Heritage preservation presents uncertainties in the global warming context. Both predicting the performance of environmental variables within the built environment and assessing the preservation conditions of heritage elements are today something of a challenge. Based on current monitorings, this study uses artificial intelligence to predict environmental performance in future scenarios. In addition, heritage elements are assessed according to threshold preservation values. The Chapel of the University of Seville was chosen as case study, applying both the threshold values from UNI 10829 and the performance index (PI) to assess the indoor environmental conditions. A total of 9 environmental treatment zones were identified according to temperature and humidity. To estimate the variables, the multilayer perceptron (MLP) was monitored and applied considering the time series of A2 future scenarios in each decade of the 21st century. The current MLP performances were analysed, thus showing a progressive reduction of the PI inside the church: at the end of the century, PI values oscillated between 6.6 and 6.8% in wooden objects, and between 1.8 and 2.3% in paintings. The results also showed a progressive increase in the number of hours and distance with respect to the optimal preservation values in the zones with a greater temperature and relative humidity, with cooling and dehumidification being the most effective strategies. This study represents a progress towards the most appropriate heritage preservation strategies by using an extrapolated methodology considering the climate change effect.

在全球变暖的背景下，遗产保护存在不确定性。如今，预测建筑环境中环境变量的性能以及评估遗产元素的保存条件都面临挑战。基于当前的监视，本研究使用人工智能来预测未来场景中的环境绩效。另外，根据阈值保存值评估遗产元素。选择了塞维利亚大学教堂作为案例研究，同时应用了UNI 10829的阈值和性能指数（PI）来评估室内环境条件。根据温度和湿度，总共确定了9个环境处理区。要估算变量，考虑到21世纪每十年A2未来情景的时间序列，对多层感知器（MLP）进行了监视和应用。分析了当前的MLP表现，从而表明教堂内的PI逐渐降低：在本世纪末，木制物体的PI值在6.6％至6.8％之间波动，而绘画中的PI值在1.8％至2.3％之间波动。结果还表明，相对于温度和相对湿度较高的区域中的最佳保存值而言，小时数和距离的逐渐增加，冷却和除湿是最有效的策略。通过考虑气候变化影响的外推方法，这项研究代表了朝着最合适的遗产保护策略迈进的一步。