High Impedance fault detection classification using Neural Network in MATLAB
We will delve into the detection of high impedance faults in distribution feeders within a distribution system. High impedance faults occur when a conductor touches the ground or other objects, creating a potential fire hazard. Our focus will be on simulating these faults in MATLAB and employing neural networks for accurate fault detection.
Understanding High Impedance Faults:
High-impedance faults manifest when conductors come into contact with vegetation, concrete surfaces, or other high-resistance paths. These faults can lead to dangerous situations, including fire, and hence, efficient detection is crucial.
Simulation Model Overview:
We've developed a simulation model to test the detection of high-impedance faults in a distribution system. The model consists of a sending end and a receiving end, both operating at 33 kilovolts, connected by a feeder. A step-down transformer reduces the voltage to 11 kilovolts, and a subsequent step-up transformer increases it back to 33 kilovolts. Our goal is to create high-impedance faults in different phases (A, B, C) and employ a neural network to accurately detect and classify these faults.
Creating High Impedance Faults:
To mimic real-world scenarios, we utilize a fault creation model based on research. The fault can be introduced in any of the three phases, representing high impedance conditions. We then monitor the system's behavior during and after the fault occurrence.
Neural Network Training:
To enable fault detection, we collect data on sending end voltage and currents under normal and faulty conditions. The neural network is trained using this dataset, with corresponding targets indicating the faulted phase. The goal is to teach the neural network to classify faults accurately based on the input data.
Neural Network Implementation:
The trained neural network is integrated into the simulation model. It receives inputs for sending end voltage and currents, providing outputs that signify the fault status. Visualization tools, such as scopes, help observe the system's response to faults and the neural network's classification results.
Simulation Results:
We demonstrate the model's effectiveness by intentionally introducing faults in different phases. The neural network accurately detects and classifies high-impedance faults, allowing for swift identification of affected lines.
Conclusion:
This simulation model showcases the potential of neural networks in efficiently detecting and classifying high-impedance faults in distribution feeders. The integration of advanced technologies like neural networks contributes to enhancing the reliability and safety of distribution systems.
Comments