Architecture of an Intelligent Training System based on Virtual Environments for Electricity Distribution Substations Yasmín Hernández 1 , Miguel Pérez Ramírez 1 , William Ingram Ramírez 1 , Eric Nava Ayala 1 , Norma J. Ontiveros Hernández 2 1 Instituto Nacional de Electricidad y Energías Limpias, Tecnologías de la Información, Cuernavaca, Mexico 2 Instituto Tecnológico de Zacatepec, Sistemas y Computación, Zacatepec de Hidalgo, Mexico {mperez, myhp, jwingram}@iie.org.mx, ericna91@gmail.com, njoh314@yahoo.com.mx Abstract. In the electrical domain, qualified electricians are required since it involves danger. Training has been based on classroom courses and practice in the real installations, but it takes a long time and is costly. We propose to complement traditional training with an intelligent training system based on virtual environments in such a way that a blended training model is composed. We are looking for adaptive and innovative training; therefore we are including the characteristics and states of the student to adapt the instruction. We are developing a system to teach novice student how an electrical substation works as well as its individual compounding equipment. The proposed architecture for the intelligent training system and the development progress of the virtual environment are presented. Keywords: Adaptive training, electrical distribution, learning environment, learning styles, virtual reality. 1 Introduction Training is very important for every productive process, since if personnel are not well trained, accidents and losses could occur. In energy sector, training becomes a strategic aspect because accidents prevent the energy provision, which in turn affects industry and services sectors. Additionally, accidents could injure people. In most areas of electrical domain, the training takes a long time and is a costly process, because personnel have to move in order to practice in the actual installations. The substations are not always available because they are in operation in a regular basis. Additionally, very often there are not enough instructors available for the training demand. We have composed a blended model with the aim of improving electrical training. The model includes the aspects of traditional training: instruction in a classroom with a certified instructor and practice in real installations; and the novel component is conformed by intelligent training systems with virtual environments which allow 63 ISSN 1870-4069 Research in Computing Science 129 (2016) pp. 63–70; rec. 2016-07-20; acc. 2016-09-15