Genetic algorithm for S-transform optimisation in the analysis and classification of electrical signal perturbations

At present, many analytical methods are used for the analysis, detection and classification of electrical signal perturbations. One of these methods, the S-transform, has proven effective under specific conditions for acquiring information and parameters of interest associated with a signal. However, depending on the nature of the signal and the input parameters, this method offers different results that sometimes negatively impact the quality of information obtained in the time and frequency domains.This paper describes the design of a genetic algorithm that optimises the S-transform for analysis and classification of the perturbations in electrical signals. This algorithm provides the best parameter values for optimising the Gaussian window, maximising the precision obtained with regard to classification and, later, analysis (via other techniques, such as neural networks or rule-based systems).This paper demonstrates the effectiveness of the S-transform (specified herein) with respect to the original S-transform and reports the best values obtained after optimisation via a comparative study that includes both typical cases and perturbations in modern electrical systems.