小脑神经网络在矿工疲劳监测控制系统中的应用

1.西安科技大学 管理学院,陕西 西安 710054; 2.西安科技大学 能源学院,陕西 西安 710054; 3.西安科技大学 安全科学与工程学院,陕西 西安 710054; 4.西安科技大学 电气与控制工程学院,陕西 西安 710054

小脑神经网络; 矿工疲劳水平; 疲劳监测信号体系; 疲劳监测与控制系统; 多元信息融合

Miner fatigue monitoring and control system based on cerebella model articulation controller neutral network
LI Hong-xia1,HUANG Yi-xin2,TIAN Shui-cheng3,HOU Yuan-bin4

(1.College of Management,Xi'an University of Science and Technology,Xi'an 710054,China; 2.College of Energy Science and Engineering,Xi'an University of Science and Technology,Xi'an 710054,China; 3.College of Safety Science and Engineering,Xi'an University of Science and Technology,Xi'an 710054,China; 4.College of Electrical and Control Engineering,Xi'an University of Science and Technology,Xi'an 710054,China)

cerebella model articulation controller; miner fatigue level; fatigue monitoring signal system; fatigue monitoring and control system; multiple information fusion

DOI: 10.13800/j.cnki.xakjdxxb.2018.0314文章编号: 1672-9315(2018)03-0443-09

备注

针对矿工井下疲劳监测与控制难题,探索矿工疲劳务工的生理指标状况,降低矿工疲劳状态下务工的可能性,指导煤矿高层管理者对矿工疲劳生产的管理与控制,并为新设计的矿工疲劳监测控制系统提供新的方法与依据。采用多元信息融合技术,提取了皮电、肌电、脑电等12个疲劳水平监测信号,对矿工务工期间进行半接触式肢体图像和信号分析,构建了矿工疲劳监测信号体系; 结合CMAC神经网络理论,提出矿工疲劳务工期间的疲劳检测控制系统模型,并通过收集矿工务工期间的多生理指标对构建的模型进行验证。结果 表明:矿工疲劳监测与控制系统监测结果与实际矿工疲劳水平的平均偏差为0.2,偏差为0.1~0.3的约占51.7%,0.4~0.7的约占32.2%,0.8~1.0的约占16.1%,比传统的单一信号监测精度更高,偏差更小,更能满足实际生产中的矿工监测与控制的精度需求。为缓解当前矿工疲劳生产的现状,依据矿工疲劳监测结果,结合金融风险管理理论,应采用的方法是:当矿工疲劳水平超出生理负荷值时,机械自启强迫停止功能; 当矿工疲劳水平严重时,提醒班组长强迫矿工换班(岗); 当矿工出现疲劳迹象时,报警提示矿工调整自身疲劳状态等。为后期中国煤矿人员安全与高效生产的电子监测与控制领域提供一种新的设计参考。

This paper is aiming to solve the difficulty regarding to a detection and control on fatigue which probably happened on underground coal miners.The paper explores the miners' physiological indicators under a circumstance of their fatigue working in order to reduce its possibility.Possibly being a guideline for senior mining managers,the paper is also hoped to provide new methods and supports for an updated designed Miner Fatigue Detection and Control System.GSR,EMG,EEG and other 12 fatigue level monitoring signals is extracted based on Multiple Information Fusion Theory.A miner fatigue detection signal system has been build through an analysis of semi-contact limb image and signal during miners' work.Moreover,the Cerebella Model Articulation Controller(CMAC)Neural Network Theory is applied to construct a model referring to fatigue detection and control system.The model has been verified through all the data and MATLAB stimulation analysis.It is shown that the average deviation between the detection results,given by the model system,and the actual fatigue level is 0.2.A deviation from 0.1 to 0.3 accounts for approximately 51.7%.0.4 to 0.7 accounts for approximately 32.2%,and 0.8 to 1.0 accounts for approximately 16.1%.With a higher accuracy and a smaller deviation,the updated system is more likely to satisfy an accuracy requirement for miner fatigue detection and control than the traditional single signal system in practical production.According to the detection result and financial risk management principles,the methods are suggested as follows:Equipment will be forced to shut down automatically when miner's fatigue exceeds his physiological limitation.Group leaders will be reminded to command a compulsive shift change when serious fatigue occurs on miners.Adjustments will be alarmed when signs of fatigue appeared on miners.Hoping this paper enables to provide a new design reference for the field of electronic detection and control,which is significant to the safety and high-efficiency production of miners in China.