选用改进高斯过程回归模型的碳排放短期预测

Short-term forecasting of carbon emission by using improved Gaussian process regression

针对于采矿过程中以电机为研究对象的碳排放来源的复杂性以及其影响因素的多样性所引起的碳排放短期预测精度不高的问题,结合灰色理论提出一种基于改进高斯过程回归模型的铅锌矿采矿过程碳排放预测方法。对碳排放来源及其影响因素进行分析,用灰色理论进行聚类分析以归并同类因素;根据灰色关联性分析得到主要影响因素;因传统高斯过程回归模型直接选定协方差函数的方式易导致与研究对象的物理过程拟合度不够高的问题,因而提出了一种依据先验知识的协方差函数选择方式,将四种常用协方差函数建模的训练结果作为反馈,结合极大似然估计法、最小二乘法和蒙特卡洛法参数估计的对比结果得到与研究对象拟合度最高即预测误差最小的协方差函数,进而得到预测效果最好的改进模型。经实验证明,基于该种方法选择协方差函数的模型相较于其他常规预测模型能更精确地预测铅锌矿采矿过程的碳排放量,其预测误差更小。

Considering the problem of low forecasting accuracy caused by the complexity of the carbon emission sources from the motor and the diversity of its impacts during the mining process of lead-zinc mine,a carbon emission forecasting method for mining process is proposed based on improved Gaussian process regression model combined with the grey theory.Firstly,the sources of carbon emission and their impacts are analyzed and the grey theory is used to cluster and merge the similar impacts.Then,the grey relational analysis is applied to obtain the main impacts.Finally,the forecasting accuracy is not satisfied due to the direct selection of the covariance function based on experience in the traditional Gaussian process regression model.In order to improve the accuracy,a covariance function selection method with prior knowledge is proposed.Based on the parameter estimation result of maximum likelihood estimation and least square method as well as Monte Carlo method,the covariance function that makes the model forecasting accuracy highest and the error minimum is obtained by comparing the forecasting results of several common-used covariance functions modeling.The result shows that improved Gaussian process regression model can be used to forecast the short-term carbon emission of leadzinc mine mining process with high accuracy and minimum error compared with other forecasting models.