摘要
基于回采工作面瓦斯涌出分源涌出,利用人工神经网络分别预测开采煤层、邻近煤层、采空区3种来源的瓦斯涌出量;因3种来源瓦斯涌出量的影响因素不同,为了避免不相关因素的干扰,提高预测精度,确定整个预测体系由开采层、邻近层、采空区等3个瓦斯涌出量预测神经网络组成,对每个涌出源分别建立神经网络预测模型;最后采用Matlab中BP神经网络算法,针对实际矿井进行应用,预测误差小.
Based on the different-source gas emission quantity prediction theory, the BP nerve network was applied to predict respectively the gas emission quantity from the mining coal seam, neighboring coal seam and goaf of working face. Because the gas-emission influencing factors of the mining coal seam, neighboring coal seam and goaf of working face were different, three gas emission prediction neural network models were established respectively to avoid the interference of the irrelevant factors and to increase the prediction accuracy. The gas emission prediction neural network model of the mining coal seam was made up of three layers and nine parameters. The prediction model of the neighboring coal seam was made up of three layers and eight parameters, and the prediction model of the goaf of working face was made up of three layers and four parameters. The different-source gas emission prediction model can improve prediction accuracy greatly. The BP neural network arithmetic of Matlab software was adopted and put into application in coal mine, and the prediction error can satisfy the demand in application.
出处
《煤炭学报》
EI
CAS
CSCD
北大核心
2007年第5期504-508,共5页
Journal of China Coal Society
基金
国家自然科学基金重点资助项目(50134040)
关键词
回采工作面
瓦斯涌出量
BP人工神经网络
分源预测
working face
gas emission quantity
BP neural network
different-source prediction
