This paper deals with the control of the blind system for building to prevent the thermal discomfort, which may occur during mid-season and summer. In the first part of this article, the problem is formulated with a Hybrid Model Predictive Control approach. However, the implementation of this solution on an industrial product remains a big challenge due to the limited computing capacity of this latter. As a result, we propose in the second part a logical controller whose rules and parameters are based on a learning from the behavior of the optimal controller using support vector machine (SVMs). In addition, we aim to demonstrate the importance of a proper selection of the measurements for the learning problem. In the last part, various simulations are performed with SIMBAD, a Matlab toolbox dedicated to building simulation, to assess the performance of the learned controller. (C) 2014 Elsevier B.V. All rights reserved.