研究目的
To address the immediate diagnosis and elimination of faults in photovoltaic systems to ensure stable operation by proposing a probabilistic neural network (PNN) as the fault classification tool.
研究成果
The probabilistic neural network (PNN) can achieve high accuracy classification for photovoltaic array faults, providing a favorable premise basis for the intelligent classification of faults. The method shows good classification effect both in simulation and field experiments, with an accuracy of 97%.
研究不足
The external field experimental data may have some noise due to environmental factors, affecting the classification accuracy, especially for multi-fault conditions.
1：Experimental Design and Method Selection:
Building a 4×3 PV array model using Matlab, proposing an efficient feature vector of five dimensions as the input of the fault diagnosis model, and using PNN for fault classification.
2：Sample Selection and Data Sources:
Simulating five fault states (normal, short-circuit fault, open-circuit fault, line-to-line fault, and multiple faults) and collecting 779 samples for each state.
3：List of Experimental Equipment and Materials:
Using Matlab for simulation and a built fault data acquisition device for field tests.
4：Experimental Procedures and Operational Workflow:
Normalizing the simulation data, dividing each type of data into training and test data, inputting into the PNN network for classification, and calculating classification accuracy.
5：Data Analysis Methods:
Using PNN for classification and analyzing the results to achieve high accuracy classification.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
