| 基于GASA-BP神经网络的煤层瓦斯含量预测方法研究* |
| |
| 引用本文： | 马磊,陆卫东,魏国营.基于GASA-BP神经网络的煤层瓦斯含量预测方法研究*[J].中国安全生产科学技术,2022,18(8):59-65. |
| |
| 作者姓名： | 马磊 陆卫东 魏国营 |
| |
| 作者单位： | （1.河南理工大学 安全科学与工程学院，河南 焦作 454000；2.新疆工程学院 安全科学与工程学院，新疆 乌鲁木齐830023；3.煤炭安全生产与清洁高效利用省部共建协同创新中心，河南 焦作 454000 ） |
| |
| 基金项目： | * 基金项目: 河南省瓦斯地质与瓦斯治理重点实验室——省部共建国家重点实验室培育基地开放基金项目(WS2018A04);河南省科技攻关项目（202102310221,202102310619） |
| |
| 摘 要： | 为提高煤层瓦斯含量预测的精准度和效率，提出1种利用遗传算法（GA）和模拟退火算法(SA)混合初始化BP神经网络（BPNN）的瓦斯含量预测新模型（GASA-BPNN模型）。利用灰色关联分析法（GRA）筛选瓦斯含量主控因素并作为GASA-BPNN预测模型的输入。为解决BPNN收敛速度慢和易陷入局部极小陷阱的问题，将GA和具有时变概率突跳性的SA整合为GASA算法协同初始化BPNN的权值和阈值，有效地提高BPNN的参数学习能力。将该模型应用于煤炭生产现场，结果表明:BPNN模型、GA-BPNN模型和GASA-BPNN模型瓦斯含量预测总平均相对误差分别为15.79%，9.03%，5.56%。相比BPNN模型和GA-BPNN模型，GASA-BPNN模型对样本的泛化能力更强，参数训练速度最快并且预测精准度最高。
|
| 关 键 词： | BP神经网络 煤层瓦斯含量 遗传算法（GA） 模拟退火算法(SA) 灰色关联分析(GRA) |
Study on prediction method of coal seam gas content based on GASA-BP neural network |
| |
| MA Lei,LU Weidong,WEI Guoying.Study on prediction method of coal seam gas content based on GASA-BP neural network[J].Journal of Safety Science and Technology,2022,18(8):59-65. |
| |
| Authors: | MA Lei LU Weidong WEI Guoying |
| |
| Affiliation: | (1.College of Safety Science and Engineering,Henan Polytechnic University,Jiaozuo Henan 454000,China;2.Department of Safety Engineering,Xinjiang Institute of Engineering,Urumqi Xinjiang 830023,China;3.Collaborative Innovation Center of Coal Work Safety and Clean High Efficiency Utilization,Jiaozuo Henan 454000,China) |
| |
| Abstract: | To improve the accuracy and efficiency of coal seam gas content prediction,a new model of gas content prediction (GASA-BPNN model) was proposed,which used the mixed genetic algorithm (GA) and simulated annealing algorithm (SA) to initialize the BP neural network (BPNN).The grey relational analysis (GRA) method was used to screen the main controlling factors of gas content,which were used as the input of the GASA-BPNN prediction model.To solve the problems of slow convergence speed and easy to fall into the local minimum trap of BPNN,the GA and SA with time-varying probability jump were integrated into the GASA algorithm to initialize the weight and threshold of BPNN,which effectively improved the parameter learning ability of BPNN.The model was applied to the coal production site,and the results showed that the total average relative errors of gas content prediction of BPNN model,GA-BPNN model and GASA-BPNN model were 15.79%,9.03% and 5.56%,respectively.Compared with the BPNN model and GA-BPNN model,the GASA-BPNN model had the stronger generalization ability to the samples,the fastest parameter training speed and the highest prediction accuracy. |
| |
| Keywords: | BP neural network coal seam gas content Genetic Algorithm (GA) Simulated Annealing Algorithm (SA) Grey Relational Analysis (GRA) |
|
| 点击此处可从《中国安全生产科学技术》浏览原始摘要信息 |
|
点击此处可从《中国安全生产科学技术》下载全文 |
