Abstract:
With the development of modern distribution networks, user-side energy storage has become a crucial component. The converter of energy storage devices serves as the interface between energy storage and the grid, with its switches being components prone to high failure rates, particularly open-circuit faults which are concealed. This paper proposes a fault diagnosis method for open-circuit faults in switches of two-level converters during charging and discharging in distribution network user-side environments with complex power quality. Based on research into the fault mechanism, a diagnostic approach combining an extended Kalman filter (EKF) and a broad learning system (BLS) is introduced. Simulations and experiments demonstrate 100% accuracy in diagnosing and locating single-tube and double-tube open-circuit faults under common disturbances such as load fluctuations, system harmonics, and frequency deviations, providing effective support for user-side energy storage system monitoring.
1. Introduction
Energy storage technology plays a pivotal role in modern power systems, particularly on the user side, where it helps balance supply and demand, enhance grid resilience, and improve power quality. However, the converters used in energy storage systems are susceptible to various faults, among which open-circuit faults in switches are particularly concealed and challenging to diagnose. This paper focuses on addressing this issue.
2. Literature Review
Non-intrusive fault diagnosis methods can be broadly classified into three categories: model-based methods, signal processing-based methods, and data-driven methods. Model-based methods establish parametric equations based on the converter’s circuit topology and identify faults through parameter estimation or state estimation. Signal processing-based methods analyze fault features by processing signals from the converter. Data-driven methods leverage large amounts of data to train models that can identify faults.
| Method Type | Description | Advantages | Disadvantages |
|---|---|---|---|
| Model-Based | Establishes parametric models to identify faults through parameter or state estimation | High accuracy when parameters are accurate | Sensitive to parameter changes |
| Signal Processing-Based | Analyzes fault features by processing signals | Can handle complex signals | May require extensive signal processing |
| Data-Driven | Uses data to train models for fault identification | Adaptable to various fault scenarios | Requires large amounts of data and computational resources |
3. Fault Diagnosis Method for Two-Level Converters in Energy Storage Systems
3.1 Fault Mechanism Analysis
In energy storage systems, two-level converters are commonly used. During charging and discharging, open-circuit faults in switches can significantly affect the converter’s output current. This paper analyzes the impact of various disturbances, such as load changes, harmonics, and frequency deviations, on fault currents.
3.2 Proposed Fault Diagnosis Method
The proposed method combines an extended Kalman filter (EKF) and a broad learning system (BLS). The EKF is used to dynamically extract fault harmonic features, while the BLS is used to identify features and diagnose faults.
3.2.1 Dynamic Harmonic Feature Extraction using EKF
The EKF estimates the amplitudes and frequencies of harmonics in the converter’s output current. The state variables and state transition matrix are defined based on the harmonic model. By updating the state estimates, the EKF can track changes in harmonic features caused by faults.
3.2.2 Fault Identification and Location using BLS
The BLS is a type of neural network that improves upon traditional deep learning methods in terms of training speed and model simplicity. In this paper, the BLS is used to identify fault features extracted by the EKF and diagnose faults.
The input to the BLS includes the harmonic features extracted by the EKF, and the output is the fault diagnosis result. The BLS model is trained using a dataset constructed from simulated fault scenarios.
4. Simulation and Experimental Results
4.1 Simulation Setup
Simulink was used to establish a grid-connected model of the energy storage PCS (Power Conversion System) and test the effectiveness of the proposed method under various conditions. The converter’s current sampling rate was set to 10 kHz, with filter capacitors and inductors of 50 uF and 5 mH, respectively. The converter was controlled using PQ control with proportional and integral coefficients of 2000 and 10, respectively.
4.2 Simulation Results
The proposed method was tested under various fault scenarios, including different types of open-circuit faults, load changes, harmonic interference, and frequency deviations. The results demonstrate that the proposed method can accurately diagnose and locate faults within 0.035 seconds, with 100% accuracy.
| Fault Type | Charging/Discharging State | Load Disturbance | Harmonic Interference | Diagnosis Accuracy |
|---|---|---|---|---|
| Type I | Discharging | Present | Present | 100% |
| Type II | Discharging | Present | Present | 100% |
| Type III | Discharging | Present | Present | 100% |
| Type IV | Discharging | Present | Present | 100% |
| Type I | Charging | Present | Present | 100% |
| Type II | Charging | Present | Present | 100% |
5. Conclusion
This paper proposes a fault diagnosis method for open-circuit faults in switches of two-level converters in user-side energy storage systems. The method combines an extended Kalman filter and a broad learning system to dynamically extract fault harmonic features and diagnose faults. Simulations and experiments demonstrate the effectiveness and accuracy of the proposed method, providing valuable support for user-side energy storage system monitoring and maintenance.
Equation for Harmonic Feature Extraction:
xk+1=Fxk+wk
Where xk is the state vector containing harmonic amplitudes and frequencies, F is the state transition matrix, and wk is the process noise.
The fault diagnosis results when the converter is operating in an environment with harmonics. The proposed method accurately identifies and locates the fault.
In conclusion, the proposed fault diagnosis method for open-circuit faults in switches of two-level converters in user-side energy storage systems demonstrates high accuracy and robustness under various conditions. This method has significant potential for improving the reliability and safety of energy storage systems.