A hybrid approach using AHP–TOPSIS methods for ranking of soft computing techniques based on their attributes for prediction of solar radiation

Abstract

Recent developments in solar equipment's have motivated researchers to formulate an accurate measurement system for solar radiation under varying environment circumstances that could prove to be economical and viable. To furnish precise estimators of solar radiation, multiple models of different background are employed which are capable of solving complex non-linear data collection and processing problems. These models are immensely proficient in predicting solar radiation by means of numerous algorithms for different functions. Current research compares the prediction capability of several prediction models on various criteria's such as accuracy, cost, time and skill requirement. The present study is accomplished in New Delhi, India using hybrid combination of two Multi Criteria Estimators methods. Weights were calculated by Analytical Hierarchical Process and assigned to each performance attribute. Accuracy is the most significant attribute as its percentage contribution is the maximum (48.28%) followed by skill requirement (31.38%), time (14.41%), and cost (5.9%). Furthermore, the research then ranks these models on the basis of their attributes with the aid of Technique for Order Preference by Similarity to Ideal Solution. Among all these models, Artificial Neural Network model was ranked as the best model for its application in the field of solar energy, followed very closely by Support Vector Machine model while Response Surface Methodology came out to be least favourite.

Keywords

• Optimization
• Artificial intelligence
• Energy systems
• Hybrid algorithms
• Energy environment
• Nature inspired optimization