A method for resource allocation and collaborative scheduling of aerial mobile charging robots
By establishing a track-space-time resource model and detecting directed waiting graphs, the driving and charging operation sequence of the vehicle is dynamically adjusted, solving the deadlock and resource utilization problems in the aerial mobile charging robot system, and realizing efficient resource collaborative scheduling and power supply capacity collaboration.
Patent Information
- Application Number
- CN202611514973.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-28
- Publication Date
- 2026-11-17
AI Technical Summary
Existing aerial mobile charging robot systems are prone to deadlock when multiple vehicles share a straight track, cannot effectively coordinate charging demand and track resources, and fail to adapt to dynamic adjustments under uncertain conditions.
By establishing a track spatiotemporal resource model, the earliest reachable time and unified time window of the vehicle are calculated, directed cycles in the directed waiting graph are detected, the vehicle's driving and charging operation sequence is dynamically adjusted, and the scheduling scheme is optimized to avoid deadlock and power overload.
It eliminates deadlock on straight tracks without the need for circularization or additional hardware, ensuring the vehicle's physical reachability, improving resource utilization, adapting to uncertainties during charging, and reducing waiting time and overload risk.
Smart Images

Figure CN123104448A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of charging resource scheduling technology for aerial mobile charging robots, and in particular to a method for resource allocation and collaborative scheduling of aerial mobile charging robots. Background Technology
[0002] With the increasing number of electric vehicles, the demand for charging in parking lots and other similar scenarios is growing. To reduce the land occupation and idleness of fixed charging piles and improve the utilization rate of charging facilities, an aerial mobile charging robot system has emerged. This system consists of a track erected in the air along the parking space arrangement, on which multiple trolleys carrying power supply compartments run. The track is equipped with a sliding contact line for external high-voltage power supply. The trolleys are equipped with current collectors that contact the sliding contact line to draw power during movement and operation. A charging gun is fixedly installed at the track position corresponding to each parking space. After the trolley moves to the target charging gun position, the power supply end of its power supply compartment connects with the charging gun, and the user then uses the charging gun to charge the vehicle. After charging is completed, the trolley disconnects and moves to the next task position or returns to the parking area at the end of the track.
[0003] In existing track-based or aerial shared charging devices, multiple charging terminals (mobile vehicles) are usually shared along a track. Their scheduling often adopts static heuristic rules such as "nearest distance and in an idle state" or "nearest idle while taking into account load balancing" to move the charging terminal to the location of the vehicle to be charged to make contact with the power source. Some solutions use a circular track or add hardware mechanisms such as track changing and track splitting to change the running path of the charging terminal in order to avoid positional interference between different charging terminals.
[0004] In the broader field of rail transit scheduling, there are already some time window routing, conflict detection and deadlock avoidance methods. The general approach is to allocate occupied time windows for nodes or sections on the proposed path of the vehicle, perform time window insertion and overlap checks on candidate paths, eliminate conflicts by detouring or re-selecting routes on two-dimensional road networks or tracks with bypasses and buffers, or perform loop deadlock detection and control on unidirectional guide networks during vehicle operation.
[0005] However, in scenarios where multiple cars share the same straight track and multiple parking spaces simultaneously initiate charging requests, the aforementioned existing technology has at least the following shortcomings: Firstly, the static allocation rule does not need to consider the time and space occupation of multiple cars on the track section, nor does it need to consider the time sequence of charging tasks; however, cars on a straight reciprocating track cannot go offline, cannot go lateral, and cannot overtake. Once multiple cars are traveling towards each other and there is no available passing zone in between, a blocking situation is formed, and multiple cars may also form a deadlock of cyclical waiting, which can only be resolved by manually towing or reversing the cars.
[0006] Secondly, existing time-window routing and deadlock avoidance methods characterize the short-term occupation of vehicles during "passage and release," where vehicles have alternative paths, can detour, or reroute during operation. However, the dwell time of the aerial mobile charging robot at the charging gun position is equal to the user's actual charging time. During the docking operation, the entire vehicle occupies the track, preventing other vehicles from crossing, effectively cutting the single-line track into two unreachable ends during that period. If allocation is still based on proximity, vehicles may be assigned to charging guns on the other side of the track, but these guns are continuously blocked by the operating vehicles before the user expects to leave, resulting in invalid assignments or even chain deadlocks. Existing "passage-type" time-window models cannot characterize the topological fragmentation caused by such long dwell times.
[0007] Third, the scheduling method for mobile charging robots is fragmented, with demand-side parameters (expected charging time, expected departure time, charging location) and motion-side resources (track section, passing zone) and power supply-side resources (sliding contact line power supply section and its power capacity) being isolated from each other. It does not form a unified collaborative allocation model, nor does it consider rolling adjustments under uncertain conditions such as users ending charging early, extended operation, robot malfunction, and power supply fluctuations.
[0008] Therefore, for application scenarios where a small number of trolleys carrying power supply compartments serve multiple fixed charging gun positions on a linear reciprocating aerial track, a multi-trolley resource allocation and collaborative scheduling method is needed that can eliminate opposing conflicts and cyclic waiting deadlocks before the scheduling scheme is issued and executed, and coordinate charging demand, track segment time and space occupation, and allow for the allocation and collaborative scheduling of resources and power capacity of the sliding contact line segments. Summary of the Invention
[0009] To address the aforementioned shortcomings, the present invention aims to propose a resource allocation and collaborative scheduling method for aerial mobile charging robots.
[0010] To achieve this objective, the present invention adopts the following technical solution: A resource allocation and collaborative scheduling method for an aerial mobile charging robot is applied to an aerial mobile charging robot system. The system includes a track for parking spaces, a sliding contact line laid on the track, multiple trolleys running along the track and carrying power supply compartments, and charging guns corresponding to the parking spaces and fixedly installed on the track. The trolleys draw power through current collectors and contact the sliding contact line, and the power supply compartments connect with the charging guns at the target charging guns for power supply. The method is executed by a scheduling server and includes the following steps: Step S1: Obtain the segment division of the track and the capacity of each segment, the location of each charging gun, the location and capacity of the passing zone, the power capacity of each power supply segment, and collect the location, running direction, working condition and power supply capacity of each car to establish a track spatiotemporal resource model. Step S2: Receive charging requests, determine the target charging gun, required power or expected charging duration and expected departure time for each charging task, and calculate the charging operation duration, deadline and time slack for each charging task. Step S3: Record the trolley in the docking charging operation as the blocking point, determine the connectivity of the track at different times based on the operation time window of the blocking point, calculate the earliest reachable time of the candidate trolley to the target charging gun, screen out feasible trolley and task candidate pairs, and obtain candidate scheduling schemes. Step S4: For the candidate scheduling scheme, along the sequence of travel segments from the current position of the vehicle to the target charging gun, the movement window of the vehicle to the segment and the operation window at the target charging gun are respectively deduced, and during the operation window, the blocking mark is set on the segment occupied by the vehicle to obtain the unified time window of each segment. Step S5: Based on the unified time window, detect the passage conflict between the vehicles, establish a directed waiting graph with the vehicles as nodes and the waiting relationship between the vehicles as directed edges, detect the directed cycles in the directed waiting graph to determine whether there is a deadlock in the scheduling scheme, and verify whether the power of simultaneous operation in the same power supply segment exceeds the limit. Step S6: For detected traffic conflicts or directed loops, determine the passing order of the yielding point and the yielding car, the critical section and the departure time of the relevant cars, or organize the tasks in batches according to the direction of travel, and return to step S4 to re-determine after correction, until the scheduling scheme has no traffic conflicts, no directed loops and the power does not exceed the limit. Step S7: With the goal of minimizing the weighted sum of user waiting costs, departure default costs, empty vehicle running costs, yielding waiting costs, unbalanced vehicle load costs, and power supply segment power limit exceeding costs, under the constraints of unique task assignment, mutually exclusive time windows of segments, dynamic reachability, operation duration, and power supply segment power capacity, the assignment relationship between tasks and vehicles, the task sequence of each vehicle, and the departure time are determined to obtain a collaborative scheduling scheme. Step S8: The collaborative scheduling plan is sent to each vehicle. The sent content includes the target charging gun, departure time, driving direction, planned time window for each section, passing point and waiting time, docking and operation instructions, and the actual operation information of the vehicle is collected and compared with the planned time window. Step S9: When any of the following triggering events are detected, such as a new charging request, early termination or extension of the charging operation, vehicle malfunction, deviation of the actual trajectory from the planned time window exceeding the threshold, or power fluctuation in the power supply segment, the time window already executed within the frozen time domain is frozen, and steps S3 to S7 are re-executed for tasks not executed outside the frozen time domain; wherein, when the charging operation ends early, the corresponding blocking point is removed, track connectivity is restored, and the earliest reachable time of the candidate vehicle is updated.
[0011] Preferably, step S1 includes: Step S11: Establish a one-dimensional coordinate along the track length and discretize the track into an ordered set of segments. Record the length, midpoint coordinates, and capacity of each segment. Only one car is allowed to occupy a normal segment at any given time. Two cars are allowed to pass each other or avoid each other in the passing area and the parking area at the end of the track. The capacity is 2. Each charging gun corresponds to a parking space and its coordinates and the segment to which it belongs are recorded. The sliding contact line is divided into several power supply segments along the track. Record the set of segments covered by each power supply segment and its power capacity. Step S12: Record the vehicle length, rated operating speed, rated output power of the power supply compartment, docking time for the power supply compartment to dock with the charging gun, and evacuation time for the evacuation channel after charging is completed for each vehicle; when the vehicle is located in any section, determine the set of sections actually covered by the vehicle according to its vehicle length and the coordinates of the midpoint of each section. Step S13: Periodically collect the real-time status of each vehicle. The real-time status includes at least the vehicle's current position, running direction, working condition, power supply compartment charge status and expected idle time. The working condition includes three states: idle, moving and docking operation. Step S14, the power supply compartment is either energy storage type or direct supply type: for energy storage type power supply compartment, the available power is determined by the product of its state of charge and energy storage capacity; for direct supply type power supply compartment, its rated output power is used as the available power supply capacity.
[0012] Preferably, step S2 includes: Step S21: Represent each charging request as a charging task and record its task parameters. The task parameters include the target charging gun, request time, required power, default expected charging duration, default expected departure time, and task priority. The vehicle, the charging gun, and the power supply compartment negotiate to determine the charging power of the task, and the charging power does not exceed the rated output power of the power supply compartment and the rated power of the charging gun. The planned charging time for charging task j is calculated using the following formula: ; When the user gives the desired charging time At that time, the planned charging time is taken ; This indicates the planned charging operation time for charging task j, which is the duration for which the vehicle is connected to the power supply at the charging gun position. This indicates the charging power negotiated and determined by charging task j. This indicates the user's stated expected charging time; This indicates the amount of electricity required for charging task j; The required electricity is calculated according to the following formula: ; , These represent the user's target state of charge and the initial and state of charge when initiating a charging request, respectively. This indicates the rated quantity of the power battery of the vehicle to be charged. Step S22: When the task does not have a given expected exit time, the task's deadline is calculated according to the following formula: ; Indicates the expected departure (charging complete) deadline for charging task j; Indicates the moment when the user initiated the charging request; This indicates the upper limit of the longest waiting time promised by the charging station. Step S23: Calculate the time relaxation of task j at time t according to the following formula. : ; Step S24: Determine the waiting cost and departure timeout cost per unit time according to the following formulas: ; ; , Let $\mathbf$ and $\mathbf$ represent the waiting cost for charging task j to start powering by 1 unit of time and the default cost for exceeding the deadline by 1 unit of time, respectively. This represents the task priority coefficient, which is determined by membership level, service commitment, or overtime compensation rules, and defaults to 1. This represents the penalty multiplier for timeout relative to waiting, and is a configurable coefficient greater than 1.
[0013] Preferably, step S3 includes: Step S31: Record the currently docking trolleys and the trolleys that are to be docked in the confirmed scheduling plan as blocking points to form a blocking point set; record the operation position, the set of sections covered by the vehicle body, and the operation time window defined by the start time and end time of the operation for each blocking point; within the operation time window of the blocking point, other trolleys are prohibited from occupying the section covered by the vehicle body. Step S32, define the connectivity between two points on the track at time t: when there is no blocking point at time t within the operation time window between the two points, the two points are determined to be connected and directly reachable; otherwise, the two points are determined to be blocked. The connectivity changes with time as the operation time window of the blocking point begins and ends. Step S33: For candidate vehicle r and task j, proceed sequentially from the current position to the target charging gun along the vehicle's travel direction: the arrival time of the vehicle traveling on the empty track to the next blocking point is calculated by dividing the travel distance by the rated speed; if the blocking point has not yet finished its operation when the vehicle arrives, the vehicle waits in front of it, updates the current time to the sum of the blocking point's operation completion time and the withdrawal time, and then continues to advance; taking into account the docking time at the target charging gun, the earliest achievable time when the candidate vehicle completes docking and can start power supply is recursively obtained; Step S34: When the sum of the earliest reachable time and the planned charging operation duration is not later than the task deadline, the corresponding vehicle-task pair is marked as a candidate feasible pair; otherwise, it is eliminated. In soft cutoff mode, only the target charging gun is required to be physically reachable, and its departure timeout is included in the optimization objective of step S7. Based on the condition that the available power is not less than the required power, or the available power supply capacity is not less than the charging power, candidate vehicles are eliminated.
[0014] Preferably, step S4 includes: Step S41: Determine the driving direction of the trolley based on the position of the target charging gun and the current position of the trolley, and determine a unique sequence of driving segments from the current position to the target charging gun on the one-dimensional track. Step S42: Starting from the departure time of the car, the time when the front of the car reaches the boundary of each section is calculated sequentially according to the length of each section and the rated speed of the car. The time from the arrival time of the front of the car to the departure time of the rear of the car is used to determine the movement window of the car for each section. The departure time of the rear of the car is included in the time required for the car body to pass through the section. Step S43: The earliest time when the candidate car reaches the target charging gun is taken as the start time of the operation, and the start time of the operation plus the planned charging operation duration is taken as the end time of the operation. The operation windows with the same start and end times are set for each section covered by the vehicle body at the target charging gun, and the blocking mark is set for the main section where the target charging gun is located. Step S44: Merge the movement window and operation window of the trolley to the section to form a unified time window for the trolley to the section; each scheduling scheme is represented as a set of unified time windows for all trolleys to all sections. Step S45, the unified time window set satisfies at least one of the following mutual exclusion requirements: The time when a following vehicle traveling in the same direction enters any section is no earlier than the sum of the time when the preceding vehicle leaves the section and the safe following interval; Two cars traveling in opposite directions are not allowed to intersect in a section with a capacity of 1; they are only allowed to meet in a passing zone with a capacity of 2. The movable windows of other trolleys must not intersect with the working windows of the docking trolley within its body coverage area.
[0015] Preferably, step S5 includes: Step S51: Count the number of four types of conflicts in the unified time window: the number of same-direction tracking conflicts that violate the safety tracking interval, the number of opposing intersection conflicts where opposing trolleys intersect in the window occupied by the trolley in the section with a capacity of 1, the number of operation cutting conflicts where the moving window and the operation window intersect in the vehicle body coverage section, and the number of power exceeding the power capacity when the power of the same power supply segment is operating simultaneously. Step S52: Establish a directed waiting graph with all cars as nodes: When a car is scheduled to enter a section that is occupied by another car and the one-dimensional driving section sequence is unique, making it impossible for the car to go around and it can only wait, add a directed edge from the waiting car to the car being waited for, and use the difference between the expected release time of the section by the car being waited for and the expected arrival time of the waiting car as the waiting time of the directed edge. Step S53: When there is a directed cycle consisting of directed edges in the directed waiting graph, it is determined that the scheduling scheme has a circular waiting deadlock and the scheme is infeasible; the directed cycle is detected by depth-first search or strongly connected component algorithm; the scheduling scheme is determined to pass the deadlock-free check only when the number of opposing intersection conflicts, the number of job cutting conflicts and the number of directed cycles are all zero. Step S54: For each power supply segment and each work period, sum the charging power of the docking trolleys within the coverage area of that power supply segment. The portion exceeding the power capacity of that power supply segment is recorded as the power exceeding the limit. If the power exceeding the limit is not zero, the power verification of the scheduling scheme is determined to be unsuccessful.
[0016] Preferably, step S6 includes: Step S61: For two cars that are in a head-on collision, calculate the sum of the travel time of each car to enter each passing zone and the time required to wait for the other car to pass in the passing zone as the passing cost of that car in the passing zone. Select the passing zone with the smallest passing cost as the passing point. Compare the minimum passing costs of the two cars. The car with the smaller minimum passing cost enters the passing point to give way and the other car passes first. Step S62: For multiple vehicles that need to pass through the same critical section, the passage order is determined in ascending order by using the time slack of the task served by each vehicle as the passage order key. If the time slack is the same, the passage order is determined in ascending order by the earliest reachable time. The departure time of the vehicle that is ranked later is postponed until after the sum of the time when the tail of the preceding vehicle leaves the critical section and the safe tracking interval. Step S63: For tasks that need to cross the same bottleneck section or the same work cut point within the rolling time domain, organize them in batches according to the direction of the vehicle's travel. First, release the task string in one direction of travel, and then release the task string in the opposite direction of travel. The direction switching position is arranged in the passing area or the end storage area. The batching scheme is determined by the minimum weighted sum of the number of direction switching in the time domain and the departure timeout increment caused by batching. In step S64, after each completion of point assignment, passage order rearrangement, departure time postponement, or travel direction batching, return to step S4 to re-analyze all unified time windows, and then execute step S5 for re-checking; when the number of opposing intersection conflicts, the number of job cutting conflicts, and the number of directed loops are all zero, and the same-direction tracking conflicts and power exceeding the limit are acceptable, a deadlock-free feasible solution is obtained; set a maximum number of iterations, and candidate solutions that still do not meet the requirements after exceeding this number of iterations are determined to be infeasible, and penalties are imposed or they are discarded in step S7.
[0017] Preferably, step S7 includes: Step S71, establish the following decision variables: assignment variable indicating whether a task is assigned to a car, sequence variable indicating the order in which the same car undertakes two tasks, departure time and start and end time of each car, passage sequence variable indicating the order in which two opposing cars pass through the critical section, and yield variable indicating whether a car enters the yielding zone to yield. Step S72, with the optimization objective of minimizing the weighted sum, the weighted sum includes: the weighted sum of waiting costs for all tasks from the request time to the actual start time of power supply, the weighted sum of departure default costs for all tasks whose work completion time exceeds the deadline, the cost of the trolley's empty driving time, the total waiting time for the trolley to give way in the passing zone, the load imbalance penalty for the deviation of each trolley's task volume from the average task volume of all trolleys, and the power limit violation penalty for power supply segments; wherein the weights of the departure default cost and the deadlock risk item are taken as the relatively larger values; Step S73, the optimization satisfies the following constraints: Assignment and vehicle sequence constraints: Each task is assigned to exactly one vehicle; multiple tasks undertaken by the same vehicle are executed sequentially, and the start time of the next task is no earlier than the end time of the previous task plus the empty travel time to the charging gun of the next task. Opposing segment passage order constraint: For two cars traveling in opposite directions and needing to pass through the same segment with a capacity of 1, set a segment passage order variable to determine which car passes first, and require that the time when the car that passes later enters the segment is not earlier than the sum of the time when the car that passes first leaves the segment and the safety interval, so that the occupancy windows of any two opposing cars on the segment with a capacity of 1 do not intersect. Power supply constraints: During any work period, the total charging power of the trolleys operating simultaneously in the same power supply segment shall not exceed the power capacity of that power supply segment; the total electricity demand of the tasks undertaken by the energy storage type power supply compartment shall not exceed the sum of its available electricity and the electricity replenished during the journey; the charging power of the tasks undertaken by the direct power supply compartment shall not exceed its rated output power. Operation duration and passing zone capacity constraints: The difference between the end time and start time of each task shall not be less than its planned charging operation duration; the number of vehicles in the same passing zone at the same time shall not exceed the capacity of the passing zone. Step S74, using a two-stage solution: In the first stage, the tasks to be assigned are arranged in ascending order of time slackness, and in descending order of task priority when the time slackness is the same. The tasks are then inserted into the task sequence of the candidate cars according to the arrangement order. The insertion result that minimizes the weighted sum increment and is verified as deadlock-free and feasible through steps S4 to S6 is selected to generate an initial feasible scheduling scheme. Phase 2 involves solving the initial feasible scheduling scheme using tabu search. The neighborhood operations include at least one of task reassignment, reordering of tasks in the same vehicle, changing passing points, and batch redrawing of driving directions. Neighborhood solutions containing deadlocks are penalized based on the number of directed cycles. The process stops when the maximum number of iterations or the computation time limit is reached, and the cooperative scheduling scheme is output.
[0018] Preferably, step S8 includes: Step S81: The collaborative scheduling scheme is parsed and sent to each vehicle. The sent content includes: the vehicle's task sequence, target charging gun, departure time and driving direction, planned time window for each section, passing points and waiting time at passing points, power supply connector docking and disconnection instructions, and estimated start and end times of the operation; and the estimated start time and estimated completion time are pushed to the corresponding user. In step S82, each vehicle periodically reports its actual position, running speed, operating condition, and charging power. The scheduling server verifies the actual entry and exit times of each vehicle into and from each segment and compares them with the planned time window for that segment, recording the time deviation. The time deviation serves as the basis for determining the trajectory deviation trigger event in step S9.
[0019] Preferably, step S9 includes: Step S91: Advance the scheduling time domain according to the rolling cycle, and start rescheduling when any of the following triggering events are detected: new charging request, early termination of charging operation, extension of charging operation, trolley failure, deviation of the actual entry or exit time of the trolley from the planned time window exceeding the preset deviation threshold, and power fluctuation of the power supply segment. During rescheduling, a frozen time zone is set. The time window of the work vehicle that is already on the way or docked within the frozen time zone remains unchanged. Only tasks that have not been executed and idle vehicles outside the frozen time zone are re-executed through steps S3 to S7. The rescheduling plan is then issued through step S8. Step S92: When the charging operation ends early, update the end time of the operation to the current time, remove the corresponding car from the blockage point set, and recalculate the track connectivity and the earliest reachable time of the affected candidate cars; if the task that was originally blocked by the blockage point and was waiting becomes reachable or no longer timed out, immediately generate a new plan and notify the relevant cars to start early. Step S93: When a trolley malfunctions, it is treated as a blockage point before the expected recovery time. The end time of its operation or occupancy window is taken as the expected recovery time. If the expected recovery time cannot be estimated, a preset maximum value is taken and a manual intervention alarm is issued. The tasks that the malfunctioning trolley has not yet started are reassigned to other trolleys, and steps S3 to S7 are executed again. Step S94: When the power supply segment fluctuates, re-verify according to the power limit and the power supply constraint, and make staggered local adjustments to the start time of the affected operation within a unified time window, or arrange the relevant trolleys to wait briefly in the passing area or end storage area before docking.
[0020] One of the above technical solutions has the following advantages or beneficial effects: 1. Before the scheduling plan is issued, this invention uses unified deduction of segment time windows, directed waiting diagram directed loop detection, and one-dimensional yield correction to ensure that the number of opposing intersection conflicts, operation cutting conflicts, and directed loops in the plan are all zero, thus eliminating multi-car deadlock and opposing blockage from the mechanism. On straight reciprocating tracks, there is no need to build the track into a loop or add track changing or lateral movement mechanisms to achieve multi-car coordination, reducing hardware investment and the difficulty of station modification. 2. Dynamic topology cutting and time-dependent reachability calculation ensure that each assigned vehicle is physically reachable within the task time window, eliminating invalid assignments that are "unable to reach the gun position before the user leaves" from the source; sorting by time slack and participating in global assignment by weighting waiting and departure defaults ensures that time-critical tasks are prioritized, which helps reduce the departure timeout rate of high-urgent tasks and the average waiting time of remote parking positions. 3. The unified time window allows the "movement and passage occupation" of the trolley and the "long-term operation occupation" at the gun position to be uniformly and accurately arranged. Combined with the load imbalance penalty, a smaller number of trolleys can serve more fixed charging gun positions, avoiding the "uneven work and idle" of the trolleys and the long-term vacancy of the gun positions, thus improving the utilization rate of trolleys and track resources. 4. By using the power capacity constraint of the sliding contact line segment, the overload, voltage drop and protection tripping caused by multiple trolleys simultaneously connecting to the same power supply segment are avoided from the dispatching level. This supports maximizing the simultaneous service capacity under the existing external power capacity of the station and realizes the coordination between motion dispatching and power supply capacity. 5. When a user ends charging early, line connectivity is immediately restored, the previously waiting tasks are instantly awakened and can be dispatched ahead of time; when a vehicle malfunctions, the malfunctioning vehicle is instantly modeled as a dynamic blocking point, and its unstarted tasks are reassigned, thereby reducing cascading delays caused by a single event; when power fluctuates, the start time of the operation can be adjusted to avoid peak periods; adapting to uncertainties in the charging process. Attached Figure Description
[0021] Figure 1This is a schematic diagram of spatiotemporal resource modeling for an aerial mobile charging robot system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the overall process of resource allocation and collaborative scheduling for an aerial mobile charging robot, according to an embodiment of the present invention. Figure 3 This is a spatiotemporal diagram of a segmented time window that is unified by a moving window and a work window in one embodiment of the present invention, and is used for segmentation and cutting / allowing. Detailed Implementation
[0022] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.
[0023] In the description of embodiments of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of embodiments of the present invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0024] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0025] This application proposes a resource allocation and collaborative scheduling method for an aerial mobile charging robot, applied to an aerial mobile charging robot system, the structure of which is shown in the attached figure. Figure 1 As shown: A straight track is erected in the air along the direction of the parking spaces. Multiple mobile trolleys run on the track, and each trolley carries a power supply compartment. The track is laid with a sliding contact line and connected to a high-voltage power supply. The trolleys are equipped with current collectors. When the trolleys move or operate, they draw power by contacting the sliding contact line through the current collectors. A charging gun is fixedly installed at the track position corresponding to each parking space (the charging gun hangs down to the side of the parking space). The charging gun can be detachably connected to the power supply end of the trolley's power supply compartment.
[0026] The basic working process is as follows: After the vehicle parks in the parking space and submits a charging request, the dispatching system assigns a trolley carrying a power supply compartment to move along the track to the target charging gun position. The current collector contacts the sliding contact line to draw power, and the power supply end of the power supply compartment extends a charging connector to complete the power supply docking with the charging gun. The user then uses the charging gun to charge the vehicle. After charging is completed, the trolley disconnects from the docking and moves to the next task position or returns to the end storage area.
[0027] To facilitate algorithm modeling, the track is discretized into several segments along its length. Only one car is allowed to occupy a normal segment at any given time (capacity 1). Passing zones can be optionally set at the end of the track and in the middle of the track. Passing zones can accommodate two cars to meet or pass each other (capacity 2). The sliding contact line is divided into several power supply segments along the track. Each power supply segment has a power limit determined by the external high-voltage capacity and the cable / transformer capacity.
[0028] Specifically, such as Figure 2 As shown, a method for resource allocation and collaborative scheduling of an aerial mobile charging robot includes the following steps: Step S1: Obtain the segment division of the track and the capacity of each segment, the location of each charging gun, the location and capacity of the passing zone, the power capacity of each power supply segment, and collect the location, running direction, working condition and power supply capacity of each car to establish a track spatiotemporal resource model. Step S2: Receive charging requests, determine the target charging gun, required power or expected charging duration and expected departure time for each charging task, and calculate the charging operation duration, deadline and time slack for each charging task. Step S3: Record the trolley in the docking charging operation as the blocking point, determine the connectivity of the track at different times based on the operation time window of the blocking point, calculate the earliest reachable time of the candidate trolley to the target charging gun, screen out feasible trolley and task candidate pairs, and obtain candidate scheduling schemes. Step S4: For the candidate scheduling scheme, along the sequence of travel segments from the current position of the vehicle to the target charging gun, the movement window of the vehicle to the segment and the operation window at the target charging gun are respectively deduced, and during the operation window, the blocking mark is set on the segment occupied by the vehicle to obtain the unified time window of each segment. Step S5: Based on the unified time window, detect the passage conflict between the vehicles, establish a directed waiting graph with the vehicles as nodes and the waiting relationship between the vehicles as directed edges, detect the directed cycles in the directed waiting graph to determine whether there is a deadlock in the scheduling scheme, and verify whether the power of simultaneous operation in the same power supply segment exceeds the limit. Step S6: For detected traffic conflicts or directed loops, determine the passing order of the yielding point and the yielding car, the critical section and the departure time of the relevant cars, or organize the tasks in batches according to the direction of travel, and return to step S4 to re-determine after correction, until the scheduling scheme has no traffic conflicts, no directed loops and the power does not exceed the limit. Step S7: With the goal of minimizing the weighted sum of user waiting costs, departure default costs, empty vehicle running costs, yielding waiting costs, unbalanced vehicle load costs, and power supply segment power limit exceeding costs, under the constraints of unique task assignment, mutually exclusive time windows of segments, dynamic reachability, operation duration, and power supply segment power capacity, the assignment relationship between tasks and vehicles, the task sequence of each vehicle, and the departure time are determined to obtain a collaborative scheduling scheme. Step S8: The collaborative scheduling plan is sent to each vehicle. The sent content includes the target charging gun, departure time, driving direction, planned time window for each section, passing point and waiting time, docking and operation instructions, and the actual operation information of the vehicle is collected and compared with the planned time window. Step S9: When any of the following triggering events are detected, such as a new charging request, early termination or extension of the charging operation, vehicle malfunction, deviation of the actual trajectory from the planned time window exceeding the threshold, or power fluctuation in the power supply segment, the time window already executed within the frozen time domain is frozen, and steps S3 to S7 are re-executed for tasks not executed outside the frozen time domain; wherein, when the charging operation ends early, the corresponding blocking point is removed, track connectivity is restored, and the earliest reachable time of the candidate vehicle is updated.
[0029] In one embodiment, this method is deployed on the scheduling server (cloud or station control edge server) of the charging station, communicating with the track controller, vehicle-mounted controller, charging gun detection unit, user charging mini-program / APP, and power supply metering unit. It operates in a closed loop following the steps of "perception modeling—demand taskization—dynamic reachability determination—time window extrapolation—deadlock detection—give way correction—multi-objective optimization—execution—rolling rescheduling," with the overall process as follows: Figure 2 As shown. The core ideas of this method are threefold: First, the "movement and passage occupancy" of the trolley on the section and the "long-term parking occupancy for charging operations" at the gun position are uniformly represented as a section time window, and the trolley is regarded as a "dynamic blocking point" that cuts the straight track into multiple connected components during the operation period, and time-dependent reachability is calculated accordingly; Second, before any scheduling plan is issued, the plan is first simulated using section time windows and directed waiting graph loop detection is performed. Conflicts and deadlocks detected are corrected using yielding rules specific to one-dimensional tracks, forming an inner loop of "simulation-detection-correction" to ensure that the issued plan is deadlock-free; Third, within the deadlock-free feasible solution space, user waiting and departure default, empty running, yielding waiting, and load are considered.
[0030] Preferred, such as Figure 1As shown, step S1 includes: Step S11: Establish a one-dimensional coordinate x along the track, and discretize the track into an ordered set of segments. Record the length of each segment s. Midpoint coordinates and car capacity Among them, ordinary section The end parking area and the passing area are denoted separately as set Y, and the coordinates of the midpoints are recorded. With the capacity of the car , ; The set of fixed charging guns is denoted as Record the coordinates of each charging gun g. and its section Each charging gun corresponds to one parking space; The sliding contact line is divided into power supply segment sets. Record the set of segments covered by segment b. and its power capacity ; exist Figure 1 In the middle, the track is discrete into sections s1~s14 (the capacity of a normal section is 1, and only one car is allowed to occupy it at the same time); the charging guns G1~G6 are fixed for each car position; the passing area / end storage area has a capacity of 2, which can be used for oncoming vehicles to pass each other; the sliding contact line is powered in sections B1~B3, and each section has a power capacity limit.
[0031] Step S12, denote the set of cars as Record the length of each car r along the track, including the power supply compartment. Rated operating speed Rated output charging power Time spent connecting or disconnecting the power supply connector from the charging gun ; The car r is in any central section The set of sections actually occupied is as follows: --Formula (1); This indicates that the car r is stopped in the section. When it is in the center position, it is covered by its body and thus constitutes the entire set of occupied sections for other cars; Indicates the coordinates of the midpoint of segment s. Indicates the central section The center coordinates; This represents the length of the car body (including the power supply compartment) along the track direction of the trolley r; S represents the set of all track segments; Step S13: Periodically collect and maintain the real-time state vector of each vehicle. : --Formula (2); This represents the position coordinates of the trolley at time t; The value represents the working condition of the vehicle r. It is 0 when it is idle and parked, 1 when it is moving, and 2 when it is docking and charging. It indicates the current direction of the car's movement. It takes +1 when the one-dimensional coordinate x is increasing along the track, -1 when the one-dimensional coordinate x is decreasing, or 0 when the car is stationary. This indicates the charging state of the power supply compartment. If the power supply compartment does not contain energy storage and only draws power in real time from the sliding contact line via the current collector, then... Take null value; This indicates the time when the car r is expected to complete its current task and return to idle time, where the idle time is equal to the current time t; Step S14, for the power supply compartment containing energy storage, the amount of electricity available for subsequent missions is: --Formula (3-1); For power supply compartments that do not contain energy storage and draw power directly from the sliding contact line, their output capacity is limited by their rated power. --Formula (3-2); , These represent the available power and rated capacity of the energy storage power supply compartment, respectively. Indicates the current charge status of the power supply compartment; , These represent the available output power and rated output charging power of the direct-supply type power supply compartment, respectively.
[0032] Preferably, step S2 includes: the user scans a code at the parking space or initiates a request through the charging order system, and the system obtains the target charging gun for the task j. (Determined by parking space, corresponding coordinates) ), Section ) Request time Electricity demand User's expected charging time Expected departure time and task priority coefficient (Based on membership level) Service commitments or overdue compensation rules are determined (default option 1), and the charging power is determined through negotiation between the vehicle, charging gun, and power supply compartment. .
[0033] Step S21: Obtain charging task j based on user charging needs, and calculate the planned charging duration for task j. --Formula (4-1); When the user specifies the desired charging time --Formula (4-2); This indicates the planned charging operation time for charging task j, which is the duration for which the vehicle is connected to the power supply at the charging gun position. This indicates the charging power negotiated and determined by charging task j. This indicates the user's stated expected charging time; The required charge for charging task j is determined by the difference between the target charge reported by the vehicle's BMS and the vehicle's current charge, or by formula (5) based on the vehicle's battery capacity and target state of charge. --Formula (5); , These represent the user's target state of charge and the initial and state of charge when initiating a charging request, respectively. This indicates the rated quantity of the power battery of the vehicle to be charged. Step S22, calculate the deadline for charging task j to complete charging: --Formula (6); Indicates the expected departure (charging complete) deadline for charging task j; Indicates the moment when the user initiated the charging request; This indicates the upper limit of the longest waiting time promised by the charging station. Step S23, define the time relaxation of charging task j at time t: --Formula (7); This represents the relaxation time of charging task j at time t, which is the remaining time before the deadline that can be used for waiting and travel after deducting the necessary charging operation time. The smaller, the more urgent. This indicates that even if power is supplied immediately, the timeout will still occur; Step S24, define the waiting cost per unit time and the exit timeout cost: --Formula (8-1); --Formula (8-2); , Let $\mathbf$ and $\mathbf$ represent the waiting cost for charging task j to start powering by 1 unit of time and the default cost for exceeding the deadline by 1 unit of time, respectively. This represents the task priority coefficient, which is determined by membership level, service commitment, or overtime compensation rules, and defaults to 1. This represents the penalty multiplier for timeout relative to waiting, and is a configurable coefficient greater than 1.
[0034] Preferably, step S3 is the key difference between this method and general time-window routing. Specifically, the trolley performing the task is a "dynamic blockage point" on the straight track during its operation period. The track topology is segmented over time, and the reachability of candidate trolleys to the task and their earliest reachable time must be calculated based on this time-varying connectivity. Specifically, this includes: Step S31: Summarize the vehicles that have already been docked for work, as well as the vehicles that are scheduled to be docked for work in the future, to form a set of blocking points A. For each blocking point a, i.e., a working vehicle, record its working position. The set of sections covered by the vehicle body With the task time window : Blocking period --Formula (9); Step S32, define two points x and y on the orbit at time t. Connectivity function between: --Formula (10); When the interval There is no content that satisfies If the blocking point is a, then If it exists, then ; This indicates connectivity; a value of 1 indicates that the track between two points at time t is unobstructed and directly accessible, while a value of 0 indicates that it is blocked by the work trolley. , The coordinates of any two points on the track can be used to represent the current position of the vehicle and the position of the target charging gun. , These represent the times when the docking point (operation trolley) begins and ends, and when the evacuation route is completed; Step S33: For a candidate vehicle r, when it completes charging task j, it moves from its current position along the direction of travel of vehicle r. To the target charging gun location Passing through the blocking points in sequence The candidate vehicle r moves segment by segment. When it encounters a blockage point where the operation has not yet been completed, it waits in front of it until the blockage point finishes its operation and leaves the passage. The earliest time that candidate vehicle r can complete charging docking and start supplying power is determined by this process. The following formula can be used for recursion: , --Formula (11-1); --Formula (11-2); , --Formula (11-3); --Formula (11-4); , These represent the "current departure time" and "current position" in the recursive process, respectively, with initial values being the scheduling time. Current position of the car ; This indicates that the car has traveled along the empty track to the location of the next blocking point 'a'. The moment; This indicates the time when the operation at point a was completed and the personnel were evacuated, and the passage was cleared again. This represents the time it takes for the work vehicle represented by the blocking point a to detach and move to a position that does not obstruct passage. The max function means that if the vehicle arrives first, it waits, and if it arrives later, it can pass directly. This represents the rated operating speed of the candidate vehicle r; This indicates that candidate car r has arrived at the charging gun location where charging task j is located. And the earliest time when power supply connection is completed and power can be started; This indicates the time it takes for the power connector of the power supply compartment of candidate vehicle r to extend out of the charging gun and complete the docking. Step S34, for each candidate team of "candidate car r-charging task j" Give the physical reachability and time feasibility flags: If and only if (Hard cutoff scenario) -- Formula (12); Indicates candidate pair The feasibility flag is set to 1, indicating that the candidate vehicle r can complete the task before the deadline of charging task j. This indicates the estimated time when power supply will be completed if candidate car r undertakes the charging task j. This indicates the deadline for charging task j. For stations with only a soft deadline, this can be relaxed to physical reach. The timeout amount is charged as a penalty for breach of contract in step S7.
[0035] In addition, the vehicle's own power supply capacity must meet the mission requirements: energy storage type needs to... (Or include energy replenishment during the time domain), direct supply type requires Candidate pairs that do not meet the requirements are directly eliminated.
[0036] Preferred, such as Figure 3 As shown, in step S4, for any candidate scheduling scheme (i.e., "task-vehicle" assignment and the task sequence of each vehicle), the occupancy time window in each segment is calculated for each vehicle. The key to this step is to unify the movement occupancy and charging operation occupancy into a segment time window. Specifically, as follows: Step S41, (Travel Segment Sequence): For any candidate pair If the path of the car r on the track from its current position (or the previous mission gun position) to the mission gun position j is unique, then the travel direction and travel segment sequence of the candidate car r are defined as follows: , --Formula (13); This indicates that the candidate car r starts from the current position. To the charging gun location corresponding to charging task j The direction of travel, when hour, ,when hour, ; This indicates that when candidate car r is performing charging task j, it starts from the current segment. To the target charging gun location section An ordered sequence consisting of all the segments traversed in sequence; Step S42, (Movement Time Window): Define the movement occupancy window of the candidate car r for segment s from the moment the car head enters the boundary of segment s until the moment the car leaves the boundary of segment s: ( The initial value is taken )--Formula (14); --Formula (15); This represents the time when the front of candidate car r reaches the boundary of segment s, with the initial value being... , This indicates the planned departure time of candidate car r; Indicates that the candidate car r passes through a distance of 100 km. The required travel time for section s; , These represent the start (entry) and end (departure) times of the candidate car r's movement window in section s, respectively. This represents the total time taken from the moment the candidate car r enters section s to the moment it leaves the parking space in section s, and is included in the length of the candidate car r. ; Step S43 (Charging operation time window and cutting mark): After the candidate vehicle r completes docking at the charging gun position, it begins to supply power and continuously occupies the set of sections near the charging gun position that are covered by the vehicle body during the operation. and the main section of the charging gun position Set blocking marker: , --Formula (16-1); , --Formula (16-2); , --Formula (16-3); , These represent the times when candidate car r starts and ends power supply for charging task j, respectively. This represents the window of operation occupied by candidate car r on segment s during the operation period, for the set of segments covered by the car body. Each segment within the range is valid; This indicates that the candidate car r is at the charging gun position. The set of sections covered by the vehicle body is obtained by formula (1); The blocking flag indicates that the segment s at time t is blocked by the work vehicle, and other vehicles are prohibited from crossing it. It is the expression of the topology cutting in step S3 at the time window level. Step S44, (Unified Segment Time Window): Set all time windows occupied by candidate car r in segment s. Positioned as the union of the mobile occupied window and the job occupied window: --Formula (17); Step S45, (Occupancy Mutual Exclusion Constraint): On the same segment, the time windows of different vehicles must meet the same-direction tracking interval, opposite-direction intersection constraint, and operation cutting constraint: --Formula (18); when and hour, --Formula (19); For any Its moving window and the work window are in The upper parts do not intersect -- Formula (20); Where: r represents the following vehicle traveling in the same direction, Indicates the vehicle ahead traveling in the same direction; This indicates the minimum safe following distance when traveling in the same direction; , These represent the driving directions of the two vehicles, which are opposite in direction and must not overlap when the segment capacity is 1. This represents the capacity of the car in section s, with 1 for a normal section and 2 for a passing section. , These represent the following vehicle (r) and the preceding vehicle traveling in the same direction, respectively. The collective term for any occupied window (moving occupied window or operation occupied window) on segment s.
[0037] Preferably, step S5 includes: Step S51, (Conflict Score Count): Based on the time window set derived from step S4. The set of all time windows occupied by candidate car r for segment s. Statistical analysis of four types of conflicts: --Formula (21); This represents the set of all time windows occupied by candidate car r for segment s. Statistical analysis of four types of conflicts was conducted. This represents the number of tracking conflicts in the same direction, that is, the number of times when the window occupied by the following vehicle and the window occupied by the preceding vehicle do not meet the safety interval of formula (18); This represents the number of head-on collisions, which is the number of times two vehicles traveling in opposite directions occupy overlapping windows in a segment with a capacity of 1, i.e., the number of times formula (19) is violated; This indicates the number of job cutting conflicts, that is, the number of times the moving occupancy window of other trolleys intersects with the job occupancy window of a certain job vehicle in its coverage area, that is, the number of times formula (20) is violated; The power over-limit number refers to the number of time periods in which the power of simultaneous operation exceeds the capacity within the same power supply segment, which is determined by formula (24). Step S52 (Directed Waiting Graph): Construct a directed waiting graph with all cars as nodes. When car r needs to enter segment s as planned, and segment s is blocked by car r during that time period... The path is occupied and unique, preventing car r from detouring; it can only wait for the other car. Add a directed edge upon release. The edge weight represents the expected waiting time: , If and only if make and There is a conflict and r has no alternative path. --Formula (22); Let V represent a directed waiting graph, where V and E represent the set of nodes (all cars) and the set of directed edges of the directed waiting graph, respectively. This means that car r must wait for car r. Release its necessary passage and wait for the dependency to point from car r to car. ; This indicates the expected waiting time for the waiting relationship; This indicates the estimated time when car r will arrive at segment s; Indicates a small car The expected release time for segment s; Step S53 (Deadlock Detection): Set the deadlock judgment criteria for the directed waiting graph: If there is a directed cycle in the directed waiting graph, then the cars on the cycle form a circular waiting, which indicates that the solution is deadlocked, as shown in formula (23): ; --Formula (23); Indicates the number of deadlocks in the solution; Directed cycles can be detected in linear time using existing techniques known to those skilled in the art, such as depth-first search (DFS) or Tarjan's strongly connected component algorithm. The feasibility criterion for the proposed solution is: , , (Same-direction tracking conflicts and power limit violations can be resolved in step S6 by delaying departure and staggering peak hours).
[0038] Step S54, (Segmented Power Exceeding Limit): Calculate the sum of the power of the trolley operating simultaneously and the amount of power exceeding the limit. , --Formula (24); This indicates that the power supply segment b exceeds the limit during time period t. A value of 0 indicates that the limit is not exceeded. b represents the number of the power supply segment, and t represents the working period, which is the discrete time period in the time window simulation. This indicates the charging task j currently undertaken by the vehicle r; The set of sections covered by the vehicle body when the trolley r docks at the target gun position of its charging task is represented by formula (1); This represents the indicator function, which takes the value 1 when the condition is true and 0 otherwise. It is used to determine whether the trolley r is in the working state at time t. These represent the sets of sections covered by power supply segment b. This indicates that the location occupied by the trolley falls within the power supply range of segment b; This indicates the upper limit of the power capacity of power supply segment b; This represents the charging power of task j served by vehicle r; The positive operator is defined only if the sum of the power of simultaneous operations exceeds [a certain value]. Record positive values that exceed the limit, and record 0 values that do not exceed the limit.
[0039] For the conflicts and directed cycles detected in step S5, corrections are made based on the characteristics of a one-dimensional track: "overtaking is not allowed, and resources are limited." After correction, the process returns to step S4 for re-analysis to form a closed loop. Specifically, step S6 includes: Step S61 (Given point selection and driving determination): For a pair of opposing conflicting vehicles r and The candidate will be the point of reference for the capacity of the car between the two cars. Calculate the total cost for car r to avoid area y when it enters the area: --Formula (25-1); --Formula (25-2); --Formula (26); This represents the total time cost for car r to enter and for area y to complete the avoidance maneuver. This represents the one-dimensional coordinate of the region (or point) that will be affected; This indicates that the opposing collision vehicle r and The coordinates of the expected intersection and conflict point; This indicates the rated operating speed of the trolley r; This represents the travel time of the car r in area y when it retreats from or enters the area before the conflict point, calculated by dividing the absolute value of the coordinate difference by the rated operating speed. Indicates the opposing car The estimated time at which the passage will be cleared again will be determined by the passage y. Indicates a small car Waiting for oncoming traffic in the designated area. This will allow for the waiting time required to clear the area and passageway; Y represents the passing area assembly, including the passing area in the middle of the track and the end storage area; This indicates the point where the car r will yield to the obstacle with the lowest cost of avoiding it. The vehicle with the smaller cost of entering its optimal yield point is determined to yield, and the other vehicle has priority to pass. This indicates two cars that collided while traveling in opposite directions. This represents the minimum total cost for car r to complete the avoidance at its respective optimal passing point; argmin means that among two conflicting cars, the one with the lower "minimum yielding cost" is chosen as the yielding car. The other party has priority to pass; Step S62, (Critical Section Passage Sequence and Departure Correction): For critical sections with a capacity of 1 that must pass through sequentially. Multiple vehicles will proceed in the order determined by their task slack. If the slack is the same, the vehicle that arrives earliest will proceed first, and vehicles that arrive later will either have their departure delayed or wait in the passing area. (Ascending order, urgent items take precedence, in case of ties) Ascending order)--Formula (27-1); --Formula (27-2); The sorting key for the trolley r passing through the critical section is given by formula (7), which determines the relaxation level it serves. The smaller the value, the higher the priority; This indicates the charging task currently undertaken by the vehicle r; The time slack of task j(r) at time t is represented by formula (7), which is the deadline minus the current time and then the task duration. Take this value and sort it in ascending order. The trolley with the smaller the slack (the more urgency) will pass first. This represents the earliest reachable time when the trolley r arrives at the target gun position and completes docking; when relaxations are tied, this time is used as the ascending order. This indicates the departure time of car r; Indicates the leading car The rear of the car leaves the critical section The moment; The safe following interval obtained by formula (18) means that the following vehicle can only enter after the preceding vehicle has cleared the critical section and left a safe interval; This indicates that the departure time of the following train is the later of the two: the "originally scheduled departure time" and the "time when the preceding train clears the critical section and leaves a safe distance". Step S63 (Directional Batching): For charging tasks that need to cross the same bottleneck section or the same work cut-off point within the rolling time domain H, organize them in batches according to the driving direction. First, release the task string in one direction, and then release the task string in the opposite direction. The batch switching is arranged to be completed in the passing zone or the end storage area to minimize the number of direction switching and the resulting default increment. --Formula (28); This indicates the number of times the driving direction changes at the bottleneck section within the rolling time domain H. Each change requires a change of direction. This represents the weighting coefficient of the default increment relative to the number of switching operations in the batch segmentation target; This indicates the timeout amount for charging task j due to batch sorting. Step S64 (Iterative Closed Loop): Each time the point assignment, passage sequence rearrangement, departure time postponement or direction batching is completed, the process returns to step S4 to re-analyze all time windows, and then step S5 is used for review, and this process is repeated. when Furthermore, when the number of collisions and power consumption in the same direction are acceptable, a deadlock-free feasible solution is obtained; however, when the maximum number of iterations is exceeded... Candidate solutions that are still infeasible are judged as infeasible solutions and are penalized or discarded in S7.
[0040] Step S6 is for one-dimensional common track with no alternative path. Before the solution is issued, deadlock is eliminated by re-deductive simulation with time window using special rules of "comparison of yield point costs - relaxation passage sequence - direction batching". Compared with the existing technology that relies on the unidirectional circulation of the circular track to naturally avoid the opposing conflict, this step is applicable to the straight reciprocating track. There is no need to build the track into a circle, and there is no need to add a track changing mechanism in the hardware.
[0041] Preferably, step S7 includes: performing joint optimization on "task-vehicle assignment, task execution order, departure time, and yielding decision" within the deadlock-free feasible solution space.
[0042] Step S71, (Decision Variables): For each deadlock-free feasible solution, define the decision variables: This indicates whether the charging task j is assigned to the car r, with 0 indicating no assignment and 1 indicating assignment. This indicates whether the car r should execute the charging task j1 before executing j2; 0 indicates no, and 1 indicates yes. , , These are represented, in order, as departure time, start time of operation, and end time of operation; Indicates the opposing car r, The order in which the trolley r passes through the critical section s is determined by h: 1 indicates that the trolley r passes through first, and 0 indicates that the trolley r passes through first. Pass it first; This indicates whether car r will be forced to yield to area y if it enters the area; 0 indicates no, 1 indicates yes. Step S72, (Objective Function): Define the objective function with the goal of minimizing the weighted sum of risks of user waiting, departure default, empty running, yielding waiting, load imbalance, power overrun, and deadlock: --Formula (29); Z represents the total cost (weighted sum) of the optimization objective, and min represents the scheduling scheme that minimizes it; j, r, b, and t are the numbers for the charging task, the vehicle, the power supply segment, and the working period, respectively. This represents the weighted waiting cost for all users from the time the request is initiated until power is supplied. Obtained from formula (8-1), This indicates the actual start time of power supply for charging task j. Indicates the moment the user initiated the request; This represents the weighted average cost of default for all tasks exceeding the expected departure time. Obtained from formula (8-2), Indicates the end time of task j. Indicates the user's desired departure time; This represents the energy consumption / time cost of the car idling. This represents a 0-1 assignment variable, where 1 is taken when task j is assigned to car r, and 0 is taken otherwise. This represents the idle travel time of the vehicle r when it undertakes the charging task j; This indicates that suppressing unnecessary things will... This represents the total yielding waiting time that the vehicle r is scheduled to take in step S6; This indicates a penalty for unbalanced load. , These represent the cumulative workload (operation time or mileage) undertaken by car r in the time domain and the average workload of all cars, respectively. This indicates that a soft penalty term is provided when power is not treated as a hard constraint; This indicates that the segmented power limit in formula (24) is exceeded; The term represents the number of directed cycles in formula (23). This term is 0 for deadlock-free feasible solutions, which is used to ensure that the optimization search tends to a deadlock-free solution. This represents the weight coefficient for each item; Step S73, (Constraints): Implement sequential constraints on vehicle assignment and charging tasks, ensuring that each charging task is undertaken by exactly one vehicle, and tasks on the same vehicle are executed sequentially. --Formula (30-1); --Formula (30-2); This indicates that charging task j must be assigned to one and only one vehicle. This means that the constraint holds for all charging tasks j; This is a 0-1 sequence variable representing the order in which car r executes task j1 before task j2. The value is 1 if car r executes task j1 first and then task j2, and 0 otherwise. , These represent the start time of the next task j2 and the end time of the previous task j1 on the same cart r, respectively. This represents the travel time required for the vehicle r to travel from the gun position of mission j1 to the gun position of mission j2 without any equipment. M represents a sufficiently large constant. When z=0, the constraint is automatically relaxed. When z=1, the start time of the next task is no earlier than the end time of the previous task plus the empty run time. Constraints on the order of the opposing vehicles passing through the segment (i.e., linearization of formula (19)): For the opposing vehicles r, with a shared capacity of 1 in the segment s, ,use A single, predetermined order is specified: --Formula (31-1); --Formula (31-2); Represents variables of 0 and 1, when When formula (31-1) takes effect, the car r passes through section s first, when At that time, formula (31-2) takes effect, and the car... First, pass through segment s; Formulas (31-1) and (31-2) together ensure that oncoming vehicles will never meet on a section with a capacity of 1.
[0043] , The numbers represent the car r, ... At the start of the occupancy window of segment s; , The numbers represent the car r, ... At the end of the occupancy window of segment s; Constraints on power supply and energy consumption: --Formula (32-1); Energy storage power supply compartment: --Formula (32-2); Direct power supply compartment: --Formula (32-3); Formula (32-1) means that the sum of the charging power of trolleys operating simultaneously within any power supply segment b shall not exceed the capacity of that segment. ; This hard constraint is used in power plants with tight power capacity, while other power plants can retain only the soft penalty in the sixth term of formula (29); This represents the charging power of the task served by the trolley r operating within segment b; The indicator function is defined as follows: r is 1 when the vehicle is within its operating time window during time period t, and 0 otherwise. The available power of the power supply compartment of the vehicle r is shown in formula (3), which is the product of the current state of charge and the energy storage capacity; This represents the amount of supplementary power that the vehicle r can obtain in the time domain (such as the return end supplementary potential); the energy storage constraint ensures that its cumulative output power does not exceed the sum of the available power and the supplementary power; Formula (32-3) indicates that the output power of the direct-supply power supply compartment does not exceed the rated value; Operation duration and yield area capacity constraints: --Formula (33-1); --Formula (33-2); Formula (33-1) indicates that the actual power supply duration is not less than the planned operation duration determined by formula (4); Formula (33-2) means that the number of cars in the passing zone y at any given time does not exceed its capacity. ; In addition, the same-direction tracking interval of formula (18), the job cutting limit of formula (20), and the candidate feasibility of formula (12) are retained as constraints.
[0044] Step S74 (Two-stage solution): In this embodiment, considering the real-time nature of online computation, a two-stage algorithm known to those skilled in the art, namely "fast and feasible initial solution + heuristic improvement," is adopted for the solution. The technology involved in the two-stage solution is existing technology and is not the focus of this solution's improvement, so it will not be elaborated on here. The solution process is simply described below, which will be clear to those skilled in the art: Phase 1: Assign tasks within the time domain according to their relaxation level Ascending order ( If they are the same, then the default weight is... In descending order, each task is ranked in the feasible candidate vehicles ( In the process of inserting the task into its task sequence, the increment of the objective function obtained through closed-loop derivation in steps S4 to S6 is calculated. ,choose The smallest cart and its insertion position; perform a deadlock-free check after inserting each task; Phase Two: (Using existing tabu search techniques or existing genetic algorithms to solve) Encoding is "assignment vector + task order of each vehicle + yield decision". Neighborhood actions include task reassignment (vehicle switching), exchanging the order of tasks within the same vehicle, changing yield points, and batch redrawing of bottleneck section directions; each neighborhood solution calls steps S4~S6 to re-deductively and verify its feasibility, and infeasible solutions are included. Penalty: Avoid roundabout search by using a tabu list, stop after reaching the maximum number of iterations, and output the current optimal feasible solution.
[0045] This step unifies user waiting / departure default (demand side), empty running and yielding waiting (movement side), load balancing (resource side), and sliding contact line segmented power capacity (power supply side) into the same optimization goal, and takes the deadlock-free closed loop of S4~S6 as the feasibility premise to realize the resource allocation of the "demand-movement-power supply" three-way coupling.
[0046] Preferably, step S8 includes: Step S81 (Instruction Issuance): The scheduling server parses the plan into execution instructions to each vehicle, including the task sequence and target charging gun, departure time and running direction, planned entry / exit time windows for each section along the route, passing points and waiting time, docking / disconnection instructions, and expected start and end times for power supply. At the same time, it pushes the expected start time for power supply and the expected completion time to the user side. Step S82 (Execution Monitoring): The trolley periodically reports its position, speed, operating condition, and actual power; the track-side positioning device verifies the actual entry / exit times of the trolley in each section; the system compares the actual trajectory with the planned time window. The discrepancies are compared and recorded, and step S9 determines whether to trigger a rescheduling.
[0047] Preferably, step S9 includes: Step S91, (Rolling Time Domain and Triggering Event): The system operates according to a fixed period. Rolling forward, and triggering rescheduling immediately upon the occurrence of any of the following events: e1: New charging demand arrives; e2: User unplugs charging gun or ends operation early; e3: Operation is extended due to changes in negotiated power; e4: Trolley malfunction or power supply compartment abnormality; e5: Actual trajectory deviates from the planned time window by more than a threshold; e6: Segmented power fluctuation of the sliding contact line; Deviation criteria and freeze time domain are: --Formula (34); Replanning is triggered when formula (34) is satisfied, where the time domain is frozen. The existing mobile or job occupancy windows are fixed, and only for [other purposes]. The task and the idle car will re-execute steps S3 to S7; , These represent the actual entry time of the car r into section s and the planned entry time, respectively. This indicates that if the time window deviates from the threshold, the plan is considered invalid and needs to be replanned. This indicates the length of the frozen time domain. Tasks that are already in transit or connected will not be interrupted within this time domain to ensure safety and user experience. Tasks outside the frozen time domain will be re-optimized. Indicates the current scheduling time when rescheduling is triggered; Indicates a frozen time domain: within this time domain, the moving window and the operation window of the trolley that are already en route or have already docked for work remain fixed and uninterrupted to ensure operational safety and user experience during charging. This indicates that the part outside the time domain is frozen: only tasks that have not been executed after that time and idle cars are re-executed through steps S3 to S7, and the replanned scheme is then issued through step S8. Step S92, (Topology restoration for early job termination): Upon receiving the e2 event, close the corresponding job window in advance, delete the blocking point from the blocking point set A, and recalculate the connectivity and earliest reachable time of the affected candidate pair according to formula (10) and formulas (11-1) to (11-4); if the task that was originally blocked and was waiting becomes reachable or no longer times out, immediately generate a new plan and notify the relevant car to start early. Step S93 (Fault Handling): Upon receiving the e4 event, the faulty trolley is processed according to the new blocking point before the expected recovery time, and the expected recovery time is taken at the end of the operation occupancy window. If the reachability cannot be estimated, set it to a sufficiently large value and issue an alarm for manual intervention. Recalculate the reachability according to formula (10) and formulas (11-1) to (11-4), and reassign other vehicles to tasks that have not yet started. Step S94 (Power Fluctuation Handling): When the e6 event is received, re-verify the segmented power according to formula (24), formula (32-1) and formula (32-2). If necessary, make local adjustments to the start time of the off-peak operation within the time window, or arrange for the trolley to wait briefly in the passing area / storage area before docking.
[0048] This step addresses the characteristic of charging operations being "flexible in length and determined by the user," by setting up three dedicated mechanisms: restoring track connectivity and waking up waiting tasks when the operation ends early, converting faulty vehicles into dynamic blocking points, and triggering peak shaving due to power fluctuations. These mechanisms ensure the continued feasibility of the scheduling scheme under uncertain environments.
[0049] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0050] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
Claims
1. A method for resource allocation and collaborative scheduling of an aerial mobile charging robot, applied to an aerial mobile charging robot system, the system comprising a track for parking spaces, a sliding contact line laid on the track, multiple trolleys running along the track and carrying power supply compartments, and charging guns corresponding to the parking spaces and fixedly installed on the track, wherein the trolleys draw power through contact with the sliding contact line via current collectors and are powered by the power supply compartments docking with the charging guns at the target charging guns; characterized in that... The method is executed by the scheduling server and includes the following steps: Step S1: Obtain the segment division of the track and the capacity of each segment, the location of each charging gun, the location and capacity of the passing zone, the power capacity of each power supply segment, and collect the location, running direction, working condition and power supply capacity of each car to establish a track spatiotemporal resource model. Step S2: Receive charging requests, determine the target charging gun, required power or expected charging time and expected departure time for each charging task, and calculate the charging operation time, deadline and time slack for each charging task. Step S3: Record the trolley in the docking charging operation as the blocking point, determine the connectivity of the track at different times based on the operation time window of the blocking point, calculate the earliest reachable time of the candidate trolley to the target charging gun, screen out feasible trolley and task candidate pairs, and obtain candidate scheduling schemes. Step S4: For the candidate scheduling scheme, along the sequence of travel segments from the current position of the vehicle to the target charging gun, the movement window of the vehicle to the segment and the operation window at the target charging gun are respectively deduced, and during the operation window, the blocking mark is set on the segment occupied by the vehicle to obtain the unified time window of each segment. Step S5: Based on the unified time window, detect the passage conflict between the vehicles, establish a directed waiting graph with the vehicles as nodes and the waiting relationship between the vehicles as directed edges, detect the directed cycles in the directed waiting graph to determine whether there is a deadlock in the scheduling scheme, and verify whether the power of simultaneous operation in the same power supply segment exceeds the limit. Step S6: For the detected traffic conflicts or directed loops, determine the passing order of the yielding point and the yielding car, the critical section and the departure time of the relevant cars, or organize the tasks in batches according to the driving direction, and return to step S4 to re-determine after correction, until the scheduling scheme has no traffic conflicts, no directed loops and the power does not exceed the limit. Step S7: With the goal of minimizing the weighted sum of user waiting costs, departure default costs, empty vehicle running costs, yielding waiting costs, unbalanced vehicle load costs, and power supply segment power limit exceeding costs, under the constraints of unique task assignment, mutually exclusive time windows of segments, dynamic reachability, operation duration, and power supply segment power capacity, the assignment relationship between tasks and vehicles, the task sequence of each vehicle, and the departure time are determined to obtain a collaborative scheduling scheme. Step S8: The collaborative scheduling plan is sent to each vehicle. The sent content includes the target charging gun, departure time, driving direction, planned time window for each section, passing point and waiting time, docking and operation instructions, and the actual operation information of the vehicle is collected and compared with the planned time window. Step S9: When any of the following triggering events are detected, such as a new charging request, early termination or extension of the charging operation, vehicle malfunction, deviation of the actual trajectory from the planned time window exceeding the threshold, or power fluctuation in the power supply segment, the time window already executed within the frozen time domain is frozen, and steps S3 to S7 are re-executed for tasks not executed outside the frozen time domain; wherein, when the charging operation ends early, the corresponding blocking point is removed, track connectivity is restored, and the earliest reachable time of the candidate vehicle is updated.
2. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 1, characterized in that, Step S1 includes: Step S11: Establish a one-dimensional coordinate along the track length and discretize the track into an ordered set of segments. Record the length, midpoint coordinates, and capacity of each segment. Only one car is allowed to occupy a normal segment at any given time. Two cars are allowed to pass each other or avoid each other in the passing area and the parking area at the end of the track. The capacity is 2. Each charging gun corresponds to a parking space and its coordinates and the segment to which it belongs are recorded. The sliding contact line is divided into several power supply segments along the track. Record the set of segments covered by each power supply segment and its power capacity. Step S12: Record the vehicle length, rated operating speed, rated output power of the power supply compartment, docking time for the power supply compartment to dock with the charging gun, and evacuation time for the evacuation channel after charging is completed for each vehicle; when the vehicle is located in any section, determine the set of sections actually covered by the vehicle according to its vehicle length and the coordinates of the midpoint of each section. Step S13: Periodically collect the real-time status of each vehicle. The real-time status includes at least the vehicle's current position, running direction, working condition, power supply compartment charge status and expected idle time. The working condition includes three states: idle, moving and docking operation. Step S14, the power supply compartment is either energy storage type or direct supply type: for energy storage type power supply compartment, the available power is determined by the product of its state of charge and energy storage capacity; for direct supply type power supply compartment, its rated output power is used as the available power supply capacity.
3. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 2, characterized in that, Step S2 includes: Step S21: Represent each charging request as a charging task and record its task parameters. The task parameters include the target charging gun, request time, required power, default expected charging duration, default expected departure time, and task priority. The vehicle, the charging gun, and the power supply compartment negotiate to determine the charging power of the task, and the charging power does not exceed the rated output power of the power supply compartment and the rated power of the charging gun. The planned charging time for charging task j is calculated using the following formula: ; When the user gives the desired charging time At that time, the planned charging time is taken ; This indicates the planned charging operation duration for charging task j, which is the duration for which the vehicle is connected to the power supply at the charging gun position. This indicates the charging power negotiated and determined by charging task j. This indicates the user's stated expected charging time; This indicates the amount of electricity required for charging task j; The required electricity is calculated according to the following formula: ; , These represent the user's target state of charge and the initial and state of charge when initiating a charging request, respectively. This indicates the rated quantity of the power battery of the vehicle to be charged. Step S22: When the task does not have a given expected exit time, the task's deadline is calculated according to the following formula: ; Indicates the expected departure (charging complete) deadline for charging task j; Indicates the moment when the user initiated the charging request; This indicates the upper limit of the longest waiting time promised by the charging station. Step S23: Calculate the time relaxation of task j at time t according to the following formula. : ; Step S24: Determine the waiting cost and departure timeout cost per unit time according to the following formulas: ; ; , Let $\mathbf$ and $\mathbf$ represent the waiting cost for charging task j to start powering by 1 unit of time and the default cost for exceeding the deadline by 1 unit of time, respectively. This represents the task priority coefficient, which is determined by membership level, service commitment, or overtime compensation rules, and defaults to 1. This represents the penalty multiplier for timeout relative to waiting, and is a configurable coefficient greater than 1.
4. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 3, characterized in that, Step S3 includes: Step S31: Record the currently docking trolleys and the trolleys that are to be docked in the confirmed scheduling plan as blocking points to form a blocking point set; record the operation position, the set of sections covered by the vehicle body, and the operation time window defined by the start time and end time of the operation for each blocking point; within the operation time window of the blocking point, other trolleys are prohibited from occupying the section covered by the vehicle body. Step S32, define the connectivity between two points on the track at time t: when there is no blocking point at time t within the operation time window between the two points, the two points are determined to be connected and directly reachable; otherwise, the two points are determined to be blocked. The connectivity changes with time as the operation time window of the blocking point begins and ends. Step S33: For candidate vehicle r and task j, proceed sequentially from the current position to the target charging gun along the vehicle's travel direction: the arrival time of the vehicle traveling on the empty track to the next blocking point is calculated by dividing the travel distance by the rated speed; if the blocking point has not yet finished its operation when the vehicle arrives, the vehicle waits in front of it, updates the current time to the sum of the blocking point's operation completion time and the withdrawal time, and then continues to advance; taking into account the docking time at the target charging gun, the earliest achievable time when the candidate vehicle completes docking and can start power supply is recursively obtained; Step S34: When the sum of the earliest reachable time and the planned charging operation duration is not later than the task deadline, the corresponding vehicle-task pair is marked as a candidate feasible pair; otherwise, it is eliminated. In soft cutoff mode, only the target charging gun is required to be physically reachable, and its departure timeout is included in the optimization objective of step S7. Based on the condition that the available power is not less than the required power, or the available power supply capacity is not less than the charging power, candidate vehicles are eliminated.
5. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 4, characterized in that, Step S4 includes: Step S41: Determine the driving direction of the trolley based on the position of the target charging gun and the current position of the trolley, and determine a unique sequence of driving segments from the current position to the target charging gun on the one-dimensional track. Step S42: Starting from the departure time of the car, the time when the front of the car reaches the boundary of each section is calculated sequentially according to the length of each section and the rated speed of the car. The time from the arrival time of the front of the car to the departure time of the rear of the car is used to determine the movement window of the car for each section. The departure time of the rear of the car is included in the time required for the car body to pass through the section. Step S43: The earliest time when the candidate car reaches the target charging gun is taken as the start time of the operation, and the start time of the operation plus the planned charging operation duration is taken as the end time of the operation. The operation windows with the same start and end times are set for each section covered by the vehicle body at the target charging gun, and the blocking mark is set for the main section where the target charging gun is located. Step S44: Merge the movement window and operation window of the trolley to the section to form a unified time window for the trolley to the section; each scheduling scheme is represented as a set of unified time windows for all trolleys to all sections. Step S45, the unified time window set satisfies at least one of the following mutual exclusion requirements: The time when a following vehicle traveling in the same direction enters any section is no earlier than the sum of the time when the preceding vehicle leaves the section and the safe following interval; Two cars traveling in opposite directions are not allowed to intersect in a section with a capacity of 1; they are only allowed to meet in a passing zone with a capacity of 2. The movable windows of other trolleys must not intersect with the working windows of the docking trolley within its body coverage area.
6. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 5, characterized in that, Step S5 includes: Step S51: Count the number of four types of conflicts in the unified time window: the number of same-direction tracking conflicts that violate the safety tracking interval, the number of opposing intersection conflicts where opposing trolleys intersect in the window occupied by the trolley in the section with a capacity of 1, the number of operation cutting conflicts where the moving window and the operation window intersect in the vehicle body coverage section, and the number of power exceeding the power capacity when the power of the same power supply segment is operating simultaneously. Step S52: Establish a directed waiting graph with all cars as nodes: When a car is scheduled to enter a section that is occupied by another car, and the one-dimensional driving section sequence is unique, making it impossible for the car to go around and it can only wait, add a directed edge from the waiting car to the car being waited for, and use the difference between the expected release time of the section by the car being waited for and the expected arrival time of the waiting car as the waiting time of the directed edge. Step S53: When there is a directed cycle consisting of directed edges in the directed waiting graph, it is determined that the scheduling scheme has a circular waiting deadlock and the scheme is infeasible; the directed cycle is detected by depth-first search or strongly connected component algorithm; the scheduling scheme is determined to pass the deadlock-free check only when the number of opposing intersection conflicts, the number of job cutting conflicts and the number of directed cycles are all zero. Step S54: For each power supply segment and each work period, sum the charging power of the docking trolleys within the coverage area of that power supply segment. The portion exceeding the power capacity of that power supply segment is recorded as the power exceeding the limit. If the power exceeding the limit is not zero, the power verification of the scheduling scheme is determined to be unsuccessful.
7. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 6, characterized in that, Step S6 includes: Step S61: For two cars that are in a head-on collision, calculate the sum of the travel time of each car to enter each passing zone and the time required to wait for the other car to pass in the passing zone as the passing cost of that car in the passing zone. Select the passing zone with the smallest passing cost as the passing point. Compare the minimum passing costs of the two cars. The car with the smaller minimum passing cost enters the passing point to give way and the other car passes first. Step S62: For multiple vehicles that need to pass through the same critical section, the passage order is determined in ascending order by using the time slack of the task served by each vehicle as the passage order key. If the time slack is the same, the passage order is determined in ascending order by the earliest reachable time. The departure time of the vehicle that is ranked later is postponed until after the sum of the time when the tail of the preceding vehicle leaves the critical section and the safe tracking interval. Step S63: For tasks that need to cross the same bottleneck section or the same work cut point within the rolling time domain, organize them in batches according to the direction of the vehicle's travel. First, release the task string in one direction of travel, and then release the task string in the opposite direction of travel. The direction switching position is arranged in the passing area or the end storage area. The batching scheme is determined by the minimum weighted sum of the number of direction switching in the time domain and the departure timeout increment caused by batching. In step S64, after each completion of point assignment, passage order rearrangement, departure time postponement, or travel direction batching, return to step S4 to re-analyze all unified time windows, and then execute step S5 for re-checking; when the number of opposing intersection conflicts, the number of job cutting conflicts, and the number of directed loops are all zero, and the same-direction tracking conflicts and power exceeding the limit are acceptable, a deadlock-free feasible solution is obtained; set a maximum number of iterations, and candidate solutions that still do not meet the requirements after exceeding this number of iterations are determined to be infeasible, and penalties are imposed or they are discarded in step S7.
8. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 7, characterized in that, Step S7 includes: Step S71, establish the following decision variables: assignment variable indicating whether a task is assigned to a car, sequence variable indicating the order in which the same car undertakes two tasks, departure time and start and end time of each car, passage sequence variable indicating the order in which two opposing cars pass through the critical section, and yielding variable indicating whether a car enters the yielding zone to yield. Step S72, with the optimization objective of minimizing the weighted sum, the weighted sum includes: the weighted sum of waiting costs for all tasks from the request time to the actual start time of power supply, the weighted sum of departure default costs for all tasks whose work completion time exceeds the deadline, the cost of the trolley's empty driving time, the total waiting time for the trolley to give way in the passing zone, the load imbalance penalty for the deviation of each trolley's task volume from the average task volume of all trolleys, and the power limit violation penalty for power supply segments; wherein the weights of the departure default cost and the deadlock risk item are taken as the relatively larger values; Step S73, the optimization satisfies the following constraints: Assignment and vehicle sequence constraints: Each task is assigned to exactly one vehicle; multiple tasks undertaken by the same vehicle are executed sequentially, and the start time of the next task is no earlier than the end time of the previous task plus the empty travel time to the charging gun of the next task. Opposing segment passage order constraint: For two cars traveling in opposite directions and needing to pass through the same segment with a capacity of 1, set a segment passage order variable to determine which car passes first, and require that the time when the car that passes later enters the segment is not earlier than the sum of the time when the car that passes first leaves the segment and the safety interval, so that the occupancy windows of any two opposing cars on the segment with a capacity of 1 do not intersect. Power supply constraints: During any work period, the total charging power of the trolleys operating simultaneously in the same power supply segment shall not exceed the power capacity of that power supply segment; the total electricity demand of the tasks undertaken by the energy storage type power supply compartment shall not exceed the sum of its available electricity and the electricity replenished during the journey; the charging power of the tasks undertaken by the direct power supply compartment shall not exceed its rated output power. Operation duration and passing zone capacity constraints: The difference between the end time and start time of each task shall not be less than its planned charging operation duration; the number of vehicles in the same passing zone at the same time shall not exceed the capacity of that passing zone; Step S74, using a two-stage solution: In the first stage, the tasks to be assigned are arranged in ascending order of time slackness, and in descending order of task priority when the time slackness is the same. The tasks are then inserted into the task sequence of the candidate cars according to the arrangement order. The insertion result that minimizes the weighted sum increment and is verified as deadlock-free through steps S4 to S6 is selected to generate an initial feasible scheduling scheme. Phase 2 involves solving the initial feasible scheduling scheme using tabu search. The neighborhood operations include at least one of task reassignment, reordering of tasks in the same vehicle, changing passing points, and batch redrawing of driving directions. Neighborhood solutions containing deadlocks are penalized based on the number of directed cycles. The process stops when the maximum number of iterations or the computation time limit is reached, and the cooperative scheduling scheme is output.
9. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 8, characterized in that, Step S8 includes: Step S81: The collaborative scheduling scheme is parsed and sent to each vehicle. The sent content includes: the vehicle's task sequence, target charging gun, departure time and driving direction, planned time window for each section, passing points and waiting time at passing points, power supply connector docking and disconnection instructions, and estimated start and end times of the operation; and the estimated start time and estimated completion time are pushed to the corresponding user. In step S82, each vehicle periodically reports its actual position, running speed, operating condition, and charging power. The scheduling server verifies the actual entry and exit times of each vehicle into and from each segment and compares them with the planned time window for that segment, recording the time deviation. The time deviation serves as the basis for determining the trajectory deviation trigger event in step S9.
10. The method for resource allocation and collaborative scheduling of an aerial mobile charging robot according to claim 9, characterized in that, Step S9 includes: Step S91: Advance the scheduling time domain according to the rolling cycle, and start rescheduling when any of the following triggering events are detected: new charging request, early termination of charging operation, extension of charging operation, trolley failure, deviation of the actual entry or exit time of the trolley from the planned time window exceeding the preset deviation threshold, and power fluctuation of the power supply segment. During rescheduling, a frozen time zone is set. The time window of the work vehicle that is already on the way or docked within the frozen time zone remains unchanged. Only tasks that have not been executed and idle vehicles outside the frozen time zone are re-executed through steps S3 to S7. The rescheduling plan is then issued through step S8. Step S92: When the charging operation ends early, update the end time of the operation to the current time, remove the corresponding car from the blockage point set, and recalculate the track connectivity and the earliest reachable time of the affected candidate cars; if the task that was originally blocked by the blockage point and was waiting becomes reachable or no longer timed out, immediately generate a new plan and notify the relevant cars to start early. Step S93: When a trolley malfunctions, it is treated as a blockage point before the expected recovery time. The end time of its operation or occupancy window is taken as the expected recovery time. If the expected recovery time cannot be estimated, a preset maximum value is taken and a manual intervention alarm is issued. The tasks that the malfunctioning trolley has not yet started are reassigned to other trolleys, and steps S3 to S7 are executed again. Step S94: When the power supply segment fluctuates, re-verify according to the power limit and the power supply constraint, and make staggered local adjustments to the start time of the affected operation within a unified time window, or arrange the relevant trolleys to wait briefly in the passing area or end storage area before docking.
