چكيده به لاتين
Renewable energy sources such as solar and wind energies are dependent on environmental and weather conditions. Therefore, the simultaneous use of several energy sources in the form of a hybrid energy system can fulfill energy demands more efficiently and reliably. Sizing the system equipment and also specifying its operation, as the prerequisites for using an energy system, are addressed in this thesis. Thus, optimization and energy management of a hybrid energy system consisting of photovoltaic and thermal collectors, wind turbines, batteries, hot water storage tanks, a water heater, reverse osmosis desalination system (RO), electric chiller, and CHP units that provide electricity, heating, cooling, and freshwater demands, is considered. The mentioned energy system was studied for four residential buildings in Negin Island (near the port of Bushehr). The effect of using excess thermal energy for preheating the RO feedwater (which reduces its electricity consumption) is also investigated. In order to determine the system configuration, a technique was proposed to model and optimize the system equipment by employing an artificial neural network (ANN) and particle swarm optimization (PSO) algorithm. Besides, for determining the operation of the system equipment, a new method based on reinforcement learning was developed for system energy management. In this method, based on the state of the energy system (the stored electrical and thermal energies, the loads, and the amount of renewable energy that the system receives), an appropriate management decision is made for the operation of the equipment. Results showed that in optimizing the system equipment, the use of ANN reduced the run time by about 10%. The use of excess heat to warm the inlet water of RO reduced the annual electricity consumption of desalination by 4.95%. Moreover, the proposed method of energy management was compared with conventional methods of energy management. It was observed that the presented method reduces 12.4% the operational cost and 13.1% the CO2 production compared to the load following (LF) method and it has 38.59% less operational cost and 48.5% lower CO2 emissions compared to the energy management with full state of charge (FSC) approach.
