Neural Network Design for State of Charge Estimation | Estimate Battery SOC With Deep Learning

Learn about the main aspects of the neural network design process for battery state of charge estimation. This video includes an overview of the training evaluation testing process, the neural network model structure, data preparation, an approach to improve robustness of the model, and SOC estimation results at multiple temperature, including -10 degrees Celsius.

- An Introduction to Battery State of Charge Estimation
- The Experiment Using Neural Networks
- Neural Networks for SOC Estimation
- Training and Prediction in MATLAB and Simulink Implementation

The focus of this video series is the application of neural networks to battery state of charge estimation. State of charge estimation is the task of the battery management system, or BMS. An accurate determination of the State of Charge (SOC) in a battery indicates to the user how long they can continue to use the battery-powered device before a recharge is needed. In a car, for example, an accurate knowledge of the time to recharge reduces anxiety and allows for appropriate trip planning.

The materials presented in this video series are the result of the work done by Carlos Vidal and - Phil Kollmeyer, both researchers at McMaster University in Hamilton, Ontario. The work was done in collaboration with engineers from FCA and published last year as an SAE paper.

Related Resources:

--------------------------------------------------------------------------------------------------------

© 2021 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc.

[fernando.liozzi.]

All good with these videos, we all know how to use TensorFlow to train neural networks, import them into simulink and generate code for microcontrollers. We want to see how they get the data set, that's what they want to show. If they do not show that, it is simply to show that they have expensive equipment and that they know, it is a demonstration of knowledge rather than teaching how to take the data to estimate the SOC. We want to see how they take the data for training. What sensors and techniques do they use. Thanks!

[ninav]

Thank you guys for sharing all of this, great and actual contribution to your peers :)

[pulkitsinghrana]

But how do we what the present SOC is for given temp, current and voltage to train the neural network ?

[QuiddelQuaddel]

This leaves the impression of being purposefully misleading to me. Not only throwing around the „nice looking“ error rates, but also suggesting wide applicability by citing the -10degC result… And not a single word on the multitude of pitfalls this approach would have in a real world application.

I know that this serves the purpose of being an advertisement rather than being educational, but this ad setup just rises red flags for me.

I think if one naively catches on to this and starts to invest, the Mathworks Consultancy Services trap is set.