Scientists in India have proposed using a multilayer neural network to find line-to-ground, line-to-line, and bypass
PV Magazine International 11:02 am on May 29, 2024
A multi-layer neural network approach has been proposed for PV string fault detection, with an accuracy of 98.76% in detecting L-G and L-L faults among others. It requires one current sensor per string, is capable of handling varied datasets, and can classify complex relationships.
- Multi-layer Neural Network Approach: Proposed technique with 98.76% accuracy in PV fault detection.
- Sensor Requirement: A single current sensor is needed per string for implementation.
- Data Variability Handling: Tested across multiple datasets with diverse conditions (temperature, irradiance, and power output).
- Complexity Management: Manages complex relationships using a hierarchical learning representation.
- Comparison to Existing Techniques: Higher than previous techniques such as PNN (96.5%), RBF (92.1%), and CNN (90%).
Category selection: Solar
support: "Solar" category as it discusses a novel technique for detecting faults in photovoltaic systems, which is directly related to solar energy technology and innovation.
https://www.pv-magazine.com/2024/05/29/pv-string-fault-detection-technique-based-on-multi-layer-neural-network/
< Previous Story - Next Story >