Electric vehicle charging scheduling method considering demand matching

Abstract

The present invention discloses an electric vehicle charging scheduling method considering demand matching, relates to the technical field of power system control, and alleviates the problems of under-response, over-response and slow solution convergence speed in the process of electric vehicle charging scheduling. The method comprises the following steps: establishing an electric vehicle charging load prediction model of a travel chain; inputting the sorted information into the electric vehicle charging load prediction model to obtain a prediction result; establishing an electric vehicle scheduling model of demand matching; inputting the prediction result in step 1 into the scheduling model in step 2, solving the problem through a fuzzy cuckoo algorithm, and obtaining a scheduling plan; the present invention utilizes an improved travel chain to perform electric vehicle load prediction, and predicts the charging load of electric vehicles at different times and spaces. The method can take into account multiple factors such as the number of electric vehicles and battery capacity, making the prediction result more realistic and reasonable.

Description

Technical Field

[0001] The present invention relates to the technical field of power system control, and in particular to an electric vehicle charging scheduling method considering demand matching. Background Art

[0002] As the scale of electric vehicles (EVs) continues to increase, the safe operation of the power grid faces increasing risks when a large number of EVs are concentrated in charging during peak load hours.

[0003] On the one hand, some studies have explored the Electric Vehicle Aggregator (EVA) mechanism for EV charging scheduling. However, due to the large fluctuations in EV load, EVA predictions of EV charging load are inaccurate, making it difficult to ensure that EV users' charging power matches the grid's required power, leading to frequent under- and over-response issues. On the other hand, when solving EV scheduling models, the algorithms used to solve them fail to converge quickly enough to find the optimal solution, resulting in uneconomical scheduling solutions. Summary of the Invention

[0004] The purpose of the present invention is to solve the above technical problems and provide an electric vehicle charging scheduling method taking demand matching into consideration.

[0005] The technical solution adopted in the present invention is as follows:

[0006] A method for scheduling electric vehicle charging considering demand matching includes the following steps:

[0007] Step 1: Establish a travel chain electric vehicle charging load prediction model; input the organized information into the electric vehicle charging load prediction model to obtain the prediction results;

[0008] Step 2: Establish an electric vehicle dispatch model with demand matching;

[0009] Step 3: Input the prediction results in step 1 into the scheduling model in step 2, and solve it through the fuzzy cuckoo algorithm to obtain the scheduling plan.

[0010] Furthermore, the step 1 specifically includes:

[0011] Step 1.1: Establish a spatiotemporal model of vehicle travel, where each user, trip end time, and trip distance of an electric private car are independent of each other;

[0012] The distribution of trip end time uses the Weibull probability function to fit the end time of each trip, that is:

[0013]

[0014] Where x is a random variable, k is a shape parameter that can represent the peak of the curve, c is a scale parameter related to the end time of the trip, and γ is a location parameter; F(x; k; c; γ) is the probability distribution function, f(x; k; c; γ) is the time probability density function, and e is the natural exponent.

[0015] The distribution of travel distance is fitted by the log-normal distribution function, and the travel distance d of electric private car users is calculated. The probability density function of the distance is:

[0016]

[0017] Where μ D is the expected value of the travel distance; σ D 2 The standard deviation of the expected value of the travel distance; the standard deviation is defined as the square of the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean, which reflects the degree of dispersion between individuals in the group; The physical meaning is a total of N dis Driving distance d p , which means taking the difference between each driving distance and its average, then summing them up and taking the square root, σ D It is the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean;

[0018] Step 1.2: Calculate the spatial transfer probability of vehicles and obtain the spatiotemporal distribution of vehicles in a day:

[0019] Step 1.3: Calculate the power consumption of electric vehicles. The power consumption per unit mileage of electric vehicles is as follows:

[0020]

[0021] Where, E ksl E zgl E cgl E z1 are the power consumption per unit mileage of expressways, main roads, secondary roads and branch roads respectively; V is the speed of electric vehicles.

[0022] The remaining power of the electric vehicle when it arrives at the destination can be obtained by calculating the mileage power consumption based on the initial power during the trip. The formula is as follows:

[0023] SOC i end =SOC i ini -E l i ×d i ,E l ∈(E ksl,E zgl ,E cgl ,E z1 )

[0024] Among them, SOC i ini represents the initial state of charge of the i-th EV, E l i represents the power consumption corresponding to the road traveled by the i-th EV, d i is the road mileage of the i-th electric vehicle, SOC i end represents the battery state of charge of the i-th electric vehicle at the end of its trip.

[0025] Step 1.4 calculates the electric vehicle charging load and obtains the electric vehicle prediction result.

[0026] Furthermore, the step 1.3 more specifically includes:

[0027] To alleviate range anxiety, users must ensure that the vehicle's state of charge is 0.25 at the last transfer. Therefore, the charging conditions at destination n are as follows:

[0028] S n C-ω n+1 l n+1 <0.25C (6)

[0029] S n C=S n-1 C-ω n l n (7)

[0030] Where S n is the battery state of charge when the vehicle arrives at the destination n; C is the battery capacity of the electric vehicle, ω n is the power consumption per kilometer of the vehicle traveling to n; l n Indicates the distance traveled to trip n; n+1 is the next trip;

[0031] Formula (6) and Formula (7) show that when the vehicle arrives at the state destination n, the remaining battery power is calculated. If the battery state of charge is greater than 0.25 when the vehicle arrives at the state destination n+1, then charging is not required at n, otherwise charging is required.

[0032] If the vehicle needs to be charged at destination n, the initial state of charge and charging power level need to be considered, and the charging time can be estimated as

[0033]

[0034] Where, T c nis the charging time of the vehicle at location n; S n-1 is the battery state of charge of the vehicle at time n-1; ω n and l n Same meaning as above; η is the charging efficiency; P c n Represents the charging power level at n.

[0035] Furthermore, the step 2 specifically includes:

[0036] Step 2.1: Establish an electric vehicle charging model; the model is expressed as

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043] are the charging power of the i-th electric vehicle in the t-scheduling period; is the rated discharge power of the i-th electric vehicle; is the state of charge of the i-th vehicle at time t; is the capacity of the battery of vehicle i; η c is the charging efficiency of electric vehicles; Δt is the time interval between two optimization dispatches; t i,arr is the time period when the i-th electric vehicle begins to accept dispatch; i,dep represents the time period when the i-th vehicle leaves; is the expected state of charge of the i-th electric vehicle when the user leaves; and Respectively represent the upper and lower limits of the EV battery state of charge; P EV c (t) represents the sum of the charging power of all electric vehicles being charged at time t, and N represents the number of all electric vehicles being charged;

[0044] Step 2.2: Propose a demand matching factor. When EV aggregators meet the response requirements of the power grid, over-response and under-response may occur. Therefore, the concept of demand matching factor is proposed to describe this situation.

[0045] The under-response factor during peak clipping is r1(P EV c(t)-P des (t)), the under-response factor when filling the valley is r2(P EV c (t)-P des (t))

[0046] The over-response factor during peak clipping is e1(P EV c (t)-P des (t)), the over-response factor when filling the valley is e2(P EV c (t)-P des (t))

[0047] Among them, P EV c (t) represents the sum of the charging power of all electric vehicles being charged at time t, P des (t) represents the expected power of the dispatching department at the dispatching time t.

[0048] r1 represents a value when the peak shaving response is insufficient. It is added to the objective function as a penalty term. The value of r1 is set according to the difference between the electric vehicle charging power and the expected power. The specific value setting is shown in Table 2.

[0049] r2 represents a value when the valley filling is insufficient, and it is added to the objective function as a penalty term. The value of r2 is set according to the difference between the electric vehicle charging power and the expected power. The specific value setting is shown in Table 2.

[0050] e1 is a value when the peak is clipped and the response is excessive. It is added to the objective function as a penalty term. The value of e1 is set according to the difference between the electric vehicle charging power and the expected power. The specific value setting is shown in Table 2.

[0051] e2 represents a value when the valley filling response is too high. It is added to the objective function as a penalty term. The value of e2 is set according to the difference between the electric vehicle charging power and the expected power. The specific value setting is shown in Table 2.

[0052] Step 2.3: Establish an electric vehicle charging scheduling model based on demand matching. The established objective function has two layers. The upper layer objective is F, which means to minimize the difference between the electric vehicle response and the expected power of the scheduling department. The charging power of each controlled electric vehicle is used as the decision variable. The model is as follows: (The goal is to minimize the difference between the electric vehicle response and the expected power of the scheduling department, and the penalty term when the electric vehicle does not fully complete the response is added to the objective. The solution is the charging power of the electric vehicle at each moment)

[0053] (Formula 15 is the objective function, P EVc It is a variable, and the result is a matrix, which is a 1*24 matrix, representing the optimal charging capacity for each of the 24 time periods)

[0054]

[0055] The meaning of the model formula has been explained in step 2.2;

[0056] The lower-level goal is to maximize EVA benefits, and its model is as follows:

[0057] maxC=C EV +C dem -C grid -C sub (16)

[0058]

[0059] Among them C EV represents the charging revenue that the aggregator gets from the user, c EV represents the charging price of electric vehicles, T represents the charging period, ΔT represents one period, and t start This is the time when charging starts.

[0060]

[0061] Among them C dem represents the subsidy for participating in grid demand response, c dem represents the subsidy standard, and α represents the response coefficient, which is related to the scheduling notification time.

[0062]

[0063] Among them C grid Represents the cost of EVA purchasing electricity from the grid, c grid Represents the electricity purchase price.

[0064]

[0065] Among them C sub Represents EVA subsidies to EV users participating in demand response, c sub c Represents the charging subsidy electricity price.

[0066] Furthermore, the fuzzy cuckoo algorithm in step 3 is:

[0067] Step 3.1, set the search stop accuracy t ol 、The upper bound of the solution u b , the lower bound of the solution l b , the dimension of the search space n dand the discovery probability P a , and randomly generate z cuckoo nest positions accordingly;

[0068] Step 3.2: Assume that in the previous generation search, the obtained optimal solution is where best represents the number of optima, dai represents the number of iterations. Update the Levy flight positions of other solutions according to the optimal position, and calculate the fitness function values of the obtained set of new solutions;

[0069] Step 3.3: Call the fuzzy controller to dynamically update the discovery probability P a ;

[0070] Step 3.4: Compare the random number r between 0 and and the discovery probability P set in Step 3.1 a . If P a < r, compare the fitness function value obtained in Step 3.2 with the fitness function value of the previous generation solution. If the result of the new solution is better than the previous generation solution, update the solution. If the result of the new solution is worse than the previous generation solution or P a ≥ r, retain the previous generation solution;

[0071] Step 3.5: Determine whether the optimal solution obtained in Step 3.3 reaches the set accuracy t ol . If it is satisfied, output the result of the optimal solution; if not, return to Step 3.2 to re - perform iterative update. represents the size of the optimal charging power of the electric vehicle obtained by solving, represents the set of solutions obtained for the optimal charging power; the fitness function represents minimizing the difference between the response quantity of the electric vehicle and the expected power of the dispatching department;

[0072] The beneficial effects of the present invention are as follows:

[0073] 1. The present invention uses an improved travel chain to predict the load of electric vehicles, predicting the charging load of electric vehicles at different times and spaces. This method can take into account various factors such as the number of electric vehicles and battery capacity, making the prediction results more realistic and reasonable;

[0074] 2. The present invention considers the electric vehicle load aggregator mechanism, proposes the concept of demand matching degree factor, and adds the demand matching degree factor to the electric vehicle charging scheduling model, which can reduce the negative impacts brought by under - / over - response;

[0075] 3. The present invention proposes an improved cuckoo algorithm to solve the electric vehicle charging scheduling model, which has the characteristics of fast convergence speed and high population diversity, alleviating the problems of under - response, over - response and slow solution convergence speed existing in the electric vehicle charging scheduling process. Brief Description of the Drawings

[0076] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the proportional relationships of the various components in the drawings of this specification do not represent the proportional relationships in actual material selection and design, and are merely schematic diagrams of structures or positions, among which:

[0077] Figure 1 It is the overall flow chart of the present invention;

[0078] Figure 2 is a flow chart of electric vehicle charging load prediction according to the present invention;

[0079] Figure 3 Schematic diagram of expected charging load of electric vehicles under different scenario modes of the present invention;

[0080] Figure 4 It is a convergence comparison diagram of the algorithm of the present invention;

[0081] Figure 5 This is a population diversity comparison chart of the algorithm of the present invention. DETAILED DESCRIPTION

[0082] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. That is, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments.

[0083] The following combination Figures 1 to 5 , the present invention is described in detail.

[0084] Example 1

[0085] A method for scheduling electric vehicle charging considering demand matching includes the following steps:

[0086] Step 1: Establish a travel chain electric vehicle charging load prediction model; input the organized information into the electric vehicle charging load prediction model to obtain the prediction results;

[0087] Step 2: Establish an electric vehicle dispatch model with demand matching;

[0088] Step 3: Input the prediction results in step 1 into the scheduling model in step 2, and solve it through the fuzzy cuckoo algorithm to obtain the scheduling plan.

[0089] The working principle / process of the present invention is as follows: first, a travel chain electric vehicle charging load prediction model and a demand matching electric vehicle scheduling model are established, and the system information required by the electric vehicle prediction model is sorted out; secondly, the sorted information is input into the electric vehicle charging load prediction model; then, the prediction result output from the prediction model is input into the electric vehicle charging scheduling model; finally, a fuzzy cuckoo algorithm is proposed to solve the electric vehicle charging scheduling model and output a scheduling plan.

[0090] Example 2

[0091] The process of establishing the electric vehicle charging load prediction model for the travel chain is as follows: Figure 2 As shown, please follow the steps below:

[0092] Step 1.1: Establish a spatiotemporal model of vehicle travel

[0093] The prediction model examines resident travel data and categorizes these trips into four main categories based on activity type: Home (H), Work (W), Shopping & Eating (SE), and Other Family / Personal Errands (O). Each electric private vehicle user randomly moves between these four categories, with independent trip end times and travel distances.

[0094] The distribution of trip end time uses the Weibull probability function to fit the end time of each trip, that is:

[0095]

[0096] Where x is a random variable, k is a shape parameter that can represent the peak of the curve, c is a scale parameter related to the end time of the trip, and γ is a location parameter; F(x; k; c; γ) is the probability distribution function, f(x; k; c; γ) is the time probability density function, and e is the natural exponent.

[0097] The distribution of travel distance is fitted by the log-normal distribution function, and the travel distance d of electric private car users is calculated. The probability density function of the distance is:

[0098]

[0099] Where μ D is the expected value of the travel distance; σ D 2 The standard deviation of the expected value of the travel distance; the standard deviation is defined as the square of the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean, which reflects the degree of dispersion between individuals in the group; The physical meaning is a total of Ndis Driving distance d p , which means taking the difference between each driving distance and its average, then summing them up and taking the square root, σ D It is the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean;

[0100] Step 1.2, the vehicle spatial transfer probability, calculates the state transition probability, and determines the spatiotemporal distribution of electric vehicles in a day;

[0101] A Markov chain is a sequence of random events, where the future value is only related to the current value. It is a discrete random process that can reasonably describe the travel patterns of electric private car users. There can only be one state at each moment, and the current state is denoted as E. u , the state at the next moment is E y , then the Markov chain can be expressed by conditional probability:

[0102] P(E u →E y )=P(E u |E y )=P uy (3)

[0103] This solution considers the location corresponding to the destination as a state, and the next state of the vehicle driven by the user is determined by the location at the current moment. uy is the probability of state transition, where subscripts u and y represent the state E u To state E y The probability of the next state transition is written in matrix form as follows:

[0104]

[0105] Assuming that the vehicle is currently in E1, then at the next moment, it may move from state E1 to any state among E1, E2,…, E4. So p uy Meet the conditions

[0106]

[0107] There are four states set in this patent. In order to express the one-step transition probability of the electric vehicle to the next state, n can be changed to 4, or used in formula (4) and formula (5). The transition probability matrix containing the four states is as follows:

[0108]

[0109] Where p uy It can be determined based on statistics of local residents’ travel survey data.

[0110] Step 1.3: Calculate the power consumption of electric vehicles. The power consumption per unit mileage of electric vehicles is as follows:

[0111]

[0112] Where, E ksl E zgl E cgl E z1 are the power consumption per unit mileage of expressways, main roads, secondary roads and branch roads respectively; V is the speed of electric vehicles.

[0113] The remaining power of the electric vehicle when it arrives at the destination can be obtained by calculating the mileage power consumption based on the initial power during the trip. The formula is as follows:

[0114] SOC i end =SOC i ini -E l i ×d i ,E l ∈(E ksl ,E zgl ,E cgl ,E z1 )

[0115] Among them, SOC i ini represents the initial state of charge of the i-th EV, E l i represents the power consumption corresponding to the road traveled by the i-th EV, d i is the road mileage of the i-th electric vehicle, SOC i end represents the battery state of charge of the i-th electric vehicle at the end of its trip.

[0116] Optionally, the step 1.3 may further include:

[0117] To alleviate range anxiety, users must ensure that the vehicle's state of charge is 0.25 at the last transfer. Therefore, the charging conditions at destination n are as follows:

[0118] S n C-ω n+1 l n+1 <0.25C (6)

[0119] S n C=S n-1 C-ω n l n (7)

[0120] Where Sn is the battery state of charge when the vehicle arrives at the destination n; C is the battery capacity of the electric vehicle, ω n is the power consumption per kilometer of the vehicle traveling to n; l n Indicates the distance traveled to trip n; n+1 is the next trip;

[0121] Formula (6) and Formula (7) show that when the vehicle arrives at the state destination n, the remaining battery power is calculated. If the battery state of charge is greater than 0.25 when the vehicle arrives at the state destination n+1, then charging is not required at n, otherwise charging is required.

[0122] If the vehicle needs to be charged at destination n, the initial state of charge and charging power level need to be considered, and the charging time can be estimated as

[0123]

[0124] Where, T c n is the charging time of the vehicle at location n; S n-1 is the battery state of charge of the vehicle at time n-1; ω n and l n Same meaning as above; η is the charging efficiency; P c n Represents the charging power level at n.

[0125] Step 1.4 calculates the EV charging load and generates the EV prediction results. EVA sets the following simulation parameters: the total number of EVs and the number of trip chains; vehicle battery capacity, charging power level, and charging efficiency; the probability distribution of the starting charging time and driving distance, and the number of simulations (all of these parameters are set by EVA itself and are considered known parameters). It simulates electric travel using the Monte Carlo simulation method (this method is already available). It then calculates the power consumption of each EV and determines the charging power based on the expected power consumption of the EV user upon leaving the charging station. Finally, it accumulates the charging power of each EV to generate the EV charging load prediction results.

[0126] Traffic conditions can be determined from the "Urban Road Engineering Design Code," which divides urban roads into four levels: expressways, trunk roads, secondary trunk roads, and branch roads. Each level of road has different travel speeds under different congestion conditions.

[0127] Its prediction effect is as follows Figure 3As shown in the figure, the daily EV charging power reaches its peak between 6:00 PM and 7:00 PM, which reasonably describes the scenario where a large number of EV users return home to charge their vehicles. However, the daily EV charging power drops to a low point between 3:00 AM and 6:00 AM because charging starts are concentrated after 6:00 PM. By 3:00 AM, most EVs have already reached the expected state of charge, meeting the charging needs of real-world EV users.

[0128] Assuming that the scale of electric vehicles is also 2,500, the charging power level is 3.3kW, and the charging mode is one day, Method 1 is the electric vehicle load prediction method based on the travel chain of this patent, and Method 2 is the BP neural network prediction method.

[0129] The calculation results of the two methods are shown in the following table: the difference between the prediction method proposed in this patent and the actual value is 13.1%, while the difference between the prediction value of method 2 and the actual value is 23.3%. Therefore, the prediction effect of the method proposed in this patent is better.

[0130] Table 1 Comparison of prediction results of methods 1 and 2 with actual values

[0131]

[0132] Example 3

[0133] Establish an electric vehicle dispatch model with demand matching, and implement it according to the following steps:

[0134] Step 2.1: Build an electric vehicle (EV) charging model

[0135] In order to give full play to the energy storage characteristics of electric vehicle batteries during optimization scheduling, this patent establishes an electric vehicle charging model. The model is expressed as

[0136]

[0137]

[0138]

[0139]

[0140]

[0141]

[0142] are the charging power of the i-th electric vehicle in the t-scheduling period; is the rated discharge power of the i-th electric vehicle; is the state of charge (SOC) of the i-th vehicle at time t; is the capacity of the battery of vehicle i; η c is the charging efficiency of electric vehicles; Δt is the time interval between two optimization dispatches; t i,arr is the time period when the i-th electric vehicle begins to accept dispatch; i, d ep represents the time period when the i-th vehicle leaves; is the expected state of charge of the i-th electric vehicle when the user leaves; and Respectively represent the upper and lower limits of the EV battery state of charge; P EV c (t) represents the sum of the charging powers of all electric vehicles being charged at time t, and N represents the number of electric vehicles being charged.

[0143] Step 2.2: Propose a demand matching factor. When EV aggregators meet the response requirements of the power grid, over-response and under-response may occur. Therefore, the concept of demand matching factor is proposed to describe this situation.

[0144] The under-response factor during peak clipping is r1(P EV c (t)-P des (t)), the under-response factor when filling the valley is r2(P EV c (t)-P des (t))

[0145] The over-response factor during peak clipping is e1(P EV c (t)-P des (t)), the over-response factor when filling the valley is e2(P EV c (t)-P des (t))

[0146] Among them, Table 2 shows the corresponding values ​​of demand matching factors under different response ratios, P EV c (t) represents the sum of the charging power of all electric vehicles being charged at time t, P des (t) represents the expected power of the dispatching department at the dispatching time t.

[0147] Table 2 Corresponding values ​​of demand matching factors under different response ratios

[0148]

[0149]

[0150] Step 2.3: Establish an electric vehicle charging scheduling model based on demand matching. The established objective function has two layers. The upper layer objective is F, which means to minimize the difference between the electric vehicle response and the expected power of the scheduling department. The charging power of each controlled electric vehicle is used as the decision variable. The model is as follows: (The goal is to minimize the difference between the electric vehicle response and the expected power of the scheduling department, and the penalty term when the electric vehicle does not fully complete the response is added to the objective. The solution is the charging power of the electric vehicle at each moment)

[0151] (Formula 15 is the objective function, P EV c It is a variable, and the result is a matrix, which is a 1*24 matrix, representing the optimal charging capacity for each of the 24 time periods)

[0152]

[0153] The meaning of the model formula has been explained in step 2.2;

[0154] The lower-level goal is to maximize EVA benefits, and its model is as follows:

[0155] maxC=C EV +C dem -C grid -C sub (16)

[0156]

[0157] Among them C EV represents the charging revenue that the aggregator gets from the user, c EV represents the charging price of electric vehicles, T represents the charging period, ΔT represents one period, and t start This is the time when charging starts.

[0158]

[0159] Among them C dem represents the subsidy for participating in grid demand response, c dem represents the subsidy standard, and α represents the response coefficient, which is related to the scheduling notification time.

[0160] Table 3 Subsidy standards and response coefficients under different response types

[0161]

[0162]

[0163]

[0164] Among them C grid Represents the cost of EVA purchasing electricity from the grid, c grid Represents the electricity purchase price.

[0165]

[0166] Among them C sub Represents EVA subsidies to EV users participating in demand response, c sub c Represents the charging subsidy electricity price.

[0167] Example 4

[0168] The prediction results from step 1 are input into the scheduling model from step 2. The fuzzy cuckoo search algorithm is used to solve the scheduling solution. The cuckoo search algorithm is a biomimetic-inspired algorithm for solving optimization problems. It is inspired by the unique way cuckoos use to find host nests with eggs similar to their own, thereby using the host birds to reproduce their offspring. This algorithm uses the related Levy flight search mechanism. Research has shown that this algorithm has advantages such as easy search, few parameters, and fast convergence, making it more effective than other swarm optimization algorithms.

[0169] Existing technology shows that the implementation of the cuckoo search algorithm needs to meet three basic conditions:

[0170] 1. Regardless of external factors such as distance and environment, cuckoos randomly select host bird nests to lay eggs, and it is assumed that each cuckoo can only lay one egg at a time;

[0171] 2. If the egg in a certain nest is the best among all the eggs currently available, the best egg will be automatically kept for the next generation;

[0172] 3. Assume that when a foreign cuckoo egg is discovered, the host bird will abandon the egg or the entire nest. To keep the total number of nests n unchanged, the host bird will build a new nest in another place. Assume that the probability of any parasitic cuckoo egg being discovered by the host bird is P a ∈[0,1]

[0173] The basic steps of the cuckoo search algorithm are:

[0174] 1) Set the search stop accuracy t ol 、The upper bound of the solution u b , the lower bound of the solution l b , the dimension of the search space n d And the probability of discovery P a , and then randomly generate n bird nest locations;

[0175] 2) Assume that in the previous generation of search, the optimal position obtained is \(x\). tb , then update the positions of other solutions according to the optimal position by Levy flight, and calculate the fitness function values of the obtained set of new solutions;

[0176] 3) Use a random number \(r\) between 0 and 1 to compare with the discovery probability \(P\) set in step 1). a If \(P\). a < \(r\), then compare the fitness function value obtained in step 2) with the fitness function value of the previous generation of solutions. If the result of the new solution is better than that of the previous generation of solutions, update the solution. If the result of the new solution is worse than that of the previous generation of solutions or \(P\). a ≥ \(r\), then retain the previous generation of solutions;

[0177] 4) Determine whether the optimal solution obtained in step 3) reaches the set accuracy \(t\). ol , if satisfied, output the result of the optimal solution; if not satisfied, return to step 2) to re - perform iterative update.

[0178] Based on the above - mentioned existing technology, the cuckoo algorithm is optimized using fuzzy logic and applied to this solution to solve the charging scheduling scheme in the scheduling model; the specific steps are as follows:

[0179] Step 3.1: Set the search stop accuracy \(t\). ol , the upper bound \(u\) of the solution. b , the lower bound \(l\) of the solution. b , the dimension \(n\) of the search space. d and the discovery probability \(P\). a , then randomly generate \(z\) cuckoo nest positions. The calculation method of the fitness value function is shown in formula (15). Let the initial value of the solved variable. represents the set initial value, where \(dai = 1\) represents the first - generation solution, and continuously perform iterative update to find the optimal solution. The specific iterative steps are shown in 3.2 - 3.5;

[0180] Step 3.2: Assume that in the previous generation of search, the optimal solution obtained is where best represents the number of optima and dai represents the number of iterations. Update the positions of other solutions according to the optimal position by Levy flight, and calculate the fitness function values of the obtained set of new solutions;

[0181] Step 3.3: Call the fuzzy controller to dynamically update the discovery probability \(P\). a ;

[0182] Step 3.4: Use a random number \(r\) between 0 and 1 to compare with the discovery probability \(P\) set in step 3.1. a If \(P\). a<r, then compare the fitness function value obtained in step 3.2 with the fitness function value of the solution in the previous generation. If the result of the new solution is better than that of the previous generation, update the solution. If the result of the new solution is worse than that of the previous generation or P a ≥r, then retain the solution of the previous generation;

[0183] Step 3.5, determine whether the optimal solution obtained in step 3.3 reaches the set accuracy t ol , if satisfied, output the result of the optimal solution; if not satisfied, return to step 3.2 to re-iterate and update. represents the size of the optimal charging power of the electric vehicle solved, <atag represents the set of solutions obtained by the optimal charging power; the fitness function represents minimizing the difference between the electric vehicle response amount and the expected power of the dispatching department.

[0184] Optionally: calculation of the input variables of the fuzzy controller, for realizing the discovery probability P of the cuckoo search algorithm a dynamic update, introducing 2 input variables of the fuzzy controller: population diversity evaluation index D x and population convergence speed evaluation index D f .

[0185] Population diversity evaluation index: use the variance of the current solution to evaluate the population

[0186] diversity, and its calculation formula is:

[0187]

[0188] In the formula, n d represents the number of solutions,; represents the average value of the individuals in the dai-th generation represents the g-th in the dai-th generation i individual;

[0189] Population convergence speed evaluation index: use the average distance of the fitness function to evaluate the convergence speed, and its calculation formula is:

[0190]

[0191] In the formula, represents the average value of the fitness function values of the individuals in the dai-th generation,

[0192] Optionally,: fuzzyification of the fuzzy controller variables, for the discovery probability P a , diversity evaluation index D x and convergence speed evaluation index D f , respectively define the fuzzy state of the corresponding fuzzy controller input and output variables as: FP a , FDx 、FD f ∈{PB, PM, PS, Z, NS, NM, NB}, where PB (positive big) represents positive big, PM (positive medium) represents positive medium, PS (positive small) represents positive small, Z (zero) represents zero, NS (negative small) represents negative small, NM (negative medium) represents negative medium, and NB (negative big) represents negative big.

[0193] In order to comprehensively consider the population diversity and population convergence speed, according to the diversity evaluation index D x And the convergence speed evaluation index D f The level of discovery probability P is appropriately adjusted. a ,The fuzzy rules of the fuzzy cuckoo algorithm designed in this paper follow the following principles:

[0194] 1. When D x and D f If it is too high, P should be appropriately lowered. a , which can speed up the population convergence and prevent premature maturity;

[0195] 2. When D x and D f If it is too low, P should be appropriately increased. a , generating more new individuals, thereby increasing the diversity of the population. The designed fuzzy controller rules are shown in Table 4.

[0196] Table 4 Fuzzy controller rules

[0197]

[0198]

[0199] Furthermore, in order to verify the superiority of the fuzzy cuckoo algorithm, the following two indicators are used to evaluate the algorithm:

[0200] The probability P is found by the cuckoo search algorithm a Dynamic update, introducing two input variables of the fuzzy controller: population diversity evaluation index D x And population convergence speed evaluation index D f ;

[0201] Population diversity evaluation index D x : Use the variance of the current solution to evaluate population diversity, and the calculation formula is:

[0202] Where n d represents the number of solutions; Represents the average value of individuals in the dai generation Indicates the daith generation g i individual;

[0203] Population convergence speed evaluation index D f : The average distance of the fitness function is used to evaluate the convergence speed, and its calculation formula is:

[0204] Where, Represents the average value of the fitness function of individuals in the dai generation,

[0205] Optionally, a fuzzy cuckoo algorithm performance test is performed to verify the performance advantage of the fuzzy cuckoo algorithm designed in the present invention. The Schaffer function is used as a test function to evaluate the population diversity and convergence of the genetic algorithm (GA), the cuckoo algorithm (CS), and the fuzzy cuckoo algorithm (FCS), and the results of each algorithm are compared. The expression of the Schaffer function is:

[0206]

[0207] from Figure 4 、 Figure 5 It can be seen that the fuzzy cuckoo algorithm performs better in terms of convergence speed. At the same time, the fuzzy cuckoo algorithm can always maintain a better level in population diversity. The advantage of the improved algorithm performance is precisely due to the addition of two evaluation indicators in its optimization process.

[0208] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for scheduling electric vehicle charging considering demand matching, characterized in that: The following steps are included: Step 1: Establish a travel chain electric vehicle charging load prediction model; input the organized information into the electric vehicle charging load prediction model to obtain the prediction results; Step 2: Establish an electric vehicle dispatch model with demand matching; Step 3: Input the prediction results in step 1 into the scheduling model in step 2 and solve it using the fuzzy cuckoo algorithm to obtain the scheduling solution; The step 2 specifically includes: Step 2.1: Establish an electric vehicle charging model; the model is expressed as: (9) (10) (11) (12) (13) (14) are the charging power of the i-th electric vehicle in the t-scheduling period; is the rated discharge power of the i-th electric vehicle; is the state of charge of the i-th vehicle at time t; is the capacity of the battery of vehicle i; Charging efficiency for electric vehicles; The time interval between two optimization schedules; The time period when the i-th electric vehicle starts to accept dispatch; represents the time period when the i-th vehicle leaves; is the expected state of charge of the i-th electric vehicle when the user leaves; and Respectively represent the upper and lower limits of the EV battery state of charge; represents the sum of the charging power of all electric vehicles being charged at time t, and N represents the number of electric vehicles being charged; Step 2.2: Propose a demand matching factor. When EV aggregators meet the grid's response requirements, they may over-respond or under-respond. Therefore, the concept of a demand matching factor is proposed to describe this situation. The under-response factor during peak clipping is , the under-response factor when filling the valley is The over-response factor during peak clipping is , the over-response factor when filling the valley is in, It represents the sum of the charging power of all electric vehicles being charged at time t. represents the expected power of the dispatching department at the dispatching time t; It represents a value when the peak shaving is under-response. It is added to the objective function as a penalty term and set according to the difference between the electric vehicle charging power and the expected power. size; It represents a value when the valley filling is insufficient, and is added to the objective function as a penalty term. It is set according to the difference between the electric vehicle charging power and the expected power. size; A value when the peak shaving response is too high, which is added to the objective function as a penalty term and set according to the difference between the electric vehicle charging power and the expected power size; It represents a value when the valley filling response is too high. It is added to the objective function as a penalty term and set according to the difference between the electric vehicle charging power and the expected power. size; Step 2.3: Establish an electric vehicle charging scheduling model based on demand matching; the established objective function has two layers, the upper layer objective is Its meaning is to minimize the difference between the response of electric vehicles and the expected power of the dispatching department, with the charging power of each controlled electric vehicle as the decision variable. The model is as follows: (15) The meaning of the model formula has been explained in step 2.2; The lower-level goal is to maximize EVA benefits, and its model is as follows: (16) (17) in Represents the charging revenue that the aggregator gets from users, represents the electricity price for electric vehicle charging, Indicates the time required for charging. Indicates 1 period, The charging start time; (18) in represents subsidies for participating in grid demand response, represents the subsidy standard, represents the response coefficient, which is related to the scheduling notification time; (19) in Represents the cost of EVA purchasing electricity from the grid, represents the electricity purchase price; (20) in On behalf of EVA, subsidies are given to EV users who participate in demand response. Represents the charging subsidy electricity price.

2. The electric vehicle charging scheduling method considering demand matching according to claim 1 is characterized in that: The step 1 specifically includes: Step 1.1: Establish a spatiotemporal model of vehicle travel, where each user, trip end time, and trip distance of an electric private car are independent of each other; The distribution of trip end time uses the Weibull probability function to fit the end time of each trip, that is: （1） Where x is a random variable, k is a shape parameter that can represent the peak of the curve, and c is a scale parameter related to the end time of the trip. is a positional parameter; is the probability distribution function, is the time probability density function, e is the natural exponential; The distribution of travel distance is fitted by the log-normal distribution function, and the travel distance d of electric private car users is calculated. The probability density function of the distance is: （2） Where, is the expected value of the travel distance; The standard deviation of the expected value of the travel distance; the standard deviation is defined as the square of the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean, which reflects the degree of dispersion between individuals in the group; The physical meaning is a total Driving distance , which means taking the difference between each driving distance and its average, then summing them up and taking the square root. It is the arithmetic mean of the squares of the deviations between the standard values ​​of each unit in the population and its mean; Step 1.2: Calculate the spatial transfer probability of the vehicle and obtain the spatiotemporal distribution of the vehicle in a day: Step 1.3: Calculate the power consumption of electric vehicles. The power consumption per unit mileage of electric vehicles is as follows: Where, are the power consumption per unit mileage of expressways, main roads, secondary roads, and branch roads respectively; V is the speed of electric vehicles; The remaining power of the electric vehicle when it arrives at the destination can be obtained by calculating the mileage power consumption based on the initial power during the trip. The formula is as follows: in represents the initial state of charge of the i-th EV, represents the power consumption corresponding to the road traveled by the i-th EV, is the road mileage of the i-th electric vehicle, represents the battery state of charge of the i-th electric vehicle at the end of its trip; Step 1.4 calculates the electric vehicle charging load and obtains the electric vehicle prediction result.

3. The electric vehicle charging scheduling method considering demand matching according to claim 2 is characterized in that: The step 1.3 more specifically includes: To alleviate range anxiety, users must ensure that the vehicle's state of charge is 0.25 at the last transfer. Therefore, the charging conditions at destination n are as follows: (6) (7) Where, is the battery state of charge when the vehicle arrives at the destination n; C is the battery capacity of the electric vehicle, is the power consumption per kilometer of the vehicle traveling to n; Indicates the distance traveled to trip n; n+1 is the next trip; Formula (6) and Formula (7) show that when the vehicle arrives at the state destination n, the remaining battery power is calculated. If the battery state of charge is greater than 0.25 when the vehicle arrives at the state destination n+1, then charging is not required at n, otherwise charging is required. If a vehicle needs to charge at destination n, the initial state of charge and charging power level need to be considered, and the charging time can be estimated as: （8） Where, is the charging time of the vehicle at location n; is the battery state of charge of the vehicle at time n-1; and Same meaning as above; is charging efficiency; Represents the charging power level at n.

4. The electric vehicle charging scheduling method considering demand matching according to claim 1 is characterized in that: The fuzzy cuckoo algorithm in step 3 is: Step 3.1: Set the search stop accuracy , the upper bound of the solution , the lower bound of the solution , the dimension of the search space and the probability of discovery , and then randomly generate z bird nest locations; Step 3.2: Assume that in the previous generation search, the optimal solution is ,in represents the optimal number, Represents the number of iterations, updates the Levy flight positions of other solutions according to the optimal position, and calculates the fitness function value of the obtained set of new solutions; Step 3.3, call the fuzzy controller to find the probability Perform dynamic updates; Step 3.4: Use a random number r between 0 and 1 and the discovery probability set in step 3.1 In contrast, if , then compare the fitness function value obtained in step 3.2 with the fitness function value of the previous generation solution. If the result of the new solution is better than the previous generation solution, the solution is updated. If the result of the new solution is worse than the previous generation solution or , then retain the previous generation solution; Step 3.5: Determine whether the optimal solution obtained in step 3.3 reaches the set accuracy If it is satisfied, the optimal solution is output; if not, return to step 3.2 and iterate again; represents the optimal charging power of the electric vehicle. It represents the set of solutions for the optimal charging power; the fitness function represents the difference between the response of electric vehicles and the power expected by the dispatching department.

Images