FLUIDMechatronix™: Connectivity, Identification, and Advanced Control via Python (Updated: April 30, 2023)

Python control and connectivity algorithms, simulations, and system identifiers for the FLUIDMechatronix™ process automation and pumps teaching system, developed by Khush Harman Singh Sekhon, Purushottama Dasari, Om Prakash, and and Oguzhan Dogru. Contact dogru@ualberta.ca for the details/datasets.

Codes:

OPC_connection:

• OPCWrapper.py : Wrapper class for the basic opc methods read/write/close for OPC_direct_readwrite.py to work with OPC-UA.
• OPC_direct_readwrite.py : Helper functions used to perform read/write operations with OpenOPC, store and plot data for Fluid Mechatronix experiment.
• OPC_expert_readwrite.py : Helper functions used to perform read/write with OPC Expert and requests module, store and plot data for Fluid Mechatronix experiment.
• opcda-to-opcua.py : Modified proxy to convert opc-da tag data (in 32-bit python env) to opc-ua tag data (in 64-bit python env). The original proxy was created by @Eskild Schroll-Fleischer
• ua_client.py : A working example for an opc-ua client in python-64-bit.

State Space Modelling:

• State_space_modelling_ex.ipynb : Working example of state space model implemented in python
• gif_plot.ipynb : A plotting function to create a gif plot from state space data saved as a csv.
• animated_plot.py : A plotting function to create a gif plot with matlab animation module. (faster than gif_plot.py)
• pid_control.py : PID controller (position and velocity) classes for controlling a simulated state space environment.
• plotter.py : Online plotting function to plot data in real-time. Works with both simulated and experimental environments.
• online_plot.py : stand-alone online plotter function (older version of plotter.py), redundant since plotter.py was made.

System Identification:

• sysID_modellign.m : A compilation of multiple system models for Fluid Mechatronix in matlab.
  1. FOPTD and SOPTD
  2. SOPTD with custom parameters
  3. State Space
  4. Linear ARX
  5. Non-Linear ARX
  6. First Principles (static)
• pysindy.ipynb : pySINDy (Sparse Indetification of Non-Linear Dynamics) based SISO and SIMO models for Fluid Mechatronix in python.
• rgs_signal.m : Function to generate RGS signal in matlab.

Automation [In Progress]:

• constants.py : Stores all global constants used by fsm_states.py
• fsm_states.py : Finite state machine based model, will be used in conjunction with RL based agents to automate Fluid Mechatronix lab machine.
• plotter.py : Almost identical to the pre-existing plotter.py in State_space_modelling/ directory with minor changes.

RL Agents [In Progress]:

• Qlearning.ipynb : Q learning based agent for simulated environment (based on Fluid Mechatronix lab machine).
• A2C.py : Advantage Actor Critic agent trained on simulated environment.
• A3C.py : Asynchronous Advantage Actor Critic agent trained on simulated environment for 3000 episodes.
• A2C/A3C Agents : Actor and Critic agent weights (as .h5 files)

Data:

Plots and Data:

State Space Modelling:

• eg.gif : Gif plot generated from either gif_plot.ipynb or animated_plot.py.

img_refs:

Contains images and figures used in the github repo.

Miscellaneous

• state_machine_diagram.png : Details the finite state machine structure used in fsm_states.py. (subject to change)
• state_transition_table_automation.xlsx : State transitions in detail. (subject to change)
• FM_tags.xlsx: All tag data for Fluid Mechatronix obtained using OpenOPC.