New energy battery cabinet life detection

How is battery life prediction based on feature engineering?

Based on feature engineering, battery degradation stage detection and physical similarity analysis are used as the first two steps of life prediction. Battery data from the same aging stage and similar physics to the test data are clustered into subgroups, and the prediction model is subsequently established. Fig. 1.

How do major manufacturers predict battery remaining life?

For SOC, currently, the major manufacturers mainly apply the current integration method. For SOH, currently, the major manufacturers mainly apply the voltage curve fitting method. For battery remaining life prediction, currently, the major manufacturers mainly apply statistical patterns based on long-term data to estimate.

How to predict EV battery life?

As an extremely important part of the current and future testing of EV batteries, there are two general methods of life prediction: (1) Empirically based prediction: empirically based RUL (remaining useful life) prediction method, mainly including cycle number method and event-oriented aging accumulation method.

Can deep learning improve battery life prediction?

In order to solve the problems of poor interpretability and huge computation resource consumption of deep learning-based life prediction models in the field of battery health management, this paper proposes a novel optimization method for remaining battery life prediction.

Can a life prediction model accurately predict battery life?

In summary, the MAE of all batteries is between 3 and 6 cycles, and the errors are within a reasonable range, which proves that the model established by fusing the CNN and LSTM in this paper can accurately predict the remaining life of batteries. 4.2. Life prediction model interpretation and analysis

Can a degradation stage detection method be used to classify retired batteries?

First, for the first time, a degradation stage detection method that does not involve accessing historical data is proposed; this method can quickly classify retired batteries, particularly by detecting whether the current cycle is in a rapid degradation stage.