Vehicle soc optimal scheduling method for smart rail line shared charging resource

Abstract

The application provides a vehicle SOC optimal scheduling method for smart rail line shared charging resources, relates to the technical field of rail transit scheduling control, and solves the unreliable charging scheduling problem of multi-line smart rail vehicles in a shared charging resource environment. The method establishes a rolling horizon scheduling framework and constructs unified state information of vehicles, depots and power grids. Then, the target SOC of the vehicle executing the next task is calculated, and the corresponding charging scheduling constraint condition is constructed. A space-time coupling model of the track and charging pile resources is further constructed to represent the space-time resource occupation constraint. Based on the obtained or constructed various information, a joint optimization model is constructed and solved to synchronously decide the parking track and time period of the vehicle, the charging pile allocation, the charging time length and the charging power, generate corresponding instructions and issue them. The application significantly improves the utilization efficiency of the charging resources under the multi-line shared depot, and improves the vehicle on-line punctuality rate and the system operation capacity guarantee ability.

Description

Technical Field

[0001] This invention relates to the field of rail transit dispatching and control technology, specifically to an optimal vehicle SOC dispatching method for shared charging resources on intelligent rail transit lines. Background Technology

[0002] As a new type of urban rail transit, the intelligent rail transit system (IRS) uses electrically powered vehicles that rely on charging facilities within the depot for power replenishment. When multiple IRS lines share the same depot, the number of charging tracks and the available DC charging stations are extremely limited. Since departure times for each line strictly adhere to the timetable, the reliability and stability of the line's capacity are directly affected by whether the vehicles can be fully charged and ready to depart before the scheduled departure time.

[0003] Currently, several scheduling schemes exist in fields such as electric buses. Their typical practices can be summarized as follows: using a first-come, first-served charging rule based on the order in which vehicles return to the depot; statically triggering charging queuing based on battery state-of-charge thresholds; and performing simple queuing scheduling based on the immediate availability of charging stations and the remaining driving range of vehicles. Simultaneously, simplified linear energy consumption estimation models are typically used, generally not taking into account the energy recovery generated by regenerative braking. If these schemes are directly transplanted to the scenario of shared depots for multiple intelligent rail lines, a series of incompatibilities will be exposed.

[0004] The operation of intelligent rail transit systems (IRS) is highly planned, with vehicle departure times strictly constrained by the timetable, unlike conventional buses which can flexibly adjust subsequent tasks. Current solutions, based on the assumption of task adjustability, rely heavily on queuing and sequencing rules that can easily cause vehicles to miss their scheduled departure times, leading to timetable disruptions. Existing solutions typically treat charging stations as independent point resources for allocation, neglecting the physical layout and lane occupancy characteristics of the depot tracks themselves. When IRS vehicles are parked on the tracks, they obstruct the entry and exit of subsequent vehicles on the same tracks. Furthermore, vehicle entry and exit routes are fixed and require sequential execution, preventing parallel processing and creating complex coupled congestion. Existing models fail to cover these resource constraints, potentially causing severe delays in entry and exit points during actual scheduling.

[0005] Existing energy consumption estimation methods are relatively crude, failing to fully consider the specific positioning of intelligent rail vehicles on different routes, variations in gradient, differences in power consumption under different operating modes, and the recovery effect of regenerative braking energy. This leads to significant deviations in the prediction of battery state of charge. In public transportation scenarios, such deviations can be partially mitigated by flexibly adjusting service schedules. However, in intelligent rail systems with fixed operating schedules, prediction errors are directly transmitted to the charging planning stage, reducing scheduling accuracy. Furthermore, when multiple routes share resources, there are significant differences in vehicle energy consumption levels, departure frequencies, and one-way mileage among the routes. Existing solutions lack a mechanism that can uniformly quantify the urgency of vehicle charging and the severity of energy shortages across routes, making it difficult to achieve fair and efficient allocation of charging resources at a global level.

[0006] Existing technologies are often static or semi-static scheduling technologies, lacking the ability to perform rolling optimization to cope with real-time disturbances. Dynamic factors such as the advance or delay of the actual return time of vehicles, the decrease in the maximum charging power of batteries due to low temperatures, and temporary adjustments to the power limit of the power grid can all cause the established scheduling scheme to fail, and existing methods are unable to make timely and effective adjustments in such situations.

[0007] Therefore, existing technologies are insufficient to meet the demands of intelligent rail transit systems for coordinated, refined, and dynamic scheduling of vehicle charging sequence, parking resource occupancy, and battery state of charge under conditions of multiple shared lines, rigid operating schedule constraints, and physical limitations of tracks. The industry urgently needs a new scheduling method that can systematically address these complex constraints and improve the utilization efficiency of shared charging resources and the reliability of line operation. Summary of the Invention

[0008] The purpose of this invention is to address the unreliable charging scheduling problems caused by rigid operating schedules, track congestion, inaccurate energy consumption prediction, and lack of dynamic adjustment mechanisms for multi-line intelligent rail transit vehicles operating in shared charging resource environments. Therefore, this invention proposes an optimal vehicle SOC scheduling method for shared charging resources on intelligent rail transit lines. This invention significantly improves the utilization efficiency of charging resources in multi-line shared depots, and enhances vehicle on-time performance and system capacity assurance.

[0009] The present invention employs the following technical solutions to achieve its objective:

[0010] A vehicle SOC optimal scheduling method for shared charging resources on intelligent rail transit lines, the method includes the following steps:

[0011] S1. Establish a rolling time-domain scheduling framework to periodically perform prediction and decision updates;

[0012] S2. Obtain vehicle operating status data and combine it with vehicle operation diagram data and real-time status data of vehicle depot and power grid to construct unified status information;

[0013] S3. Based on the operating status data and the operating diagram data, calculate the target SOC for the vehicle to perform the next operating task, and construct charging scheduling constraints including the target SOC, rigid constraints on departure time, battery characteristic constraints, and grid power limits.

[0014] S4. Construct a spatiotemporal coupling model of track and charging pile resources within the depot. The spatiotemporal coupling model is used to characterize the spatiotemporal resource occupancy constraints of vehicles on track and charging piles.

[0015] S5. Based on the unified state information, the charging scheduling constraints, and the spatiotemporal coupling model, construct and solve the joint optimization model to simultaneously decide on the vehicle's parking lane and time period, charging pile allocation, charging duration, and charging power.

[0016] S6. Based on the solution results of the joint optimization model, generate vehicle entry allocation instructions, charging instructions and online release instructions, and after each rolling cycle, re-execute the process of steps S2 to S5 according to the latest status, and regenerate new instructions of various types.

[0017] Specifically, in step S1, establishing the rolling time-domain scheduling framework includes: setting the rolling prediction window length, rolling execution cycle, and number of prediction levels; discretizing the prediction window into a time slot sequence and defining the boundary time of each time slot; at the beginning of each rolling cycle, updating the real-time status of the depot based on the latest current status, and performing forward predictions on the vehicle return time, track availability, and grid power limitations within the prediction window; and executing the scheduling decision corresponding to the first time slot in each cycle.

[0018] This step constructs a rolling prediction structure for multi-cycle iterative updates. By dividing the scheduling problem into short-cycle decision-making and long-cycle prediction, depot scheduling can be continuously updated under external uncertainties such as state changes, vehicle arrival time deviations, and power grid constraint fluctuations. The rolling time-domain structure ensures that each optimization is based on the latest state information, enabling the scheduling process to adapt in real time.

[0019] Specifically, in step S2, the construction of unified status information includes: collecting the vehicle's current SOC, the vehicle's estimated return time, and the vehicle's location information; collecting the departure schedule, line length, section energy consumption parameters, inter-station gradient information, and regenerative braking capability parameters of the line to which the vehicle belongs; collecting the track occupancy status of the depot, the availability status of charging piles, and the health status of power modules; and collecting the maximum power supply capacity of the power grid and its power limitation information that changes over time.

[0020] This step collects multi-source online data from vehicles, lines, depots, charging systems, and the power grid to form a unified state vector that can be used for constraint generation, SOC derivation, and resource occupancy analysis. The data involved can be discretized over time to ensure that the data structures from different systems are highly aligned in the periodic dimension, providing a consistent data foundation for the subsequent construction of joint optimization models.

[0021] Preferably, in step S3, calculating the target SOC for the vehicle to perform the next running task specifically includes: estimating the energy requirement required to complete the next running task based on the route length, gradient distribution, stop characteristics, and regenerative braking characteristics of the vehicle's next running task; calculating the minimum SOC value required for the vehicle to complete the task based on the energy requirement; and constructing a safety margin model by combining the vehicle's battery aging state, battery temperature conditions, and route operation delay risk, and adding a safety margin to the minimum SOC value to form the target SOC.

[0022] The target SOC in this step serves as the core input for subsequent scheduling, guiding charging duration, charging power allocation, and parking window planning, ensuring that the SOC of the vehicle before it goes online meets the requirements of the operation plan and guarantees operational reliability.

[0023] Preferably, when estimating the energy demand required to complete the next operation task, a regenerative braking energy prediction model is integrated; the regenerative braking energy prediction model predicts the energy that the vehicle can recover during braking based on the inter-station gradient information, the regenerative braking capacity parameters, and the vehicle's interval operating speed mode, and uses the obtained prediction value to correct the energy demand estimation result.

[0024] Specifically, in step S4, the spatiotemporal coupling model is used to describe the following constraints: a vehicle must occupy a lane before it can use a charging pile configured under that lane; at any given time, only one vehicle is allowed to occupy any single lane; at any given time, only one vehicle is allowed to access any single charging pile; the entry and exit routes of vehicles between lanes follow a preset fixed path sequence rule within the vehicle depot; the vehicle's occupation behavior within a lane will couple and affect the entry and exit process of subsequent vehicles on that lane or associated path.

[0025] This step treats both lanes and charging stations as finite resources with exclusivity and time dimensions, and uses a unified occupancy variable to characterize the parking and charging status of vehicles. By unifying the spatial selection and temporal exclusivity of lanes and charging stations, the resource competition process within real vehicle depots can be accurately simulated.

[0026] Furthermore, in step S5, the joint optimization model unifies the target SOC, the rigid constraint of departure time, the battery characteristic constraint, the grid power limit, and the spatiotemporal resource occupancy constraint represented by the spatiotemporal coupling model into a single mathematical framework; the objective function of the joint optimization model includes at least one of the following: minimizing the sum of the deviations between the actual SOC of all vehicles before they go online and their respective target SOCs; minimizing the congestion cost caused by resource occupancy within the vehicle depot; smoothing the fluctuation of the total grid power; and minimizing the total parking time of vehicles within the vehicle depot.

[0027] This step can fundamentally eliminate the suboptimal solution problem caused by the segmented processing of "parking planning - charging planning - online release" in traditional technologies.

[0028] Specifically, the battery characteristic constraints include the maximum charging power constraint of the battery, and the maximum charging power constraint is a variable that is dynamically adjusted according to the battery temperature conditions; in the process of solving the joint optimization model, the maximum acceptable charging power value is determined according to the real-time or predicted battery temperature of each vehicle, and the charging power allocation decision is made accordingly.

[0029] Preferably, in step S6, the generated vehicle entry allocation instruction, charging instruction, and online release instruction are issued in stages; wherein, the vehicle entry allocation instruction is first issued to the vehicle depot signal control system to arrange the route, the charging instruction is then issued to the corresponding charging pile controller to start the charging process, and the online release instruction is issued to the vehicle and exit signal control system when the vehicle has completed charging and meets the departure conditions.

[0030] Preferably, in step S6, the process of re-executing step S5 based on the latest state after each rolling cycle corresponds to at least one of the following triggering conditions: the actual return time of a vehicle is detected to be earlier or later than the predicted value; the actual congestion level in the vehicle segment exceeds the preset congestion threshold; a charging pile malfunctions or its performance degrades; the power limit issued by the grid side undergoes a sudden change exceeding the preset power threshold; or the operation plan is modified.

[0031] The rolling re-optimization mechanism in this step can continuously meet the frequently changing conditions in the actual operating environment, maintaining the feasibility and global optimality of the scheduling scheme. The entire method process has high reliability, adaptability, and real-time performance.

[0032] In summary, due to the adoption of this technical solution, the beneficial effects of this invention are as follows:

[0033] This invention significantly reduces energy consumption prediction errors caused by factors such as line gradient, operating mode, and regenerative braking by integrating a rolling time-domain optimization framework with a high-precision SOC prediction model. This ensures accurate replenishment of vehicle power before deployment, fundamentally avoiding the problem of delayed departure due to insufficient power and guaranteeing the rigidity and reliability of the operation schedule.

[0034] This invention, by constructing a spatiotemporal coupling model of lane and charging pile resources, is the first to incorporate the spatial congestion effect of vehicle parking and the time occupation of charging into a unified optimization framework. This enables the system to simultaneously handle the complex constraints of multi-dimensional scarce resources such as lane occupancy conflicts, charging pile resource competition, and total power limitation of the power grid, significantly alleviating coupled congestion within the vehicle depot and reducing scheduling conflicts during vehicle entry / exit and charging processes.

[0035] Compared to traditional segmented or sequential scheduling, the joint optimization model employed in this invention can make decisions on vehicle parking order, charging pile allocation, charging power curve, and online schedule in a single, coordinated manner. This optimization mechanism eliminates information fragmentation and inconsistency in objectives caused by segmented decision-making, avoids suboptimal solutions, and thus achieves an overall improvement in charging scheduling quality at the global level.

[0036] Due to its rolling time-domain execution mechanism and real-time state updates, this invention possesses strong adaptability and anti-disturbance capabilities. When faced with uncertainties such as vehicle arrival time deviations from predictions, charging facility power fluctuations, battery performance affected by temperature, or temporary adjustments to grid power supply limits, the system can quickly respond and regenerate optimized scheduling schemes, ensuring that scheduling instructions always match the dynamically changing operating environment.

[0037] This invention achieves efficient, fair, and intelligent allocation of limited charging resources under complex conditions where multiple lines share depots. This not only significantly improves the utilization efficiency of infrastructure such as charging piles and tracks, but also significantly improves the on-time rate of vehicle deployment and the overall capacity guarantee of the system by ensuring that every intelligent rail vehicle can be put into operation on time and with full charge. Attached Figure Description

[0038] The present invention is described in detail with reference to the following figures, which include two figures as follows:

[0039] Figure 1 This is a schematic diagram illustrating the overall process of the vehicle SOC optimal scheduling method of the present invention;

[0040] Figure 2 This is a schematic diagram illustrating the application of the method of the present invention to the intelligent rail multi-line SOC optimal scheduling system. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0042] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0043] Example 1

[0044] An optimal vehicle SOC scheduling method for shared charging resources on intelligent rail transit lines. Figure 1 The overall process of this method is briefly described below and can be viewed concurrently; the key steps of this method can be summarized as follows:

[0045] S1. Establish a rolling time-domain scheduling framework to periodically perform prediction and decision updates;

[0046] S2. Obtain vehicle operating status data and combine it with vehicle operation diagram data and real-time status data of vehicle depot and power grid to construct unified status information;

[0047] S3. Based on the operating status data and the operating diagram data, calculate the target SOC of the vehicle to execute the next operating task, and construct charging scheduling constraints including target SOC, rigid constraints on departure time, battery characteristic constraints and grid power limits.

[0048] S4. Construct a spatiotemporal coupling model of track and charging pile resources within the depot. The spatiotemporal coupling model is used to characterize the spatiotemporal resource occupancy constraints of vehicles on track and charging piles.

[0049] S5. Based on unified state information, charging scheduling constraints, and spatiotemporal coupling model, a joint optimization model is constructed and solved to simultaneously determine the vehicle's parking lane and time period, charging pile allocation, charging duration, and charging power.

[0050] S6. Based on the solution results of the joint optimization model, generate vehicle entry allocation instructions, charging instructions and online release instructions, and after each rolling cycle, re-execute the process of steps S2 to S5 according to the latest status, and regenerate new instructions of various types.

[0051] In this embodiment, the above method and process can be applied to the intelligent rail multi-line SOC optimal scheduling system, which can be referred to as follows. Figure 2 This is a schematic diagram. The system includes an external system, a multi-line SOC optimal scheduling platform for intelligent rail transit, and controlled objects. The external system includes vehicle controllers, depot monitoring systems, operation diagrams / scheduling systems, energy / carbon management platforms, etc.; the controlled objects are intelligent rail transit vehicles, depot charging tracks, charging piles / power supply units, etc. The multi-line SOC optimal scheduling platform for intelligent rail transit can be divided into a status perception module, an energy and operation prediction module, a joint optimization module, and a rolling scheduling execution module. The status perception module is used to collect SOC, operation diagram, and track status; the energy and operation prediction module is used to predict operating energy consumption and regenerative energy; the joint optimization module is used for joint optimization of charging power, track occupancy, and parking windows; and the rolling scheduling execution module is used to issue charging power and parking instructions.

[0052] This embodiment will describe in detail the specific implementation and preferred details of the method in application.

[0053] In step S1, when establishing the rolling time-domain scheduling framework, the length of the rolling prediction window is set. Set the rolling execution cycle Set the number of prediction levels The relationship is as follows:

[0054]

[0055] Discretize the prediction window into time slot sequences. Starting from serial number 0, and define the time slot boundary time. :

[0056]

[0057] In this embodiment, at the beginning of each rolling cycle... Update the real-time status of the depot and perform... Forward predictions are made regarding vehicle return to the section, track availability, and power grid limitations. The predictions here can use existing and mature conventional prediction methods, and this embodiment will not improve them; it is sufficient that the relevant prediction tasks can be completed.

[0058] In this embodiment, only the scheduling decision for the first time slot is executed in each rolling cycle. The decisions for subsequent time slots are used as forward references, and the system is recalculated with the latest state in the next rolling cycle. This enables the update process proposed in step S6 of the method.

[0059] Step S2 mainly involves the process of collecting data from multiple sources, as follows:

[0060] Collect vehicle-side status data, including: the vehicle's current status. subscript Represents the vehicle serial number, subscript Represents time slot sequence Time; estimated return time of the vehicle Next task execution identifier Battery temperature Maximum allowed charging power of the battery .

[0061] Collect data from the line and train schedule, including: departure schedules. Line length and section energy consumption parameters; station gradient information and regenerative braking capacity parameters.

[0062] Collect the resource status of the depot, including: track occupancy status; track-charging pile binding relationship; charging pile availability status and power module health status; and depot route conflict relationship set.

[0063] Collect the status of the power grid, including: maximum power supply capacity and its power limit information over time, specifically the upper limit of power grid power in each time slot. Information on the periods when power decreases or recovers, as well as the time periods during which power decreases or recovers.

[0064] Align the above data to time slots. Then, a unified state vector is formed. This is represented by a concatenation of vehicle and resource dimensions. This embodiment uses several parameters as examples, and the concatenation is as follows:

[0065]

[0066] In this formula, Indicates stock market In the time slot The occupancy status; Indicates charging station In the time slot The occupancy status can then be determined. Following this, the target SOC can be calculated and a safety margin model can be performed based on the task's energy requirements.

[0067] In step S3, the vehicle The next task will calculate the energy requirements, assuming the task route length is... Energy consumption per unit distance is The equivalent energy consumption for slope correction is The total energy consumption requirement of the task for:

[0068]

[0069] As a preferred embodiment, the regenerative braking recoverable energy is estimated; let the regeneration efficiency be... The potential energy change is equivalent to the energy of Regenerative braking can recover energy. for:

[0070]

[0071] Based on the total energy consumption requirement of the task and regenerative braking can recover energy To obtain net energy requirements And calculate the minimum task requirements. The usable battery capacity is assumed to be... The formulas are as follows:

[0072]

[0073]

[0074] Next, build a safety margin. This safety margin is determined by temperature, aging, and operational risks, and can be expressed as follows:

[0075]

[0076] In the formula, Similarly, regarding the degree to which temperature determines the safety margin, and The degree to which aging and operational risks determine the safety margin.

[0077] Finally, the target SOC is calculated, i.e. The calculation is as follows:

[0078]

[0079] The target SOC will serve as the core input for subsequent joint optimization, but it needs to be combined with rigid constraints on departure time, battery characteristics, and grid power limitations to form charging scheduling constraints.

[0080] In this embodiment, a charging power decision variable is defined. And establish the SOC kinetic equation. Assume the charging efficiency is... The time slot length is The SOC dynamic equation is as follows:

[0081]

[0082] Continue to establish power and resource constraints, including:

[0083]

[0084] In the formula, Vehicles In the time slot The maximum allowable charging power of the battery at that time. Simultaneously, a "charging only allowed upon connection" constraint is introduced when needed, making the charging power decision variable... The binary variable defined in step S4 for connecting the vehicle to the charging station Interconnected.

[0085] In addition, rigid constraints on vehicle departure are established in the operation schedule to constrain the tolerance window between the vehicle's online time and the planned time, and to ensure that the vehicle's SOC at the online time reaches or is as close as possible to the target SOC. After establishing the above-mentioned constraints, the charging scheduling constraints are obtained, which are used to construct the subsequent joint optimization model.

[0086] In step S4, in order to construct the spatiotemporal coupling model, the following operations will be performed:

[0087] Define a binary variable representing the lane occupancy of a vehicle. , indicating vehicle In the time slot Does it occupy the lane? A value of 0 indicates that the space is not occupied, while a value of 1 indicates that the space is occupied.

[0088] Define a binary variable for vehicle connection to a charging station. , indicating vehicle In the time slot Is it connected to a charging station? A value of 0 indicates no connection, while a value of 1 indicates connection.

[0089] In this embodiment, the following formula is used to represent the spatiotemporal resource occupancy constraint of vehicles on the track and charging pile. This constraint is an exclusive constraint and includes:

[0090]

[0091] Then, establish a binding coupling constraint between the track and the charging pile, and set the charging pile... Binding to Stock Channel Then the constraint is:

[0092]

[0093] Continue to establish path conflict and action continuity constraints. Path conflicts are addressed through conflict sets. Constraints on mutually exclusive passage within the same time slot and continuity of action ensure the feasibility of the vehicle's "entering the section—stopping—charging—leaving the section" sequence. All constraints of the above optimal combination will serve as constraints characterizing the spatiotemporal resource occupancy of vehicles on the track and charging piles, thus forming a spatiotemporal coupling model; the spatiotemporal coupling model also participates in the construction and solution process of the joint optimization model in step S5.

[0094] In step S5, the joint optimization model unifies the target SOC, departure time rigid constraints, battery characteristic constraints, grid power limits, and spatiotemporal resource occupancy constraints represented by the spatiotemporal coupling model into a mathematical framework and solves them; thereby, joint optimization of parking sequence planning, charging sequence optimization, power allocation decision and lane flow arrangement can be achieved.

[0095] As a preferred embodiment, the objective function of the joint optimization model includes at least one of the following: minimizing the sum of the deviations between the actual SOC of all vehicles before they go online and their respective target SOCs; minimizing the congestion costs caused by resource occupation within the depot; smoothing the fluctuations in the total power of the power grid; and minimizing the total parking time of vehicles within the depot.

[0096] Specifically, in this embodiment, a joint objective function is constructed. The preferred example formed by fusing SOC bias, resource congestion, and power smoothing terms is as follows:

[0097]

[0098] In the formula, , , These represent the weighting coefficients for different sub-items; Representative vehicle SOC just before launch; Representatives for each of the vehicles The summation, i.e. ;symbol This represents the positive part function. Various mature solution methods within the field can be used to solve the joint optimization model. This embodiment does not make specific improvements or limitations on the solution methods; they can be selected and executed as needed during actual solution processing.

[0099] Finally, in step S6, based on the solution results of the joint optimization model, various specific scheduling instructions are generated, including:

[0100] Generate vehicle entry and track allocation instructions, including the track number corresponding to the vehicle and the time window for occupancy;

[0101] Generate a charging access command, including the charging pile number, access time, and exit time;

[0102] The command to generate a charging power curve is the specific charging power decision variable. And distribute the data to the charging control system or energy management system according to time slots or stages;

[0103] Generate vehicle launch release instructions to ensure that the vehicle meets the operation diagram constraints at the time of launch and that the SOC before launch meets or is as close as possible to the target SOC.

[0104] Generate a total power plan for the depot and coordinate its execution under power grid limitations to ensure that the overall power does not exceed the limit.

[0105] As a preferred embodiment, the entire method process, after generating specific instructions and running continuously, can be updated and re-optimized in real time in the rolling time domain. The conditions that trigger this update may include: detecting that the actual return time of a vehicle is earlier or later than the predicted value; the actual congestion level in the vehicle depot exceeds a preset congestion threshold; a charging pile malfunctions or its performance degrades; the power limit issued by the grid side undergoes a sudden change exceeding a preset power threshold; and the operation plan is modified.

[0106] In this embodiment, when monitoring the vehicle return time deviation, if the following conditions are met:

[0107]

[0108] This triggers a re-optimization process; where, For vehicles The actual return period For vehicles The prediction will be made in a certain period of time. A time difference threshold is preset to reach a certain level.

[0109] In this embodiment, monitoring the availability of charging stations, the health status of power modules, and battery temperature are crucial factors. Changes, such as failures, recovery, or throttling, will trigger re-optimization.

[0110] In this embodiment, the upper limit of grid power is monitored. Mutations or changes to the run graph will trigger re-optimization if they occur.

[0111] When re-optimization is triggered, steps S2 to S5 will be re-executed. Based on the latest state, i.e., unified state information will be constructed from step S2, and new charging scheduling constraints and new spatiotemporal coupling models will be considered to jointly complete the construction and solution of a new joint optimization model. New instructions will be replaced or issued in parallel to maintain the feasibility and global optimality of the scheduling scheme.

[0112] Example 2

[0113] Based on the method of Example 1, this example verifies the calculation and scheduling process of Example 1 through a concise and specific example.

[0114] Assume there are 4 tracks in the depot, each track is equipped with 150kW DC charging pile, the three operating lines (L1 to L3) share the depot, the number of vehicles is 8 (V1 to V8), the initial SOC of the vehicles may be different levels such as 31%, 48%, 52%, the vehicle battery capacity is taken as 180kWh, and the grid power limit is taken as 600kW.

[0115] This embodiment uses the rolling execution cycle in the rolling time domain setting. Minutes, Length of the rolling prediction window minutes, and meet ,but It is 6.

[0116] Taking vehicle V3 as an example, assuming its initial SOC is 31%, the estimated return time is 10:05, and the next task is route L2. Assuming the route length is 21km, the energy consumption per unit area is 5.5kWh / km, and the gradient correction is 0.4kWh / km, then the task's energy consumption requirement is:

[0117]

[0118] Assuming the slope potential energy is equivalent to 12 kWh, and the regeneration efficiency is taken as... The recoverable energy is:

[0119]

[0120] Net energy requirement is The minimum SOC is:

[0121]

[0122] If the safety margin is 10%, then the target SOC is As a result, vehicle V3 will be given a higher energy gap drive weight in joint optimization, thus gaining priority in lane allocation and power allocation.

[0123] During charging, if vehicle V3 is connected to a 150kW charging station and the charging process is completed... Minutes are used to calculate SOC boost and measure charging efficiency. If we make an estimate, then:

[0124]

[0125] When the battery temperature rises during operation, causing the maximum charging power to drop from 150kW to 130kW, or when the grid-side power limit changes, the re-optimization is triggered according to step S6 of Example 1, and the process of steps S2 to S5 is re-executed to maintain the scheduling feasibility and on-time performance under the condition of multiple lines sharing the depot.

Claims

1. A vehicle SOC optimal scheduling method for shared charging resources on intelligent rail transit lines, characterized in that, The method includes the following steps: S1. Establish a rolling time-domain scheduling framework to periodically perform prediction and decision updates; S2. Obtain vehicle operating status data and combine it with vehicle operation diagram data and real-time status data of vehicle depot and power grid to construct unified status information; S3. Based on the operating status data and the operating diagram data, calculate the target SOC for the vehicle to perform the next operating task, and construct charging scheduling constraints including the target SOC, rigid constraints on departure time, battery characteristic constraints, and grid power limits. S4. Construct a spatiotemporal coupling model of track and charging pile resources within the depot. The spatiotemporal coupling model is used to characterize the spatiotemporal resource occupancy constraints of vehicles on track and charging piles. S5. Based on the unified state information, the charging scheduling constraints, and the spatiotemporal coupling model, construct and solve the joint optimization model to simultaneously decide on the vehicle's parking lane and time period, charging pile allocation, charging duration, and charging power. S6. Based on the solution results of the joint optimization model, generate vehicle entry allocation instructions, charging instructions and online release instructions, and after each rolling cycle, re-execute the process of steps S2 to S5 according to the latest status, and regenerate new instructions of various types. In step S3, calculating the target SOC for the vehicle to perform the next running task specifically includes: estimating the energy requirement required to complete the next running task based on the route length, gradient distribution, stop characteristics, and regenerative braking characteristics of the vehicle's next running task; calculating the minimum SOC value required for the vehicle to complete the task based on the energy requirement; and constructing a safety margin model by combining the vehicle's battery aging state, battery temperature conditions, and route operation delay risk, and adding a safety margin to the minimum SOC value to form the target SOC. In step S4, the spatiotemporal coupling model is used to describe the following constraints: a vehicle must occupy a lane before it can use a charging pile configured under that lane; at any given time, only one vehicle is allowed to occupy any single lane; at any given time, only one vehicle is allowed to access any single charging pile; the entry and exit routes of vehicles between lanes follow a preset fixed path sequence rule within the vehicle depot; the vehicle's lane occupation behavior will couple and affect the entry and exit process of subsequent vehicles on that lane or associated path. In step S5, the joint optimization model unifies the target SOC, the rigid constraint of departure time, the battery characteristic constraint, the grid power limit, and the spatiotemporal resource occupancy constraint represented by the spatiotemporal coupling model into a single mathematical framework. The objective function of the joint optimization model includes at least one of the following: minimizing the sum of the deviations between the actual SOC of all vehicles before they go online and their respective target SOCs; minimizing the congestion cost caused by resource occupancy within the vehicle depot; smoothing the fluctuation of the total grid power; and minimizing the total parking time of vehicles within the vehicle depot.

2. The vehicle SOC optimal scheduling method according to claim 1, characterized in that, In step S1, establishing the rolling time-domain scheduling framework specifically includes: setting the rolling prediction window length, rolling execution cycle, and number of prediction levels; discretizing the prediction window into a time slot sequence and defining the boundary time of each time slot; at the beginning of each rolling cycle, updating the real-time status of the depot based on the latest current status, and performing forward predictions on the vehicle return time, track availability, and grid power limitations within the prediction window; and executing the scheduling decision corresponding to the first time slot in each cycle.

3. The vehicle SOC optimal scheduling method according to claim 1, characterized in that, In step S2, the construction of unified status information specifically includes: collecting the vehicle's current SOC, the vehicle's estimated return time, and the vehicle's location information; collecting the departure schedule, line length, section energy consumption parameters, inter-station gradient information, and regenerative braking capability parameters of the line to which the vehicle belongs; collecting the track occupancy status of the depot, the availability status of charging piles, and the health status of power modules; and collecting the maximum power supply capacity of the power grid and its power limitation information that changes over time.

4. The vehicle SOC optimal scheduling method according to claim 1, characterized in that: When estimating the energy requirements for completing the next operational task, a regenerative braking energy prediction model is integrated. The regenerative braking energy prediction model predicts the energy that the vehicle can recover during braking based on the inter-station gradient information, the regenerative braking capacity parameters, and the vehicle's interval operating speed mode, and uses the obtained prediction value to correct the energy requirement estimation result.

5. The vehicle SOC optimal scheduling method according to claim 1, characterized in that: The battery characteristic constraints include the maximum charging power constraint of the battery, and the maximum charging power constraint is a variable that is dynamically adjusted according to the battery temperature conditions; in the process of solving the joint optimization model, the maximum acceptable charging power value is determined according to the real-time or predicted battery temperature of each vehicle, and the charging power allocation decision is made accordingly.

6. The vehicle SOC optimal scheduling method according to claim 1, characterized in that: In step S6, the generated vehicle entry allocation instruction, charging instruction, and online release instruction are issued in stages. The vehicle entry allocation instruction is first issued to the vehicle depot signal control system to arrange the route. The charging instruction is then issued to the corresponding charging pile controller to start the charging process. The online release instruction is issued to the vehicle and exit signal control systems when the vehicle has finished charging and meets the departure conditions.

7. The vehicle SOC optimal scheduling method according to claim 1, characterized in that, In step S6, after each rolling cycle, the process of step S5 is re-executed based on the latest status, corresponding to at least one of the following triggering conditions: the actual return time of a vehicle is detected to be earlier or later than the predicted value; the actual congestion level in the vehicle segment exceeds the preset congestion threshold; a charging pile malfunctions or its performance degrades; the power limit issued by the grid side undergoes a sudden change exceeding the preset power threshold; the operation plan is modified.

Images