Clear Filters
Clear Filters

Hey, how can i import and replace PCM model by trained neural state-space model in EV system level model

2 views (last 30 days)
Hi i am trying to replace PCM controller by AI - Neural state space model i trained the model with data taken by EV model (PCM Control input and output) and it working fine, now i want to import tarined NSSM into system level EV model i need to replace PCM by NSS model i replaced successfully but not getting expected output....

Accepted Answer

Sam Chak
Sam Chak on 15 May 2024
If you're referring to the performance of the PCM controller in the EV system, it depends on what you mean by 'working fine.' If the Neural State-Space Model (NSSM) is effectively trained to learn the dynamics of the PCM controller and accurately predict its output, then feeding the exact same PCM control input to the NSSM should result in an EV output that doesn't deviate significantly from the expected output as it would with the PCM controller.
In other words, if the NSSM is able to accurately replicate the behavior of the PCM controller, the EV output should closely match the expected output when using the NSSM.
However, if you're not achieving the expected EV output, there are a couple of possibilities to consider. One possibility is that you may have fed in control input that is significantly different from the training data used to train the NSSM. In such cases, the NSSM may struggle to accurately predict the output.
Another possibility is that the NSSM may fail to learn the dynamics of the PCM controller within certain ranges of the data. This could lead to discrepancies between the EV output predicted by the NSSM and the actual output produced by the PCM controller.
To address these issues, it's important to carefully analyze the training data and ensure that it adequately represents the range of control inputs expected in real-world scenarios. Additionally, it may be necessary to refine the training process or explore alternative modeling approaches to better capture the dynamics of the PCM controller.
  3 Comments
Sam Chak
Sam Chak on 15 May 2024
In that case, it is likely that the NSSM is struggling to learn the dynamics within the critical 5% range of the PCM controller. This range is crucial as it has the potential to significantly impact the system output and cause deviations from the expected output.
To address this issue, I recommend focusing on refining the training process for the NSSM. This could involve adjusting the training parameters, increasing the amount of training data within the critical range, or exploring alternative training techniques to improve the model's ability to capture the desired dynamics.

Sign in to comment.

More Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!