PRECISION RANGE IN SALT INTRUSION FORECASTING RESULT OF THE ESTUARY OF BENGAWAN SOLO USING ADAPTIVE NEURO FUZZY INFERENCE SYSTEM MODEL APPROACH
Abstract
Scientists have conducted many researches and developed models of salt intrusion due to tidal influence which collide with discharge of the river upstream. Most of developed models for forecasting the salt intrusion were physical or mathematical. Physical model are inflexible and very expensive when applied to salt intrusion cases, because every developed models only match with in the particular estuary on the other hand, mathematical models have a human limitation range when entering all of the variables to construct a model that represent the comprehensive natural phenomenon. In the last decade, the softcomputing model as branch of the artificial intelligence science were introduction as a forecast tool beside knowledge based system expert system, fuzzy logic, artificial neural network, and genetic algorithm. This reseach choose softcomputing model as an aid program in the system modeling because it has several advantages,which are operate in non linear system that can hardly be modeled mathematically, and have a parameter flexibility as an input of the model. Using all of the spesific advantages that the stated above, this research have a main purpose to develop the estuary salt intrusion forecasting model in dry season due to the influence of daily maximum tidal wave which collide with discharge of the river upstream using softcomputing approach. The method that used in this research was a combination between fuzzy logic and artificial neural network which usually called neuro fuzzy system of adaptive neuro fuzzy inference system algorithm (ANFIS). The main of research proved that the forecast result of the model that use M-SINFES as an aid program software were very sensitive to changes in range of influence parameter and were also have an accurate forecast range for a day ahead ( L ) when used a secondary data of measurement between 12 August – 7 October 1988 in estuary of Bengawan Solo. The configuration of the model can be described in a relationship pattern will be stated : L = ( Input : height of the daily maximum tidal wave (Ht), discharge of river upstream (Qt), and length of salt intrusion (L1+t1+tt) ; adaptive neuro fuzzy inference system : change in the value of range of influence parameter compared with amount of rules of fuzzy inference system, and mean square error of the learned and tested data ). The performances of the ANFIS model both training and testing data were evaluated and the best training/testing data selected according to stastical parameter such as mean square error (MSE) is 1.24 .10-6 (Scheme 7).