Battery State Anomaly Detection With Selective Cloud Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current battery management systems for electric vehicles face inefficiencies and increased costs due to the need to analyze vast amounts of battery data in real-time, particularly when high-performance GPUs or NPUs are used, and the transmission of this data to cloud servers, which degrades efficiency and incurs significant infrastructure and communication costs.
Innovation Solution
A battery management apparatus and system that employs a first controller to obtain state data, a second controller using machine learning (specifically an LSTM algorithm) to predict battery states, and a communication unit to transmit only compressed data when an anomaly is detected, thereby reducing unnecessary data transmission and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-performance GPU or NPU is mounted on the battery management apparatus to precisely analyze the state of the battery in real time, then the analysis precision is improved, but the device complexity and cost increase
Solution Approach 1:
The patent extracts and separates the heavy computational tasks (machine learning model execution, anomaly detection algorithms) from the battery management apparatus to external servers. The BMS only performs lightweight preprocessing and transmits necessary data, while complex analysis is performed remotely, thus improving analysis precision without increasing local device complexity.
Solution Approach 2:
The patent introduces communication units and data transmission protocols as intermediaries between the battery management apparatus and external servers. This intermediary layer enables precise analysis by leveraging external computational resources while keeping the BMS itself relatively simple in structure.
2Measurement precision
If the entire battery data measured by the battery management apparatus is transmitted to a cloud server for analysis, then the analysis precision is improved, but the communication cost and infra installation cost increase
Solution Approach 1:
The patent extracts only the essential and anomaly-related data for transmission to cloud servers, rather than transmitting entire battery datasets. This selective data extraction maintains analysis precision by focusing on critical information while dramatically reducing communication costs and data transmission volume.
Solution Approach 2:
The patent applies different data processing qualities at different locations: lightweight preprocessing and anomaly detection are performed locally at the BMS, while comprehensive analysis is performed remotely at servers. This local-quality differentiation optimizes resource allocation and reduces unnecessary data transmission.
3Measurement precision
If the entire battery data measured by the battery management apparatus is transmitted to a cloud server for analysis, then the analysis precision is improved, but the processing efficiency is degraded
Solution Approach 1:
The patent extracts and transmits only essential real-time data and anomaly-related information to cloud servers, enabling timely analysis without the overhead of processing complete historical datasets. This extraction approach maintains high processing efficiency while achieving precise anomaly detection.
Solution Approach 2:
The patent performs preliminary preprocessing, filtering, and anomaly detection at the battery management apparatus before data transmission. This preliminary action reduces the data volume requiring remote processing and enables faster overall processing efficiency while maintaining analysis precision.
Data Source
AI summary
A battery management apparatus includes a first controller obtaining state data comprising a measurement value corresponding to a state of a battery. The battery management apparatus also includes a second controller generating prediction data for predicting the state of the battery by applying at least a part of the state data to machine learning. The second controller determines the state of the battery by comparing the prediction data with the state data. The battery management apparatus further includes a communication unit transmitting the state data to a server based on a result of determining the state of the battery.