Battery State of Charge Estimation via Twin Cell Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the state of charge of a battery require significant computing power, which can lead to increased risks of overcharging or over-discharging, especially when voltage thresholds are exceeded, and do not efficiently manage computing resources.
Innovation Solution
The method reduces computing power by limiting the frequency of state of charge estimation for cells with stable voltages, using prediction calculations instead of full corrections, and identifying twin cells to execute the estimation algorithm for only one cell, thereby conserving resources and enhancing precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the state of charge of each cell is estimated at each instant using a complete estimation algorithm, then the accuracy of the state of charge determination is improved, but the computing power required increases significantly
Solution Approach 1:
The patent divides the battery into groups based on voltage similarity, and further segments the estimation process by identifying twin cells (cells with identical voltage values). Only one cell from each twin group undergoes complete estimation algorithm execution, while other twin cells reuse the estimation results. This segmentation reduces the number of complete algorithm executions from N (total cells) to N/M (where M is the average number of twins per group), significantly reducing computing power requirements while maintaining accuracy.
Solution Approach 2:
The patent changes the estimation frequency parameter dynamically based on voltage stability. When voltage varies little between instants, the complete estimation algorithm is executed less frequently. Instead, prediction calculations are performed at each instant, with corrections applied only when voltage changes exceed a threshold or at scheduled intervals. This parameter adjustment maintains accuracy for stable cells while reducing computational burden.
2Power
If the frequency of state of charge estimation is reduced to save computing power, then the computing resources are conserved, but the risk of overcharging or over-discharging increases
Solution Approach 1:
The patent applies different estimation frequencies and algorithms to different cells based on their local characteristics. Cells with voltages above a high threshold or below a low threshold (critical cells at risk of overcharging or over-discharging) have their state of charge estimated at every instant using the complete algorithm. Cells with stable voltages in the intermediate range use reduced-frequency estimation. This localized quality adjustment ensures high reliability for critical cells while conserving computing resources for stable cells.
Solution Approach 2:
The patent implements feedback mechanisms where voltage measurements trigger adaptive estimation frequency adjustments. When voltage exceeds thresholds or changes rapidly, the system increases estimation frequency for affected cells. The prediction-correction mechanism also provides feedback, where prediction errors trigger additional correction steps. This feedback ensures that reliability is maintained dynamically based on actual battery conditions while optimizing resource usage.
3Measurement precision
If the state of charge of every cell is estimated independently at each instant, then the precision of individual cell estimation is maintained, but the time required for computation increases
Solution Approach 1:
The patent merges the estimation process for twin cells by executing the complete estimation algorithm only once per twin group and reusing the result for all cells in that group. Since twin cells have identical voltage values and similar characteristics, their state of charge estimates are effectively the same. This merging reduces computation time from proportional to N (number of cells) to proportional to N/M, while maintaining precision through the prediction-correction mechanism for non-twin cells.
Solution Approach 2:
The patent performs preliminary grouping of cells by voltage value before execution, identifying twin cells in advance. This preliminary action allows the system to optimize the estimation schedule beforehand, knowing which cells can share results. The prediction step is also performed preliminarily at each instant for all cells, with corrections applied only when necessary, reducing overall computation time while maintaining precision.
Data Source
Figure 1~3
Figure 4~9
Figure 10~13
AI summary
The invention relates to a method for determining the state of charge of a battery having several cells comprising: - for each cell, calculating (204) a difference in voltage between a value yk measured across the terminals thereof at a time k and a value yk-X measured at a previous time, - assigning (206) to each cell a priority level that increases commensurate with the calculated difference in voltage, - scheduling (208) times for refreshing the estimation of the state of charge of each cell according to the assigned priority levels, wherein this scheduling comprises assigning, during a same time interval, a greater number of refresh times to cells having higher priority levels compared to the number of refresh times assigned to the other cells, and - for each cell, fully executing (210) an algorithm for estimating the state of charge of said cell at the scheduled refresh times.