Learning to Predict and Control with Sparse Model Discovery and Deep Temporal Difference Reinforcement Learning
MAPTD agent controlling deflection of a cantilever beam MAPTD agent controlling wind-induced vibration
Cantilever beam |
76DOF skyscraper |
📂 src
|_📁 EQD # Equation discovery files
|_📂 MAPTD # MAPTD algorithm with numerical predictor
|_📁 algorithm # contains main files for reinforcement learning
|_📁 cfgs # contains yaml files to configure the test examples
|_📁 data # contains model information of 76DOF structure
|_📁 logs # contains trained agents and logs
|_📁 results # directory to save trajectories
|_📄 cfg.py # file to purge yaml files
|_📄 dynamics_gym.py # script for accumulating all the environments
|_📄 env_beam.py # cantilever beam environment setup
|_📄 envs.py # environment wrapper
|_📄 env_tallstorey.py # 76DOF environment setup
|_📄 logger.py # file to log training information
|_📄 train_maptd_76dof.py # training script
|_📄 train_maptd_beam.py # training script
|_📄 test_76dof.py # testing script
|_📄 test_beam.py # testing script
|_📁 MAPTD_hybrid # MAPTD algorithm with hybrid Real2Sim strategy
|_📁 MAPTD_NN # MAPTD algorithm with ANN as world model
|_📁 MAPTD_oml # MAPTD using hybrid Real2Sim strategy with online NO update
|_📁 MPC # Model Predictive Control algorithm true physics
|_📁 NO # Neural Operator surrogate
|_📁 TDMPC # TDMPC algorithm
|_📂 data
|_📄 B76_inp.mat # 76DOF benchmark model
|_📂 images # Result directory
|_📄 beam_solver.py # script of finite element method
|_📄 piezoelectric.py # script for estimating the voltage supply in the piezoelectric patch
|_📄 Systems_76dof.py # script for 76 DOF structure
|_📄 Systems_76dof_rom.py # script for ROM of 76 DOF structure
|_📄 Systems_cantilever.py # script for cantilever beam
|_📄 utils_data.py # script for generating data
|_📄 wind_pressure.py # script for generating wind pressure
|_📄 TT_rlc.yml # Anaconda environment configuration details
- Install the library dependencies from the
TT_rlc.ymlfile




