精准的分布式光伏短期发电功率预测有助于电力系统运行与功率就地平衡。该文提出一种基于BIRCH(balanced iterative reducing and clustering using hierarchies)相似日聚类的L-Transformer(LSTM-Transformer)模型进行短期光伏功率预测...精准的分布式光伏短期发电功率预测有助于电力系统运行与功率就地平衡。该文提出一种基于BIRCH(balanced iterative reducing and clustering using hierarchies)相似日聚类的L-Transformer(LSTM-Transformer)模型进行短期光伏功率预测。首先使用BIRCH无监督聚类算法对历史数据聚类得到3种典型天气,根据聚类结果划分测试集对模型进行训练。为提高不同天气类型下的预测精度,采用双层架构的L-Transformer模型,首层通过长短期记忆网络(long short term memory,LSTM)的门控单元机制捕捉时间序列中的长期依赖关系;次层结合Transformer模型的自注意力机制聚焦于当前任务更关键的特征量,通过多注意力头与光伏数据特征量相结合生成向量,注意力头并行计算,从而高效、精确地预测短期光伏功率。实测数据验证结果表明L-Transformer模型对于不同天气类型功率预测泛化性优异、精确度高,气象数据波动大时鲁棒性强。展开更多
BIRCH算法是一种适合处理大规模数值型的聚类算法,但现实生活中的数据往往是混合型数据,导致了BIRCH算法的局限性;此外,在使用BIRCH算法进行聚类分析的过程中存在隐私泄露的风险,而传统的中心化差分隐私算法存在需要可信第三方的缺点。...BIRCH算法是一种适合处理大规模数值型的聚类算法,但现实生活中的数据往往是混合型数据,导致了BIRCH算法的局限性;此外,在使用BIRCH算法进行聚类分析的过程中存在隐私泄露的风险,而传统的中心化差分隐私算法存在需要可信第三方的缺点。针对以上缺陷,提出了基于本地差分隐私的BIRCH混合数据(LDP-BIRCH)算法,对混合型数据中的非数值型数据进行编码处理,并使用本地差分隐私对数据集进行扰动,将扰动后的数据集发给第三方进行BIRCH算法聚类分析。研究结果表明,LDP-BIRCH算法在adult和Facebook Live Sellers in Thailand数据集上满足隐私保护性和聚类可用性。展开更多
当前分布式光伏系统短期发电功率预测结构多设定为目标式,预测范围在实际发电环境下受限,导致平均绝对预测误差增加。为此设计基于利用层次方法的平衡迭代规约和聚类(Balanced Iterative Reducing and Clustering Using Hierarchies,BIR...当前分布式光伏系统短期发电功率预测结构多设定为目标式,预测范围在实际发电环境下受限,导致平均绝对预测误差增加。为此设计基于利用层次方法的平衡迭代规约和聚类(Balanced Iterative Reducing and Clustering Using Hierarchies,BIRCH)的分布式光伏系统短期发电功率预测方法。首先,明确预测指标,采用多层级的方式设计预测结构;其次,结合BIRCH原理,设计发电功率预测模型;最后,采用梯度回归处理的方式来实现最终预测。测试结果表明,对比传统变分模态分解-麻雀搜索算法-反向传播(Variational Mode Decomposition-Sparrow Search Algorithm-Back Propagation,VMD-SSA-BP)光伏系统短期发电功率预测小组、传统时序动态回归光伏系统短期发电功率预测小组,此次所设计的方法得出的平均绝对预测误差被较好地控制在2.1以下,预测效果更佳,针对性更强,误差可控,具有实际的应用价值。展开更多
Genetic relationships of eight species of genus Betula were evaluated using ISSR marks. A total of 236 loci were generated from 17 ISSR primers. Percentage of polymorphic bands (PPB) varied from 5.93 to 19.92. The hig...Genetic relationships of eight species of genus Betula were evaluated using ISSR marks. A total of 236 loci were generated from 17 ISSR primers. Percentage of polymorphic bands (PPB) varied from 5.93 to 19.92. The highest and the lowest level of genetic differentiation were detected in B. ovalifolia and B. maximowicziana Regel respectively. In these eight species, genetic diversity of birch (HT) was 24.38 %, and the genetic variation (GST ) interspecies was accounting for 79.36% of total genetic variation. According to the cluster results of genetic distance, the eight species were classified into three groups as B. davurica, B. ovalifolia, B. platyphylla and B. pendula for one group; B. schmidtii, B. costata and B. ermanii Cham. var. communis for one group, and B. maximowicziana Regel for another group. The result of cluster is consistent with traditional morphological classification.展开更多
文摘精准的分布式光伏短期发电功率预测有助于电力系统运行与功率就地平衡。该文提出一种基于BIRCH(balanced iterative reducing and clustering using hierarchies)相似日聚类的L-Transformer(LSTM-Transformer)模型进行短期光伏功率预测。首先使用BIRCH无监督聚类算法对历史数据聚类得到3种典型天气,根据聚类结果划分测试集对模型进行训练。为提高不同天气类型下的预测精度,采用双层架构的L-Transformer模型,首层通过长短期记忆网络(long short term memory,LSTM)的门控单元机制捕捉时间序列中的长期依赖关系;次层结合Transformer模型的自注意力机制聚焦于当前任务更关键的特征量,通过多注意力头与光伏数据特征量相结合生成向量,注意力头并行计算,从而高效、精确地预测短期光伏功率。实测数据验证结果表明L-Transformer模型对于不同天气类型功率预测泛化性优异、精确度高,气象数据波动大时鲁棒性强。
文摘BIRCH算法是一种适合处理大规模数值型的聚类算法,但现实生活中的数据往往是混合型数据,导致了BIRCH算法的局限性;此外,在使用BIRCH算法进行聚类分析的过程中存在隐私泄露的风险,而传统的中心化差分隐私算法存在需要可信第三方的缺点。针对以上缺陷,提出了基于本地差分隐私的BIRCH混合数据(LDP-BIRCH)算法,对混合型数据中的非数值型数据进行编码处理,并使用本地差分隐私对数据集进行扰动,将扰动后的数据集发给第三方进行BIRCH算法聚类分析。研究结果表明,LDP-BIRCH算法在adult和Facebook Live Sellers in Thailand数据集上满足隐私保护性和聚类可用性。
基金国家自然科学基金(the National Natural Science Foundation of China under Grant No.60273043)安徽省自然科学基金(the Natural Science Foundation of Anhui Province of China under Grant No.050460402)
基金This paper was supported by Northeast Forestry University.
文摘Genetic relationships of eight species of genus Betula were evaluated using ISSR marks. A total of 236 loci were generated from 17 ISSR primers. Percentage of polymorphic bands (PPB) varied from 5.93 to 19.92. The highest and the lowest level of genetic differentiation were detected in B. ovalifolia and B. maximowicziana Regel respectively. In these eight species, genetic diversity of birch (HT) was 24.38 %, and the genetic variation (GST ) interspecies was accounting for 79.36% of total genetic variation. According to the cluster results of genetic distance, the eight species were classified into three groups as B. davurica, B. ovalifolia, B. platyphylla and B. pendula for one group; B. schmidtii, B. costata and B. ermanii Cham. var. communis for one group, and B. maximowicziana Regel for another group. The result of cluster is consistent with traditional morphological classification.