Battery Abnormality Diagnosis Using Vehicle-Specific Autoencoder Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery monitoring technologies face challenges in accurately detecting abnormalities in vehicle batteries due to variations in battery characteristics and vehicle states, leading to inefficiencies and potential power loss.
Innovation Solution
A server-based system that classifies learning battery information into groups based on vehicle models and additional information, generates learning data, and constructs a diagnostic model using autoencoders to detect battery abnormalities by reflecting vehicle characteristics and states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If battery cell balancing technology is used to adjust voltages based on the battery with the lowest voltage, then voltage differences between batteries are reduced, but power loss occurs
Solution Approach 1:
The system performs preliminary classification of battery data by vehicle model and state before diagnostic analysis, enabling proactive identification of abnormal batteries before they cause power loss or failure. This allows for early intervention through targeted charging adjustments rather than reactive balancing that causes energy waste.
Solution Approach 2:
The diagnostic model is constructed separately for each vehicle model and state condition, creating localized diagnostic criteria tailored to specific battery characteristics. This enables precise identification of abnormal batteries without affecting the charging strategy of normal batteries, thereby avoiding unnecessary power loss from blanket balancing operations.
2Measurement precision
If substantial abnormal data is secured to achieve accurate detection of battery abnormalities, then detection accuracy improves, but data acquisition becomes challenging due to scarcity of abnormal data
Solution Approach 1:
The patent segments battery data by vehicle model and operational state, creating distinct data groups for each combination. This segmentation allows the system to build specialized diagnostic models for each segment, improving detection accuracy even with limited abnormal data per segment by leveraging the structure and patterns within each specific group.
Solution Approach 2:
The system uses vehicle model and state information as intermediary variables to bridge the gap between scarce abnormal data and accurate detection. These intermediaries enable the construction of diagnostic models that can generalize from limited abnormal examples by contextualizing them within specific vehicle and operational conditions.
3Device complexity
If battery diagnostic technology does not reflect vehicle characteristics and states, then system complexity is reduced, but detection accuracy of battery abnormalities decreases
Solution Approach 1:
The system creates a family of diagnostic models that are universal across different vehicle models and states rather than requiring completely separate systems for each condition. Each model in the family follows the same structural framework but is trained on data specific to its vehicle model and state, achieving both universality in approach and specificity in application.
Solution Approach 2:
The system changes the parameters used for diagnosis based on vehicle model and state conditions. By adapting the diagnostic parameters and thresholds to match specific vehicle characteristics and operational states, the system achieves high detection accuracy without requiring a fundamentally different diagnostic approach for each scenario.
Data Source
AI summary
Disclosed is a server for diagnosing a battery abnormality. The server includes a memory that stores a diagnostic model for determining an abnormal state of a target battery, and a processor that is connected to the memory, wherein the processor classifies learning battery information in a form of time series data provided from each of a plurality of vehicles into groups according to a preset condition, generates learning data in units of groups based on the learning battery information, and learns the learning data to construct the diagnostic model.