Battery SoH Estimation Using AI and Equivalent Circuit Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for accurately determining battery state of health (SoH) values are hindered by the need for full charge/discharge tests, which are unsafe and impractical in certain scenarios, and existing models require extensive workload and long periods, making it difficult to monitor battery capacity in applications like data centers and backup power supplies.
Innovation Solution
An energy storage apparatus using an AI-based model trained by historical equivalent circuit parameters and actual SoH values, estimating SoH without deep charging or discharging, by sending alternating current signals and analyzing impedance spectrum information to update the model continuously.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full charge/discharge tests are performed to obtain accurate battery SoH values, then measurement precision is improved, but safety risks and operational feasibility deteriorate
Solution Approach 1:
The patent introduces an equivalent circuit model as an intermediary between the battery and the measurement process. Instead of directly measuring SoH through full charge/discharge tests, the system uses the equivalent circuit model to estimate SoH based on voltage, current, and temperature data collected during normal operation, thereby avoiding safety risks while maintaining measurement accuracy
Solution Approach 2:
The patent replaces the mechanical/physical process of full charge/discharge testing with an electrical modeling approach. By substituting the physical test process with an equivalent circuit model and AI-based estimation algorithm, the system achieves SoH measurement without subjecting the battery to extreme charging/discharging conditions
2Ease of operation
If traditional charge/discharge curve methods are used to estimate battery SoH, then SoH value can be obtained without full testing, but the method cannot be applied in float charging scenarios where voltage and current remain unchanged
Solution Approach 1:
The patent makes the SoH estimation method dynamic by using a living equivalent circuit model that can adapt to different operating conditions. The system continuously updates model parameters based on real-time data from float charging, partial charging, and discharge scenarios, enabling accurate SoH estimation across varying operational states rather than relying on static charge/discharge curves
Solution Approach 2:
The patent changes the estimation parameters from traditional charge/discharge curve parameters to equivalent circuit model parameters (such as internal resistance, capacitance, and inductance). These parameters can be extracted and updated during float charging operations where voltage and current are stable, allowing the system to estimate SoH in previously unsuitable operating conditions
3Measurement precision
If prior aging tests are conducted to establish aging models for SoH estimation, then estimation accuracy is improved, but the workload and time required increase significantly
Solution Approach 1:
The patent performs preliminary actions by pre-building the equivalent circuit model structure and preparing the AI estimation framework in advance. However, unlike traditional methods that require extensive prior aging tests to calibrate models, this system prepares a flexible model framework that can be quickly adapted to specific battery types using minimal initial data, significantly reducing preparation time and workload
Solution Approach 2:
The system implements self-service by automatically updating and refining the equivalent circuit model parameters through continuous data collection during normal battery operation. The AI algorithm autonomously learns from operational data to improve estimation accuracy over time, eliminating the need for manual model calibration and reducing ongoing maintenance workload
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
Accurately obtains SoH values in various scenarios, identifying potential capacity abnormalities, enhancing system stability by reducing the need for unsafe tests and improving model precision through continuous iteration.
Implementation Method 1
obtain a sampling signal from the first cell, where the sampling signal is used to obtain the real-time equivalent circuit model parameter of the first cell, and the sampling signal includes at least one of voltage information, current information, and temperature information that are of the first cell
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This application provides an energy storage apparatus, a method for obtaining a battery state of health value, and a battery management system. The energy storage apparatus, the method for obtaining the battery state of health value, and the battery management system may be used in an application scenario, for example, a data center power supply, a station backup power supply, an in-vehicle power supply, or a photovoltaic power supply, in which a battery is in a float charging state or a shallow charging and shallow discharging state for long time. The energy storage apparatus includes a battery control unit, a battery monitor unit, and a plurality of cells. The battery control unit is configured to receive a battery state of health SoH value estimation model from a cloud device, and the battery monitor unit is configured to obtain an estimated SoH value of a first cell based on a real-time equivalent circuit model parameter of the first cell in the plurality of cells and the battery SoH value estimation model. According to the energy storage apparatus disclosed in this application, a battery SoH value is estimated by using an artificial intelligence model, and deep charging and discharging do not need to be performed or a battery, so that the battery SoH value can be accurately obtained in a plurality of application scenarios, and a battery capacity abnormality risk can be identified in advance, thereby improving stability of an energy storage system.