摘要
通过人工神经网络方法,将影响工作面来压的主要因素作为输入层,构建BP神经网络模型,应用黄金分割算法确定最优隐含层节点数,得到最优模型,并对工作面支架平均来压阻力、平均非来压阻力、平均来压步距进行预测,分析可知预测误差在±10%,且符合正态分布,可通过多种方法提高预测精度,控制在±5%,取得了较好的预测结果精度,对于指导工作面的安全生产具有重要意义。
Based on the amficial neural network, the factors affecting the face pressure was set as input layers, and the BP neural network was set up, through gold segmentation algorithm defined the proper nodes, got the optimal model , and predicted the resistance pressure strength, the resistance of un-pressure strength and pressure step, analysis showed that the error controlled in the range of -10%--10%, and the error were normally distributed, through many ways to improve the prediction accuracy, the average can control the prediction accuracy in the range of -5% --5%, which can obtain a better predict accuracy, and possessed great significance to the safety production of the working face.
出处
《煤炭与化工》
CAS
2014年第8期37-40,共4页
Coal and Chemical Industry
关键词
BP网络
来压影响因素
来压预测
误差分析
BP network
factors affecting the pressure
pressure prediction
error analysis
