Multi-objective fuzzy fractional-order BELBIC for semi-active seismic control in non-linear building structures

This study proposes a novel model-free Multi-Objective Fuzzy Fractional-Order BELBIC (MOFFO-BELBIC) algorithm to address the key gaps in semi-active seismic control of civil structures, unlike conventional single-input single-output (SISO) Brain emotional learning (BEL) controllers, which lack a holistic understanding of multi-loop structural control system dynamics. To bridge this gap, we introduce: (1) A novel multi-input multi-output (MIMO) BEL controller cascaded with a fractional order Proportional-Integral-Derivative controller (FOPID). (2): Offline hybrid multi-objective non-sorting genetic algorithm-III (NSGA-III) optimized Takagi-Sugeno-Kang (TSK) Fuzzy Inference System (FIS) tuning, (3): Online tuning through Fuzzy-to-Lookup table transformations, the trained FIS models are obtained adaptively to include the online training input-output pairs with optimal structural control performance, with 30 % cut down in computational cost. Therefore, MOFFO-BELBIC proposed an autonomous uncertainty handling solution specifically for nonlinear building structures without requiring retraining. To examine the efficacy of the proposed MOFFO-BELBIC algorithm, an illustrative example of a non-linear three-story benchmark building equipped with magnetorheological dampers is presented. The proposed MOFFO-BELBIC has shown a notable performance, 45–60 % superior response reduction compared with the modal-based and model-free control strategies presented in the past few decades. • MOFFO-BELBIC: model-free adaptive control for MIMO feedback and BEL networks in seismic control of nonlinear buildings. • Integrates FOPID and BELBIC with multi-objective optimization for robust real-time control under dynamic uncertainties. • Hybrid offline-online tuning with fuzzy lookup tables cuts computational cost while ensuring optimal structural control. • The system adapts to real-time seismic events without retraining, ensuring robustness against dynamic uncertainties. • Applications show superior seismic vibration control over classical model-based and model-free methods.