Battery Deterioration Judgment Using Reliability-Weighted AI Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery deterioration judgment systems face challenges in accurately assessing battery deterioration, especially when insufficient data is available, and often have low accuracy due to reliance on single calculation models and limited data handling.
Innovation Solution
A battery deterioration judging system that utilizes multiple calculation models, including physical and AI-based models, to derive and combine deterioration probabilities, with a reliability degree assessment to enhance accuracy, allowing for accurate judgment even with limited data. The system transitions from a physical model to an AI model as data accumulates, effectively handling dispersion in later stages of deterioration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single calculation model is used to judge battery deterioration, then the device complexity is low, but the measurement precision of deterioration judgment is insufficient
Solution Approach 1:
The patent combines multiple calculation models (first calculation model and second calculation model) to derive deterioration probabilities. The judging section integrates results from both models, weighting them by reliability degree, to achieve more accurate deterioration judgment than a single model could provide alone.
Solution Approach 2:
The patent creates a composite judgment system that combines different types of calculation models (physical models and AI models) into a unified deterioration assessment framework. This composite approach leverages the strengths of different model types to improve overall judgment accuracy.
2Measurement precision
If judgment is made with limited state amount data, then the productivity is high (early judgment possible), but the measurement precision is low
Solution Approach 1:
The patent dynamically adjusts the reliability degree of the second calculation model based on the number of accumulated state amounts. When data is limited, the system adaptively weights the first calculation model higher; as data accumulates, it transitions to relying more on the second calculation model, optimizing accuracy at each stage of data availability.
Solution Approach 2:
The patent prepares multiple calculation models in advance, with the first calculation model designed to provide reliable judgments even with minimal data. This preliminary preparation enables accurate deterioration assessment from the early stages of battery operation, before extensive data accumulation is required.
3Measurement precision
If only physical models are used for deterioration judgment, then the reliability is stable, but the measurement precision is insufficient for later stages of deterioration
Solution Approach 1:
The patent changes the weighting parameters (reliability degree) assigned to different calculation models based on the accumulation of state amounts. As more data becomes available, the system transitions from relying on physical models to incorporating AI-based models more heavily, adapting to the changing reliability characteristics at different deterioration stages.
Solution Approach 2:
The patent introduces the reliability degree as an intermediary parameter that mediates between multiple calculation models with different reliability characteristics. This intermediary weighting mechanism allows the system to balance between the stability of physical models and the improving accuracy of AI models as data accumulates.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A battery deterioration judging system (10) comprises an acquiring section (22, 30), a first deriving section (41), a reliability degree deriving section (36), a second deriving section (42), and a judging section (44). The acquiring section (22, 30) acquires a state amount of a battery (20). The first deriving section (41) derives a first deterioration probability of the battery (20), based on the state amount of the battery (20) acquired by the acquiring section (22, 30) and a predetermined first calculation model. The reliability degree deriving section (36) derives a reliability degree of a second calculation model, which is different than the first calculation model, based on a number of state amounts acquired by the acquiring section (22, 30). The second deriving section (42) derives a second deterioration probability of the battery (20), based on the state amount of the battery (20) acquired by the acquiring section (22, 30) and the second calculation model. The judging section (44) judges deterioration of the battery (20) based on the reliability degree, and at least one of the first deterioration probability or the second deterioration probability.