AI powered tool to predict failure modes, reduce testing time and improve R&D and production testing of batteries and supercapacitors.
Current lifetime testing for devices such as batteries and supercapacitor cells can take anywhere between 3 months to 1 year. These long duration of manual testing procedures in the R&D and production stages, produce unnecessary time and cost consequences.
Voltx provides a machine learning product to predict lifetime using previous test data and as a result we can accelerate this testing process by 85-96%. We’ve developed a product, working with top manufacturers in the energy storage space, to make lifetime predictions using machine learning.
Using just 4% of data from the typical lifetime tests per cell, we can reduce the time it takes to test a cell from 3 months on average to 3 days. Our predictions of lifetime have accuracies as high as 90-95%. Our web application seamlessly integrates into your existing testing pipeline.
Our model uses data from previous lifetime tests done on various cells to effectively learn how variables such as voltage and temperature impact capacitance and ESR of cells. For the prediction, we require very little input data to make a full prediction on when the cell will degrade and under what conditions. For ex., for a lifetime prediction of 5000hrs, we only require an initial 500hrs of test data on that cell.
“This product would be of immense value for battery and supercapacitor companies that don’t have a built in machine learning platform.”
VP at Tesla
Cell Degradation Results
Using the initial test data, we are able to predict the entire lifetime of a cell including ESR and capacitance changes. We showcase how different values of voltage and temperature for example can impact the lifetime of the cell. These results are displayed is both an interactable graph and table for further analysis.
“Our company has been looking at a solution for this problem for years and we have yet to find much that uses machine learning. This product is of huge value-add for us as it will help us save money and costs spent on manual testing.”
Engineer at Nanoramic
Instead of having to do seperate tests of each cell or having long duration of tests, we allow you to test and predict multiple cells all at once. This makes it very easy to compare the degradation curve of various cells against each other and much faster to do all the predictions at once rather than seperately.
“This is definitely a problem when developing supercapacitors. If we can solve this, it can help our company save a lot of money and time.”
Head of Engineering at AVX
Testing time reduced by 85-95% per batch which will result in more products being commercialized each year.
On average, our models are able to predict lifetime and cycle life with an accuracy of 90-95+%.
Only around 4% of the testing data for a cell needed from what companies would usually collect to run our models.
These are some of the products that we have right now and what we hope to develop in the future.
Our current product is able to make highly accurate full lifetime predictions for various supercapacitor and battery cell.
Cycle Life Prediction
We understand that cycle life testing is just as important as lifetime testing. This is why we are working to develop functionally that allows you to do predict outcomes for a cell for both.
We are currently focused on battery and supercapacitor/capacitor testing. As our model is highly adaptable, in the future, we hope to expand to battery recyling remaining life prediction, and other electronics globally.