Adaptive Battery Charging Using Anode Potential to Limit Lithium Plating
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Lithium plating in Li-ion batteries during charging, which reduces battery lifespan and hinders fast-charging capabilities, is not effectively addressed by existing technologies, leading to premature failures and increased recycling demands.
Innovation Solution
A computer-implemented method that predicts anode potential and adjusts charging policies using machine learning models to prevent lithium plating by modifying anode potential offsets, leveraging cloud and edge processors for real-time monitoring and predictive maintenance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fast charging is applied to Li-ion batteries, then charging speed is improved, but lithium plating occurs which reduces battery lifespan
Solution Approach 1:
The system performs preliminary assessment of battery state (temperature, charge level, age) before initiating fast charging, and preemptively adjusts charging parameters to prevent lithium plating before it occurs. The machine learning model predicts potential plating risks in advance and modifies charging policy proactively.
Solution Approach 2:
The charging policy is made dynamic and adaptive rather than fixed. The system continuously monitors battery conditions and adjusts charging current, voltage, and temperature parameters in real-time based on predicted lithium plating risk, allowing optimal balancing of charging speed and battery protection throughout the charging process.
2Reliability
If traditional charging policies are used, then battery lifespan is maintained, but fast-charging capabilities are hindered
Solution Approach 1:
The system dynamically changes charging parameters (current, voltage, temperature thresholds) based on real-time battery state assessment. Machine learning models predict optimal parameter combinations that enable fast charging while maintaining battery lifespan by preventing lithium plating through adaptive parameter adjustment.
Solution Approach 2:
The system implements continuous feedback loops where battery performance data, temperature readings, and charge level information are fed back to the machine learning model, which then refines charging policy predictions. This closed-loop control enables the system to learn from actual battery responses and optimize fast charging parameters while protecting battery lifespan.
3Loss of time
If lithium plating is allowed to occur, then charging time is reduced, but battery performance degrades and premature failures increase
Solution Approach 1:
The system applies preliminary anti-action by predicting lithium plating risk before it occurs and counteracting it through preventive charging policy adjustments. The machine learning model identifies conditions that would lead to plating and modifies charging parameters in advance to prevent the harmful effect while still enabling fast charging.
Solution Approach 2:
The system converts the potential harm of lithium plating into a benefit by using the prediction of plating risk as a signal to optimize charging parameters. The machine learning model transforms what would be a damaging condition into an opportunity to adjust charging strategy, achieving both fast charging and battery protection simultaneously.
Data Source
AI summary
In one aspect, a computer-implemented method may include determining whether an anode potential offset of a battery pack has reached an upper threshold of a target range, responsive to determining that the anode potential offset of the battery pack has reached the upper threshold of the target range, receiving tear-down data associated with the battery pack, training, based on the tear-down data associated with the battery pack, cloud prediction models that are trained to predict a likelihood of occurrence of lithium plating associated with battery packs, and transmitting, to edge processors of fleet vehicles, parameters associated with the trained prediction models to cause the edge processors to train edge prediction models using the edge processors, wherein the edge processors modify a battery charging policy based on a prediction from the edge prediction models.