Early-Cycle HPPC Battery Diagnostics for Catastrophic Fade
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery lifespan prediction requires long-term data collection and knowledge of degradation mechanisms, making it inefficient and impractical for identifying catastrophic fade.
Innovation Solution
A machine learning model trained on early-cycle battery test data, particularly from Hybrid Pulse Power Characterization (HPPC) tests, predicts catastrophic fade in batteries without prior knowledge of degradation mechanisms, using nonlinear algorithms like MLP, SVM, Decision Trees, Random Forests, or Gradient Boosting Machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If long-term data collection is used for battery lifespan prediction, then prediction accuracy is improved, but time consumption and operational complexity increase
Solution Approach 1:
The patent applies preliminary action by performing battery defect prediction during early cycles (first 5-10 cycles) rather than waiting for long-term data collection. The machine learning model is trained in advance on historical data and then used to predict catastrophic fade based on early-cycle HPPC test results, enabling early identification of defective batteries before they are deployed or shortly after deployment begins
Solution Approach 2:
The patent uses partial action by collecting and analyzing only a subset of battery cycle data (early cycles with HPPC tests) rather than requiring complete long-term operational data. This partial data collection approach achieves sufficient prediction accuracy while significantly reducing time consumption and operational burden
2Measurement precision
If long-term data collection is used for battery lifespan prediction, then prediction accuracy is improved, but operational complexity increases
Solution Approach 1:
The machine learning model is pre-trained on historical battery data before deployment. This preliminary training phase captures degradation patterns and relationships, so that during actual battery testing, only early-cycle HPPC data needs to be collected and input to the already-trained model, greatly simplifying operational complexity while maintaining prediction accuracy
Solution Approach 2:
The patent uses copying by applying the same machine learning prediction framework and HPPC test protocol across multiple batteries and production batches. Once the model is trained on historical data, it can be replicated and applied to predict defects in new batteries using the same early-cycle testing approach, reducing operational complexity through standardization
3Productivity
If early-cycle data is used for prediction, then time efficiency is improved, but prediction accuracy may deteriorate
Solution Approach 1:
The patent applies parameter changes by using Hybrid Pulse Power Characterization (HPPC) tests during early cycles, which apply specific current pulses at defined state-of-charge intervals (every 10% SOC from 90% to 0%). This standardized parameterized testing approach extracts meaningful degradation signals from early-cycle data that are sufficient for accurate prediction, overcoming the limitation of using only early-cycle data
Solution Approach 2:
The patent replaces traditional mechanical/physical battery testing approaches (which require long-term cycling and monitoring) with a machine learning-based predictive system. The ML model substitutes for prolonged physical observation by identifying patterns in early-cycle HPPC data that correlate with future catastrophic fade, achieving both time efficiency and accuracy
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for battery defect identification. One of the methods includes receiving battery test data of a battery cell. The battery test data includes data of at least one battery cell property in a battery test during at least one portion of a battery cycle. The battery test includes applying one or more pulses on the battery cell. The battery test data of the battery cell is provided as input to a machine learning model running on the computing system to predict whether the battery cell will experience catastrophic fade. The machine learning model has been trained using training data including battery test data of battery cells that experienced catastrophic fade. A prediction result for the battery cell is automatically generated by the machine learning model. An action is taken based on the prediction result for the battery cell.