基于云理论的知识表示，首次将云理论与电力系统有效结合来解决配电系统原始数据不足的问题。在对原始数据进行模糊聚类的基础上，提出了一种组合云发生器，将正向云发生器和逆向云发生器有效组合成一种闭环结构。结合配电网负荷数据的实际特点，在组合云发生器中加入了约束方程组和系统工况数据补充两个单元，保证了生成的负荷数据既能够包含系统的大部分情况，又不会出现实际不会发生的不可能数据，生成的云滴很好的反映了负荷数据所具有的模糊性和随机性。并在此基础上进行了基于T-S模糊模型的负荷辨识，将辨识结果与当前通用的几类模型对比显示了所提出方法的有效性和实用性。

During the system fault period, usually the explosive growth signals including fuzziness and randomness are too redundant to make right decision for the dispatcher. So intelligent methods must be developed to aid users in maintaining and using this abundance of information effectively. An important issue in fault diagnosis system (FDS) is to allow the discovered knowledge to be as close as possible to natural languages to satisfy user needs with tractability, and to offer FDS robustness. At this junction, the cloud theory is introduced. The mathematical description of cloud has effectively integrated the fuzziness and randomness of linguistic terms in a unified way. A cloud-rough model is put forward Based on it, a method of knowledge representation in FDS is developed which bridges the gap between quantitative knowledge and qualitative knowledge. In relation to classical rough set, the cloud-rough model can deal with the uncertainty of the attribute and make a soft discretization for continuous ones. A novel approach, including discretization, attribute reduction, valve reduction and data complement, is presented. The data redundancy is greatly reduced based on an integrated use fo cloud theory and rough set theory. Illustrated with a power distribution FDS shows the effectiveness and practicality of the proposed approach.

The volume of data with a few uncertainties overwhelms classic information systems in the distribution control center and exacerbates the existing knowledge acquisition process of expert systems. The paper describes a systematic approach for detecting superfluous data. It is considered as a “white box” rather than a “black box” like in the case of neural network. The approach therefore could offer user both the opportunity to learn about the data and to validate the extracted knowledge. To deal with the uncertainty and deferent structures of the system, rough sets and fuzzy sets are introduced. The reduction algorithm based on uncertainty rough sets is improved. The rule reliability is deduced using fuzzy sets and probability. The worked example and simulation result of a power distribution system shows the effectiveness and usefulness of the approach.

提出了一种基于断点重要性的配电网连续属性离散化方法，证明了该方法的有效性。综合考虑了线路的选择系数、灵敏系数和支持权值，应用属性离散指标作为离散化的评价标准，证明该指标可以作为离散化彻底的充分条件。并且在时间复杂度和空间复杂度方面分析了算法的有效性，与同类算法的比较可以发现算法在基本不损失分类信息的基础上有效降低了时间复杂度和空间复杂度。首次在配电系统故障诊断中将连续和离散信号统一的应用粗糙集进行约简，使电压、电流等配电系统的重要模拟量可以参与到故障诊断系统中，提高了故障辨识的精度。同时，也使对于配电系统的临界稳态运行情况的判定提供了一个可能的解决办法。

针对在实际的配电系统故障诊断中，往往要面对从海量数据中找到真正对于诊断结果有帮助的关键数据问题，提出了一类基于遗传算法的粗糙集约简算法，可以在较少信息的情况下获得正确的诊断结果。在分析了遗传算法中各个主要参数对算法结果的影响基础上，重点讨论了适应值函数中目标函数、惩罚函数以及惩罚因子的构造方法，并对于其他关键参数及算法进化过程针对粗糙集约简特点进行了修正。与传统的粗糙集约简算法相比，文中的方法能够有效找到最优的属性约简结果，同时大大提高了算法的效率和实用性。将算法应用于美国PG&E的69节点配电系统进行仿真，对于202个属性，319条记录的复杂数据进行了有效的约简，结果表明算法对于实际的复杂配电系统能够进行故障诊断。