论文部分内容阅读
本论文针对电能量数据缺失值处理技术,提出了一种引入时间序列的基于贝叶斯常均值模型的DA多重插补法。首先使用EM插补算法计算缺失值的插补值,将得到的插补值作为插补的初始值;然后根据电能量数据随时间变化的特点,构建基于常均值模型的多重插补模型;最后通过贝叶斯方法预测每个缺失值的多次插补值,得到多个完整数据集合,综合分析观测误差方差和状态误差方差得到最终插补值。并通过与基于贝叶斯线性回归的DA多重插补结果比较,得出本论文改进的插补方法更加充分考虑到电能量数据的时间波动特性,构建的模型更符合客观实际,插补结果更加科学合理的结论。
In this paper, aiming at the lack of value processing technology of electric energy data, a new DA multiple interpolation method based on Bayesian constant mean-time model is introduced. Firstly, EM interpolation algorithm is used to calculate the interpolation value of the missing value, and the interpolation value is taken as the initial value of interpolation. Then, based on the characteristics of time-varying electrical energy data, a multiple interpolation model based on the normal average model is constructed. Finally, The Bayesian method is used to predict the multiple interpolation values of each missing value to obtain multiple complete data sets. The final interpolation values are obtained by comprehensively analyzing the observed error variance and the state error variance. And compared with the results of DA multiple interpolation based on Bayesian linear regression, it is concluded that the improved interpolation method in this paper considers the time fluctuation characteristics of electric energy data more fully, and the model is more objective and practical, and the interpolation result is more Scientific and reasonable conclusion.