MATLAB Implementation of ANFIS Based Fault Classification Location and Detection in Power System
Introduction to ANFIS
ANFIS is a powerful hybrid system that leverages the learning capabilities of neural networks alongside the reasoning capabilities of fuzzy inference systems. This combination makes ANFIS particularly suited for detecting and classifying faults in electrical networks, ensuring swift responses to issues that may disrupt power supply.
System Setup
In our model, we establish a power system comprising three buses, interconnected through transmission lines. Here are the key specifications:
Bus Configurations:
Bus 1 and Bus 2 are connected to a source feeder rated at 11 kV and 30 MVA.
A distribution line of 10 km connects Bus 1 to Bus 3.
We simulate various faults, including line-to-ground faults and double faults, across this setup to evaluate the system's performance.
Data Collection and Training Process
To train the ANFIS model effectively, we gather critical data from Bus 1, including:
RMS voltage
RMS current
Zero-sequence voltage
Zero-sequence current
This data is categorized based on different fault conditions, and corresponding binary outputs are established for each scenario. For example, a specific binary value indicates a fault type, which helps the system learn to classify and locate faults accurately.
Simulation and Results
Upon simulating the model, we collect data for both normal and fault conditions. The results demonstrate the system's ability to accurately classify and locate faults. For instance, when we introduce a fault, the model processes the incoming data and produces output indicating the fault type and location.
The system's accuracy is notable, achieving approximately 90-91% in classifying various fault types. This high level of precision underscores the effectiveness of the ANFIS approach in real-time fault management.
Conclusion
The implementation of ANFIS in power systems proves to be a robust solution for fault detection, classification, and location. By efficiently combining neural networks and fuzzy logic, this system enhances the reliability of electrical networks, ensuring timely responses to faults.
Comments