Abstract:
The progress in technology is increasing the use of machines and that increases the demand for
electrical energy. The Demand Response (DR) programs in Demand Side Management (DSM) are
used in Smart Grids (SG) to reduce load consumption. The monitoring tools in DSM give the
ability to make a decision and control the load in peak demand. This thesis consists of two phases,
Message Queue Telemetry Transport (MQTT) Implementation and DSM implementation based
on DR. In the first phase we proposed the IoT-oriented MQTT protocol as the communication
network between load appliances and DSM devices and a medium to monitor the appliances for
DSM. The MQTT protocol is implemented in a Smart Home (SH) environment where
consumption values of major loads were sent to MQTT-based Home Gateway (HG). The
performance of the MQTT protocol was analyzed in a real-time environment. We proposed the
HG equipped with an Artificial Neural Network (ANN) model to estimate the total load
consumption for an SH. The activation function for the ANN model was selected by Trial-andError learning and the model was trained with the real-time dataset. The performance and accuracy
of the ANN model in terms of estimation were evaluated by comparing the model with the Support
Vector Regression (SVR) model. In the second stage, we propose a simple Load Scheduling (LS)
algorithm that will work as a DR program in DSM. The same network topology used in the first
phase was adopted in the second phase. The proposed MQTT protocol provides the
communication platform to send the data wirelessly from the first to the second stage and vice
versa. The proposed LS algorithm was tested with the real-time data sent from the first stage HG.
The LS algorithm in terms of TOU was also implemented and tested with the real-time dataset.
The research work proves MQTT can become an efficient communication tool to use in DR for
DSM
