With the sustainable development of the energy industry, the excessive exploitation of traditional fossil energy has brought increasingly severe environmental pollution problems such as energy resource shortage and climate change, making it extremely urgent to utilize renewable energy as an alternative for fossil fuels (Brockway et al., 2019; Xiong et al., 2020). In recent years, the large-scale penetration of renewable resources and new energy equipment alleviates the energy crisis, while also bringing rigorous challenges to the power system optimizing configuration and secondary energy rational utilization (Peidong et al., 2009). For example, the randomness and intermittency of power generated from wind turbines (WTs) or photovoltaic cells (PVs) and the time-space dimensional decentralization of electric vehicle (EV) charging, induce more uncertainty for present distribution network (DN) operations than ever before, as well as problems involving harmonic pollution (Siahroodi et al., 2021), three-phase voltage imbalance (Islam et al., 2020), and transformers aging (Elbatawy and Morsi, 2022). Particularly, the enlarged peak-to-valley difference (PVD) of the load curve puts forward higher requirements for the power system. The safety and economy, expressed as node voltage excursion and active power loss respectively, are commonly regarded as crucial indices reflecting the operation state of the DN (Dong et al., 2019). The serious peak load arisen from the aggregated charging behaviors of EV owners probably causes low-voltage even blackout, concurrently increasing system loss. Hence, guiding measures for coordinated regulation should be taken to minimize the negative influences of integrated renewable energy and EVs while satisfying the power and travel demands of customers.

Recently, the integrated energy system (IES) has been greatly supported by the Chinese government and turned into a research hotspot, since it benefits the integration of renewable energy and the coordinated development of the multi-energy system. Under the background of the IES, the management idea for the DN has changed from supply-oriented to demand-oriented (Zhang et al., 2018), which integrates the functions of distributed generation control, real-time monitoring, information sharing, and market transactions. This makes for better accommodation and scheduling abilities to diversify distributed energy equipment.

To stabilize the output fluctuation of renewable power generation, the battery energy storage system (BESS) has become a key supporting unit to improve the compatibility level for the WT and PV of the DN. By power transferring through electrochemical charging/discharging, the BESS not only enhances the utilization of clean energies, bringing higher economic benefits to the regional grid but further improves the reliability of the power supply. A hierarchical coordinated control strategy with the fast-response BESS to suppress high fluctuations associated with DGs outputs has been developed based on the MPC framework (Zhang et al., 2021). Moreover, as part of a vehicle-to-grid (V2G) system (Luo et al., 2020), EV can also be regarded as a controllable power resource, to realize the bidirectional power flow between the building cluster and DN. According to previous studies (Zeng et al., 2020; Fang et al., 2021), the time-of-use (TOU) price mechanism is an effective method for integrated demand side management, which creates an economic incentive for users to adjust their charging/discharging time. For example, residential customers plug in their vehicles during off-peak hours and discharge in peak regions to produce V2G profits. This helps to reduce the risks of distribution transformer overloading and power outage and stabilizes the load fluctuation by valley filling and peak shaving, thus balancing the demand and supply. These findings could enlighten us about solving the optimizing configuration problem of the EV-integrated DN.

Tremendous efforts have been devoted to DN optimization referring to multi-source cooperation. A voltage regulation model was proposed in a study by Mahmoudi et al. (2021) to enhance the voltage quality of the power system with EVs penetration; nevertheless, the active loss is not included in the objective function. Contrarily, Chen G. et al. (2013) analyzed the model for DN reconfiguration, with the single optimization function of power loss. Zarei M et al. designed a multi-objective optimization model for DN reconfiguration, considering the uncertainties of load variations and power production of DGs (Zarei and Zangeneh, 2017). For simultaneous optimization of power loss and voltage stability of the DN, reactive compensation measures, carried out through static Var compensator (SVC) capacitor generally in studies by (Xu et al., 2021) and (GholamiFarkoush et al., 2019), are always taken. Studies by (Wu et al., 2021) and (Tan and Chen, 2020) mainly focus on power management under the predictive control strategy for charging EVs, while they are only regarded as stochastic charging loads rather than controllable units which could participate in energy dispatch, and the V2G functions are not included. In a study by Jiao et al. (2021), a multi-objective optimization model emphasizing the EV integration to a smart DN is described, with the EVs scheduled in response to TOU price. Since the charging or discharging plan is generated with fixed peak and valley periods, it may not be suitable for the time-varying loads.

As analyzed above, earlier reports on the DN optimization operation concentrated on single or combinative objective functions of node voltage fluctuation and active power loss; however, these works entailed some deficiencies including the following aspects: (1) optimization scheduling did not involve wide-range types of energy resources, such as EVs or DGs, which were disregarded in some previous works; (2) utilization efficiency perfection of DGs was ignored; (3) EVs were only treated as stochastic charging loads; and (4) the dynamic dispatch of EVs was not considered.

To overcome the shortcomings of previous works, a hierarchical optimization strategy is proposed in this study, which combines the maximum power point tracking (MPPT) control of the DGs (Stage 1), coordinated V2G service, and BESS dispatch for minimized PVD via the positive guidance of the TOU mechanism (Stage 2), with dynamic reactive compensation for maintaining the reliable and economic operation of the DN (Stage 3). The tradeoff between node voltage excursion and active power loss is selected as the objective function to be optimized, and an improved genetic algorithm (GA) with elite preservation strategy (EPS) owning fast convergence is adopted to solve the best daily scheduling plan for the SVC component while meeting all constraints of the DN. An IEEE 33-bus system involving WT, PV, BESS, and EVs accessed to the building cluster is taken as the research object, to validate the effectiveness of the optimization model. In Stage 1, MPPT control improves the energy conversion efficiency of DGs and reduces the power supply pressure of the DN. In Stage 2, EVs participate in energy management as flexible loads, and functions including stabilization of the DGs’ output fluctuation and power load shifting are realized through coordinated dispatch of EVs and the BESS. In Stage 3, the multi-objective economic-secure optimization problem is dynamically worked out through SVC being put into operation, and the significant advantage of load shifting is highlighted by the statistical result comparison of relative voltage fluctuation and power loss, further confirming the effectiveness of the hierarchical scheduling approach.

A smart DN is mostly configured as an energy cascade utilization system, enabling free power flow among multiple energy sources, and provides power to residents through building cluster. Figure 1 briefly illustrates the structure of the DN containing dispersed multi-source. Specifically, the demand side involves basic power load, BESS to be charged, and scattered EVs connected to the building cluster. The schedulable supply side includes diversified distributed energy sources (such as WT, PV, bidirectional EV chargers, and BESS), to satisfy the power demand of residents. EVs are not regarded as traditional loads since they can also act as power suppliers. For the V2G application, EV devices not only provide customers with charging services but also allow the owners to interact with the power grid and make V2G profits by selling excessive energy back to the grid at an appointed electricity price level. Overall, the optimization strategy of the multi-source DN is intended to maintain stability and improve the energy efficiency of the whole system through power control over the distributed energy resources based on their operation features. On the premise of that daily power demand, information is collected, and the impacts of EV traveling and distributed energy equipment working are confirmed; the efficient planning and operation of the supply side can be carried out according to users’ willingness, under the guidance of distributed cooperative control, to finally achieve the comprehensive optimization dispatch from the demand side response, energy interaction, and DN operation perspectives. In this study, a hierarchical optimization strategy for the smart DN with multi-source integration is proposed, and the framework is presented in Figure 2.

In Stage 1, the optimal operation control of distributed renewable energy generators is implemented. MPPT measures are taken to maximize the working efficiency of WT and PV, to make full use of clean energies to alleviate the power supply pressure of the grid. In Stage 2, from the economic and security views of the power system, minimizing the PVD concurrently suppressing the power fluctuation created by DGs is selected to be the optimization objective in this stage, and an incentive method for EVs coordinated scheduling is introduced, wherein TOU measure is utilized to provide an economic incentive to make EV owners charge during less congested regions and discharge in peak hours, combined with the variable power charge/discharge control of the BESS. Resultantly, the charge/discharge guidance of the connected EVs could be determined by figuring out the optimal peak and valley periods, and the real-time output plan of the BESS configuration associated with load shifting could also be solved. In Stage 3, based on the power load after peak shaking and valley filling function of EVs and BESS collaborative services, SVC is put into operation for reactive power compensation. The coordinated structure formulated as a comprehensive objective function aiming to concurrently minimize the active power loss and node voltage excursion of the DN is established, and the optimal daily output schedule of SVC capacitors can be dynamically obtained. The strategy not only maintains the balance of power supply and demand but favors the running costs and voltage quality, merging the high energy efficiency, economy, and security for system operation. Analysis of mathematical modeling of the hierarchical optimization problem is given next.

Renewable energy power generation such as WT and PV has the characteristics of non-pollution and sustainability but is intermittent and time-varying. The MPPT control method is employed to maximize the output efficiency and improve the response speed of the renewable power generation system, which is directly related to the effective utilization of wind and solar energy and the safe operation performances of the generation system. Specifically for the WT running characteristics, as provided in Figure 3, under a constant blade pitch angle, the wind energy utilization coefficient changes with the tip speed ratio (Luo and Niu, 2016), and each wind velocity corresponds to an optimal mechanical speed of WT ω _max, for example, S for wind velocity V ₁ and P for V ₂, which makes the WT operate at MPP. By depicting all MPPs under different wind velocities, it forms a maximum power curve. Hence, it can be seen that satisfying the equation ω _m = ω _max is the essential task for the MPPT control of WT. Similarly, for the PV generation system, which usually covers the PV panel, Boost circuit, inverter, and load, the duty cycle D of the switching device in the Boost circuit can be taken as the control parameter for MPPT to well reduce the complexity of the system. More precisely, the temperature changes mainly affect the PV output voltage, while the irradiation changes mainly affect the PV output current, and under a given temperature and sunlight intensity, the intersection point of the load-line with the PV characteristic determines the operating point (Koutroulis et al., 2001). Consequently, the maximum power production is based on the load-line regulation by means of adjusting D to realize impedance matching, that is, the equivalent input resistance of the power converter equals the internal resistance of the PV cell (Podder et al., 2019), to impel the working point to move toward the MPP correctly, and the process is declared in Figure 4.

Due to the connection through a gear box, the WT rotational speed ω _m is directly proportional to the wind generator speed ω _r ; thus, the optimal ω _max of WT corresponds to a reference ω _r * of the generator. According to Figure 3, if the detected ω _r is equal to ω _r *, which indicates that the generation system operates at the MPP as expected, the excitation current value i _c of the wind power generator can be maintained. If the detected ω _r <ω _r *, manifesting that the generation system works at Region І, and it is necessary to increase ω _r to approach the MPP; while ω _r >ω _r * represents the operation state of the WT generation system at Region Ⅱ, and ω _r should be decreased to improve the conversion efficiency. Under such a non-equilibrium condition, regulation principles are established:

1) Region І: a positive speed variation Δω _r illustrates that the working point is close to the MPP along the path 1 direction, and ω _m increasing indicates that the output of the WT P _m may be greater than the generator P _g , and the excitation current i _c of the generator should be enhanced for power matching; contrastively, a negative Δω _r signifies that the working point moves far away from the MPP along the path 3 direction, and the excitation current i _c needs to be rapidly reduced to satisfy P _m > P _g , which drives the increase of ω _m and ω _r , and guarantee the correct moving direction.

Similar to WT analyzed above, the MPP tracking principle for PV generation also can be summarized as follows: if the PV generation system operates in Region І, the duty cycle D of Boost circuit should be increased, while D must be reduced as a response to the operating points in Region Ⅱ.

The explosive growth of these DGs’ penetration brings power rush and uncertainty problems to the DN. Since the connected EVs and BESS can both function as power buffer for intermittent DGs and backup power supply, matching the bidirectional EVs and BESS equipment for power smoothing is always considered as a simple and effective measure to solve the above problems.

The guiding ideology for the whole DN optimization can be concluded as follows: on the premise that the parameters including branches and loads are known and various constraints are satisfied, dynamically adjust the output of the SVC capacitor banks to optimize the comprehensive indices of active power loss and voltage excursion. Finally, the economic and safe operation of the multi-source DN can be achieved.

The standard IEEE 33-node power DN system shown in Figure 9 is taken for verification, and its load information is detailed in Reference (Huang et al., 2020). The basic parameters of the network for the per-unit system are 10 MVA and 12.66 kV. The proposed hierarchical optimization strategy is implemented on the MATLAB platform, and the research cycle is 24 h. The IEEE 33-node DN mainly contains: (1) residential building load (original power load without DGs and EVs integration), which is assumed to evenly distribute on four nodes: 3, 10, 18, and 32. Considering the time-varying characteristics of the daily load, a series of residential building load data for a certain day are collected from the study by Huang et al. (2020), with the PVD of 1763.8 kW, as shown in Figure 10. (2) Renewable energy power generation. DG1 (WT) and DG2 (PV) are installed on nodes 2 and 5, respectively, with the optimal output power under MPPT control. For the day studied, meteorological data involving wind velocity, temperature, and sunlight intensity (Hosseinalizadeh et al., 2016) associated with renewable energy power generation are depicted in Figures 11, 12, with sampling per hour in view of the randomness and uncertainty of climate. (3) BESS, positioned at nodes 3 and 10, with individually rated power of 100 kW and capacity of 500 kWh. According to (Yao et al., 2017), the charge/discharge efficiencies are selected as η _c = 0.95, η _d = 0.9, respectively. (4) SVC, placed on the weakest two nodes for compensation, and each capacitor bank is supposed to provide the maximum capacity C _jmax = 1 Mvar. (5) EV charging loads, which can also be deemed as power suppliers under the V2G mode. It is assumed that 500 EVs are accessed to the building cluster, with the charge/discharge power of 2.5 kW. Furthermore, the convergence accuracy of power flow calculation is 10^–4.

As a demand-oriented energy supply system, a smart DN with multi-source coupling is a major strategic measure for domestic energy development and the trend of the global power system. However, owing to the hybrid access of user resources, natural resources, and time-space dispersed resources making the power supply diversified and unstable, the DN system faces great changes in architecture, which brings great challenges to the upper collaborative dispatch and optimization. Consequently, establishing on the fundamental principle of multi-source integration and DN optimization control, the hierarchical multi-source coordinated regulation strategy considering the penetration of DGs, EVs, and BESS is proposed in this study. Combining the efficient utilization of DGs, load shifting function of EVs and BESS, and the dynamic reactive compensation of SVC capacitor banks, the optimization strategy could effectively balance the tradeoff between active power loss and node voltage fluctuation, and its feasibility and effectiveness are illustrated through MATLAB simulation results. First, based on MPPT control, the self-optimizing processes of WT and PV output power are implemented for improving the power production efficiency of the DGs. Next, according to the dynamic load change, the TOU price mechanism is introduced to improve the initiative and utilization of bidirectional EVs and BESS by guiding the charging/discharging plans of the owners and achieving the peak shaving and valley filling function, which is significant for alleviating the power supply pressure of the DN. Finally, both total active power loss and node voltage fluctuation are selected as the optimization objectives, under the basic constraints, and the improved GA with EPS method is employed to determine the daily optimal SVC capacitor bank compensation plan for the multi-objective economic-secure scheduling problem. After the above hierarchical dispatch steps, the entire power loss and node voltage out-of-limit risk in the DN can be improved simultaneously, which promotes the economic and reliable operation of the regional DN. The research results in this study have good extensibility and may provide certain reference significance for exploring effective flexible load modeling and synergetic management technology of “source-grid-storage-load” system in the subsequent construction of smart DNs. Nevertheless, the application would be subject to some limitations. First, if other objective functions are also paid attention to, it is better to adopt efficient multi-objective solvers instead of a single-objective optimizer by regulating the weight coefficients ω₁ and ω₂. Moreover, influences of parameter uncertainties, such as DGs and power load are not taken into account in this study. Finally, if different types of distributed energy equipment are to be surveyed, mathematical models should be modified according to the actual running characteristics.