Detection of power quality disturbances with overcomplete dictionary matrix and ℓ1-norm minimization

In this paper, we present a new automatic transient detection and localization for analysis of power quality disturbances. The proposed method is based on an over-complete dictionary (OCD) matrix and an ℓ₁-norm minimization algorithm. The OCD matrix is constructed using the spike-like (or identity) bases and discrete-cosine bases. The proposed method is validated using the four types of transient events and the results are compared with wavelet-based method. Experiment results show that the proposed method provides accurate time-information of impulsive and oscillatory transients under high levels of noise, and also preserves signature of transient events.