A lot of us are probably more interested in knowing the battery lifespan of our smartphones over our actual lifespan, and US scientists have recently thought of an idea that utilizes machine learning to project the life cycle of batteries.
The lifespan of batteries is dependent on utilization, as well as design and chemical composition, and there are no optimal approaches in indicating the remaining duration of batteries as of now, which is a problem because users are always wondering when should they buy new batteries.
Noah Paulson, a computational scientist at the Argonne National Laboratory of DOE, commented that the lifespan of batteries is essential to various applications from smartphones to EVs and grid-level energy storage systems. Susan “Sue” Babinec, a chemist also at the Argonne National Laboratory, added that the only method in evaluating the capacity of batteries at the current stage is to conduct actual charging and discharging cycles, which is both time and cost consuming.
As a result, the laboratory wishes to forecast the lifespan of batteries formed with various chemical substances through machine learning. The team first collected 300 pieces of battery experiment statistics, including batteries containing six chemicals, which allows scientists in confirming the lifespan cycle of different batteries. Machine learning is then applied, where the computer program first infers the preliminary data that would be transferred to another set of data.
Babinec pointed out that further studies are likely to derive the future of lithium-ion batteries, since computers would be able to predict the applications of batteries formed with other chemicals from the algorithm of known chemical substances. The system, in a way, may provide a path for subsequent battery materials, as well as offer orientations for new materials and improvement.
Babinec believes that machine learning can accelerate development and testing of battery materials, and the aforementioned system is able to first forecast the lifespan of batteries, which confirms on the decision of further investment, rather than actually testing the cycles of new materials.
(Cover photo source: pixabay)