An Optimal Scheduling Method for Electric Vehicles Based on User Characteristics

Abstract

The present invention relates to the technical field of power system optimization and dispatching, and relates to an electric vehicle optimization and dispatching method based on user characteristics, including: S1, collecting and screening user electric vehicle data, and eliminating error values ​​and wrong value user data; S2, constructing a CF feature tree for user electric vehicle data through a second-order clustering algorithm, and dividing user electric vehicle data into different cluster subgroups; S3, dividing electric vehicle users in different cluster groups into adjustable and unadjustable vehicle groups; S4, establishing an objective function with the minimum total cost of system operation. The present invention uses a second-order clustering algorithm to perform cluster analysis on electric vehicle user characteristics under big data, and then combines the power grid optimization dispatching strategy to establish an objective function with the minimum total cost of system operation, balance the charging and discharging costs, reduce the total cost of system operation, and achieve the effect of peak shaving and valley filling.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization dispatching, and in particular to an electric vehicle optimization dispatching method based on user characteristics. Background Art

[0002] User characteristics refer to the travel characteristics and charging characteristics of electric vehicle users in their daily lives. Large-scale access to the grid for electric vehicles will affect the safe and stable operation of the power system. The charging behaviors and daily travel behaviors of different types of users are different and similar. Zhang Yanjuan et al.'s "Charging Load Forecasting Based on Spatiotemporal Characteristics of Electric Vehicles" refines and integrates the influencing factors such as the selection of charging areas for electric vehicle charging, the classification of electric vehicles, the charging methods of electric vehicles, and the spatiotemporal distribution of charging into a model study based on the charging characteristics of electric vehicles, and finally obtains the charging load model of electric vehicles between different spatiotemporal regions. Zhang Hongcai et al.'s "Electric Vehicle Charging Load Forecasting Method Considering Spatiotemporal Distribution" distinguishes the charging behaviors of electric vehicles based on the driving characteristics and parking characteristics of electric vehicle users, and simulates the charging load through Monte Carlo simulation.

[0003] By mining and analyzing user characteristics, we can find out their characteristics in charging location selection, charging time period selection, and SOC state restrictions. Electric vehicle platform companies can adopt different scheduling strategies based on the analysis results to control the scale and time of electric vehicles entering the grid, so that the user side can become a scheduling participant that enables the power supply side to operate smoothly, so that the power system can operate safely, stably, economically, efficiently, and reliably. After clustering the data, analyze the characteristics of the subgroups, and distinguish between controllable and uncontrollable subgroups based on their characteristics, count their needs, and participate in the scheduling of the power grid. This helps electric vehicle platform companies to develop optimized scheduling plans for electric vehicles. Summary of the invention

[0004] The technical problems to be solved by the present invention are:

[0005] The technical solution adopted by the present invention is: an electric vehicle optimization scheduling method based on user characteristics, comprising the following steps:

[0006] S1. User data preprocessing, including: collecting and screening user electric vehicle data, and removing error values ​​and wrong value user data;

[0007] The user electric vehicle data obtained through the automobile supervision platform and the automobile charging pile supervision platform include: vehicle type (Cartype), charging start time (Timestart), charging end time (Timeend), travel distance (Range), charging start battery state of charge (SOCstart), charging end battery state of charge (SOCend), charging method (Chargeway), charging location (Chargeplace) and charging period (Chargingperiod), a total of 9 user feature data. Table 1 is a sample data of a user feature:

[0008] Table 1 Characteristic data of a user

[0009]

[0010] The vehicle types are divided into: Taxi, Business, Family car, Bus; the travel distance is divided into: Slong Distince (more than 200 km), Long Distince (100-200 km), Short Distince (0-50 km) and Medium Distance (50-100 km); the charging mode is divided into: Longtime, Shorttime, Normal; the charging time period is divided into: Night and Day;

[0011] S2. Take the 9 user feature data as input, and use the second-order clustering algorithm of SPSS software to build a CF feature tree. Perform cyclic pre-clustering based on whether the number of leaf nodes of the clustering algorithm reaches the maximum allowed number of clusters. When the required conditions are met, that is, clustering is completed according to different driving characteristics and charging characteristics, and the clustering effect is judged by BIC (Schwarz Bayesian). According to the clustering results, the similarity and regularity of user characteristics are studied, and whether the user's satisfaction with the choice of charging location, charging time period, charging start SOC, and charging end SOC meets the travel needs is considered. The clustering results show that the choice of charging location and the choice of charging method of different electric vehicle users in the same time period are restricted by the charging time period, and the charging time of the same electric vehicle user in the same time period is affected by the battery capacity. The order of clustering importance is vehicle type>charging location>travel distance>charging method>charging time period; if Figure 2As shown in the figure, the importance of clustering data increases from top to bottom, and the most important one is Cartype. The data shows that when users choose to charge, the vehicle type, charging location, mileage and charging method are the decision points when users choose to charge. Finally, private cars, buses, taxis and official cars are clustered into 6 cluster subgroups. The behavioral characteristics of the 6 cluster groups are described in Table 2:

[0012] Table 2 Summary of cluster group behavior characteristics

[0013]

[0014] The distribution of characteristics of private cars, buses, taxis and official cars has crossover and separation; the characteristics of electric vehicle users in the six clusters are divided into controllable or not, and are divided into controllable vehicle group M1 and non-controllable vehicle group M2, among which the controllable vehicle group M1 includes cluster group 3, cluster group 4 and cluster group 6; the non-controllable vehicle group M2 includes cluster group 1, cluster group 2 and cluster group 5; the controllable vehicle group M1 participates in the interaction of the power grid as a load or a power source. In the interaction, the vehicle group consumes electric energy as a load and provides electric energy as a power source;

[0015] Combination Figure 3-8 As can be seen from Table 2, clusters 1 and 2 are clusters composed of taxi users. When choosing to charge, they need to consider the operating time. They only choose regular charging in the middle of the night, and choose fast charging during the operating intervals at other times. Their charging choices are relatively arbitrary. Therefore, clusters 1 and 2 are divided into the M2 vehicle group for scheduling, and they perform normal charging and discharging behaviors as the uncontrollable vehicle group M2.

[0016] In clusters 5 and 6, buses and private cars dominate. The charging behavior of buses in cluster 5 during daytime operation is to choose fast charging. The choice of charging time is affected by SOC and driving distance, and they mostly choose to charge in bus parking areas. Private car owners in cluster 6 choose to connect their electric vehicles to the power grid for charging after getting off work at night. At this time, the controllability of buses is not as high as that of private cars. Therefore, cluster 5 is divided into the uncontrollable vehicle group M2, and cluster 6 is divided into the controllable vehicle group M1.

[0017] Cluster 3 includes official vehicles that charge at night and some buses that replace batteries for charging. The charging behavior of this type of users can be controlled. Official vehicles participate in the interaction of the power grid as loads or power sources, and the cost of replacing batteries for buses is lower than the charging cost and can be used as a backup power source. Therefore, cluster 3 is divided into the controllable vehicle group M1.

[0018] Cluster 4 includes private car users and some taxi users. From the perspective of their behavior, the behavior of these taxi users is similar to that of private car users. This is because in the research process, online car-hailing vehicles are classified as taxi users. Some online car-hailing vehicle owners actually drive their vehicles back to residential areas and charge them in residential areas. Their working hours are relatively short, similar to private cars. Therefore, when dividing this type of cluster, it is divided into the controllable vehicle group M1.

[0019] S3. Establish an objective function to minimize the total cost of system operation, and optimize the scheduling according to the characteristics of adjustable and unadjustable users. The objective function minAC (minAll cost) is expressed as:

[0020]

[0021] Where i represents the i-th unit; T is the start-up time period of the selected thermal power unit; the total number of thermal power units is N; P i (t) represents the active output of the i-th unit at time t; V i (t) represents the state of the thermal power unit, 1 represents the on state, and 0 represents the off state; C i is the total fossil fuel cost of thermal power unit i at time t; S i (t) is the cost incurred when thermal power unit i is turned on in time period t, DP(t) is the discharge price at time t, OP(t) is the charging price at time t, P vdis(n) (t) Electric vehicle discharge power, P vc h (n) (t) is the charging power of the electric vehicle; N v The total number of electric vehicles participating in the dispatch;

[0022] C i (P i (t)) and P i (t) is expressed as a function:

[0023] C i (P i (t)) = a i +b i P i (t)+c i [P i (t)] 2 (2)

[0024] In the formula, a i 、b i 、c i is the cost coefficient of the fuel for thermal power generation units;

[0025] Constraints of thermal power units

[0026] 1) Power balance constraints

[0027]

[0028] Among them, P L (t) is the active load of the system during period t;

[0029] In the formula, the charging power P of a single electric vehicle is vch (t) can be expressed as:

[0030]

[0031] P vdis (t) The discharge power can be expressed as:

[0032]

[0033] Among them, N vch Refers to the number of vehicles charged per hour, N vdis Refers to the number of vehicles discharging per hour, Indicates the current SOC of the electric vehicle. Indicates the amount of electricity consumed by the electric vehicle battery; P v Indicates the rated power of the electric vehicle;

[0034] 2) Charge and discharge constraints

[0035] EV Charging Constraints

[0036]

[0037] Among them, P vchLimit(n) (t) is the upper limit of the charging capacity of the nth electric vehicle at time t;

[0038] EV Discharge Constraints

[0039]

[0040]

[0041] Where P vdisLimit(n) (t) is the discharge limit of the nth electric vehicle at time t;

[0042] EV quantity constraints

[0043]

[0044] Where N vdis 、N vch Indicates the number of electric vehicles currently being charged and discharged;

[0045] A n (t), Bn (t) represents the choice between charging and discharging of the nth electric vehicle at time t. Because charging and discharging are not performed at the same time, binary variables 0 and 1 are selected as the identifiers of their states.

[0046] 3) EV balance constraints

[0047]

[0048] 4) Output constraints of conventional thermal power units

[0049] P i min (t)≤P i (t)≤P i max (t) (10)

[0050] Where P i max , P i min They are respectively the upper and lower limits of active output of thermal power unit i.

[0051] 5) Spinning reserve constraints

[0052]

[0053] Where P i max (t) is the maximum output of unit i in period t, P R (t) is the spinning reserve capacity of the system during the corresponding period t.

[0054] 6) Unit start and stop time constraints

[0055]

[0056]

[0057] Where, T i off (t), T i on (t) are the continuous shutdown time and continuous startup time of unit i in time period t, respectively; They are the minimum continuous shutdown time limit and the minimum continuous startup time limit of unit i respectively.

[0058] The beneficial effects produced by the present invention are:

[0059] 1. A second-order clustering algorithm was used to perform cluster analysis on the characteristics of electric vehicle users under big data. Combined with the grid optimization dispatching strategy, an objective function with the minimum total cost of system operation was established to balance the charging and discharging costs, reduce the total cost of system operation, and achieve the effect of peak shaving and valley filling. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a basic flow chart of the electric vehicle optimization dispatching method based on user characteristics of the present invention;

[0061] Figure 2 is the importance of different variables obtained from the clustering results;

[0062] Figure 3-8 It is a distribution diagram of cluster subgroups obtained by clustering different characteristics of the present invention;

[0063] Fig. 9 is a flow chart of the system of the present invention;

[0064] Fig.10 This is a comparison chart of load consumption before and after the optimization system of the present invention, taking one day as an example. DETAILED DESCRIPTION

[0065] The present invention is further described below in conjunction with the accompanying drawings and embodiments. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore it only shows the components related to the present invention.

[0066] The present invention provides an electric vehicle optimization scheduling method based on user characteristics, such as Figure 1 As shown, including:

[0067] S1. User data preprocessing, including: collecting and screening user electric vehicle data, eliminating error values ​​and wrong value user data, where the user electric vehicle data comes from a certain automobile manufacturer's electric vehicle supervision platform and a certain city's electric vehicle charging pile supervision platform;

[0068] S2. Take the 9 user feature data as input, use the second-order clustering algorithm of SPSS software to build a CF feature tree, judge whether the number of leaf nodes of the clustering algorithm reaches the maximum allowed number of clusters as a condition for cyclic pre-clustering, complete clustering when the condition is met, and judge the clustering effect by BIC (Schwarz Bayesian); the results of user feature clustering of an electric vehicle regulatory platform of an automobile manufacturer and an electric vehicle charging pile regulatory platform of a city within a certain period of time are shown in Table 3:

[0069] Table 3 Second-order clustering result output table

[0070]

[0071] S3, dividing the electric vehicle user characteristics of different cluster groups into controllable or not, and dividing them into controllable vehicle group M1 and non-controllable vehicle group M2;

[0072] The electric vehicle user characteristics of the six clusters are divided into controllable and uncontrollable vehicle group M1 and uncontrollable vehicle group M2. The controllable vehicle group M1 includes cluster group 3, cluster group 4 and cluster group 6; the uncontrollable vehicle group M2 includes cluster group 1, cluster group 2 and cluster group 5; the controllable vehicle group M1 participates in the interaction of the power grid as a load or a power source;

[0073] S4. Establish an objective function to minimize the total cost of system operation, optimize scheduling according to the characteristics of controllable and uncontrollable users, and impose power balance constraints, charge and discharge constraints, EV balance constraints, conventional thermal power unit output constraints, spinning reserve constraints, and unit start and stop time constraints on the objective function.

[0074] like Fig. 9 For the entire uncontrollable M2 vehicle group and the controllable M1 vehicle group participating in the interactive process of the power grid, when participating in the optimization dispatch, it is necessary to first determine whether the vehicle is controllable. If it belongs to the controllable vehicle group M1, the discharge electricity price is determined and optimized; if it does not belong to the cluster group, a second determination is made to determine whether it belongs to the uncontrollable vehicle group M2. If it meets the requirements, the charging demand of the vehicle is counted to optimize the charging of the uncontrollable charging vehicle group. The vehicle cannot be in a state of both charging and discharging at the same time. Electric vehicles can only be charged and discharged once a day; the discharge depth is set to 10%P v , P v is the rated capacity of the electric vehicle; when the charge and discharge reaches the constraint condition, t is incremented by 1 to enter the next cycle. At this time, the electric vehicle is in the charging state, and the charging of the electric vehicle is optimized. In the entire process of charge and discharge optimization scheduling, the constraints on the unit side should always be maintained; including: the start and stop time constraints of the unit, the rotation reserve constraints of the unit, and the balance constraints of the system operation and the electric vehicle; after optimizing the scheduling of the entire system on this basis, all costs are superimposed, and the total cost is minimized. If there is room for the total cost to be reduced, the number of iterations X is incremented by 1 for the next iteration process.

[0075] Table 4 Generator set parameters

[0076]

[0077] In the above example, after orderly optimization scheduling of electric vehicles, the results show that the total charging capacity of EVs decreased by 1144.96MW and the total discharge capacity increased by 711.91MW; the load comparison between peak and valley periods within a single day decreased significantly, and the peak-valley difference of disorderly charging was greater than the difference of orderly charging and discharging loads. The results show that orderly charging and discharging can achieve the effect of "peak shaving and valley filling"; it is worth noting that the load in the valley period at this time is reduced. This is because when the controllable vehicle group M1 and the uncontrollable vehicle group M2 are divided according to the clustering results, the vehicle group M2 is still charged according to the plan that it cannot be charged disorderly in the interaction with the power grid; the cluster subgroups show that in cluster group 1, cluster group 2 and cluster group 5, taxis are mainly charged randomly, and the proportion of this vehicle group in the number of cases is 33.33%. Therefore, the obvious reason for the increase in order but not disorder during the load valley period is this. The charging cost in a single day was reduced by 742,893.1 yuan, while the discharging cost increased by 579,130.65 yuan. This shows that this method can not only reduce the total cost of system operation, but also achieve a good peak-shaving and valley-filling effect.

[0078] The charging and discharging demands of controllable and uncontrollable vehicle groups are counted, and the EV vehicle groups are interacted with the power grid in disorderly and orderly charging modes. The comparison of the optimal scheduling of disorderly EV vehicle groups and orderly EV vehicle groups is obtained, such as Fig.10 As shown, this scheme has been well verified for the optimal dispatch of the power grid.

[0079] The present invention uses a second-order clustering algorithm to perform cluster analysis on the characteristics of electric vehicle users under big data, and then combines it with the power grid optimization scheduling strategy to establish an objective function with the minimum total cost of system operation, balance the charging and discharging costs, reduce the total cost of system operation, and achieve the effect of peak shaving and valley filling.

[0080] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An electric vehicle optimization scheduling method based on user characteristics, characterized in that: The following steps are involved: S1. Collect and screen the user's electric vehicle data, and remove the error value and wrong value user data; S2, constructing a CF feature tree using the second-order clustering algorithm for the user's electric vehicle data, and performing cyclic pre-clustering based on whether the number of leaf nodes of the clustering algorithm reaches the maximum allowed number of clusters, and dividing the user's electric vehicle data into different cluster subgroups; S3, dividing the electric vehicle user characteristics of different cluster groups into controllable or not, and dividing them into controllable vehicle group M1 and non-controllable vehicle group M2; S4. Establish an objective function to minimize the total cost of system operation, optimize scheduling according to the characteristics of adjustable and unadjustable users, and impose power balance constraints, charge and discharge constraints, EV balance constraints, conventional thermal power unit output constraints, spinning reserve constraints, and unit start and stop time constraints on the objective function; The objective function formula is: Where i represents the i-th unit; T is the start-up time period of the selected thermal power unit; the total number of thermal power units is N; P i (t) represents the active output of the i-th unit at time t; V i (t) represents the state of the thermal power unit, 1 represents the on state, and 0 represents the off state; C i is the total fossil fuel cost of thermal power unit i at time t; S i (t) is the cost incurred when thermal power unit i is turned on in time period t, DP(t) is the discharge price at time t, OP(t) is the charging price at time t, P vdis(n) (t) Electric vehicle discharge power, P vch(n) (t) is the charging power of the electric vehicle; N v The total number of electric vehicles participating in the dispatch; C i (P i (t)) and P i (t) is expressed as a function: C i (P i (t))=a i +b i P i (t)+c i [P i (t)] 2 (2) In the formula, a i 、b i 、c i is the cost coefficient of the fuel for thermal power units.

2. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The user electric vehicle data includes: vehicle type, charging start time, charging end time, travel distance, battery state of charge at the start of charging, battery state of charge at the end of charging, charging method, charging location and charging time period.

3. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The power balance constraint formula is: Among them, P L (t) is the active load of the system during period t; In the formula, the charging power P of a single electric vehicle is vch (t) can be expressed as: P vdis (t) The discharge power can be expressed as: Among them, N vch Refers to the number of vehicles charged per hour, N vdis Refers to the number of vehicles discharging per hour, Indicates the current SOC of the electric vehicle. Indicates the amount of electricity consumed by the electric vehicle battery; P v Indicates the rated power of the electric vehicle.

4. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The charging and discharging constraints include EV charging constraints, EV discharging constraints and EV quantity constraints. The EV charging constraint formula is: Among them, P vchLimit(n) (t) is the upper limit of the charging capacity of the nth electric vehicle at time t; The EV discharge constraint formula is: Where P vdisLimit(n) (t) is the discharge limit of the nth electric vehicle at time t; The EV quantity constraint formula is: Where N vdis 、N vch Indicates the number of electric vehicles currently being charged and discharged; A n (t), B n (t) represents the choice between charging and discharging of the nth electric vehicle at time t; because charging and discharging are not performed at the same time, binary variables 0 and 1 are selected as identifiers of their states.

5. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The EV balance constraint formula is:

6. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The output constraint formula of the conventional thermal power unit is: P i min ≤P i (t)≤P i max (10) Where P i max , P i min They are respectively the upper and lower limits of active output of thermal power unit i.

7. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The spinning reserve constraint formula is: Where P i max (t) is the maximum output of unit i in period t, P R (t) is the spinning reserve capacity of the system during the corresponding period t.

8. The electric vehicle optimization scheduling method based on user characteristics as claimed in claim 1, characterized in that: The start and stop time constraint formula of the unit is: Where, T i off (t), T i on (t) are the continuous shutdown time and continuous startup time of unit i in time period t, respectively; They are the minimum continuous shutdown time limit and the minimum continuous startup time limit of unit i respectively.

Images