imc

基于神经网络的控制工具箱

                   -----------------------------------------------
                      
                   OVERVIEW OF FILES ASSOCIATED WITH SIMULATION OF
                             INTERNAL MODEL CONTROL
                   
                   -----------------------------------------------
                    
MATLAB tools for simulating neural network based internal model control (IMC).


                                 --OO--
                                  

DESCRIPTION                             
Assume that a process has been modelled by a neural network:
     ^      ^
     y(t) = f(y(t-1),...,y(t-n),u(t-1),...,u(t-m))
     
and that the inverse exists and has been modelled by another network:

     ^      ^ -1
     u(t) = f   (y(t+1),y(t),...,y(t-n+1),u(t-1),...,u(t-m+1))


If network weights and architecture definitions for these two networks have
been saved in mat-files, 'imccon' can be used for simulation of
internal model control of the process under consideration. The
"process" can be a SIMULINK model or a model described by a set of differential
equations specified in a MATLAB-function. If the training data has been
obtained from a "real world" experiment (i.e., a MATLAB/SIMULINK model is not
available) model is not available the closed-loop system can also be
simulated by using the network model for simulating the process.

The model of the process can be generated with the 'nnarx' function from the
NNSYSID toolbox. In order to train an inverse model, one of the functions
'general', 'special1', or 'special2' can be applied.


                                --OO--

                                 
REFERENCES
o K.J. Hunt, D. Sbarbaro: "Neural Networks for Nonlinear Internal Model Control"
  IEE Proceedings-D, Vol. 138, No. 5, pp. 431-438, 1991.


                                --OO--
                                 

FILE OVERVIEW:
Imc      - This file.
imccon   - Program for simulating internal model control.
imcinit  - File containing design parameters for 'imccon'.


The following functions must be available as well:
NNSYSID toolbox
shift, siggener                    - Functions used by imccon


                                --OO--
                                 
                                 
SUGGESTED WORKING PROCEDURE FOR TESTING THE CONCEPT ON A KNOWN PROCESS

1: BUILD THE PROCESS
Build the process in SIMULINK or write the differential equations in a
MATLAB-function. If SIMULINK is used, specify input to and output from the
model by an inport and an outport, respectively. If SIMULINK is not available, 
write the differential equations in a MATLAB-function in the format required
by the ODE45 differential equation solver.

2: MAKE AN EXPERIMENT
Use 'experim' for generating a data set to be used for identifying neural
network models.

3. IDENTIFY A "FORWARD" MODEL OF THE PROCESS
Train a model of the process to be controlled using the 'nnarx' function found
in the NNSYSID toolbox. 

4: IDENTIFY AN INVERSE MODEL
The inverse model can be identified with generalized or specialized training
by following the guidelines given in the text file 'Inverse'. 

5: SIMULATION OF CLOSED-LOOP SYSTEM
The closed-loop system is simulated with the program 'imccon'. The design
parameters necessary for running this program must be defined in the file
'imcinit'. For example, the user must specify the name of the MATLAB/SIMULINK
model, sampling period, reference signal, and the names of the files
containing architecture and weights of forward and inverse model networks.