During the last few years, multi-objective optimization processes have become one of the main challenges for energy efficiency in buildings. In this work, a new efficient multi-objective optimization method, based on the Building Performance Optimization (BPO) technique, has been developed to improve the indoor thermal comfort and energy performance of residential buildings, i.e. a Moroccan ground floor + first floor (GFFF) house located in Marrakech region (5th climatic zone according to the Thermal Building Code in Morocco). The most influential design variables have been well explored in order to find the optimal trade-off between these two objectives. Indeed, this technique is based on the integration of Artificial Neural Networks (ANNs), in particular Multilayer Feedforward Neural Networks (MFNN), coupled with the most commonly used metaheuristic algorithms, i.e. Non-dominated Sorting Genetic Algorithm (NSGA-II), Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Genetic Algorithm (MOGA), in order to minimize computation time as much as possible. The TRNSYS software was used to establish the various dynamic thermal simulations required to create the database, from which the ANNs were able to set up their learning. The results show that this methodology is being used successfully, leading to different proposed solutions in terms of building envelope design. However, only the solutions using MOPSO are finally retained, as they have shown the greatest desired performance compared to the others. Thus, the thermal needs, particularly those for heating and cooling, have been significantly reduced to 74.52% of the total, while improving the indoor thermal comfort by 4.32% compared to the base design. Finally, we strongly recommend this methodology to the different actors in this field, including designers, engineers, architects, engineering offices, etc., when several objectives need to be contrasted while simultaneously considering several design variables.

在过去的几年中，多目标优化过程已成为建筑物节能的主要挑战之一。在这项工作中，已经开发了一种基于建筑性能优化（BPO）技术的新型高效多目标优化方法，以改善住宅建筑的室内热舒适性和能源性能，即摩洛哥底层+底层（GFFF） ）的房屋位于马拉喀什地区（根据摩洛哥《热力建筑规范》第5个气候带）。为了找到这两个目标之间的最佳平衡，已经对最有影响力的设计变量进行了充分的探索。确实，这项技术基于人工神经网络（ANN），特别是多层前馈神经网络（MFNN）的集成，结合最常用的元启发式算法，即非支配排序遗传算法（NSGA-II），多目标粒子群优化（MOPSO）和多目标遗传算法（MOGA），以最大程度地减少计算时间可能的。TRNSYS软件用于建立创建数据库所需的各种动态热仿真，ANN可以从该数据库进行学习。结果表明，该方法已被成功使用，从而导致在建筑围护结构设计方面提出了不同的解决方案。但是，仅保留了使用MOPSO的解决方案，因为与其他解决方案相比，它们表现出了最大的期望性能。因此，热能需求，特别是用于加热和冷却的热能需求，已大大减少至总需求的74.52％，与基础设计相比，室内热舒适度提高了4.32％。最后，当需要对比几个目标并同时考虑多个设计变量时，我们强烈建议将此方法推荐给该领域的不同参与者，包括设计师，工程师，建筑师，工程办公室等。