Читать книгу Microgrid Technologies - Группа авторов - Страница 32

Energy management in a micro-grid is addressed by applying different approaches. All the approaches have the common aim to optimize the MG operation. Some methods are supported on linear or non-linear programming such as in Ref. [49] where a MILP is used to optimize the system. The cost function solution is obtained by linear programming, which is based on GAMS (general algebraic modeling system).

In MG, the energy obtained from RES, load demands and market rate of energy are considered as stochastic variables because of uncertainty. So, it is good to use stochastic modeling to analyze any energy management strategies. Generally, researches have gone through stochastic algorithms or metaheuristic algorithms to solve problems of optimum power balancing in MG. According to Ref. [50], in stochastic programming, the data is stochastic, and the result or solution is dependent on the collection of variables that arbitrarily generated. Recently researchers are considering that the management system of energy in micro-grids stand on the implementation of advanced technology such as collaboration of a variety of optimization technologies or improving classical algorithms, to get the most suitable solution of a problem for MSE in MG.

In Ref. [51] a multi-objective genetic algorithm was applied to a standalone system having an internal combustion engine and gas turbine with the PV module. In Ref. [52] the author represented a dynamic programming technique for a standalone micro-grid. The micro-grid is consisting of DG, PV panel and battery. Here the constraints of the problem are supply–load balancing and the capability of the supply generators. The main goal is to minimize the functioning cost and emission.

The authors in Ref. [53] represented a relative analysis of the various objectives of the optimization methods for MSE of standalone micro-grids. The comparison is based on linear programming and genetic algorithms. The result was found out that the controllable power consumption can reduce the cost with renewable energies.

In Ref. [54], the weight factor has been analyzed to increase the ability of PSO (Particle Swarm Optimization) technique and to balance the convergence. Even though a large amount of the internal weighted factor can create a limitation to the algorithm to discover the best possible solution locally, the convergence can be achieved at a prolonged rate. The author has recommended enhancing the PSO technique which adjusts and decreases the weight factor linearly through iterations to solve the problem. Thus, the enhanced PSO can get the optimum solution universally without fixing at local minima. This method is utilized in an MSE for hybrid power sources. In some highly developed means, the chaotic sequence is used in place of arbitrary numbers to optimize the action.

The authors in Ref. [45] have suggested CBA (Chaotic Bat Algorithm) optimize the financial send-off of the system under study. To get better performance of the BA (Bat Algorithm) to get the universal optimal solution, the chaotic sequences is applied in the primary BA. Authors of Ref. [44] have suggested Ant Colony Optimization technology for actual time functioning of MSE with a new Multi-Layer technique (MACO) in standalone micro-grid. The MACO is an enhanced version of the basic Ant Colony Optimization (ACO) algorithm. In this algorithm, the numerical quantity of levels is same as that of variables in designing on the problem and quantity of nodes in every level is equal to the numeric quantity of satisfactory values of every variable. The author investigated an MSE in Ref. [55] using a two-layer predictive control, and the degradation cost of SSE is taken as the main consideration.In Ref. [56] the author has suggested a new technology to optimize the performance of fireworks algorithm (FA) as a novel crossbreed Multi-goal-based FA and Gravitational Search Operator (MFAGSO) to resolve the non-linear trouble with several variables. There are also multiple hybrid constraints. This recommended algorithm uses gravitational explorer to lead the flash into the collection area to swap position information with optimal solutions to reach the best results.

Authors of Ref. [57] have explained an advanced algorithm called Improved Artificial Bee Colony algorithm (IABC) to get an optimized result in a hybrid grid-connected micro-grid. The author has rectified the basic ABC by generating the scrutinize bee using Gravitational Search Operator, which optimises the finding accuracy, so the universal best possible solution can be enhanced. The authors in ref. [58] have suggested a new algorithm such as Enhanced Bee Colony Optimization (EBCO), which gives a better performance of MSE for MGs with verity RESs and several SSE. EBCO operator is unlike the classical BCO, as it has the self- adaptation revulsion factor in the bee swarm, for getting the better performance of each bee swarm and that is why the finding accuracy enhances effectively in more dimensional problems.

The authors in Ref. [59] described an MSE applying Artificial Bee Colony (ABC) algorithm for an isolated MG. In Ref. [60] it proposed an MSE, which is based on utilization of fuzzy logic controller in a micro-grid that employs around 25 sets of laws. The objective function is to lower the deviation of power with maintaining battery SoC. In Ref. [61] the MSE is for an interconnected system of micro-grid using an advanced algorithm based on fuzzy logic called Mamdani algorithm. The optimization is done with the scheme combination of fuzzy logic and genetic algorithms. The authors in Ref. [62] provide an algorithm for MSE based on game theory to maximize the gain available during consumption of energy. Ref. [63] represents an adaptable neural fuzzy interference system, with the help a predictor of the echo state network. In Ref. [64] the author proposed a new approach, a Stackelberg game approach for managing the flow of energy in MG. The author of Ref. [65] suggested an MSE model for a smart micro-grid using game theory, where maximization of profit to the overall cost and satisfactory power utilization is selected as the strategy. It is a distributed energy management model.

Figure 1.8 summarizes the energy management technologies for micro-grids. Among them, some methodologies are classical techniques such as MILP, linear programming and non-linear programming. These programmings may be an excellent move towards optimization depending on the goal function and limitations. But the artificial intelligence (AI) processes are dedicated to approaches towards the situation while the classical methods come into unsatisfactory results.

Figure 1.8 Energy management methodology.