煤与瓦斯突出是煤矿的五大灾害之一，预防煤与瓦斯突出是煤炭企业安全生产的关键。瓦斯突出机理模型尚不明确，且瓦斯涌出量为非线性时间序列数据，使得传统的预测方法难以应对瓦斯突出这样复杂的非线性过程。由于神经网络有强大的非线性拟合能力，所以可以通过构建神经网络瓦斯涌出量预测模型，对瓦斯灾害进行有效的防治。为提高瓦斯涌出量的预测精度，本文采用改进的烟花算法(WOFWA)对回声状态网络(ESN)储备池参数选择过程进行优化的方法，构建基于WOFWA-ESN网络的煤与瓦斯突出预测系统。本文主要研究内容如下：

首先，为提高烟花算法的寻优能力，提出一种基于对立学习以及指数递减策略的改进烟花算法。该算法对传统烟花算法主要有两方面的改进：一是用对立学习策略来生成初始烟花种群，即同时产生当前烟花个体和对立烟花个体，然后通过比较将适应度值较好的烟花选作初始种群个体。二是针对原始FWA中爆炸火花的位移方式，引入增强烟花算法中的爆炸火花产生方式以及一个惯性权重因子，从而提高算法的寻优精度和收敛速度。选取4种经典的复杂非线性基准函数作为测试函数，验证WOFWA算法的优越性。

其次，针对传统ESN网络储备池相关参数难以确定，导致预测模型精度低的问题，采用WOFWA算法对储备池的4个参数值设置过程进行优化，从而构建基于WOFWA-ESN网络的瓦斯突出预测模型。模型构建的核心是采用WOFWA算法对ESN储备池参数选择过程进行优化，用训练的误差函数代替WOFWA算法中的目标函数进行寻优，然后将算法求得的最优解设置为ESN网络的储备池参数值。实验结果表明，与其他模型相比，本文提出的WOFWA-ESN预测模型的准确性有了显著提高，从而证实了该模型的有效性。

最后，在WOFWA-ESN算法研究的基础上，通过PyCharm建立系统界面，调用预测模型，建立煤与瓦斯突出风险等级预测平台，实现瓦斯浓度序列分析以及突出风险等级预测的功能。本平台界面舒适美观，系统使用方便，实用价值和应用前景良好。

The five major disasters in coal mine including coal and gas outburst. Preventing coal and gas outburst is the key to ensure safe mining of the coal industry. The mechanism model of gas outburst is not clear yet, and the gas emission amount is nonlinear time series data, which makes the traditional prediction method difficult to deal with the complex nonlinear process of gas outburst. Because of the strong nonlinear fitting ability of neural network, the prediction model of gas emission quantity of neural network can be built to effectively prevent and control gas disasters. In order to improve the prediction accuracy of gas emission, the improved Fireworks Algorithm (WOFWA) was adopted in this paper to optimize the parameter selection process of ESN reserve pool, and a coal and gas outburst prediction system based on WOFWA-ESN network was established. The main research contents of this paper are as follows:

Firstly, to improve the optimization ability of the fireworks algorithm, an improved fireworks algorithm based on opposition-based learning and exponential decline strategy was proposed. The algorithm for traditional fireworks algorithm improved basically has two aspects: one is to use opposite learning strategies to generate initial fireworks populations, which at the same time to produce the current individual fireworks and corresponding opposing the fireworks, and then through the comparison and choose better fitness value of fireworks as the initial populations of individuals, thus improve the searching efficiency of the algorithm. Secondly, aiming at the displacement mode of explosive spark in the original FWA, the explosive spark generation mode in the enhanced fireworks algorithm and an inertia weight factor which decreases non-linearly with the increase of iteration times are introduced to improve the optimization accuracy and convergence speed of the algorithm. Four classical complex nonlinear reference functions are selected as test functions to verify the superiority of WOFWA algorithm.

Secondly, it’s difficult to determine relevant parameters of the reserve pool in the traditional ESN network, which leads to the low accuracy of the prediction model. WOFWA algorithm is used to optimize the setting process of four parameters of the reserve pool, so as to build a gas outburst prediction model based on WOFWA-ESN network. The core of the model construction is to optimize the selection process of ESN reserve pool parameters by using WOFWA algorithm, replacing the objective function of WOFWA algorithm with the training error function for optimization, and then setting the optimal solution obtained by the algorithm as the reserve pool parameter value of ESN network. Experimental results show that compared with other models, the accuracy of the proposed WOFWA-ESN prediction model is significantly improved, thus confirming the effectiveness of the model.

Finally, based on the study of WOFWA-ESN algorithm, the system interface was established through PyCharm, the prediction model was called, and the prediction and early warning platform of coal and gas outburst was established to realize the function of gas concentration sequence analysis and outburst prediction and early warning. The platform interface is comfortable and beautiful, the system is convenient to use, practical value and application prospects are good.

[1] 滕吉文,乔勇虎,宋鹏汉.我国煤炭需求、探查潜力与高效利用分析[J].地球物理学报,2016,59(12):4633-4653.

[2] 赵亚军,孟祥铭,陈丽林.对煤炭业年死亡人数和百万吨死亡率的模拟预测研究[J].煤矿现代化,2011(06):7-10.

[3] 姜福兴,杨光宇,魏全德,王存文,曲效成,朱斯陶.煤矿复合动力灾害危险性实时预警平台研究与展望. 煤炭学报, 2018, 43(2):333-339.

[4] 孙利源.基于灰色关联分析和PSO-SVM的煤与瓦斯突出预测应用研究[D].中国矿业大学;中国矿业大学(江苏),2019.

[5] Ridong Zhang, Jili Tal, Renquan Lu, and Qibing Jin. Decoupled ARX and RBF Neural Network Modeling Using PCA and GA Optimization for Nonlinear Distributed Parameter Systems. IEEE Transactions on Neural Networks and Learning Systems, 2018,29(2):457-469.

[6] 樊栓保.国内外煤与瓦斯突出预测的新方法[J].矿业安全与环保,2000,27(5):17-19.

[7] 秦汝祥,张国枢,杨应迪.瓦斯涌出异常预报煤与瓦斯突出[J].煤炭学报,2006(05):599-602.

[8] 福岛笃,孙重旭.日本煤矿瓦斯突出的预测与防治措施[J].煤炭工程师,1991(01):46-54.

[9] 薛生,谢军.澳大利亚煤矿瓦斯治理及利用[J].中国煤炭,2012,38(03):117-119+122.

[10] 孙忠强,苏昭桂,张金锋.煤与瓦斯突出预测预报技术研究现状及发展趋势[J].能源技术与管理,2008(02):56-57+65.

[11] 漆旺生,凌标灿,蔡嗣经.煤与瓦斯突出预测研究动态及展望[J].中国安全科学学报,2003,13(12):4-7.

[12] 赵鹏伟.矿山统计法预测综放工作面瓦斯涌出量[J].机械管理开发,2006,(3):18-19.

[13] 王佑安.煤和瓦斯突出危险性预测方法[J].煤矿安全,1984(04):1-7.

[14] 胡千庭,邹银辉,文光才,等. 瓦斯含量法预测突出危险新技术[J]. 煤炭学报,2007,32(3):276-280.

[15] 王刚,程卫民,王海洋,等.瓦斯含量法预测煤与瓦斯突出的应用与实践[C]//第九届全国煤炭工业生产一线青年技术创新大会. 2014:334-342.

[16] 关维娟,张国枢,陈清华,邓明.基于瓦斯涌出时间序列的煤与瓦斯突出预测方法研究[J].安全与环境学报,2011,11(03):170-173.

[17] 邓明,张国枢,陈清华.基于瓦斯涌出时间序列的煤与瓦斯突出预报[J].煤炭学报,2010,35(02):260-263.

[18] 刘硕.K_1值预测突出敏感性的实验研究[D].中国矿业大学,2016.

[19] 张浪.突出煤体变形破坏过程声发射演化特征综合分析[J].煤炭学报,2018,43(z1):130-139.

[20] 王雨虹,刘璐璐,付华,等.基于声发射多参数时间序列的瓦斯突出预测[J].中国安全科学学报,2018,28(5):129-134.

[21] 白慧敏,李忠辉,沈荣喜,等.电磁辐射技术在煤与瓦斯突出预测中的应用[J].煤矿安全,2010,41(6):26-28.

[22] 李洋,刘晓悦.基于FOA-BP的电磁辐射法预测煤与瓦斯突出[J].华北理工大学学报（自然科学版）,2018,40(1):56-63.

[23] 钟晓晖,朱令起,郭立稳,等.煤体破裂过程辐射温度场的研究[J].煤炭科学技术,2006,34(2):57-59.

[24] 杨敏,汪云甲,程远平.煤与瓦斯突出预测的改进差分进化神经网络模型研究[J].中国矿业大学学报,2009,38(3):439-444.

[25] 周西华,孙家正.基于主因子分析的改进BP神经网络瓦斯涌出量预测[J].矿业安全与环保,2018,45(6):43-47,52.

[26] Yaqin Wu,Ronglei Gao,Jinzhen Yang. Prediction of coal and gas outburst: A method based on the BP neural network optimized by GASA[J]. Transactions of The Institution of Chemical Engineers. Process Safety and Environmental Protection, Part B,2020,133:64-72.

[27] Bhattacherjee Ashis, Ramani R.V., Natarajan R.. Time series analysis of coal mine accident experience[J]. Journal of Safety Research ,1994,18(4):229-234.

[28] 张晓铭,张爱绒,郭勇义.基于PSO算法优化的自组织竞争神经网络在煤与瓦斯突出预测中的应用研究[J].中国煤炭,2013,39(01):106-109.

[29] Shi-yue Wu, Yong-yi Guo, Yuan-xing Li, Guo-liang Yang, Yu Niu. Research on the mechanism of coal and gas outburst and the screening of prediction indices[C]//The 6th International Conference on Mining Science & Technology ICMST 2009. 2009:1-6.

[30] 陈全秋,郭勇义,吴世跃,徐玉胜,王灿召.遗传算法与神经网络的结合在煤与瓦斯突出危险性预测中的应用研究[J].中国煤炭,2010,36(03):86-89.

[31] 俞启香.矿井瓦斯防治[M],徐州：中国矿业大学出版社，1992.

[32] 王汉斌.煤与瓦斯突出的分形预测理论及应用[D].山西:太原理工大学,2009.

[33] 鲁宗厚.“瓦斯放散初速度”ΔP测定中的问题与改进[J].工业安全与防尘, 1992，(09):39-41.

[34] 李建铭.煤与瓦斯突出防治技术手册[M].徐州：中国矿业大学出版社，2006.

[35] Tan Y,Zhu Y. Fireworks algorithm for optimization[C]//International Conference on Advances in Swarm Intelligence, Berlin, 2010:355-364.

[36] 谭营,郑少秋.烟花算法研究进展[J].智能系统学报,2014,9(05):515-528.

[37] Hany M. Hasanien. Performance improvement of photovoltaic power systems using an optimal control strategy based on whale optimization algorithm[J].Electric Power Systems Research,2018,157(04):168-176.

[38] W. Z. Sun, J. S. Wang. Elman neural network soft-sensor model of conversion velocity in polymerization process optimized by chaos whale optimization algorithm[C]//IEEE success,2017(05):13062-13076.

[39] Randy L. Haupt, Sue Ellen Hauput. Practical genetic algorithms[J]. Discrete Applied Mathematics,2004,146(1):119.

[40] Shubham Gupta, Kusum Deep, Ali Asghar Heidari, Hossein Moayedi, Mingjing Wang. Opposition-based learning Harris hawks optimization with advanced transition rules: principles and analysis[J]. Expert Systems With Applications,2020,158(1):1-23.

[41] Hamid R. Tizhoosh. Opposition-Based Reinforcement Learning[J]. Journal of Advanced Computational Intelligence, 2006,10(4):578-585.

[42] Liu, Y., Wang, H., Rahnamayan, S., et al. Enhancing particle swarm optimization using generalized opposition-based learning[J]. Information Sciences: An International Journal,2011,181(20):4699-4714.

[43] Eberhart RC, Shi Yuhui. Tracking and optimizing dynamic systems with particle swarms[C]//Proc of Congress on Evolutionary Computation. Piscataway, NJ:IEEE Press,2001:94-100.

[44] Shi Yuhui, Eberhart RC. Empirical study of particle swarm optimization [C]//Proc of Congress on Evolutionary Computation. Piscataway, NJ:IEEE Press,1999:1945-1949.

[45] Shi Yuhui, Eberhart RC. Fuzzy adaptive particle swarm optimization [C]//Proc of Congress on Evolutionary Computation. Piscataway, NJ:IEEE Press,2001:101-106.

[46] 陈贵敏,贾建援,韩琪.粒子群优化算法的惯性权值递减策略研究[J].西安交通大学学报,2006(01):53-56+61.

[47] Zheng S, Janecek A, Tan Y. Enhanced Fireworks Algorithm[C]//IEEE Congress on Evolutionary Computation.:Institute of Electrical and Electronics Engineers, 2013:2069-2077.

[48] Jaeger H．Echo state network[J]. Scholarpedia,2007,2(9)：1479-1482．

[49] Herbert Jaeger. Adaptive Nonlinear System Identification with Echo State Networks[C]//Advances in Neural Information Processing Systems 15. 2002:609-616.

[50] M.C.Ozturk, D.Xu and J.Ncipe. Analysis and design of echo state networks[J]. NeuralComputation,2007,19(01),111-138.

[51] Braverman, E., Idels, L., Berezansky, L.. The Mackey-glass model of respiratory dynamics: Review and new results[J].Nonlinear Analysis: An International Multidisciplinary Journal,2012,75(16):6034-6052.