In this paper, forecasting analysis to Box Cox transformation models with a practical example is considered. Based on chosen generalized functional form, variables influencing passenger are selected by statistic mech...In this paper, forecasting analysis to Box Cox transformation models with a practical example is considered. Based on chosen generalized functional form, variables influencing passenger are selected by statistic mechanism, not just by subjective judgment or dependent on certain specified model, and forecasting models are constructed. Comparing with typical linear regression forecasting models, nonlinear forecasting models are more effective and precise. Based on collecting data and final forecasting models, forecasting results are obtained and forecasting errors are analyzed. Finally, some helpful conclusions can be drawn from this study.展开更多
风电场准确的风速预测可以减轻或避免风电对电网的不利影响,有利于在开放的电力市场环境下正确制定电能交换计划,提高风电竞争力。基于风速序列的时序性,使用极大似然法对风速序列进行了Box-Cox最优变换,建立了ARMA(p,q)风速预测模型。...风电场准确的风速预测可以减轻或避免风电对电网的不利影响,有利于在开放的电力市场环境下正确制定电能交换计划,提高风电竞争力。基于风速序列的时序性,使用极大似然法对风速序列进行了Box-Cox最优变换,建立了ARMA(p,q)风速预测模型。为检验时间序列模型的有效性,利用最小信息准则中的BIC(Bayesian Information Criterion)函数对ARMA(p,q)模型进行识别,并通过风速频率曲线对预测结果进行了修正。仿真结果和算例验证了该方法在风电场风速预测中的适用性,具有一定的实用价值。展开更多
针对贝叶斯概率预报模型(Bayesian processor of forecasts,BPF)中输入数据的正态转换问题,探讨了Meta-Gaussian模型(MG)和Box-Cox变换(BC)对BPF模型性能的影响。首先利用MG和BC分别对BPF模型输入数据进行正态转换,然后分别建立BPF-MG和...针对贝叶斯概率预报模型(Bayesian processor of forecasts,BPF)中输入数据的正态转换问题,探讨了Meta-Gaussian模型(MG)和Box-Cox变换(BC)对BPF模型性能的影响。首先利用MG和BC分别对BPF模型输入数据进行正态转换,然后分别建立BPF-MG和BPF-BC模型进行概率预报,最后对BPF-MG和BPF-BC在不同预见期和不同数据样本条件下的预报能力进行了分析。结果表明,当数据样本较少时,BPF-MG具有较高的稳定性,但BC转换比MG更简单,BC变换系数非常敏感;当数据样本增多后,BC变换的转换系数稳定,BPFBC预报质量提高。展开更多
文摘In this paper, forecasting analysis to Box Cox transformation models with a practical example is considered. Based on chosen generalized functional form, variables influencing passenger are selected by statistic mechanism, not just by subjective judgment or dependent on certain specified model, and forecasting models are constructed. Comparing with typical linear regression forecasting models, nonlinear forecasting models are more effective and precise. Based on collecting data and final forecasting models, forecasting results are obtained and forecasting errors are analyzed. Finally, some helpful conclusions can be drawn from this study.
文摘风电场准确的风速预测可以减轻或避免风电对电网的不利影响,有利于在开放的电力市场环境下正确制定电能交换计划,提高风电竞争力。基于风速序列的时序性,使用极大似然法对风速序列进行了Box-Cox最优变换,建立了ARMA(p,q)风速预测模型。为检验时间序列模型的有效性,利用最小信息准则中的BIC(Bayesian Information Criterion)函数对ARMA(p,q)模型进行识别,并通过风速频率曲线对预测结果进行了修正。仿真结果和算例验证了该方法在风电场风速预测中的适用性,具有一定的实用价值。
文摘针对贝叶斯概率预报模型(Bayesian processor of forecasts,BPF)中输入数据的正态转换问题,探讨了Meta-Gaussian模型(MG)和Box-Cox变换(BC)对BPF模型性能的影响。首先利用MG和BC分别对BPF模型输入数据进行正态转换,然后分别建立BPF-MG和BPF-BC模型进行概率预报,最后对BPF-MG和BPF-BC在不同预见期和不同数据样本条件下的预报能力进行了分析。结果表明,当数据样本较少时,BPF-MG具有较高的稳定性,但BC转换比MG更简单,BC变换系数非常敏感;当数据样本增多后,BC变换的转换系数稳定,BPFBC预报质量提高。