为了提高光伏发电功率预测精度,提出了一种基于长短期时序数据融合的Transformer生成式预测模型:LSTformer,能准确有效地预测光伏发电功率。LSTformer创新性地提出了时序分析模块(time series analysis,TSA)、时序特征融合模块(time ser...为了提高光伏发电功率预测精度,提出了一种基于长短期时序数据融合的Transformer生成式预测模型:LSTformer,能准确有效地预测光伏发电功率。LSTformer创新性地提出了时序分析模块(time series analysis,TSA)、时序特征融合模块(time series feature fusion,TSFF)和多周期嵌入模块(cycleEmbed),利用数据融合解决难以提取多时间尺度时序特征问题。设计时间卷积前馈(time convolution feedforward,TCNforward)单元,在编解码的过程中进一步提取时序特征。利用某光伏电站实际历史发电数据,通过实验验证LSTformer模型在光伏发电功率预测领域得到最低的均方误差(mean squared error,MSE)、平均绝对误差(mean absolute error,MAE),并通过消融实验验证了各模块的有效性。展开更多
针对目前机场群发展不平衡,国际枢纽机场的延误率居高不下,航班时刻短缺,资源紧张,而区域枢纽机场却存在资源空闲的问题,提出一种基于跳过门的长短时记忆网络(Skip-LSTM,Skip Long Short Term Memory)的机场群延误预测模型。该模型首先...针对目前机场群发展不平衡,国际枢纽机场的延误率居高不下,航班时刻短缺,资源紧张,而区域枢纽机场却存在资源空闲的问题,提出一种基于跳过门的长短时记忆网络(Skip-LSTM,Skip Long Short Term Memory)的机场群延误预测模型。该模型首先将机场群中各个机场的信息,机场群航班信息以及机场群地区的气象信息进行融合及处理,然后搭建Skip-LSTM网络对融合后的数据信息进行特征提取,最后利用Softmax分类器对机场群的延误状况进行分类预测。Skip-LSTM网络在传统的长短时记忆网络(LSTM,Long Short Term Memory)的基础上增加了Skip门,能更加充分地提取机场群数据信息的时间相关性,获得更高的准确率。实验结果表明,基于Skip-LSTM的机场群延误预测模型的准确率可达95.35%,预测性能优于传统的网络模型,能对机场群的延误状况进行有效的预测。展开更多
With the growing trend toward using cloud storage,the problem of efficiently checking and proving data integrity needs more consideration.Many cryptography and security schemes,such as PDP(Provable Data Possession) an...With the growing trend toward using cloud storage,the problem of efficiently checking and proving data integrity needs more consideration.Many cryptography and security schemes,such as PDP(Provable Data Possession) and POR(Proofs of Retrievability) were proposed for this problem.Although many efficient schemes for static data have been constructed,only a few dynamic schemes exist,such as DPDP(Dynamic Provable Data Possession).But the DPDP scheme falls short when updates are not proportional to a fixed block size.The FlexList-based Dynamic Provable Data Possession(FlexDPDP) was an optimized scheme for DPDP.However,the update operations(insertion,remove,modification)in Flex DPDP scheme only apply to single node at a time,while multiple consecutive nodes operation is more common in practice.To solve this problem,we propose optimized algorithms for multiple consecutive nodes,which including MultiNodes Insert and Verification,MultiNodes Remove and Verification,MultiNodes Modify and Verification.The cost of our optimized algorithms is also analyzed.For m consecutive nodes,an insertion takes O(m) + O(log N) + O(log m),where N is the number of leaf nodes of FlexList,a remove takes O(log/V),and a modification is the same as the original algorithm.Finally,we compare the optimized algorithms with original FlexList through experiences,and the results show that our scheme has the higher efficiency of time and space.展开更多
利用认证数据结构(ADS,authenticated data structures)的安全特性,分析并设计了面向云计算的基于ADS的数据外包认证模型,给出了模型的形式化定义、数据查询认证协议与数据更新认证协议;对ADS在模型实际应用时遇到的关键问题进行分析,...利用认证数据结构(ADS,authenticated data structures)的安全特性,分析并设计了面向云计算的基于ADS的数据外包认证模型,给出了模型的形式化定义、数据查询认证协议与数据更新认证协议;对ADS在模型实际应用时遇到的关键问题进行分析,设计了扩展数据一致性证据生成算法和扩展验证算法,从而实现了ADS在模型中的有效融入。最后从安全性和效率两方面对模型的性能进行分析比较,结果表明模型以较高效率实现了数据的正确性与一致性认证。展开更多
文摘针对目前机场群发展不平衡,国际枢纽机场的延误率居高不下,航班时刻短缺,资源紧张,而区域枢纽机场却存在资源空闲的问题,提出一种基于跳过门的长短时记忆网络(Skip-LSTM,Skip Long Short Term Memory)的机场群延误预测模型。该模型首先将机场群中各个机场的信息,机场群航班信息以及机场群地区的气象信息进行融合及处理,然后搭建Skip-LSTM网络对融合后的数据信息进行特征提取,最后利用Softmax分类器对机场群的延误状况进行分类预测。Skip-LSTM网络在传统的长短时记忆网络(LSTM,Long Short Term Memory)的基础上增加了Skip门,能更加充分地提取机场群数据信息的时间相关性,获得更高的准确率。实验结果表明,基于Skip-LSTM的机场群延误预测模型的准确率可达95.35%,预测性能优于传统的网络模型,能对机场群的延误状况进行有效的预测。
基金supported in part by the National Natural Science Foundation of China under Grant No.61440014&&No.61300196the Liaoning Province Doctor Startup Fundunder Grant No.20141012+2 种基金the Liaoning Province Science and Technology Projects under Grant No.2013217004the Shenyang Province Science and Technology Projects under Grant Nothe Fundamental Research Funds for the Central Universities under Grant No.N130317002 and No.N130317003
文摘With the growing trend toward using cloud storage,the problem of efficiently checking and proving data integrity needs more consideration.Many cryptography and security schemes,such as PDP(Provable Data Possession) and POR(Proofs of Retrievability) were proposed for this problem.Although many efficient schemes for static data have been constructed,only a few dynamic schemes exist,such as DPDP(Dynamic Provable Data Possession).But the DPDP scheme falls short when updates are not proportional to a fixed block size.The FlexList-based Dynamic Provable Data Possession(FlexDPDP) was an optimized scheme for DPDP.However,the update operations(insertion,remove,modification)in Flex DPDP scheme only apply to single node at a time,while multiple consecutive nodes operation is more common in practice.To solve this problem,we propose optimized algorithms for multiple consecutive nodes,which including MultiNodes Insert and Verification,MultiNodes Remove and Verification,MultiNodes Modify and Verification.The cost of our optimized algorithms is also analyzed.For m consecutive nodes,an insertion takes O(m) + O(log N) + O(log m),where N is the number of leaf nodes of FlexList,a remove takes O(log/V),and a modification is the same as the original algorithm.Finally,we compare the optimized algorithms with original FlexList through experiences,and the results show that our scheme has the higher efficiency of time and space.
文摘利用认证数据结构(ADS,authenticated data structures)的安全特性,分析并设计了面向云计算的基于ADS的数据外包认证模型,给出了模型的形式化定义、数据查询认证协议与数据更新认证协议;对ADS在模型实际应用时遇到的关键问题进行分析,设计了扩展数据一致性证据生成算法和扩展验证算法,从而实现了ADS在模型中的有效融入。最后从安全性和效率两方面对模型的性能进行分析比较,结果表明模型以较高效率实现了数据的正确性与一致性认证。
