Laboratory experiments were conducted to simulate oil weathering process, a medium to long term weathering process for 210-d, using samples collected from five different oil resources. Based on relative deviation and ...Laboratory experiments were conducted to simulate oil weathering process, a medium to long term weathering process for 210-d, using samples collected from five different oil resources. Based on relative deviation and repeatability limit analysis about indexes of these samples, the results show there had been significant changes in diagnostic ratios among the initial and weathered samples of different oils during this process. Changes of selected n-alkane diagnostic ratios of all oil samples displayed more obviously than diagnostic ratios of terpanes,steranes and PAHs in this process. Almost all selected diagnostic ratios of terpanes, steranes and PAHs can be efficiently used in tracking sources of hydrocarbon pollution, differentiating from the n-alkane diagnostic ratios.In these efficient diagnostic ratios, only four ratios maintained good stability in the weathering processes and are more suitable because their relative deviation（RSD） are lower than 5%.展开更多
针对中长期电力负荷序列噪声含量高、难以直接提取序列周期规律从而影响预测精度的问题,提出了一种基于完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)和奇异谱分析(sin...针对中长期电力负荷序列噪声含量高、难以直接提取序列周期规律从而影响预测精度的问题,提出了一种基于完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)和奇异谱分析(singular spectrum analysis,SSA)双重分解的双向长短时记忆网络(bidirectional long and short time memory,BiLSTM)预测模型。首先,采用CEEMDAN对历史负荷进行分解,以得到若干个周期规律更为清晰的子序列;再利用多尺度熵(multiscale entropy,MSE)计算所有子序列的复杂程度,根据不同时间尺度上的样本熵值将相似的子序列重构聚合;然后,利用SSA去噪的功能,对高度复杂的新序列进行二次分解,去除序列中的噪声并提取更为主要的规律,从而进一步提高中长序列预测精度;再将得到的最终一组子序列输入BiLSTM进行预测;最后,考虑到天气、节假日等外部因素对电力负荷的影响,提出了一种误差修正技术。选取了巴拿马某地区的用电负荷进行实验,实验结果表明,经过双重分解可以将均方根误差降低87.4%;预测未来一年的负荷序列时,采用的BiLSTM模型将拟合系数最高提高2.5%;所提出的误差修正技术可将均方根误差降低9.7%。展开更多
高效准确的短期电力负荷预测对提升新型电力系统经济运行十分重要。针对极端天气场景下负荷预测数据量较少、随机性较强的特点,提出一种基于张量低秩补全算法的短期负荷预测模型,并选取极端高温场景展开研究。首先,给出极端天气定义,并...高效准确的短期电力负荷预测对提升新型电力系统经济运行十分重要。针对极端天气场景下负荷预测数据量较少、随机性较强的特点,提出一种基于张量低秩补全算法的短期负荷预测模型,并选取极端高温场景展开研究。首先,给出极端天气定义,并基于改进型炎热指数和气温两项指标进行数据筛选;其次,提出一种基于张量的负荷数据补全模型,补全缺失数据;然后,通过Pearson相关性分析筛选输入特征量,构建基于长短时记忆(long short term memory, LSTM)网络和粗糙集理论(rough set theory, RST)的LSTM-RST短期负荷预测模型;最后,以苏州某地实际负荷数据设置算例进行验证,仿真结果表明,所提短期负荷预测方法具有较高的准确性。展开更多
基金The National Natural Science Foundation of China under contract No.41206089Project of on-site sediment microbial remediation of public area of central Bohai Sea,North China Sea Branch of State Oceanic Administration under contract No.QDZC20150420-002Program of Science and Technology Service Network Initiative,Chinese Academy of Sciences under contract No.KFJ-EW-STS-127
文摘Laboratory experiments were conducted to simulate oil weathering process, a medium to long term weathering process for 210-d, using samples collected from five different oil resources. Based on relative deviation and repeatability limit analysis about indexes of these samples, the results show there had been significant changes in diagnostic ratios among the initial and weathered samples of different oils during this process. Changes of selected n-alkane diagnostic ratios of all oil samples displayed more obviously than diagnostic ratios of terpanes,steranes and PAHs in this process. Almost all selected diagnostic ratios of terpanes, steranes and PAHs can be efficiently used in tracking sources of hydrocarbon pollution, differentiating from the n-alkane diagnostic ratios.In these efficient diagnostic ratios, only four ratios maintained good stability in the weathering processes and are more suitable because their relative deviation（RSD） are lower than 5%.
文摘针对中长期电力负荷序列噪声含量高、难以直接提取序列周期规律从而影响预测精度的问题,提出了一种基于完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)和奇异谱分析(singular spectrum analysis,SSA)双重分解的双向长短时记忆网络(bidirectional long and short time memory,BiLSTM)预测模型。首先,采用CEEMDAN对历史负荷进行分解,以得到若干个周期规律更为清晰的子序列;再利用多尺度熵(multiscale entropy,MSE)计算所有子序列的复杂程度,根据不同时间尺度上的样本熵值将相似的子序列重构聚合;然后,利用SSA去噪的功能,对高度复杂的新序列进行二次分解,去除序列中的噪声并提取更为主要的规律,从而进一步提高中长序列预测精度;再将得到的最终一组子序列输入BiLSTM进行预测;最后,考虑到天气、节假日等外部因素对电力负荷的影响,提出了一种误差修正技术。选取了巴拿马某地区的用电负荷进行实验,实验结果表明,经过双重分解可以将均方根误差降低87.4%;预测未来一年的负荷序列时,采用的BiLSTM模型将拟合系数最高提高2.5%;所提出的误差修正技术可将均方根误差降低9.7%。
文摘高效准确的短期电力负荷预测对提升新型电力系统经济运行十分重要。针对极端天气场景下负荷预测数据量较少、随机性较强的特点,提出一种基于张量低秩补全算法的短期负荷预测模型,并选取极端高温场景展开研究。首先,给出极端天气定义,并基于改进型炎热指数和气温两项指标进行数据筛选;其次,提出一种基于张量的负荷数据补全模型,补全缺失数据;然后,通过Pearson相关性分析筛选输入特征量,构建基于长短时记忆(long short term memory, LSTM)网络和粗糙集理论(rough set theory, RST)的LSTM-RST短期负荷预测模型;最后,以苏州某地实际负荷数据设置算例进行验证,仿真结果表明,所提短期负荷预测方法具有较高的准确性。
