Fuzzy Modeling of Water Flow on Lower Indus River Basin Using a New Defuzzification Method Based on Soft Computing

There are many physical systems where linguistic descriptions (Fuzzy Sets) allow better assessments because analyses of real systems are familiar with fuzziness for instance we cannot measure the stream flow accurately because of the fluctuations. A feature of fuzzy system is that it gives an efficient base for measure uncertainty. The analysis of precipitation in Pakistan plays a major part in water flow of Indus basin. Therefore, it is of the fundamental nature to store up its water and utilize it economically. The scientific approach to get this preservation is to build up the stream flow forecasting model. This study here proposes a data-driven algorithm of fuzzy model build on a new parameterized defuzzification method, called Mid Variation Method (MVM), with fuzzy time series that integrates a soft computing technique, termed as Particle Swarm Optimization (PSO). The performance of the algorithm of both training and testing phases are evaluated with adequate estimates of accuracy and less than one errors measurement which exhibit the suitability of the fuzzy model. This paper also presents the adequacy of the proposed model for extreme events and the numerical results exhibit this proficiency over the evaluations of Low Flow Error Criteria (LFEC) and Peak Flow Error Criteria (PFEC) for both seasons of Low and High water flows on average of 0.23% error which do better for fuzzy model adequacy; similarly, there is no Low Flows observations and sufficient accuracy for High Flows in testing phase. The other aspect of this paper is that it provides a major preference if only the observations of water flow s are obtainable, which can be useful for water reserve planners as well as Sukkur Barrage monitoring operators.