An Ordered Charging Control Method for Electric Vehicles Based on Dynamic Time-of-Use Electricity Price

Abstract

An orderly charging control method for electric vehicles based on dynamic time-of-use electricity price, which relates to an orderly charging control method for electric vehicles. The present invention aims to solve the problems of weak pertinence of the existing orderly charging model for electric vehicles and difficulty in solving intelligent optimization algorithms. The present invention proposes an orderly charging model for electric vehicles, sets up an objective function that meets actual needs and constraint conditions that conform to the actual situation, uses the Monte Carlo method to predict the charging load of electric vehicles, and proposes a dynamic time-of-use electricity price strategy based on the day-ahead load prediction results. Combining with the improved NSGA-II algorithm to solve the model, the formulation of an orderly charging strategy for electric vehicles that simultaneously considers the optimization of charging costs and load variance is realized. The present invention belongs to the technical field of new energy vehicle charging.

Description

Technical Field

[0001] The present invention relates to a method for controlling the orderly charging of electric vehicles, belonging to the technical field of new energy vehicle charging. Background Art

[0002] With the continuous development of social technology, human exploration in various fields has been continuously deepened, and the consumption and demand for energy have also been continuously climbing. The rapid growth of carbon dioxide emissions has caused alarming changes in the global environment, and there has been a fierce collision between industrial civilization and ecological civilization during the process of industrial development.

[0003] As a flexible load, electric vehicles can give full play to their flexibility characteristics, guiding electric vehicle users to charge during off-peak hours such as at night, so as to achieve "peak shaving and valley filling" of the grid load curve, improve the utilization rate of power equipment, and enhance the economic benefits of the operation of the power system. Therefore, the popularization and coverage of electric vehicles to the general public users is an important way to realize the transformation of a low-carbon economy society and an important direction for the future development of automobiles, which has also received great attention from governments around the world and automobile enterprises worldwide.

[0004] Problems existing in the prior art: In the problem of orderly charging of electric vehicles, different electric vehicles have different charging characteristics, and the prediction accuracy of the access time of a large number of electric vehicles to the grid is relatively low. The formulation strategy of traditional time-of-use electricity prices is relatively general, and it is difficult to consider the impact on the grid load after a large number of electric vehicles are connected to the grid. The unreasonable formulation of time-of-use electricity prices will also cause phenomena such as reverse peak shaving during the process of regulating the orderly charging of electric vehicles. The establishment of the charging scheduling model for electric vehicles is not accurate enough, and the constraint conditions are not considered comprehensively. During the multi-objective solution process of the model, traditional solution algorithms face problems such as slow solution convergence speed and small search space. Summary of the Invention

[0005] In order to solve the problems of weak pertinence of the existing orderly charging model for electric vehicles and difficult solution of intelligent optimization algorithms, the present invention further provides a method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices.

[0006] The technical solution adopted by the present invention to solve the above problems is: The steps of the present invention include:

[0007] Step 1, sort out the probability density function of the travel characteristics of electric vehicles;

[0008] Step 2, adopt the Monte Carlo simulation method to simulate the situation of a large number of electric vehicles accessing the grid, and obtain the disorderly charging load of electric vehicles;

[0009] Step 3, based on the disorderly charging load of electric vehicles obtained in Step 2, formulate a dynamic time-of-use electricity price based on the day-ahead load;

[0010] Step 4: Input the dynamic time-of-use electricity price obtained in Step 3 into the electric vehicle charging model;

[0011] Step 5: For each connected electric vehicle, simulate its travel characteristics using the Monte Carlo method;

[0012] Step 6: Determine whether the scheduling conditions are met and decide whether the electric vehicle participates in the scheduling;

[0013] Step 7: Set the objective function and constraint conditions;

[0014] Step 8: Use the improved NSGA-II algorithm for optimization and solution;

[0015] Step 9: After reaching the set number of iterations, obtain a set of Pareto optimal solutions, select the solution with the best comprehensive effect according to the actual requirements, and obtain the optimal electric vehicle orderly charging plan.

[0016] Further, the specific steps for predicting the unordered charging load of electric vehicles in Step 2 are as follows:

[0017] Step 201: Determine the basic parameters for participating in orderly charging according to the simulation environment settings;

[0018] Step 202: Calculate the user's homecoming time, obtain the daily driving mileage of the user, calculate the charging duration and the initial SOC of the electric vehicle, and eliminate the unreasonable data generated;

[0019] Step 203: Determine whether the battery levels of all vehicles reach 100%. If the charging load calculation for all is completed, superimpose the basic load and the electric vehicle charging load, and go to Step 204; otherwise, return to Step 202;

[0020] Step 204: Determine whether the set number of simulation times is completed. If so, go to Step 205; otherwise, return to Step 202;

[0021] Step 205: After completing the iteration, calculate the mean value and superimpose it with the basic load.

[0022] Further, the basic parameters described in Step 201 include the number of electric vehicles, battery capacity, power consumption per unit mileage, and the charging power of the charging facilities.

[0023] Further, the formula for calculating the user's homecoming time in Step 202 is:

[0024]

[0025] In formula (1), t represents the homecoming time, with the unit of hour; σ s represents the probability density standard deviation, with a value of 3.4; μ sDenotes the expected value of the probability density, with a value of 17.6;

[0026] The formula for calculating the daily driving mileage of a user is:

[0027]

[0028] In formula (3), z represents the daily driving mileage of the vehicle, with the unit of kilometers; σ d Denotes the standard deviation of the probability density, with a value of 0.88; μ d Denotes the expected value of the probability density, with a value of 3.2;

[0029] The formula for calculating the charging duration is:

[0030]

[0031] In formula (4), T e Denotes the charging duration required for the electric vehicle, with the unit of h, D denotes the daily driving mileage of the electric vehicle, with the unit of km, E 100 Denotes the power consumption per 100 kilometers of the electric vehicle, with the unit of kw·h. Under approximate calculation conditions, the household electric vehicle is 17 kWh, P e Denotes the charging power of the electric vehicle, with the unit of kW, α denotes the charging station efficiency, α = 0.9;

[0032] The formula for calculating the initial SOC of the electric vehicle is:

[0033]

[0034] In formula (5), C represents the expected charging power of the electric vehicle, kw·h; B is the designed capacity of the electric vehicle battery, with the unit of kw·h.

[0035] Furthermore, the formula for formulating the dynamic time-of-use electricity price based on the day-ahead load in step 3 is:

[0036]

[0037] In formula (6), R(j) represents the time-of-use dynamic electricity price at time j, R g Denotes the off-peak electricity price of the local time-of-use electricity price, R f Denotes the peak electricity price of the local time-of-use electricity price, P O (j) represents the base load at time j, P e (j) represents the charging load of the electric vehicle at time j, P max Denotes the magnitude of the load peak in the day-ahead load forecast.

[0038] Furthermore, in step 6, it is judged whether the scheduling condition is satisfied. The scheduling condition is T stay >T e, the specific meaning is:

[0039] T stay = T r - T o (7),

[0040] In formula (7), T stay is the residence time of the electric vehicle at the charging station, T o is the user's homecoming time generated according to formula (1), that is, the start time of residence, T r is the user's leaving home time generated according to formula That is, the end time of residence. In formula (2), t is the leaving home time, with the unit of h; σ end is the standard deviation of the probability density, with a value of 3.24; μ end is the expected value of the probability density, with a value of 7.92.

[0041] Furthermore, the objective function set in step 7 is:

[0042]

[0043] In formulas (8) and (9), P O (j) is the base load in the jth period, P e (j) is the charging load of the electric vehicle in the jth period, P avg is the average load within the t period, P e (j) is the charging power of the electric vehicle within the jth time period, and R(j) is the dynamic time-of-use electricity price formulated according to formula (6).

[0044] The set constraint conditions are:

[0045] P O (j)+P e (j) < P allow (10),

[0046] 0 ≤ P e，i (j) ≤ P cmax (11),

[0047] T o ≤ T ≤ T r (12),

[0048] In formulas (10), (11), and (12), P O (j) is the base load in the jth period, P e (j) is the charging load of the electric vehicle in the jth period, P allow is the maximum power allowed by the transformer; P e，i (j) is the charging power of the ith electric vehicle at the jth moment, P cmaxis the rated maximum charging power of the electric vehicle; T stay is the residence time of the electric vehicle at the charging station, T o is the user's homecoming time generated according to formula (1), that is, the start time of the stay, T r is the user's leaving home time generated according to formula (2), that is, the end time of the stay.

[0049] Furthermore, the improved NSGA-II algorithm in step 8 specifically includes:

[0050] Step 801, Selection; Sort the population individuals in ascending order of fitness value. In NSGA-II, first sort the ranks in descending order, and then sort the crowding distances in ascending order. The higher the rank, that is, the smaller the fitness value, the higher the selection probability. Then use a linear function to allocate the selection probability of each individual;

[0051] Among them, x1 has the highest rank, and x n has the lowest rank. The individual x i The selection probability is:

[0052]

[0053] In formula (13), i = 1, 2,..., N, η + , η - is a constant, and 0 ≤ η - ≤ 1,

[0054] According to the selection probability, it can be seen that if η + = 2, η - = 0, the selection pressure of the population is the largest at this time. When η + = η - = 1, this is a random selection, and the selection pressure of the population is the smallest;

[0055] In order to effectively strengthen the selection pressure of the parent population during the calculation process, an exponential term is added to gradually adjust the values of η + , η - as the generations progress

[0056] Step 802, Crossover;

[0057] Step 803, Mutation; The improved self-adaptive adjustment mutation operator defines the mutation probability and fitness of an individual as:

[0058]

[0059] From formulas (18) and (19), it can be obtained that:

[0060]

[0061] In formulas (18), (19), (20), and (21), e(X i ) is the fitness of individual X i , E(X i ) is the fitness evaluation function of X i , P m is the mutation probability, is the average mutation probability, and M is the number of individuals.

[0062] It can be obtained from formula (18) that the larger the fitness of an individual, the smaller the value of its mutation probability; the smaller the fitness of an individual, the larger the value of its mutation probability. It can be seen from formula (20) that the self-adaptive adjustment of the mutation method keeps the sum of the mutation probabilities of the population individuals unchanged, which not only ensures their diversity but also speeds up the optimization rate;

[0063] Step 804, fast non-dominated sorting;

[0064] Step 805, crowding degree calculation;

[0065] Step 806, elitist retention strategy.

[0066] Furthermore, the specific process of the crossover in step 802 includes:

[0067] Step 8021, generate a random number t ∈ (0, 1];

[0068] Step 8022, if t ≤ 0.5, then:

[0069]

[0070] Step 8023, if t > 0.5, then:

[0071]

[0072] In formula (17), |N(0, 1)| is a normally distributed random variable.

[0073] Furthermore, the specific steps of the fast non-dominated sorting in step 804 include:

[0074] Step 8041, initialization: First, initialize each individual x, including the number of individuals it is dominated by and the set of individuals that dominate this individual;

[0075] Step 8042, then divide all individuals into multiple non-dominated levels. For each level, traverse each individual x in the population i , and compare its dominance relationship with other individuals x j . If individual x i is dominated by individual x j , then increase the number of individuals that x i is dominated by, and at the same time add xj Add x i to the dominating set. If individual x i is not dominated by any other individual, add it to the first non-dominated layer;

[0076] Step 8043: Repeat the above steps until all individuals are assigned to the corresponding non-dominated levels. In each iteration, only consider individuals with a domination count of 0, remove them from the population, and update the domination relationships of the remaining individuals. Through this process, the individuals in the population are divided into multiple non-dominated levels, where the first layer contains individuals not dominated by other individuals, the second layer contains individuals dominated only by the individuals in the first layer, and so on.

[0077] The beneficial effects of the present invention are as follows:

[0078] 1. The present invention proposes a dynamic time-of-use electricity price formulation strategy based on day-ahead load forecasting and an improved NSGA-II algorithm, moderately adjusts the traditional time-of-use electricity price, and regulates users' charging behaviors by regulating the electricity price to avoid the occurrence of reverse peak regulation phenomena. Improvements are made to the NSGA-II algorithm in terms of selection, crossover, mutation, and retention strategies;

[0079] 2. In terms of selection, the present invention adopts a parent selection method based on linear ranking, proposes a crossover operator based on normal distribution in terms of crossover, proposes an adaptive mutation strategy in terms of mutation strategy, and makes certain improvements to the traditional elite retention strategy;

[0080] 3. The present invention formulates an orderly charging strategy for electric vehicles by combining a day-ahead electric vehicle charging load forecasting model based on Monte Carlo and an electric vehicle charging model;

[0081] 4. The present invention proposes an orderly charging model for electric vehicles, sets up an objective function that meets actual requirements and constraint conditions that conform to the actual situation, uses the Monte Carlo method to predict the electric vehicle charging load, and proposes a dynamic time-of-use electricity price strategy based on the day-ahead load forecasting results. The model is solved by combining the improved NSGA-II algorithm, realizing the formulation of an orderly charging strategy for electric vehicles that simultaneously considers the optimization of charging costs and load variance. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 is a framework diagram for formulating an orderly charging strategy for electric vehicles;

[0083] Figure 2 is a flowchart for forecasting the electric vehicle charging load;

[0084] Figure 3 is a flowchart for orderly charging;

[0085] Figure 4It is a Pareto rank schematic diagram;

[0086] Figure 5 It is a congestion degree schematic diagram;

[0087] Figure 6 It is a flowchart of the NSGA-II algorithm. Specific implementation manners

[0088] Specific implementation manner one: In combination with Figures 1 to 6 To illustrate this implementation manner, the steps of a method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices described in this implementation manner include:

[0089] Step 1: Organize the probability density function of the travel characteristics of electric vehicles;

[0090] Step 2: Adopt the Monte Carlo simulation method to simulate the situation of a large number of electric vehicles accessing the power grid, and obtain the unordered charging load of electric vehicles;

[0091] Step 3: Based on the unordered charging load of electric vehicles obtained in Step 2, formulate a dynamic time-of-use electricity price based on the daily load;

[0092] Step 4: Input the dynamic time-of-use electricity price obtained in Step 3 into the electric vehicle charging model;

[0093] Step 5: For each connected electric vehicle, use the Monte Carlo method to simulate its travel characteristics;

[0094] Step 6: Judge whether the scheduling conditions are met, and judge whether the electric vehicle participates in the scheduling;

[0095] Step 7: Set the objective function and constraint conditions;

[0096] Step 8: Use the improved NSGA-II algorithm for optimization and solution;

[0097] Step 9: After reaching the set number of iterations, obtain a set of Pareto optimal solutions, and select the solution with the best comprehensive effect according to the actual needs to obtain the optimal electric vehicle orderly charging scheme.

[0098] Specific implementation manner two: In combination with Figures 1 to 6 To illustrate this implementation manner, the specific steps for predicting the unordered charging load of electric vehicles in Step 2 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices described in this implementation manner are:

[0099] Step 201: According to the simulation environment settings, determine the basic parameters for participating in the orderly charging;

[0100] Step 202: Calculate the user's homecoming time, obtain the user's daily driving mileage, calculate the charging duration and the initial SOC of the electric vehicle, and eliminate the generated unreasonable data;

[0101] Step 203: Determine whether the battery levels of all vehicles reach 100%. If the charging load calculation is completed for all, superimpose the basic load and the electric vehicle charging load, and proceed to Step 204; otherwise, return to Step 202.

[0102] Step 204: Determine whether the set number of simulation times is completed. If so, proceed to Step 205; otherwise, return to Step 202.

[0103] Step 205: After the iteration is completed, calculate the mean value and superimpose it with the basic load.

[0104] Specific Embodiment 3: Combined Figures 1 to 6 To illustrate this embodiment, the basic parameters described in Step 201 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices in this embodiment include the number of electric vehicles, battery capacity, power consumption per unit mileage, and the charging power of the charging facilities.

[0105] Specific Embodiment 4: Combined Figures 1 to 6 To illustrate this embodiment, the formula for calculating the user's homecoming time in Step 202 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices in this embodiment is:

[0106]

[0107] In formula (1), t represents the homecoming time, with the unit of hour; σ s represents the probability density standard deviation, with a value of 3.4; μ s represents the probability density expected value, with a value of 17.6;

[0108] The formula for calculating the user's daily driving mileage is:

[0109]

[0110] In formula (3), z represents the daily driving mileage of the vehicle, with the unit of kilometer; σ d represents the probability density standard deviation, with a value of 0.88; μ d represents the probability density expected value, with a value of 3.2;

[0111] The formula for calculating the charging duration is:

[0112]

[0113] In formula (4), T e represents the charging duration required for the electric vehicle, with the unit of h, D represents the daily driving mileage of the electric vehicle, with the unit of km, E 100Indicates the power consumption of an electric vehicle per 100 kilometers, with the unit of kw·h. Under approximate calculation conditions, the power consumption of a household electric vehicle is 17 kWh, P e Indicates the charging power of the electric vehicle, with the unit of kW. α represents the charging station efficiency, and α = 0.9;

[0114] The formula for calculating the initial SOC of an electric vehicle is:

[0115]

[0116] In formula (5), C represents the expected charging power of the electric vehicle, in kw·h; B is the designed capacity of the electric vehicle battery, with the unit of kw·h.

[0117] Specific implementation method five: Combining Figures 1 to 6 To illustrate this implementation method, in step 3 of the method for controlling the orderly charging of an electric vehicle based on dynamic time-of-use electricity prices described in this implementation method, the formula for formulating the dynamic time-of-use electricity price based on the daily load is:

[0118]

[0119] In formula (6), R(j) represents the time-of-use dynamic electricity price at time j, R g Represents the valley electricity price of the local time-of-use electricity price, R f Represents the peak electricity price of the local time-of-use electricity price, P O (j) represents the base load at time j, P e (j) represents the charging load of the electric vehicle at time j, P max Represents the magnitude of the load peak in the daily load forecast.

[0120] Specific implementation method six: Combining Figures 1 to 6 To illustrate this implementation method, in step 6 of the method for controlling the orderly charging of an electric vehicle based on dynamic time-of-use electricity prices described in this implementation method, it is determined whether the scheduling condition is met. The scheduling condition is T stay >T e , and the specific meaning is:

[0121] T stay =T r -T o (7),

[0122] In formula (7), T stay Is the residence time of the electric vehicle at the charging station, T o Is the user's homecoming time generated according to formula (1), that is, the start time of the stay, T r Is the user's leaving home time generated according to the formula Generated, that is, the end time of the stay. In formula (2), t is the leaving home time, with the unit of h; σ endis the standard deviation of the probability density, with a value of 3.24; μ end is the expected value of the probability density, with a value of 7.92.

[0123] Specific Embodiment Seven: In combination with Figures 1 to 6 to illustrate this embodiment, the objective function set in step 7 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices described in this embodiment is:

[0124]

[0125] In formulas (8) and (9), P O (j) is the base load in the j-th period, P e (j) is the charging load of electric vehicles in the j-th period, P avg is the average load within the t-th period, P e (j) is the charging power of electric vehicles within the j-th time period, and R(j) is the dynamic time-of-use electricity price formulated according to formula (6);

[0126] The set constraint conditions are:

[0127] P O (j) + P e (j) < P allow (10),

[0128] 0 ≤ P e，i (j) ≤ P cmax (11),

[0129] T o ≤ T ≤ T r (12),

[0130] In formulas (10), (11), and (12), P O (j) is the base load in the j-th period, P e (j) is the charging load of electric vehicles in the j-th period, P allow is the maximum power allowed by the transformer; P e,i (j) is the charging power of the i-th electric vehicle at the j-th moment, P cmax is the rated maximum charging power of the electric vehicle; T stay is the residence time of the electric vehicle at the charging station, T o is the user's homecoming moment generated according to formula (1), that is, the start time of residence, T r is the user's leaving home moment generated according to formula (2), that is, the end time of residence.

[0131] Specific Embodiment Eight: In combination with Figures 1 to 6Regarding this embodiment, the improved NSGA-II algorithm in step 8 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices described in this embodiment specifically includes:

[0132] Step 801, Selection: Sort the population individuals in ascending order of fitness value. In NSGA-II, first sort the ranks in descending order, and then sort the crowding distances in ascending order. The higher the rank, that is, the smaller the fitness value, the higher the selection probability. Then use a linear function to allocate the selection probability of each individual;

[0133] Among them, x1 has the highest rank, and x n has the lowest rank. The selection probability of individual x i is:

[0134]

[0135] In formula (13), i = 1, 2,..., N, η + , η - is a constant, and 0 ≤ η - ≤ 1,

[0136] According to the selection probability, it can be known that if η + = 2, η - = 0, the selection pressure of the population is the largest at this time. When η + = η - = 1, this is a random selection, and the selection pressure of the population is the smallest;

[0137] In order to effectively strengthen the selection pressure of the parent population during the calculation process, an exponential term is added to gradually adjust the values of η + , η - as the generations progress

[0138] Step 802, Crossover;

[0139] Step 803, Mutation; The improved self-adaptive adjustment mutation operator defines the mutation probability and fitness of an individual as:

[0140]

[0141] From formulas (18) and (19), it can be obtained that:

[0142]

[0143] In formulas (18), (19), (20), and (21), e(X i ) is the fitness of individual X i , E(X i ) is the fitness evaluation function of X i , P m is the mutation probability, is the average mutation probability, and M is the number of individuals;

[0144] It can be obtained from formula (18) that the larger the fitness of an individual, the smaller the value of its mutation probability; the smaller the fitness of an individual, the larger the value of its mutation probability; it can be seen from formula (20) that the adaptive adjustment of the mutation method keeps the sum of the mutation probabilities of the population individuals unchanged, which not only ensures its diversity but also speeds up the optimization rate;

[0145] Step 804, fast non-dominated sorting;

[0146] Step 805, crowding degree calculation;

[0147] Step 806, elitist retention strategy.

[0148] In this embodiment, in order to ensure diversity, the concept of crowding degree is introduced. The crowding degree of solution n is related to the positions of the n - 1 point and the n + 1 point. The crowding degree of the n point is the perimeter of the square; the crowding degrees of the leftmost and rightmost points are set to infinity; points with a larger crowding degree are more capable of maintaining population diversity and are more likely to develop new populations;

[0149] In step 806, the parent population and the offspring population are combined into a new population. The entire layer of the population is listed in the next-generation parent population in ascending order of Pareto rank until a certain layer cannot be completely placed in the next-generation parent population. Then, the individuals in this layer are sorted in descending order according to the crowding degree and are placed in the next-generation parent population in turn until it is full;

[0150] In order to better help the algorithm select more excellent individuals to form the elite set, the concept of "pseudo-fitness value" is proposed. This concept allows increasing the number of elite individuals saved. Among them, individuals with a smaller pseudo-fitness value are considered more excellent. This method helps to improve the performance of the algorithm and ensures that those more excellent solutions can be effectively retained during the evolution process. The calculation is as follows:

[0151]

[0152] In formula (22), N i is the maximum number of individuals in the i-th non-dominated layer: N is the population size; K is the order of the Pareto front; r ∈ [0, 1] is the decay rate; after optimization, when the number of non-dominated solution sets is small in the early stage of population iteration, elite individuals are well retained, and in the later stage when the number of non-dominated solution sets is large, the number of elite individuals retained can be effectively expanded, making the population fast.

[0153] Specific Embodiment Nine: Combined Figures 1 to 6 To illustrate this embodiment, the process of crossover in step 802 of the method for controlling the orderly charging of electric vehicles based on dynamic time-of-use electricity prices described in this embodiment specifically includes:

[0154] Step 8021: Generate a random number \(t\in(0, 1]\);

[0155] Step 8022: If \(t\leq0.5\), then:

[0156]

[0157] Step 8023: If \(t > 0.5\), then:

[0158]

[0159] In formula (17), \(|N(0, 1)|\) is a normally distributed random variable.

[0160] Specific Embodiment Ten: Combining Figures 1 to 6 To illustrate this embodiment, the specific steps of the fast non - dominated sorting in step 804 of the method for controlling the orderly charging of electric vehicles based on dynamic time - of - use electricity prices described in this embodiment are as follows:

[0161] Step 8041: Initialization: First, initialize each individual \(x\), including the number of individuals it is dominated by and the set of individuals that dominate this individual;

[0162] Step 8042: Then divide all individuals into multiple non - dominated levels. For each level, traverse each individual \(x\) in the population i , and compare its dominance relationship with other individuals \(x\) j ; If the individual \(x\) i is dominated by the individual \(x\) j , then increase the number of individuals that \(x\) i is dominated by, and at the same time add \(x\) j to the set of individuals that dominate \(x\) i ; If the individual \(x\) i is not dominated by any other individual, then add it to the first - layer non - dominated level;

[0163] Step 8043: Repeat the above steps until all individuals are assigned to the corresponding non - dominated levels. In each round of iteration, only consider individuals with a dominance count of 0, remove them from the population, and update the dominance relationships of the remaining individuals. Through this process, the individuals in the population are divided into multiple non - dominated levels, where the first layer contains individuals that are not dominated by other individuals, the second layer contains individuals that are only dominated by the individuals in the first layer, and so on.

[0164] The above are only the preferred embodiments of the present invention and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the above-disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention and is based on the technical essence of the present invention, any simple modification, equivalent replacement, and improvement made to the above embodiments still fall within the protection scope of the technical solution of the present invention.

Claims

1. An orderly charging control method for electric vehicles based on dynamic time-of-use electricity prices, characterized in that: The steps of the orderly charging control method of an electric vehicle based on dynamic time-of-use electricity price include: Step 1: Arrange the probability density function of electric vehicle travel characteristics; Step 2: Use the Monte Carlo simulation method to simulate the situation where a large number of electric vehicles are connected to the power grid to obtain the disorderly charging load of electric vehicles; the specific steps are: Step 201: Determine basic parameters for orderly charging according to simulation environment settings; Step 202: Calculate the user's homecoming time, obtain the user's daily mileage, calculate the charging time and the electric vehicle's starting SOC, and remove the unreasonable data generated; Step 203, determine whether the battery power of any vehicle has reached 100%, if the charging load calculation is completed, superimpose the basic load and the electric vehicle charging load, and go to step 204, otherwise return to step 202; Step 204, determine whether the set number of simulations is completed, if so, go to step 205, otherwise return to step 202; Step 205: after the iteration is completed, the average is calculated and superimposed with the base load; Step 3: Based on the disorderly charging load of electric vehicles obtained in step 2, a dynamic time-of-use electricity price based on the day-ahead load is formulated; Step 4: Input the dynamic time-of-use electricity price obtained in step 3 into the electric vehicle charging model; Step 5: For each connected tram, use the Monte Carlo method to simulate its travel characteristics; Step 6: Determine whether the dispatching conditions are met and whether the tram participates in the dispatching; Step 7: Set the objective function and constraints; Step 8: Use the improved NSGA-Ⅱ algorithm to optimize and solve; Step 9: After reaching the set number of iterations, a set of Pareto optimal solutions are obtained. The solution with the best comprehensive effect is selected according to actual needs to obtain the optimal orderly charging plan for electric vehicles.

2. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: The basic parameters described in step 201 include the number of electric vehicles, battery capacity, power consumption per unit mileage, and charging power of charging facilities.

3. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: The formula for calculating the user's homecoming time in step 202 is: In formula (1), t represents the time of returning home, in hours; σ s represents the standard deviation of probability density, with a value of 3.4; μ s represents the expected value of probability density, which is 17.6; The formula for calculating a user's daily mileage is: In formula (3), z represents the daily mileage of the car, in kilometers; σ d represents the standard deviation of probability density, with a value of 0.88; μ d represents the expected value of probability density, which is 3.2; The formula for calculating charging time is: In formula (4), T e It represents the time required to charge an electric vehicle in h, D represents the daily mileage of an electric vehicle in km, and E 100 It represents the power consumption of electric vehicles per 100 kilometers, in kw·h. Under approximate calculation conditions, household electric vehicles consume 17 kWh. e represents the charging power of electric vehicles, in kW, α represents the efficiency of the charging station, α=0.9; The formula for calculating the starting SOC of an electric vehicle is: In formula (5), C represents the expected charging power of the electric vehicle, in kw·h; B is the design capacity of the electric vehicle battery, in kw·h.

4. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: The formula for formulating the dynamic time-of-use electricity price based on the day-ahead load in step 3 is: In formula (6), R(j) represents the time-of-use dynamic electricity price in period j, R g represents the local time-of-use electricity price and the valley electricity price, R f represents the local time-of-use peak electricity price, P O (j) represents the base load in period j, P e (j) represents the electric vehicle charging load during period j, P max Indicates the load peak size in the day-ahead load forecast.

5. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: In step 6, it is determined whether the scheduling condition is met, and the scheduling condition is T stay >T e , T e is the time required for electric vehicle charging generated according to formula (4), and its specific meaning is: T stay =T r -T o (7), In formula (7), T stay is the time that the electric vehicle stays at the charging station, T o is the user's homecoming time generated according to formula (1), i.e., the start time of stay, T r According to the formula The generated time when the user leaves home, that is, the end time of the stay, in formula (2), t is the time of leaving home, the unit is h; σ end is the standard deviation of the probability density, which is 3.24; μ end is the expected value of the probability density, which is 7.

92.

6. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: The objective function set in step 7 is: In formulas (8) and (9), P O (j) is the base load for period j, P e (j) is the electric vehicle charging load in period j, P avg is the average load in period t, R(j) is the dynamic time-of-use electricity price formulated according to formula (6); The constraints set are: P O (j)+P e (j)<P allow (10), 0≤P e,i (j)≤P cmax (11), T o ≤T≤T r (12), In formulas (10), (11), and (12), P O (j) is the base load for period j, P e (j) is the electric vehicle charging load in period j, P allow The maximum power allowed by the transformer; P e，i (j) is the charging power of the i-th electric vehicle at time j, P cmax is the rated maximum charging power of the electric vehicle; T o is the user's homecoming time generated according to formula (1), i.e., the start time of stay, T r is the time when the user leaves home generated according to formula (2), that is, the end time of the stay.

7. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 1 is characterized in that: The improved NSGA-Ⅱ algorithm in step 8 specifically includes: Step 801, selection; sort the individuals in the population from small to large according to the fitness value. In NSGA-Ⅱ, the ranks are first sorted in descending order, and then the crowding distance is sorted in ascending order. The higher the ranking, that is, the smaller the fitness value, the higher the selection probability. Then, a linear function is used to allocate the selection probability of each individual. Among them, x1 ranks the highest, n The lowest ranking, individual x i The probability of selection is: In formula (13), i = 1, 2, ..., N, η + , η - is a constant, and 0≤η - ≤1, According to the selection probability, if η + =2, η - = 0, the selection pressure of the population is the largest at this time, when η + =η - =1, it is a random selection, and the pressure of population selection is the smallest; In order to effectively strengthen the selection pressure of the parent population during the calculation process, an exponential term is added to gradually adjust η as the generations progress. + , η - The value of Step 802, cross; Step 803, mutation: The improved adaptive adjustment mutation operator defines the mutation probability and fitness of an individual as: From formulas (18) and (19), we can get: In formulas (18), (19), (20), and (21), e(x i ) is individual x i The fitness of E(x i ) is x i The fitness evaluation function, P m is the mutation probability, is the average mutation probability, M is the number of individuals; From formula (18), it can be concluded that the greater the fitness of an individual, the smaller the value of its mutation probability; the smaller the fitness of an individual, the larger the value of its mutation probability; from formula (20), it can be seen that the adaptive adjustment of the mutation mode makes the mutation probability and of the individuals in the population unchanged, which not only ensures its diversity, but also speeds up the optimization rate; Step 804: fast non-dominated sorting; Step 805: Calculate the congestion degree; Step 806: Elite retention strategy.

8. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 7 is characterized in that: The crossover process in step 802 specifically includes: Step 8021: Generate a random number t∈(0,1]; Step 8022: If t≤0.5, then: Step 8023: If t>0.5, then: In formula (17), |N(0,1)| is a normally distributed random variable.

9. The method for controlling orderly charging of electric vehicles based on dynamic time-of-use electricity prices according to claim 7 is characterized in that: The specific steps of the fast non-dominated sorting in step 804 include: Step 8041, initialization: First, each individual x is initialized, including the number of individuals it dominates and the set of individuals that dominate the individual; Step 8042: Divide all individuals into multiple non-dominated levels. For each level, traverse each individual x in the population. i , and compare it with other individuals x j The dominance relationship; if individual x i By individual x j Dominate, then x i The number of dominance increases, and x j Join x i In the dominating set of i If it is not dominated by any other individual, it is added to the first non-dominated layer; Step 8043, repeat the above steps until all individuals are divided into corresponding non-dominated levels; in each round of iteration, only individuals with 0 dominance number are considered, removed from the population, and the dominance relationship of the remaining individuals is updated; through this process, the individuals in the population are divided into multiple non-dominated levels, where the first level contains individuals that are not dominated by other individuals, the second level contains individuals that are only dominated by individuals in the first level, and so on.

Images