Last update: May 18, 2012

8. Evaluation of the Adaptive Controller

The Simulink control scheme provided by the participants will be firstly evaluated on the simulator using the given sequences of disturbances.

The step changes in the frequencies of the disturbances can be generated with the block "multi sinusoidal disturbances" already included in the Simulink simulator. The description of these sequences can be found in the files simple_step_tests.pdf and step_changes_tests.pdf.

The chirp change in the frequencies of the disturbances can be generated with the Simulink block "test_chirp" given in the file test_chirp.zip. The description of the chirp tests are given in the file chirp_test.pdf

Provided that the basic requirements are satisfied we will proceed to the next step, i.e., compilation using xPC Target. If compilation problems occur, these will be indicated to the contributors for making appropriate modifications.

Once the compilation is OK the scheme will be tested on the real system by the GIPSA-Lab staff.

The tests will be done with the sequences of disturbances described in the files simple_step_tests.pdf, step_changes_tests.pdf, and chirp_test.pdf as well. Once these tests are achieved, test on similar sequences of disturbance but with values of frequencies chosen randomly (within the given range) will be done.

For measurement of performances in the time domain, both transients and steady state will be evaluated. The transients will be compared in terms of maximum value, norm and duration. The steady state performance in time domain will be evaluated by measuring the mean square value of the residual force which will be compared with the value of the residual force in open loop (providing a measure of the global attenuation).

For chirp disturbances, the maximum value of the residual force during the chirp will be measured as well as the mean square value of the residual force (during the chirp).

Complexity of the control scheme will be evaluated in terms of execution time (the organizers are aware that this is not a fully objective criterion - any suggestion is welcome).

For the calculation of the different required values, one can use the following Matlab functions, written especially for the Benchmark: