A user-willingness-driven electric vehicle charging optimization method

Abstract

The present application relates to electric vehicle charging scheduling technical field, specifically, it is a kind of electric vehicle charging optimization method based on user willingness drive, the problem solved by the present application: how to establish accurate user willingness prediction model, realize the problem of differentiating subsidy strategy, to solve the above problems, the present application provides a kind of charging optimization method, comprising: generating agent individual according to multi-agent model, and establishing the complete behavior characteristics of agent individual;User willingness discrimination is carried out based on scene parameter and psychological threshold, and a reference quantitative model for user participation decision is established;According to user individual characteristics, implement subsidy allocation, establish the basic allocation rule of differentiating subsidy;According to external environmental parameters, the schedulable space of target area is predicted.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging scheduling technology, and more specifically, to an electric vehicle charging optimization method driven by user willingness. Background Technology

[0002] With the rapid development of electric vehicles, they are connected to the power grid in large numbers during certain periods of daily life. If their charging load is directly superimposed on the existing power load, it may cause overload of the existing power distribution network and have a great impact on the safety of the power grid. Therefore, it is necessary to implement orderly charging control for electric vehicles.

[0003] Electric vehicle charging scheduling methods primarily focus on technical optimization, neglecting the crucial factor of user participation. Actual user participation decisions involve a comprehensive consideration of multiple influencing factors, including charging time, battery status, and subsidy prices. Therefore, the decision-making characteristics are not entirely equivalent to pure economic optimization. In a power market environment, improving user participation in demand response through reasonable incentive mechanisms is an important means to achieve peak shaving and valley filling in the power grid and improve system operating efficiency. Furthermore, after electric vehicle charging scheduling, it is necessary to monitor the completed scheduling work to avoid resource waste and increased scheduling costs due to excessive scheduling. Summary of the Invention

[0004] The problem this invention addresses is how to establish an accurate user willingness prediction model to implement a differentiated subsidy strategy.

[0005] To address the aforementioned issues, this invention provides a user-intention-driven electric vehicle charging optimization method. The method includes: establishing a multi-agent model based on questionnaire data to analyze user behavior within a target area; generating agent individuals based on the multi-agent model and establishing complete behavioral characteristics of each agent individual; determining user participation intention based on scenario parameters and psychological thresholds, handling boundary conditions with relaxed discrimination conditions, and establishing a reference quantitative model for user participation decision-making; implementing subsidy allocation based on individual user characteristics and establishing basic allocation rules for differentiated subsidies; predicting the schedulable space of the target area based on external environmental parameters, and determining the initial user set participating in scheduling within the target area based on the schedulable space and basic allocation rules; obtaining the actual scheduling space of the target area, and determining whether the target area is saturated based on the scheduling usage of the initial user set; if not, establishing a user-oriented optimized scheduling system based on the initial user set; if so, adjusting the basic allocation rules based on the reference quantitative model, changing the number of users in the initial user set to change the scheduling usage.

[0006] Compared with existing technologies, the technical effects achieved by this solution are as follows: the multi-model agent generation method makes the agent individuals more consistent with the real people in the questionnaire data, forming an accurate user intention prediction model; the setting of scene parameters and psychological thresholds makes user intentions more consistent with the actual scheduling situation; the setting of differentiated subsidies can better attract different users to participate in the scheduling work; the acquisition of scheduling space can make adjustments in real time according to the power consumption during the scheduling process; the determination of saturation state allows the scheduling work to make more economical decisions based on subsidy costs and scheduling space, while also avoiding the impact of secondary peaks caused by scheduling work, thus improving the practicality of the optimization method.

[0007] In one embodiment of the present invention, a multi-agent model is established based on questionnaire data to target user behavior within a target area. Specifically, this includes: constructing a probability distribution of user types based on questionnaire data to establish a statistical basis for user behavior characteristics; constructing a joint probability distribution of the importance ranking of reference factors based on questionnaire data, statistically analyzing the frequency distribution of different reference factors at different ranking positions to form a quantitative representation of user preferences; constructing a psychological threshold conditional probability distribution to establish a quantitative model of the degree of user acceptance of different reference factors, and setting psychological threshold levels.

[0008] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: By setting up a questionnaire data matrix, the reference standards of users regarding charging time, on-grid SOC, off-grid SOC, and subsidy price are fully considered when charging. The joint probability distribution of importance ranking can be obtained to obtain the degree of preference of each user for different reference factors. The conditional probability distribution of psychological threshold can better simulate the degree of user's active response to power dispatch under different conditions. Furthermore, users are divided by psychological threshold levels, which improves the accuracy of the quantitative model.

[0009] In one embodiment of the present invention, agent individuals are generated according to a multi-agent model, and complete behavioral characteristics of the agent individuals are established. Specifically, this includes: generating an importance ranking of the agent individuals, randomly selecting complete ranking combinations from the questionnaire data, and maintaining the internal logical consistency of the ranking; generating psychological thresholds for the agent individuals, independently sampling from the marginal probability distributions of different reference factors, and establishing a complete psychological profile of the agent individuals.

[0010] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by ranking the importance, it is possible to accurately obtain the degree of attention different users pay to different factors when participating in power dispatch. The setting of psychological characteristic profiles and psychological thresholds can obtain the differences between different factors and the user's ideal state, and obtain more accurate user participation intentions.

[0011] In one embodiment of the present invention, user participation intention is judged based on scenario parameters and psychological thresholds, and boundary cases are handled by relaxed discrimination conditions to establish a reference quantitative model for user participation decision-making. Specifically, this includes: determining piecewise linear transformation functions for on-grid SOC, off-grid SOC, charging time, and subsidy price based on scenario parameters; analyzing questionnaire data to obtain transformation rules for each scenario parameter; and predicting user participation behavior based on psychological thresholds and importance ranking to obtain the reference quantitative model.

[0012] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by setting a conversion function to map continuous physical quantities into discrete satisfaction levels, it is easier to compare with psychological thresholds. The conversion boundary is determined based on quantile analysis of large-sample user surveys, ensuring the balance of user distribution at each level. The method for judging participation intention also reflects the conservative and risk-averse characteristics of user decision-making, thus improving the accuracy of prediction.

[0013] In one embodiment of the present invention, subsidies are allocated based on individual user characteristics, and a basic allocation rule for differentiated subsidies is established. Specifically, this includes: recording reference factors that meet the psychological threshold as satisfaction factors and reference factors that do not meet the psychological threshold as dissatisfaction factors; providing differentiated subsidies based on the user's dissatisfaction factors and their importance ranking; and establishing conversion functions between different comprehensive threshold levels and subsidy prices.

[0014] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: the design of relaxation conditions avoids overly strict discrimination criteria, improves the model's fit to actual user behavior, and the setting of the transformation function ensures that the subsidy level matches user expectations, thereby increasing users' enthusiasm for participating in power dispatch.

[0015] In one embodiment of the present invention, the dispatchable space of the target area is predicted based on external environmental parameters, and the initial set of users participating in dispatch within the target area is determined based on the dispatchable space and basic allocation rules. Specifically, this includes: predicting the peak excess power and valley surplus power of the power grid based on external environmental parameters, and determining the dispatchable space based on the difference between the peak excess power and valley surplus power; obtaining a theoretical set of users with dispatch intentions based on basic allocation rules, and calculating the dispatchable power of the theoretical user set participating in dispatch; when the dispatchable power is less than or equal to the dispatchable space, the theoretical user set is recorded as the initial user set; when the dispatchable power is greater than the dispatchable space, the theoretical user set is filtered based on the subsidy price to obtain the initial user set.

[0016] Compared with existing technologies, the technical effects achieved by adopting this solution are as follows: by predicting the peak excess power and valley surplus power of the power grid through external environmental parameters, the power dispatching work is more in line with the current working environment of the power grid. The setting of dispatchable space can not only reduce the execution cost of dispatching work, but also avoid the occurrence of secondary peak power consumption after the dispatching work is carried out. By filtering the theoretical user set through the dispatching space, the number of users participating in the power dispatching work can be accurately reduced, thereby improving the adaptability of the initial user set to the power dispatching work.

[0017] In one embodiment of the present invention, obtaining the actual scheduling space of the target area and determining whether the target area is in a saturated state based on the scheduling usage of the initial user set specifically includes: comparing the actual scheduling space with the scalable space to obtain a variation coefficient; recording users who have performed scheduling in the initial user set as completed users and those who have not performed scheduling as remaining users, with the current electricity consumption of completed users as the scheduling usage and the subsequent electricity consumption of remaining users as the scheduling amount to be optimized; obtaining a scheduling difference based on the actual scheduling space and the scheduling usage, and obtaining a saturation threshold based on the scheduling difference and the variation coefficient; when the scheduling amount to be optimized is greater than or equal to the saturation threshold, the target area is in a saturated state; when the scheduling amount to be optimized is less than the saturation threshold, the target area is not in a saturated state.

[0018] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: the setting of the variation coefficient fully considers the difference between the current power grid operating state and the power grid operating state in historical data; the setting of the saturation threshold intuitively reflects the maximum dispatchable power required by the power grid under the current operating state; and the determination of the saturation state provides a reference standard for the subsequent execution of dispatching work, which facilitates timely adjustments to the dispatching work.

[0019] In one embodiment of the present invention, if so, the basic allocation rules are adjusted according to the reference quantification model to change the number of users in the initial user set to change the scheduling usage. Specifically, this includes: analyzing the satisfaction and dissatisfaction factors corresponding to the remaining users and establishing a correction model for the remaining users; adjusting the comprehensive threshold levels corresponding to the satisfaction and dissatisfaction factors to reduce the number of users among the remaining users who continue to perform scheduling until the scheduling amount to be optimized is lower than the saturation threshold.

[0020] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by lowering the threshold level, the optimal power dispatching method can be selected in real time according to the working status of the power grid. Compared with directly performing dispatching work based on the initial user set, it avoids the occurrence of ineffective dispatching as much as possible and reduces the expenditure of ineffective subsidies. Attached Figure Description

[0021] Figure 1This is one of the flowcharts for the electric vehicle charging optimization method of the present invention;

[0022] Figure 2 This is the second flowchart of the electric vehicle charging optimization method of the present invention;

[0023] Figure 3 This is the third flowchart of the electric vehicle charging optimization method of the present invention;

[0024] Figure 4 This is the fourth flowchart of the electric vehicle charging optimization method of the present invention;

[0025] Figure 5 This is the fifth flowchart of the electric vehicle charging optimization method of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0027] [First Embodiment]

[0028] See Figure 1 In one specific embodiment, this application provides a user-will-driven electric vehicle charging optimization method, which includes:

[0029] S100. Establish a multi-agent model based on questionnaire data to analyze user behavior within the target area;

[0030] S200. Generate agent individuals based on the multi-agent model and establish complete behavioral characteristics of the agent individuals;

[0031] S300: Based on scene parameters and psychological thresholds, user participation intention is judged, and relaxed discrimination conditions are used to handle boundary cases, and a reference quantitative model for user participation decision-making is established.

[0032] S400. Implement subsidy allocation based on individual user characteristics and establish basic allocation rules for differentiated subsidies;

[0033] S500: Predict the schedulable space of the target area based on external environmental parameters, and determine the initial set of users participating in scheduling within the target area based on the schedulable space and basic allocation rules;

[0034] S600: Obtain the actual scheduling space of the target area, and determine whether the target area is saturated based on the scheduling usage of the initial user set;

[0035] S610. If not, then establish a user-oriented optimized scheduling system based on the initial user set.

[0036] S620. If so, the basic allocation rules are adjusted according to the reference quantization model, and the number of users in the initial user set is changed to change the scheduling usage.

[0037] In step S100, the target area is usually an administrative planning area, but it can also be a city, a town or a village. The questionnaire data is obtained through offline surveys, which are conducted regularly to collect the charging scheduling intentions of electric vehicle users in the target area. The multi-agent model is an agent model that combines user participation willingness, subsidy strategy and user individual difference identification.

[0038] In step S200, a virtual user agent is generated based on the questionnaire data, and the results obtained from the virtual user agent are used to simulate the impact on each user after the scheduling work changes.

[0039] In step S300, the scenario parameters include the network access SOC ( S arr Off-grid SOC ( S dep ), charging time ( T ), subsidized price ( p The system sets psychological thresholds for each of these four parameters and compares these thresholds with the actual parameters to determine whether a user will participate in the scheduling process.

[0040] In step S500, the external environmental parameters include, but are not limited to, weather, temperature and current date. The schedulable space refers to the time period that needs peak shaving and valley filling, which is obtained based on the historical data of the power grid under the current external environmental parameters, and the amount of electricity consumption that needs to be adjusted during this time period. The initial user set is the electric vehicle users participating in the scheduling, which are obtained from the basic allocation rules published according to the amount of electricity consumption that needs to be adjusted.

[0041] In steps S600 to S620, the actual scheduling space is the remaining power supply that needs to be adjusted during the peak period during the gradual scheduling process. If a large number of electric vehicles perform scheduling during the peak period, it will cause the power supply to drop too much during the peak period. Since the scheduling work is based on subsidies, in order to avoid unnecessary scheduling, the scheduling usage is used as a criterion. When the scheduling work has achieved the desired effect, the subsequent scheduling is stopped in time. This can not only save subsidy costs, but also prevent a new peak period from being formed during the off-peak period after a large number of electric vehicles perform scheduling.

[0042] When the target area is saturated, it means that the power supply that needs to be adjusted during peak hours has not been completely reduced, and electric vehicles need to continue to reduce their power consumption. At this time, the initial user set will execute the original scheduling plan. When the target area is not saturated, it means that electric vehicles do not need to reduce their power consumption. However, if a stop scheduling instruction is issued to electric vehicles that have not yet executed, it will make it difficult for subsequent scheduling work to continue. Therefore, the basic allocation rules can be adjusted to allow some initial user sets to withdraw on their own.

[0043] It should be noted that electric vehicles that have already undergone scheduling will not be affected by the temporary basic allocation rule adjustments.

[0044] The multi-model agent generation method makes the agent individuals more consistent with the real people in the questionnaire data, forming an accurate user intention prediction model. The setting of scenario parameters and psychological thresholds makes user intentions more consistent with the actual scheduling situation. The setting of differentiated subsidies can better attract different users to participate in the scheduling work. The acquisition of scheduling space can make adjustments in real time according to the power consumption during the scheduling process. The saturation state determination allows the scheduling work to make more economical decisions based on subsidy costs and scheduling space, while also avoiding the impact of secondary peaks caused by scheduling work, thus improving the practicality of the optimization method.

[0045] [Second Embodiment]

[0046] See Figure 2 In one specific embodiment, a multi-agent model is established based on questionnaire data to analyze user behavior within a target area, specifically including:

[0047] S110. Construct a probability distribution of user types based on questionnaire data to establish a statistical basis for user behavior characteristics;

[0048] S120. Construct a joint probability distribution of the importance ranking of reference factors based on questionnaire data, and statistically analyze the frequency distribution of different reference factors in different ranking positions to form a quantitative representation of user preferences.

[0049] S130. Construct a psychological threshold conditional probability distribution, establish a quantitative model of the user's acceptance of different reference factors, and set psychological threshold levels.

[0050] In step S110, the questionnaire data matrix is ​​as follows: n samples ×9 dimensions, column definition:

[0051] .

[0052] In the formula, nsamples is the number of samples; Number the user; The importance of charging time, on-grid SOC, off-grid SOC, and subsidy price are ranked in order, with values ​​ranging from 1 to 4, where a smaller value indicates higher importance. These are the psychological thresholds for the corresponding factors, also graded from 1 to 4.

[0053] In step S120, based on the actual questionnaire data, the frequency distribution of each factor in different ranking positions is statistically analyzed, and an importance ranking probability matrix of the four influencing factors is established to form a quantitative representation of user preferences:

[0054] .

[0055] In the formula, For the first k The influencing factor ranked in the first place i The probability of the position. k ∈{1,2,3,4} respectively correspond to the charging time T Network Access SOC S arr Off-network SOC S dep Subsidized prices p , i ∈{1,2,3,4} represents the importance ranking position, with ranking position 1 indicating the most important and ranking position 4 indicating the least important. This probability distribution reflects the differences in the importance that different user groups place on various influencing factors, providing a statistical basis for the generation of proxy individuals, while also satisfying the uniqueness constraint of the ranking. And for any user j Its sorting vector R j It is a permutation of {1,2,3,4}.

[0056] In step S130, the psychological threshold distribution is a four-level grading system, and the specific calculation formula is as follows:

[0057] .

[0058] in: Let δ be the probability that the psychological threshold of the k-th influencing factor is at level j, where j∈{1,2,3,4} is the threshold level. The higher the level, the stricter the user's requirements for that factor. Level 1 is the minimum requirement, and level 4 is the maximum requirement. The relaxation parameter δ is set based on the expected value of the binomial distribution and comes from the compromise probability of users in boundary situations in actual surveys.

[0059] By setting up a questionnaire data matrix, we fully considered users' reference standards regarding charging time, on-grid SOC, off-grid SOC, and subsidy price when charging. The joint probability distribution of importance ranking can be obtained to show each user's preference for different reference factors. The conditional probability distribution of psychological thresholds can better simulate the user's active response to power dispatch under different conditions. Furthermore, by segmenting users through psychological threshold levels, we can improve the accuracy of the quantitative model.

[0060] [Third Embodiment]

[0061] See Figure 3 In one specific embodiment, agent individuals are generated based on a multi-agent model, and the complete behavioral characteristics of the agent individuals are established, specifically including:

[0062] S210. Generate an importance ranking of the agent individuals, randomly select complete ranking combinations from the questionnaire data, and maintain the consistency of the internal logic of the ranking.

[0063] S220. Generate the psychological threshold of the agent individual, and independently sample from the marginal probability distribution of different reference factors to establish a complete psychological profile of the agent individual.

[0064] In step S210, the extracted questionnaire data is historical questionnaire data. Complete sorted combinations are randomly selected to maintain the internal logical consistency of the sorting. The extraction method is as follows:

[0065] .

[0066] in, R i For the first i An importance ranking vector for each agent. r i,k Indicates the first k The factors are ranked in order of importance to the agent, with values ​​ranging from 1 to 4 and all being different; R hist The dataset contains historical sorted data and all valid sorting combinations. This method avoids logical contradictions that may result from independently generating sorts of each factor, ensuring that the generated sorts conform to actual user behavior patterns. The sorting vectors satisfy the following constraints: r i,k ∈ {1,2,3,4} and for a fixed i , { r i,1 , r i,2 , r i,3 , r i,4The expression {1,2,3,4} reflects the relativity and exclusivity of users' perception of the importance of different factors.

[0067] In step S220, the method for establishing a complete psychological profile of an individual user is as follows:

[0068] .

[0069] in, T i,k For the first i The agent for the first k The psychological thresholds for each factor, ranging from 1 to 4, are generated through sampling from a categorical distribution (a special case of the multinomial distribution). T i as an agent i The complete psychological threshold vector; the psychological thresholds of each factor are independent of each other, reflecting the differences in users' acceptance of different factors; when the actual conditions reach or exceed the psychological threshold, users feel satisfied and are willing to participate.

[0070] By prioritizing importance, we can accurately determine the degree of attention different users pay to different factors when participating in power dispatch. By setting psychological profiles and psychological thresholds, we can obtain the differences between different factors and users' ideal state, and obtain more accurate user participation intentions.

[0071] [Fourth Embodiment]

[0072] See Figure 4 In one specific embodiment, user participation intention is determined based on scenario parameters and psychological thresholds. Relaxed discriminant conditions are used to handle boundary cases, and a reference quantitative model for user participation decision-making is established, specifically including:

[0073] S310. Determine the piecewise linear transformation function of on-grid SOC, off-grid SOC, charging time and subsidy price based on scenario parameters;

[0074] S320. Analyze the questionnaire data to obtain the conversion rules for parameters in each scenario;

[0075] S330. Based on psychological thresholds and importance ranking, user participation behavior is predicted to obtain a reference quantitative model.

[0076] In steps S310 and S320, the piecewise linear transformation function is established as follows:

[0077] .

[0078] in, L i,k For the first i The user on the firstk The actual level of each factor. X i,k For the first k The actual values ​​of each factor f k For the first k The conversion function for each factor; the specific conversion rules are determined based on user satisfaction surveys and statistical analysis, and the determination method is as follows:

[0079] Charging time conversion: ;

[0080] Network access SOC conversion: ;

[0081] Off-grid SOC conversion: ;

[0082] Subsidy price conversion: .

[0083] In step S330, based on psychological thresholds and importance ranking, user participation intention is determined, and a multi-factor comprehensive decision-making model is established to achieve accurate prediction of user participation behavior. The prediction method is as follows:

[0084] ;

[0085] .

[0086] In the formula, W i For the first i The participation intention of each user can be either 0 (unwilling to participate) or 1 (willing to participate). T i,k Psychological threshold; U i The set of unsatisfactory factors includes all factors whose actual levels are below the psychological threshold. This model is based on the "barrel theory," which states that a user's willingness to participate is determined by the least satisfied factor. When the actual levels of all factors are not lower than the corresponding psychological thresholds, users are willing to participate in scheduling.

[0087] By setting a conversion function to map continuous physical quantities into discrete satisfaction levels, it is easier to compare with psychological thresholds. The conversion boundary is determined based on quantile analysis of a large sample user survey to ensure the balance of user distribution at each level. The method for judging participation intention also reflects the conservative and risk-averse characteristics of user decision-making, thus improving the accuracy of prediction.

[0088] [Fifth Embodiment]

[0089] See Figure 5In one specific embodiment, subsidies are allocated based on individual user characteristics, establishing basic allocation rules for differentiated subsidies, specifically including:

[0090] S410. Reference factors that meet the psychological threshold are recorded as satisfactory factors, and reference factors that do not meet the psychological threshold are recorded as unsatisfactory factors.

[0091] S420. Differentiated subsidies based on the factors of user dissatisfaction and their importance ranking;

[0092] S430. Establish conversion functions between different comprehensive threshold levels and subsidy prices.

[0093] In step S420, differentiated subsidies are applied based on four factors: grid-connected SOC, off-grid SOC, charging time, and subsidy price. The threshold values ​​of these four factors are combined to obtain a comprehensive threshold level. When only the least important factor is not met and the difference is ≤2 levels, a probabilistic judgment mechanism is used. The specific judgment method is as follows:

[0094] .

[0095] In the formula, δ These are relaxation parameters; R i,k To rank by importance, when R i,k = 4 indicates the first k The least important factor is the one mentioned above. B (1, δ ) is the parameter δ Bernoulli distribution; T i,k - L i,k This refers to the gap between the psychological threshold and the actual level; the relaxation mechanism is based on the theory of "bounded rationality," which believes that users will adopt simplified heuristic strategies when faced with complex decisions and will show a certain tolerance for less important factors.

[0096] In step S430, the subsidy method generates a subsidy strategy through conditional constraints. The specific subsidy strategy is as follows:

[0097] Condition 1: If the subsidy price alone is insufficient and is the least important factor, a downgraded satisfaction strategy will be adopted.

[0098] .

[0099] Condition 2: If the subsidy price alone is not satisfied and the subsidy price is not the least important factor, a complete satisfaction strategy shall be adopted:

[0100] .

[0101] Condition 3: If neither the subsidy price nor the other least important factor is satisfied, a strategy that fully satisfies the subsidy price is adopted.

[0102] .

[0103] In the formula, s i For the first i Percentage of subsidies per user T i,4 To subsidize the psychological threshold of price, g This is a conversion function from comprehensive threshold levels to prices.

[0104] Establish a conversion function from comprehensive threshold levels to subsidy prices to map the threshold levels to specific subsidy ratios:

[0105] .

[0106] The design of relaxation conditions avoids overly strict discrimination criteria, improves the model's fit to actual user behavior, and the setting of the transformation function ensures that the subsidy level matches user expectations, thereby increasing users' enthusiasm for participating in power dispatch.

[0107] [Sixth Embodiment]

[0108] In one specific embodiment, the schedulable space of the target area is predicted based on external environmental parameters, and the initial set of users participating in scheduling within the target area is determined based on the schedulable space and basic allocation rules, specifically including:

[0109] S510. Predict the peak excess power and valley surplus power of the power grid based on external environmental parameters, and determine the dispatchable space based on the difference between the peak excess power and the valley surplus power.

[0110] S520. Obtain the set of theoretical users with scheduling intentions according to the basic allocation rules, and calculate the scheduling power of the theoretical user set participating in the scheduling.

[0111] S530. When the scheduled power is less than or equal to the schedulable space, the theoretical user set is recorded as the initial user set.

[0112] S540. When the dispatched electricity volume is greater than the dispatchable space, the theoretical user set is filtered according to the subsidy price to obtain the initial user set.

[0113] In step S510, based on today's weather, temperature, and date, the corresponding power grid data is retrieved from historical data. Corresponding weather refers to the same weather conditions, corresponding temperature refers to a temperature difference within 1°C during the time period when power dispatch is required, and corresponding date refers to dates within half a month of each other, even if they are from different years. Through this historical data, specific data on peak excess power and valley surplus power are obtained, providing data support for dispatching work. The maximum dispatchable space is the difference between peak excess power and valley surplus power.

[0114] In steps S520 to S540, based on the prediction model in steps S100 to S400, users who intend to participate in the scheduling can be identified in advance before the peak period arrives. However, in order to reduce unnecessary subsidies and avoid the occurrence of secondary peaks, not all users who intend to participate in the scheduling can participate. The scheduled power can be calculated based on the charging habits of each theoretical user and the remaining power of the electric vehicle.

[0115] When the scheduled power is less than or equal to the available scheduling space, all theoretical users can participate in this scheduling work, meaning that the set of theoretical users is the same as the initial set of users.

[0116] When the dispatched power exceeds the available dispatch space, it means that if all theoretical users participate in this dispatching work, the excess power may disappear, or a new peak in electricity consumption may be generated during subsequent charging. To avoid this situation, it is necessary to change the selection method of the theoretical user set, that is, to change one of the parameters of charging time, grid-connected SOC, off-grid SOC and subsidy price, so that some theoretical users in the original theoretical user set give up participating in this dispatching work, thereby completing the power dispatching work.

[0117] By predicting peak excess power and valley surplus power of the power grid through external environmental parameters, the power dispatching work can be more in line with the current working environment of the power grid. The setting of dispatchable space can not only reduce the execution cost of dispatching work, but also avoid the occurrence of secondary peak power consumption after the dispatching work is carried out. By filtering the theoretical user set through the dispatch space, the number of users participating in the power dispatching work can be accurately reduced, thereby improving the adaptability of the initial user set to the power dispatching work.

[0118] [Seventh Embodiment]

[0119] In one specific embodiment, the actual scheduling space of the target area is obtained, and the saturation status of the target area is determined based on the scheduling usage of the initial user set. This specifically includes:

[0120] S601. Compare the actual scheduling space with the schedulable space to obtain the variation coefficient;

[0121] S602. In the initial user set, users who have performed scheduling are recorded as completed users, and those who have not performed scheduling are recorded as remaining users. The current electricity consumption of completed users is the scheduling consumption, and the subsequent electricity consumption of remaining users is the scheduling quantity to be optimized.

[0122] S603. Obtain the scheduling difference based on the actual scheduling space and scheduling usage, and obtain the saturation threshold based on the scheduling difference and the change coefficient. When the amount of scheduling to be optimized is greater than or equal to the saturation threshold, the target area is in a saturated state. When the amount of scheduling to be optimized is less than the saturation threshold, the target area is not in a saturated state.

[0123] In step S601, when the peak electricity consumption period begins, the actual dispatch space can be obtained based on the current working status of the power grid. By comparing the actual dispatch space with the expected dispatchable space, the overall change in the power grid's electricity consumption for the day can be obtained.

[0124] In step S602, although users in the initial user set need to charge their electric vehicles during off-peak hours, some users have their own charging stations and can set the charging time through the charging stations. Users who have completed the charging settings are considered to have completed the scheduling and are recorded as completed users. Users who have not set the charging settings are considered as remaining users.

[0125] In step S603, the actual dispatch quantity corresponding to the actual dispatch space is subtracted from the dispatch usage quantity to obtain the dispatch difference. The larger the dispatch difference, the more users need to continue to perform dispatch work. At the same time, the overall power consumption trend of the power grid needs to be considered through the change coefficient. The saturation threshold refers to the maximum dispatch quantity that still needs to be dispatched to meet the power grid dispatch work. The change coefficient is calculated as follows:

[0126] V0 = 1 + (V1 - V2) ÷ V1.

[0127] Where V0 is the variation coefficient, V1 is the schedulable space, V2 is the actual schedulable space, the saturation threshold is denoted as V3, and the scheduling difference is ΔV. The saturation threshold and the scheduling difference satisfy the following relationship:

[0128] V3 = △V × V0.

[0129] When the amount of scheduling to be optimized is greater than or equal to the saturation threshold, it means that if all remaining users continue to perform scheduling, it will lead to excessive power scheduling. It is necessary to appropriately reduce the number of remaining users performing scheduling. When the amount of scheduling to be optimized is less than or equal to the saturation threshold, if all remaining users continue to perform scheduling, it will not lead to excessive scheduling.

[0130] The setting of the variation coefficient fully considers the difference between the current power grid operating status and the power grid operating status in historical data. The setting of the saturation threshold intuitively reflects the maximum dispatchable power required by the power grid under the current operating status. The determination of the saturation state provides a reference standard for the subsequent execution of dispatching work, which facilitates timely adjustments to the dispatching work.

[0131] [Eighth Embodiment]

[0132] In a specific embodiment, if so, the basic allocation rules are adjusted according to the reference quantization model, changing the number of users in the initial user set and the scheduling usage, specifically including:

[0133] S621. Analyze the satisfaction and dissatisfaction factors of the remaining users and establish a correction model for the remaining users;

[0134] S622. Adjust the comprehensive threshold levels corresponding to the satisfaction factors and the unsatisfaction factors, and reduce the number of users who continue to perform scheduling among the remaining users until the amount of scheduling to be optimized is lower than the saturation threshold.

[0135] In steps S621 and S622, the factors of satisfaction and dissatisfaction are analyzed, the change in the number of remaining users is determined for the change of each reference factor, and a modified model is obtained based on the relationship between the change in the number of different factors and the corresponding change in the number of remaining users.

[0136] Based on the saturation threshold, the adjustment range corresponding to the remaining users is obtained. The method of reducing the subsidy price is used first to reduce the number of remaining users. When reducing the subsidy price cannot achieve satisfactory results, the method of reducing the comprehensive threshold level is then considered in combination with multiple reference factors.

[0137] By lowering the comprehensive threshold level, the optimal power dispatching method can be selected in real time based on the power grid's operating status. Compared to directly executing dispatching work based on the initial user set, this approach minimizes the occurrence of ineffective dispatching and reduces the expenditure of ineffective subsidies.

[0138] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A user-driven electric vehicle charging optimization method, characterized in that, The charging optimization method includes: A multi-agent model is established based on questionnaire data to analyze user behavior within the target area. Based on the multi-agent model, agent individuals are generated, and the complete behavioral characteristics of the agent individuals are established; Based on scenario parameters and psychological thresholds, user participation intention is judged, and relaxed discrimination conditions are used to handle boundary cases, establishing a reference quantitative model for user participation decision-making; Subsidies are allocated based on individual user characteristics, and basic allocation rules for differentiated subsidies are established. The schedulable space of the target area is predicted based on external environmental parameters, and the initial set of users participating in scheduling within the target area is determined based on the schedulable space and the basic allocation rules. Obtain the actual scheduling space of the target area, and determine whether the target area is saturated based on the scheduling usage of the initial user set; If not, then a user-oriented optimization scheduling system is established based on the initial user set; If so, the basic allocation rules are adjusted according to the reference quantization model to change the number of users in the initial user set in order to change the scheduling usage. The provision of subsidy allocation based on individual user characteristics, establishing basic allocation rules for differentiated subsidies, specifically includes: Reference factors that meet the psychological threshold are recorded as satisfactory factors, and reference factors that do not meet the psychological threshold are recorded as unsatisfactory factors. The differentiated subsidy is based on the user's dissatisfaction factors and their importance ranking; Establish conversion functions between different comprehensive threshold levels and subsidy prices; Differentiated subsidies are implemented by referencing four factors: on-grid SOC, off-grid SOC, charging time, and the aforementioned subsidy price. The threshold values ​​of these four factors are combined to obtain a comprehensive threshold tier. When only the least important factor is not met and the difference is ≤2 tiers, a probabilistic judgment mechanism is used. The specific judgment method is as follows: ； In the formula, δ These are relaxation parameters; R i,k To rank by importance, when R i,k = 4 indicates the first k The least important factor is the one mentioned above. B (1, δ ) is the parameter δ Bernoulli distribution; T i,k - L i,k The difference between the psychological threshold and the actual gear level. W i For the first i The willingness of individual users to participate U i This is the set of factors that do not meet the requirements. The subsidy method generates a subsidy strategy through conditional restrictions, and the subsidy strategy is as follows: If only the subsidy price is not met, and the subsidy price is the least important factor, a downgraded satisfaction strategy will be adopted: ； If the subsidy price is not satisfied and the subsidy price is not the least important factor, a complete satisfaction strategy is adopted: ； When neither the aforementioned subsidy price nor the other least important factor is satisfied, the strategy of fully satisfying the subsidy price is adopted: ； s i For the first i Percentage of subsidies per user T i,4 To subsidize the psychological threshold of price, g This is a conversion function from comprehensive threshold levels to prices; Establish a conversion function from comprehensive threshold levels to subsidy prices to map the threshold levels to specific subsidy ratios: 。 2. The charging optimization method according to claim 1, characterized in that, The step of generating agent individuals based on the multi-agent model and establishing complete behavioral characteristics of the agent individuals specifically includes: Generate an importance ranking of the agent individuals, randomly extract complete ranking combinations from the questionnaire data, and maintain the internal logic consistency of the ranking. The psychological threshold of the agent individual is generated, and independent sampling is performed on the marginal probability distribution of different reference factors to establish a complete psychological profile of the agent individual.

3. The charging optimization method according to claim 2, characterized in that, The method for judging user participation intention based on scene parameters and psychological thresholds, employing relaxed discriminant conditions to handle boundary cases, and establishing a reference quantitative model for user participation decision-making specifically includes: Based on the scenario parameters, a piecewise linear transformation function is determined for the on-grid SOC, off-grid SOC, charging time, and subsidy price. The questionnaire data is analyzed to obtain the conversion rules for each scenario parameter; Based on the psychological threshold and the importance ranking, user participation behavior is predicted to obtain the reference quantification model.

4. The charging optimization method according to claim 3, characterized in that, The step of predicting the schedulable space of the target area based on external environmental parameters, and determining the initial set of users participating in scheduling within the target area based on the schedulable space and the basic allocation rules, specifically includes: The peak excess power and valley surplus power of the power grid are predicted based on the external environmental parameters, and the dispatchable space is determined based on the difference between the peak excess power and the valley surplus power. Based on the basic allocation rules, a set of theoretical users with scheduling intentions is obtained, and the scheduling power of the theoretical user set participating in the scheduling is calculated. When the scheduled power is less than or equal to the schedulable space, the theoretical user set is recorded as the initial user set; When the scheduled power is greater than the available space, the theoretical user set is filtered according to the subsidy price to obtain the initial user set.

5. The charging optimization method according to claim 4, characterized in that, The step of obtaining the actual scheduling space of the target region and determining whether the target region is saturated based on the scheduling usage of the initial user set specifically includes: The actual scheduling space is compared with the schedulable space to obtain the variation coefficient; Users who have been scheduled in the initial user set are recorded as completed users, and those who have not been scheduled are recorded as remaining users. The current electricity consumption of the completed users is the scheduled consumption, and the subsequent electricity consumption of the remaining users is the scheduling amount to be optimized. The scheduling difference is obtained based on the actual scheduling space and the scheduling usage, and the saturation threshold is obtained based on the scheduling difference and the change coefficient. When the amount of scheduling to be optimized is greater than or equal to the saturation threshold, the target region is in the saturation state; When the amount of scheduling to be optimized is less than the saturation threshold, the target region is not in the saturation state.

6. The charging optimization method according to claim 5, characterized in that, If so, the basic allocation rules are adjusted according to the reference quantization model to change the number of users in the initial user set and thus change the scheduling usage, specifically including: Analyze the satisfaction factors and dissatisfaction factors corresponding to the remaining users, and establish a correction model for the remaining users; Adjust the comprehensive threshold levels corresponding to the satisfaction factors and the dissatisfaction factors to reduce the number of users among the remaining users who continue to perform scheduling until the amount of scheduling to be optimized is lower than the saturation threshold.

Images