Elaboration d’un système pour le monitoring et la prédiction de la consommation électrique avec les méthodes Machine Learning et Deep Learning.

Abstract #

Electrical energy is a very important factor for the evolution of Morocco, hence the interest of

monitoring systems for controlling electricity consumption. This concept remains new in

Moroccan companies, this framework is a part of our project of graduation studies, done within

the company “Syntechnology”, with the aim of designing and implementing a system of

supervision which is based on monitoring and forecasting of electricity consumption. During

this internship we identified the following objectives:

ü Diagnose existing, identify the need of society.

ü Model a monitoring system for monitoring and predicting the consumption of electrical

energy.

ü Implement a monitoring system.

To achieve this work, we first began by clarifying the needs of society at the end of the system

for the monitoring and the prediction of the electricity consumption, then we proceeded to its

conception and its implementation. To improve the performance of the prediction we have

developed a benchmark between several models of the Learning machine, time series and deep

learning, we have created hybrid models such as Arima-SVR, ANFIS. The results of the

evaluation show that the ANFIS model exceeds all the models developed which makes it the

best model of our study. The system we propose contains a billing module which is based on

tariffs of the ranges specified by the admin. It also contains a module for forecasting electricity

consumption. The implementation of the system was done with the Python language for the

Application Backend and Html -CSS for the interfaces, the Plotly library for the visualization

of Machine Learning and Deep Learning data and models, and the hybrid models for the

forecast of electricity consumption.