Closed Roy027 closed 2 weeks ago
Hi @Roy027 Yes, the current soh task code is still empty, we will update it as soon as possible!
Hi @Roy027 Yes, the current soh task code is still empty, we will update it as soon as possible!
Hi @agiamason thanks so much for the reply. May I also ask if currently all the model in BatteryML are trained with RUL annotator as target value? I am not sure if I understand it correct but seems the application of this BatteryML is rather limited as RUL annotator itself is a just Boolean value which is strongly linked to the readily available capacity value. To just get this result do not add much significance to the power for this BatteryML framework.
Would it be possible (in the future) to output multiple performance metric ( V, Current, Capacity ) by BatteryML ( say forecast the performance profile of battery after a few cycles? Or just to prediction the degradation trend after say 500 cycles? Please let me know if those application is available (or in the plan).
Glad to run the baseline.ipynb which gives a good introduction about BatteryML but hope to see more application ready example tutorials where we can use the model for really life application predictions. Thank you!
Hi @Roy027, thanks for your questions and advice.
Currently, the BatteryML framework can only predict a single value, such as RUL or the SOH of a specific cycle.
Your suggestion is excellent, and we highly appreciate industry-driven requirements. We genuinely hope that BatteryML can assist in real-world battery development, not just in academic research.
If possible, could you provide more details about your specific use case? This would help us better understand and develop the necessary features.
Based on your request, here is my understanding of the desired functionality. Please correct me if I am mistaken:
Feature 1: The input consists of electrical signals from the first N cycles (e.g., Voltage, Current, Capacity), and the output is the electrical signals for the M-th cycle (e.g., Voltage, Current, Capacity), where M > N. Feature 2: The input consists of electrical signals from the first N cycles (e.g., Voltage, Current, Capacity), and the output is the Capacity for cycles M to M+k, where M > N.
hi @agiamason Thanks so much again for the answers. You have understand the requirement very well. Basically it is the ability to forecast the degradation trend that is pretty hard to be directly calculated.
is soh.py supposed to be empty?