The project contains three folders dedicated to each technique. Please refer to the correspoding section for running instructions.

Before running simulink model, please run "parameters.m" file to load required parameters into simulink block.

In the simulink model, the manual switches in the left-hand side are for selection of road disturbances. The manual switch in the right-hand side are for switching between the passive suspension system and active suspension system. Specifically, when the switch is on the constant '0', the system is operating as a passive suspension system.

Files "PID_opt" and "RMS_opt" should be placed in the same directory. The "PID_opt" file is the main code for optimization and "RMS_opt" is a sub-function of "PID_opt". To run the optimization for a specific road profile (specific out of three available), the input should be selected manually and the system should be in the mode active suspension system + (the manual switch in the right-hand side should be switched to the output of the PID controller).

To run the code please execute the following extractions:

1. Open the files "GENA_LQR2_run.m", "GENA_LQR2.m", "Quarter_LQRGENA_wh2"
2. Run "GENA_LQR2_run.m"
3. Select a road disturbance in the simulink model "Quarter_LQRGENA_wh2" and run the simulink model
4. Click "scope" and "scope2" in the right-hand side of the simulink model to check the suspension stroke and verticle acceleration compared with the passive suspension system

Please note:

• "GENA_LQR2.m" creates a function "GENA_LQR2" that will be called in "GENA_LQR2_run.m"
• You can change the fitness function in "GENA_LQR2.m" for your application
• You can modify the parameters of genetic algorithm in "GENA_LQR2_run.m"

Please perform the following actions to run the code:

• Open DDPG.slx file
• Load parameters.m file
• Run DDPG_Train.m file

After this sequence, please run the following command on command window for training the agent:

trainingStats = agent.train(env,trainOpts);

Once the training is completed, please run the following commands on command window:

simOpts = rlSimulationOptions(MaxSteps=ceil(Tf/Ts),StopOnError="on");
experiences = env.sim(agent,simOpts);

Alternatively, you can already load the pre-trained agent:

load("trained_agent","agent")

Finally, please run the DDPG simulink file