Battery Impedance Testing for Capacity Assessment Without BMS History
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery state estimation methods require specific training for each battery type and rely on knowledge of battery history or management systems, making them complex and time-consuming for assessing multiple batteries of the same kind.
Innovation Solution
A battery test system that includes a battery test bench and a server, using a machine learning algorithm trained on initial measurements of a group of batteries to quickly assess the state of further batteries of the same kind through electrical impedance spectroscopy, without needing individual battery history or management system data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional battery state estimation methods are used, then accurate battery state assessment can be achieved, but the process becomes complex and time-consuming due to requirements for battery history knowledge and management system data
Solution Approach 1:
The patent extracts and removes the requirement for complex battery history data and management system interactions from the testing process. By using only electrical impedance spectroscopy measurements combined with a trained machine learning model, the system eliminates the need for additional battery information inputs, thereby reducing testing complexity while maintaining assessment accuracy
Solution Approach 2:
The patent replaces the conventional mechanical/information-intensive approach (requiring battery history data, management system access, and multiple measurement types) with an electrical measurement approach using electrical impedance spectroscopy combined with machine learning. This substitution simplifies the testing process by replacing complex data collection requirements with a streamlined electrical measurement and computational analysis system
2Reliability
If conventional battery state estimation methods are used, then reliable battery assessment can be achieved, but significant time is required for training and measuring each individual battery
Solution Approach 1:
The patent applies preliminary action by training the machine learning model once on a dataset of electrical impedance spectra and corresponding battery states before deployment. This preliminary training phase creates a reusable assessment system that can quickly evaluate individual batteries without requiring time-consuming retraining for each battery, thereby reducing the time loss for individual battery assessments while maintaining reliability
Solution Approach 2:
The patent changes the measurement approach from requiring multiple measurement types and extensive battery information to using only electrical impedance spectroscopy parameters. By transforming the assessment methodology to rely on a single measurement type (electrical impedance) processed through a pre-trained machine learning model, the system reduces testing time while preserving assessment reliability
3Loss of information
If detailed battery history and management system data are collected, then comprehensive battery state information can be obtained, but the testing process becomes more complex and requires more resources
Solution Approach 1:
The patent creates a computational model (machine learning algorithm) that copies and learns the relationship between electrical impedance characteristics and battery states from training data. This model then serves as a virtual expert system that can assess battery states using only electrical impedance measurements, eliminating the need for physical access to battery management systems or historical data while maintaining information completeness
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables rapid, reliable, and cost-effective assessment of battery state for multiple batteries of the same kind, simplifying the testing process and eliminating the need for detailed battery history or management system interaction.
Implementation Method 1
a measurement device configured for performing a battery capacity measurement and an electrical impedance spectrum (EIS) measurement on an electrochemical battery connected to the measurement device
Data Source
AI summary
A battery test bench, server, and battery test system, and a method for assessing a battery state of electrochemical batteries, the battery test bench having a measurement device for performing a battery capacity measurement and an electrical impedance spectrum measurement on an electrochemical battery, a machine learning algorithm of the server being configured for processing a measured electrical impedance spectrum. A battery capacity and an electrical impedance spectrum are measured on a number of batteries of a same kind. The obtained first measurement data are transmitted to the server via a communication network, and the machine learning algorithm is trained based on the first measurement data. Then, an electrical impedance spectrum is measured on at least one further battery of the same kind. The obtained second measurement data are transmitted to the server and evaluated by the trained machine learning algorithm, including processing the measured electrical impedance spectrum by the machine learning algorithm, and generating an output that represents battery state information relating to a battery capacity.