A neural-network-based model predictive control scheme for grain dryers

Abstract

Aiming to improve the precision of grain moisture content discharged out of grain dryers and the degree of automation and intelligence of grain drying process, this article proposed a model predictive control scheme based on neural network for grain dryers. In the scheme, a mathematical model on basis of grain drying theory is built for an actual grain dryer system. With this model, sufficient input and output time series of grain dryer under different conditions can be obtained expediently via simulation. As a training set, the data series is utilized to train a nonlinear autoregressive neural network which will be used as a predictive model instead of the mathematical model. Finally, based on the neural network model, MPC is designed to realize an accurate control of grain moisture content, in which, the particle swarm optimization algorithm is employed to optimize the cost function. Furthermore, MATLAB simulations are carried out to fully validate the effectiveness of the proposed control strategy, with which it is found that the use of the neural-network-based model predictive scheme is able to enhance the control precision of grain dryers.