Prediction of coal and gas outburst intensity based on LLE-FOA-BP model
-
摘要: 针对目前煤与瓦斯突出强度预测精度低、稳定性差及训练速度慢等问题，提出了一种基于局部线性嵌入法-果蝇优化算法-BP神经网络（LLE-FOA-BP）模型的煤与瓦斯突出强度预测方法。借助LLE算法的非线性数据特征提取优势，提取煤与瓦斯突出影响因素原始数据的本质特征，形成重构有效因子，降低数据间的冗余信息及噪声；利用FOA算法较强的全局寻优能力优化BP神经网络的权值和阈值，避免陷入局部极小，提高参数寻优效率；将重构有效因子输入优化后的BP神经网络进行训练，实现煤与瓦斯突出强度快速、准确预测。测试结果表明，LLE-FOA-BP模型的平均相对误差为8.06%，相对误差的方差为3.69，经过24次迭代训练就达到10-8的训练精度，能够在保证预测精度的基础上，提高鲁棒性和学习效率。
-
关键词:
- 煤与瓦斯突出强度预测 /
- 局部线性嵌入 /
- 果蝇算法 /
- BP神经网络 /
- 大数据处理
Abstract: In view of problems of low prediction accuracy, poor stability and slow training of current coal and gas outburst intensity prediction, a prediction method of coal and gas outburst intensity based on LLE-FOA-BP model was proposed. Essential characteristics of raw data of coal and gas outburst influencing factors are extracted taking use of the advantage of nonlinear data feature extraction of LLE algorithm, effective reconstruction factors are formed, and redundant information and noise between data are reduced. The weight and threshold of BP neural network are optimized by using FOA algorithm's strong global optimization ability to avoid falling into local minima and improve parameter optimization efficiency. Effective reconstruction factors are input into the optimized BP neural network for training, so as to realize quick and accurate prediction of coal and gas outburst intensity.The test results show that the average relative error of LLE-FOA-BP model is 8.06%, variance of relative error is 3.69, and training accuracy of 10-8 is achieved after 24 iterations, which verifies the model can improve robustness and learning efficiency while ensuring prediction accuracy. -
计量
- 文章访问数: 76
- HTML全文浏览量: 10
- PDF下载量: 15
下载:
