摘要：模糊聚類是通過模糊數(shù)學(xué)方法對已知類別的各參數(shù)進行比較，找出規(guī)律來推斷和預(yù)報未知類別的屬性。本文以轉(zhuǎn)子故障數(shù)據(jù)為對象，運用了模糊聚類方法和C-均值聚類方法對轉(zhuǎn)子故障數(shù)據(jù)進行了處理，利用Matlab工具對轉(zhuǎn)子故障診斷的系統(tǒng)進行了仿真和調(diào)試，并進行了可行性驗證。
完成設(shè)計內(nèi)容如下：
1、用模糊聚類方法對轉(zhuǎn)子故障數(shù)據(jù)進行處理，建立模糊矩陣，從而求得相似系數(shù)，來進行數(shù)據(jù)的標定。然后使用傳遞閉包法對標定的模糊相似矩陣用平方法計算，利用Matlab求得傳遞閉包，最后進行聚類。
2、用C-均值聚類方法利用公式計算轉(zhuǎn)子故障數(shù)據(jù)的相似矩陣，利用Matlab從初始矩陣算出迭代次數(shù)，得到模糊矩陣，并對此進行聚類。將方法應(yīng)用于抽風(fēng)機故障數(shù)據(jù)，再次進行驗證。

關(guān)鍵詞：轉(zhuǎn)子故障診斷；模糊聚類；相似矩陣；傳遞閉包


Rotor Fault Diagnosis and Analysis
Based on Fuzzy clustering Method
Abstract: Fuzzy clustering is comparing the parameters of the known types through fuzzy mathematics to find out the laws which could be used in indentifying and forecasting the properties of the unknown categories. As rotor fault data as the object, this thesis has used both fuzzy clustering method and C-means clustering method to process the rotor fault data. At the same time, the system of rotor fault diagnosis has been simulated and debugged by Matlab, which has verified the feasibility of fuzzy clustering method.
Its basic design process is as follows:
1, Using fuzzy clustering method for rotor fault data processing, fuzzy matrix, and thus achieve a similar coefficient to the calibration data. Then use the transitive closure method of calibration of the fuzzy similarity matrix calculated using Matlab to seek transitive closure, the final cluster.
2, Using C-means clustering method of rotor fault using a formula similar to the data matrix, use of Matlab matrix calculated from the initial iteration, the fuzzy matrix, and this cluster. Method has been applied to fault data exhaust fan once again to verify.

Keywords: Rotor fault diagnosis; Fuzzy clustering; Similarity matrix; Transitive closure.

Classification:TP2


目 次
摘 要 II
目 錄 III
1.引言 1
1.1國內(nèi)外研究現(xiàn)狀 1
1.1.1模糊聚類應(yīng)用領(lǐng)域 2
1.1.2 其他用于故障診斷的方法 3
1.2 研究思路 3
1.2.1 理論基礎(chǔ)和分析 3
1.2.2 研究設(shè)想、方法 6
1.3研究的意義 6
2. 模糊聚類的原理 7
2.1 模糊聚類 7
2.1.1模糊聚類的理論基礎(chǔ) 7
2.1.2 聚類的方法 9
2.2 模糊邏輯原理 12
3.轉(zhuǎn)子故障的診斷與分析 14
3.1模糊聚類法對故障數(shù)據(jù)進行診斷與分析 17
3.1.1聚類步驟 18
3.1.2 利用Matlab程序得到的聚類譜系圖 19
3.2模糊C-均值聚類方法對故障數(shù)據(jù)進行診斷分析 19
3.2.1 模糊C-均值聚類步驟 20
3.2.2 聚類譜系圖的獲得 22
3.3 聚類與分析 24
3.3.1二十組數(shù)據(jù)測試 24
3.3.2四十組數(shù)據(jù)測試 25
3.3.3應(yīng)用于抽風(fēng)機故障數(shù)據(jù)驗證 26
4.總結(jié) 29
參考文獻 29
附錄 主要程序和矩陣 31



參考資料
[1] Zhao zhenyun, Xu yong mou,The basic and applications of fuzzy theory and neural networks，tsinghua university publish，1996.
[2] Wang dazhong et al.，“fuzzy theory ,expert systems, neural networks and their applications in fault diagnosis of the fourier transform”，Proceedings of the Chinese society for electrical engineering ，Vol.16, No.5, pp.349-353,1996.
[3] S.K.Bhattacharyya, et al.,“A neural network approach to transformer fault diagnosis using dissolved gaanalysis data”, NAPS’93, 1993.
[4] Wang caisheng，“The fault diagnosis method BPNN in detecting for fourier transform ”，Proceedings of the Chinese society for electricaengineering，Vol.17, No.5, pp.322-325, 1997．
[5] Baldi,“Neural networks and principal component analysis:learning from examples and local minima”,Neural Networks, 2:53-58, 1989.