Transparent Processing of Neural Networks in Industrial Control
Download as PDF
DOI: 10.23977/cii2019.50
Author(s)
Jiaren Xu, Chun Dong
Corresponding Author
Chun Dong
ABSTRACT
This paper introduces a method that can approximate the neural network as a variable-gain linear feedback controller, which enables the neural network to achieve parameter transparency and parameter interpretability in the application phase. The approximation method is independent of the structure of the neural network, and the learning process is completely consistent with the deep reinforcement learning. Only the learned neural network is approximated in multiple intervals, and the number and range of approximate intervals can be arbitrarily selected.
KEYWORDS
DRL, Linear interpolation, Control