Analysis Method for Uncertainty of Electric Vehicle Charging Response Based on Three-Way Decision Theory

Abstract

The present invention relates to a method for analyzing the uncertainty of electric vehicle charging response based on the three-way decision theory, including: constructing the cost losses of different electric vehicles taking different decision-making behaviors in different states; calculating the three-way decision thresholds according to the cost losses; calculating the decision state values of electric vehicles; judging the decision-making behaviors taken by electric vehicles according to the decision state values and the three-way decision thresholds, and judging the type of demand response participated by electric vehicles according to the decision-making behaviors: calculating the charging loads of each electric vehicle one by one according to the decision-making behaviors taken by electric vehicles; and accumulating the charging loads of all electric vehicles to obtain the total charging load.

Description

Technical Field

[0001] The present invention relates to a method for analyzing the uncertainty of electric vehicle charging response based on the three-way decision theory, and belongs to the field of electric vehicle charging response analysis. Background Art

[0002] The access of a large number of electric vehicles will bring new challenges to the planning, design and operation of the power grid. Due to the strong subjectivity of user charging, it will further increase the randomness and uncertainty of the power grid operation, causing certain troubles to the load forecasting, dispatching operation, etc. of the power grid. Therefore, it is necessary to analyze the change of the power grid load after the access of electric vehicles. At present, the research on large-scale electric vehicle charging load mainly adopts the method of Monte Carlo simulation of electric vehicle charging, calculates the charging probability of electric vehicles at a certain moment, determines the charging time and charging energy of electric vehicles, and obtains the charging load curve of electric vehicles.

[0003] However, the charging behavior of users is uncertain. In response to the uncertainty of electric vehicle response, existing research mainly focuses on the key factor of "user behavior" and conducts qualitative analysis. There are the following several strategies: (1) Considering the uncertainty of electric vehicle user charging behavior and energy consumption demand, the capacity and risk of electric vehicles participating in the frequency modulation auxiliary service market are analyzed. (2) A polyhedron uncertain set is used to describe the volatility and uncertainty of electric vehicle charging load. (3) Fuzzy rules are used to analyze the uncertain relationship among the driving mileage, departure time and arrival time of electric vehicles. (4) An electric vehicle user response model based on Weber-Fechner's law is established to participate in frequency response in the form of a virtual power plant. (5) On the basis of the above strategies, considering the influence of factors such as user benefits and vehicle use requirements, the sigmoid cloud model is used to analyze the stochastic mapping relationship between the user response benefits of electric vehicle users and the user response behavior. The above methods first make more subjective qualitative assumptions about the user response decision-making behavior, and there is less quantitative analysis of the uncertainty of electric vehicle charging. Secondly, the analysis of user uncertainty factors mostly adopts probability models, which have relatively high requirements for the quantity and quality of data. Therefore, using probability distribution to fit the uncertain factors in user response behavior, its accuracy and credibility are still lacking.

[0004] The paper "Calculation Method of Electric Vehicle Charging Load with Coupling Characteristics" by Yang Bing, Wang Lifang, Liao Chenglin, etc. discloses that: using a kernel density function to replace a deterministic probability distribution function to fit the driving law of electric vehicles, using a multi-dimensional probability distribution function to generate random numbers of driving laws with coupling characteristics, and using the charging probability to represent the uncertain charging behavior to predict the future charging load of large-scale residential private electric vehicles. In this paper, the binomial distribution probability is used to represent the uncertain charging behavior of electric vehicle users, and its accuracy and credibility need to be improved. Summary of the Invention

[0005] To overcome the problems existing in the prior art, the present invention designs an analysis method for the uncertainty of electric vehicle charging response based on the three-way decision theory, constructs a clustered decision-making information system set for electric vehicle charging behavior, quantifies and classifies the uncertainty of electric vehicle user decisions using the three-way decision threshold, determines the type of demand response participated by the electric vehicle according to the decision of the electric vehicle user, quantitatively analyzes the charging load after the response of various electric vehicle users, and promotes the efficient allocation of electric vehicle charging resources.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] An analysis method for the uncertainty of electric vehicle charging response based on the three-way decision theory includes the following steps:

[0008] Construct the cost losses of different electric vehicles taking different decision-making behaviors in different states;

[0009] According to the cost losses, calculate the three-way decision thresholds α and β;

[0010] Calculate the decision state value P(X|[x]) of the electric vehicle;

[0011] According to the decision state value and the three-way decision thresholds α and β, judge the decision-making behavior of the electric vehicle, and according to the decision-making behavior, judge the type of demand response participated by the electric vehicle:

[0012] If P(X|[x])≥α, the electric vehicle takes an active response behavior, that is, charges, and it is considered that the electric vehicle participates in the price-based demand response;

[0013] If β<P(X|[x])<α, the electric vehicle takes a delayed response behavior, it is considered that the electric vehicle participates in the incentive-based demand response, and it is judged whether the electric vehicle charges by solving a multi-objective function with the goal of maximizing the interests of electric vehicles participating in different types of demand responses and minimizing the grid load fluctuation;

[0014] If P(X|[x])≤β, the electric vehicle takes a rejection response behavior and does not participate in the demand response, that is, does not charge, and it is considered that the electric vehicle does not participate in the demand response and charges disorderly;

[0015] According to the decision-making behavior of the electric vehicle, calculate the charging load of each electric vehicle one by one; accumulate the charging loads of all electric vehicles to obtain the total charging load.

[0016] Furthermore, the calculation of the three-way decision thresholds α and β is expressed by the formula:

[0017]

[0018]

[0019] In the formula, λ PP , λ BP and λ NP are the cost losses of taking decision-making actions a P , a B and a N when the object belongs to state X; λ PN , λ BN and λ NN are the cost losses of taking decision-making actions a P , a B and a N when the object does not belong to state X; λ PP ≤λ BP ≤λ NP and λ NN ≤λ BN ≤λ PN and (λ BP -λ PP )(λ BN -λ NN ) < (λ PN -λ BN )(λ NP -λ BP ).

[0020] Furthermore, it also includes: iteratively changing the cost loss, recalculating the three-way decision thresholds α and β, so as to obtain the grid load calculation results under different decision boundary regions; taking the three-way decision thresholds α and β corresponding to the optimal one among the grid load calculation results.

[0021] Furthermore, statistically calculate the probability that the electric vehicle is in the grid-connected state as the decision state value of the electric vehicle.

[0022] Furthermore, construct the multi-objective function, which is expressed by the formula as:

[0023] max(F P +F I -F B ) + minD(L(t))

[0024]

[0025]

[0026]

[0027] In the formula, F P is the cost saved by the electric vehicle participating in the price-based demand response; F I is the cost saved by the electric vehicle participating in the incentive-based demand response; FB The loss cost of the electric vehicle battery after the electric vehicle participates in demand response; D() represents the variance of load fluctuation; L(t) represents the total load of the distribution network at time t; N represents the number of electric vehicles; F o,i represents the cost (required) generated by the unordered charging of vehicle i; T s 、T t respectively represent the time when the electric vehicle connects to the power grid and leaves the power grid; η c 、η d respectively represent the charging and discharging efficiency of the electric vehicle;

[0028] P c,i (t), P d,i (t) respectively represent the charging power and discharging power of vehicle i at time t; p(t) represents the time-of-use electricity price at time t; Q I,i represents the load transfer amount generated by the electric vehicle participating in the incentive-based demand response compared with unordered charging; ω1 and ω2 respectively represent the first incentive amount coefficient and the second incentive amount coefficient; μ1, μ2, and μ3 respectively represent the first battery loss coefficient, the second battery loss coefficient, and the third battery loss coefficient; y i (t) represents the charging and discharging state of vehicle i at time t, y i (t) = ±1 indicates charging or discharging, y i (t) = 0 indicates not working.

[0029] Compared with the prior art, the present invention has the following characteristics and beneficial effects:

[0030] In the prior art, probability distribution fitting is mostly used to handle the uncertain factors in user response behavior, and its accuracy and credibility are still lacking. The present invention constructs a clustered decision-making information system set for electric vehicle charging behavior, quantifies and classifies the uncertainty of electric vehicle user decisions using three-way decision thresholds, determines the type of demand response they participate in according to the decisions of electric vehicle users, and quantitatively analyzes the charging load after various electric vehicle users respond, promoting the efficient allocation of electric vehicle charging resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is the flowchart of charging load calculation;

[0032] Figure 2 is the simulation flowchart;

[0033] Figure 3 is the schematic diagram of load curves of five typical scenarios;

[0034] Figure 4 is the schematic diagram of load curves of various models;

[0035] Figure 5 is the schematic diagram of load transfer amounts of each decision branch;

[0036] Figure 6 It is a schematic diagram of the load transfer amount in each functional area. Detailed implementation manners

[0037] The present invention will be described in more detail below in conjunction with embodiments.

[0038] Embodiment 1

[0039] An analysis method for the uncertainty of electric vehicle charging response based on the three-way decision theory includes the following steps:

[0040] S1. Construct a charging information system based on Pawlak rough sets:

[0041] Define the electric vehicle object set U and the charging behavior attribute set A. U and A are finite sets, and the charging information system of the electric vehicle is (U, A). Define an information function for each charging behavior attribute a ∈ A: U → V a , where V a = {a(x): x ∈ U} is the information function value set of the charging behavior attribute a. A = C ∪ {d}, C is the conditional attribute value, and d is the decision attribute set. In this embodiment, the electric vehicle object set U is obtained from the historical data of electric vehicles.

[0042] S2. Construct the cost loss of user decision-making behavior based on the three-way decision:

[0043] Given the charging information system (U, C ∪ {d}), set the state set indicating that the object belongs to state X or the complement of state X. In this embodiment, state X is that the electric vehicle is connected to the power grid; if the electric vehicle charges disorderly, it will start charging immediately after being connected to the power grid; if the electric vehicle participates in demand response, it will decide whether to charge according to the specific situation of the load, and more electric vehicles will charge during the load valley. Construct the decision behavior set Λ = {a P , a B , a N} to represent three decision-making behaviors: active response, delayed response, and rejection response.

[0044] Construct the cost loss generated by taking different decision-making behaviors in different states, as shown in Table 1. In Table 1, λ PP , λ BP and λ NP are the cost losses of taking decision-making behaviors a P , a B and a N when the object belongs to state X; λ PN , λ BN and λ NN are the cost losses of taking decision-making behaviors a when the object does not belong to state X.P , a B and a N cost loss.

[0045] Table 1 Decision Cost Matrix

[0046]

[0047] For the object [x], the cost losses generated by taking different actions are calculated through Equation (1):

[0048]

[0049] In the formula, R(a P |[x]) represents the cost loss generated by taking the decision behavior a P for the object [x]; R(a B |[x]) represents the cost loss generated by taking the decision behavior a B for the object [x]; R(a N |[x]) represents the cost loss generated by taking the decision behavior a N for the object [x]; P(X|[x]) is the evaluation function representing the probability that the object [x] is in the state X; represents the probability that the object [x] is in the state .

[0050] According to the Bayesian minimum risk decision principle, the three-way decision rules (P positive region, B boundary region, N negative region) are shown in Table 2.

[0051] Table 2 Three-way Decision Rule Table

[0052]

[0053] In the table, POS(X) represents the set of electric vehicles that take positive response actions; BND(X) represents the set of electric vehicles that take delayed response actions; NEG(X) represents the set of electric vehicles that reject responses.

[0054] Considering the most likely situation in practice, when the electric vehicle is in the state X, it takes a positive response; when it is in the state , it takes a rejection response; that is, λ PP ≤λ BP ≤λ NP and λ NN ≤λ BN ≤λ PN . Assuming that the decision boundary region exists, that is, it satisfies: (λ BP -λ PP )(λ BN -λ NN )<(λ PN -λBN )(λ NP -λ BP ),then the above decision rules P to N can be further simplified to Table 3, obtaining the three-way decision thresholds α and β, which are expressed by the formula as follows:

[0055]

[0056] Table 3 Simplified Table of Three-Way Decision Rules

[0057]

[0058] S3. According to the historical charging information of electric vehicles, count the probability that the electric vehicle [x] is in state X as the decision state value of the electric vehicle;

[0059] S4. According to the decision state value, judge the decision behavior taken by the electric vehicle, and according to the decision behavior, judge the type of demand response participated by the electric vehicle:

[0060] If P(X|[x])≥α, then x∈POS(X), and the electric vehicle takes an active response behavior, i.e., charging, and it is considered that the electric vehicle participates in price-based demand response.

[0061] If β<P(X|[x])<α, then x∈BND(X), and the electric vehicle takes a delayed response behavior. By solving the multi-objective function with the goal of maximizing the interests of the electric vehicle and minimizing the grid load fluctuation, judge whether the electric vehicle charges, and it is considered that the electric vehicle participates in incentive-based demand response.

[0062] If P(X|[x])≤β, then x∈NEG(X), that is, the electric vehicle takes a rejection response behavior and does not participate in demand response, i.e., does not charge, and it is considered that the electric vehicle does not participate in demand response and charges disorderly.

[0063] S5. Judge whether each electric vehicle charges one by one; accumulate the charging loads of all the charging electric vehicles to obtain the total charging load; superimpose the baseline load and the total charging load to obtain the total grid load.

[0064] Embodiment 2

[0065] Furthermore, set the maximum number of iterations, iteratively change the cost loss, and recalculate the three-way decision thresholds α and β through Equation 2, so as to adjust the decision boundary domain to obtain the grid load calculation results under different decision boundary domains until the maximum number of iterations is reached; compare the multiple grid load calculation results obtained through multiple iterations, and select the optimal one (with the smallest error) among them.

[0066] Embodiment 3

[0067] Construct a multi-objective function with the goal of maximizing the benefits of electric vehicles and minimizing the grid load fluctuation, which is expressed by the formula as follows:

[0068] max(F P +F I -F B )+minD(L(t))

[0069]

[0070]

[0071]

[0072] In the formula, F P is the cost saved by electric vehicles participating in price-based demand response; F I is the cost saved by electric vehicles participating in incentive-based demand response; F B is the loss cost of the electric vehicle battery after the electric vehicle participates in demand response; N represents the number of electric vehicles; F o,i represents the (required) cost generated by the unordered charging of vehicle i; T s and T t respectively represent the time when the electric vehicle connects to the grid and leaves the grid; η c and η d respectively represent the charging and discharging efficiencies of the electric vehicle; P c,i (t) and P d,i (t) respectively represent the charging power and discharging power of vehicle i at time t; p(t) represents the time-of-use electricity price at time t; Q I,i represents the load transfer amount generated by the electric vehicle participating in incentive-based demand response compared to unordered charging; ω1 and ω2 respectively represent the first incentive amount coefficient and the second incentive amount coefficient; μ1, μ2, and μ3 respectively represent the first battery loss coefficient, the second battery loss coefficient, and the third battery loss coefficient; y i (t) represents the charging and discharging state of vehicle i at time t. When y i (t) = ±1, it represents charging or discharging. When y i (t) = 0, it represents not working.

[0073]

[0074] In the formula, L(t) represents the total load of the distribution network at time t, E() represents the variance; D() represents the load fluctuation variance.

[0075] Set the following constraint conditions, including the electric vehicle battery target charge state (SOC, State of Charge) constraint, charging and discharging power constraint, charging demand constraint, and transformer capacity constraint:

[0076] S i S(t) = S i (t - 1)+ [P c,i (t)η c - P d,i (t - 1) / η d Δt / C i

[0077] S min ≤ S i (t) ≤ S max

[0078] y i (t)P cmin ≤ P c,i (t) ≤ y i (t)P cmax , y i (t) ∈ {0, 1}

[0079] y i (t)P dmin ≤ P d,i (t) ≤ y i (t)P dmax , y i (t) ∈ {0, -1}

[0080] S i (t) ≥ S l,i

[0081] L(t) ≤ k T A T , t = 1, 2, …, 96

[0082] Wherein, S i (t) represents the SOC level of vehicle i at time t; C i represents the battery capacity of vehicle i; S min , S max represent the minimum and maximum values of the SOC of the electric vehicle; P cmin , P cmax respectively represent the minimum rated charging power and the maximum rated charging power of the electric vehicle; P dmin , P dmax respectively represent the minimum rated discharging power and the maximum rated discharging power of the electric vehicle; S l,i represents the expected SOC value when vehicle i is off - grid; k T represents the transformer efficiency; A T represents the rated capacity of the transformer.

[0083] Solve the multi-objective function to obtain parameters such as the charging and discharging states, grid connection time, grid disconnection time, charging and discharging efficiency, and charging and discharging power of electric vehicles participating in incentive-based demand response at time t, so as to determine whether the electric vehicle charges or not.

[0084] Example 4

[0085] A method for calculating the charging load of electric vehicles based on the three-way decision theory includes the following steps:

[0086] 1. Adopt the probability distribution of the driving characteristics of electric vehicles (the time of vehicle grid connection and disconnection follows a normal distribution of (17.1, 3.3) and (8.91, 3.24), and the daily driving mileage of the vehicle follows a lognormal distribution of (3.31, 0.87)). Through Monte Carlo sampling, obtain the battery capacity, charging power, grid connection and disconnection times, and target SOC levels of each electric vehicle; calculate the initial SOC level of the electric vehicle:

[0087]

[0088] In the formula, S 0,i represents the initial SOC level of vehicle i; R i represents the daily driving mileage of vehicle i; W i represents the energy consumption per kilometer of vehicle i.

[0089] 2. According to the historical charging data of electric vehicle charging piles, establish a charging information system, determine the three-way decision threshold, and calculate the decision state value of the current electric vehicle, that is, the probability that the grid is in the grid-connected state.

[0090] 3. According to the decision state value, judge the decision behavior of the current electric vehicle:

[0091] If α ≤ P(X|[x]), it is considered that the electric vehicle implements an active response behavior, that is, charging;

[0092] If α < P(X|[x] < β, it is considered that the electric vehicle implements a delayed response behavior, and judge whether to charge through the multi-objective function.

[0093] If P(X|[x]) ≤ β, it is considered that the electric vehicle implements a rejection response behavior, that is, not charging.

[0094] 4. According to whether the electric vehicle charges judged in step 3 and the battery capacity, charging power, grid connection and disconnection times, initial SOC level, and target SOC level of the electric vehicle obtained in step 1, calculate the charging load situation of each electric vehicle; accumulate the charging loads of all electric vehicles to obtain the total charging load curve.

[0095] Example 5

[0096] Taking the actual load of charging piles in a certain area of Shanghai as historical data. Assume that the load of a certain area's distribution network consists of two parts: the baseline load and the electric vehicle cluster load. Among them, the baseline load refers to the typical daily load curve of residents in a certain place, the rated capacity of the transformer is 35000 kVA, and the efficiency is 0.95. There are 500 electric vehicles in the area. Assume that the charging and discharging power of electric vehicles are 4 kW for slow charging and 12 kW for fast charging, the charging and discharging efficiency is 0.95, the maximum and minimum values of SOC are 1 and 0 respectively, the battery capacity follows a uniform distribution U(30, 40) kWh, and the expected value of SOC for electric vehicle users to complete charging and leave the grid is 0.95. The incentive amount coefficients are taken as 1 yuan / (MWh) 2 and 85 yuan / MWh respectively, and the battery loss coefficients are taken as 0.3, 0.2, -0.2 yuan respectively. The time-of-use electricity price is shown in Table 4. The simulation time granularity is 15 minutes.

[0097] Table 4 Time-of-use electricity price

[0098]

[0099] In the process of electric vehicle users choosing whether to participate in demand response, the judgment indicators come from various factors such as the vehicle's SOC level, charging access time, driving mileage, charging power, etc. This article mainly considers the following main factors: charging access time D1, starting SOC level D2, vehicle leaving time D3, battery capacity D4, charging power D5. Construct a decision attribute set through the above charging behavior information. By gradually changing the cost loss, changing the values of α and β to narrow the boundary domain, obtaining decision domains of different sizes, and finally becoming a two-branch decision. In this process, analyze the load changes of various scenarios step by step to find the optimal threshold. Therefore, first set multiple user selection application scenarios for simulation, compare and analyze the simulation results, set five groups of cost losses, and calculate the decision threshold pairs (α, β). The specific parameters are shown in Table 5. When it is less than 0.3, reject the response and adopt an unordered charging strategy; when the probability evaluation value of the electric vehicle is greater than 0.7, it is an active response and adopts a price-based demand response; the remaining electric vehicles are delayed responses and adopt an incentive-based demand response.

[0100] Table 5 Scenario parameters

[0101]

[0102] Using different cost loss coefficients and under different decision threshold settings, the comparison of the electric vehicle charging load prediction curves is as Figure 3 shown: Five typical scenarios are selected, which are: 1) α = 0.5, β = 0.5 (i.e., two-branch decision); 2) α = 0.6, β = 0.4; 3) α = 0.8, β = 0.2; 4) α = 0.9, β = 0.1; 5) α = 0.7, β = 0.3.

[0103] The maximum peak-valley difference percentage is used as an index to evaluate the peak-valley difference level of each load curve:

[0104]

[0105] where L max is the maximum value of the load curve, and L min is the minimum value of the load curve.

[0106] It can be seen from the simulation results that as the cost loss coefficient increases, α gradually increases and β gradually decreases, indicating that the user accuracy of three-way decision classification is greater. In the parameter scenario where α is relatively large and β is relatively small, a large number of users match the incentive response, and the peak-valley difference of the load curve further increases; in the parameter scenario where α is relatively small and β is relatively large, a large number of users participate in disorderly charging and respond to price signals, and the load curve tends to two-way decision-making, weakening the significance of delayed response.

[0107] Table 6 Evaluation of Load Curves in Application Scenarios

[0108]

[0109] As shown in Table 6, when α = 0.7 and β = 0.3, the peak and valley values of the grid load are 2.83 MW and 1.72 MW respectively, the maximum peak-valley difference is 39.2%, and the load fluctuation variance is 0.1159. In this scenario, the total grid load fluctuation and the peak-valley difference of the load are the smallest. When α = 0.6 and β = 0.4, a large number of users complete charging under the stimulation of price signals, and a small number of users intermittently connect to the grid under the stimulation of incentive signals, resulting in multiple load fluctuations; when α = 0.8 and β = 0.2, the load curve has no obvious effect of peak shaving and valley filling, and the peak and valley loads are still relatively obvious; when α = 0.9 and β = 0.1, it indicates that a large number of electric vehicle users are classified as delayed response states, respond to incentive signals, and almost lose the influence of off-peak electricity prices, resulting in a further increase in the peak-valley difference; when α = 0.5 and β = 0.5, it is a two-way decision. At this time, without the stimulation of incentive response, the price factor plays a dominant role, resulting in a new load peak at 4 am. Therefore, in different scenarios, the load curve will change greatly, which shows that the decision-making of electric vehicle user response will directly affect the amount of load transfer in demand response. The application of the three-way decision theory has a guiding role in the analysis of the uncertainty of electric vehicle user response.

[0110] Compare the electric vehicle charging load curve based on the proposed method with the disorderly charging strategy and the time-of-use electricity price strategy, as Figure 4 shown. The maximum peak-valley differences of each model are shown in Table 7.

[0111] Table 7 Maximum Peak-Valley Difference Percentage

[0112]

[0113] The simulation results show that under the condition of disordered charging, the load curve has obvious fluctuations with large peak-valley differences. The charging peak of electric vehicles overlaps with the peak period of the basic load, resulting in an overloaded distribution network and even the situation of transformer over-limit. The load curve based on the time-of-use electricity price strategy alleviates the phenomenon of transformer over-limit under disordered charging and reduces the peak-valley difference. However, within the electricity price range during the valley period from 0:00 to 7:00, a large number of electric vehicles are connected to the grid, resulting in a load peak. In the simulation results of the charging load curve based on three-way decisions, some electric vehicle users with delayed decisions are motivated by the economic signals of demand response and actively participate in demand response under guidance, further reducing the peak-valley difference, and the total load curve is the most stable.

[0114] In the model of this paper, the case with the smallest peak-valley difference and the most stable load curve is α = 0.7 and β = 0.3. Based on this case, the charging load generated by users of each decision branch is analyzed. As Figure 5 shown, a small number of electric vehicles are in a disordered state, and the disordered charging load is only 8.39 MW, accounting for 3.69% of the total load. Under the action of price and incentive policies, the remaining 47.61 MW of electric vehicle charging load participates in demand response, reducing the maximum peak-valley difference from 79.9% of the baseline load to 39.2%, playing a certain role in peak shaving and valley filling. From 0:00 to 7:00, the baseline load is at the bottom of the valley, generating a large load gap. Under the influence of the response policy, a large number of active responding users choose to charge during the valley period with a lower electricity price, filling a 19.04 MW load gap. However, there is still a large peak-valley difference in the load. Therefore, the system mobilizes more users with delayed decisions to participate in the response. Stimulated by the incentive compensation, it plays a role in the time periods from 0:00 to 7:00 and from 21:00 to 24:00, providing load transfer amounts of 10.47 MW and 4.52 MW respectively, and the peak shaving and valley filling effect is relatively obvious.

[0115] To further verify the universality of the proposed method, three different functional areas, namely residential areas, commercial areas, and office areas, are selected as application scenarios. The simulation results are as Figure 6 shown. The charging range of residential area users is concentrated from 19:00 to 24:00 at night and from 0:00 to 7:30 in the early morning. They respond more actively to the price signal, that is, they choose to charge during the flat and valley periods with lower prices, generating a total load of 31.03 MW. Due to the charging behavior of users in commercial areas being mostly disordered plug-and-charge, an unordered flow of 5.5 MW is generated. In addition, the charging behavior of electric vehicle users in office areas is mainly concentrated during the day and is greatly affected by the incentive response. Under the stimulation of incentive compensation, users generate a response load of 8.21 MW.

[0116] It should be noted that the above-mentioned electric vehicle charging response uncertainty analysis system based on the three-way decision theory is also used to implement the method steps corresponding to the embodiments in the method of the electric vehicle charging response uncertainty analysis method based on the three-way decision theory as described above. This application does not repeat the description here. Figure 1 shown, and the method steps corresponding to the embodiments in the method of the electric vehicle charging response uncertainty analysis method based on the three-way decision theory are not repeated in this application.

[0117] It should be noted that in each embodiment of the present invention, each functional unit / module may be integrated in a processing unit / module, or each unit / module may exist physically alone, or two or more units / modules may be integrated in one unit / module. The above-mentioned integrated unit / module may be implemented in the form of hardware or in the form of a software functional unit / module.

[0118] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments described here can be implemented by hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described here, or a combination thereof. For software implementation, part or all of the processes of the embodiments can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. The computer-readable medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transmission of a computer program from one place to another. The storage media can be any available medium that can be accessed by a computer. The computer-readable medium may include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disc storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. An analysis method for the uncertainty of electric vehicle charging response based on the three-way decision theory, characterized in that, Including the following steps: Construct the cost losses of different electric vehicles taking different decision-making behaviors in different states; Calculate three decision thresholds α and β according to the cost losses; Calculate the decision state value P(X|[x]) of the electric vehicle; According to the decision state value and the three decision thresholds α and β, judge the decision-making behavior of the electric vehicle, and according to the decision-making behavior, judge the type of demand response participated by the electric vehicle: If P(X|[x]) ≥ α, the electric vehicle takes an active response behavior, that is, charging, and it is considered that the electric vehicle participates in price-based demand response; If β < P(X|[x]) < α, the electric vehicle takes a delayed response behavior, it is considered that the electric vehicle participates in incentive-based demand response, and whether the electric vehicle charges is judged by solving a multi-objective function aiming at maximizing the interests of electric vehicles participating in different types of demand response and minimizing the grid load fluctuation; If P(X|[x]) ≤ β, the electric vehicle takes a rejection response behavior and does not participate in demand response, and conducts unordered charging; According to the decision-making behavior of the electric vehicle, calculate the charging load of each electric vehicle one by one, and accumulate the charging loads of all electric vehicles to obtain the total charging load; Among them, the probability of the electric vehicle being in the grid-connected state is statistically used as the decision state value of the electric vehicle; Among them, the construction of the multi-objective function is expressed by the formula: max(F P +F I -F B )+minD(L(t)) Where, F P is the cost saved by electric vehicles participating in price-based demand response; F I is the cost saved by electric vehicles participating in incentive-based demand response; F B is the loss cost of the electric vehicle battery after the electric vehicle participates in demand response; D() represents the load fluctuation variance calculation function; L(t) represents the total load of the distribution network at time t; N represents the number of electric vehicles; F o,i represents the cost generated by the unordered charging of vehicle i; T s and T t represent the time when the electric vehicle connects to the grid and leaves the grid respectively; η c and η d represent the charging and discharging efficiencies of the electric vehicle respectively; P c,i (t) and P d,i (t) represent the charging power and discharging power of vehicle i at time t respectively; p(t) represents the time-of-use electricity price at time t; Q I,i represents the load transfer amount generated by the electric vehicle participating in incentive-based demand response compared with unordered charging; ω1 and ω2 represent the first incentive amount coefficient and the second incentive amount coefficient respectively; μ1, μ2, and μ3 represent the first battery loss coefficient, the second battery loss coefficient, and the third battery loss coefficient respectively; y i (t) represents the charging and discharging state of vehicle i at time t. When y i (t)=1, it means it is in the charging state. When y i (t)=-1, it means it is in the discharging state. When y i (t)=0, it means it is not working.

2. The method for analyzing the uncertainty of electric vehicle charging response based on the three-way decision theory according to claim 1, wherein The calculation of the three decision thresholds α and β is expressed by the formula: where λ PP , λ BP and λ NP are the cost losses of taking decision-making actions a P , a B and a N when the object belongs to state X; λ PN , λ BN and λ NN are the cost losses of taking decision-making actions a P , a B and a N when the object does not belong to state X; λ PP ≤λ BP ≤λ NP and λ NN ≤λ BN ≤λ PN and (λ BP -λ PP )(λ BN -λ NN ) < (λ PN -λ BN )(λ NP -λ BP ).

3. The method for analyzing the uncertainty of electric vehicle charging response based on the three-way decision theory according to claim 1, wherein It also includes: iteratively changing the cost losses, recalculating the three decision thresholds α and β, so as to obtain the grid load calculation results under different decision boundary domains; taking the three decision thresholds α and β corresponding to the optimal one in the grid load calculation results.

4. Electric vehicle charging response uncertainty analysis system based on three-way decision theory, characterized in that Including: A database, which is provided with the cost losses of different electric vehicles taking different decision-making behaviors in different states; A data processing unit, which is used to calculate the three decision thresholds α and β according to the cost losses; calculate the decision state value P(X|[x]) of the electric vehicle; according to the decision state value and the three decision thresholds α and β, judge the decision-making behavior of the electric vehicle, and according to the decision-making behavior, judge the type of demand response participated by the electric vehicle, calculate the charging load of each electric vehicle one by one, and accumulate the charging loads of all electric vehicles to obtain the total charging load; Among them, judging the decision-making behavior of the electric vehicle and the type of demand response participated by the electric vehicle is specifically: If P(X|[x]) ≥ α, the electric vehicle takes an active response behavior, that is, charging, and it is considered that the electric vehicle participates in price-based demand response; If β < P(X|[x]) < α, the electric vehicle takes a delayed response behavior, judges whether the electric vehicle charges by solving a multi-objective function aiming at maximizing the interests of electric vehicles participating in different types of demand response and minimizing the grid load fluctuation, and it is considered that the electric vehicle participates in incentive-based demand response; If P(X|[x]) ≤ β, the electric vehicle takes a rejection response behavior and does not participate in demand response, and conducts unordered charging; Among them, the probability of the electric vehicle being in the grid-connected state is statistically used as the decision state value of the electric vehicle; Among them, the construction of the multi-objective function is expressed by the formula: max(F P +F I -F B )+minD(L(t)) Where, F P is the cost saved by electric vehicles participating in price-based demand response; F I is the cost saved by electric vehicles participating in incentive-based demand response; F B is the loss cost of the electric vehicle battery after the electric vehicle participates in demand response; D() represents the load fluctuation variance calculation function; L(t) represents the total load of the distribution network at time t; N represents the number of electric vehicles; F o,i represents the cost generated by the unordered charging of vehicle i; T s and T t respectively represent the time when the electric vehicle connects to the grid and leaves the grid; η c and η d respectively represent the charging and discharging efficiencies of the electric vehicle; P c,i (t) and P d,i (t) respectively represent the charging power and discharging power of vehicle i at time t; p(t) represents the time-of-use electricity price at time t; Q I,i represents the load transfer amount generated by the electric vehicle participating in incentive-based demand response compared with unordered charging; ω1 and ω2 respectively represent the first incentive amount coefficient and the second incentive amount coefficient; μ1, μ2, and μ3 respectively represent the first battery loss coefficient, the second battery loss coefficient, and the third battery loss coefficient; y i (t) represents the charging and discharging state of vehicle i at time t. When y i (t) = 1, it means it is in the charging state. When y i (t) = -1, it means it is in the discharging state. When y i (t) = 0, it means it is not working.

5. The uncertainty analysis system for electric vehicle charging response based on the three-way decision theory according to claim 4, wherein It further includes: iteratively changing the cost loss, recalculating the three-way decision thresholds α and β, so as to obtain the power grid load calculation results under different decision boundary regions; and taking the three-way decision thresholds α and β corresponding to the optimal one among the power grid load calculation results.

6. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1-3 above.

7. An electronic device, characterized in that, The electronic device includes: a processor; a memory for storing executable instructions of the processor; The processor is configured to read the executable instructions from the memory and execute the instructions to implement the method according to any one of claims 1-3 above.

Images