2023-04-11 Sajjad Meeting Notes

@Taim Al-Dabbagh @Mostafa Hussein

Sajjad’s Slides (Matlab Workshop):

• Simulink model: watertank_comp_design (from matlab documents)

  • Proposed problem: Tank system where the water level as to be changed, and it can be changed using a water pump controlled with a voltage V.

  • Using Matlab blocks, the differential equation representing the problem can be modeled / built out.

• Drone Example:

  • You should have a plan and a good model

    • Open

      

    • Equations for flight dynamics should be organized. These need to be researched and can be found from reading research papers.

    • Also good model for the motors and sensors.

    • This model would be like a control loop.

  • Controller submodule:

    • Open

      

    • There also would have to be an extra set of 2 PID for roll and pitch since they affect position as well:

    • MMA (motor mixing algorithm)

  • Estimation Modules that convert sensor inputs and converted them into measurable states:

    • Should have good equations for the sensors. Noise should be simulated in a good manner.

      • Should be using Kalman filters to better estimate the drone state.

  • We already have PID inside ZP3, do we need to still create models for the Roll, pitch yaw thrust?

    • It is good that we have premade controllers, but we should implement them in the plant

  • How to know the equations found online are the correct ones:

    • When finding a research paper, scan what type of vehicle they are controlling. How many motors, the orientation and placement of the motors. Based on that, use equations that apply to our vehicle.

  • Inner loop vs Outer loop controller:

    • Inner loop is faster.

      • Inner loop only is used when affecting roll, pitch and yaw doesn’t affect other modes of movement.

    • However, outer loop would be needed as it is responsible for controlling position which would be affected by roll and pitch for example.

      • Reaction time of outer loop is slower than the inner loop

        • Not constantly checking position, and it takes longer to calculate positions, which is why outer loop tends to be slower.

  • Providing Simulation with outer data:

    • The data has to be in a .mat file and then it has to be loaded to the workplace

    • Use module in simulink called “workspace”?

      • Helps put data from outside of simulation into simulink.

Post Questions:

• Examples of papers to think about for getting equations:

  • Course on Udemy that starts from developing the equations to going into simulation. Uses Python though, but it could be a good starting point. Sajjad can share the link to it.

  • Could also use books, but that can be time wasting. Sajjad can however can share a few books for aircraft flight dynamics.

• When do you know to make assumptions, and which assumptions do you usually take:

  • Follow research papers. Use the assumptions they use with their models, especially as a beginner.

• Workflow:

  • Rough, not perfect, but complete loop at the start is good to have.

    • Find some dynamics equations for the vehicles and use them (although not perfect)

  • Then place sensors and have equations for them.

    • After simulate noises even if not perfect.

  • Finally go through the controller

    • Still not perfect just a rough simulations

  • After a rough simulation is completed, go through it again to perfect it.

    • Constantly researching to improve model.

    • Design is a loop, and going through the loop multiple times is a must.

  • Cant work on a single section, make it perfect and move on.

    • After you move on you will probably find something missing in the pervious section.

  • Take your time, the process takes time so do not rush.

• Is it possible to simulate real time systems in simulink

  • Yes, that is one of the most important things to have a simulation. If things can’t be simulated in real time, then the simulation becomes somewhat obsolete.