Prediction of the Energy Requirements in a Refrigeration Plant Using Advanced Simulation Techniques

In this work the efficiency of the Artificial Neural Network technology for the evaluation of the energy requirements in an actual refrigeration plant has been evaluated. A good prediction capability of the energy demand has been obtained by correlating the refrigeration load with macroscopic quantities (external air temperature and relative humidity, and time). This result has been achieved avoiding any detailed time-wasting analysis of the production plant, in terms of refrigeration demand from each department. Several net architectures have been set up and verified as regards their response to this strongly non-linear problem.