Comparative study of artificial intelligence-based building thermal control methods – Application of fuzzy, adaptive neuro-fuzzy inference system, and artificial neural network

This study’s aim is to develop diverse Artificial Intelligence-based (AI-based) thermal control logics and to compare their performances for identifying potentials as an advanced thermal control method in buildings. Towards that aim, three AI-based control logics have been developed: i) Fuzzy-based control; ii) ANFIS-based (Adaptive Neuro-Fuzzy Inference System-based) control; and iii) ANN-based (Artificial Neural Network-based) control. The last-mentioned two were adaptive methods employing iterative self-tuning process during system operation. Each method’s performance was tested in a typical two-story residential building in USA, via computer simulation incorporating IBPT (International Building Physics Toolbox) and MATLAB. In analysis of test results for indoor air temperature, thermal comfort profiles, and amount of heat supply and removal, two adaptive control methods – ANFIS-based and ANN-based – significantly stabilized thermal conditions by the increased comfort period and the decreased deviations from the set-point compared to the Fuzzy-based non-adaptive method. No control method showed significant energy saving effects over the other. In conclusion, adaptive AI-based control methods have potential to maintain interior air temperature more comfortably.

► We develop AI-based thermal control logics and test their performances.
► FL, ANFIS, ANN are applied as AI-based methods.
► Tests are conducted in a residential buildings using computer simulation.
► Two adaptive ANFIS- and ANN-based logics stabilize thermal conditions.
► No control method shows significant energy saving effects over the other.