摘要
为提高矿井涌水量预测精度,解决矿井涌水量预测无法及时响应动态变化的问题,构建一种基于模态分解和深度学习的矿井涌水量多因素时间序列组合预测模型。使用变分模态分解和灰色关联分析筛选主控因素,通过双向长短期记忆网络和卷积长短期记忆网络对高、低频模态分量进行预测。结果表明:对比不同时序预测模型,变分模态分解可以有效捕捉时序数据中的长期依赖关系,提供了更加准确的长期时序数据预测能力;经过鲸鱼优化、贝叶斯优化算法对不同频率模态分量的处理,有效降低了高频部分的无序性、复杂性并优化了较为线性、缓慢的低频部分;验证了矿井涌水量时序预测中的变分模态深度学习组合模型的有效性和适用性,预测精度满足生产需求。该理论丰富了矿井涌水量时序预测方法,对煤矿水害预防具有一定的理论意义。
To enhance the accuracy of mine water inflow predictions and address the current inability of these forecasts so as to respond promptly to dynamic changes,a multifactorial timeseries combination forecast model for mine water inflow was developed,based on modal decomposition and deep learning.The model employs variational modal decomposition and grey relational analysis to select the main controlling factors,and predicts high and low frequency modal components through bidirectional long short-term memory networks and convolutional long short-term memory networks.The results that:compared to different time series prediction models,variational modal decomposition can effectively capture long-term dependencies in time series data,thus exhibting a more accurate long-term time series prediction capability;after processing different frequency modal components with whale optimization and Bayesian optimization algorithms,the disorder and complexity of the highfrequency part were effectively reduced,and the more linear and slow low-frequency part was optimized;through error assessment,the effectiveness and applicability of the variational modal deep learning combination model in mine water inflow time series prediction were verified,and its predictive accuracy meets production requirements.The model has better predictive performance and higher accuracy in the field of mine water inflow prediction,and one more method of mine water inflow time series prediction has been added,a theoretical significance for the prevention of water hazards in coal mines.
作者
侯恩科
徐林啸
荣统瑞
HOU Enke;XU Linxiao;RONG Tongrui(College of Geology and Environment,Xi’an University of Science and Technology,Xi’an 710054,China;Shaanxi Provincial Key Laboratory of Geological Support for Coal Green Exploitation,Xi’an University of Science and Technology,Xi’an 710054,China;Qinghai Salt Lake Industry Co.,Ltd.,Golmud 816000,China)
出处
《西安科技大学学报》
CAS
北大核心
2024年第3期490-500,共11页
Journal of Xi’an University of Science and Technology
基金
国家自然科学基金项目(42177174)。
关键词
模态分解
深度学习
时间序列
多因素序列降维
矿井涌水量预测
modal decomposition
depth learning
time series
multi-factor series dimensionality reduction
mine water inflow prediction
