Lithium-Ion Battery Pack SOC/SOH Estimation With NSSR-LSTM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating State of Charge (SOC) and State of Health (SOH) of lithium-ion batteries in electric vehicles face challenges due to inconsistencies among cells, nonlinear aging, and environmental factors, leading to inaccurate and unreliable estimates that impact battery performance and safety.
Innovation Solution
A machine learning method using Nonlinear State Space Reconstruction (NSSR) combined with Long Short-Term Memory (LSTM) neural networks to reconstruct phase state spaces for joint estimation of SOC and SOH, incorporating data normalization and iterative training to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional estimation algorithms are used for SOC and SOH, then the system complexity remains low, but the measurement precision and reliability deteriorate due to cell inconsistencies and nonlinear aging
Solution Approach 1:
The patent segments the estimation problem into two separate sequential estimations: first estimating SOH using NSSR-LSTM, then using the SOH result as input for SOC estimation. This segmentation allows each estimator to focus on specific characteristics (SOH on capacity degradation, SOC on charge state) while accounting for cell inconsistencies and nonlinear aging independently, thereby improving measurement precision without overwhelming system complexity
Solution Approach 2:
The patent introduces SOH estimation as an intermediary component between the battery system and SOC estimation. The SOH value acts as a mediator that captures the effects of aging and cell degradation, which are then fed into the SOC estimation process. This intermediary approach allows the system to account for nonlinear aging and cell inconsistencies indirectly, improving SOC estimation accuracy without requiring direct complex modeling of all degradation factors
2Reliability
If simple estimation algorithms are used, then the computational resources required are low, but the reliability deteriorates under varying operational conditions and aging
Solution Approach 1:
The patent performs preliminary SOH estimation that captures the effects of aging, temperature, and usage patterns before proceeding to SOC estimation. By pre-processing the degradation effects through SOH estimation, the system establishes a reliable baseline that improves subsequent SOC estimation reliability under varying operational conditions, while the computational energy is distributed across two focused tasks rather than one complex task
3Measurement precision
If separate estimation methods are used for SOC and SOH, then the measurement precision for each parameter improves, but the device complexity increases due to interdependence requirements
Solution Approach 1:
The patent merges the SOC and SOH estimation systems by making them sequential and interdependent, where SOH estimation feeds into SOC estimation. This merging allows both parameters to be estimated with high precision using specialized algorithms for each, while the integration complexity is managed through a clear sequential architecture where the output of one becomes the input of the other, avoiding the need for complex simultaneous multi-parameter optimization
Data Source
AI summary
A machine learning method of estimating State of Charge (SOC) and State of Health (SOH) of a battery pack, the method including collecting time-series data of a battery pack; processing the collected data utilizing a reconstruction algorithm to reconstruct a first phase state space reconstruction; training a first neural network model using the first phase state space reconstruction for predicting the SOH; feeding the time-series data to the first neural network to predict an estimated SOH value; processing the time-series data utilizing the reconstruction algorithm to reconstruct a second phase state space reconstruction; training a second neural network model using the second phase state space reconstruction, taking into account the estimated SOH value, for predicting the SOC, and feeding the time-series data and the estimated SOH value to the second neural network to obtain an estimated SOC value. A system configured to perform the above method is also disclosed.