【作者】 郭景涛；

【导师】 金志刚；

【作者基本信息】 天津大学 ， 计算机应用技术， 2012， 博士

【摘要】 进入智能电网时代,大量电能表分散固定安装所导致的计量性能维护难题依然存在。AMI计量数据采集和处理的边际成本极低,然而对其挖掘和应用的研究尚在起步阶段,致使大量宝贵资源归于闲置。分时阶梯电价体系的推行使得经济用电问题成为只有专家才能解决的复杂规划问题。而智能电网AMI以其丰富的测量、通信、计算和存储资源,为这些问题的解决提供了广阔平台。论文各部分围绕AMI计量功能,并由此出发向下延伸研究了计量性能的维护,向上研究了计量数据的挖掘应用。研究内容涉及四个问题:(1)时间同步,此为分布式测控系统的基本问题;(2)AMI远程校准,旨在解决特定电能表的计量校准问题;(3)基于AMI计量数据的电能误差分析挖掘,实现对AMI的整体巡检,发现超差电表,使校准有的放矢;(4)家庭用电规划,为分时阶梯电价体系下家庭电力用户提供专业的用电经济规划服务。其中(1)的解决是前提,(2)(3)协同保障AMI计量性能的准确可靠,(3)(4)是对AMI计量数据的挖掘应用,体现AMI的内在价值。在时间同步问题的研究中,论文提出了二阶漂移补偿(SODC)误差修正算法和指数增长线性调整(EGLR)对钟间隔自适应算法相结合的时间同步方法。在处理非线性时钟漂移时,SODC算法具有较线性误差修正算法更小的残余误差和更大的漂移抑制比,从而加大对钟间隔,从而减小系统开销。EGLR算法在确定对钟间隔上具有更小的系统开销。在AMI远程校准系统的研究中,论文提出了通用远程校准平台设计的一般原则以及实现方案,给出了硬件和软件构成。实现了将互联网模块与标准仪器共同作为传递单元,通过互联网在远程主机的监控下完成校准。为实现通用性摆脱具体接口的束缚,将边缘分布函数(MDF)特征提取与神经元网络分类器相结合,很好地实现了图像字符数据的识别和采集。还对通用平台进行优化,得到AMI远程校准系统。并利用该平台实现对所提出的基于矢量合成的外磁场影响试验装置进行远程校准。论文将电能表、水表、气表抽象为一类具有树形拓扑且满足守恒关系的广义流量仪表。并证明若已知集群中任意仪表误差,无需借助外部标准仪器,仅通过对若干次集群的读数的分析挖掘,即可确定其余电能表的误差。在考虑了AMI中各种损耗后,给出了针对AMI的修正算法,并仿真验证了算法的有效性。在家庭用电规划研究中,将电费优化问题抽象为一类含积分函数的优化,给出了完整数学表达。提出了解析法与启发式方法相结合的混合优化方法以及基于全概率公式的优化效果评估方案。仿真表明,该方法能够显著降低电费。更多还原

【Abstract】 In the era of the smart grid, the challenge of calibration of large number of distributed fixed-installation electricity meters still remains. The marginal cost of AMI metering data acquisition and processing is very low, yet the study on mining and application of these data is still in its infancy, resulting in the idle of large number of valuable resources. The introducing of time-of-use and step tariff system makes using power efficiently out of ordinary user’s ability and the skeduling problem complex only solvable to experts. AMI and smart grid with rich resources of measuring, communicating, computing and storage provide a broad platform for the problems above.This dissertation focuses on the measuring function of AMI, underlying maintainence of measuring function and high level application of AMI data mining. The dissertation covers the 4 problems below. (1) Time synchronization. As is the basic problem in distributed measuring and control system. (2) AMI remote calibration, which aimes to solve the problem of calibrating specific electricity meter. (3) Meter error mining and analysis based on AMI data, which is to screen all the meters in AMI and locate the suspicious meter. (4) Household power consumption scheduling, providing professional service of economic planning for power consumer. Solving of problem (1) is the precondition of solving other problems. (2) and (3) together ensure the measurement accurate and reliable. (3) and (4) are AMI data mining and embody the value of AMI.In the study of time synchronization, the dissertation proposes a synchronizing algorithm which is the combination of second order drift compensation (SODC) algorithm for error correcting and exponential growing linear regulation (EGLR) self adaptive algorithm. When dealing nonlinear clock drift, SODC has a minor residual error and a greater drift rejection ratio. So the SODC can enlarge the interval between clock comparisons and reduce the system overhead. EGLR reduces the system overhead in determining the interval of two clock comparison operations.In the study of AMI remote calibration system, this dissertation presents the principles and scheme for desigination a general remote calibration platform, and the construction of hardware and software. The Internet module and standard instrument are combined to be the transmiting unit. The calibration is conducted under the monitoring by the remote host through Internet. To achieve maximum versatility and get rid of the constraint of certain interface, the image character realization method is introduced. The method combines marginal distribution function feature extraction and BP neural network classification. By optimizing the general platform, the AMI remote calibration system is achieved. The system is used in the remote calibration of magnetic fields test equipment which is based on synthetic vectors and detailed in the dissertation.This dissertation develops an autonomas algorithm for relative errors of meters in tree topology, and demonstrates that, giving the premise of energy conservation, providing any meter’s relative error, all the other meters’relative errors can be accomplished according to a group of meter readings. Considering the energy wastage in AMI, a revised algorithm is presented and proven to be effective by computer simulation.In the study of scheduling of household power consumption for Step and Time of Use (TOU) tariff system, this dissertation treats the problem of cost optimization of electricity as an optimization problem of function containing integration and provides the complete mathematic presentation. This dissertation proposes a hybrid optimization method which contains analysis stage and heuristic stage. An evaluation methodology is deduced to evaluate the optimization. The computer simulation demonstrates that the proposed approach can reduce the cost of electricity evidently in the sense of probability.更多还原