In the analysis and classification of signals from massive databases, it is highly desirable to use automatic mechanisms. The synergy of artificial intelligence and advanced signal processing techniques is becoming very efficient in developing this kind of task. In this work we employ a signal processing strategy based on the wavelet transform and then genetic algorithms for classification purposes. An in-depth analysis of the waveforms has been carried out, and an analytical preprocessing has been applied to prepare the signals for their classification. Each individual of the simulated population represents a classifying rule, composed of an antecedent and a consequent. The codification of the knowledge is one of the main contributions of this paper. This genetic classification system has been applied to six different classes of plasma signals of the TJ-II stellarator database at CIEMAT in Spain with satisfactory results.