A novel synergistic fibroblast optimization based Kalman estimation model for forecasting time-series data

摘要

Evolution of a new computational model for estimating the time-varying data is a highly empirical task in scientific computing theory. Using various evolutionary computation techniques, the identified parameters and functions of the estimation model are optimized to improve the performance of the execution process. The objective of this paper is to introduce a novel Synergistic Fibroblast Optimization (SFO) algorithm in Kalman filter, to develop an optimal estimation model, for forecasting the future state variables of time series data. The proposed model is evaluated using the water samples collected from Ukkadam Periyakulam Lake, Coimbatore, India, where water quality forecasting is done. Fisher score method is applied to choose optimal features subset from the specified high dimensional dataset. Standard performance metrics such as root mean square error (RMSE), mean absolute error (MAE) and regression equation of Sum of Squared Error (SSE) are measured to evaluate the performance of the estimation model, and it is also compared with the actual measurements. Experimental results illustrate that SFO based estimation model produces better promising results than conventional estimating methods.