Summary :: Control Systems (Neural Network Toolbox)

Neural Network Toolbox

Summary

The following table summarizes the controllers discussed in this chapter.

Block
Description

NN Predictive Control
Uses a neural network plant model to predict future plant behavior. An optimization algorithm determines the control input that optimizes plant performance over a finite time horizon. The plant training requires only a batch algorithm for static networks and is reasonably fast. The controller requires an online optimization algorithm, which requires more computation than the other controllers.

NARMA-L2 Control
An approximate plant model is in companion form. The next control input is computed to force the plant output to follow a reference signal. The neural network plant model is trained with static backpropagation and is reasonably fast. The controller is a rearrangement of the plant model, and requires minimal online computation.

Model Reference Control
A neural network plant model is first developed. The plant model is then used to train a neural network controller to force the plant output to follow the output of a reference model. This control architecture requires the use of dynamic backpropagation for training the controller. This generally takes more time than training static networks with the standard backpropagation algorithm. However, this approach applies to a more general class of plant than does the NARMA-L2 control architecture. The controller requires minimal online computation.

Block	Description
NN Predictive Control	Uses a neural network plant model to predict future plant behavior. An optimization algorithm determines the control input that optimizes plant performance over a finite time horizon. The plant training requires only a batch algorithm for static networks and is reasonably fast. The controller requires an online optimization algorithm, which requires more computation than the other controllers.
NARMA-L2 Control	An approximate plant model is in companion form. The next control input is computed to force the plant output to follow a reference signal. The neural network plant model is trained with static backpropagation and is reasonably fast. The controller is a rearrangement of the plant model, and requires minimal online computation.
Model Reference Control	A neural network plant model is first developed. The plant model is then used to train a neural network controller to force the plant output to follow the output of a reference model. This control architecture requires the use of dynamic backpropagation for training the controller. This generally takes more time than training static networks with the standard backpropagation algorithm. However, this approach applies to a more general class of plant than does the NARMA-L2 control architecture. The controller requires minimal online computation.

Importing and Exporting Training Data Radial Basis Networks