Technical, economic and environmental advantages of distributed generations especially renewable-based ones has increased the installation of these sources in many power systems. By the high penetration of distributed generations in power system, there would be some new challenges for system operation. Some these challenges are possibility of changing power flow direction on some of distribution feeders, non-predictable renewablesoutput and difficulty of the DG direct participation into electricity market. The challenges not only reduce motivation of DG owners for more investment but also the potential of DG can not be completely exploited and a number of new problems are created in power system operation. Virtual power plant has been introduced as a new concept that overcome the challenges and provide better exploiting DG merits. VPP concept provides a plan for coordinated operation and management of DG on a geographic wide area. Virtual power plant is a set of controllable distributed units that each unit can include consumers or independent producers generation equipment, controllable loads or energy storage systems. The VPP manager exploits units by using suitable communication infrastructure, management softwares and operation algorithms to attain the maximum benefits and the best system performance. For this purpose,designing the efficient operation algorithms has a special importance formaximizing the profit of virtual power plant and other objects. Accordingly the purpose of this thesis is to review the definitions and concepts of virtual power plant and explore various aspects of that , and also introducing some models, provided for virtual power plant operation optimization. Moreover, in this thesis a comprehensive model for optimized operatio of a large scale virtual power plant is proposed by integration and completion of theprevious models, In this model by simultaneously considering of electrical and thermal loads and sources, electrical and thermal energy storages, different kind of distributed generations and electrical vehicles and virtual power plant participate in both energy and reserve markets, the profit is maximized subjected to unit constraints and security constraint of system. The comprehensive proposed model is formulated in two deterministic and stochastic frames and is tested on 33-bus distribution network test and the results are discussed. In probabilistic programming the uncertainty in load forecasting and output power of solar resources are considered. Keywords: Virtual Power Plant, Profit Optimization, Probabilistic Programming.