Abstract:
Science is looking for ways to slow down worldwide warming and lessen the
effects of long-term fossil fuel use in the production of electricity. In this regard, one of
the best answers is to deploy and promote renewable energy in many ways. A
significant obstacle faces to run power plants through the economic dispatch function
of connected electric networks. Analysis of Advancement in solar power generation is
crucial for increasing the effectiveness of economic dispatch, reducing reliance on
fossil fuels, and supporting energy management systems.
To forecasting time series solar energy based on upcoming meteorological conditions,
two complementing models are proposed in this work. With offline training, a model
based on linear regression (LR) and a knowledge-based neural network (KBNN) is
used to forecast solar power. Under the guidance of the specified input parameter
selection approach, LR is constructed after there is sufficient training data. KBNN is
utilized to make use of the existing prediction models when there is a lack of training
data. A KBNN model can be a useful technique to increase the predicted accuracy
produced by any models used for brief training data, according to test findings using
real data sets. An LR model can deal effectively with linear data, but a KBNN model
can cope effectively with nonlinear behaviour. Additionally, the results demonstrate the
effectiveness of LR shows correlation coefficient (R2) is 98% with root mean square
error 45 and KBNN shows correlation coefficient (R2) is 99% with root mean square
error 44 in providing a reliable version, particularly when a small training set is
available. The results further demonstrate that KBNN can create a reliable model even
there is not enough data for training.
