Evolving Resilient Back-Propagation Algorithm for Energy Efficiency Problem

Abstract

Energy efficiency is one of our most economical sources of new energy. When it comes to efficient building design, the computation of the heating load (HL) and cooling load (CL) is required to determine the specifications of the heating and cooling equipment. The objective of this paper is to model heating load and cooling load buildings using neural networks in order to predict HL load and CL load. Rprop with genetic algorithm was proposed to increase the global convergence capability of Rprop by modifying a corresponding weight. Comparison results show that Rprop with GA can successfully improve the global convergence capability of Rprop and achieve lower MSE than other perceptron training algorithms, such as Back-Propagation or original Rprop. In addition, the trained network has better generalization ability and stabilization performance.