Influencing Factor-Decoupled Battery Aging Assessment for Real-World Electric Vehicles Based on Fusion of Fuzzy Logic and Neural Network

The incremental capacity analysis (ICA) method is a typical data-driven method with great potential in battery aging assessment for electric vehicles (EVs). However, the battery health features generated through the ICA method are subject to battery state-of-health (SOH) and environmental factors, which compromises the accuracy of battery aging assessment in real-world situations. This article proposes a novel model structure that combines the fuzzy logic and the radial basis function neural network (RBFNN) to decouple the influencing factors of battery aging using operating data collected from real-world EVs. First, the distortion phenomenon of the battery aging trajectory is discussed, and the relationships between influencing factors and battery health features are carefully analyzed. Second, a Fuzzy-RBFNN model for battery aging assessment is constructed, considering two influencing factors as inputs. Finally, employing an artificially adjusted method, the influence of temperature on battery aging assessment is decoupled using the trained Fuzzy-RBFNN model. The comparison results with the sole RBFNN model demonstrate the effectiveness and necessity of combining with fuzzy logic for battery aging assessment.