【作者】 曲朝阳；

【导师】 穆钢；

【作者基本信息】 华北电力大学（河北） ， 电力系统及其自动化， 2010， 博士

【摘要】 智能电网是未来电网的一个主要发展方向,被认为是21世纪电力系统的重大科技创新和发展趋势。智能电网技术应包括先进的相量测量和广域测量技术;先进的三维、动态、可视化技术;分布式发电技术和电力储能技术等。其中先进的基于知识的三维、动态、可视化技术采用有别于传统二维显示的三维显示模式,以实物为模型,建立智能电网电力设备三维虚拟模型,逼真重现实际场景,给人以视觉上的冲击,使电网信息的表现更真实、丰富、具体、精确;从静态数据处理向动态、图形化发展,从根本上保证了整个电力系统稳定运行。本文在智能电网可视化方面做了如下研究:⑴研究并设计了“智能电网知识可视化引擎”的体系结构。针对传统可视化系统开放性差,二次开发与系统集成困难等弊端进行了深入研究,提出了基于SOA架构的知识可视化引擎的体系结构,利用引擎的知识推理功能实现了智能电网中海量数据的整合、挖掘、映射、虚拟展示,并将所有功能模块以构件方式进行集成,使得开发人员不必关心实现知识可视化所需各种关键技术的具体细节,直接就能够以引擎为核心开发智能电网中的各种可视化系统,从而大大减少开发人员的工作难度和工作量,加快智能电网可视化的建设速度,有效解决电力企业“信息孤岛”问题。⑵研究并提出了电网知识提取的方法。深入分析了电网海量流数据和历史数据的特点,并根据各自特点设计了电网知识分类模型,给出了电网知识提取的方法。对于流数据,提出采用滑动窗口模型、离散小波变换和改进的SSWAT算法进行预处理;并根据预处理后的流数据信息选取样本集,建立条件属性表以及决策表的可辨识矩阵,根据运算形成电网运行知识单元;建立基于PFCM的数学模型,并在此基础上完成知识推理。对于历史数据,提出了E-Grp算法对电网数据进行概化处理;并采用改进的Apriori算法和粗糙集算法完成电网关联知识和预测型知识的提取。⑶研究并提出了基于一体化模型的知识三维可视化方法。针对电力系统可视化应用的特点,提出了建立基于图元的知识可视化载体,并采用改进的BP神经网络算法进行一体化模型纹理特征的提取和模型的生成;同时提出了基于电气特性的场景组织方法和基于空间划分的动态负载均衡算法;实现了基于语义理解的虚拟现实交互机制。⑷设计了基于软件服务总线的知识可视化引擎控制机制。引擎服务总线控制机制包括传输适配器、服务适配器、核心服务、元数据库和系统管理五部分,它们按照对服务的需求变化划分服务粒度,并通过通用的传输方式进行发布、查找和捆绑等操作;彼此之间相互协作,按照业务规则共同完成整个业务流程。⑸基于智能电网知识可视引擎开发了变电站知识可视化应用系统,应用结果表明,知识可视化引擎一方面提高了智能电网相关可视化应用系统的开发效率,另一方面把底层的电网数据转变为电网知识,并基于虚拟现实技术进行三维展示,为电力系统安全运行提供了有力保障。更多还原

【Abstract】 Smart Grid is one of the main development direction of power grid in the future, which is considered as major technological innovation and development tendency in the 21st century. Smart-Grid technology includes advanced phasor measurement technique and widearea measurement technology; advanced three-dimensional dynamic visualization technology; Distributed Power Generation Systems and Electric Power energy management etc.A kind of advanced technology of 3-Dimension and dynamic visualization based on knowledge is different from the traditional 2-D or 3-D display mode, the advanced technology which is based on material objects, creats 3-D virtual models of electrical equipment in the smart grid and replicates real world scenarios realisticly in order to give people a visual impact and make the information expression of power system so real, rich, specific and accurate. From the static data processing to dynamic, graphical development, fundamentally, it guarantees the stable operation of the power system. In this paper, the following research has been done in the smart grid visualization.⑴The ’’smart power grid knowledge visualization engine’’ architecture is designed. It makes a deep research on the weak opennesses and difficulties for a new development and system integration with the traditional visualization system and suggestes a architecture that based on SOA knowledge visualization engine. It makes the massive data integration, mining, mapping, virtually shows in the smart power grid using the konwledge reasoning function of the engine. What’s more, it integrates the functional modules in a component way so that the developers don’t need to care about the details of key technology realizing the knowledge visualization, making it possible to use the engine as a core implement to develope a variety of visualization system in the smart power grid, thus greatly reducing the developer’s work and difficulties and speeding up the construction of the smart power grid visualization, effectively solve the problem -- ’’information islet’’ in many power enterprises.⑵The method of the knowledge extraction in the smart grid is presented in this paper. With the deep analysis of the features of massive data-flow of power grid and historical data, the classification model of knowledge is designed in accordance with the characteristics of themselves and the extraction method of knowledge is presented.For the flow of data, it proposes a sliding window model, discretes wavelet transform and improves SSWAT (approximation tree) algorithm to preprocess and selects sample set according to the flow of the data which has been preprocessed. Then it establishes condition attributes and decision-making table discernibility matrix with running knowledge unit according to the formation of grid. The completion of knowledge inference bases on the establishment of a mathematical model which relys on PFCM. For historical data, the paper introduces E-Grp algorithm for data pre-processing; and adopts improved Apriori algorithm and decision-tree algorithm to complete the extraction of association and predictive knowledge of the power system.⑶A model of knowledge three-dimensional visualization based on Integration Model is proposed. It visualizs the characteristics of applications for the power system, proposes the establishment of carrier based on the pixel visual, and uses the improved BP neural network algorithm for the integration model of texture feature extraction and model generation; It also proposes the scene organization method based on electrical characteristics and the dynamic load balancing algorithm based on space division, So that it achieves a virtual reality interaction mechanism based on semantic understanding .⑷The control mechanism of knowledge visualization engine which based on software service bus is design.Service bus includes Transport Adapter、Service Adapterr、Core Service、MetaData Base and System Management.They divide service granularity,distribute service,find service and bind service which cooperate each other is to accomplish the whole operation.⑸The substation experiment system of knowledge visualization which based on the knowledge-visualization engine of smart grid is developed. Application results show that on the one hand, the knowledge visualization engine improve the visualization of a smart grid-related application systems development efficiency, on the other, it transfomrs data that is from the bottom of the power grid operation into knowledge. At the same time,the knowledge is shown by three-dimensional images that are based on virtual reality technology. The engine make easy monitoring and maintenance of power system monitoring.更多还原