Optimization algorithms as training approaches for prediction of reference evapotranspiration using adaptive neuro fuzzy inference system

Reference evapotranspiration (ET 0 ), widely used in efficient and meaningful scheduling of irrigation events, is an essential component agricultural water management strategy for proper utilization limited resources. Accurate early prediction ET can provide the basis designing effective help resourceful agriculture. This study aims to evaluate compare performances different hybridized Adaptive Neuro Fuzzy Inference System (ANFIS) models with optimization algorithms predicting daily . The FAO-56 Penman-Monteith method was estimate values using historical weather data obtained from a station Bangladesh. climatic variables estimated form input-output training patterns ANFIS models. these were compared classical model tuned combined Gradient Descent Least Squares Estimate (GD-LSE) algorithm. Performance ranking performed Shannon’s Entropy (SE), Variation Coefficient (VC), Grey Relational Analysis (GRA) based decision theories supported by eight statistical indices. Results indicate that both SE VC provided similar though numeric weights differed. On other hand, GRA slightly sequence ranking. Both identified Firefly Algorithm-ANFIS (FA-ANFIS) as best performing followed Particle Swarm Optimization-ANFIS. In contrast, FA-ANFIS found be second-best according negligible difference weight between (GD-LSE-ANFIS). Therefore, considered model, which utilized predict areas conditions. findings this research great importance planning scheduling. • comparison algorithm demonstrated. Fifteen are parameter tuning. Benefit cost indices incorporated within Entropy, Coefficient,