Multi-Level Battery State Estimation for Voltage Flat Zones
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional neural network models, such as LSTM, FNN, CNN-LSTM, TCN, and Transformer, face challenges in accurately estimating the state of charge (SOC) and health of batteries, particularly in the voltage flat zone of lithium-iron batteries, due to high estimation errors and limited generalization ability across different battery types and conditions.
Innovation Solution
A multi-level neural network architecture is proposed, comprising a denoising autoencoder as the first-level network, a temporal convolution model as the second-level network, and an attention model as the third-level network. This architecture enhances local features and reduces noise, improving estimation accuracy in the voltage flat zone and enabling quick adaptation to different battery types through transfer learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional neural network models (LSTM, FNN, CNN-LSTM, TCN, Transformer) are used to estimate battery state, then the model can process battery data, but the estimation error in voltage flat zone exceeds 10%
Solution Approach 1:
The patent segments the battery state estimation task into multiple levels: a first-level neural network extracts initial features from raw battery data, a second-level neural network processes temporal dependencies, and a third-level neural network performs final state estimation. This multi-level segmentation allows each network to specialize in specific aspects, improving accuracy in voltage flat zones while maintaining adaptability across different battery types through modular architecture.
2Measurement precision
If TCN model is used to enhance local features, then estimation accuracy in voltage flat zone improves, but computing time increases significantly
Solution Approach 1:
The patent applies local quality by having the second-level neural network specifically focus on extracting temporal features from voltage flat zone data, while the first and third levels handle general feature extraction and state estimation. This localized specialization allows the model to achieve high accuracy in voltage flat zones without requiring the entire system to be computationally intensive, thus reducing overall computing time compared to using TCN throughout.
3Adaptability or versatility
If Transformer model is used for better generalization, then the model can estimate batteries of different brands, but it lacks local feature information and cannot accurately estimate lithium-iron batteries
Solution Approach 1:
The patent merges the strengths of different neural network architectures by combining a Transformer-based first-level network for general feature extraction (providing good generalization across battery brands) with a TCN-based second-level network for temporal feature processing (providing local feature information for lithium-iron batteries). This fusion allows the model to achieve both broad adaptability and specialized accuracy simultaneously.
Data Source
AI summary
A method is to estimate the state of batteries by using a multi-level neural network formed with at least three neural networks. The method comprises steps of: extracting features from the charging and discharging data of a battery through a first-level neural network to form a first-stage output data, and inputting the first-stage output data into a second-level neural network; enhancing local features in the first-stage output data through the second-level neural network to form a second-stage output data; combining the first-stage output data with the second-stage output data to form a combination result to be input into a third-level neural network for data modeling, to generate a state estimation result of the battery. The present invention improves the accuracy of estimation for a flat zone in the charge/discharge curve of the battery, and quickly adjusts the multi-level neural network to achieve accurate estimation of different types of batteries.