Skip to main content
Open Access Publications from the University of California


UCLA Electronic Theses and Dissertations bannerUCLA

Economic MPC of Nonlinear Processes via Recurrent Neural Networks Using Structural Process Knowledge


This work discusses three methods that incorporate a priori process knowledge into recurrent neural network (RNN) modeling of nonlinear processes to get increased prediction accuracy and provide information on how the neural network models are structured. The first method proposes a hybrid model that integrates first-principles models and RNN models together. The second method proposes a partially-connected RNN model which its structure is based on a priori structural process knowledge. The third method proposes a weight-constrained RNN model that integrates weight constraints into the training of the RNN model. The proposed RNN models are used in an economic model predictive control system and then applied to a chemical process example to validate the improved approximation performance compared to a fully-connected RNN model that is treated as a black box model.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View