Uncorrected Author Proof Journal of Intelligent & Fuzzy Systems xx (20xx) x–xx DOI:10.3233/JIFS-191923 IOS Press 1 Recognition of human emotions based on user context and brain signals applied to electrical power systems operators evaluation 1 2 3 4 Taciana Saad Rached, Maria de F´ atima Queiroz Vieira, Danilo Santos, Angelo Perkusich ∗ and Hyggo Almeida 5 6 Embedded and Pervasive Computing Laboratory, Electrical Engineering Department, Electrical Engineering and Informatics Center, Federal University of Campina Grande, Campina Grande, PB, Brazil 7 8 Abstract. In this article, we propose a method to recognize human emotions based on user context and brain signals. We evaluated the method through an experiment during which individuals performed tasks using a simulator for electrical power systems operator training. We collected user context through log data retrieval and brain signals using an Electroencephalog- raphy (EEG) portable monitor. The experimental results demonstrated that the method could be successfully applied to recognize the emotional states based on EEG signals and user context. 9 10 11 12 13 Keywords: Emotion recognition, electroencephalography, signal processing, context-awareness 14 1. Introduction 15 The application of emotion recognition to a variety 16 of fields such as elderly care, robotics, medicine, avi- 17 ation, and automation systems, among others [–3], is 18 increasing in the last years. Also, a variety of tech- 19 niques to acquire emotion information, such as facial 20 expressions [4], voice signals [5], body language [6], 21 and physiological signals [7], have been reported. The 22 central problem in emotion recognition is its suscep- 23 tibility to ambiguous interpretation by analysts and 24 dissimulation from the subject. Such a situation is 25 mainly the case when using systems based on facial 26 analysis, spoken language, or body language, com- 27 ∗ Corresponding author. Angelo Perkusich, Embedded and Per- vasive Computing Laboratory, Electrical Engineering Department, Electrical Engineering and Informatics Center, Federal University of Campina Grande, CP 10105, 58429-970, Campina Grande, PB, Brazil. E-mail: perkusic@embedded.ufcg.edu.br. promising the accuracy of the data [8]. As an example, 28 an actor can simulate several emotions regardless of 29 the true emotional state. Besides, such systems are 30 not universal, depending on culture, gender, and age 31 of the individual [9]. 32 In contrast, even if a person does not express 33 his emotions through voice, facial expressions, or 34 gestures, changes in physiological state are uncon- 35 trollable and thus detectable [10], because the 36 sympathetic nerves of the autonomic nervous sys- 37 tem are activated when an individual is positively or 38 negatively excited. Another advantage is the impossi- 39 bility to manipulate physiological signals in emotion 40 recognition to simulate a false emotional state, which 41 makes these systems more reliable [11]. 42 Despite being immune to dissimulation, biological 43 signals may present a low temporal resolution, which 44 prevents their use in real-time systems. An exception 45 to this case is the brain electrical signal acquired by 46 ISSN 1064-1246/20/$35.00 © 2020 – IOS Press and the authors. All rights reserved