Optimizing Fuzzy Logic Control-Based Weather Forecasting through Optimal Antecedent Selection Using the Fuzzy Analytical Hierarchy Process Model

The numerical weather forecasts rely largely on the amount of precipitation available, and the use of statistical and empirical methods, but fall short of higher accuracy and relatively short-time required. Recently, the fuzzy AHP (that combined AHP with fuzzy logic) for the purpose of arriving at better outcomes from fuzzy logic control (FLC) rules-list. While evolutionary computing and fuzzy logic techniques are known to guarantee better accuracy and reliability of the outcomes when applied to weather uncertainty problems. Though, the fuzzy logic approach has low accuracy, which needs to be improved with rules-list refinement. This paper pulls on these approaches to develop a weather forecasting model for cities. First of all, the outcomes of the FAHP model revealed that, Wind Direction (WND) and Relative Humidity (HUM) as contributing 30.01% and 19.97% influence to the decisionmaking process against air temperature, windspeed, WND, HUM, and air pressure identified earlier. Secondly, the select FAHP parameters served as antecedents for the FLC model, in which five fuzzy rules were included in rule-base. Upon validation with the standard and local datasets, the proposed model achieved lower error rates of 0.0010, 0.0317 and 0.0319 for MSE, RMSE and MAPE respectively when treated with the Kaggle standard dataset. By comparing the proposed FLC model outcomes to the unoptimized FLC model in term of error rates, MSE of 0.0010, RMSE of 0.0317, and MAPE of 0.0355 were achieved attained by former indicative of its superiority