摘要
针对煤尘固体表面接触角测定过程繁琐和煤尘润湿等级划分不合理的问题,以煤质化学组成及其结构参数共13个影响因子为输入参数,采用两层双曲正切S形函数为激励函数,构建有关煤尘接触角估算及润湿性分级的3层BP神经网络.结果表明,隐含层节点数为10时,估算结果相对误差为0.19%~13.99%,平均相对误差为5.18%,煤尘润湿接触角估算结果与实测结果相关性系数为R2=0.933,煤尘润湿分级正确率达91.67%.BP神经网络模型的接触角估算结果和润湿性分级结果可用于指导煤矿井选择降尘措施.
This paper built a three layers BP neural network to solve the problem of the complicated measurement of wetting contact angle on the solid surface and unreasonable grade division of coal dust wettability. Whose input parameters included thirteen factors, such as chemical composition and characteristic parameter of coal dust, and driving function adopted hyperbolic tangent sigrnoid transfer function. The results show that when the nodes number of hidden layer is 10, the relative errors are at the range from 0.19 % to 13.99 %, and the average relative error is as low as 5.18 %. The correlation coefficent between measured values and estimate values of coal dust wetting contact angle is 0.933, and the accuracy of wettability classification is 91.67 %. Thus, the coal dust wetting contact angle and wettability classification results with BP neural network can be used to guide the selection of effective dust-reducing measures for coal mine well.
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2017年第6期593-597,共5页
Journal of Liaoning Technical University (Natural Science)
基金
国家自然科学基金(U1261121)
关键词
煤尘
接触角
润湿性分级
BP神经网络
双曲正切S形函数
coal dust
contact angle
wettability classification
BP neural network
hyperbolic tangent sigrnoid transfer function
