Impact of Membership Functions on the Performance of AI‐Assisted Neuro‐Fuzzy System for Analysis of Anion‐Responsive Behaviours of Polypyridyl‐Imidazole Based Ru(II) Receptors

This study explores the anion sensing properties of mono-, bi-, and trinuclear Ru(II) complexes based on polypyridyl-imidazole ligands with varying numbers of dissociable protons in their secondary coordination sphere. The imidazole protons, influenced by the Ru²⁺ centers, facilitate interactions with anions like F^- and CN^- in acetonitrile, and CN^- in water, through hydrogen bonding or via proton abstraction. The spectral properties of the metalloreceptors change in the presence of these anions, and the initial state can be restored with acid, allowing recycling. The complexes exhibit Boolean logic functions (Implication logic gates) using spectral outputs in the presence of specific anions and acids. To address the time and cost constraints of executing comprehensive sensing experiments, an adaptive neuro-fuzzy inference system (ANFIS) with five membership functions (triangular, trapezoidal, generalized bell-shaped, Gaussian, and pi-shaped) was used to predict experimental data and appropriate modeling of the sensing characteristics. Comparison of ANFIS model outputs with experimental data showed a good correlation, with the Gaussian and generalized bell-shaped functions providing high accuracy and close resemblance to experimental data. This neuro-fuzzification approach promises accurate prediction and modeling of sensing data for Ru(II)-based metalloreceptors.