矿井胶带火灾通常由胶带摩擦、电火花和明火等原因造成，具有发展速度快、危害性强及扑灭难等特点。传统胶带火灾检测技术主要根据传感器实时采集的环境参数，再通过环境参数阈值来进行分析和判断，存在数据利用率低、准确率不高和难以判断火灾危险程度等问题。为尽早判别火情并及时诊断其危险程度和发展趋势，本文通过数据挖掘等方法研究多参数传感器监测数据，将机器学习算法应用于矿井胶带火灾危险程度判定与趋势预测，致力于研发高准确率、低漏报率的胶带火灾检测应用软件。

针对传统胶带火灾危险程度判定方法漏报率和误报率高的问题，通过改进降噪自动编码器，提出了一种长短时记忆降噪自编码神经网络（LSTM-DAE）的火灾危险程度判定方法。该方法结合了自动编码器和长短期记忆网络的优点，同时在解码器内引入注意力机制，提高了特征提取的有效性。实验仿真结果表明，所提出的方法与传统的神经网络危险等级判定方法相比准确率更高，达到了90.1%。该判定方法在矿井胶带火灾检测过程具有重要的应用价值。

针对传统胶带火灾趋势预测方法存在趋势预测精度低和算法模型训练过程中人为调参困难的问题，提出了一种基于粒子群优化的卷积双向长短期记忆（PSO-CNN-Bi-LSTM）胶带火灾趋势预测方法。同时采用t-SNE方法旨在降低训练的难度。实验结果表明改进后的Bi-LSTM预测模型均方误差降低了15.5%，预测精度大幅度提高。

通过使用SpringBoot框架结合LSTM-DAE与PSO-CNN-Bi-LSTM模型设计了胶带火灾危险程度判定和趋势预测系统软件。实现了胶带运输巷道环境参数的实时采集与监测，并可以准确进行胶带火灾报警预警。经过测试验证了系统软件的功能和性能，证明了该软件在保障矿井胶带运输巷道安全方面具有极高的应用价值。

Mine belt fires are usually caused by belt friction, electric sparks and open flames. They have the characteristics of fast development, strong hazards and difficult to extinguish. The traditional belt fire detection technology mainly analyzes and judges according to the environmental parameters collected by the sensor in real time, and then analyzes and judges by the environmental parameter threshold. In order to identify the fire situation as early as possible and diagnose its danger degree and development trend in time, this paper studies the monitoring data of multi-parameter sensors through data mining and other methods, and applies the machine learning algorithm to the determination and trend prediction of the fire danger degree of mine belts, and is committed to the research and development of high accuracy. , Low false alarm rate of belt fire detection application software.

Aiming at the problem of high false alarm rate and high false alarm rate in the traditional method for determining the fire danger degree of belts, a long-short-term memory noise reduction autoencoder neural network (LSTM-DAE) is proposed to determine the fire danger degree by improving the noise reduction autoencoder. method. This method combines the advantages of autoencoders and long short-term memory networks, while introducing an attention mechanism within the decoder to improve the effectiveness of feature extraction. The experimental simulation results show that the proposed method has a higher accuracy than the traditional neural network risk level judgment method, reaching 90.1%. The determination method has important application value in the process of mine belt fire detection.

Aiming at the problems of low trend prediction accuracy and difficulty in manual parameter adjustment in the process of algorithm model training in traditional belt fire trend prediction methods, a convolutional bidirectional long short-term memory (PSO-CNN-Bi-LSTM) belt based on particle swarm optimization was proposed. Fire trend prediction methods. At the same time, the t-SNE method is adopted to reduce the difficulty of training. The experimental results show that the mean square error of the improved Bi-LSTM prediction model is reduced by 15.5%, and the prediction accuracy is greatly improved.

By using SpringBoot framework combined with LSTM-DAE and PSO-CNN-Bi-LSTM model, the software of belt fire danger degree judgment and trend prediction system is designed. The real-time collection and monitoring of environmental parameters of the belt transport roadway is realized, and the belt fire alarm and early warning can be accurately carried out. The function and performance of the system software have been verified through tests, and it has been proved that the software has extremely high application value in ensuring the safety of mine belt transportation tunnels.

[1] 张宏.煤炭行业“十四五”发展面临的形势与任务[J].企业观察家,2020(11):56-60.

[2] 王国法,任世华,庞义辉,等.煤炭工业“十三五”发展成效与“双碳”目标实施路径[J].煤炭科学技术,2021,49(09):1-8.

[3] 史亮亮.矿井火灾的人为失误分析及防范对策研究[J].当代化工研究,2022(03):183-185.

[4] 赵志强.矿用带式输送机的使用及故障分析[J].当代化工研究,2019(12):17-18.

[5] 王烨. 矿井胶带火灾烟流蔓延特性与监测方法研究[D].西安科技大学,2021.

[6] 张美金. 基于RS-SVM算法的矿井主运输系统火灾诊断及灾害控制研究[D].辽宁工程技术大学,2016.

[7] 王凯,蔡炜垚,高士伟,等.矿井火灾风烟流应急调控系统联动可靠性研究[J].中国矿业大学学报,2021,50(04):744-754.

[8] 李志宪,陈江.日本三井三池有明煤矿井下特大火灾事故[J].工业安全与防尘,1990(03):42+21.

[9] 郑宝莲.煤矿外因火灾的原因及其防控[J].科技创新导报,2017,14(02):55+57.

[10] 卫德俊. 复杂矿井火灾烟气蔓延三维仿真研究[J]. 能源与环保, 2020, 42(08):55-59.

[11] 王德明, 程远平, 周福宝, 等. 矿井火灾火源燃烧特性的实验研究[J]. 中国矿业大学学报, 2002, 31(1):30-33.

[12] 王玉怀, 马尚权, 潘德祥. 煤矿井下可燃材料燃烧特性的试验研究[J]. 华北科技学院学报, 2006(01):1-4.

[13] 季经纬, 程远平. PVC输送带点燃温度判据及应用[J]. 燃烧科学与技术, 2006(05):438-441.

[14] 马洪亮, 周心权. 矿井火灾燃烧特性曲线的研究与应用[J]. 煤炭学报, 2008(7):780-783.

[15] Perera I E, Litton C D. Impact of Air Velocity on the Detection of Fires in Conveyor Belt Haulageways [J]. Fire Technology, 2012, 48(2):405-418.

[16] 王文才, 姜宇鸿, 张博, 等. 巷道中煤燃烧时热释放速率的研究[J]. 煤矿安全, 2013, 44(12):40-42.

[17] 苏墨.矿井胶带火灾灾变规律数值模拟及自动灭火系统设计研究[D].太原理工大学,2017.

[18] 郭军,刘荫,金永飞,等.矿井胶带火灾巷道环境多参数时空演化规律[J].西安科技大学学报,2019,39(01):21-27.

[19] 肖国强,张鹏宇,马砺,等.矿用胶带火灾燃烧特性预测模型[J].湖南科技大学学报(自然科学版),2021,36(03):9-15.

[20] Krawiec P, Warguła Ł, Czarnecka-Komorowska D, et al. Chemical compounds released by combustion of polymer composites flat belts[J]. Scientific Reports, 2021, 11(1): 1-10.

[21] 黄民, 顾玉华, 魏任之. 矿用胶带输送机火灾监测系统的研究[J]. 煤炭学报, 2002(01):78-82.

[22] 黄民, 魏任之. 矿用钢绳芯带式输送机实时工况监测与故障诊断技术[J]. 煤炭学报, 2005(02):245-250.

[23] 孙继平, 任慧, 吕朝辉. 矿井红外火灾图像处理方法的研究[J]. 煤炭科学技术, 2007(06):64-66.

[24] 邓军, 肖旸, 陈晓坤, 等. 矿井火灾多源信息融合预警方法的研究[J]. 采矿与安全工程学报, 2011, 28(04):638-643.

[25] 张鹏鹏,俞阿龙,孙诗裕,等.多传感器数据融合在矿井安全监测中的应用[J].工矿自动化,2015,41(12):5-8.

[26] 颜国强,黄启铭,杜文州,等.煤矿胶带巷火灾预警与控制系统设计[J].煤炭工程,2015,47(03):20-22+25.

[27] 马宏伟, 杨文娟, 张旭辉. 带式输送机托辊红外图像分割与定位算法[J]. 西安科技大学学报, 2017, 37(06):892-898.

[28] Hoff H. Using distributed fibre optic sensors for detecting fires and hot rollers on conveyor belts[C]//2017 2nd International Conference for Fibre-optic and Photonic Sensors for Industrial and Safety Applications (OFSIS). IEEE, 2017: 70-76.

[29] 边丽颖.光纤测温监测预警系统在带式输送机上的开发[J].选煤技术,2019(03):111-115.

[30] 孙继平,孙雁宇.矿井火灾监测与趋势预测方法研究[J].工矿自动化,2019,45(03):1-4.

[31] Yang Y, Hou C, Qiao T, et al. Longitudinal tear early-warning method for conveyor belt based on infrared vision[J]. Measurement, 2019, 147: 106817.

[32] Yang R, Qiao T, Pang Y, et al. Infrared spectrum analysis method for detection and early warning of longitudinal tear of mine conveyor belt[J]. Measurement, 2020, 165: 107856.

[33] Amruthnath N, Gupta T. A research study on unsupervised machine learning algorithms for early fault detection in predictive maintenance[C]//2018 5th International Conference on Industrial Engineering and Applications (ICIEA),2018: 355-361.

[34] Ahmad S, Lavin A, Purdy S, et al. Unsupervised real-time anomaly detection for streaming data[J]. Neurocomputing, 2017, 262: 134-147.

[35] Dey P , Saurabh K , Kumar C , et al. t-SNE and variational auto-encoder with a bi-LSTM neural network-based model for prediction of gas concentration in a sealed-off area of underground coal mines[J]. Soft Computing, 2021, 25(22):14183-14207.

[36] 张美金,王一,贾立军.基于支持向量机的传输线故障类型识别算法[J].辽宁工程技术大学学报(自然科学版),2017,36(05):528-532.

[37] 何永勃,李明伟.基于循环神经网络的飞机货舱火灾快速识别算法[J].消防科学与技术,2020,39(11):1490-1494.

[38] 蔡钟山.基于模糊聚类算法的早期火灾预警巡检机器人[J].机电工程技术,2020,49(10):149-150+250.

[39] Barros-Daza M J, Luxbacher K D, Lattimer B Y, et al. Mine conveyor belt fire classification[J]. Journal of fire sciences, 2022, 40(1): 44-69.

[40] 黄伟建,李永涛,黄远.基于混合神经网络和注意力机制的混沌时间序列预测[J].物理学报,2021,70(01):235-243.

[41] 尹云飞,林跃江,黄发良,等.基于趋势特征向量的火灾烟气流动与温度分布预测[J].计算机科学,2021,48(07):299-307.

[42] Sharma R, Rani S, Memon I. A smart approach for fire prediction under uncertain conditions using machine learning[J]. Multimedia Tools and Applications, 2020, 79(37): 28155-28168.

[43] Barros-Daza M J, Luxbacher K D, Lattimer B Y, et al. Fire Size and Response Time Predictions in Underground Coal Mines Using Neural Networks[J]. Mining, Metallurgy & Exploration, 2022: 1-12.

[44] 王亭亭. 矿用带式输送机安全监测系统研究[D]. 江苏:中国矿业大学,2016.

[45] 柳青秀,马红占,褚学宁,等.基于长短时记忆—自编码神经网络的风电机组性能评估及异常监测[J].计算机集成制造系统,2019,25(12):3209-3219.

[46] 陈彩华.矿井安全监控平台中的智能决策系统研究[J].贵州师范大学学报(自然科学版),2019,37(05):105-109.

[47] 王鹏,伍永平,王栓林,等.矿井瓦斯浓度Lagrange-ARIMA实时预测模型研究[J].煤炭科学技术,2019,47(04):141-146.

[48] 孙继平,孙雁宇.矿井火灾监测与趋势预测方法研究[J].工矿自动化,2019,45(03):1-4.

[49] 王猛,赵金升,丁宇辉,等.基于AI智能的煤矿井筒隐患识别系统[J].中国高新科技,2021(11):72-73.

[50] 王希予.智能算法在工业数据预处理上的应用与研究[D].北京工业大学,2019.

[51] 游东东,黎家良,刘高俊,等.基于贝叶斯BiLSTM模型的核电阀位传感器故障预警方法[J].华南理工大学学报(自然科学版),2021,49(12):43-52.

[52] 李俊,陈嘉杰,金捷,等.道岔电器时序监测和异常自动分析实现方案[J].轻型汽车技术,2019(07):15-18.

[53] 杨乐,马驰,胡辉,等.基于改进LSTMs模型的区域中长期气温预测方法研究[J].惠州学院学报,2021,41(06):75-79+99.

[54] Carreno A, Inza I, Lozano J A. Analyzing rare event, anomaly, novelty and outlier detection terms under the supervised classification framework[J]. Artificial Intelligence Review, 2019: 1-20.

[55] 丁小欧, 于晟健, 王沐贤,等. 基于相关性分析的工业时序数据异常监测[J]. 软件学报, 2020(3):726-747.

[56] 温粉莲.一种混合模型的时序数据异常监测方法[J].数字通信世界,2020(01):15-16+34.

[57] Raffei A F M, Awang N S, Rahman N S A, et al. Internet of Things (IoT) Based Fire Alert Monitoring System for Car Parking[C]//2020 7th International Conference on Electrical and Electronics Engineering (ICEEE). IEEE, 2020: 290-293.

[58] 苟怡.分布式光纤多参数监测及煤矿井下应用趋势探讨[J].自动化技术与应用, 2020, 039(004):114-118,128.

[59] Sarwar B, Bajwa I S, Jamil N, et al. An intelligent fire warning application using IoT and an adaptive neuro-fuzzy inference system[J]. Sensors, 2019, 19(14): 3150.