An optimal prediction model for time series prediction in manufacturing systems

Abstract

Gray model GM(1, 1), a single variable first-order gray differential equation model, which is based on gray system theory, has been proposed as a prediction model to solve efficiently the prediction problems in manufacturing systems. However, the prediction accuracy of this model is unsatisfying when original data set shows great randomness. In this paper, in order to improve the prediction capability of GM(1, 1), the exponential smoothing method is integrated into GM(1, 1) through the preprocessing for original data set. Then the particle swarm optimization algorithm is employed so that the prediction power can be further enhanced. Finally, a residual compensation approach based on artificial neural network is proposed to acquire the best prediction performance.The real case of time series prediction in a product manufacturing process is used to validate the effectiveness of the proposed model.