基于频域分割的电能质量复合扰动识别技术研究

项目名称： 基于频域分割的电能质量复合扰动识别技术研究

项目编号： No.51307020

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 电工技术

项目作者： 黄南天

作者单位： 东北电力大学

项目金额： 26万元

中文摘要： 电力系统电能质量复合扰动识别是暂态识别领域中的重点与难点问题。现有复合扰动识别研究尚存在信号处理方法特征表现能力不足，特征选择困难，复合扰动识别种类有限、识别率低等问题。针对上述问题，本项目重点展开以下研究：1)建立S变换窗函数的自适应调整方法，提高其时-频特征表现能力；2)对原始信号类型集合进行基于扰动成分频域分布的粗分类，设计针对不同扰动类型子集的特征选择算法，以满足含多类复合扰动信号识别的特征选择要求；3)建立基于直觉模糊理论的频域分割理论，S变换结果将被精确分割为若干频域，并针对不同频域设计分类器，分别建立基于概率神经网络、可变形模型模板和直觉模糊聚类的频域分类器；4)设计各分类器分类结果信息融合方法，融合不同分类器分类结论，实现复合扰动的准确识别。本项研究将实现电能质量复合扰动的精确识别，对暂态扰动治理与电能质量控制具有重要意义。

中文关键词： 电能质量；复合扰动；多分辨率广义S变换；频域分割；模式识别

英文摘要： The complex power quality disturbances recognition is one of the most important and difficult issues in short duration disturbances recognition. However, the existing researches still have shortages, such as poor time-frequency presentation abilities in signal processing, difficulties in feature selection, and low recognition accuracies for limited types of complex disturbances. Aiming to solve these problems, this project will focus on the following issues. 1) An adaptive algorithm to adjust the factors of window function for S-transform will be constructed to get better time-frequency presentation abilities. 2) A rough classification method will be proposed to separate the original disturbances into subsets by exploring the frequency distributions of disturbance components. Feature selections will be realized respectively in different subsets to satisfy the requirements of multiple types of complex disturbances recognition. 3) A frequency segmentation algorithm based on Intuitionistic Fuzzy Sets theory will be constructed to segment the S-transform results into several frequency domains. The classifier in each domain will be separately designed by using probabilistic neural network, deformable model template and intuitionistic fuzzy clustering methods.4) An information fusion method will be proposed to combine

英文关键词： power quality；complex disturbance；a multi-resolution generalized S transform；frequency segmentation；pattern recognition