In order to extract disturbance features accurately in noisy environment, a power quality disturbance detection and location algorithm based on improved wavelet threshold function denoising and singular value decomposition is proposed. The improved wavelet threshold function is used to denoise the noisy power quality disturbance signal. The frequency band division ability of empirical mode decomposition is used to separate the various modes of the disturbance signal after denoising. Hilbert transform is used to extract the characteristic information such as the amplitude and frequency of each mode. Meanwhile, the effective location of the start and stop time of the disturbance signal is realized by the principle of singular value decomposition. The simulation results of different kinds of disturbance signals show the effectiveness of the algorithm. The experiments show that the proposed algorithm not only has good anti-noise performance, but also has high positioning accuracy and good robustness.