Photovoltaic backboard fault diagnosis method based on layered Softmax

Abstract

The invention discloses a photovoltaic backboard fault diagnosis method based on layered Softmax, and the method comprises the steps: firstly obtaining electrical characteristics, working states and duration of a photovoltaic backboard under normal conditions and various types of faults, and carrying out maximum and minimum normalization of data; then, building a BP neural network model, buildinga Huffman tree according to the total duration of the photovoltaic backboard under all working conditions, connecting an output layer of the neural network with a root node of the Huffman tree, and building a layered Softmax; and taking the preprocessed electrical characteristics as a training sample, training a BP neural network and layering Softmax, and obtaining a fault diagnosis model; detecting four electrical characteristic values of the photovoltaic backboard in real time, taking the four electrical characteristic values as model input, and judging whether the photovoltaic backboard hasa fault or not and judging the type of the fault. The calculation amount of the model is reduced, the fault detection efficiency of the photovoltaic power station is improved, and the operation and maintenance cost of the photovoltaic power station is reduced.

Description

technical field [0001] The invention relates to the field of solar power generation, in particular to a photovoltaic backplane fault diagnosis method based on layered Softmax. Background technique [0002] After nearly ten years of development of photovoltaic power generation, major breakthroughs have been made in aspects including photovoltaic cell efficiency improvement, MPPT algorithm research, inverter algorithm research, and grid-connected mode; Combining with each other and cooperating with energy management systems and other methods, the shortcomings of solar power generation fluctuations and instability are minimized, and the impact of photovoltaic power generation on grid connection is weakened. [0003] Nevertheless, the operation and maintenance of photovoltaic power plants has always been a difficult point in the industry. In particular, photovoltaic backplanes are prone to various failures due to long-term deployment outdoors and are affected by bad weather, wh...

AI Technical Summary

Problems solved by technology

When the output feature dimension of the classification model is large, the corresponding calculation amount will be large, resulting in slow training

Images