Battery Life Prediction🔋

The Hawaii Natural Energy Institute conducted an analysis on 14 NMC-LCO 18650 batteries, each with a nominal capacity of 2.8 Ah. These batteries underwent over 1000 charge-discharge cycles at a temperature of 25°C, using a constant current-constant voltage (CC-CV) charging method at a C/2 rate and a discharge rate of 1.5C.

Battery Life Prediction Notebook | Google Colab

Objective: Develop a predictive model for the remaining battery life of the batteries based on several features such as discharge time, voltage, charging time, etc.

Data:

Battery Data (CSV)

The data includes measurements from charging and discharging 14 batteries until the remaining useful life is reached.

  • Cycle Index: number of cycle
  • F1: Discharge Time (s)
  • F2: Time at 4.15V (s)
  • F3: Time Constant Current (s)
  • F4: Decrement 3.6-3.4V (s)
  • F5: Max. Voltage Discharge (V)
  • F6: Min. Voltage Charge (V)
  • F7: Charging Time (s)
  • Total time (s)

The cycle count is recorded for each charge/discharge cycle. When the battery reaches a predefined lower charge capacity, the remaining useful life is set to zero and preceding cycles are recorded as 1, 2, 3...as the target variable RUL (Remaining Useful Life).

  • RUL: target

Methods:

  • Data cleansing, visualization (e.g., pair plots, joint plots), and exploration.
  • Application of multiple regression methods including Linear Regression, Multiple Linear Regression, PyTorch, and TensorFlow models.
  • Splitting data into training and testing sets.
  • Regression comparison
  • View Remaining Useful Life prediction versus measured values

References

Solution

Battery Life Prediction Solution | Google Colab
Model Adjusted R-Squared R-Squared RMSE Time Taken
KernelRidge 0.96 0.96 0.20 7.81
ElasticNetCV 0.96 0.96 0.20 0.16
RidgeCV 0.96 0.96 0.20 0.01
Ridge 0.96 0.96 0.20 0.01
BayesianRidge 0.96 0.96 0.20 0.01
TransformedTargetRegressor 0.96 0.96 0.20 0.01
LinearRegression 0.96 0.96 0.20 0.01

Training Results (11 Batteries)

Test Results (3 Batteries)