Photovoltaic day-ahead prediction method and system based on convolution transformer architecture, and equipment

Abstract

The invention discloses a photovoltaic day-ahead prediction method and system based on a convolution transformer architecture, and equipment.The method comprises the steps: obtaining meteorological historical data and photovoltaic system operation historical data, and taking the data as a data set; dividing the data set into a training set and a test set, extracting a plurality of different training sets from the original time sequence, and sequentially or randomly dividing the training set in the first year into a plurality of parts; performing irradiance filtering on the training set of the first year, wherein the data set after irradiance filtering is used for training machine learning models under different irradiances; calculating self-attention layers in the encoder and the decoder, and performing convolution operation by adopting a convolution kernel greater than 1 when calculating the previous output of the decoder and the output of the encoder; stacking the encoder and the decoder to obtain a photovoltaic power prediction model; and obtaining current meteorological data and photovoltaic system operation data, inputting the data into the photovoltaic power prediction model, and obtaining an output result of the photovoltaic power prediction model as a prediction result of photovoltaic day-ahead power.

Description

technical field [0001] The invention belongs to the field of photovoltaic power prediction, and relates to a photovoltaic day-ahead prediction method, system and equipment based on a convolution transformer architecture. Background technique [0002] Photovoltaic technology is becoming a major source of future electricity demand. A higher share of renewable energy technologies is essential to meet the needs of future new power system grids, but it also brings new grid operation challenges. Power companies need to predict the power of photovoltaic power generation in order to carry out power generation dispatching operations. Forecasting is a major enabler to ensure safe and economical PV grid integration, while creating links between many flexibility innovations at different levels of the power system to achieve synergies. Accurate PV power forecasting is an important and cost-effective energy management element, which also helps PV plants and aggregation systems to partic...

AI Technical Summary

Problems solved by technology

As mentioned earlier, the method based on the recurrent neural network cannot completely eliminate the problem of gradient disappearance and gradient explosion when facing long sequences, and the Transformer architecture can solve this problem, and the effect is better on long sequences, but the original Transformer architecture. The self-attention calculation method is not sensitive to local information, making the model vulnerable to outliers

Images