GROUNDWATER LEVEL SIMULATION USING ANN FOR GANDHINAGAR DISTRICT
SHAH SHUBHAM NILESHBHAI, Prof. Megha B. Bhatt, Prof. Kinnari R. Mishra
Rainfall, Maximum and Minimum temperature, Humidity, Evaporation, MATLAB and M.S. EXCEL.
In this research, Artificial Neural Network is used to predict groundwater level in Gandhinagar district. Model considers precipitation, maximum and minimum temperature, humidity, and Evaporation data as input parameter and groundwater levels as an output parameter. The past 09 year data was utilized for modeling. The dataset was divided into different training, testing and validation ratio to find the best model. The efficiencies of the Levenberg-Marquardt (LM), the Bayesian regularization (BR) and the scaled conjugate gradient (SCG) algorithms are compared. The model efficiency and accuracy were measured based on Mean square error (MSE) and correlation coefficient (R). Multiple Linear Regression (MLR) technique is also used to prepare a model. The performance of MLR model is determined using Correlation coefficient (R) and coefficient of determination (R2). The results revealed that the best model is composed of the feedforward networks, trained by the Levenberg-Marquardt algorithm.