近年来,图像信息的传输安全性已经成为互联网领域的重要研究方向.本文提出了一种基于量子长短期记忆(quantum long-short term memory,QLSTM)网络的量子图像混沌加密方案.结果发现,因为QLSTM网络具有复杂的结构和较多的参数,应用QLSTM...近年来,图像信息的传输安全性已经成为互联网领域的重要研究方向.本文提出了一种基于量子长短期记忆(quantum long-short term memory,QLSTM)网络的量子图像混沌加密方案.结果发现,因为QLSTM网络具有复杂的结构和较多的参数,应用QLSTM网络对Lorenz混沌序列进行改进,其最大Lyapunov指数比原序列提高2.5465%,比经典长短期记忆(long-short term memory,LSTM)网络改进的序列提高0.2844%,同时在0—1测试中结果更接近1且更稳定,因此QLSTM网络改进的序列具备更优异的混沌性能,更难以被预测,提高了单一混沌系统加密的安全性.运用NCQI(novel quantum representation of color digital images)量子图像表示模型,将原始图像存储为量子态形式,利用QLSTM网络改进的序列分别控制三级径向扩散、量子广义Arnold变换和量子W变换,改变量子图像的灰度值与像素位置,生成最终的加密图像.本文提出的加密方案在统计学特性测试中,实现了RGB三通道平均信息熵均大于7.999,像素数改变率的平均值达99.6047%,统一平均变化强度的平均值为33.4613%,平均相关性为0.0038等,比其他一些传统方法具有更高的安全性,能够抵抗常见的攻击方式.展开更多
The three-dimensional internal flow field of centrifugal pump is complex and variable with design parameters and operation conditions. The post-processing technique named differential amplification method was proposed...The three-dimensional internal flow field of centrifugal pump is complex and variable with design parameters and operation conditions. The post-processing technique named differential amplification method was proposed for the comparison study of different flow structures. The full steady flow fields of an industrial centrifugal pump working on-design and off-design points were numerically investigated by solving Reynolds average Navier-Stokes equations together with a shear-stress transport(SST) k-? turbulence model. And the numerically predicted performance curves of the studied pump agree well with test measurement results. Compared with the flow flied on design point under the help of differential amplification method, the disturbance caused by interaction between blade and volute tongue is very obvious and it extends to the diffuser pipe on the working point with 0.8 times rated flux. While on the point with 1.2 times rated flux, the flow distribution in impeller region is roughly even and it flows more to the bottom section of the diffuser pipe. The above method was proved to be good at displaying the subtle secondary flow structure changes with a higher resolution effect relative to single isolated case observation, which helps the optimization decision-making from multiple design cases.展开更多
文摘近年来,图像信息的传输安全性已经成为互联网领域的重要研究方向.本文提出了一种基于量子长短期记忆(quantum long-short term memory,QLSTM)网络的量子图像混沌加密方案.结果发现,因为QLSTM网络具有复杂的结构和较多的参数,应用QLSTM网络对Lorenz混沌序列进行改进,其最大Lyapunov指数比原序列提高2.5465%,比经典长短期记忆(long-short term memory,LSTM)网络改进的序列提高0.2844%,同时在0—1测试中结果更接近1且更稳定,因此QLSTM网络改进的序列具备更优异的混沌性能,更难以被预测,提高了单一混沌系统加密的安全性.运用NCQI(novel quantum representation of color digital images)量子图像表示模型,将原始图像存储为量子态形式,利用QLSTM网络改进的序列分别控制三级径向扩散、量子广义Arnold变换和量子W变换,改变量子图像的灰度值与像素位置,生成最终的加密图像.本文提出的加密方案在统计学特性测试中,实现了RGB三通道平均信息熵均大于7.999,像素数改变率的平均值达99.6047%,统一平均变化强度的平均值为33.4613%,平均相关性为0.0038等,比其他一些传统方法具有更高的安全性,能够抵抗常见的攻击方式.
基金Project(2014GK3150)supported by Science and Technology Plan of Hunan Province,China
文摘The three-dimensional internal flow field of centrifugal pump is complex and variable with design parameters and operation conditions. The post-processing technique named differential amplification method was proposed for the comparison study of different flow structures. The full steady flow fields of an industrial centrifugal pump working on-design and off-design points were numerically investigated by solving Reynolds average Navier-Stokes equations together with a shear-stress transport(SST) k-? turbulence model. And the numerically predicted performance curves of the studied pump agree well with test measurement results. Compared with the flow flied on design point under the help of differential amplification method, the disturbance caused by interaction between blade and volute tongue is very obvious and it extends to the diffuser pipe on the working point with 0.8 times rated flux. While on the point with 1.2 times rated flux, the flow distribution in impeller region is roughly even and it flows more to the bottom section of the diffuser pipe. The above method was proved to be good at displaying the subtle secondary flow structure changes with a higher resolution effect relative to single isolated case observation, which helps the optimization decision-making from multiple design cases.