Forecasting by designing Mamdani general type-2 fuzzy logic systems optimized with quantum particle swarm optimization algorithms

Much more attention has been focused on studying and applying general type-2 fuzzy logic systems (GT2 FLSs) in recent years. The paper designs a type of Mamdani GT2 FLS for studying forecasting problems based on the data of permanent magnetic drive (PMD) loss. During the system design process, we choose the primary membership functions (MFs) of antecedent, consequent and input measurement general type-2 fuzzy sets (GT2 FSs) as Gaussian type MFs with uncertain standard deviations. The corresponding vertical slices (secondary MFs) are chosen as the triangle MFs. All the parameters of Mamdani GT2 FLSs are optimized by the quantum particle swarm optimization (QPSO) algorithms. Noisy data of PMD loss are adopted for both training and testing the proposed FLSs forecasting approaches. Simulation studies and convergence analysis are employed to show the effectiveness and feasibility of the proposed GT2 FLSs forecasting methods compared with their T1 and IT2 counterparts.