Skip to content

Latest commit

 

History

History
54 lines (43 loc) · 1.79 KB

README.md

File metadata and controls

54 lines (43 loc) · 1.79 KB
Raw

Chain Mass System

The classic chain mass benchmark system for MPC.

MPC ingredients

To compute the MPC ingredients, run the offlinempcingredients.m file in MATLAB. The output of the MATLAB file are already available in the folder mpc_parameters.

matlab -nodisplay -r "run('offlinempcingredients.m')"

The mpc parameters should be saved in human readable .txt form in the folder mpc_parameters.

MPC dataset generation

To generate samples of the MPC, call

python3 samplempc.py sample_mpc \\
    --n_mass=3 \\
    --numberofsamples=10

The results of this would be saved in a folder called datasets/chain_mass_{n_mass}_N_{numberofsamples}_{date}-{time}.

You can similarly create a larger dataset, by calling this function in parallel

python3 samplempc.py parallel_sample_mpc \\
    --instances=24 \\
    --samplesperinstance=10 \\
    --prefix=Cluster_test

The results of this would be saved in a folder called datasets/chain_mass_{n_mass}_N_{instances*samplesperinstance}_merged_{prefix}_{date}-{time}

If you downloaded the precomputed dataset for this example, you should find it under datasets/chain_mass_3_N_19080000.

Training a NN

If you want to train an approximator on the precomputed dataset for this example, call

python3 approximatempc.py find_approximate_mpc
    --dataset=chain_mass_3_N_19080000

the models will be saved in a models folder.

If you downloaded the pretrained NN, you should find it under models/9-200-400-600-600-400-200-30_mu=0.09_20230131-021341

Testing the NN

You can run closed loop test with the model calling

python3 safeonlineevaluation.py closed_loop_test_on_dataset \\
    --dataset=chain_mass_3_N_120000_test \\
    --model_name=9-200-400-600-600-400-200-30_mu=0.09_20230131-021341 \\
    --N_samples=10000 \\
    --N_sim=200