采煤机是综采工作面的核心装备，复杂煤层条件下，其工况恶劣、环境复杂，采掘装备智能化程度不高，导致我国煤矿开采灾害多、煤机适应性不强、故障率高、效率低，提高煤机装备的可靠性与适应性是煤矿智能化发展的主要任务之一。采煤机工作机构与复杂煤层耦合作用机理及煤岩截割状态与动力传递系统的导控机制，是实现采煤机智能高效截割的关键。基于虚拟样机技术、模糊控制技术，结合数据自适应加权融合算法、深度强化学习算法，采用多领域建模与协同仿真及试验分析相结合的方法，构建机-电-液-控一体化的采煤机自适应截割系统模型，研究其自适应截割控制策略。利用AMEsim建立调高液压系统模型，并与EDEM-RecurDyn煤岩截割双向耦合模型集成；根据煤层实际赋存条件划分煤岩坚固性系数等级范围，以采煤机综合性能最优为目标，利用改进的MOGWO（Multi-Objective Grey Wolf Optimizer）算法对采煤机的牵引速度和滚筒转速进行分级优化。以采煤机截割部的时域振动信号作为煤岩截割状态识别的特征参数，运用数据自适应加权融合算法对其进行融合处理；以特征参数融合值为依据利用模糊控制实现煤岩截割状态的智能识别；利用Simulink搭建基于深度确定性策略梯度算法DDPG（Deep Deterministic Policy Gradient）的采煤机牵引速度-滚筒转速（vq-n）协同调速和自适应调高控制系统模型，利用接口技术实现EDEM-RecurDyn-AMEsim-Simulink耦合，构建机-电-液-控一体化的采煤机自适应截割控制系统模型并进行仿真。研究结果表明：系统以煤岩截割仿真数据流为主线，能够实现对煤岩截割动态过程的感知分析、信号特征处理和自适应调节的决策控制。利用物理试验验证了基于EDEM-RecurDyn耦合仿真的可行性与结果可靠性；在保证煤机综合性能最优且动态可靠的前提下，当螺旋滚筒位于上位，且识别到煤岩体坚固性系数f>7时，首先按滚筒截顶工况界定，采用vq-n协同调速及自适应调高控制策略，并可根据调高过程中采样时间（2 s）内滚筒截割阻力方向振动加速度波动的变化趋势，进一步判断其是处于截顶亦或截割坚硬煤岩层或硬结核（f>7且非顶底板），若识别结果为后者或煤岩体f≤7时，仅采用vq-n协同调速策略；当识别到煤岩体坚固性系数f值减小的工况时，选用vq-n同时调控策略可全面考虑采煤机各性能指标；当识别到煤岩体坚固性系数f值增大的工况时，为保证采煤机的动态可靠性，选用牵引速度优先于滚筒转速的顺序调控策略，其相比于同时调控策略能够使滚筒受载降低23.7%、载荷波动减小28.1%；仿真过程验证了系统能够按照预期的调控策略对采煤机牵引速度、滚筒转速及滚筒高度进行精准调控，最长仅经0.64 s即能感知到截割工况的变化，具有调节的实时性和响应的快速性，实现了复杂煤层条件下的采煤机自适应截割，并通过物理试验验证了所搭建的采煤机自适应截割控制系统及仿真结果的正确性，可有效提高采煤机对复杂煤层的适应性。

Shearer is the core equipment of fully mechanized mining face. Under some complex coal seam conditions, the working conditions of shearer are harsh, the environment is complex, and the mining equipment is not very intelligent. Those have led to many coal mining disasters, and the poor adaptability, high failure rate and low efficiency of shearer. Improving the reliability and adaptability of coal mining equipment is one of the main tasks for the intelligent development of coal mines. The coupling mechanism between the working mechanism of shearer and the complex coal seam, and the guidance and control mechanism of the cutting state of coal and rock and the power transmission system are the keys to realize the intelligent and efficient cutting of shearer. Based on virtual prototype technology, fuzzy control technology, combined with data adaptive weighted fusion algorithm and deep reinforcement learning algorithm, the method of combining multi-domain modeling with collaborative simulation and experimental analysis is used to build a machine-electric-hydraulic control integrated self adaptive cutting system model for shearer and investigate the self-adaptive cutting control strategy of shearer. The AMEsim is used to build a height-adjusting hydraulic system model which integrates with the EDEM-RecurDyn coal and rock cutting bi-directional coupling model. According to the actual occurrence conditions of coal seams, the range of coal rock solidity coefficient grades is divided. Aiming to optimize the overall performance of shearer, the improved MOGWO algorithm is used to optimize the traction speed and drum speed of shearer. The time-domain vibration signal of the shearer cutting part is used as the characteristic parameter of coal and rock cutting state recognition, and the data adaptive weighted fusion algorithm is used for fusion processing. Based on the fusion value of characteristic parameters, the fuzzy control is used to realize the intelligent recognition of the cutting state of coal and rock. Using Simulink, a shearer traction speed-drum speed (vq-n) coordinated speed regulation and self-adaptive height control system model is built based on the DDPG. Using the interface technology to realize EDEM-RecurDyn-AMEsim-Simulink coupling, a machine-electric hydraulic-control integrated shearer self-adaptive cutting control system model is constructed and the simulation is performed. The research results show that the system takes the coal and rock cutting simulation data stream as the main focus, and can realize the perceptual analysis of the coal and rock cutting dynamic process, signal feature processing and self-adaptive adjustment decision-making control. Physical experiments are used to verify the feasibility and reliability of the results based on the EDEM-Recur Dyn coupling simulation. Under the premise of ensuring the optimal overall performance of the shearer and its dynamic reliability, when the spiral drum is located at the upper position and the firmness coefficient of coal and rock mass f>7 is identified, the vq-n cooperative speed regulation and adaptive height regulation control strategy are adopted according to the working conditions of cutting the roof by the drum. According to the variation trend of vibration acceleration fluctuation in the direction of drum cutting resistance within the sampling time (2 s) during the height adjustment process, it can be further assessed whether it is in the state of cutting the roof, cutting the hard coal and rock or hard nodules (f>7 and not the top and bottom plate). If the identification result is the latter or the firmness coefficient of coal and rock mass f≤7, only the vq-n cooperative speed regulation strategy is adopted. When the coal and rock mass firmness coefficient f value is reduced, the selection of the vq-n simultaneous control strategy can fully consider the performance indicators of the shearer. When the coal and rock mass firmness coefficient f value is increased, in order to ensure the dynamic reliability of the shearer, the sequential control strategy in which the traction speed takes precedence over the drum speed is selected. Compared with the simultaneous control strategy, the drum load can be reduced by 23.7% and the load fluctuation can be reduced by 28.1%. The simulation process verifies that the system can accurately adjust the shearer traction speed, drum speed and drum height according to the expected control strategy. The change in cutting conditions can be sensed after only 0.64 s, and the adjustment is real-time showing a rapidity of response. The technology realizes the shearer self adaptive cutting under complex coal seam conditions. The study verifies the correctness of the built shearer self-adaptive cutting control system and simulation results through physical tests, which can effectively improve the shearer’s adaptability to operate in complex coal seams.

1 采煤机自适应截割系统架构

2 采煤机自适应截割控制系统多领域建模

   2.1 采煤机自适应截割机-液系统模型构建

   2.2 EDEM煤壁模型构建

   2.3 采煤机刚柔耦合虚拟样机截割煤壁双向耦合模型构建

   2.4 双向耦合仿真可行性验证

3 采煤机自适应截割最优决策与控制

   3.1 基于改进的MOGWO算法的最优决策方案

   3.2 基于模糊控制的煤岩截割状态识别

   3.3 采煤机自适应截割控制系统模型构建

   3.4 自适应截割控制策略制定

4 采煤机自适应截割系统性能分析

   4.1 采煤机自适应截割工况设计

   4.2 自适应截割仿真及性能分析

5 试验验证

6 结论