Design of an ecological growth room using internal temperature modeling in neural network interface

In this research, an ecological growth room made from recycled materials such as polyethylene terephthalate (PET) was built. The ecological growth room was compared with a growth room made from traditional materials (cement, blocks, etc.). The research was developed setting fifteen temperature sensors in both buildings from which samples were taken every five minutes for a period of three months. The results showed that the ecological building had better temperature behavior than that of the traditional materials; reducing energy costs for heating and cooling environment and advantage of being portable. As part of the study a predictor based in artificial neural networks (ANN) multi-layer perceptron (MLP) was developed and trained by means of the Levenbergh-Marquardt back propagation algorithm. The predictor inputs are external temperature, external relative humidity, wind velocity and solar radiation, and the outputs are the internal temperature of the ecological construction. Different neural network models were tested and the best behavior was the structure 8-8-1 with a coefficient of determination R 2 = 0.913; these model can be used for intelligent control strategies. The goal of this study was to develop an ecological growth room made from recyclable materials to obtain favorable thermal conditions for plant growth.