Greenhouse indoor temperature prediction based on extreme learning machines for resource-constrained control devices implementation

Abstract

In this paper we present an Extreme Learning Machine approach for a real problem of indoor temperature prediction in greenhouses. In this specific problem, the computational cost of the forecasting algorithm is capital, since it should be implemented in resource-constrained devices, typically an embedded controller. We show that the ELM algorithm is extremely fast, and obtains a reasonable performance in this problem, so it is a very good option for a real implementation of the temperature forecasting system in greenhouses. © 2011 Springer-Verlag Berlin Heidelberg.