论文部分内容阅读
以较少的模糊规则在全局范围内获得满意的辨识精度,是模糊辨识追求的重要目标。该文提出了一种基于最差子空间分解聚类的非线性系统模糊辨识方法。根据各子空间的“可线性化”程度,对聚类的有效性进行评判,进一步对有效性最差的子集进行重新分解聚类,并辨识新增子空间的模型参数,以此逐步完成整个样本空间的模糊划分和模型辨识过程,直至模型满足既定要求。文中给出了所提出的模糊辨识方法与其他相关模糊辨识方法的对比结果,并利用该方法对2个典型热工对象进行了模糊辨识。
Obtaining satisfactory identification accuracy with fewer fuzzy rules in the global scope is an important goal of fuzzy identification. This paper presents a fuzzy system identification method based on the worst subspace decomposition clustering. According to the degree of “linearizability” in each subspace, the validity of the clustering is judged, the subsets with the weakest validity are further decomposed and clustered, and the model parameters of the newly added subspace are identified, Gradually complete the fuzzy partition of the entire sample space and model identification process until the model meets the established requirements. In this paper, the contrast between the proposed fuzzy identification method and other related fuzzy identification methods is given. By using this method, fuzzy identification of two typical thermal objects is carried out.