为解决石油行业大数据高效安全传输难题,将图像转化为三通道矩阵数据并对矩阵进行奇异值分解,综合考虑Laplace算子、灰度方差函数、离散余弦变换系数、图像相关系数、熵函数、图像结构相似度和图像信噪比等7个评价因素,利用熵权逼近理...为解决石油行业大数据高效安全传输难题,将图像转化为三通道矩阵数据并对矩阵进行奇异值分解,综合考虑Laplace算子、灰度方差函数、离散余弦变换系数、图像相关系数、熵函数、图像结构相似度和图像信噪比等7个评价因素,利用熵权逼近理想解排序(technique for order preference by similarity to an ideal solution,TOPSIS)法对分解后的奇异值进行优选,在确保数据真实性的前提下用少量奇异值表征原始图像,进行图像压缩,降低数据大小,提高传输效率;提出多通道猫脸分割加密方法,分别对每个颜色通道的图像进行随机分割、随机加密和随机排序,解决了传统猫脸加密算法颜色通道线性相关度高,整体置乱度低的问题.结果表明:改进奇异值压缩技术在保证图像清晰的情况下仅利用15%的奇异值数据完成对图像的压缩,最大图像压缩比可达4.43,平均压缩后所占用的存储空间仅为原空间的26.29%,数据传输控制协议通信平均传输效率提高86.39%.在加密图像达到0相关的前提下,多通道猫脸分割加密算法加密图像在像素点处三通道颜色值完全不同,新方法颜色通道相关系数分别为0.20、0.22和0.25,对比传统猫脸加密方法,分别降低0.78、0.75和0.71.新方法加密效果好、难破解,可为石油行业数字化转型提供理论和技术支撑.展开更多
在过程监控中,使用现代工业系统中的变量进行准确有效的监控诊断仍然是一个具有挑战性的任务.本文以多元指数加权移动平均(MEWMA)策略结合一种有监督分类器(“one plus epsilon”,简称OPE分类器),提出OPE-MEWMA控制图.在考虑不同模型、...在过程监控中,使用现代工业系统中的变量进行准确有效的监控诊断仍然是一个具有挑战性的任务.本文以多元指数加权移动平均(MEWMA)策略结合一种有监督分类器(“one plus epsilon”,简称OPE分类器),提出OPE-MEWMA控制图.在考虑不同模型、偏移模式和偏移大小的情况下,探究了控制图对均值偏移的检测能力,通过比较平均运行长度等多个指标衡量控制图的性能表现.仿真结果表明,所开发的OPE-MEWMA控制图能够快速检测到均值偏移,灵敏度较高.展开更多
Some typical structural schemes of Fuzzy control have been surveyed. Besides general structure of fuzzy logic controller (FLC), the structural schemes include PID fuzzy controller, self-organizing fuzzy controller, se...Some typical structural schemes of Fuzzy control have been surveyed. Besides general structure of fuzzy logic controller (FLC), the structural schemes include PID fuzzy controller, self-organizing fuzzy controller, selftuning fuzzy controller, self-learning fuzzy controller, and expect fuzzy controller, etc. This survey focuses on the control principle, and provides a basis for potential applications. Most of the structures have been used in various control fields, one of application areas is in the metallurgy industry, e. g., the temperature control of the electric furnace, the control of the aluminum smelting process, etc. According to the application requirements, one can choose a structural scheme for special use.展开更多
文摘为解决石油行业大数据高效安全传输难题,将图像转化为三通道矩阵数据并对矩阵进行奇异值分解,综合考虑Laplace算子、灰度方差函数、离散余弦变换系数、图像相关系数、熵函数、图像结构相似度和图像信噪比等7个评价因素,利用熵权逼近理想解排序(technique for order preference by similarity to an ideal solution,TOPSIS)法对分解后的奇异值进行优选,在确保数据真实性的前提下用少量奇异值表征原始图像,进行图像压缩,降低数据大小,提高传输效率;提出多通道猫脸分割加密方法,分别对每个颜色通道的图像进行随机分割、随机加密和随机排序,解决了传统猫脸加密算法颜色通道线性相关度高,整体置乱度低的问题.结果表明:改进奇异值压缩技术在保证图像清晰的情况下仅利用15%的奇异值数据完成对图像的压缩,最大图像压缩比可达4.43,平均压缩后所占用的存储空间仅为原空间的26.29%,数据传输控制协议通信平均传输效率提高86.39%.在加密图像达到0相关的前提下,多通道猫脸分割加密算法加密图像在像素点处三通道颜色值完全不同,新方法颜色通道相关系数分别为0.20、0.22和0.25,对比传统猫脸加密方法,分别降低0.78、0.75和0.71.新方法加密效果好、难破解,可为石油行业数字化转型提供理论和技术支撑.
文摘在过程监控中,使用现代工业系统中的变量进行准确有效的监控诊断仍然是一个具有挑战性的任务.本文以多元指数加权移动平均(MEWMA)策略结合一种有监督分类器(“one plus epsilon”,简称OPE分类器),提出OPE-MEWMA控制图.在考虑不同模型、偏移模式和偏移大小的情况下,探究了控制图对均值偏移的检测能力,通过比较平均运行长度等多个指标衡量控制图的性能表现.仿真结果表明,所开发的OPE-MEWMA控制图能够快速检测到均值偏移,灵敏度较高.
基金Supported by National Natural Science Foundation of China (60474002, 60674041) and National High Technology Research and Development Program of China (863 Program) (2006AA04Z173)
文摘Some typical structural schemes of Fuzzy control have been surveyed. Besides general structure of fuzzy logic controller (FLC), the structural schemes include PID fuzzy controller, self-organizing fuzzy controller, selftuning fuzzy controller, self-learning fuzzy controller, and expect fuzzy controller, etc. This survey focuses on the control principle, and provides a basis for potential applications. Most of the structures have been used in various control fields, one of application areas is in the metallurgy industry, e. g., the temperature control of the electric furnace, the control of the aluminum smelting process, etc. According to the application requirements, one can choose a structural scheme for special use.