基于多分辨率技术及奇异值理论的故障检测方法

提出了以小波多分辨率技术与矩阵奇异值理论相结合的故障检测方法。根据小波变换的多分辨率分解特性 ,提出了系统状态观测信号的二初始特征向量矩阵———粗分辨逼近矩阵和边缘细节信息矩阵。利用矩阵奇异值分解理论得到初始特征向量矩阵的奇异值 ,将其作为状态信号的特征向量。针对提取出的系统状态信号奇异值特征 ,设计出相应的故障检测算法 ,并将该方法用于某武器平台上精密弹簧阻尼器的故障检测。仿真结果证实了该方法的正确性和有效性。

Fault Detection Method Based on Wavelet Multiresolution Technology and Singular Value Theory

A fault detection method based on wavelet multiresolution analysis and singular value theory is presented, and its feasibility and advantages are demonstrated. By using wavelet multiresolution function, two initial feature vector matrixes of system state observation signal, i.e.approximation approach component matrix and detailed information component matrix, are proposed. By using matrix singular value decomposition theory, two groups of singular values of initial feature vector matrixes are obtained, which are regarded as the state signal feature vectors. Based on the extracted system state signal singular value feature, a new fault detection algorithm is worked out. The algorithm is applied to the fault detection of the precision spring damper which is installed on a weapon platform. The validity of the fault detection method has been demonstrated by a simulation.