An optimized scheduling method considering differences among EV owners

Abstract

The present invention relates to the technical field of electric vehicle scheduling, and particularly to an optimized scheduling method considering differences among EV owners, including performing clustering analysis on EV data respectively; establishing price demand response models for various types of EV owners; guiding the charging load and V2G of various types of EV owners through prices, and combining with the mathematical model of the power generation cost of the generating unit, with the minimum power generation cost as the optimization goal, to solve the optimal power generation plan. The present invention performs clustering analysis based on EV owner data, establishes demand response models for various types of EV owners, and conducts price guidance based on the demand response models, combines with the power generation cost model of the generating unit, and takes the minimization of the power generation cost as the goal to solve the optimal power generation plan, selling electricity price, and the price at which the power grid buys electricity from EV owners. The present invention considers the differences among various types of EV owners during scheduling by clustering the owners, and has obvious advantages in the scheduling effect of EV charging load and V2G.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle scheduling, and in particular to an optimized scheduling method considering the differences of EV owners. Background Art

[0002] With the innovation of the automotive industry, a large number of internal combustion locomotives have been replaced by EVs. It can be foreseen that the load of EVs will be a major growth point of future power loads. Transferring a large part of the load in commuting traffic to the power system poses a great challenge to the power system. It is imperative to build a safe, effective, and sustainable energy system. The EV load is a very typical dispatchable load. The storage battery of an EV can store electrical energy. In recent years, V2G (Vehicle-to-grid) has developed very rapidly. EVs with V2G function can deliver the stored electrical energy to the power grid to fill the gap in power generation. Considering that EVs belong to individual EV owners, the driving behaviors of EV owners vary greatly. The driving, charging load, and V2G size at different time periods will directly affect the scheduling results. In order to achieve better scheduling results, it is necessary to pre-extract the driving characteristics of EV owners and then perform scheduling based on the characteristics of EV owners, so as to increase the number of EV owners willing to participate in scheduling and obtain better scheduling results.

[0003] Patent CN202010292515.9, an electric vehicle scheduling method and scheduling system, obtains a set of EV loads by establishing a mathematical model of power generation and EV power consumption and performing optimization and solution. However, this solution only calculates the EV load that is most beneficial to the power generation side from the perspective of the power generation side, and does not fully consider whether EV owners are willing to cooperate with the scheduling. The literature "Modeling and prioritizing demand response programs in power markets" proposes a price-guided EV scheduling scheme, which considers the impact of electricity price changes at different time periods of a day on the EV load, but does not fully analyze the acceptance degree of different EV owners for the scheduling scheme and cannot achieve the optimal scheduling of various EV owners. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the technical solution adopted by the present invention is: an optimized scheduling method considering the differences of EV owners, including the following steps:

[0005] S1. Perform clustering analysis on the driving time, charging, and V2G data of EV owners respectively to construct several sets of EV owners with different driving characteristics;

[0006] Furthermore, based on the charging data and vehicle usage time data of EV owners, EV owner charging classification is achieved through the K-Means clustering algorithm; based on the V2G data and vehicle usage time data of EV owners, EV owner V2G classification is achieved through the K-Means clustering algorithm; the K-Means clustering algorithm divides EV owners into K clusters based on the distances between data points without a supervision signal; the K-Means clustering algorithm first randomly selects K points as the initial centers of each cluster, divides the data into the nearest clusters according to the distances between points and the cluster centers, recalculates the center points of each cluster after all points are divided, and then re-divides based on the new center points until the results of each division remain unchanged; in the final result, the distances between points within the clusters are as small as possible, while the distances between different clusters are as large as possible;

[0007] The vehicle usage time determines whether EV owners have free time to participate in scheduling during that period; the charging load can distinguish whether EV owners are light users or heavy users of EVs. Light EV users have sufficient electric energy reserves and greater application potential in V2G, while heavy EV users have a high demand for charging and greater application potential in charging scheduling;

[0008] S2. According to the clustering results, establish a demand response model for various types of EV owners to quantify the sensitivity of different EV owners to electricity prices;

[0009] Furthermore, the demand response model includes the price elasticity matrix of EV owners, which is used to quantify the sensitivity of EV owners to electricity prices at different time periods;

[0010] The formula for the price elasticity matrix is:

[0011]

[0012] where, E a (t, j) represents the price elasticity coefficient of the demand at time t with respect to the price change at time j for the a-th type of EV owner;

[0013] When the price changes in other time periods are proportional to the demand, it is represented by the cross-price elasticity coefficient E a (t, j), and the calculation formula is:

[0014]

[0015] where, ρ 0 a (j), d 0 a (t) are the original electricity price at time j and the original demand at time t for the a-th type of EV owner respectively;

[0016] On the premise of being small enough, the infinitesimal Approximately equal to Δd(t), Δρ(j), the demand quantity d is transformed into f(ρ) through the price-demand function d = f(ρ), and the calculation method of the price elasticity coefficient can be equivalently replaced by:

[0017]

[0018] Among them, the price-demand function f(ρ) adopts a linear form, f a (ρ(t)) = a a Lin +b a Lin *ρ(t), where a a Lin , b a Lin are constant coefficients. Because the characteristics of EV owners are different, the response degrees of EV owners to prices are different, so the price-demand functions of different clustered EV owners are different;

[0019] When the price change in this period is inversely proportional to the demand, a decrease in the price in this period will promote an increase in the demand quantity of the commodity, and vice versa, an increase in the price will lead to a decrease in the demand quantity of the commodity. It is represented by the self-elasticity coefficient E a (t,t), and the formula is:

[0020]

[0021] According to the characteristics of various EV owners, a demand response model for various EV owners is formulated; different EV owners have different sensitivities to prices. The characteristics of EV owners can reflect the sensitivity of EV owners to prices, and then the price-demand function of EV owners is calculated. Finally, based on the price-demand function of EV owners, a price elasticity matrix of EV owners is formulated to complete the establishment of the demand response model;

[0022] Furthermore, the demand response model of EV owners includes the charging demand response model and the V2G demand response model of EV owners;

[0023] Furthermore, the price elasticity matrix of EV owners' charging is as follows:

[0024]

[0025] By multiplying the matrix of the price elasticity matrices of various EV owners by the vector of the charging price change, the change amount of the load quantity in each period can be obtained. The formula is as follows:

[0026]

[0027] Among them, Δd a (t) represents the change amount of the demand quantity of the a-th type of EV owner in the t-th period, Δρa (j) represents the price change of the j-th moment of the a-th type of EV owner; Δρ a (j) is in percentage form, and the calculation formula is:

[0028]

[0029] where ρ a (j) is the electricity price of the a-th type of EV owner at the j-th moment after optimized calculation, ρ 0 a (j) is the initial electricity price of the a-th type of EV owner at the j-th moment;

[0030] The calculation formula for the change in the demand of the a-th type of EV owner is:

[0031]

[0032] The change in the demand Δd a (t) at the t-th moment obtained through Formula 8 is equal to the dot product of the elements in the t-th row of the elasticity matrix and the price changes in each time period, reflecting that the charging volume of the EV owner at this moment will be affected by the electricity prices at other moments of the day;

[0033] The optimized charging load of the EV is equal to the original charging load plus the load change:

[0034]

[0035] where, is the charging load of the a-th type of EV owner before optimization;

[0036] Furthermore, the V2G demand response model is:

[0037] The price elasticity matrix of EV owners for V2G is as follows:

[0038]

[0039] where, represents the price elasticity coefficient of the a-th type of EV owner for V2G at the t-th moment with respect to the price change at the j-th moment;

[0040] By multiplying the matrix of the price elasticity of each type of EV owner by the vector of the V2G price change, the change in V2G for each time period can be obtained. The formula is as follows:

[0041]

[0042] where, Δd in av (t) represents the change in V2G of the a-th type of EV owner in the t-th time period, Δρ inav (j) represents the change in the V2G compensation electricity price for the j-th moment of the av-type EV owner;

[0043] The V2G amount of the av-type EV owner after optimization The calculation method is equal to the original V2G amount plus the V2G change amount:

[0044]

[0045] S3. Based on the demand response models of various types of vehicle owners, guide the charging loads and V2G of various EV owners through prices, and combine the mathematical model of the power generation cost of the generator set, with the minimum power generation cost that meets the constraint conditions as the optimization goal, and solve the optimal power generation plan, electricity selling price, and the electricity purchase price of the power grid from EV owners;

[0046] Furthermore, construct the mathematical model of the power generation cost of the generator set in combination with the unit output cost model, including: the output characteristics and constraint conditions of each generator set, and the output characteristic formula of the generator set is:

[0047]

[0048] Among them, U i (t) is the state variable of unit i, which is 1 when the unit is turned on and 0 when the unit is turned off, and S i (t) is the start-up cost of unit i, and P i (t) is the output of unit i.

[0049] The optimization goal is to minimize the power generation cost, including the unit output cost, V2G compensation cost, and the power grid's electricity selling revenue to EVs. The optimization goal is:

[0050]

[0051] Among them, C is the power generation cost of the power grid, including the unit output cost, V2G compensation cost, and the power grid's electricity selling revenue to EVs, is the power grid's electricity selling revenue to EV owners, is the compensation cost of V2G, and ρ in av (t) = (1 + Δρ in av (j)) * ρ in,0 av (t).

[0052] Furthermore, the constraint conditions include: unit output upper and lower limit constraints, load balance constraints, spinning reserve constraints, unit ramp rate constraints, unit start-stop time constraints, electricity price constraints. Among them, the unit output upper and lower limit constraints: The operation of the generator set has power limitations and cannot exceed a certain power or be lower than a certain power. The formula for the output upper and lower limit constraints is:

[0053]

[0054] Among them, represents the minimum output of unit i, represents the maximum output of unit i.

[0055] The load balance constraint is that the power generated by the generator set is equal to the load. The load balance constraint formula is:

[0056]

[0057] Among them, n is the number of generator sets, represents the load of the EV of the a-th type of EV owner at time t, k is the number of clusters for EV owner charging, P Load (t) represents the base load at time t, is the V2G output of the av-th type of EV owner at time t; kv is the number of clusters for EV owner V2G;

[0058] The unit spinning reserve constraint is that the load in the power grid is a continuously changing quantity. To cope with sudden load changes, the unit needs to have a certain additional power generation capacity. The spinning reserve constraint formula is:

[0059]

[0060] Among them, R i (t) represents the spinning reserve of unit i at time t;

[0061] The unit ramp rate constraint is due to the mechanical characteristics of the generator set. There is an upper limit on the rate at which the unit changes its output. The unit ramp rate formula is:

[0062]

[0063] Among them, is the maximum rate of decrease in the output of unit i. ΔP i up is the maximum rate of increase in the output of unit i.

[0064] When starting and stopping large generator sets, the minimum shutdown time and minimum operation time constraints of the unit need to be considered. The unit start-stop time constraint formula is:

[0065]

[0066]

[0067] Among them, are the time when unit i has been shut down and the time when unit i has been started up within time period t, respectively; They are the minimum continuous outage time limit and the minimum continuous startup time limit for unit i, respectively.

[0068] As a special commodity, the price of electricity cannot change too much. The constraint formula for the change in the selling price of electricity is:

[0069] -0.4 < Δρ a (j) < 0.8 (21)

[0070] Among them, Δρ a (j) represents the change in the selling price of electricity from the power grid to the j-th moment of the a-type EV owner.

[0071] The constraint formula for the change in the V2G power grid's electricity purchase price is:

[0072] -0.45 < Δρ in av (j) < 1 (22)

[0073] Among them, Δρ in av (j) represents the change in the electricity purchase price of the power grid from the av-type EV owner at the j-th moment.

[0074] The beneficial effects of the present invention are:

[0075] 1. Classify EV owners based on EV owner charging and V2G data. Use the clustering algorithm to perform clustering analysis on the EV usage time, EV charging load, and V2G of EV owners respectively. Analyze the participation ability and price sensitivity of each type of EV owner in dispatching according to the clustering results. The classification of EV owners effectively reflects the characteristics of each type of EV owner and can more specifically analyze the relationship between the load and price of each type of EV owner.

[0076] 2. Conduct price guidance on EV owner charging and V2G. Based on the mathematical model of the dispatching ability and price sensitivity of EV owners, establish a demand response model for EV charging and V2G. Use price guidance to calculate the optimal selling / purchasing electricity price, considering the differences of each type of EV owner during dispatching, and better dispatching effects can be achieved.

[0077] 3. Calculate the optimal power grid operation plan. The EV owner demand response model is combined with the generator set power generation cost model, considering various constraints, and solve the optimal unit operation, electricity sales, and power grid electricity purchase plan from EV owners. Description of the Drawings

[0078] Figure 1 It is the system energy structure block diagram of an optimized dispatching method considering the differences of EV owners in an embodiment of the present invention;

[0079] Figure 2Flow chart of an optimized scheduling method considering differences among EV owners according to an embodiment of the present invention;

[0080] Figure 3 Basic load diagram of the IEEE ten - machine system according to the present invention;

[0081] Figure 4 Comparison diagram of EV charging loads between classified scheduling and unclassified scheduling when EV owners charge according to the present invention;

[0082] Figure 5 V2G comparison diagram of EV loads between classified scheduling and unclassified scheduling during V2G of the present invention. Detailed implementation manners

[0083] The present invention will be 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. Therefore, it only shows the components related to the present invention.

[0084] In the following description, terms are used only to describe specific embodiments; time - related terms, such as "peak time" means that the load is at a high point at this moment, "normal time" means that the load is near the daily average load at this moment, "valley time" means that the load is at a low point at this moment, etc.

[0085] As Figure 1 、 2 shown, an optimized scheduling method considering differences among EV owners includes the following steps:

[0086] S1. The classification of EV owners for charging is based on clustering analysis of two types of data, namely the driving time and charging data of 7000 EV owners; the classification of EV owners for V2G is based on clustering analysis of the driving time of 7000 EV owners and V2G data, and several sets of EV owners with different driving characteristics are constructed;

[0087] The driving time of an EV owner can reflect the time period during which the EV owner often drives. During the time period when the EV owner often drives, the EV owner's response degree to charge - discharge scheduling is low. Corresponding to the load time period, it can be distinguished which EV owners are peak - time driving EV owners, which are normal - time driving EV owners, and which are valley - time driving EV owners;

[0088] The charging and V2G data of an EV owner reflect the vehicle's demand for electric energy. EV owners with a large demand for electric energy generally consume more electric energy and need to supplement more electric energy, and have greater application potential in charging scheduling. EV owners with a large V2G generally have sufficient electric energy reserves in the battery and have greater application potential in V2G;

[0089] Considering that the K-Means clustering algorithm classifies data based on the distances between data points, and there are differences in the data dimensions of different classes. For example, in this application, the time data range is [0, 24], while the charging data can be much larger than 24. Therefore, directly using the original data for clustering analysis will result in insufficient influence of the time dimension data on the clustering results. To avoid this problem, this application performs standardization processing on all data, and the standardization formula is as follows:

[0090]

[0091] Among them, represents the average value of the variable, and σ represents the standard deviation of the variable.

[0092] After comparative analysis, the charging data of EV owners is most suitable to be divided into 8 categories, and the data of 8 cluster centers are shown in Table 1:

[0093] Table 1: Corresponding relationship table of 8 types of characteristic data of EV owners' charging, EV owners' vehicle usage time, and charging data:

[0094]

[0095] The 8 types of EV owners are divided into: The 1st and 8th types of EV owners are EV owners with large charging loads, the 5th type of EV owner is an EV owner with a relatively large charging load, the 2nd, 6th, and 7th types of EV owners are EV owners with small charging loads, and the 3rd and 4th types of EV owners are EV owners with small charging loads.

[0096] The V2G data of EV owners is most suitable to be divided into 4 categories, and the data of 4 cluster centers are shown in Table 2:

[0097] Table 2: Corresponding relationship table of 4 types of characteristic data of EV owners' discharging, EV owners' vehicle usage time, and discharging data:

[0098]

[0099] The 4 types of EV owners are divided into: The 1st, 2nd, and 3rd types of EV owners are EV owners with large V2G types, and the 4th type of EV owner is an EV owner with a small V2G type.

[0100] S2. Based on the clustering results, establish a demand response model for various types of EV owners to quantify the sensitivity of different EV owners to electricity prices;

[0101] Furthermore, the response model includes the price elasticity matrix of EV owners, which is used to quantify the sensitivity of EV owners to electricity prices in each time period;

[0102] Time-of-use electricity price is currently adopted in China. In this application, one hour is taken as a node, and each node has a corresponding EV load, electricity selling price, and V2G compensation electricity price. Since the time interval is small, the load at a certain moment is affected by the price changes at each moment within a day. Therefore, the responsiveness of EV owners to price can be quantified as a price elasticity matrix, and the formula for the price elasticity matrix is:

[0103]

[0104] Among them, E a (t, j) represents the price elasticity coefficient of the demand at time t with respect to the price change at time j for the a-th type of EV owner;

[0105] When the price change in other periods is proportional to the demand, it is represented by the cross-price elasticity coefficient E a (t, j), and the calculation formula is:

[0106]

[0107] Among them, ρ 0 a (j), d 0 a (t) are respectively the original electricity price at time j and the original demand at time i for the a-th type of EV owner;

[0108] On the premise of being small enough, the differential element is approximately equal to Δd(i), Δρ(j), and the demand quantity d is transformed into f(ρ) through the price-demand function d = f(ρ).

[0109] The calculation method of the price elasticity coefficient can be equivalently replaced by:

[0110]

[0111] Among them, the price-demand function f(ρ) adopts a linear form, f a (ρ(t)) = a a Lin +b a Lin *ρ(t), where a a Lin , b a Lin are constant coefficients. Since the characteristics of EV owners are different and the responsiveness of EV owners to price is different, the price-demand functions of different clustered EV owners are different;

[0112] When the price change in this period is inversely proportional to the demand, a decrease in the price in this period will promote an increase in the demand quantity of goods, and conversely, an increase in the price will lead to a decrease in the demand quantity of goods. It is represented by the self-elasticity coefficient E a (t, t), and the formula is:

[0113]

[0114] According to the classification results of the K-Means clustering algorithm, analyze the characteristics of EV owners in each category through the data analysis of each cluster center. During scheduling, EV owners with high loads have greater scheduling potential, and because of the large demand, these EV owners are also more sensitive to price. The constant coefficients of each type of EV owner are shown in Table 3:

[0115] Table 3: Relationship Table of Charging Constant Coefficients of EV Owners in Each Category

[0116]

[0117] Based on the characteristics of each type of EV owner, establish a demand response model for each type of EV owner; different EV owners have different sensitivities to price. The characteristics of EV owners can reflect their sensitivity to price, and then calculate the price-demand function of EV owners. Finally, based on the price-demand function of EV owners, establish the price elasticity matrix of EV owners;

[0118] The demand response model of EV owners includes a charging demand response model and a V2G demand response model;

[0119] Furthermore, the price elasticity matrix of EV owners for charging is as follows:

[0120]

[0121] By multiplying the matrix of the price elasticity matrix of each type of EV owner obtained by the vector of price changes, the change in the load volume at each time period can be obtained. The formula is as follows:

[0122]

[0123] where, Δd a (i) represents the change in demand of the a-th type of EV owner at the i-th time period, and Δρ a (j) represents the change in price of the a-th type of EV owner at the j-th moment; in order to separate the coefficients, Δρ a (j) is in percentage form, and the calculation formula is:

[0124]

[0125] where, ρ a (j) is the electricity price of the a-th type of EV owner at the j-th moment after optimization calculation, and ρ 0 a (j) is the initial electricity price of the a-th type of EV owner at the j-th moment;

[0126] The calculation formula for the change in demand of the a-th type of EV owner is:

[0127]

[0128] The change in the demand at the t-th moment is obtained through Equation 8 and is equal to the elements in the t-th row of the elasticity matrix multiplied by the changes in the prices of each time period, reflecting that the charging amount of EV owners at this moment will be affected by the electricity prices at other moments of the day;

[0129] The optimized EV load is equal to the original load plus the load change:

[0130]

[0131] Furthermore, the V2G demand response model:

[0132] The price elasticity matrix of EV owners for V2G is as follows:

[0133]

[0134] where represents the price elasticity coefficient of the av-type EV owners for the V2G at the t-th moment with respect to the price change at the j-th moment;

[0135] By multiplying the matrix of price elasticities of various types of EV owners by the vector of V2G price changes, the changes in V2G for each time period can be obtained. The formula is as follows:

[0136]

[0137] where Δd in av (t) represents the change in V2G of the av-type EV owners at the t-th time period, and Δρ in av (j) represents the change in the V2G compensation electricity price of the av-type EV owners at the j-th moment;

[0138] The V2G amount of the av-type EV owners after optimization is calculated as equal to the original V2G amount plus the V2G change:

[0139]

[0140] S3. Based on the demand response models of various types of vehicle owners, guide the charging loads and V2G of various EV owners through prices, and combine with the mathematical model of the power generation cost of the generator set. With the minimum power generation cost that meets the constraint conditions as the optimization goal, solve the optimal power generation plan, electricity selling price, and the price at which the power grid buys electricity from EV owners;

[0141] Furthermore, based on the demand response models of various EV owners, an optimal scheduling model considering the differences of EV owners is constructed by combining the unit output cost model. The unit model represents the output characteristics of each generator set;

[0142] The fuel cost of the unit is represented by the output characteristics of each generator set, and the formula is:

[0143]

[0144] Among them, U i (t) is the state variable of unit i, which is 1 when the unit is on and 0 when the unit is off. S i (t) is the start-up cost of unit i, and P i (t) is the output of unit i;

[0145] The optimization objective is to minimize the power generation cost C, including the unit output cost, V2G compensation cost, and the power grid's revenue from selling electricity to EVs. The optimization objective is:

[0146]

[0147] Among them, C is the power grid power generation cost, including the unit output cost, V2G compensation cost, and the power grid's revenue from selling electricity to EVs. is the power grid's revenue from selling electricity to EV owners. is the cost of the power grid buying electricity from EV owners. ρ in av (t)=(1 + Δρ in av (j)) * ρ in,0 av (t).

[0148] Furthermore, the constraint conditions include: unit output upper and lower limit constraints, load balance constraints, spinning reserve constraints, unit ramp rate constraints, unit start-stop time constraints, and electricity price constraints. Among them, the unit output upper and lower limit constraints: the operation of the generator set has power limitations, and it cannot exceed a certain power or be lower than a certain power. The formula for the output upper and lower limit constraints is:

[0149]

[0150] Among them, represents the minimum output of unit i. represents the maximum output of unit i.

[0151] The load balance constraint is that the electricity generated by the generator set is equal to the load. The formula for the load balance constraint is:

[0152]

[0153] Among them, The V2G output of EV owners of class av at time t, P Load (t) represents the base load at time t, represents the load of the EV of the a - type EV owner at time t;

[0154] The unit spinning reserve constraint: Since the load in the power grid is a continuously changing quantity, in order to cope with sudden load changes, the unit needs to have a certain additional power generation capacity. The spinning reserve constraint formula is:

[0155]

[0156] Among them, R i (t) represents the spinning reserve of unit i at time t;

[0157] The unit ramp rate constraint: Due to the mechanical characteristics of the generator set, there is an upper limit on the rate at which the unit changes its output. The unit ramp rate formula is:

[0158]

[0159] Among them, is the maximum rate of decrease in the output of unit i. ΔP i up is the maximum rate of increase in the output of unit i.

[0160] For large - scale generator sets, starting and stopping need to consider the minimum shutdown time and minimum running time constraints of the unit. The unit start - stop time constraint formula is:

[0161]

[0162]

[0163] Among them, are respectively the time that unit i has been shut down and the time that unit i has been started up within time period t; are respectively the minimum continuous shutdown time limit and the minimum continuous startup time limit for unit i;

[0164] As a special commodity, the price of electricity cannot change too much. The selling - electricity price change constraint formula is:

[0165] - 0.4 < Δρ a (j) < 0.8 (21)

[0166] Among them, Δρ a (j) represents the change in the selling - electricity price of the power grid at time j for the a - type EV owner;

[0167] The V2G compensation electricity - price change constraint formula is:

[0168] - 0.45 < Δρin av (j) < 1 (22)

[0169] where Δρ in av (j) represents the price change of the grid power receiving price at the j-th moment for the EV owners of the av-th category;

[0170] The base load of the IEEE ten-machine system adopted is as Figure 3 shown:

[0171] It can be seen that there are two peaks at 12 o'clock and 20 o'clock, the lowest point appears at 1 o'clock, and a trough is formed at 17 o'clock.

[0172] The EV load has strong schedulability. Appropriate guidance for the EV load can effectively reduce the gap between the peak and valley of the grid load. The charging scheduling effect is as Figure 4 shown:

[0173] Figure 4 It can be seen that the classified scheduling significantly reduces the load during the period from 6 o'clock to 12 o'clock, and this time period coincides with the first peak period of the base load, indicating that the classified scheduling has an obvious effect on peak shaving.

[0174] Reverse power transmission from the vehicle to the grid, V2G, can relieve the power generation pressure of the generator sets during the peak of the grid load. The guiding effect is as Figure 5 shown. From the V2G curves of classified guidance and unclassified guidance for EV owners, it can be seen that the classified guidance has a better peak shaving and valley filling effect, and can guide the discharge of EV owners during the valley period to the peak period; especially at the peak at 20 o'clock, the classified guidance can enable EV owners to share a greater load pressure.

[0175] Inspired by the ideal embodiments of the present invention described above, through the above description, relevant staff can completely make various changes and modifications within the scope not deviating from the technical idea of this invention. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An optimized scheduling method considering the differences of EV owners, characterized in that, It includes the following steps: S1. Conduct cluster analysis on the driving time, charging, and V2G data of EV owners respectively to construct several sets of EV owners with different driving characteristics; S2. Based on the clustering results, establish a demand response model for each type of EV owner to quantify the sensitivity of different EV owners to electricity prices; S3. Based on the demand response models of each type of owner, guide the charging load and V2G of each type of EV owner through price, and combine the mathematical model of the power generation cost of the generator set. Taking the minimum power generation cost that meets the constraint conditions as the optimization goal, solve the optimal power generation plan, electricity selling price, and V2G compensation electricity price; The demand response model is used to quantify the impact of electricity price changes on the charging load and V2G of EV owners through the price elasticity matrix of EV owners; The demand response model includes an EV owner charging demand response model and a V2G demand response model; The EV owner charging demand response model is based on the price elasticity matrix. The charging of EV owners is divided into a categories, with a total of a charging price elasticity matrices: Among them, E a (t, j) represents the price elasticity coefficient of the demand at time t with respect to the price change at time j for EV owners of type a; To obtain the price elasticity matrix of each type of EV owner, multiply the matrix by the vector of the change in the electricity selling price to obtain the change in the load volume at each time period. The formula is as follows: where, Δd a (t) represents the change in demand of the a-th type of EV owners in the t-th period, and Δρ a (j) represents the change in price of the a-th type of EV owners at the j-th moment; Δρ a (j) is in percentage form, and the calculation formula is: Among them, ρ a (j) is the electricity price of the a-th type of EV owner at the j-th moment after optimized calculation, and ρ 0 a (j) is the initial electricity price of the a-th type of EV owner at the j-th moment; The calculation formula for the change in the demand of the a-th type of EV owner is: The change in the demand volume Δd a (t) is equal to the product of the elements in the t-th row of the elasticity matrix and the price changes in each time period, reflecting the impact of electricity prices at other times of the day on the charging volume of EV owners at this moment; Optimized EV load Equal to the original load plus the load change amount: Among them, is the load of Class A EV owners before optimization; The V2G demand response model is based on the price elasticity matrix. The V2G of EV owners is divided into av categories, with a total of av V2G elasticity matrices: To obtain the price elasticity matrix of each type of EV owner, multiply the matrix by the vector of the change in the V2G compensation electricity price to obtain the change in V2G at each time period. The formula is as follows: where, Δd in av (t) represents the change in V2G of the av - type EV owners in the t - th time period, Δρ in av (j) represents the change in the V2G compensation electricity price of the av - type EV owners at the j - th moment, also in percentage form; V2G volume of Class av EV owners after optimization The calculation method is equal to the original V2G volume plus the V2G change volume: Among them, is the original V2G amount of Class av EV owners; The mathematical model of the power generation cost of the generator set includes: the output characteristics and constraint conditions of each generator set. The formula for the output characteristics of the generator set is: Among them, U i (t) is the state variable of unit i, which is 1 when the unit is on and 0 when the unit is off. S i (t) is the start-up cost of unit i, and P i (t) is the output of unit i; The optimization goal of the minimum power generation cost includes: the unit output cost, the V2G compensation cost, and the revenue from selling electricity from the power grid to EVs. The optimization goal is: Among them, C is the power generation cost of the power grid, including the unit output cost, the V2G compensation cost, and the power selling revenue from the power grid to EVs. is the power selling revenue from the power grid to EV owners, ρ a (t) = (1 + Δρ a (t)) * ρ 0 a (t), is the V2G compensation cost, ρ in av (t) = (1 + Δρ in av (j)) * ρ in,0 av (t); The constraint conditions include: the upper and lower limits of unit output constraints, load balance constraints, spinning reserve constraints, unit ramp rate constraints, unit start-stop time constraints, and electricity price constraints; The load balance constraint is that the electricity generated by the generator set is equal to the load volume. The formula for the load balance constraint is: where n is the number of generating units, represents the load of the EV of the ath type of EV owner at time t, k is the number of clusters of EV owners' charging, P Load (t) represents the base load at time t, is the V2G output of the avth type of EV owner at time t, and kv is the number of clusters of EV owners' V2G.

2. The optimized scheduling method considering EV owner differences according to claim 1, characterized in that: The cluster analysis classifies EV owners according to the driving time and charging data of EV owners through the K-Means clustering algorithm.

Images