The impact of power quality disturbance is one of the major factors of power quality. S-transform (ST) is a very effective method for power quality disturbances analysis which provides frequency-dependent resolution while maintaining a direct relationship with the Fourier spectrum. This paper proposed a new approach for power quality disturbances recognition using modified S-transform with hyperbolic Gaussian window known as HS-transform and automatic pattern classifier is carried out using rule-based decision tree. Various non-stationary power signals are processed by HS-transform to generate time-frequency features in order to generate rules for disturbances pattern classification. Compare to traditional S-transform, HS-transform is more precisely localized in the time domain. 6 types of disturbances are classified by the rule-based decision tree and there is no need to use other complicated classifiers. Simulation results show that the proposed method is feasible and promising for real applications.