Abstract
This article focuses on the safe operation of lithium battery energy storage power stations and develops a data monitoring and safety warning platform for energy storage systems. The background, architecture, implementation methods, and main functions of the platform development are introduced in sequence. This platform significantly improves the safety of energy storage stations by implementing active safety monitoring and early warning, which is of great significance for the large-scale application and promotion of lithium battery energy storage stations.
Introduction
As the world moves toward "dual carbon" goals, lithium-ion energy storage systems (ESS) have become critical for balancing grid fluctuations from renewable energy. However, these systems face a significant hurdle: safety. Lithium-ion batteries are prone to combustion and explosion, often with little warning.
A major gap exists in current safety protocols. Most measures are concentrated either in the design and construction stage—which cannot account for aging—or the fire protection stage, which only acts after an accident has already begun. There is a dire need for an "active" safety system that monitors battery health during operation to identify early-stage failures and prevent accidents before they occur.
System Architecture
The proposed intelligent operation and maintenance system is built upon a multi-layered big data architecture designed to handle massive telemetry from energy storage stations.
- Data Acquisition Layer: The system pulls second-level data from the Battery Management System (BMS) and Power Conversion System (PCS). This data is resampled at intervals (10s, 1m, 5m) to balance granularity with processing efficiency.
- Platform Layer (Big Data System): Utilizes an internet-scale data processing architecture for massive storage and parallel analysis. Raw data is retained for one year, while processed algorithm results are stored for three years.
- Core Algorithm Engine: This serves as the platform's "brain," performing health evaluations, inconsistency analysis (identifying outliers in voltage and temperature), remaining life prediction, and abnormal failure analysis.
- Advanced Application Layer: A user interface providing health scores, risk "heat maps," and a fault expert diagnosis module to suggest specific maintenance actions.
Key Innovations
The "External-Internal" Diagnostic Link
A core conceptual move in this platform is establishing a diagnostic mechanism that maps "external features" (observable data points like voltage variance) to "internal mechanisms" (chemical degradation paths). This allows the system to diagnose what is happening inside a sealed battery cell without requiring intrusive physical sensors.
Closed-Loop Maintenance
The platform does more than just alert; it integrates with a comprehensive "Work Order" system. Using an expert knowledge base, the system suggests a cause for detected anomalies and allows operators to issue maintenance orders directly from the diagnostic interface, ensuring that warnings lead to timely physical repairs.
Results and Discussion
The platform has been successfully implemented, demonstrating several key functional capabilities:
- Multi-level Messaging: The system generates warnings based on a multi-dimensional health assessment, allowing for tiered responses to different risk levels.
- Risk Identification: The platform effectively identifies three distinct types of battery safety risk sources.
- Operational Efficiency: By utilizing existing BMS/PCS data channels, the platform improves safety without increasing hardware costs or complexity.
- Visualization: Health scores are presented as curve graphs, enabling operators to monitor long-term health trends rather than just reacting to snapshots.
Limitations and Future Work
While the architecture is robust, certain areas require further exploration. The current documentation focuses on functional capabilities rather than quantitative performance metrics, such as false-positive rates or the specific lead time provided before a predicted failure. Additionally, while the system is built for "big data," specific computational scaling limits for monitoring thousands of units simultaneously remain to be detailed. Future research will focus on refining the diagnostic accuracy of the "external features" model by studying the internal mechanisms of lithium iron phosphate batteries under various aging paths.
Summary updated