hydraulic drive system are conducted, laying a foundation for subsequent research on trajectory tracking, valve-controlled motor speed regulation, and other control issues.

Addressing issues such as poor reliability and low safety in manual control for deviation correction in roadheaders, a trajectory deviation correction control method based on a dual-layer collaborative control strategy is proposed. In the upper-level position control, a trajectory tracking control method using improved sliding mode control is adopted. By designing a novel reaching law combined with the boundary layer method, rapid convergence of system deviations and effective suppression of chattering phenomena are achieved. Theoretical analysis verifies the existence, reachability, and stability of this reaching law, while deriving the steady-state error bound for disturbances. In the lower-level speed control, considering that conventional PID is difficult to meet the control requirements of the roadheader's hydraulic drive system, an improved RBF-PID valve-controlled motor speed control method is adopted. The gradient update is optimized through a momentum optimization algorithm, and on this basis, an adaptive adjustment algorithm for learning rate and momentum factor is proposed to improve the neural network learning process. This addresses issues such as slow network convergence and local extrema. To ensure the stability of the controlled system, the adjustment range of control parameters is limited. Finally, simulation analysis verifies the effectiveness of the proposed method.

To address issues such as poor visualization and low efficiency in the remote monitoring and control method that combines video monitoring with manual machine tracking, a twin-driven control system platform for roadheaders has been developed. A digital twin model for autonomous deviation correction control of the roadheader has been constructed. Through steps such as "virtual space establishment, physical space state perception, and twin data interaction," the real-time status of the excavation face and the roadheader is mapped into a digital virtual space, enabling visual monitoring of the environment and equipment status. Additionally, a trajectory tracking control algorithm is embedded in the system platform, allowing the virtual end to generate and issue deviation correction control commands. At the same time, sensor data from the physical space drives synchronous changes in the virtual space, enabling intelligent remote control.

Finally, an experimental platform was set up to test and verify the main functions of the system platform, the deviation correction control performance of the roadheader, and the walking speed control performance of the roadheader. The results show that the system platform can achieve remote control functions such as command issuance and data upload. Based on this, the feasibility of the improved sliding mode control method for deviation correction control of the roadheader was verified, and the system can ensure the synchronization and consistency of the virtual and physical spaces during operation. The improved RBF-PID control method enables the roadheader to quickly respond to the given speed value and meet the speed control requirements. This research provides a new approach for intelligent deviation correction control of roadheaders.