Intra ELM variants ensemble based model to predict energy performance in residential buildings. (December 2018)

• Record Type: Journal Article
• Authors: Kumar, Sachin; Pal, Saibal K.; Singh, Ram Pal
• Is Part Of: Sustainable energy, grids and networks. Volume 16(2018)
• Page Start: 177
• Page End: 187
• Publication Date: 2018-12
• Subjects: B-ELM -- Cooling load -- Energy performance prediction -- ELM -- Heating load -- OSELM -- Ensemble model; Renewable energy sources -- Periodicals; Smart power grids -- Periodicals; Electric power systems -- Periodicals; 333.794
• DOI: 10.1016/j.segan.2018.07.001
• Languages: English
• ISSNs: 2352-4677

Abstract

Abstract: Optimization of energy performance of the buildings has been a great interest of research. Energy performance monitoring and maintenance depend on the several parameters such as temperature, humidity, sunlight, roof area, wall area and mostly on heating ventilation and air conditioning(HVAC). Each building has its own patterns and settings obtained through monitoring which depends on many parameters such as users activities, environmental related attributes and building design structural parameters etc. In this respect, an estimation of energy load of the building in real-time efficiently for optimization becomes an important task for cost-effective energy management. The paper proposes an Intra ensemble model based on variants of emerging machine learning approach which includes extreme learning machine(ELM), Online Sequential ELM(OSELM) and Bidirectional ELM(B-ELM). The energy performance estimation requires the model to be real-time and efficient. This asks for use of highly correlated parameters and a very efficient model. For this, OSELM based model for real-time prediction of energy performance has been used. ELM variants are used because of their fast computation and efficiency of prediction over conventional machine learning models. The proposed model has been compared with few state of art methods on accuracy and efficiency criteria and proposed models outperformed existing methods.