Development of an Artificial Neural Network Model for Daily Electrical Energy Management

Abstract

Efficient monitoring and control of electrical energy do not only prevent fire out-breaks caused by electrical appliances, but can also reduce excessive billings and prevent electrical installations. Most Energy Management Systems (EMS) for remote controlling of electrical appliances rely mostly on sensors, data and GSM networks which are un-reliable or even un-available in most part of developing world, this makes them less reliable. Therefore, there is need for an intelligent system that can manage electrical consumption intelligently using user-appliance interactive pattern over a period of time for intelligent control of users’ appliances in his/her absence. The model parameters (number of neurons and training algorithms) that affects its performance were first investigated and adopted. The performance of the developed model was evaluated using Regression analysis (R) and Mean Square Error (MSE) using ANN and Simulink tool boxes in Matlab R2015b. A good model can be used for real time control when deployed. Also, Scale Conjugate Gradient (SCG) training algorithm should also be used because of its high performance for pattern recognition problems. This work will go a long way in efficiently controlling household electrical appliances in the absence of the users thereby preventing fire disasters caused by electrical appliances, reducing the tariffs of consumers while increasing lifespan of electrical installations.