测控技术专业 毕业设计 英文翻译 用于模拟电路故障诊断的特征提取和多层次分类的算法比较.pdfVIP

Proceedings of the 2007 International Conference on Wavelet Analysis and Pattern Recognition, Beijing, China, 2-4 Nov. 2007 ALGORITHMS COMPARATION OF FEATURE EXTRACTION AND MULTI-CLASS CLASSIFICATION FOR FAULT DIAGNOSIS OF ANALOG CIRCUIT Abstract recognition methods. The estimation methods require mathematical process models that represent the real process For the novelties or anomalies of faulty signals occur satisfactorily, and if the model is complex, the application of in a damage circuit and fault signals vary with different the estimation methods is very limited in practice. However, circuit damages. To ensure the accuracy and reliability of no mathematical model of the process is required in the diagnosis, it is very important to extract the characteristic pattern recognition methods as the operation of the process features of fault signals. Two feature extraction methods is classified by matching the measurement data [1]. based on wavelet packet transform is proposed to treat The process of fault diagnosis of analog circuit consists transient signals: optimal wavelet packet transform (OWPT) of two stages: the feature extraction and the pattern and incomplete wavelet packet transform (IWPT). For the recognition. Automatic fault diagnosis and testing of analog fault signals decomposed, the energy in each frequency circuits is subjected to many difficult problems, such as poor band may be heightened or be reduced, so a novel fault models, noise, nonlinearity and tolerance effects [2]. ‘energy-fault’ method is put forward to extract fault f