Abstract:In order to improve the reconstruction effect of the vibration signal in the process of blocked compressed sensing, an adaptive compressed sensing algorithm is proposed. Firstly, the signal is transformed into matrix by devising the signal into blocks, and the redundant diction is constructed by KSVD adaptively. Secondly, the complexity weight is defined by the attenuation rate, which comes from the sparse coefficients under the redundant dictionary. At last, based on the complexity weight, the observation number of different signal blocks is allotted adaptively in the guarantee of the same total sampling rate. While this algorithm is applied in the compressed sensing of machinery vibration signal, the reconstruction precision is higher than other compressed sensing algorithms.