Recently I am trying to use machine learning in the field of battery. I have gone through several research articles related to this. Most of the papers predicted voltage, State of Charge (SOC), State of Health (SOH) and Remaining useful life (RUL). But I have decided to predict capacity of the battery. Is there any drawback for the prediction of this particular parameter ? If anyone can suggest anything regarding this, I will be grateful. Thanks in advance.
One of the best advantages of machine learning techniques are that they can be used for any sort of data sets. For your case you can conveniently use any machine learning tool for battery capacity prediction. If you have any further query you can contact me for the same.
Hope this helps.
Yes Sir. Thank you very much for your response. I agree with you. But when I have gone through several literatures, I couldn't find any article predicting capacity directly. But they have predicted similar type of parameters like SOC, SOH and RUL which are actually related to capacity. So, for that reason, I was thinking whether any inherent problem it could face for the battery capacity prediction. And along with this, could you please help me regarding error range (it can be in rmse, mse or mae) for capacity prediction. Because I am getting very high error (rmse in the range of 20-30) for capacity prediction and I have found some similar range of error for SOC or RUL prediction in some literatures. Thanks in advance and thanks again for your response.
The error range can be analysed using any of the three parameters you have mentioned. You can decide the parameters based on literature. 20-30% error for RMSE is quite high which signifies that the model is not capable for true prediction. You should check your data properly and verify if there is any unacceptable discrepancy in them.
- Badges
- Science topic