Control method using neural networks and a voltage/reactive-power controller for a power system using the control method
First Claim
1. A control method for controlling a control object by means of a neuro controller made up of a recurrent neural network, said control method comprising the steps of:
- connecting said neuro controller in parallel to a control unit having functions for controlling said control object and connecting in parallel to said control object a neuro control-object simulator made up of a recurrent neural network for simulating said control object;
operating said neural network of said neuro controller and a neural network making up said neuro control-object simulator to perform a pre-learning phase so as to make input-output relations of said neuro controller match input-output relations of said control unit, and so as to make in said pre-learning phase input-output relations of said neuro control-object simulator match input-output relations of said control object;
connecting the output and the input of said neuro controller to the input and the output, respectively of said neuro control-object simulator;
operating said neural network of said neuro controller to perform a re-learning phase so as to make the output of said neuro control-object simulator match the input of said neuro controller; and
thereafter connecting said neuro controller to control said control object after completing said re-learning phase and thereafter controlling said control object using only said neuro controller.
0 Assignments
0 Petitions
Accused Products
Abstract
A neural network apparatus and method for use in applications such as in a voltage/reactive-power controller in which a neuro control-object simulator and a neuro controller pre-learn so as to make input-output relations of the controller match the input-output relations of a control unit and so as to make input-output relations of the simulator match input-output relations of a control object. The controller re-learns so as to make the output of the simulator match an input corresponding to a desired output of the control object. After re-learning, the controller controls the control-object.
-
Citations
6 Claims
-
1. A control method for controlling a control object by means of a neuro controller made up of a recurrent neural network, said control method comprising the steps of:
-
connecting said neuro controller in parallel to a control unit having functions for controlling said control object and connecting in parallel to said control object a neuro control-object simulator made up of a recurrent neural network for simulating said control object; operating said neural network of said neuro controller and a neural network making up said neuro control-object simulator to perform a pre-learning phase so as to make input-output relations of said neuro controller match input-output relations of said control unit, and so as to make in said pre-learning phase input-output relations of said neuro control-object simulator match input-output relations of said control object; connecting the output and the input of said neuro controller to the input and the output, respectively of said neuro control-object simulator; operating said neural network of said neuro controller to perform a re-learning phase so as to make the output of said neuro control-object simulator match the input of said neuro controller; and thereafter connecting said neuro controller to control said control object after completing said re-learning phase and thereafter controlling said control object using only said neuro controller. - View Dependent Claims (2)
-
-
3. A voltage/reactive power controller for controlling voltages and reactive power of a power system comprising:
-
a voltage/reactive-power control unit having functions for controlling a power system; a neuro control-object simulator made up of a first recurrent neural network, wherein said first recurrent neural network has undergone a pre-learning process so as to make input-output relations of said neuro control-object simulator match input-output relations of said power system; and a neuro controller made up of a second recurrent neural network, wherein said second recurrent neural network has undergone a pre-learning process so as to make input-output relations of said second recurrent neural network match input-output relations of said voltage/reactive-power control unit and has further undergone a re-learning process so as to make the output of said neuro control-object simulator match the input of said neuro controller, said power system being connected to be controlled only by said neuro controller after said pre-learning and re-learning processes have been completed. - View Dependent Claims (4, 5, 6)
-
Specification