Anomaly Detection for Secondary Battery Using Kalman Filter and LSTM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods struggle to accurately predict and detect anomalies in secondary batteries used in electric vehicles, leading to potential safety issues due to degradation and sudden failures, which are difficult to predict and require precise remaining capacity information.
Innovation Solution
A control system utilizing a nonlinear Kalman filter and neural networks, specifically LSTM, to estimate internal resistance and State Of Charge (SOC), predicting anomalies by comparing estimated and predicted values, thereby cautioning users and adjusting charging conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to monitor battery status, then the system is simple to operate, but anomaly detection precision is insufficient
Solution Approach 1:
The patent introduces a neural network model as an intermediary between the battery monitoring system and the anomaly detection process. The neural network processes voltage, current, and temperature data to predict future battery states, enabling precise anomaly detection without requiring complex manual analysis or direct intervention in the battery system itself.
Solution Approach 2:
The system performs preliminary actions by training the neural network model in advance using historical battery data. This pre-training enables the model to predict future battery states accurately, allowing anomalies to be detected before they actually occur by comparing predicted states with actual measurements.
2Measurement precision
If neural network methods are used for anomaly detection, then detection precision is improved, but computational resources and time are increased
Solution Approach 1:
The neural network model is trained in advance using historical battery data containing voltage, current, and temperature information. This pre-training phase performs the computationally intensive work beforehand, so that during actual operation, the model can quickly predict future battery states and detect anomalies with minimal real-time computation.
Solution Approach 2:
The system continuously monitors battery voltage, current, and temperature while the neural network continuously predicts future states. This continuous operation allows the system to maintain accurate predictions and detect anomalies in real-time without periodic interruptions or batch processing delays.
3Reliability
If traditional monitoring methods are used, then the system is easy to manufacture, but reliability for long-term use is insufficient
Solution Approach 1:
A neural network model serves as an intermediary layer between the simple battery sensors and the safety monitoring system. This model enhances reliability by accurately predicting future battery states and detecting anomalies that traditional methods would miss, while the underlying hardware remains relatively simple and manufacturable.
Solution Approach 2:
The system performs preliminary anomaly detection by comparing actual battery states with predicted states from the neural network. This advance detection capability identifies potential safety issues before they manifest as actual problems, significantly improving long-term reliability without requiring complex hardware changes.
Data Source
AI summary
An anomaly detection system for a secondary battery which detects the remaining capacity of the secondary battery on an electric vehicle, cautions against the secondary battery with anomalous characteristics, stops using the secondary battery, changes the secondary battery, or changes charging conditions of the secondary battery is provided. The anomaly detection system is provided; the system compares a value obtained by estimating internal resistance or SOC of a secondary battery based on the measured value of a current or a voltage of the secondary battery with the use of a nonlinear Kalman filter and a value input to an anomaly detection system (network) of AI to predict a change in the internal resistance; the system regards a case where the difference is large as an anomaly; and the system detects an anomaly.