Muhammad Imran Munawar , Imtiaz Alam, Muhammad Sharjeel Ali

An efficient energy prediction model for solar energy power system using Artificial Intelligence technique

Introduction

An efficient energy prediction model for solar energy power system using artificial intelligence technique. This paper proposes an efficient AI model for solar energy prediction using short weather data. Combining LR & KBNN, it achieves up to 99% accuracy, enhancing energy management.

Abstract

Prediction of Solar power generation plays an important role to improve the efficiency of economic dispatch function and reduce the dependence on fossil fuels and help in the energy management system. For time series solar energy prediction multiple models were introduced but these model trains are based on yearly historical data. A big data collection containing many missing values makes these model training more complicated that’s why In this paper, an efficient energy prediction model is proposed for the prediction of time series solar energy based on short predicted weather training data. Two complimentary models are based on linear regression and a knowledge based neural network is exploited to predict future solar power, with offline training. The LR is structured under the direction of the proposed input method parameter selection and used when training data is enough. KBNN is used for existing advantages predictive models are also very important when training data is not enough. 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 behavior. Additionally, the results demonstrate the effectiveness of LR showing a correlation coefficient (R2) is 98% with a root mean square error of 45 and KBNN shows a correlation coefficient (R2) is 99% with a root mean square error of 44 in providing a reliable version, The results additionally show the functionality of LR and KBNN in imparting a dependable version, especially when the short training dataset is available. 

Review

The paper, titled "An efficient energy prediction model for solar energy power system using Artificial Intelligence technique," addresses a critical challenge in renewable energy management: accurate solar power generation prediction. The abstract clearly articulates the importance of this task for improving economic dispatch, reducing fossil fuel dependence, and enhancing energy management systems. The authors identify a common limitation of existing models, which often rely on yearly historical data plagued by missing values, making training complex. To overcome this, the paper proposes an efficient prediction model specifically designed for time series solar energy, leveraging short predicted weather training data. This focus on short-term data to potentially simplify the training process and improve real-world applicability is a commendable objective. The core of the proposed solution involves two complementary models: Linear Regression (LR) and a Knowledge-Based Neural Network (KBNN), both trained offline. The abstract states that LR is utilized when sufficient training data is available, structured under a novel input method parameter selection, while KBNN is deployed when data scarcity is an issue, capitalizing on its advantages for limited datasets. The reported test findings, using real datasets, suggest promising performance, with the LR model achieving an R2 of 98% and RMSE of 45 for linear data, and the KBNN model showing an R2 of 99% and RMSE of 44 for non-linear behavior. These results, particularly the high correlation coefficients, indicate the models' effectiveness in providing a reliable version, especially highlighted for scenarios with short training datasets. While the abstract presents a compelling case for the proposed models, a more detailed understanding of several aspects would strengthen the paper's claims. For instance, the specifics of the "knowledge-based neural network" and the "input method parameter selection" for LR are not elaborated, which are crucial for assessing the novelty and technical depth. Furthermore, although the paper states its aim to mitigate issues with "big data collection containing many missing values," it's not explicitly clear *how* the proposed approach, beyond simply focusing on short data, directly handles or bypasses these missing values. A more direct comparison with other established state-of-the-art time series prediction models, especially those operating on short-term data or designed for real-time applications, would significantly validate the "efficiency" claim. Future work could also explore the computational overhead of this dual-model approach and its robustness to varying weather conditions and geographical locations. Despite these points, the initial results suggest a potentially valuable contribution to solar energy forecasting.

Full Text

Comments

You need to be logged in to post a comment.