An evolutionary Morphological-Rank-Linear approach for time series prediction

In this paper, a hybrid evolutionary Morphological-Rank-Linear (MRL) approach is proposed for time series forecasting. The proposed method consists of an Intelligent Hybrid Evolutionary MRL (IHEMRL) model composed of an MRL filter and a modified Genetic Algorithm (GA) that employs optimal genetic operators that accelerate its search convergence. The modified GA searches for the particular time lags capable of a fine tuned characterization of the time series and estimates the initial (sub-optimal) parameters of the MRL filter (mixing parameter (lambda), rank (r), linear Finite Impulse Response (FIR) filter (6) and the Morphological-Rank (MR) filter (a) coefficients). Thus, each individual of the GA population is trained by the averaged Least Mean Squares (LMS) algorithm to further improve the MRL filter parameters supplied by the GA. Experiments are conducted with the proposed approach using three real world time series according to a group of relevant performance metrics and the results are compared both to ARIMA models and MultiLayer Perceptrons (MLP).