Battery Charging Control Using Neural Network Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery charging methods, such as CC/CV and multi-step charging, are not suitable for fast charging as they lead to battery degradation and limit charging time, which is a concern for increasing demands for quick charging in electric vehicles and mobile devices.
Innovation Solution
A battery charging method that generates charging currents and profiles based on a desired charging time and an electrochemical model, incorporating internal state conditions and maximum charging times to prevent degradation, by estimating internal states and adjusting charging parameters to minimize battery degradation while ensuring fast charging.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If CC/CV charging scheme is used, then charging voltage is maintained, but charging time is excessively long
Solution Approach 1:
The patent implements dynamic charging control by continuously monitoring battery internal states (SOC, temperature, voltage) and adjusting charging current in real-time. The charging current is dynamically modified based on predicted future states to prevent degradation while maintaining fast charging, rather than using fixed CC/CV stages.
Solution Approach 2:
The system performs preliminary prediction of battery internal states using a trained neural network model before actual charging occurs. By predicting future SOC, temperature, and voltage states based on current conditions and proposed charging currents, the system proactively adjusts charging parameters to avoid degradation conditions before they occur.
2Loss of time
If fast charging is performed, then charging time is reduced, but battery degradation occurs
Solution Approach 1:
The patent employs closed-loop feedback control where the neural network model continuously predicts battery response to charging currents, and the system adjusts charging current based on these predictions. The predicted internal states (SOC, temperature, voltage) feed back into the control algorithm to optimize charging current, preventing degradation while enabling fast charging.
Solution Approach 2:
The system changes charging current parameters dynamically based on predicted battery states. Instead of fixed current levels, the charging current is continuously adjusted according to predicted future states of the battery, allowing fast charging while staying within safe operational boundaries that prevent degradation.
3Reliability
If multi-step charging scheme is used, then charging current is reduced in steps, but charging time remains long and control complexity increases
Solution Approach 1:
The patent replaces traditional mechanical/stage-based charging control with an intelligent predictive control system using neural networks. Instead of predefined charging stages with fixed current reductions, the system uses data-driven prediction to continuously optimize charging current, achieving both fast charging and battery protection without complex multi-step transitions.
4Device complexity
If experience-based charging scheme is used, then simple control is provided, but charging time cannot be sufficiently reduced and degradation control is limited
Solution Approach 1:
The patent implements self-service charging control where the neural network model autonomously predicts battery behavior and determines optimal charging currents without requiring complex external control logic. The system serves itself by using the trained model to automatically adjust charging parameters, achieving fast charging with simplified control architecture.
Data Source
AI summary
A battery charging method and apparatus are provided. The battery charging apparatus receives a desired charging time of a battery, generates charging currents of charging steps to charge the battery based on the desired charging time, acquires a charging limit condition including an internal state condition and a maximum charging time for each of the charging steps based on the desired charging time and an electrochemical model of the battery, and generates a charging profile including the charging currents and charging times of the charging currents based on the charging limit condition.