This technology provides an early warning system to predict and detect thermal runaway in rechargeable batteries. This system performs minimal-frequency electrochemical impedance spectroscopy (EIS) and then uses a physics-informed machine learning algorithm to process the data and detect thermal runaway before it escalates. The use of machine learning enables this system to reconstruct a full EIS response from sparse measurements that can be performed during standard battery operation. This system allows batteries to continuously quietly self-asses for thermal danger and detect thermal runaway risk long before it becomes visible. Background: Thermal runaway is a vicious cycle in batteries in which heat generation exceeds the system’s ability to disperse it, triggering further heat-producing reactions, leading to a dangerous battery failure. Thermal runaway remains the foremost safety concern associated with rechargeable batteries, and a major limiting factor to their growth. With the rise of electric vehicles and consumer electronic devices, it’s increasingly important to be able to detect and prevent thermal runaway events. Thermal runaway prediction technologies based on electrochemical impedance spectroscopy (EIS) have been researched, as have thermal runaway prediction technologies enhanced by machine learning. This technology combines the two approaches to achieve a system to detect thermal runaway events at the earliest possible stage. Applications:
Advantages: