Battery RUL Estimation Using Neural Network Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery diagnostic systems face challenges in accurately estimating the remaining useful life (RUL) of batteries for second-life applications without relying on previous usage and performance data, often requiring numerous test cycles that waste battery life.
Innovation Solution
A data-driven method using a trained neural network to estimate RUL from a limited number of test charge and discharge cycles, extracting features like capacity, internal resistance, and voltage differences, processed through machine learning algorithms to classify batteries into short or long RUL classes, without increasing the number of observed test cycles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of test cycles are performed to obtain measurements for RUL estimation, then measurement precision is improved, but loss of time increases due to battery life wastage
Solution Approach 1:
The patent applies partial action by performing only a limited number of test cycles (e.g., 5-10 cycles) instead of extensive testing, extracting sufficient features from these partial cycles to achieve acceptable RUL estimation accuracy while minimizing battery life consumption
Solution Approach 2:
The patent replaces traditional mechanical/physical testing approaches with data-driven machine learning methods, using neural networks to predict RUL from limited test data, thereby reducing the need for extensive physical test cycles
2Adaptability or versatility
If estimation models are trained based on measurements of test cycles of different types of batteries, then adaptability is improved, but loss of time increases due to extensive testing requirements
Solution Approach 1:
The patent develops a universal RUL estimation model that can be applied across different battery types (Lithium-ion, LFP, etc.) by training on diverse test cycle data from multiple battery types, enabling the same model to generalize to various battery chemistries without type-specific customization
Solution Approach 2:
The patent uses data from test cycles of different battery types as training data to create a generalized model, effectively copying patterns from various battery behaviors to build a universal prediction system that works across battery types
3Measurement precision
If previous usage and performance information of the battery is required for RUL estimation, then measurement precision is improved, but ease of operation deteriorates due to data collection complexity
Solution Approach 1:
The patent enables the battery diagnostic system to perform self-assessment by estimating RUL based solely on test cycle measurements from the battery itself, without requiring external historical usage data, making the system autonomous and easier to deploy
Solution Approach 2:
The patent extracts only the essential features needed for RUL estimation (capacity, voltage, current, temperature) from test cycles, discarding the need for comprehensive historical usage data, thereby simplifying the input requirements while maintaining estimation capability
Data Source
AI summary
A battery diagnostic system is disclosed for an RUL estimation of a battery. The battery diagnostic system comprises a memory configured to store a neural network trained to estimate a remaining useful life (RUL) of a test battery from a predetermined set of features indicative of a battery cycle of the test battery and a capacity of the test battery; a charging system configured to charge and discharge the test battery to provide measurements of the battery cycle and the capacity of the test battery; a processor configured to extract the predetermined set of features from the measurements to submit the extracted set of features to the neural network and an output interface configured to output the estimated RUL of the test battery.