A city-level electric vehicle charging / exchanging feedforward collaborative scheduling method fusing threshold state of charge prediction and user decision update

Abstract

A city-level electric vehicle charging / exchanging feedforward collaborative scheduling method fusing threshold state of charge prediction and user decision updating, comprising: acquiring multi-source data of electric vehicles in a city area; identifying vehicles that may trigger energy supplement behavior in a rolling window to form expected arrival rate and arrival time distribution of each site in each period; calculating the comprehensive value of the candidate site; when detecting that the road traffic flow rate is lower than the road level threshold, re-estimating the value function in real time and judging whether to replace the original planned path; superimposing the obtained vehicle selection probability and arrival rate to form short-term load prediction of each site; jointly optimizing the guide parameters in the rolling window; collecting user actual arrival rate, guide acceptance rate and queuing time data to periodically update key parameters in the feedforward model. The invention simultaneously describes the coupling mechanism of multi-vehicle type threshold triggering, decision updating in driving and queuing backflow, effectively improving the accuracy of city-level charging / exchanging load prediction and the coordination of site resource allocation.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging / swapping scheduling technology, specifically to a city-level electric vehicle charging / swapping feedforward collaborative scheduling method that integrates threshold state of charge prediction and user decision updates. Background Technology

[0002] With the rapid popularization of new energy vehicles in cities, the demand for electric vehicle charging and swapping exhibits obvious tidal and regional characteristics. Different users (such as private cars, taxis, and buses) show strong imbalances in their energy replenishment needs in both time and space due to differences in travel patterns and operational behaviors. Without an effective scheduling and guidance mechanism, this can lead to localized battery shortages, surges in queuing times, load imbalances between charging stations, and even affect the safe and stable operation of the power distribution network. Therefore, establishing an efficient city-level electric vehicle charging and swapping scheduling mechanism is crucial to ensuring user experience and energy system security.

[0003] Existing electric vehicle (EV) dispatching schemes largely rely on historical load statistics and simplified behavioral assumptions, neglecting the diversity and dynamism of EV user charging behavior. Some methods rely solely on static predictions based on average SOC thresholds, lacking segmentation for different vehicle models and user groups; others use shortest path or lowest price as the site selection criterion, failing to fully consider the re-decision-making process of EV users after perceiving changes during driving. Furthermore, queuing feedback, dynamic inventory allocation, and traffic diversion mechanisms are often handled in isolation, making it difficult to achieve coordinated optimization of charging and swapping resources at the city level. Therefore, a holistic dispatching method integrating user behavior modeling, feedforward prediction, and real-time feedback control is urgently needed to improve the operational efficiency and responsiveness of city-level EV charging systems. Summary of the Invention

[0004] To address the problems of low demand forecasting accuracy, simplistic user behavior assumptions, uneven site resource allocation, and delayed queuing feedback in existing charging / swapping scheduling technologies, this invention provides a city-level electric vehicle charging / swapping feedforward collaborative scheduling method that integrates threshold state of charge prediction and user decision updates. This method enables coordinated and optimized energy replenishment scheduling for multiple types of electric vehicles across various time periods and scenarios. Compared to existing schemes based solely on static thresholds or one-time site selection, this invention simultaneously characterizes the coupling mechanism of multi-vehicle threshold triggering, in-driving decision updates, and queuing backfeedback, effectively improving the accuracy of city-level charging / swapping load forecasting and the coordination of site resource allocation.

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

[0006] A city-level electric vehicle charging / swapping feedforward collaborative scheduling method that integrates threshold state of charge prediction and user decision updates includes the following steps: Step 1: Data Acquisition: Real-time acquisition of multi-source data on the state of charge (SOC) of electric vehicles in the urban area, road traffic flow, battery inventory and queuing time at stations, and time-of-use electricity prices; Step 2: Prior demand modeling: Based on vehicle type classification, set the state of charge (SOC) threshold, identify vehicles that may trigger recharging behavior in the rolling window, and form the expected arrival rate and arrival time distribution of each station in each time period. Step 3: Behavioral Intention Inference: Construct a site selection probability model and calculate the comprehensive value of candidate sites based on current traffic conditions, time-of-use electricity prices, and queuing time; Step 4: Dynamic adjustment during the trip: When the road traffic flow rate is detected to be lower than the road grade threshold, or when the station price / waiting conditions change significantly, the value function is re-evaluated in real time and it is determined whether to replace the original planned route; Step 5: Queue Reinjection and Feedback: The vehicle selection probabilities and arrival rates obtained in Steps 2 to 4 are superimposed to form short-term load forecasts for each station. The queuing time is estimated based on the approximate queuing time estimation method (segmented waiting time estimation and minimum remaining service time correction), and then reinjected into the energy replenishment cost model to adjust the station selection logic. Step 6: Joint optimization and feedforward pre-deployment unit: Jointly optimize the site battery inventory allocation, battery transfer route, in-site charging power scheduling, and navigation service fee guidance parameters within the rolling window; Rigid user guarantee mechanism: This mechanism sets hard constraints on capacity reservation or queuing priority for users with rigid energy replenishment needs, such as public transportation, to ensure their energy replenishment services are not interfered with by users with flexible guidance. This module relies on pre-set operational strategies and business rules on the dispatching side. Step 7: Feedback closed-loop correction: Collect data such as actual user arrival rate, guidance acceptance rate and queuing time, and periodically update the key parameters in the feedforward model.

[0007] In step 1, multi-source data on the state of charge (SOC) of electric vehicles in the urban area, road traffic flow speed, battery inventory and queuing time at stations, and time-of-use electricity prices are obtained. 1.1: For the i-th electric vehicle, the remaining battery charge at time t is determined by the remaining charge at the previous time t-1 and the charge consumed during the time period t due to vehicle operation. The iterative model for its remaining battery charge can be expressed as: ; In the formula, and Let represent the remaining battery power of the i-th electric vehicle at times t and t-1, respectively; This represents the proportion of the distance traveled within time period t to the vehicle's daily mileage. This represents the daily mileage of the i-th electric vehicle; This indicates the energy consumption of an electric vehicle per 100 kilometers.

[0008] 1.2: Traffic flow in the actual road network at different times affects the driving speed of electric vehicles, thereby changing their energy consumption and travel time. Therefore, this invention introduces a speed-flow model to construct a dynamic traffic network model to simulate the actual operating state of the traffic network at different times. Its expression is: ; ; In the above formula: This represents the speed of travel on road ij within time period t; This indicates the zero-flow travel speed of road ij; This represents the traffic flow on road ij within time period t; This indicates the traffic capacity of road ij. An adaptive exponent for the speed-flow relationship; , and These are the model parameters.

[0009] 1.3: Battery inventory data at the site (taking a battery swapping station as an example) is categorized into three types based on their working status: batteries awaiting charging, batteries being charged, and fully charged batteries. (The data is then distributed according to time period.) They are respectively denoted as , , And meet the constraints of total battery count and charging compartment capacity: , ; in: This represents the total number of batteries at the battery swapping station. Here, $t$ represents the number of charging bays within the station, and $t$ is the index for a discrete time period. To depict the evolution of inventory as battery swapping demand changes, battery state transitions are updated by time period: ; ; ; in: express The number of batteries swapped during the time period (i.e., the battery consumption at the site). express The number of batteries that transitioned from "awaiting charging" to "charging" during the specified time period. If a situation arises where "fully charged batteries are insufficient to meet battery swapping needs," then batteries that can meet the demand will be selected from the rechargeable batteries. Battery replenishment and swapping services; among which For the SOC of the selected rechargeable battery, This represents the lower limit of the SOC (State of Charge) of the battery for user-acceptable swapping.

[0010] 1.4: Station queuing time data is represented by "arrival waiting time", denoted as... Due to queuing time With energy replenishment service time composition: The site during the time period The queuing time is approximated by parallel service: when the number of vehicles in the station hour: ; when hour: ; in: For the site Number of vehicles at the station (including those in queue and those currently being served). This refers to the number of energy replenishment facilities (number of charging piles or number of battery swapping stations). For average refueling time, To round down, This represents the minimum remaining charging time for vehicles currently receiving service at the station. Charging service time is determined by station type: charging station A constant value can be taken for the battery swapping station. ;in, For the arriving SOC, To replenish energy and end SOC, For the battery's rated capacity, For charging efficiency, This refers to the power output of the charging station.

[0011] 1.5: Time-of-use electricity price data includes grid-side time-of-use electricity prices. And the time-sharing service fee at the site (charging service fee) Battery swapping service fee This is used to calculate the charging costs for different time periods and participate in subsequent value assessments. The charging cost is: ; The battery swapping price is: ; This can be used to determine the cost of battery swapping: ; in: The difference in battery capacity before and after the battery swap. This is a compensation item for non-fully charged battery swapping. Time-of-use pricing and charging / swapping prices can be plotted as curves based on time periods and mapped to the corresponding times of vehicle arrival. The service fee parameter serves as the input for the energy replenishment cost.

[0012] In step 2, the state of charge (SOC) threshold and triggering mechanism are set for different vehicle models respectively: 2.1: Private cars generate daily mileage based on a log-normal distribution. The trigger point is determined by iterating the State of Occurrence (SOC) hourly using the Monte Carlo method. Specifically, this includes: According to NHTS (National Household Mobility Survey) data, the daily mileage of electric private vehicles... It follows a log-normal distribution, and its probability density function is: ; In the formula: The daily mileage is represented as a value. The probability density function value at time; The value representing the daily mileage; Used to characterize the degree of dispersion of daily mileage in the logarithmic field; Used to characterize the concentration level of daily mileage in the logarithmic field; Represented by natural constant The operation of exponential functions with base 0.05, the parentheses containing "..." "" indicates the expression of the independent variable of the exponential function (i.e., the overall term that needs to be substituted into the exponential operation). 2.2: The taxi area is divided into grids, and the battery swapping time is estimated based on OD (Original Destination) trips and route steps, as detailed below: The planning area is divided into a square grid with side length h, centered at coordinates (x, y). Assume the distance from the passenger's travel demand point to their destination is... Then the number of grids that need to be traversed is .

[0013] The power consumption for a single battery swap is: ; In the formula: The amount of electricity consumed by electric taxis; Number of grid cells; The side length of a single cell; 2.3: The first battery swap time for buses is set daily based on departure intervals, route length, and fixed energy consumption budget, as detailed below: The routes and stopping times of electric buses follow a strong regularity every day. (The sentence is incomplete and ends abruptly.) Taking a bus route as an example: the straight-line distance between the bus's starting station and its terminal station is... The start and end times are respectively and The departure interval is The bus travels at a speed of The time required for a bus to make one round trip between the starting station and the terminal station is: ; In the formula: This indicates the time required for a bus to make one round trip between its starting and ending points. Indicates rounding down; This indicates the number of departure intervals covered in one round trip; The number of buses required for this route is: ; In the formula: Indicates the first The number of departures on a single route within a single operating day; 'b' indicates rounding down, and 'b' indicates bus scheduling.

[0014] The formula for the remaining battery power is: ; In the formula: Indicates vehicle After driving to the Remaining battery power at each road segment Indicates the first Remaining power in each section Used to depict the power consumption of a vehicle during driving.

[0015] In step 2, the typical energy replenishment behavior is triggered by the following conditions: ; In the formula: Let be the remaining battery power of vehicle i at time t; The rated capacity of the vehicle's battery; This is the threshold coefficient.

[0016] In step 2, the expected arrival rates and arrival time distributions for each station at each time period are generated, as detailed below: Within the scrolling window, time is discretized into equal-length time intervals. , This indicates the total number of time periods the scrolling window is divided into, with each period having a length of [length value missing]. (e.g., 1 hour). The core of step 2 is: first determine the time period when the vehicle "triggers recharging", then calculate its "arrival time at the station", and finally aggregate statistics by "station-time period" to obtain the expected arrival rate and arrival time distribution.

[0017] When a vehicle meets the conditions for triggering a charging service, it is considered to have triggered a charging service. To determine the trigger point, the battery level can be estimated by using daily mileage and the proportion of trips during different time periods within the day. ; ; In the formula: This refers to the vehicle's daily mileage. For the first The percentage of mileage traveled during different time periods meets the requirements. ; For time period Mileage; Indicates vehicle During the period Remaining battery power at the end Indicates the remaining battery power in the previous time period; This gives us the initial trigger point: ; In the formula: Indicates vehicle The first trigger point for energy replenishment; Indicates vehicle At any moment The remaining battery power; Represents the threshold coefficient ; indicates the threshold percentage for triggering energy replenishment; The larger the value, the earlier the energy replenishment is triggered. Indicates the baseline value of the electricity. This indicates taking the minimum time that satisfies the condition; After the trigger, the vehicle calculates its arrival time for the set of reachable candidate stations, without determining the final target station in step 2; the final target station is determined by the probability selection model in step 3. The arrival time is calculated for each of the candidate reachable stations in the prior demand pool. ; In the formula: Indicates vehicle The "nearest site" at the moment of triggering power replenishment; Indicates a collection of sites Find the station that minimizes the distance to the next station. This represents the set of all available charging / swapping stations within the city. Indicates vehicle At the moment of triggering energy replenishment The current position (coordinates / location point). Indicates the vehicle's current location to the station. The distance function (which can be the shortest path distance or equivalent travel distance of the road network); The speed of a road segment can be expressed as: ; In the formula: Indicates the speed of the road segment; Indicates the free-flow velocity; Indicates traffic flow on a road segment; Indicates traffic capacity; These are model parameters; And the travel time for each section: ; In the formula: This refers to the length of the road segment.

[0018] Route travel time: ; In the formula: Indicates the travel time of the route; Indicates road segment At any moment Passage time; Indicates at time A defined driving route; Therefore, the arrival time is: ; In the formula: Indicates the arrival time; Finally, the arrival volume is aggregated and counted by "site - time period": ; In the formula: Site In the Number of vehicles arriving in each time period } is an indicator function; Indicates the first Each time period; Indicates the length of each time period; Then the site During the period The expected arrival rate is: ; In the formula: This indicates the site During the period Expected arrival rate; The desired result can be approximated by averaging multiple simulations; Site The arrival time distribution is as follows: ; In the formula: Indicates site During the period The probability of arrival; This indicates the total number of time periods divided within the scrolling window.

[0019] The expected arrivals at the station in each time period are normalized to obtain the probability distribution of arrival times in each time period.

[0020] In step 3, a comprehensive value function is constructed that includes driving costs and refueling costs: ; In the formula: A comprehensive value function that includes both driving costs and refueling costs; For normalized weights, and satisfying ; The cost of operation includes the distance traveled, the time spent traveling, and the energy consumed. The cost of replenishing energy includes time-of-use pricing and queuing time; It is a value mapping function, which can be defined according to the piecewise value function of prospect theory.

[0021] If foreground theory mapping is not used, then let , This represents the input cost value (which could be the normalized driving cost, the normalized refueling cost, or a component thereof). The model degenerates into a linear weighted average. ; Vehicle stop selection uses a bounded rationality model (such as prospect theory or Logit) to generate a probability distribution. The input to the Logit model is the aforementioned comprehensive value. ; The value mapping of prospect theory can be used for calculation; otherwise, it can be calculated according to the linear weighted degradation case.

[0022] For the Vehicles that trigger a refueling demand, at any time Obtain its set of reachable candidate sites and for each candidate site Calculate the comprehensive value Such as the comprehensive value function consisting of driving costs and refueling costs. .

[0023] Then the vehicle At any moment Select site The probability can be expressed as ,in: Indicates vehicle At any moment Select candidate sites The probability of each station is equal to the sum of the probabilities of all stations being 1.

[0024] Indicates vehicle At any moment The set of reachable candidate sites (e.g., selected under SOC and reachability constraints).

[0025] Indicates vehicle At any moment Select site The comprehensive value (which can be understood as "comprehensive cost value", which is obtained by weighting the driving cost value and the refueling cost value).

[0026] Represents the coefficient of rationality intensity. The larger the value, the more sensitive it is to value differences, and the more likely it is to make "optimal / certain" choices; The smaller the value, the more dispersed / random the selection.

[0027] This is how it was obtained. It is the "probability distribution of site selection", which can be directly used for subsequent arrival rate superposition and load forecasting.

[0028] In step 3, the overall value of the candidate sites is calculated: For the vehicle at time For vehicles that trigger refueling needs, the available candidate stations are... any of the stations Overall value (cost). The calculation can be divided into two parts: "driving cost" and "refueling cost". Firstly, under a dynamic traffic network, the routes leading to stations... The candidate paths are then normalized and weighted according to distance, energy consumption, and time to obtain the path travel cost: x .; In the formula: Indicates the cost of traveling the route. This represents the number of road segments in the path. The first Road segment length, driving energy consumption, and driving time; These represent the minimum and maximum values ​​of the corresponding indicators for each road segment across the entire road network. The minimum / maximum values ​​for energy consumption indicators; The minimum / maximum value of the time indicator; The weights are distance, energy consumption, and time, respectively, and satisfy the following conditions: .

[0029] by As edge weights in shortest path algorithms (such as Dijkstra's algorithm), the distance from the vehicle's current location to the station is obtained. The shortest path, whose segment weights are summed, is the travel cost. : ; On the energy replenishment side, first calculate the station At any moment The cost of recharging and the waiting time are calculated, and then normalized and summed to obtain the comprehensive cost of recharging. ; If charging is required, the charging cost is as follows: ; In the formula: This indicates the charging cost; This is expressed as battery capacity (or rated capacity). Indicates time-of-use electricity pricing; This indicates the charging service fee; This indicates that charging is complete and the system is fully charged (SOC). Indicates the state of charge (SOC) before energy replenishment; If it is a battery swap, the battery consumption will be as follows: ; In the formula: Indicates the amount of electricity exchanged; Indicates the battery's SOC after battery swapping; The battery swapping costs are as follows: ; In the formula: This indicates the cost of battery swapping; , For battery swapping service fees; For corrections such as discounts / offers; The waiting time is as follows: ; In the formula: This indicates the total waiting time of the vehicle within the station; Indicates the waiting time in the queue; Indicates the duration of the power replenishment service; Power replenishment service hours For charging stations:

[0030] in, This indicates that charging is complete and the system is at state of charge (SOC). Indicates the State of Charge (SOC) upon arrival at the station; Indicates the battery's rated capacity; For charging efficiency; The power of the charging pile; The battery swapping station can be approximated by a constant duration, i.e. ;in: For battery swapping service time, This indicates that the battery swapping operation takes approximately 5 minutes.

[0031] When the number of vehicles in the station Less than the number of facilities If the queue time is 0, then proceed as follows: ; In the formula: This represents the average charging time. To round down; The shortest remaining refueling time for vehicles currently in service at the station; Indicates site Current number of vehicles at the station; Energy replenishment costs (Pick or ) and waiting time After normalization and summing, the comprehensive power replenishment cost is obtained: ; In the formula: This indicates the overall cost of electricity replenishment; These are the minimum and maximum charging costs for all stations, respectively. These are the minimum and maximum waiting times for all stations, respectively.

[0032] The comprehensive value function is as described in the aforementioned formula. Calculate, where, These represent the value mappings of driving costs and comprehensive recharging costs, expressed as piecewise value functions according to prospect theory: ; In the formula: This serves as a reference point for the highest acceptable driving cost for users. This serves as a reference point for the highest overall cost of power replenishment that is acceptable to users. , These are the risk preference coefficients under the scenarios of gain and loss, respectively. This represents the loss aversion coefficient.

[0033] In step 4: when the road traffic flow speed is detected to be lower than the road grade threshold, the road grade speed threshold is set according to the road type: including expressways (not less than 50km / h), arterial roads (not less than 30km / h), secondary arterial roads (not less than 25km / h), and local roads (not less than 20km / h); the value function is re-evaluated in real time and it is determined whether to replace the original planned route; as detailed below: Let the vehicle's current planned route be Composed of several road sections The system consists of components that obtain the real-time average traffic flow speed of each road segment along the path at a fixed refresh cycle. If it exists Make ,in Given the speed threshold for each road class, If the type is a road segment, then "real-time reassessment" is triggered. Dynamic road network weights are updated based on the latest speed, and the road segment travel time is taken as... , The length of the road segment; And update the time item of the candidate path accordingly. Energy consumption items Among them, the time item From each section Accumulated; Energy consumption item Desirable Alternatively, a speed-related energy consumption model may be used; Substitute this back into the route travel cost calculation from step 3: ; In the formula: ; These are respectively: road distance, energy consumption, and time; Find the shortest path from the current location to the candidate site. Updated driving costs: ; In the formula: Indicates the cost of updating driving; At the same time, the arrival time is obtained according to the new route: ; in, Indicates the departure time. Indicates vehicle Arrive at the station The planned route (consisting of several road segments) (Composition) and re-evaluate the replenishment side at that arrival time: Energy replenishment costs are recorded as Billing is based on the time-of-use electricity price and service fee corresponding to the arrival time. Specifically: first, the arrival time is calculated... Mapped to its corresponding time slot Calculate the corresponding electricity price and service fee. , , If the vehicle On the site If you choose to charge, then: ; If battery swapping is selected, the battery swapping unit price should be defined first. Battery swapping has poor power output. ,but: ; in: For the battery's rated capacity, For the arriving SOC, To complete the SOC charging process, The battery's state of charge (SOC) after battery swapping. This is a compensation item for batteries that are not fully charged (used to correct situations where the SOC of the swapped battery is not at full charge).

[0034] Waiting time is recorded as ,in , These are queue time and service time, respectively; queue time can be calculated by station. During the period Number of vehicles at the station With parallel service capabilities estimate: when hour ,when hour

[0035] Service Hours Values ​​are taken according to the energy replenishment method during charging. A constant can be used during battery swapping. ;in For average service duration, This represents the shortest remaining service time for vehicles currently in service at the station. For charging efficiency, This refers to the power output of the charging station.

[0036] The overall energy replenishment cost is still calculated using the normalized summation method: ; In the formula: vehicle At any moment Evaluation site The overall cost of energy replenishment; This represents a normalization function that transforms indices with different dimensions to the same dimensionless interval. Indicates energy replenishment costs Indicates the waiting time after arrival at the station; in, , These represent taking the minimum and maximum values ​​among the candidate sites, respectively.

[0037] Finally, use the updated version and Recalculate the value function and synthesize the composite value: ; In the formula: Indicates vehicle At any moment Candidate sites Comprehensive value This represents the value mapping result corresponding to the travel cost, used to characterize the journey of a vehicle from its current location to a station. The impact of travel costs, such as distance, time, and energy consumption, incurred during the process; This represents the value mapping result corresponding to the energy replenishment cost, used to characterize the vehicle's performance at the station. The costs and waiting time incurred during the refueling process, and other refueling costs. and These are the weighting coefficients for driving cost value and refueling cost value, used to characterize the relative importance users place on driving cost and refueling cost, and typically satisfy the following conditions: . Still following step 3, the foreground piecewise function is derived from the reference point. Calculate and select the new optimal solution: ; In the formula: Indicates vehicle At any moment The target site has been redefined; Indicates the relationship with the site The corresponding optimal driving route; This indicates the decision variable selected from the candidate set that maximizes the comprehensive value function; Indicates vehicle At any moment The set of reachable candidate stations (obtained by SOC, road network reachability and time constraints); Indicates vehicle At any moment Candidate sites The comprehensive value evaluation result is given. This formula indicates that at a given time, the vehicle selects the station with the highest comprehensive value and its corresponding route as the new travel decision scheme. The result is then compared with the original scheme: like ; If necessary, replace the original planned path and simultaneously replace the target site; otherwise, maintain the original plan. Where: Indicates at time The overall value gain of adopting the new solution compared to the original solution; This represents the comprehensive value of the new station / route plan recalculated based on current road conditions, electricity prices, and queuing status. This indicates the overall value corresponding to the originally planned sites and routes; The minimum value improvement threshold is used to characterize the lowest level of benefit that users perceive and are willing to accept in accepting a change in their decision. If the new plan has significant advantages in overall value, it will trigger station and route updates; otherwise, the original travel and refueling plans will remain unchanged.

[0038] In step 5, the vehicle selection probabilities and arrival rates obtained in steps 2 to 4 are summed, as follows: Step 2 provides the arrival time distribution of vehicles at candidate stations, and the probability of vehicle (i) arriving at station (s) is denoted as... , If vehicle Arrive at the station Arrival time To determine the value, its probability of arrival during a specific time period can be reduced to an indicative form, at which point the comprehensive value function... According to the corresponding arrival time The site status is calculated.

[0039] }; Steps 3 and 4 provide the vehicle's latest decision-making moment. (Probability of selecting site(s) at the trigger time or re-evaluation update time) Then station(s) in time period The expected arrival quantity and expected arrival rate are respectively: , ; In the formula: Indicates site During the period The expected number of arriving vehicles is calculated as follows: for all vehicles Summation, at the decision moment of the vehicle Select site The probability, Indicates site During the period Expected arrival rate; This represents the duration of the time period. It reflects the "expected number of arrivals after the arrival flow is diverted according to the station selection probability", where s is the short-term arrival rate of "station-time period".

[0040] If we overlay by vehicle group / source unit, let source (g) be in the time period The triggering strength is Its average station selection probability is ,but ; In the formula: Average station selection probability; When generating short-term load forecasts for each site, the "expected arrival amount" is further mapped to "energy demand / power demand". Taking charging as an example, the single charging amount for vehicle (i) can be written as: ; In the formula: The amount of energy that the vehicle (i) can charge on a single charge; Indicates battery capacity (rated capacity); For the arriving SOC, The charging process is complete at SOC; the corresponding charging time is as follows: ; In the formula: Indicates charging time For charging efficiency; This refers to the charging power. Then station(s) in time period The expected energy requirement is: ; In the formula: This indicates the expected power demand for recharging at the site during different time periods; Indicates the vehicle at the moment of decision-making Select site The probability of; Indicates that the vehicle has arrived at the station. and fall into the time period The probability of; This indicates the amount of electricity that needs to be replenished. Average power prediction can be obtained using "energy / duration length": ; In the formula: Indicates site During the period The expected power of the supplementary energy; Indicates site During the period The expected replenishment power demand within the area; Indicates the duration of a single time period.

[0041] In step 5, the queuing time at the site is estimated using an approximate queuing time estimation method based on parallel service workstations (segmented waiting time estimation and minimum remaining service time correction). To obtain the station queue time, first in the time period The system internally determines whether the vehicle has a need for additional power, for example: when the vehicle's remaining battery power is sufficient. This triggers the battery swapping demand; and iterates based on the remaining battery capacity: ; In the formula: and Representing vehicles During the period and The remaining battery power; Indicates road During the period The traffic condition influence coefficient is used to characterize the effect of changes in traffic flow speed on energy consumption per unit distance (such as increased energy consumption under congested conditions). Indicates vehicle On the road The driving distance on; By continuously analyzing the demand sample size for each time period, the predicted arrival scale of the station during that time period can be formed, which can be further converted into the number of vehicles within the station. Based on this, the waiting time is estimated by approximating the queuing of "parallel service workstations" within the station: The waiting time after arriving at the station is: ; In the formula: This indicates the total waiting time of the vehicle within the station; Indicates the waiting time in the queue; Indicates the duration of the power replenishment service; Queuing time is segmented: ; In the formula: Indicates the number of vehicles in the station; Indicates the number of facilities; as well as: ; In the formula: Indicates the average refueling time; To round down; For the site The number of vehicles inside; The number of warehouses that can be served in parallel at a site; The "shortest remaining refueling time" for vehicles currently receiving service at the station; The energy replenishment service hours are as follows: (Charging station) or (Battery swapping station); In the formula: Indicates the duration of the energy replenishment service; For arrival SOC; To replenish energy and complete SOC; This refers to the battery's rated capacity. For charging efficiency; The power of the charging pile; This indicates the battery swapping operation time (in hours).

[0042] When the above-mentioned queuing estimate is "reinjected" into the energy replenishment cost and participates in the comprehensive value update, the waiting time and cost will be included in the comprehensive energy replenishment cost: ; in: As input to the subsequent value function, the comprehensive supplementary power cost is used. For energy replenishment costs; The waiting time after the vehicle arrives at the station; These are the normalized upper and lower bounds for the energy replenishment costs of candidate sites; These are the normalized upper and lower bounds for the waiting time of candidate sites, respectively.

[0043] In the cost-value mapping of energy replenishment, let And the value mapping corresponding to the driving cost. Together they constitute a comprehensive value function: in, The output of the energy replenishment cost function (comprehensive energy replenishment cost). The value mapping function for energy replenishment costs, For driving costs, The value mapping function for driving costs, and These are the weighting coefficients for the two types of costs.

[0044] In step 6, the objective function is: ; in: This indicates the weight of the objective of "reducing peak user queuing times and improving service levels" in multi-objective optimization, reflecting the system's emphasis on user queuing experience (waiting time distribution).

[0045] This indicates the weight of the objective of "suppressing peak power in the charging and swapping network and alleviating pressure on the distribution network" in multi-objective optimization, and is used to reflect the system's emphasis on grid security and peak load constraints.

[0046] This indicates the weight of the objective of "controlling operation and maintenance and scheduling costs" in multi-objective optimization, and is used to reflect the system's cost control preferences in battery allocation, inventory turnover and related operation and maintenance behaviors.

[0047] Indicates queuing time The quantile (median) is a typical value used to characterize the level of queuing time at each station within a rolling time window, thus reflecting the overall severity of queuing peaks.

[0048] It represents the maximum concurrent power value of the charging and swapping network (or distribution network node aggregation) within the rolling forecast window, that is, the maximum value of the sum of the power demand of the sites in each time period, which is used to measure the peak load level of the power grid.

[0049] This represents the operation and maintenance / logistics cost metric generated to achieve demand guidance and peak shaving, and is used to depict the comprehensive costs brought about by battery allocation, inventory adjustment and related scheduling operations.

[0050] Under the condition that the constraints are satisfied, reducing queuing peaks and suppressing network peak power are achieved by coordinating the arrival-service-selection process, as follows: On the one hand, reducing peak queuing times refers to decreasing the queuing time within the station during peak hours without violating the station's parallel service capacity constraints. The station's waiting time is defined as: ; in: This indicates the total waiting time of the vehicle within the station; Indicates the waiting time in the queue; Indicates the duration of the power replenishment service; Queuing time is segmented: ; In the formula: Indicates the number of vehicles in the station; Indicates the number of facilities; when hour: ; In the formula: Indicates the waiting time in the queue; The number of workstations that can be served in parallel at a site. For average refueling time, The shortest remaining service time for vehicles currently in service.

[0051] The site reach rate obtained in step 5 Influence The scale, and And the probability of vehicle selection The result is obtained by superposition.

[0052] When optimization decisions (such as diverting some demand to less congested stations or during off-peak hours) make During peak hours when prices drop and its statistical quantile This reduces queuing peaks while constraints are met. This change is achieved through a comprehensive reduction in power supply costs. Feedback to value function ; Further reduce the probability of high-queue sites being selected, forming a closed-loop regulation.

[0053] On the other hand, suppressing network peak power refers to reducing the peak concurrent power of the entire network (or distribution node aggregation) while meeting the constraints of maximum charging power per station and number of devices. The duration and power of single-vehicle charging service must meet the following requirements: ; in: Indicates the duration of the power replenishment service; For charging power, For battery capacity, and These represent the arrival and departure SOCs, respectively. (Site) During the period Concurrent power can be determined by the number of vehicles in service and The summation is then used to define the network peak power: ; in: This represents the maximum instantaneous total power of the charging and swapping network within the entire time window (i.e., the network peak power). Indicates site During the period The supplementary power load; Indicates the time period The total power of the entire network is obtained by summing the power of all stations; Indicates all time periods The maximum value of the total power of the entire network is taken.

[0054] This indicator is used to characterize the peak load pressure that the charging and swapping network exerts on the power grid, and serves as one of the important objectives for subsequent optimization and scheduling.

[0055] When adjusting arrival rate When reducing the number of vehicles in service during peak hours by adjusting service rhythms (such as guiding some demand to arrive at off-peak times or transferring it to other stations), The maximum value decreases, thereby inhibiting This suppression effect is also reflected in the "power-wait-cost" link. and This allows high-power, high-congestion conditions to be further suppressed in subsequent decision-making.

[0056] In step 6, the rolling optimization is a weighted multi-objective minimization, which means that within each rolling time window, the predicted arrival, queuing, and power status are used as inputs, and the three objectives of "queuing peak", "network peak power", and "operation / logistics cost" are combined into a scalar objective with weights and minimized. ; In the formula: The median (50th percentile) of station queuing time is used to characterize the typical queuing experience at the user level and suppress queuing peaks. This represents the peak power of the entire charging and swapping network within the rolling window, used to reflect the peak load impact on the power grid; This represents the operation and maintenance and logistics-related costs incurred by the system during the current scheduling cycle (such as battery allocation, inventory adjustment and service guidance costs). , , These are the weighting coefficients for the three types of objectives mentioned above, used to adjust the relative importance of user service quality, power grid operation safety, and system operation cost. Their values ​​can be set according to the scheduling strategy and operation scenario.

[0057] The window first displays the waiting time for each station. Queue time statistics are generated, and: , (when )or (when ); in: This refers to the number of vehicles within the station. To provide the number of workstations for parallel service; This refers to the average service duration. The shortest remaining service time.

[0058] Therefore, taking the median of the set of peak-hour queuing times yields... ; The network peak power is obtained by summing the concurrent power of each site, and is defined as: ; Among them: the service duration of a single bicycle meets the requirements Therefore, under the constraints of arrival rate and number of parallel services, With the number of concurrent vehicles on service Common changes; It is used to describe the operational costs of allocation and turnover required to achieve traffic diversion, peak shaving and queuing reduction.

[0059] The rolling optimization process is as follows: The result is obtained within the current window based on the superposition of the predicted arrival rate and the selection probability. ,renew , and ,calculate and Then, controllable decisions (such as diversion / diversion parameters, concurrent service power within the station, or allocation of available workstations / inventory) are adjusted to minimize the objective function; after the window is scrolled forward, the calculation is recalculated based on the latest observations and forecasts. , and The solution is repeated to continuously reduce queuing peaks and suppress network peak power while the constraints are satisfied.

[0060] Weights can be adjusted according to the actual strategy. This means changing the weight coefficients of each objective in the overall objective function to reflect the emphasis under different operational stages or management strategies. Let the objective function be: ,in, ; When the system focuses on improving the user service experience, improving To strengthen the suppression of queuing time quantiles; to improve the efficiency of queuing time quantiles when the distribution network is under high load or constrained conditions. Prioritize reducing peak power; when it is necessary to control scheduling frequency and inventory turnover costs, increase This limits operation and maintenance / logistics expenses. Weights can be set in segments based on runtime, load level, or scheduling mode, and are dynamically updated during rolling optimization.

[0061] Optimization objectives include: Among the optimization objectives of site queuing time quantile, network peak power, and operation / logistics costs, site queuing time quantile is the most important factor. Used to characterize peak queuing levels, it consists of a set of station queuing times. Take within the scrolling window The quantile value is obtained, that is ,in: Based on the number of vehicles in the station With service workstations Decide; Peak power of the entire network Used to characterize the maximum impact on the power grid, it is defined as follows: ,in, For the site During the period The concurrent power supply power varies with the number of vehicles and the power per vehicle. change.

[0062] Operation and maintenance / logistics costs The costs of allocation, inventory turnover, and scheduling operations incurred in implementing demand-driven, peak-shaving, and queuing reduction can be expressed as: ; in: For unit operation and maintenance / schedule costs, This corresponds to the scheduling decision quantity.

[0063] Setting non-weakeable service guarantee constraints for users with rigid energy replenishment needs, such as public transportation, means that their energy replenishment needs must be met according to the predetermined sequence and capacity during the optimization process, and they will not be subject to reduction or weakening. This can be achieved through the following constraint forms: Firstly, there is a limit on queuing time, requiring buses to wait at bus stops. Time period The waiting time satisfies: ; Secondly, service priority or capacity reservation constraints require reserving a minimum service capacity for public transport users. or ; Third, arrival must meet constraints, requiring a certain amount of public transport demand to reach the destination. The optimized value remains unchanged or is no lower than the planned value. These constraints are embedded as hard constraints in the rolling optimization problem, ensuring that adjustments to the objective function only apply to flexible users such as private vehicles. This reduces queuing peaks and suppresses peak network power while ensuring service reliability for users with rigid energy needs, such as public transportation.

[0064] The inventory update equation describes the dynamic evolution of the fully charged battery inventory at site s within adjacent time steps. The inventory at the next time step is determined by the current inventory, the amount of charging replenishment within the site, the amount of replenishment through cross-site transfers, and the consumption of battery swapping services. ; in: This represents the amount of fully charged batteries at station s at time t+1, i.e., the inventory status after charging, allocation, and consumption within time period t. This indicates the fully charged battery inventory at station s at time t; Indicates the charging power within the station; Indicates the time step for inventory updates (e.g., 15 minutes or 1 hour), used to convert continuous power P(s,t) into the amount of new charging power or the equivalent number of fully charged batteries during that period. This indicates the amount of batteries transferred from other sites; This represents the index of sites other than site s, and typically involves traversing the set of all sites that can participate in the transfer, used to indicate the source of the transfer; This indicates the amount of electricity consumed for replenishment.

[0065] In step 6, optimizing the site battery inventory allocation, battery transfer routes, in-site charging power scheduling, and navigation service fees means that within a rolling time window, the above variables are used as joint decision variables and coordinated to minimize the comprehensive objective function under the conditions of satisfying inventory, service, and power constraints.

[0066] Specifically, the allocation of site battery inventory is constrained and optimized through an inventory update equation: ; in: Indicates at time From other sites s' to site The sum of the number of batteries allocated is used to characterize cross-site battery scheduling behavior; Decision variables include site At any moment charging power and cross-site transfer volume And meet the minimum inventory level and capacity constraints. This is to ensure that high-demand sites have sufficient available battery inventory during peak periods.

[0067] Battery allocation path optimization is reflected in the allocation quantity In the selection, the principle of conservation of allocation is applied. Combined with allocation cost item Prioritize replenishing inventory from sites with surplus inventory and lower transfer costs to sites with shortages, thereby balancing the inventory distribution among sites.

[0068] In-station charging power scheduling via constraints And optimize based on peak power performance; ; While meeting the needs of vehicle energy replenishment and inventory replenishment, the concurrent charging power during peak hours is reduced, and some charging load is shifted to off-peak hours, thereby suppressing the peak power of the network.

[0069] Navigation service fee guidance parameters, as a soft moderating variable influencing user site selection, are adjusted by the site... At any moment service fee or Enter the energy replenishment fee item This will further impact the overall cost of energy replenishment. and comprehensive value This changes the probability of vehicle selection. To achieve arrival rate The guidance and diversion of demand. Through the joint adjustment of the above four types of variables in rolling optimization, inventory balance, orderly allocation, power peak shaving and demand guidance can be achieved simultaneously under the condition that constraints are met.

[0070] In step 6, hard constraints on capacity reservation or queuing priority are set for users with rigid energy replenishment needs, such as public transport, to ensure that their energy replenishment services are not interfered with by users with flexible guidance. This means that during the rolling optimization and guidance decision-making process, their energy replenishment needs are embedded into the model as rigid constraints that cannot be reduced or delayed. This ensures that flexible guidance (such as station diversion, power peak shaving, and service fee adjustment) only affects private vehicles and does not impact the service guarantee for users with rigid needs, such as public transport. Specifically, this can be achieved through capacity reservation constraints and queuing priority constraints. Capacity reservation constraints represent the minimum service capacity or inventory reserved for public transport users within a station(s) and time period(t) to meet the following requirements: Alternatively, a lower limit could be set for the parallel service workstations. , in: The charging power allocated to public transport users is the minimum guaranteed value determined according to the public transport operation plan. Indicates site At any moment Reserve charging power (or battery swapping service capacity) for buses or that can be used for bus energy replenishment. This indicates the minimum power limit set to ensure the normal operation of public transportation. Indicates site At any moment The amount of available batteries reserved for public transport vehicles; This represents the minimum inventory level required for the refueling needs of public transport vehicles at a given time.

[0071] These constraints ensure that while optimizing and guiding the behavior of flexible users such as private cars, the charging service capacity and battery inventory of rigid charging users such as public transportation are not squeezed out, thereby guaranteeing the reliability and continuity of their operation. These constraints directly limit inventory updates and power scheduling, ensuring that any peak shaving or inventory transfer during the optimization process does not occupy this reserved capacity.

[0072] At the same time, a hard constraint on queuing priority is introduced, meaning that in the calculation of queuing and waiting time within the station, public transport users are not treated the same as flexible users, and their waiting time meets the following requirements. Meanwhile, the queuing time for private vehicles remains the same. Make an estimate.

[0073] In practice, this can be achieved through a "priority access" rule or an "independent service channel," meaning that once a bus arrives, it will occupy an available or dedicated workstation first, and its service will not be included in the queue of flexible users.

[0074] In rolling optimization, the aforementioned capacity reservation and priority constraints are directly added to the feasible region as hard constraints, making the objective function... ; Optimization will only be carried out under the premise of meeting the service guarantee requirements for public transport users; Therefore, even if the station selection probability of flexible users is adjusted by guiding the navigation service fee or value function, the arrival, queuing and refueling process of rigid refueling users such as buses remains unchanged, thus ensuring that their refueling service is not affected by flexible guided users.

[0075] In step 7: data such as actual user arrival rate, guidance acceptance rate, and queuing time are collected, and key parameters in the feedforward model are periodically updated, including the value function weights. This is used to characterize users' relative preferences for driving costs and refueling costs; the selected model coefficients are represented as... This is used to characterize the user's sensitivity to differences in overall value when choosing a site, and the expected arrival rate of each site at different times. It uses actual observed station data and historical forecast results for rolling correction, thereby improving the accuracy of subsequent load forecasting and scheduling decisions.

[0076] The guidance acceptance rate, used by the service fee to describe the user's responsiveness to the system's guidance strategy, is modeled as a function of navigation service fee adjustments, additional detour costs, and vehicle state of charge. ; in, This indicates the probability that a user will accept system guidance and travel according to recommended stations or routes; A mapping function representing the traffic acceptance rate, used to characterize the overall response relationship of users to changes in price and travel costs; This indicates the extent of the adjustment to the navigation service fee (such as the amount of service fee reduction or subsidy) implemented to achieve traffic acquisition. This indicates the additional driving distance or equivalent driving cost incurred due to guidance. This indicates the vehicle's current state of charge, reflecting the impact of remaining driving range on user willingness to accept guidance. This is achieved through actual observation... By comparing the results with the model's predictions, we can... The parameters are updated to achieve adaptive matching between the guidance strategy and user behavior.

[0077] This invention discloses a city-level electric vehicle charging / swapping feedforward collaborative scheduling method that integrates threshold state of charge prediction and user decision updates. The technical advantages are as follows: 1) This invention constructs a city-level multi-vehicle collaborative scheduling method, realizing a closed-loop process from prediction, behavior modeling, feedback and reinjection to joint optimization, and has good engineering feasibility and promotion value.

[0078] 2) At the demand forecasting level, this invention introduces a vehicle-specific SOC threshold triggering mechanism for the first time, improving the accuracy of battery swapping demand forecasting for various types of electric vehicles; at the behavioral modeling level, it considers bounded rationality and state updates during driving to achieve dynamic decision correction and differentiated traffic diversion at stations; at the resource scheduling level, it alleviates peak queuing and balances the power load of the entire network through feedforward inventory and scheduling optimization; at the system control level, it constructs a rolling closed-loop learning mechanism to improve the robustness and adaptability of the model, demonstrating good engineering feasibility and promotional value.

[0079] 3) This invention proposes a holistic scheduling system that combines feedforward demand prediction based on vehicle-specific State of Charge (SOC) thresholds, bounded rationality user behavior modeling, dynamic decision-making updates during driving, and a queue feedback mechanism. This method constructs a closed-loop scheduling logic of "prior prediction—behavioral response—queue correction—rolling optimization," achieving efficient operation and dynamic balance of city-level electric vehicle charging and swapping systems. It is compatible with existing vehicle networking platforms, scheduling systems, and battery management platforms, possessing good scalability and deployment flexibility. Attached Figure Description

[0080] The present invention will be further described below with reference to the accompanying drawings and examples; Figure 1 This is a schematic diagram of the electric vehicle travel chain in the method of the present invention.

[0081] Figure 2 This is a map showing the distribution of first-time travel times for electric vehicles.

[0082] Figure 3 This is a flowchart of the overall process for predicting charging / swapping load.

[0083] Figure 4(a) is a comparison of the site load distribution with and without considering decision updates (charging station A). Figure 4(b) is a comparison of the site load distribution with and without considering decision updates (charging station B). Figure 4(c) is a comparison of site load distribution with and without considering decision updates (charging station C). Figure 4(d) is a comparison of the site load distribution with and without considering decision updates (charging station D).

[0084] Figure 5 This is the daily battery swapping demand curve for electric vehicles in a typical urban area. Detailed Implementation

[0085] A city-level charging / swapping feedforward collaborative scheduling method is proposed, integrating threshold state of charge (SOC) prediction with user decision updates. First, multi-source data is collected within the city, including electric vehicle SOC, location, road time-of-use speed, battery inventory and queuing information at charging stations, time-of-use electricity prices, and service fees. Based on different vehicle models, SOC thresholds are set, and vehicles that may trigger charging within the prediction window are identified, forming prior arrival rates categorized by station and time period. On this basis, a comprehensive value function incorporating driving and charging costs is constructed, and a bounded rationality decision-making model, preferably a Logit model, is used to calculate the selection probability and shortest path for each candidate station. When road speed falls below the tiered threshold or station prices or waiting conditions change, the comprehensive value is recalculated online, and the vehicle travel plan is updated. The vehicle selection probability and prior arrival rate are superimposed to form the short-term load of the station. The waiting time is estimated by using an approximate queuing time estimation method and fed back into the energy replenishment cost to achieve a closed loop of "prediction-selection-queuing-reselection". Within the rolling window, the station inventory pre-allocation, in-station charging power scheduling, inter-station battery transfer and differentiated diversion parameters are jointly optimized to reduce peak queuing and suppress peak power of the entire network under the constraints of inventory balance, grid capacity, transfer time window and price smoothing. Capacity reservation and queuing priority are set for rigid users such as buses.

[0086] like Figures 1-5 As shown, Figure 1 This is a schematic diagram of the electric vehicle travel chain in the method of the present invention, showing the travel patterns of vehicles in the three dimensions of time, space and energy. Figure 1 This invention illustrates that when constructing a city-level energy replenishment demand prediction model, the travel chain analysis method is used to characterize user behavior features, providing a foundation for the formation of a priori demand pool.

[0087] Figure 2 This is a map showing the distribution of first-time trip times for electric vehicles, illustrating the patterns of start-up times for different types of vehicles throughout the day. Figure 2 This invention is used to illustrate how, when identifying the set of vehicles that will trigger refueling within a rolling prediction window, the prior arrival rate distribution is determined based on the travel probability at different times.

[0088] Figure 3 This diagram illustrates the overall process of charging / swapping load forecasting, from the input of vehicle travel information to the formation of site load. It serves to explain the overall methodological structure of this invention, including key stages such as "prior prediction—behavioral decision-making—load overlay—feedback update."

[0089] Figures 4(a) to 4(d) show a comparison of site load distribution with and without decision update considerations, illustrating the optimization effect of the proposed "queueing-based backfeeding + rolling closed-loop" mechanism on the overall network load balance. When queuing time backfeeding is used for calculation... The site load is more balanced, and the overall peak load is suppressed.

[0090] Figure 5 The curve showing the daily battery swapping demand for electric vehicles in a typical urban area serves as an input example to illustrate the method of this invention in an overall application scenario. Figure 5 It shows the temporal distribution of battery swapping demand for various types of vehicles throughout the day, providing basic data for feedforward scheduling and joint optimization.

[0091] This invention provides a city-level charging / swapping feedforward collaborative scheduling method that integrates threshold SOC prediction and user decision updates. It is applicable to city-level scenarios involving multiple types of electric vehicles, multiple charging / swapping stations, and time-of-use pricing and dynamic road networks. The method employs a hierarchical, feedforward plus feedback collaborative scheduling structure, mainly comprising the following functional units: a data aggregation unit, a priori demand pool construction unit, a behavior decision and path selection unit, a driving decision update unit, a station queuing feedback unit, a joint optimization and feedforward pre-deployment unit, a rigid priority guarantee unit, and a rolling feedback and adaptive calibration unit.

[0092] Data aggregation unit: This unit is used to periodically or in real-time collect on-board SOC information, current road network node or geographical location, vehicle type identification, and travel direction of electric vehicles within the target area; collect the graded average vehicle speed of the urban road network at different times; collect the battery inventory, number of vehicles queuing at each charging station and battery swapping station, and number of charging or battery swapping positions within the station; and simultaneously collect time-of-use electricity prices, time-of-use service fees, and guidance strategy parameters released by the operator. The above data serves as the input basis for subsequent prior predictions and joint optimizations in this invention.

[0093] Prior Demand Pool Construction Unit: This unit identifies the set of vehicles that will trigger refueling within the rolling prediction window based on the driving and energy consumption characteristics of different vehicle types. For private car users, SOC iteration is performed based on actual daily travel patterns, and when the threshold refueling condition is met, the vehicle is added to the refueling pool. ; according to Figure 2 The time-sharing trip probability allocates daily mileage to different time periods; then the vehicle's State of Charge (SOC) is iterated, and when the threshold refueling condition is met: At that time, the vehicle will be added to the energy replenishment pool.

[0094] For commercial vehicles such as taxis, energy consumption and the probability of triggering a battery swap are calculated based on regional divisions and passenger travel patterns. This, combined with the current battery SOC, determines whether to trigger a battery swap within the specified window. ; The energy consumption in this mission is: ; Determine whether to trigger a battery swap within the window based on the current battery SOC.

[0095] For buses and other vehicles operating on fixed routes, refer to... Figure 3 The loop operation mode shown calculates the daily turnover number based on the route length, round-trip time, and scheduling timetable, and directly regards the SOC replenishment action before each departure as a deterministic replenishment event, thereby obtaining the deterministic triggering time of public transport vehicles.

[0096] By identifying the above three types of vehicles, the prior arrival rate, arrival volume and arrival time distribution can be generated in a rolling prediction window according to the "station-time period" statistics, which constitutes the "prior demand pool" of the present invention.

[0097] Behavioral decision-making and path selection unit: such as Figure 1 , Figure 3 and Figure 5 As shown, after obtaining the prior trigger vehicle, the present invention needs to estimate the actual landing point of the vehicle under the conditions of dynamic road network and current price.

[0098] Therefore, a comprehensive value function is established: Among them, operating costs The cost of refueling is calculated by weighting the shortest path distance, travel time, and energy consumption per unit distance. It is derived from the time-of-use electricity price, service fee, and current predicted queue time for the corresponding station; , For normalized weights.

[0099] In actual travel, users often cannot make completely rational choices. Therefore, this invention uses a bounded rationality discrete choice model to probabilistically select stations, preferably the Logit model, whose station selection probability is: ; in: This represents the rationality strength coefficient. The system assigns target stations to vehicles based on this probability and generates corresponding travel routes in the dynamic road network using the shortest path algorithm.

[0100] In-driving decision update unit: such as Figure 1 , Figure 3As shown in Figures 4(a) to 4(d), when a vehicle is traveling along the initial path, if the actual average speed of its road segment is detected to be lower than the graded speed thresholds set by this invention (e.g., different lower limits for expressways, arterial roads, secondary arterial roads, and local roads), or if changes are detected in the time-sharing price, service fee, or queuing time of its target station, the system will recalculate the comprehensive value Vcom in the above formula and resolve the Logit selection probability. If the comprehensive value of the recalculated target station or path is better than the original scheme, the vehicle terminal will be notified to update the station selection and path, realizing dynamic decision updates during the journey to more realistically reflect the user's behavioral characteristics of updating travel decisions after information perception.

[0101] This invention combines the prior arrival rate with the behavioral choice probability to form the predicted load of each station in each time period, and then uses an approximate queuing time estimation model to estimate the waiting time of the stations: ; in: , For arrival rate, For service rate, The number of charging or battery swapping workstations. The resulting waiting time will be written back into the energy replenishment cost. As input for the next rolling cycle of behavioral decisions, it realizes the feedback loop of "arrival - queuing - cognition - re-decision".

[0102] Joint optimization and feedforward pre-arrangement of units: such as Figure 3 and Figure 5 As shown, after obtaining the time-sharing predicted load of a site, this invention simultaneously optimizes the pre-allocation of battery inventory within the site, the scheduling of charging power within the site, the battery transfer between sites, and the differentiated diversion parameters within the rolling prediction window. The optimization objective is: ; in: This represents the quantile of peak-hour queuing time. This represents the peak charging power across the entire network. This is to cover battery allocation and maintenance costs. The inventory evolution at each site must satisfy the following balance: ; in: Let be the number of fully charged batteries at station s at time t. The charging capacity corresponding to the charging power within the station. The amount of battery power transferred from station s′ to station s; This represents the amount of electricity exchanged at that moment.

[0103] During the optimization process, constraints are set on grid capacity, station capacity, allocation window, and price smoothness to prevent excessive battery concentration or price jumps.

[0104] Rigid Priority Guarantee Unit: For rigid users such as buses and sanitation vehicles that must be recharged within a fixed time, this invention sets dedicated capacity reservation and / or queuing priority hard constraints in the aforementioned joint optimization to prevent them from being affected by the diversion of flexible users. When the system determines that the station inventory is insufficient to simultaneously meet rigid and flexible demands, it prioritizes ensuring the queuing order and battery allocation for rigid users.

[0105] Rolling feedback and adaptive calibration unit: This invention is based on time step The above steps are executed periodically. At the end of each cycle, actual arrival volume, actual waiting time, and the proportion of guided vehicles are collected and compared with prior predictions. Based on the deviation, key parameters are adjusted online, including the weights of driving cost and refueling cost, the rationality coefficient of the Logit model, arrival prior a priori, service rate, and referral acceptance rate function. The referral acceptance rate can be expressed as: ; in: For the adjustment range of service fees, The SOC (State of Charge) represents the additional travel distance caused by traffic diversion. By continuously calibrating within the rolling domain, the method of this invention can adapt to urban traffic fluctuations, price adjustments, and changes in vehicle behavior, maintaining the stability of the scheduling effect.

[0106] By employing the feedforward collaborative scheduling method of this invention, it is possible to schedule events when the daily demand curve is known, such as... Figure 5 As shown, the actual behavior characteristics of vehicles, decision changes during driving, and the reaction of station queuing are further unified and regionalized in a model, thereby realizing a collaborative operation mode of "predicting first, then pre-arranging, and then rolling correction" at the city scale, which significantly reduces queuing during peak hours, suppresses the peak power of the entire network, and improves battery turnover efficiency.

Claims

1. A city-level electric vehicle charging / swapping feedforward collaborative scheduling method that integrates threshold state of charge prediction and user decision updates, characterized in that... Includes the following steps: Step 1: Real-time acquisition of multi-source data on the state of charge (SOC) of electric vehicles in the urban area, road traffic flow, battery inventory and queuing time at stations, and time-of-use electricity prices; Step 2: Based on vehicle type classification, set the state of charge (SOC) threshold, identify vehicles that may trigger recharging behavior in the rolling window, and form the expected arrival rate and arrival time distribution of each station in each time period. Step 3: Construct a site selection probability model and calculate the comprehensive value of candidate sites based on the current traffic conditions, time-of-use electricity prices, and queuing time; Step 4: When the road traffic flow rate is detected to be lower than the road grade threshold, or when the station price / waiting conditions change significantly, the value function is re-evaluated in real time and it is determined whether to replace the original planned route. Step 5: Superimpose the obtained vehicle selection probabilities and arrival rates to form short-term load forecasts for each station, estimate the queuing time based on the approximate queuing time estimation method, and feed it back into the energy replenishment cost model to adjust the station selection logic. Step 6: In the scrolling window, jointly optimize the site battery inventory allocation, battery transfer routes, in-site charging power scheduling, and navigation service fee guidance parameters; Step 7: Collect data on actual user arrival rate, guidance acceptance rate, and queuing time, and periodically update key parameters in the feedforward model.

2. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 1, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 1, multi-source data on the state of charge (SOC) of electric vehicles in the urban area, road traffic flow speed, battery inventory and queuing time at stations, and time-of-use electricity prices are obtained. 1.1: For the i-th electric vehicle, the remaining battery charge at time t is determined by the remaining charge at the previous time t-1 and the charge consumed during the time period t due to vehicle operation. The iterative model for its remaining battery charge is expressed as follows: ； In the formula, and Let represent the remaining battery power of the i-th electric vehicle at times t and t-1, respectively; This represents the proportion of the distance traveled within time period t to the vehicle's daily mileage. This represents the daily mileage of the i-th electric vehicle; This indicates the energy consumption of an electric vehicle per 100 kilometers. 1.2: Introducing a speed-flow model to construct a dynamic traffic network model to simulate the actual operating state of the traffic network at different times. Its expression is: ； ； In the above formula: This represents the speed of travel on road ij within time period t; This indicates the zero-flow travel speed of road ij; This represents the traffic flow on road ij within time period t; This indicates the traffic capacity of road ij. An adaptive exponent for the speed-flow relationship; , and These are model parameters; 1.3: The site battery inventory data is divided into three categories according to the working status: batteries to be charged, charging batteries, and fully charged batteries. During the time period They are respectively recorded as , , , and satisfy the constraints of the total number of batteries and the capacity of the charging bin: 、 ； in: This represents the total number of batteries at the battery swapping station. The number of charging bays within the station; to depict the evolution of inventory as battery swapping demand changes, battery status transitions are updated by time period: ； ； ； in: express The number of batteries swapped during the time period, i.e., the battery consumption at the site. express The number of batteries that transitioned from "waiting to charge" to "charging" during the specified time period; if a situation arises where "fully charged batteries are insufficient to meet battery swapping needs," then batteries that meet the needs will be selected from the rechargeable batteries. Battery replenishment and swapping services; among which For the SOC of the selected rechargeable battery, The lower limit of the SOC of the swappable battery that is acceptable to users; 1.4: Station queuing time data is represented by "arrival waiting time", denoted as... Due to queuing time With energy replenishment service time composition: The site during the time period The queuing time is approximated by parallel service: when the number of vehicles in the station hour: ； when hour: ； in: For the site Number of vehicles at the station, including those in queue and those currently in service; This refers to the number of energy replenishment facilities, i.e., the number of charging piles or battery swapping stations. For average refueling time, To round down, This refers to the minimum remaining charging time for vehicles currently receiving service at the station; the charging service time is determined by the station type: charging station A constant value can be taken for the battery swapping station. ;in, For the arriving SOC, To replenish energy and end SOC, For the battery's rated capacity, For charging efficiency, The power of the charging pile; 1.5: Time-of-use electricity price data includes grid-side time-of-use electricity prices. In addition, there is a time-sharing service fee for the charging station, used to calculate the charging cost for different time periods and participate in subsequent value assessments; the charging cost is: ； The battery swapping price is: ； This can be used to determine the cost of battery swapping: ； in: The difference in battery capacity before and after the battery swap. This is a compensation item for non-fully charged battery swapping; time-of-use electricity prices and charging / swapping prices can be plotted as curves based on time periods and mapped to the corresponding times of vehicle arrival. The service fee parameter serves as the input for the energy replenishment cost.

3. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 2, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 2, the state of charge (SOC) threshold and triggering mechanism are set for different vehicle models respectively: 2.1: Private cars generate daily mileage based on a log-normal distribution. The trigger point is determined by iterating the State of Occurrence (SOC) hourly using the Monte Carlo method. Specifically, this includes: Daily mileage of electric private cars It follows a log-normal distribution, and its probability density function is: ； In the formula: The daily mileage is represented as a value. The probability density function value at time; The value representing the daily mileage; Used to characterize the degree of dispersion of daily mileage in the logarithmic field; Used to characterize the concentration level of daily mileage in the logarithmic field; Represented by natural constant Operations on exponential functions with base 0; 2.2: The taxi area is divided into grids, and the battery swapping time is estimated based on OD (Original Destination) trips and route steps, as detailed below: The planning area is divided into a square grid with side length h, centered at coordinates (x, y). Assume the distance from the passenger's travel demand point to their destination is... Then the number of grids that need to be traversed is ; The power consumption for a single battery swap is: ； In the formula: The amount of electricity consumed by electric taxis; Number of grid cells; The side length of a single cell; 2.3: The first battery swap time for buses is set daily based on departure intervals, route length, and fixed energy consumption budget, as detailed below: The electric buses follow regular routes and stop times every day; taking the first... Taking a bus route as an example: the straight-line distance between the bus's starting station and its terminal station is... The start and end times are respectively and The departure interval is The bus travels at a speed of The time required for a bus to make one round trip between the starting station and the terminal station is: ； In the formula: This indicates the time required for a bus to make one round trip between its starting and ending points. Indicates rounding down; This indicates the number of departure intervals covered in one round trip; The number of buses required for this route is: ； In the formula: Indicates the first The number of departures on a single route within a single operating day; b indicates rounding down; b indicates bus scheduling. The formula for the remaining battery power is: ； In the formula: Indicates vehicle After driving to the Remaining battery power at each road segment Indicates the first Remaining power in each section Used to depict the power consumption of a vehicle during driving.

4. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 3, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 2, the expected arrival rates and arrival time distributions for each station at each time period are generated, as detailed below: Within the scrolling window, time is discretized into equal-length time intervals. , This indicates the total number of time periods the scrolling window is divided into, with each period having a length of [length value missing]. First, determine the time period when the vehicle "triggers recharging", then calculate its "arrival time at the station", and finally aggregate statistics by "station-time period" to obtain the expected arrival rate and arrival time distribution. When a vehicle meets the charging trigger conditions, charging is considered triggered. To determine the trigger point, the battery level is recursively calculated using daily mileage and the proportion of trips during different time periods within the day. ； ； In the formula: This refers to the vehicle's daily mileage. For the first The percentage of mileage traveled during different time periods meets the requirements. ; For time period Mileage; Indicates vehicle During the period Remaining battery power at the end Indicates the remaining battery power in the previous time period; This gives us the initial trigger point: ； In the formula: Indicates vehicle The first trigger point for energy replenishment; Indicates vehicle At any moment The remaining battery power; Represents the threshold coefficient ; The larger the value, the earlier the energy replenishment is triggered; Indicates the baseline value of the electricity. This indicates taking the minimum time that satisfies the condition; After the trigger, the vehicle calculates its arrival time for the set of reachable candidate stations, without determining the final target station in step 2; the final target station is determined by the probability selection model in step 3. The arrival time is calculated for each of the candidate reachable stations in the prior demand pool. ； In the formula: Indicates vehicle The "nearest site" at the moment of triggering power replenishment; Indicates a collection of sites Find the station that minimizes the distance to the next station; This represents the set of all available charging / swapping stations within the city. Indicates vehicle At the moment of triggering energy replenishment The current location; Indicates the vehicle's current location to the station. The distance function; The speed limit for a road segment is expressed as follows: ； In the formula: Indicates the speed of the road segment; Indicates the free-flow velocity; Indicates traffic flow on a road segment; Indicates traffic capacity; These are model parameters; And the travel time for each section: ； In the formula: The length of the road segment; Route travel time: ； In the formula: Indicates the travel time of the route; Indicates road segment At any moment Passage time; Indicates at time A defined driving route; Therefore, the arrival time is: ； In the formula: Indicates the arrival time; Finally, the arrival volume is aggregated and counted by "site - time period": ； In the formula: Site In the Number of vehicles arriving in each time period } is an indicator function; Indicates the first Each time period; Indicates the length of each time period; Then the site During the period The expected arrival rate is: ； In the formula: This indicates the site During the period Expected arrival rate; The desired result can be approximated by averaging multiple simulations; Site The arrival time distribution is as follows: ； In the formula: Indicates site During the period The probability of arrival; This indicates the total number of time periods divided within the scrolling window; The expected arrivals at the station in each time period are normalized to obtain the probability distribution of arrival times in each time period.

5. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 4, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 3, a comprehensive value function is constructed that includes driving costs and refueling costs: ； In the formula: A comprehensive value function that includes both driving costs and refueling costs; For normalized weights, and satisfying ; The cost of operation includes the distance traveled, the time spent traveling, and the energy consumed. The cost of replenishing energy includes time-of-use pricing and queuing time; It is a value mapping function; If foreground theory mapping is not used, then let , The input cost value degenerates into a linear weighted average in the model. ； The probability distribution for vehicle stop selection is generated using a bounded rationality model, and the input to the Logit model is the aforementioned comprehensive value. ; For the Vehicles that trigger a refueling demand, at any time Obtain its set of reachable candidate sites and for each candidate site Calculate the comprehensive value The comprehensive value function consists of driving costs and refueling costs: ; Then the vehicle At any moment Select site The probability is expressed as ,in: Indicates vehicle At any moment Select candidate sites The probability of each station is 1; Indicates vehicle At any moment The set of reachable candidate sites; Indicates vehicle At any moment Select site The comprehensive value; Represents the coefficient of rationality intensity. The larger the value, the more sensitive it is to differences in value, and the more likely it is to make "optimal / certain" choices. The smaller the value, the more dispersed / random the selection. This is how it was obtained. It is the "probability distribution of site selection", which is directly used for subsequent arrival rate superposition and load forecasting.

6. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 5, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 3, the overall value of the candidate sites is calculated: For the vehicle at time For vehicles that trigger refueling needs, the available candidate stations are... any of the stations Comprehensive value cost The calculation is divided into two parts: "driving cost" and "refueling cost". Firstly, under a dynamic traffic network, the routes leading to stations... The candidate paths are then normalized and weighted according to distance, energy consumption, and time to obtain the path travel cost: x .； In the formula: Indicates the cost of traveling the route. This represents the number of road segments in the path. The first Road segment length, driving energy consumption, and driving time; These represent the minimum and maximum values ​​of the corresponding indicators for each road segment across the entire road network. The minimum / maximum values ​​for energy consumption indicators; The minimum / maximum value of the time indicator; The weights are distance, energy consumption, and time, respectively, and satisfy the following conditions: ; by As the edge weights in the shortest path algorithm, we obtain the distance from the vehicle's current location to the station. The shortest path, whose segment weights are summed, is the travel cost. : ; On the energy replenishment side, first calculate the station At any moment The cost of recharging and the waiting time are calculated, and then normalized and summed to obtain the comprehensive cost of recharging. ; If charging is required, the charging cost is as follows: ； In the formula: This indicates the charging cost; This is expressed as battery capacity (or rated capacity). Indicates time-of-use electricity pricing; This indicates the charging service fee; This indicates that charging is complete and the system is fully charged (SOC). Indicates the state of charge (SOC) before energy replenishment; If it is a battery swap, the battery consumption will be as follows: ； In the formula: Indicates the amount of electricity exchanged; Indicates the battery's SOC after battery swapping; The battery swapping costs are as follows: ； In the formula: This indicates the cost of battery swapping; , For battery swapping service fees; For corrections such as discounts / offers; The waiting time is as follows: ； In the formula: This indicates the total waiting time of the vehicle within the station; Indicates the waiting time in the queue; Indicates the duration of the power replenishment service; Power replenishment service hours For charging stations: in, This indicates that charging is complete and the system is at state of charge (SOC). Indicates the State of Charge (SOC) upon arrival at the station; Indicates the battery's rated capacity; For charging efficiency; The power of the charging pile; The battery swapping station can be approximated by a constant duration, i.e. ;in: For battery swapping service time, The battery swapping operation takes approximately 5 minutes. When the number of vehicles in the station Less than the number of facilities If the queue time is 0, then proceed as follows: ； In the formula: This represents the average charging time. To round down; The shortest remaining refueling time for vehicles currently in service at the station; Indicates site Current number of vehicles at the station; Energy replenishment costs With waiting time After normalization and summing, the comprehensive power replenishment cost is obtained: ； In the formula: This indicates the overall cost of electricity replenishment; These are the minimum and maximum charging costs for all stations, respectively. These are the minimum and maximum waiting times for all stations, respectively. The comprehensive value function is as described in the aforementioned formula. Calculate, where, These represent the value mappings of driving costs and comprehensive recharging costs, expressed as piecewise value functions according to prospect theory: ； In the formula: This serves as a reference point for the highest acceptable driving cost for users. This serves as a reference point for the highest overall cost of power replenishment that is acceptable to users. , These are the risk preference coefficients under the scenarios of gain and loss, respectively. This represents the loss aversion coefficient.

7. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 6, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 4: when the road traffic flow speed is detected to be lower than the road grade speed threshold, the road grade speed threshold is set according to the road type: including expressways, arterial roads, secondary arterial roads, and local roads; the value function is re-evaluated in real time and it is determined whether to replace the original planned route; specifically as follows: Let the vehicle's current planned route be Composed of several road sections The system consists of components that obtain the real-time average traffic flow speed of each road segment along the path at a fixed refresh cycle. If it exists Make ,in Given the speed threshold for each road class, If it is a road segment type, then "real-time reassessment" will be triggered; Based on the latest speed-updated dynamic road network weights, the road segment travel time is taken as... , The length of the road segment; And update the time item of the candidate path accordingly. Energy consumption items ; Substitute this back into the route travel cost calculation from step 3: ； In the formula: ; These are respectively: road distance, energy consumption, and time; Find the shortest path from the current location to the candidate site. Updated driving costs: ； In the formula: Indicates the cost of updating driving; At the same time, the arrival time is obtained according to the new route: ； in, Indicates the departure time. Indicates vehicle Arrive at the station The planned route (consisting of several road segments) (Composition) and re-evaluate the replenishment side at that arrival time: Energy replenishment costs are recorded as Billing is based on the time-of-use electricity price and service fee corresponding to the arrival time. Specifically: first, the arrival time is calculated... Mapped to its corresponding time slot Calculate the corresponding electricity price and service fee. , , If the vehicle On the site If you choose to charge, then: ； If battery swapping is selected, the battery swapping unit price should be defined first. Battery swapping has poor power output. ,but: ； in: For the battery's rated capacity, For the arriving SOC, To complete the SOC charging process, The battery's state of charge (SOC) after battery swapping. This is a compensation item for non-fully charged battery swaps, used to correct situations where the swapped battery's SOC is not at full charge. Waiting time is recorded as ,in , These are queue time and service time, respectively; queue time can be calculated by station. During the period Number of vehicles at the station With parallel service capabilities estimate: when hour ,when hour Service Hours Values ​​are taken according to the energy replenishment method during charging. A constant can be used during battery swapping. ;in For average service duration, This represents the shortest remaining service time for vehicles currently in service at the station. For charging efficiency, The power of the charging pile; The overall energy replenishment cost is still calculated using the normalized summation method: ； In the formula: vehicle At any moment Evaluation site The overall cost of energy replenishment; This represents a normalization function that transforms indices with different dimensions to the same dimensionless interval. Indicates energy replenishment costs Indicates the waiting time after arrival at the station. in, , These represent taking the minimum and maximum values ​​among the candidate sites, respectively. Finally, use the updated version and Recalculate the value function and synthesize the composite value: ； In the formula: Indicates vehicle At any moment Candidate sites Comprehensive value This represents the value mapping result corresponding to the travel cost, used to characterize the journey of a vehicle from its current location to a station. The impact of travel costs, such as distance, time, and energy consumption, incurred during the process; This represents the value mapping result corresponding to the energy replenishment cost, used to characterize the vehicle's performance at the station. The costs and waiting time incurred during the refueling process, and other refueling costs. and These are the weighting coefficients for driving cost value and refueling cost value, used to characterize the relative importance users place on driving cost and refueling cost, and satisfying the following conditions: ; Still following step 3, the foreground piecewise function is derived from the reference point. Calculate and select the new optimal solution: ； In the formula: Indicates vehicle At any moment The target site has been redefined; Indicates the relationship with the site The corresponding optimal driving route; This indicates the decision variable selected from the candidate set that maximizes the comprehensive value function; Indicates vehicle At any moment The set of reachable candidate sites; Indicates vehicle At any moment Candidate sites The comprehensive value evaluation results; this formula indicates that at a given time, the vehicle selects the station with the highest comprehensive value and its corresponding route as the new travel decision scheme; and compares it with the original scheme: like ; Replace the original planned path, and if necessary, replace the target site simultaneously; otherwise, keep the original plan. In the above formula: Indicates at time The overall value gain of adopting the new solution compared to the original solution; This represents the comprehensive value of the new station / route plan recalculated based on current road conditions, electricity prices, and queuing status. This indicates the overall value corresponding to the originally planned sites and routes; The minimum value improvement threshold is used to characterize the lowest level of benefit that users perceive and are willing to accept in accepting a change in their decision; when If the new plan has significant advantages in overall value, it will trigger station and route updates; otherwise, the original travel and refueling plans will remain unchanged.

8. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 7, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 5, the vehicle selection probabilities and arrival rates obtained in steps 2 to 4 are summed, as follows: Step 2 provides the arrival time distribution of vehicles at candidate stations, and the probability of vehicle (i) arriving at station (s) is denoted as... If the vehicle Arrive at the station Arrival time To determine the value, its probability of arrival during a specific time period can be reduced to an indicative form, at which point the comprehensive value function... According to the corresponding arrival time The site status is calculated; }； Steps 3 and 4 provide the vehicle's latest decision-making moment. The probability of selecting site(s) Then station(s) in time period The expected arrival quantity and expected arrival rate are respectively: ， ； In the formula: Indicates site During the period The expected number of arriving vehicles is calculated as follows: for all vehicles Summation, at the decision moment of the vehicle Select site The probability, Indicates site During the period Expected arrival rate; The duration of the time period; If we overlay by vehicle group / source unit, let source (g) be in the time period The triggering strength is Its average station selection probability is ,but ; In the formula: Average station selection probability; When generating short-term load forecasts for each site, the "expected arrival amount" is further mapped to "energy demand / power demand"; taking charging as an example, the single charging amount for vehicle (i) can be written as: ； In the formula: The amount of energy that the vehicle (i) can charge on a single charge; Indicates battery capacity (rated capacity); For the arriving SOC, The charging process is complete at SOC; the corresponding charging time is as follows: ； In the formula: Indicates charging time For charging efficiency; This refers to the charging power. Then station(s) in time period The expected energy requirement is: ； In the formula: This indicates the expected power demand for recharging at the site during different time periods; Indicates the vehicle at the moment of decision-making Select site The probability of; Indicates that the vehicle has arrived at the station. and fall into the time period The probability of; This indicates the amount of electricity that needs to be replenished. Average power prediction can be obtained using "energy / duration length": ； In the formula: Indicates site During the period The expected power of the supplementary energy; Indicates site During the period The expected replenishment power demand within the area; Indicates the duration of a single time period.

9. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 8, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 5, the queuing time at the site is estimated using an approximate queuing time estimation method based on parallel service workstations. To obtain the station queue time, first in the time period The system internally determines whether the vehicle has a need for additional power, for example: when the vehicle's remaining battery power is sufficient. This triggers the battery swapping demand; and iterates based on the remaining battery capacity: ； In the formula: and Representing vehicles During the period and The remaining battery power; Indicates road During the period The traffic condition influence coefficient is used to characterize the corrective effect of traffic flow speed changes on energy consumption per unit mileage. Indicates vehicle On the road The driving distance on; By continuously analyzing the demand sample size for each time period, the predicted arrival scale of the station during that time period can be formed, which can be further converted into the number of vehicles within the station. Based on this, the waiting time is estimated by approximating the queuing of "parallel service workstations" within the station: The waiting time after arriving at the station is: ； In the formula: This indicates the total waiting time of the vehicle within the station; Indicates the waiting time in the queue; Indicates the duration of the power replenishment service; Queuing time is segmented: ； In the formula: Indicates the number of vehicles in the station; Indicates the number of facilities; as well as: ； In the formula: Indicates the average refueling time; To round down; For the site The number of vehicles inside; The number of warehouses that can be served in parallel at a site; The "shortest remaining refueling time" for vehicles currently receiving service at the station; The energy replenishment service hours are as follows: (Charging station) or (Battery swapping station); In the formula: Indicates the duration of the energy replenishment service; For arrival SOC; To replenish energy and complete SOC; This refers to the battery's rated capacity. For charging efficiency; The power of the charging pile; Indicates the battery swapping operation time; When the above-mentioned queuing estimate is "reinjected" into the energy replenishment cost and participates in the comprehensive value update, the waiting time and cost will be included in the comprehensive energy replenishment cost: ； in: As input to the subsequent value function, the comprehensive supplementary power cost is used. For energy replenishment costs; The waiting time after the vehicle arrives at the station; These are the normalized upper and lower bounds for the energy replenishment costs of candidate sites; These are the normalized upper and lower bounds for the waiting time at candidate sites, respectively. In the cost-value mapping of energy replenishment, let And the value mapping corresponding to the driving cost. Together they constitute a comprehensive value function: in, The output of the energy cost function, The value mapping function for energy replenishment costs, For driving costs, The value mapping function for driving costs, and These are the weighting coefficients for the two types of costs.

10. The city-level electric vehicle charging / swapping feedforward collaborative scheduling method according to claim 9, which integrates threshold state of charge prediction and user decision update, is characterized in that: In step 6, the objective function is: ； in: This indicates the weight of the objective of "reducing peak user queuing times and improving service levels" in multi-objective optimization, reflecting the system's emphasis on user queuing experience; This indicates the weight of the objective of "suppressing peak power in the charging and swapping network and alleviating pressure on the distribution network" in multi-objective optimization, and is used to reflect the system's emphasis on grid security and peak load constraints; This indicates the weight of the "controlling operation and maintenance and scheduling costs" objective in multi-objective optimization, reflecting the system's cost control preferences in battery allocation, inventory turnover, and related operation and maintenance behaviors; Indicates queuing time Quantile values ​​are used to characterize the typical level of queuing time at each station within a rolling time window, thereby reflecting the overall severity of queuing peaks. It represents the maximum concurrent power value aggregated at nodes of the charging and swapping network or distribution network within the rolling forecast window, that is, the maximum value of the sum of power demand at each time period, used to measure the peak load level of the power grid. This represents the operation and maintenance / logistics cost metric generated to achieve demand guidance and peak shaving, and is used to depict the comprehensive costs brought about by battery allocation, inventory adjustment and related scheduling operations.

Images