Decoupling Control for DACF Pressure Based on Neural Networks and Prediction Principle.

Dengfeng Dong,Xiaofeng Meng,Fan Liang

DOI: https://doi.org/10.1007/978-3-642-23235-0_76

2011-01-01

Abstract:Overcoming the coupling impact is the premise to achieve rapid, precise and especially independent control for the two concatenate pressures of the double-level air current field (DACF) system. Due to the nonlinearity, time lag, and strong coupling characteristics of the system, a decoupling method based on neural networks and prediction principle is presented in this paper. With the neural networks, a nonlinear mathematical model of the relationship describing air flow rate and other variations including the upstream pressure, the downstream pressure and valve opening is developed. With the prediction principle, the predicted pressure state formula is derived. On the basis of them, the predictive expressions of disturbances between the upstream and downstream pressure are obtained by the ideal gas equation. Thereby the controller outputs are regulated on line properly in advance, and the coupling disturbances and time lag effect are weakened notably. Experimental results show the method is effective to achieve the system decoupling.