综采工作面矿压显现的分析与预测对于复杂地质条件下工作面顶板管理,保证矿井生产安全具有重要意义。采用关系型数据库储存液压支架工作阻力数据以及利用工作面推进过程中矿压显现的时序特性,采用SQL语言,运用长短时记忆网络(Long Short...综采工作面矿压显现的分析与预测对于复杂地质条件下工作面顶板管理,保证矿井生产安全具有重要意义。采用关系型数据库储存液压支架工作阻力数据以及利用工作面推进过程中矿压显现的时序特性,采用SQL语言,运用长短时记忆网络(Long Short Time Memory,LSTM)深度学习方法,以红庆河矿31101大采高综采工作面矿压规律为研究对象,对工作面支架工作阻力、支架不平衡力、支架安全阀开启情况及初次来压与周期来压等矿压显现规律进行分析;基于建立的数据库,预测了红庆河大采高工作面矿山压力,预测结果表明LSTM方法较BP神经网络预测更具准确性。为进一步讨论本研究采用的LSTM网络模型的泛化能力,在采用布尔台42103大采高工作面、上湾矿12401大采高工作面少量矿压数据的前提下,使用迁移学习方法,对矿压数据进行预测检验,结果表明:LSTM模型具有很好的泛化能力,相比于不使用迁移学习方法,迁移学习可提高模型的泛化能力。最后,探讨了模型在3个大采高工作面矿压预测表现的差别,发现数据量本身对模型预测行为影响较大,增大数据量可弥补原始数据缺失等问题。在预测模型基础上设计了周期来压预警模型,集成形成相应矿压分析与预警系统,经工程验证判定预警系统分析效果良好。展开更多
The ongoing need to deliver improved safety, productivity and environmental benefit in coal mining presents an open challenge as well as a powerful incentive to develop new and improved solutions. This paper assesses ...The ongoing need to deliver improved safety, productivity and environmental benefit in coal mining presents an open challenge as well as a powerful incentive to develop new and improved solutions. This paper assesses the critical role that enabling technologies have played in the delivery of remote and automated capability for longwall mining. A brief historical account is given to highlight key technical contributions which have influenced the direction and development of present-day longwall technology. The current state of longwall automation is discussed with particular attention drawn to the technologies that enable automated capability. Outcomes are presented from an independently conducted case study that assessed the impact that CSIRO＇s LASC longwall automation research has made to the longwall mining industry in Australia. Importantly, this study reveals how uptake of this innova- tive technology has significantly benefitted coal mine productivity, improved working conditions for personnel and enhanced environmental outcomes. These benefits have been widely adopted with CSIRO automation technology being used in 60 per cent of all Australian underground operations. International deployment of the technology is also emerging. The paper concludes with future challenges and opportunities to highfight the ongoing scope for longwall automation research and development.展开更多
Coal bursts involve the sudden, violent ejection of coal or rock into the mine workings. They are a particular hazard because they typically occur without warning. During the past 2 years three US coal miners were kil...Coal bursts involve the sudden, violent ejection of coal or rock into the mine workings. They are a particular hazard because they typically occur without warning. During the past 2 years three US coal miners were killed in two coal bursts, following a 6-year period during which there were zero burst fatalities. This paper puts the US experience in the context of worldwide research into coal bursts. It focuses on two major longwall mining coalfields which have struggled with bursts for decades. The Utah experience displays many of the ＂classic＂ burst characteristics, including deep cover, strong roof and floor rock, and a direct association between bursts and mining activity. In Colorado, the longwalls of the North Fork Valley （NFV） also work at great depth, but their roof and floor strengths are moderate, and most bursts have occurred during entry development or in headgates, bleeders, or other outby locations. The NFV bursts also are more likely to be associated with geologic structures and large magnitude seismic events. The paper provides a detailed case history to illustrate the experience in each of these coalfields. The paper closes with a brief discussion of how US longwalls have managed the burst risk.展开更多
文摘综采工作面矿压显现的分析与预测对于复杂地质条件下工作面顶板管理,保证矿井生产安全具有重要意义。采用关系型数据库储存液压支架工作阻力数据以及利用工作面推进过程中矿压显现的时序特性,采用SQL语言,运用长短时记忆网络(Long Short Time Memory,LSTM)深度学习方法,以红庆河矿31101大采高综采工作面矿压规律为研究对象,对工作面支架工作阻力、支架不平衡力、支架安全阀开启情况及初次来压与周期来压等矿压显现规律进行分析;基于建立的数据库,预测了红庆河大采高工作面矿山压力,预测结果表明LSTM方法较BP神经网络预测更具准确性。为进一步讨论本研究采用的LSTM网络模型的泛化能力,在采用布尔台42103大采高工作面、上湾矿12401大采高工作面少量矿压数据的前提下,使用迁移学习方法,对矿压数据进行预测检验,结果表明:LSTM模型具有很好的泛化能力,相比于不使用迁移学习方法,迁移学习可提高模型的泛化能力。最后,探讨了模型在3个大采高工作面矿压预测表现的差别,发现数据量本身对模型预测行为影响较大,增大数据量可弥补原始数据缺失等问题。在预测模型基础上设计了周期来压预警模型,集成形成相应矿压分析与预警系统,经工程验证判定预警系统分析效果良好。
文摘The ongoing need to deliver improved safety, productivity and environmental benefit in coal mining presents an open challenge as well as a powerful incentive to develop new and improved solutions. This paper assesses the critical role that enabling technologies have played in the delivery of remote and automated capability for longwall mining. A brief historical account is given to highlight key technical contributions which have influenced the direction and development of present-day longwall technology. The current state of longwall automation is discussed with particular attention drawn to the technologies that enable automated capability. Outcomes are presented from an independently conducted case study that assessed the impact that CSIRO＇s LASC longwall automation research has made to the longwall mining industry in Australia. Importantly, this study reveals how uptake of this innova- tive technology has significantly benefitted coal mine productivity, improved working conditions for personnel and enhanced environmental outcomes. These benefits have been widely adopted with CSIRO automation technology being used in 60 per cent of all Australian underground operations. International deployment of the technology is also emerging. The paper concludes with future challenges and opportunities to highfight the ongoing scope for longwall automation research and development.
文摘Coal bursts involve the sudden, violent ejection of coal or rock into the mine workings. They are a particular hazard because they typically occur without warning. During the past 2 years three US coal miners were killed in two coal bursts, following a 6-year period during which there were zero burst fatalities. This paper puts the US experience in the context of worldwide research into coal bursts. It focuses on two major longwall mining coalfields which have struggled with bursts for decades. The Utah experience displays many of the ＂classic＂ burst characteristics, including deep cover, strong roof and floor rock, and a direct association between bursts and mining activity. In Colorado, the longwalls of the North Fork Valley （NFV） also work at great depth, but their roof and floor strengths are moderate, and most bursts have occurred during entry development or in headgates, bleeders, or other outby locations. The NFV bursts also are more likely to be associated with geologic structures and large magnitude seismic events. The paper provides a detailed case history to illustrate the experience in each of these coalfields. The paper closes with a brief discussion of how US longwalls have managed the burst risk.
