Research on Active Distribution Network Planning with Electric Vehicles and Distributed Generation

With the increasing number of Electric Vehicle(EV) and clean energy generation year by year, EV and distributed generation (DG) have become issues that have to be considered in active distribution network planning. Firstly, considering the time series characteristics of DG, the output time series model of DG is established; Secondly, the parking demand and space-time movement model of EV is established, and the Monta Carlo method is used to simulate the space-time distribution of EV charging load in different planning areas; Finally, taking the system investment and annual operation and maintenance cost, voltage index and environmental index as the objective function, and considering the node voltage, node current and DG installation capacity as constraints. The improved particle swarm optimization algorithm is used to solve the planning model, and the access location and capacity of EV charging station and DG are obtained. Taking a distribution network as an example, the rationality and effectiveness of the proposed model and algorithm are verified.


Introduction
Affiliations of authors should be typed in 9-point Times. They should be preceded by a numerical superscript corresponding to the same superscript after the name of the author concerned. Please ensure that affiliations are as full and complete as possible and include the country.
At present, the research of comprehensive planning considering EV and DG can be roughly divided into two categories: one is to consider the planning of EV and DG separately, and treat them as relatively independent subjects, without considering the complementary characteristics between them to plan the location and capacity of EV charging station and DG location and capacity. For example, in reference [1][2][3], the model of distribution network investment operation and maintenance cost, environmental cost and random expected value is established, and genetic algorithm is used to solve the location and capacity of DG and EV charging stations. In reference [2][3][4][5], the objective function of investment operation and maintenance cost, minimum network loss and optimal voltage quality is established, and the comprehensive optimization planning scheme is obtained. The other is to consider the coordination and complementarity between EV and DG, and make full use of EV charging to increase the local consumption of DG. Literature [3][4][5][6][7][8] established a planning model that considered the correlation between EV travel characteristics and wind turbine output characteristics, which effectively reduced the load peakvalley difference.
To sum up, the current research results at home and abroad are mainly for DG or EV to be considered separately, and less overall consideration of the complementary characteristics of DG and EV for site selection and capacity planning. Therefore, this paper fully considers the randomness of the two, establishes a model including system investment and annual operation and maintenance cost, voltage index and environmental index, and uses the improved particle swarm optimization algorithm to solve the model. Finally, the IEEE33-node system is used to verify the effectiveness of the proposed model and method.

Photovoltaic (PV)output analysis
The magnitude of PV output can be approximately determined by the intensity of light. The relationship between PV output and light intensity can be approximately expressed as: is the rated installation capacity of the PV power source, I is the actual light intensity, and N I is the rated light intensity of the PV power source.

Wind Turbine (WT) output analysis
Wind power generation has natural volatility and randomness. Even the output at the same time in the same month has strong randomness, which hinders the development of wind power generation to a certain extent [37]. The function expression of WT output WT P and wind speedV is as follows: is the rated installed capacity of WT, i V is the cut in wind speed of WT, N V is the rated wind speed of WT and O V is the cut out wind speed of WT.
3 Temporal and spatial distribution of EV charging load

Classification of EV
By analyzing the current development status of EVs in the public sector, the following trends can be seen: From 2010 to 2015, the initial development of EVs will start demonstration operations in official vehicles, buses and taxis; From 2016 to 2020, although the number of electric private cars has gradually increased, the total amount is still small. Most users are willing to choose to rent and use EVs, and the number of electric buses and electric taxis has increased year by year; It is estimated that the trend of EV development from 2021 to 2030 is the rapid increase in the number of private cars. Therefore, this article takes electric private cars, electric taxis and electric buses as the main research objects.

EV charging mode
According to the "Electric Vehicle Conductive Charging Interface" (QC/T841-2010) approved by the Ministry of Industry and Information Technology in 2010, EV charging modes are divided into three types: slow charging, regular charging and fast charging. The specific data are as follows:

Selection of EV charging strategy
(1) Disordered charging mode: Charging starts immediately after the return of the EV arrives, and the initial charging time is the grid-connected time.
(2) Orderly charging mode: this mode is mainly affected by the market incentive system, such as time of use electricity price mechanism, which divides the electricity price into peak electricity price, valley electricity price and flat section electricity price. When the load is low, the cheaper valley electricity price is adopted to encourage users to charge in the valley period, so as to achieve the effect of cutting peak and filling valley.

Space-time simulation of EV charging load
When establishing the parking demand model of EV, it is necessary to divide the regional economic form into residential area, commercial area and industrial area in terms of land use function. The charging load of EV is determined by EV type, charging mode and charging time. These factors are essentially affected by the travel rules of users and belong to random variables. Monte Carlo method is a random sampling method, that is, when the problem to be solved can be expressed in the form of probability, the "test" method can be adopted to obtain the frequency of the event as the solution of the problem [9][10]. The specific process is as follows: (1) Determine the EV ownership and the proportion of various EV in the planned area; (2) The EV in the planning area is divided into private cars, taxis and buses, and the charging load is simulated according to the travel law, parking demand and space-time movement model of different types of EV; (3) Assuming that the bus charging pile is set in the industrial zone, the EV charging load of residential area, commercial area and industrial zone can be obtained by superposition of charging load in different areas of three types of vehicles.

Objective Function
In this section, the objective function of active distribution network planning with EV and DG is constructed. The time sequence characteristics of DG output and basic load are fully considered. Three sub objective function models including system investment and annual operation and maintenance cost, voltage index and environmental index are constructed. The expression of each sub objective function is as follows: (1) System investment and annual operation and maintenance cost: Where, K is the total number of scenarios, ( ) P s is the probability of the corresponding scenario, r is the discount rate, n is the planning period, INV C is the investment cost, and OPE C is the annual operation and maintenance cost.
1) Investment cost 1, , , Where, N is the number of distribution network nodes in the area to be planned; , DG i N is the type of DG installed at node i . when the value is 1, it means that the WT is installed here; when the value is 2, it means that the PV is installed here; 1, In the formula, 2.
k i c is the maintenance cost of the unit capacity of the k -type DG at the node i ; , k i P is the power generated by the k -type DG at the node i ; , mEV i c is the maintenance cost of the EV charging station per unit capacity at the node i ; mline c is the maintenance cost of the unit line; loss c is the cost of the unit power loss; ( ) ij I t is the current between node ij ; ij R is the resistance between node ij ; P P is the annual electricity purchase cost of distribution network in peak period; V P is the annual electricity purchase cost of distribution network in valley period; F P is the annual electricity purchase cost of distribution network in horizontal section; The voltage index is as follows: The annual average value of voltage deviation of each node in distribution network is: The annual average value of voltage stability index of each branch in distribution network is as follows: 2 8760 , Where, LOAD E is the annual total electricity consumption; DG E is the annual power generation of DG; p is the pollutant type of traditional power generation; k e is the emission coefficient of pollutant of traditional power generation; k V is the environmental value of pollutant k ; k V  is the penalty standard of pollutant k .

Solution algorithm
PSO algorithm is simple and easy to implement, with high accuracy and fast convergence speed, so it is often used to solve optimization problems. However, the traditional particle swarm optimization algorithm has poor variety, easy to premature and fall into local optimal solution. This paper uses the following methods to improve the traditional particle swarm optimization algorithm: (1) Changing the fixed  into the random value between [0.5,1.0] will make the optimization ability stronger. The update formula of  after changing is as N is a random number between [0,1] . (2) Using crossover and mutation operation in genetic algorithm can improve the diversity of solutions, so as to ensure the high convergence speed of the algorithm and avoid premature phenomenon.

Overview of calculation examples
In order to verify the feasibility of the planning model, the IEEE-33 node power distribution system is used for simulation analysis. The power distribution system is shown in the figure below. Considering the time sequence characteristics of traditional load, the distribution network nodes are divided into residential load, commercial load and industrial load. The voltage level of the modified distribution network system is still 12.66kV, and the maximum normal load is 5.084+j2.547MW. The load nodes of the residential, commercial and industrial areas are assumed to be as follows in Because the fan needs to be installed in an open place to ensure its normal output, it is not suitable to install the fan in the residential area with dense buildings, and the node to be installed is selected as the node in the commercial area and industrial area; The photovoltaic installation nodes to be selected are all nodes; the charging station installation nodes to be selected are{7 9 12 14,15,17, 20 23, 24, 28,32} ，， ， ， 。The DG node is treated as PQ node. The planning period is 20 years and the discount rate is 8%. The cut in wind speed of fan is 3 / m s , cut out wind speed is 20 / m s , rated wind speed is 13.5 / m s ; the rated light intensity of photovoltaic is 2 1000 / W m . The fixed investment cost of EV charging station is 3 million yuan, the variable investment cost is 6000 yuan / kW, and the annual operation and maintenance cost is 6000 yuan / kW. It is assumed that the number of electric private cars, electric taxis and electric buses in the planned area is 500, 100 and 25 respectively. The population size is 100 and the maximum number of iterations is 100. The following three schemes are simulated and analyzed: (1) Regardless of DG, consider the expansion plan of the active distribution network under the EV disorder charging mode; (2) Consider the expansion plan of the active distribution network under the disorderly charging mode of DG and EV; (3) Consider the expansion plan of the active distribution network under the orderly charging mode of DG and EV.

Simulation result analysis
The planning results of the three planning schemes are shown in the following table:  Table 2 shows the planning results under the above three schemes, and Table 3 shows the comparison of various costs and indicators under the three different planning schemes. The specific analysis is as follows: From the perspective of cost comparison: Although the access of DG causes the system investment cost and operation and maintenance cost of the scheme (2) and scheme (3) to be higher than that of scheme (1). However, the access of DG has also brought about a reduction in power purchase costs, a reduction in environmental costs, a reduction in network loss costs, and a better voltage index. Therefore, the comprehensive cost of scheme (2) and scheme (3) is lower than that of scheme (1).
From the point of view of the voltage index: Scheme (1) only considers the access of EV charging station, but does not consider the DG, the voltage index is the largest, that is, the power quality is the worst in this case, and the large-scale access of EV leads to the increase of voltage deviation, which also means that the reliability of power supply is difficult to be guaranteed.
From the perspective of environmental indicators: Compared with scheme (2) and scheme (3), although the access of DG and EV is considered, the environmental index of scheme (3) is better than that of scheme (2) because scheme (3) adopts orderly charging strategy and increases the consumption of DG power generation.

Conclusion
In this paper, the randomness of DG output and the temporal and spatial characteristics of EV charging load are studied. On this basis, an active distribution network planning model with EV and DG is constructed, and the improved particle swarm optimization algorithm is used to solve the model.
(1) In this paper, the space-time distribution model of EV charging load is established, which not only takes into account the randomness of EV charging load time, but also fully considers the space characteristics of EV charging load.
(2) Compared with the traditional distribution network planning, this paper fully considers the temporal and spatial distribution characteristics of EV and the timing characteristics of DG, and the planning results are more suitable for engineering practice.
(3) In this paper, the active distribution network planning with EV fully considers the adoption of EV orderly charging strategy to absorb more distributed generation output, and effectively improves the power supply reliability.