郭卫 贺海涛 赵青青 赵栓峰 王渊

摘 要：針对综采煤流控制过程可以归结为复杂非线性开放系统，无法采用单一数学模型进行建模的特点，提出了采用“Petri网模型+基于数据驱动模型+在线更新”双轮驱动的方式建立综采煤流负荷平衡自适应控制系统模型。通过研究煤矿综采工作面不同工况下的“三机”运动状态，特别是采煤机空间运动规律以及“三机”联动过程，建立了煤流自适应控制模型参数随环境变化的参数辨识方法及其控制模型，突破了综采面煤流负荷平衡的协同耦合动态更新的瓶颈，解决了传统数学模型处理参数计算量大、速度慢、效率低的缺点，更好地适应了开采环境的变化，可以更加精确的捕捉开采异常状态下的特征，提高综采煤流负荷平衡系统的响应特性。为了验证基于Petri网的煤流平衡自适应控制方法的有效性，利用在榆家梁煤矿采集的采煤机及刮板输送机的状态数据，实时分析现场操作工况下的作业参数并进行模拟，并在实验中进行验证。最后将加入Petri网模型与没有加入Petri网的模型处理参数的误差率进行了实验对比。结果表明煤流负荷平衡的动态自适应控制模型预测结果良好，增强了煤流平衡自适应控制系统的预测泛化能力。关键词：深度学习;Petri网;综采煤流;自适应控制;采煤机;数据驱动中图分类号：TD 712

文献标志码：A

文章编号：1672-9315（2020）06-01071-10

DOI：10.13800/j.cnki.xakjdxxb.2020.0618开放科学（资源服务）标识码（OSID）：

Construction of an adaptive control system for fully mechanized

coal flow load balance based on intelligent Petri net

GUO Wei1，HE Hai-tao1，2，ZHAO Qing-qing2，ZHAO Shuan-feng2，WANG Yuan2

（1.College of Mechanical and Engineering，Xian University of Science and Technology，Xian 710054，China;

2.State Energy Group Shendong Coal Group Company，Shenmu 719315，China）

Abstract：With the idea of “continuous exploration，continuous improvement” and the method of “Petri net model，data-driven model and online update”in view，the “three-machine” movement state under different working conditions of the coal mining face is studied.Especially when the shearer spatial motion law and the “three-machine” linkage model are constructed，the movement state  is discussed

that the model makes timely feedback and controls the parameters with the changes of the environment.the collaborative coupling dynamic adaptive control model for coal flow load balance in fully mechanized mining face was established，which broke through the shortcomings of traditional mathematical model，such as large calculation amount，slow speed and low efficiency.According to the variation of parameters in different environments，it is possible to  determine the operation rules more accurately under the abnormal environment conditions to provide efficient feedback to coal miners and to make them react in time.In order to verify the effectiveness of the coal flow equilibrium adaptive control method based on Petri net，the state data of shearer and scraper conveyor collected in the Yujialiang coal mine was used to analyze，simulate and verify the operating parameters under the field operating conditions in real time.Finally，the error rate of the processing parameters of the model with the Petri nets and the model without the Petri net was compared experimentally.The results indicate that the dynamic adaptive control model of coal flow load balance has a better prediction result，which enhances the prediction generalization ability of coal flow balance adaptive control system.

Key words：deep learning;Petri net;fully mechanized coal mining flow;adaptive control;shearer;data-driven

0 引 言

采煤机、刮板运输机的煤流负荷平衡自适应控制是实现综采面三机（采煤机、刮板输送机、液压支架）协同控制的关键一环[1-4]。……