Abstract: In the "double high" power system, the probability of wide-band oscillations is greatly increased. However, it is difficult for the traditional methods based on the wide area measurement system (WAMS) to adapt to the complex power grid due to their narrow ranges of the oscillation frequency and their restricted accuracy and rapidity. Therefore, a wide-band oscillations monitoring method based on the diagonalized low-density parity-check codes (LDPC) check matrix and the k-nearest neighbor (KNN) algorithm is proposed. Firstly, the power system signals are compressed based on the check matrix to reduce the transmission data, which is beneficial to achieve hundreds of Hz wide-band oscillation data transmission under the existing upload frequency of phasor measurement unit (PMU). On this basis, the master station uses the weighted KNN to detect the oscillation directly based on the compressed value as the input feature, avoiding the false judgment caused by the artificial setting threshold and improving the rapidity and accuracy of the detection. Finally, according to the detection results, the orthogonal matching pursuit (OMP) algorithm is used to accurately reconstruct the signals at the master station, which is convenient for the global analysis of the wide area system oscillations. The simulation results show that the proposed method can realize the fast and accurate monitoring of the wide-band oscillation in the case of noise, data missing and data error.