I'm Alishba, a 17 year old who's worked with IBM on a platform to track medications in supply chain using blockchain. I've interned at top banks in Canada on AI enterprise solutions and I've lead a data science at a startup Pngme.
Our vision is to accelerate the speed of ageing testing for energy storage systems and predict failure modes through the use of machine learning.
With our novel solution, we're finding failure modes, reducing testing time and improving R&D and production of batteries and supercapacitors.
Validated By Industry Experts:
Energy storage systems are the biggest challenge towards mass adoption of grid scale renewable energy. One of the biggest bottlenecks to the development of these novel energy storage systems is the time it takes to test them.
That's where we come in. We're working on applying machine learning to help accelerate the testing processes of various energy storage systems, namely supercapacitors and batteries.
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 95+%.
Only around 4% of the testing data (~100 hours) for a cell needed from what companies would usually collect to run our models.
With machine learning models, we're able to predict the ageing of supercapacitors based off of factors like voltage, temperature and current. With just a fraction of data routinely collected by manufacturers, we can cut down testing times by 96% — from 3 months to 3.5 days. We are able to predict capacitance, ESR, leakage current and more for both lifetime and cyclelife.
Given a capacitor's voltage, current, temperature, our architecture is able to make predictions for cycle life and lifetime by finding ESR and capacitance. Previous test data are fed through a three-layer artificial neural network and an estimate for cycle life and lifetime is calculated, allowing us to more easily forecast how a supercapacitor will age.
Bayesian models can use both past and real-time data to make predictions. Our model is non-parametric — it can be fed an infinite amount of parameters and determines the relationships between them against time. By inputting ranges of voltage, temperature and current, we're able to successfully predict capacitance and ESR over time.
Within the app, we have a feature that allows you to easily collect and upload your current and past testing data in CSV format. This data is then pre-processed, cleaned and run through our machine learning models in the back-end.Learn More About Data Needed >
The prediction can be viewed on the dashboard where you can also select what type of prediction you want to make and select the parameters and variables you want to view for your results. The key results are shown and interpreted in graphical and numerical form.Start your prediction >
Learn more about the founders of the company and what we are passionate about.
Address : Toronto, Canada
Mail : email@example.com