论文部分内容阅读
针对矿井中瓦斯浓度预测问题,提出一种具有层级结构的多模型预测方法。该模型不仅能够对不同区域的数据选择不同的子模型进行处理,而且每个数据都是由不同子模型中多个亚子模型协同处理。由于实测瓦斯浓度时间序列数据中含大量的噪声,采用经验模态分解将该时间序列数据分解成若干个独立的本征模函数,并将小尺度函数经低通滤波自适应除噪后进行相空间重构建立时间序列预测模型。用矿井实测瓦斯浓度数据进行试验,结果表明该模型较其他模型的预测精度有明显的提高。
Aimed at the gas concentration prediction in mine, a multi-model prediction method with hierarchical structure is proposed. The model not only can select different sub-models for data processing in different regions, but also each data is co-processed by multiple sub-models in different sub-models. Due to the large amount of noise in the measured time series data of gas concentration, the time series data are decomposed into several independent eigenmode functions by empirical mode decomposition, and the small scale function is adaptively denoised by low-pass filtering and then phase-shifted Spatial reconstruction to establish time series forecasting model. Experiments were carried out by using measured gas concentration data from coal mines. The results show that the prediction accuracy of this model is obviously improved compared with other models.