Cell Capacity Prediction

Acquire data that may affect cell capacity in each production step (cell winding, electrolyte filling, formation and aging data, etc.) to construct a cell capacity calculation model, so as to predict cell capacity.
During production, the capacity of most cells are predicted through the model, and a small portion of cells still go through full-capacity grading to provide model training data and verify model effect.

Key Features

The temperature compensation method converts the temperature of all batteries during discharge to a uniform temperature. The data labels before and after temperature compensation have a significant deviation, resulting in a relative narrowing of the overall battery capacity and improving the accuracy of the results

The AI anomaly detection model is used for anomaly handling, which automatically identifies missing abnormal data and improves the accuracy of the results

Data extraction: Extract massive process data from key processes such as cell pairing, primary injection, chemical transformation, and secondary injection

Feature data extraction: obtained from time-series data extraction (statistical indicators), with a total of 70+dimensions, each row representing the entire process data of a single battery

The plan adopts a combination of feature extraction, data preprocessing, and predictive modeling

Benefits of the Product

Customized development + implementation + algorithm services based on standard functions Functions

1. Additional customer demand function