Battery Peak Power Prediction Under SOC and HPPC Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting battery peak power, such as offline test table methods and dynamic electrochemical models, fail to accurately consider state of charge (SOC) constraints, leading to overcharge and overdischarge phenomena, safety issues, and increased computational complexity.
Innovation Solution
A method using a first-order RC power battery model and a 10s equivalent resistance calculation formula, combined with a recursive least squares algorithm, to predict peak power by calculating a 10s equivalent resistance value and selecting the smaller of two predicted peak powers, thereby avoiding overcharge and overdischarge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If offline test table method is used for peak power prediction, then the prediction principle is well-established, but the model is too simple considering only voltage factor without SOC constraint, causing predicted power to exceed actual value and leading to overcharge and overdischarge phenomena
Solution Approach 1:
The patent transforms the static offline test table into a dynamic prediction model by introducing SOC-dependent parameters. The equivalent resistance and peak power are expressed as functions of SOC through polynomial fitting, allowing the model to adapt to different battery states while maintaining computational simplicity.
Solution Approach 2:
The patent converts the static offline test data into a dynamic model where parameters change with SOC. The equivalent resistance Re(SOC) and peak power PP(SOC) are dynamically adjusted based on the current state of charge, enabling real-time adaptation without requiring complex real-time measurements.
2Reliability
If offline test table method is used, then prediction framework is established, but large amount of experimental work including numerous charge-discharge tests under different life states is required which is particularly difficult to complete
Solution Approach 1:
The patent performs the complex experimental work offline to build the Re(SOC) and PP(SOC) polynomial relationships beforehand. Once these polynomial parameters are established through initial experiments, they can be reused for predictions without requiring repeated extensive testing, significantly reducing future experimental time requirements.
Solution Approach 2:
The patent creates a simplified polynomial representation that copies the essential characteristics of the complex battery behavior. Instead of performing full charge-discharge tests each time, the system uses the pre-fitted polynomial equations to replicate battery performance predictions, saving substantial experimental effort.
3Ease of manufacture
If offline test table method is used, then charge-discharge power table is created, but adaptability in online applications is poor as temperature change of battery and battery attenuation are not considered
Solution Approach 1:
The patent introduces temperature-dependent polynomial parameters to adjust Re(SOC) and PP(SOC) based on operating temperature. The model uses temperature compensation factors to modify the polynomial coefficients, enabling adaptation to different temperature conditions while maintaining the simple polynomial calculation structure.
4Reliability
If dynamic electrochemical model method is used for peak power prediction, then SOC constraint is considered, but parameter identification is not dynamic and integral calculations need to be introduced causing calculations to be more complicated and amount of calculations to be substantially increased
Solution Approach 1:
The patent extracts the essential SOC-dependent characteristics from the complex dynamic electrochemical model and represents them through simple polynomial equations. By taking out only the critical SOC-power relationship and approximating it with polynomials, the model maintains SOC constraint consideration while eliminating the need for complex integral calculations and dynamic parameter identification.
Data Source
Figure 1
Figure 2~4
Figure 5
AI summary
The present application discloses a method and device for predicting a peak power of a battery. The method comprises, upon obtaining a target battery to be predicted, calculating a 10s equivalent resistance value of the target battery using a pre-built 10s equivalent resistance calculation formula, predicting a first peak power of the target battery based on the 10s equivalent resistance value, acquiring the peak power of the target battery under a HPPC condition as a second peak power at the same time, and selecting the smaller one of the first peak power and the second peak power as the peak power of the target battery. It can be seen that the present application predicts the 10s peak power of the target battery based on the pre-built 10s equivalent resistance calculation formula, and compares the predicted result with the peak power of the target battery under the HPPC condition to obtain a final 10s peak power of the target battery. This can avoid over charge and over discharge phenomena of the target battery and protect the target battery better.