Recently, the sizing of renewable energy systems has attracted the attention of many scholars worldwide. Economically, the total annual cost of the system should be minimized by means of optimizing the combination of wind turbines, PV panels and the batteries system [
4]. That is to say, the aim of obtaining the right amount optimized of wind turbines, photovoltaic panels, and storage is to minimize the annual value of total cost for the entire system under certain constraints. Many scholars have studied in optimizing sizing of hybrid wind and photovoltaic systems. For example, on the base of a simple graphical construction, the optimal combination of a hybrid PV-wind system was investigated without taking account of the actual scale of the battery in [
5]. Taking the per-unit cost of the power produced and the total system cost into consideration, a method was present to optimize a hybrid wind/PV system in [
6]. The load considered in the optimization was constant. Ref. [
7] proposed a novel method to study the optimal sizing of stand-alone wind and photovoltaic systems using GA. Ref. [
8] discussed a general concept for search the optimal combination of hybrid wind and photovoltaic systems to implement with grid and without grid. An algorithm was applied in [
9] to find the the optimal combination of wind and photovoltaic systems without loss of power supply probabilities using load, solar insolation, and hourly mean of wind speed. Based on either integer or binary variables, mixed integer linear programming was used in [
10] to search the optimal plan of hybrid wind-PV system in rural area. The authors in [
11] used a multi-objective evolutionary and genetic algorithm to find the optimum design of hybrid systems minimizing three targets including the total cost of the service life of the installation, the unmet power demand and pollution. The technic-economic estimation of a stand-alone hybrid PV-wind system equipped with battery storage was studied considering cost of energy and system net present cost in [
12]. The analysis for the influence of wind turbines, photovoltaic panels, and battery storage sizing on the system’s economic performance were studied as well. Power generation and consumption missions of mixed energy systems were optimized simultaneously using a discrete-time MILP method to maximize the total profit in [
13]. The Hybrid Optimization Model for Electric Renewable (HOMER) software have been applied to simulation, modeling, optimum sizing, and economic analysis of renewable energy systems for micro-grid (e.g., [
14,
15,
16,
17,
18]). For example, in Ref. [
18], with grid and without grid, HOMER was used to study the practicability of application of renewable energy generations of a small hotel. The option of stand-alone mode was superior to grid-connected mode to meet the load economically. Ref. [
19] investigated methods and computer tools of optimal sizing for renewable energy systems, in which HOMER shows advantages in simulation over other available softwares. Sizing for renewable energy generation system of micro-grid is based on the dispatch of power. Jun Xiao [
20] proposed a approach for coordinated sizing of energy storage (ES) and diesel generators in an off-grid micro-grid on base of discrete Fourier transform (DFT). The DFT-based coordinated control method distributes balance energy between the two sections through frequency-time domain transform. In addition, the effectiveness of the suggested method is proved by a practical case. Raji Atia and Noboru Yamada [
21] proposed a novel method on base of MILP to find the optimization for a mixed renewable energy system equipped with a battery storage system in residential micro-grids in which the demand response of available controllable appliances is taken into account in the suggested optimization task with decreasing calculation burdens. A methodology to optimize combination of a self-governing mixed photovoltaic and wind system was proposed in [
22]. The goal of the proposed method is to achieve the configuration to ensure system reliability and a most optimal value of levelled cost of energy among a set of systems components. A trade-off analysis on the sizing of micro-grid components aiming at autonomous application in remote areas was studied in [
23]. The results show that there is a certain front of trade-off between the battery scale and other performance indexes, due to dynamic performance of the battery storage and the variation of photovoltaic output. The authors in [
24] carry out a study for a Seawater reverse osmosis desalination (SWRO) system on a small scale, which is powered by photovoltaic. The system uses hydraulic energy recovery on the base of a DC micro-grid idea and combines a short-term electric energy storage with a short-term hydraulic energy storage. The outcome of the experiment reveals that it is available to effectively run the SWRO system powered by photovoltaic. In addition, the SWRO system operates continuously and smoothly, which is to say the feed water pressure changes a little when the solar irradiation intensity varies rapidly and sharply owning to the existence of short-term storage. However, the literatures mentioned above only considered power dispatch within one single MG without considering power exchange with other MGs.
With multiple neighbouring MGs appearing in a local area, single micro-grids can be interconnected as a multiple micro-grid (MMG) system to seek for a better performance of energy [
25]. There also have been an increasing number of researches focusing on MMG operation approaches. A method based on memetic algorithm to create a micro-grid cluster for buildings obtaining local share of cooling energy is studied in [
26]. The same system was studied in [
27], using an augmented multi-objective particle swarm optimization to share thermal energy among micro-grids. The authors in [
28] studied optimal scheduling of MMG to make power sharing between MGs possible under interconnected operation mode. The literatures shows that cooperative operation between MGs can be beneficial in improving both economy and operation performance. In [
29], the authors used an interactive energy game matrix to investigate operational interaction of networked MGs. Ref. [
30] built a centralized control model for cooperative network of smart micro-grids with power exchanges between the neighbors, aiming at maximizing the total profit of all micro-grid operators. In [
31] economic operation of a multiple micro-grid system is investigated with a stochastic decision model, considering uncertainties in load and RESs. The authors in [
32] modeled the stochastic nature of renewable energy systems, loads, and prices according to their probability density functions to achieve the optimal operation of distribution networks in the notion of MMGs. Kou et al. [
33] studied that every MG tried to get the optimal operation cost working with other MGs keeping track of the power reference delivered by distribution network operator. However, it is hard to apply this method because MG operators have no intention to follow the scheduling of distribution network operator. Christos-Spyridon Karavas [
34] studied the plan and investigation of a decentralized energy management system for the self-governing polygeneration micro-grid topology to gives the feasibility to manage every unit of the micro-grid separately. Compared with the centralized solution, the decentralized solution presents promotions in economical and operational terms with nearly same technical performance. A multi-agent decentralized energy management system was investigated in [
35], in which the energy management problem was modelled using game theory. The Nash equilibrium was used to balance the possible diverging aims of the agents by maximizing their preferences.