Article Preview
Top1. Introduction
A microgrid is a collection of electrical loads and generation spread within a small geographical location. The generators in the microgrid may be micro turbines, fuel cells, reciprocating engines, or any number of alternate power sources (Hatziargyriou, Asano, Iravani, & Marnay, 2007). Countries such as USA, Germany, Greece, Japan has seen the deployment of microgrids and have availed the benefits of it (Barnes et al., 2007). The effective management of the microgrid to supply the required power without facing shortcomings and with a minimized economy is the problem which requires attention (Koutsopoulos & Tassiulas, 2012) (Khodayar, Barati, & Shahidehpour, 2012). The Energy Management makes decisions regarding the best use of the generators for producing electric power and heat, based upon the heat requirements of the local equipment, the weather conditions, the price of electric power, the cost of fuel and many other considerations. Such type of study to understand various types of long time load forecasting on a remote and isolated geographical area was performed by (Panda, Jagadev & Mohanty, 2018) considering the primary factors such as demand side factor, price and income effect.
Extensive literature survey on microgrid energy management has given an idea about the different methodologies employed so far to solve these problems. The demand of electrical power is rising at a very fast rate and to accommodate these high raising demands, electrical power sectors are growing up significantly. In India most of the electrical energy is derived from fossil fuel-based power plants. In this framework, smart grids and microgrids are the key in the near future where a decentralization of energy generation is expected. An advantage of these types of grids is the efficient balancing between energy generation, storage, and consumption. This reduces the need for centralized communication, enables autonomous operations of increasingly smaller sections of the distribution grid and decreases the losses by distant distribution. From the point of view of microgrid energy management, economic scheduling for generation devices, storage systems and loads is a crucial problem. Performance of an optimization process is necessary to minimize the operating costs while several operational constraints are taken into account.
The cost of fuel is a major part of the running expenses of these small power stations and this can be diminished if the efficiency of the plants is improved. Hence, the minimization of production cost has attracted a great deal of attention from the power engineers. Mostly conventional classical based dual decomposition optimization techniques were employed to solve the basic energy management based microgrid problems. (Marmolejo, Litvinchev, Aceves & Ramirez, 2011) used general cross decomposition method for unit commitment problem of thermal generation. This method was found to be less time consuming as it exploits the problem structure and decomposes a dual sub problem to a primal sub problem. An adaptive random search method which controls the noise level while exploration and exploitation of the search space was implemented by (Marmolejo, Velasco & Selley, 2017) for short term generation scheduling considering various network constraints.
However, these techniques resulted in solutions with reduced accuracy and hence, huge revenue loss over time. Moreover, in the above mentioned conventional classical optimization techniques, the numbers of control parameters (which control the performance of the algorithm) are large. Therefore, a time consuming tuning procedure of the control parameter is required to attain an optimized solution of the production cost problems.
Various loads and distributed energy resources (DERs) are controlled and maintained by the microgrid energy manager (MGEM). All the DERs and loads have dedicated local controllers that coordinate with the MGEM for the timely operation of resources in a distributed fashion. These disciplined and distributed fashion of functioning of microgrids face the challenge when the uncertainty and stochastic nature of RES comes into view. An ED problem with unit commitment was done for IEEE 24 bus system, IEEE 118 bus system and 104 bus electric energy system of mainland Spain using fat tail model by (Marmolejo & Rodriguez, 2015). This estimation of the stable random variables was done by regression, quartile and maximum likelihood methods and then choosing the best fit of the tails.