Systems and methods for forecasting battery state of charge

1. A system for forecasting state of charge (SOC) of a battery, the system comprising:

a battery analyzer configured to connect to the battery;

a processor in operable communication with the battery analyzer; and

a machine-readable medium in operable communication with the processor and the battery analyzer, the machine-readable medium having instructions stored thereon that, when executed by the processor, perform the following steps:

obtaining data about the battery from the battery analyzer;

forecasting a first C-rate, a voltage, a current, and a SOC percentage (SOC %) of the battery using a univariate model with parameters from at least one second C-rate, each second C-rate of the at least one second C-rate being higher than the first C-rate; and

providing the forecasted first C-rate, voltage, current, and SOC % to a machine learning model to obtain a forecasted SOC of the battery,

the providing of the forecasted first C-rate, voltage, current, and SOC % to the machine learning model to obtain the forecasted SOC of the battery comprising using an optimizer together with the machine learning model,

the optimizer being a Variable Learning Rate Gradient Descent (GDX) optimizer or an AdaGrad optimizer.

2. The system according to claim 1 , the univariate model being an Auto Regressive Integrated Moving Average (ARIMA) model.

3. The system according to claim 1 , the univariate model being a Holt Winters Exponential Smoothing (HWES) model.

4. The system according to claim 1 , the machine learning model being a Multilayer Perceptron (MLP) model.

5. The system according to claim 1 , the machine learning model being a nonlinear autoregressive neural network with external input (NARX-net) model.

6. The system according to claim 1 , the first C-rate being a C/10 C-rate.

7. The system according to claim 6 , the at least one second C-rate comprising a C/2 C-rate, a C/4 C-rate, a C/6 C-rate, and a C/8 C-rate.

8. The system according to claim 1 , the at least one second C-rate comprising a C/2 C-rate, a C/4 C-rate, a C/6 C-rate, and a C/8 C-rate.

9. The system according to claim 1 , the parameters from the at least one second C-rate being stored on the machine-readable medium after being obtained by the battery analyzer.

10. The system according to claim 1 , the battery being a lithium ion battery.

11. A method for forecasting state of charge (SOC) of a battery, the method comprising:

obtaining, by a processor, data about the battery from a battery analyzer connected to the battery;

forecasting, by the processor, a first C-rate, a voltage, a current, and a SOC percentage (SOC %) of the battery using a univariate model with parameters from at least one second C-rate, each second C-rate of the at least one second C-rate being higher than the first C-rate; and

providing, by the processor, the forecasted first C-rate, voltage, current, and SOC % to a machine learning model to obtain a forecasted SOC of the battery,

the providing of the forecasted first C-rate, voltage, current, and SOC % to the machine learning model to obtain the forecasted SOC of the battery comprising using an optimizer together with the machine learning model,

the optimizer being a Variable Learning Rate Gradient Descent (GDX) optimizer or an AdaGrad optimizer.

12. The method according to claim 11 , the univariate model being an Auto Regressive Integrated Moving Average (ARIMA) model.

13. The method according to claim 11 , the machine learning model being a Multilayer Perceptron (MLP) model.

14. The method according to claim 11 , the machine learning model being a nonlinear autoregressive neural network with external input (NARX-net) model.

15. The method according to claim 11 , the first C-rate being a C/10 C-rate, and

the at least one second C-rate comprising a C/2 C-rate, a C/4 C-rate, a C/6 C-rate, and a C/8 C-rate.

16. The method according to claim 11 , the parameters from the at least one second C-rate being stored on a machine-readable medium in operable communication with the processor after being obtained by the battery analyzer.

17. The method according to claim 11 , the battery being a lithium ion battery.

18. A system for forecasting state of charge (SOC) of a battery, the system comprising:

a battery analyzer configured to connect to the battery;

a processor in operable communication with the battery analyzer; and

a machine-readable medium in operable communication with the processor and the battery analyzer, the machine-readable medium having instructions stored thereon that, when executed by the processor, perform the following steps:

obtaining data about the battery from the battery analyzer;

forecasting a first C-rate, a voltage, a current, and a SOC percentage (SOC %) of the battery using a univariate model with parameters from at least one second C-rate, each second C-rate of the at least one second C-rate being higher than the first C-rate; and

providing the forecasted first C-rate, voltage, current, and SOC % to a machine learning model to obtain a forecasted SOC of the battery,

the univariate model being an Auto Regressive Integrated Moving Average (ARIMA) model,

the machine learning model being a Multilayer Perceptron (MLP) model or a nonlinear autoregressive neural network with external input (NARX-net) model,

the first C-rate being a C/10 C-rate,

the at least one second C-rate comprising a C/2 C-rate, a C/4 C-rate, a C/6 C-rate, and a C/8 C-rate,

the providing of the forecasted first C-rate, voltage, current, and SOC % to the machine learning model to obtain the forecasted SOC of the battery comprising using an optimizer together with the machine learning model,

the optimizer being a Variable Learning Rate Gradient Descent (GDX) optimizer or an AdaGrad optimizer,

the parameters from the at least one second C-rate being stored on the machine-readable medium after being obtained by the battery analyzer, and

the battery being a lithium ion battery.