A resource scheduling method and device based on a virtual time axis and related media

Abstract

The application discloses a resource scheduling method and device based on a virtual time axis and related media, which comprises collecting real-time state information of each target resource to generate a global state snapshot, constructing a multi-dimensional virtual time axis model based on the global state snapshot, screening candidate task steps and sorting, using the multi-dimensional virtual time axis model to perform resource trial allocation and conflict detection, combining recursive deduction to verify feasibility, rolling back and reselecting candidate task steps when verification fails, and circulating until the resource scheduling result is output. The application solves the technical problems that the scheduling decision falls into a deadlock or the task step with time constraints cannot obtain the required resources within the specified time limit and fails to execute by performing resource trial allocation and time occupation segment conflict detection on the candidate task steps in the multi-dimensional virtual time axis model and verifying the feasibility of subsequent task steps by combining the recursive deduction mechanism.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a resource scheduling method, apparatus and related media based on a virtual time axis. Background Technology

[0002] Existing resource scheduling technologies mainly include two types of schemes: static timetable scheduling and priority queue-based dynamic scheduling. Static timetable scheduling pre-calculates the execution timetable of all task steps and executes them according to the table; priority queue-based dynamic scheduling preempts resources according to the priority of the current task step and allocates resources when they are available.

[0003] However, the existing scheduling techniques mentioned above suffer from the following technical shortcomings: they all lack a forward-looking trial allocation mechanism for multiple types of resources and the ability to recursively verify feasibility. When allocating resources for the current task step, existing scheduling methods make allocation decisions solely based on the immediate availability of resources. They neither structurally model the occupancy of various resources in the time dimension nor predict or verify potential time-related conflicts in subsequent task steps caused by the current allocation decision. When an allocation decision leads to irresolvable conflicts in the time-related occupancy of multiple types of resources in subsequent task steps, the system lacks corresponding rollback and reselection mechanisms. This can result in a deadlock in the overall scheduling or cause time-constrained subsequent task steps to fail to acquire the required resources within the specified time limit. Summary of the Invention

[0004] This invention provides a resource scheduling method, apparatus, and related medium based on a virtual time axis, aiming to solve the technical problem in the prior art that the lack of a forward-looking trial allocation mechanism and recursive feasibility verification capability for multiple types of resources leads to deadlock in scheduling decisions or failure of time-constrained task steps to obtain the required resources within the specified time limit.

[0005] In a first aspect, embodiments of the present invention provide a resource scheduling method based on a virtual time axis, comprising: Collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and their corresponding dependencies; Based on the global state snapshot, a virtual timeline is constructed for each type of target resource, and resource occupation is abstracted into time occupation segments to obtain a multi-dimensional virtual timeline model. Based on the dependencies, candidate task steps that meet the execution conditions are selected from the set of task steps and sorted according to a preset priority rule; The multidimensional virtual timeline model is used to perform resource allocation for the current candidate task steps, and conflict detection is performed on the time-occupied segments to obtain conflict detection results. Based on the conflict detection results, the feasibility of the task steps is verified by recursive deduction. If the verification fails, the process rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. This process is repeated until the verification is successful and the resource scheduling result is output.

[0006] Secondly, embodiments of the present invention provide a resource scheduling device based on a virtual time axis, comprising: The data acquisition unit is used to collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and their corresponding dependencies. The model building unit is used to construct virtual timelines for various target resources based on the global state snapshot, and abstract resource occupancy into time occupancy segments to obtain a multi-dimensional virtual timeline model. The step sorting unit is used to filter candidate task steps that meet the execution conditions from the task step set according to the dependency relationship, and sort them according to a preset priority rule. The conflict detection unit is used to perform resource allocation for the current candidate task steps using the multi-dimensional virtual time axis model, and at the same time to perform conflict detection on the time-occupied segments to obtain conflict detection results. The loop calculation unit is used to verify the feasibility of the task steps based on the conflict detection results using recursive deduction. If the verification fails, it rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. The loop continues until the verification is successful and the resource scheduling result is output.

[0007] Thirdly, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the resource scheduling method based on a virtual time axis of the first aspect.

[0008] Fourthly, embodiments of the present invention provide a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, it implements the resource scheduling method based on a virtual time axis of the first aspect.

[0009] This invention provides a resource scheduling method based on a virtual timeline, comprising: collecting real-time status information of each target resource to generate a global state snapshot, and loading a set of task steps to be executed and their corresponding dependencies; constructing a virtual timeline for each type of target resource based on the global state snapshot, and abstracting resource occupation into time occupation segments to obtain a multi-dimensional virtual timeline model; selecting candidate task steps with execution conditions from the task step set according to the dependencies, and sorting them according to a preset priority rule; using the multi-dimensional virtual timeline model to perform resource trial allocation on the current candidate task steps, and simultaneously performing conflict detection on the time occupation segments to obtain conflict detection results; verifying the feasibility of the task steps using recursive deduction based on the conflict detection results, and if the verification fails, rolling back to the resource trial allocation stage, selecting the next candidate task step to re-perform the resource trial allocation, and repeating the process until verification is successful to output the resource scheduling result. This invention solves the technical problem of scheduling decisions getting stuck in deadlock or time-constrained task steps failing to acquire the required resources within the specified time limit due to the inability of time-constrained task steps to execute. By performing resource trial allocation and time segment conflict detection on candidate task steps on a multi-dimensional virtual time axis model, and combining a recursive deduction mechanism to verify the feasibility of subsequent task steps, this invention solves the technical problem of scheduling decisions getting stuck in deadlock or time-constrained task steps failing to execute due to the inability to acquire the required resources within the specified time limit.

[0010] The present invention also provides a resource scheduling device, computer equipment and storage medium based on a virtual time axis, which have the same beneficial effects as described above. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating a resource scheduling method based on a virtual time axis provided in an embodiment of the present invention; Figure 2 This is a schematic block diagram of a resource scheduling device based on a virtual time axis, provided for an embodiment of the present invention.

[0013] Explanation of reference numerals in the attached figures: 200. Resource scheduling device based on virtual time axis; 201. Data acquisition unit; 202. Model building unit; 203. Step sorting unit; 204. Conflict detection unit; 205. Loop calculation unit. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0016] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0017] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0018] Please see below. Figure 1 , Figure 1 The flowchart of a resource scheduling method based on a virtual time axis provided in an embodiment of the present invention specifically includes steps S101 to S105.

[0019] S101. Collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and their corresponding dependencies; S102. Based on the global state snapshot, construct a virtual timeline for each type of target resource, and abstract resource occupation into time occupation segments to obtain a multi-dimensional virtual timeline model; S103. Based on the dependency relationship, select candidate task steps that meet the execution conditions from the task step set, and sort them according to a preset priority rule; S104. Using the multi-dimensional virtual timeline model, perform resource allocation for the current candidate task steps, and simultaneously perform conflict detection on the time-occupied segments to obtain conflict detection results. S105. Based on the conflict detection results, the feasibility of the task steps is verified by recursive deduction. If the verification fails, the process rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. This process is repeated until the verification is successful, and the resource scheduling result is output.

[0020] In step S101, at the start of each scheduling cycle, real-time status information of each target resource is collected, including its current occupancy, location, and remaining occupancy time. This information is then aggregated and encapsulated to generate a global status snapshot reflecting the overall resource status of the current system. Simultaneously, all task steps to be executed are read, and the prerequisite dependencies between each task step are parsed to establish a task dependency topology, resulting in a set of task steps and their corresponding dependencies, providing a data foundation for subsequent scheduling simulations.

[0021] In one embodiment, step S101 includes: At the start of each scheduling cycle, the occupancy status, current position and remaining time of each target resource are collected and integrated to obtain the real-time status information corresponding to each target resource. The real-time status information corresponding to each of the target resources is aggregated and encapsulated to generate the global status snapshot. Read all task steps to be executed, parse the prerequisite dependencies between each task step to establish a task dependency topology, and obtain the set of task steps and their corresponding dependencies.

[0022] In this embodiment, the scheduling cycle refers to the smallest time unit in which the system performs a complete scan and deduction of the current resource status and task queue, and outputs a scheduling decision. At the start of each scheduling cycle, the system first enters the global status acquisition phase, performing real-time status acquisition operations on all target resources in the system one by one. Target resources refer to various physical or logical resource entities participating in task execution, such as robotic arms, conveyor belts, reaction vessels, heating positions, etc. For each target resource, the system collects the following three types of information: first, occupancy status, i.e., whether the target resource is currently occupied, and the identifier of the task step occupying the resource; second, current position, i.e., the current position of the target resource in the physical space or logical state space; and third, remaining time, i.e., if the target resource is currently occupied, the remaining time that its currently occupied task step is expected to continue executing. These three types of information together constitute the real-time status information of a single target resource, fully reflecting the availability and status characteristics of the resource at the current moment.

[0023] After collecting data on each target resource, an aggregation and encapsulation operation is performed on the real-time status information corresponding to all target resources. This integrates the scattered individual resource status information into a structured data object, namely the Global State Snapshot. The Global State Snapshot is timestamped and represents a consistent view of the overall resource status of the system at the start of the current scheduling cycle. This ensures that the resource information used in subsequent scheduling simulations comes from the same time base, avoiding inconsistencies in status caused by differences in data collection timing.

[0024] After generating the global state snapshot, the task loading phase begins. All task steps to be executed in the task queue are read, and the prerequisite dependencies between each task step are resolved. A prerequisite dependency is a constraint that a task step can only begin execution after one or more of its specified prerequisite task steps have been completed. Based on these dependencies, all task steps to be executed and their mutual prerequisite dependencies are organized into a task dependency topology, which is a directed acyclic graph structure with each task step as a node and prerequisite dependencies as directed edges. After this resolution and organization, a set of task steps containing all task steps to be executed and their complete dependencies is obtained, providing a structured task data foundation for subsequent candidate task step selection and scheduling deduction.

[0025] In a preferred embodiment, taking a laboratory automation scenario as an example, at the start of scheduling cycle T, the status of four types of target resources—robotic arm A, conveyor belt B, reaction vessel C, and heating station D—is collected: robotic arm A is currently idle, its current position is the initial reset point, and the remaining time is 0; conveyor belt B is currently occupied, executing sample transfer task step-03, with 8 seconds remaining; reaction vessel C is currently idle, its current position is open and ready, with 0 seconds remaining; and heating station D is currently occupied, executing heating and heat preservation task step-01, with 15 seconds remaining. The system aggregates and encapsulates the real-time status information of the above four types of target resources to generate a global status snapshot with a timestamp T. At the same time, the system loads the current task queue, reads a total of 8 task steps to be executed, and parses the pre-dependencies between each step. For example, Step-04 depends on the completion of Step-02, Step-05 depends on the completion of both Step-03 and Step-04, and so on. Finally, a task dependency topology containing the above 8 task steps and their complete pre-dependency constraints is established, and a set of task steps and their corresponding dependencies are obtained, providing complete input data for subsequent scheduling and deduction based on the multi-dimensional virtual time axis model.

[0026] In step S102, a global state snapshot is read, the resource categories of various key resources are identified, and an independent virtual timeline is constructed in memory for each resource category. Each virtual timeline records the occupancy of the corresponding resource category in the time dimension. Each instance of occupancy of each target resource is abstracted into a time occupancy segment, including the start and end times of the occupancy and the occupant information. The virtual timelines corresponding to each resource category and the associated time occupancy segments are integrated to obtain a multi-dimensional virtual timeline model, which is used to simulate the future occupancy status of various resources in a virtual environment.

[0027] In one embodiment, step S102 includes: Read the global state snapshot to identify the resource categories of various key resources in the global state snapshot; An independent virtual timeline is constructed for each of the resource categories, and each virtual timeline corresponds to the time occupancy record of each resource category. Each instance of the occupation of the target resources is abstracted into a time occupation segment; By integrating the virtual timelines corresponding to each resource category and the associated time occupancy segments, a multi-dimensional virtual timeline model is obtained.

[0028] In this embodiment, after generating the global state snapshot, the process proceeds to the construction phase of the multi-dimensional virtual timeline model. The multi-dimensional virtual timeline model refers to a virtual timeline structure built in memory. This structure maintains independent time occupancy records for each type of critical resource in the system, thereby providing a structured representation of the occupancy status of various resources over a future period in a virtual environment. This provides a unified basis for subsequent resource allocation and conflict detection.

[0029] First, read the global state snapshot generated in step S101, classify and identify all target resources recorded in the snapshot, and extract the resource category to which each target resource belongs. A resource category refers to the type affiliation of a group of target resources with the same functional attributes or scheduling characteristics, such as transfer resources and processing unit resources. Different resource categories undertake different functional responsibilities during task execution, and their time consumption characteristics also differ; therefore, classification modeling is necessary to support subsequent multi-dimensional conflict detection.

[0030] After resource category identification, an independent virtual timeline is constructed for each resource category. Each virtual timeline uses time as the horizontal axis and the occupancy status of each target resource under that resource category as the vertical dimension, recording all time periods of confirmed and pending occupancy information for that resource category within the current scheduling cycle. The virtual timelines of different resource categories are independent and do not interfere with each other, thus ensuring that the time occupancy modeling of each type of resource has clear boundaries and independent operability.

[0031] After the virtual timelines for each resource category are constructed, each instance of resource occupation is abstracted and represented as a time segment. A time segment is the basic data unit in the multi-dimensional virtual timeline model, containing four attributes: start time (the expected start time of the occupation); end time (the expected end time of the occupation); occupant ID (the identifier of the task step initiating the occupation); and priority information (the scheduling priority of the task step corresponding to the occupation). By abstracting resource occupation into time segments, precise temporal modeling of resource occupation behavior is performed in the virtual environment, thus supporting subsequent conflict detection operations for overlapping time intervals.

[0032] Finally, the virtual timelines corresponding to all resource categories are integrated with their respective associated time occupancy segments to form a unified multi-dimensional virtual timeline model. This model uses multiple independent virtual timelines as its framework and time occupancy segments as its basic elements, comprehensively depicting the overall time occupancy of various resources in the system within the current scheduling cycle. It serves as the core data structure for subsequent scheduling simulations, trial allocation operations, and conflict detection.

[0033] In a preferred embodiment, the aforementioned laboratory automation scenario is continued. The system reads a global state snapshot with timestamp T and identifies two resource categories: transmission resources and processing unit resources. Transmission resources include conveyor belt B, and processing unit resources include robotic arm A, reaction vessel C, and heating station D. The system constructs an independent virtual time axis VT-Transfer for transmission resources and an independent virtual time axis VT-Processing for processing unit resources. On VT-Transfer, the system abstracts the sample transfer operation currently being performed by conveyor belt B as a time segment, with a start time of T-8 seconds (i.e., the segment starts 8 seconds before time T) and an end time of T+8 seconds. The occupant ID is Step-03, and the priority information is standard priority. In VT-Processing, the system abstracts the heating and heat preservation operation currently being performed at heating position D as a time-occupied segment, with a start time of T - several seconds and an end time of T + 15 seconds. The occupant ID is Step-01, and the priority information is standard priority. Robotic arm A and reaction vessel C are currently in an idle state, and there are no registered time-occupied segments on their corresponding time axes. The system integrates VT-Transfer and VT-Processing, along with all their time-occupied segments, to obtain a multi-dimensional virtual time axis model. This model fully reflects the resource occupancy status of the system at time T, providing an accurate virtual simulation basis for the subsequent screening and sorting of candidate task steps in step S103 and the resource trial allocation and conflict detection in step S104.

[0034] In step S103, the task step set is traversed, and each task step's prerequisites are checked against dependencies to determine if they have all been completed. Task steps that meet the execution conditions are then selected as candidate task steps. During the sorting phase, all candidate task steps are sorted according to scheduling priority to ensure that critical processes receive priority resource verification during scheduling simulations.

[0035] In one embodiment, step S103 includes: Traverse each task step in the set of task steps, and determine whether the prerequisites of each task step have been completed according to the dependency relationship. Select the task steps whose prerequisites have been completed as the candidate task steps. Identify the task steps marked as atomic tasks or integration steps among the candidate task steps, and set the scheduling priority of the atomic task or integration step to be higher than that of the other candidate task steps; All candidate task steps are sorted according to the scheduling priority to obtain a sorted sequence of candidate task steps.

[0036] In this embodiment, after the construction of the multi-dimensional virtual timeline model is completed, the candidate task steps are screened and sorted. The goal of this stage is to identify task steps with the conditions for execution from the current set of task steps, and sort them according to the scheduling priority rules to form an ordered sequence of candidate task steps for sequential processing in the subsequent resource allocation stage.

[0037] First, a traversal operation is performed on all task steps in the task step set. Based on the dependency relationships parsed in step S101, it is determined whether all prerequisite dependencies of each task step have been completed. Completed prerequisite dependencies mean that the execution status of all prerequisite task steps on which the task step depends has been marked as completed; that is, in the task dependency topology, all directed incoming edges pointing to this step correspond to the prerequisite nodes in a completed state. For task steps that meet the above conditions, they are included in the candidate task step set as candidate objects with execution conditions in the current scheduling cycle. For task steps whose prerequisite dependencies are not yet fully completed, the system does not consider them in the current scheduling cycle; they will be included in the screening after their prerequisite dependencies are completed in subsequent cycles. Through the above candidate screening mechanism, the system ensures that all task steps participating in the resource allocation simulation have legal execution qualifications at the dependency level, thereby avoiding invalid scheduling decisions that violate task timing constraints.

[0038] After the candidate task steps are determined, the task attribute tags of each candidate task step are identified, with a focus on detecting task steps marked as atomic tasks or integrated steps. An atomic task is a set of task steps that must be executed sequentially and cannot be interrupted. Once its first step is started, subsequent steps must continuously obtain the necessary resources to ensure the integrity of the process, such as a series of chemical reaction steps that must be completed consecutively in a laboratory scenario. An integrated step is a task step with cross-resource collaboration characteristics that has a critical dependency on multiple subsequent task steps. Its completion directly affects the feasibility of the overall scheduling path. The system sets the scheduling priority of identified atomic tasks or integrated steps higher than that of other candidate task steps to ensure that such critical processes receive priority resource verification in scheduling simulations, guaranteeing that once started, they can continuously obtain the necessary resources without interruption.

[0039] After the priority setting is completed, the system sorts all candidate task steps according to their scheduling priority. Task steps with higher scheduling priority are placed at the front of the sequence, while task steps with the same scheduling priority are further sorted according to the preset secondary sorting rules. Finally, the sorted sequence of candidate task steps is obtained, which serves as the input for the resource trial allocation operation in the subsequent step S104.

[0040] In a preferred embodiment, the aforementioned laboratory automation scenario continues. During scheduling period T, the system iterates through the eight task steps in the task step set, checking the completion status of each step's prerequisites based on the task dependency topology: Step-02's prerequisite, Step-01, is still in execution and not yet completed; therefore, Step-02 does not meet the execution conditions and is not included in the candidate set. Step-04 depends on Step-02 and also does not meet the execution conditions. Step-06 and Step-07 have no prerequisites or all prerequisites have been completed, thus meeting the execution conditions and are included in the candidate task step set. Assume that the candidate task step set obtained after screening includes three steps: Step-06, Step-07, and Step-08. The system then identifies the task attribute tags of these three candidate steps and finds that Step-07 is marked as an atomic task, corresponding to a set of sample processing operations that must be executed continuously; interruption will result in material loss. The system sets the scheduling priority of Step-07 to be higher than that of Step-06 and Step-08. After sorting according to scheduling priority, the candidate task step sequence is: Step-07, Step-06, Step-08. The system passes this sequence to step S104 to execute the resource trial allocation and conflict detection operations on the multi-dimensional virtual time axis model in sequence.

[0041] In step S104, the multidimensional virtual timeline model is used to perform resource allocation for the current candidate task steps, and conflict detection is performed on the time-occupied segments to obtain conflict detection results.

[0042] In one embodiment, step S104 includes: Select the current candidate task step from the candidate task step sequence, and attempt to allocate the future time occupancy segment to the current candidate task step through the multi-dimensional virtual time axis model to obtain the trial allocation time occupancy segment; The trial-allocated time segment is compared one by one with the preset time segment in the multi-dimensional virtual time axis model to determine whether there is any overlap in time intervals; If there is no overlap in the time intervals, it is further determined whether the next task step can obtain a usable resource time interval on the multidimensional virtual time axis model after the current candidate task step is completed, and a conflict detection result is generated.

[0043] In this embodiment, after sorting the candidate task step sequence, the system enters the resource trial allocation and conflict detection phase. Resource trial allocation refers to the system attempting to pre-register the future time occupancy segments required by candidate task steps on a multi-dimensional virtual timeline model without actually occupying physical resources, and verifying whether the registration operation conflicts with existing occupancy records through a conflict detection mechanism, thereby completing a forward-looking verification of resource feasibility in the virtual environment.

[0044] First, from the sorted candidate task step sequence output in step S103, the task step at the top of the sequence is selected as the current candidate task step. Based on the resource requirement description of the current candidate task step, the system constructs a trial allocation time segment on the multi-dimensional virtual timeline model. Specifically, on the virtual timeline of the corresponding resource category, starting from the earliest available time point, a time segment data object containing start time, end time, occupant ID, and priority information is generated according to the resource occupancy duration required by the task step. This object serves as a candidate registration record for this trial resource allocation, i.e., the trial allocation time segment. This process only operates within the memory structure of the virtual timeline model and does not trigger any actual physical scheduling actions for resources.

[0045] After obtaining the trial allocation time segment, the system compares this segment with all pre-set time segments in the multi-dimensional virtual timeline model to determine if there is any overlap in time intervals. Time interval overlap refers to the intersection of the time interval of the trial allocation time segment and the time interval of a pre-set time segment on the same resource dimension; that is, the intervals defined by their start and end times coincide. If the system detects any overlap in time intervals, it determines that there is a direct conflict in the resource trial allocation of the current candidate task step, and the conflict detection result is "fail." The system then passes this result to step S105 to trigger the rollback process.

[0046] If the system confirms that there are no overlapping time intervals after comparing each step, then the direct resource requirements of the current candidate task step do not conflict on the virtual timeline. However, conflict detection does not stop at verifying the current candidate task step itself. The system must further perform a future conflict prediction operation, that is, assuming that the current candidate task step has been successfully completed, predict whether its directly subsequent task steps can also obtain available resource time intervals on the multi-dimensional virtual timeline model. The design logic of this prediction mechanism is that only verifying the resource feasibility of the current step without considering the continuous availability of subsequent steps may cause the system to fall into a waiting or failure state after the current step is completed due to the inability to obtain resources. This will cause serious process loss, especially for uninterrupted critical processes such as atomic tasks. Therefore, based on the absence of overlapping time intervals, the system further checks the availability of resources required by subsequent task steps in the current virtual timeline state. Combining the results of the above two judgments, the final conflict detection result is generated and passed to step S105 to enter the recursive deduction verification stage.

[0047] In a preferred embodiment, the aforementioned laboratory automation scenario is continued. The system selects Step-07, which is at the beginning of the candidate task step sequence, as the current candidate task step. Step-07 is an atomic task that requires robotic arm A to perform a sample grasping operation, with an estimated duration of 10 seconds. On the VT-Processing virtual time axis of the multi-dimensional virtual time axis model, the system constructs a trial allocation time segment for Step-07, starting at time T, with a start time of T and an end time of T+10 seconds. The occupant ID is Step-07, and the priority information is high priority. The system then compares this trial allocation time segment with all the preset time segments on VT-Processing one by one: the existing time segment of the heating position D is from T to T+15 seconds, but its occupant is the heating position D, not robotic arm A, which are different target resources and do not constitute an overlap; robotic arm A has no preset occupant segment on the virtual time axis at time T, so there is no time interval overlap, and the direct conflict detection passes. The system further performs future conflict prediction: After Step-07 is completed, the subsequent step Step-07b needs to occupy conveyor belt B to perform sample transfer at T+10 seconds. The system checks the VT-Transfer virtual time axis and finds that the current occupied segment of conveyor belt B ends at T+8 seconds, meaning that by T+10 seconds, conveyor belt B will be in an idle state, and the subsequent steps can obtain the available resource time interval. Combining the above two judgments, the system generates a conflict detection result of "pass" and passes this result along with the trial allocation time segment to step S105 to perform recursive deduction verification.

[0048] In step S105, the feasibility of the task steps is verified by recursive deduction based on the conflict detection results. If the verification fails, the process rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. This process is repeated until the verification is successful, and the resource scheduling result is output.

[0049] In one embodiment, step S105 includes: When the conflict detection result is passed, the trial allocation time segment is written into the multidimensional virtual time axis model to update the virtual system state; Based on the updated virtual system state, the recursive call scheduling deduction continues to perform resource trial allocation and conflict detection on the next candidate task step of the current candidate task step, and determines whether each subsequent candidate task step can obtain a feasible solution. If the next candidate task step can obtain a feasible solution, then the resource allocation of the current candidate task step is confirmed to be valid, and all the candidate task steps that pass the verification and their corresponding time occupancy segments are output as the resource scheduling result. If the next candidate task step has a deadlock or an unsolvable conflict in the recursive deduction, then the trial allocation time segment corresponding to the current candidate task step is cancelled, and the multidimensional virtual time axis model is restored to the state before the resource trial allocation. Select new candidate task steps from the candidate task step sequence and re-execute the resource allocation, conflict detection, and recursive deduction. Repeat this process until successful verification is achieved and the resource scheduling result is output.

[0050] In this embodiment, after completing resource trial allocation and conflict detection and generating conflict detection results in step S104, the feasibility verification stage based on recursive deduction is entered. This stage adopts the core processing logic of trial-verification-rollback, and performs full-link verification of the resource scheduling feasibility of each step in the candidate task step sequence through recursion, ensuring that the final output resource scheduling result has complete feasibility guarantee in terms of time occupation.

[0051] When the conflict detection result output in step S104 is successful, the trial allocation time segment corresponding to the current candidate task step is formally written into the multi-dimensional virtual timeline model, completing the update operation on the model's memory structure and obtaining the updated virtual system state. The virtual system state refers to the set of all registered time segments recorded by the multi-dimensional virtual timeline model at the current simulation moment, representing a complete description of future resource usage from a virtual simulation perspective. The write operation only acts on the virtual data structure in memory and does not trigger any actual physical resource scheduling actions; therefore, it can be completely revoked without producing any actual side effects in the event of verification failure.

[0052] After updating the virtual system state, a recursive verification operation is performed. Based on the updated virtual system state, the scheduling deduction logic is recursively invoked to sequentially perform resource allocation and conflict detection for the next candidate task step after the current candidate task step, and then the verification process in step S105 is repeated. The core logic of recursive verification is that it assumes the current candidate task step has successfully executed and released resources, updates the virtual system state based on this premise, and then verifies whether each subsequent candidate task step can find a feasible resource time interval under this assumption. Only when each subsequent critical task step finds a feasible solution in the recursive deduction can the resource allocation of the current candidate task step be considered effective from a global scheduling perspective. Through this recursive mechanism, the system can perform forward-looking end-to-end verification of the scheduling decision chain, rather than just verifying the local feasibility of the current single step.

[0053] If all subsequent candidate task steps in the recursive deduction process can obtain feasible solutions, that is, the resource trial allocation of all subsequent steps passes the conflict detection and recursive verification is successful, then the system confirms that the resource trial allocation of the current candidate task step is valid, and summarizes all candidate task steps that have passed the verification in this recursive deduction process and their corresponding time occupancy segments as resource scheduling results output for use in the subsequent actual execution stage.

[0054] If a feasible solution cannot be found for a subsequent candidate task step during the recursive deduction process, i.e., a deadlock or unsolvable conflict exists in the multidimensional virtual timeline model, the system immediately triggers a failure rollback mechanism. Rollback refers to the system revoking the trial allocation time segment corresponding to the current candidate task step, completely restoring the memory state of the multidimensional virtual timeline model to its state before the execution of this resource trial allocation operation, eliminating all impacts of the current allocation decision on the virtual system state, and ensuring the model recovers to a consistent, retryable baseline state. After completing the rollback, the system selects the next candidate task step from the candidate task step sequence and re-executes the complete process of resource trial allocation, conflict detection, and recursive deduction. The above selection-trial allocation-verification-rollback loop will continue to execute until a candidate task step passes full-link recursive verification, and the system outputs the resource scheduling result.

[0055] In a preferred embodiment, the aforementioned laboratory automation scenario continues. Step S104 confirms that the conflict detection result for Step-07 is passed. The system writes the trial allocation time segment (robotic arm A, T to T+10 seconds) corresponding to Step-07 into the multi-dimensional virtual time axis model and updates the virtual system state. The system then recursively calls the scheduling simulation to perform resource trial allocation for the next candidate task step Step-07b (which requires occupying conveyor belt B to perform sample transfer at T+10 seconds, with an estimated duration of 6 seconds). A trial allocation time segment (conveyor belt B, T+10 seconds to T+16 seconds) is constructed on the VT-Transfer virtual time axis. After comparison with the existing occupied segment of conveyor belt B (T to T+8 seconds), it is confirmed that there is no time interval overlap, and the conflict detection passes. The system further recursively simulates Step-07c, which requires occupying reaction container C at T+16 seconds. On the VT-Processing virtual time axis, reaction container C has no registered occupied segment, and the trial allocation passes. At this point, all subsequent key steps associated with Step-07 have found feasible solutions through recursive deduction. The system confirms that the resource allocation trial in Step-07 is effective and summarizes Step-07, Step-07b, Step-07c and their corresponding time-occupied segments as the resource scheduling result output.

[0056] In another scenario, if the recursive deduction reaches Step-07b and discovers that the existing occupied segment of conveyor belt B extends to T+12 seconds, causing the trial allocation time segment of Step-07b (T+10 seconds to T+16 seconds) to overlap with its time interval, the conflict detection fails, and subsequent deductions fall into an unsolvable conflict. The system immediately triggers a rollback mechanism, canceling the trial allocation time segment corresponding to Step-07, restoring the multi-dimensional virtual timeline model to the state before the execution of the trial allocation operation in Step-07, and then selecting the next candidate task step Step-06 from the candidate task step sequence to re-execute the resource trial allocation, conflict detection, and recursive deduction, repeating the process until verification is successful, and outputting the resource scheduling result.

[0057] In one embodiment, the resource scheduling method based on a virtual time axis further includes: When the resource scheduling result cannot be output within the preset time limit, switch to the backup scheduling strategy; Based on the current time window and the urgency of each candidate task step, the most urgent candidate task step is selected for downgrade scheduling in order to output the resource scheduling result. After outputting the resource scheduling result, the consistency of the global state snapshot with the current real-time state information of each target resource is checked. If the check is consistent, the resource scheduling result is executed; if the check is inconsistent, the resource scheduling result is discarded and scheduling is retried.

[0058] In this embodiment, the scheduling architecture adopted in this application designs two parallel scheduling decision paths: a primary strategy (deep deduction) and a backup strategy (degraded operation), to address scheduling requirements under different system load conditions. The primary strategy is the aforementioned recursive deduction scheduling process based on a multi-dimensional virtual time axis model, which verifies the feasibility of candidate task steps through resource trial allocation, conflict detection, and recursive verification to output the optimal resource scheduling result. The backup strategy is a degraded operation mechanism that is automatically activated when the primary strategy cannot complete the deduction calculation within the specified time, ensuring that the system can still maintain basic operation under extreme conditions.

[0059] At the start of each scheduling cycle, the system sets a preset time limit for the main strategy's deduction calculation. The preset time limit refers to the maximum allowable upper bound of the time during which the main strategy can complete all recursive deductions and output resource scheduling results. If the main strategy successfully completes the deduction and outputs resource scheduling results within the preset time limit, the system uses the main strategy's results to proceed to the subsequent execution phase. If the main strategy fails to output resource scheduling results due to excessive recursive deduction depth, too many candidate task steps, or extreme congestion of target resources, resulting in a calculation time exceeding the preset time limit, the system automatically and seamlessly switches to the backup strategy and no longer waits for the main strategy's deduction results.

[0060] After the backup strategy is activated, the system no longer performs deep future projections. Instead, it adopts a degraded scheduling logic based on the current time window and the urgency of tasks. The current time window refers to a finite time range extending into the future from the scheduling cycle trigger time. Within this time window, the system evaluates the urgency of each candidate task step. Urgency refers to a quantitative assessment of the urgency of a step's execution, taking into account factors such as time constraints, waiting time, and task attribute markers. Based on this assessment, the system selects the candidate task step with the highest urgency from the current candidate task steps and performs degraded scheduling on it. That is, without recursive projection verification, the system directly allocates currently available resources to this candidate task step and outputs it as the resource scheduling result for this scheduling cycle. Through this degraded operation mechanism, the system ensures that even in extreme cases where computing resources are limited or target resources are extremely congested, the execution of the most urgent tasks is prioritized, preventing the system from crashing due to scheduling projection timeouts.

[0061] After the primary or backup strategy outputs the resource scheduling result, the system does not immediately execute the result. Instead, it first performs a consistency check. The purpose of the consistency check is to verify whether the global state snapshot on which the current resource scheduling result is based is consistent with the real-time state information of each target resource at the current moment. The design logic of this check mechanism is that the scheduling simulation process has a certain computation time. During this period, the real-time state of each target resource may change due to external disturbances (such as equipment suspension, sudden occupation, etc.). If the scheduling result generated based on the old snapshot is executed directly, it may lead to the actual execution not matching the expected simulation result, thereby causing resource conflicts or task failure.

[0062] Specifically, after outputting the resource scheduling result, the system re-collects the current real-time status information of each target resource and compares it item by item with the global status snapshot generated in step S101. If the two are consistent, that is, the current real-time status of each target resource completely matches the status information recorded in the snapshot, the system determines that the basis for the deduction of this resource scheduling result is still valid, executes the resource scheduling result, and drives each candidate task step to start actual execution according to the allocated time segment. If the two are inconsistent, that is, the current real-time status of one or more target resources deviates from the snapshot record, the system determines that the resource scheduling result has failed, immediately discards the result, and re-triggers a new round of scheduling process, regenerating the global status snapshot based on the latest collected real-time status information, and re-executing the complete scheduling deduction from step S101 to ensure the effectiveness and reliability of the final execution plan.

[0063] In a preferred embodiment, the aforementioned laboratory automation scenario is continued. The system initiates the main strategy deduction during scheduling period T, with a preset time limit of 500 milliseconds. Assuming that the number of task steps to be executed in the current system is large and the target resource occupancy is complex, if the main strategy recursive deduction fails to complete the full-link verification and output resource scheduling results within 500 milliseconds, the preset time limit is triggered. The system automatically switches to the backup strategy and evaluates the current candidate task step set (Step-07, Step-06, Step-08) based on the current time window and urgency: Step-07 is an atomic task, and its corresponding sample processing operation has strict time constraints, with the highest urgency; the system directly allocates currently available resources (robotic arm A, from time T) to Step-07 and outputs the degraded scheduling result. Subsequently, the system collects the current real-time status information of each target resource and performs consistency verification with the global status snapshot: robotic arm A is still in an idle state, consistent with the snapshot record; the remaining time of conveyor belt B changes from 8 seconds in the snapshot record to 6 seconds, which deviates from the snapshot record. If the system determines that the consistency check fails, it immediately discards the current downgraded scheduling result and re-triggers the scheduling process based on the latest real-time status, ensuring that subsequent execution plans are generated based on accurate resource status information and guaranteeing the validity of the scheduling results.

[0064] In summary, this application constructs a multi-dimensional virtual timeline model to structurally model the occupancy of various target resources in the time dimension. It then performs forward-looking resource allocation and time segment conflict detection on candidate task steps in a virtual environment, fundamentally solving the problem of existing scheduling technologies lacking forward-looking allocation mechanisms and recursive feasibility verification capabilities, effectively avoiding deadlocks in scheduling decisions. In a high-concurrency stress test simulating more than 10 concurrent task batches involving dozens of mutually exclusive resource requests in a simulation environment, the scheduling method of this application effectively avoids resource deadlocks, achieving a 100% task completion rate, fully verifying the resource conflict avoidance capability of the multi-dimensional virtual timeline model in complex concurrent scenarios.

[0065] This application employs a recursive simulation mechanism to verify the full-link feasibility of candidate task step sequences. Upon verification failure, a rollback operation is triggered, and new candidate task steps are selected, ensuring the final resource scheduling result is fully feasible in terms of global time consumption. Analysis of the detailed simulation log generated by the system verifies that the backtracking mechanism correctly executes the rollback operation and finds alternative scheduling paths when resource conflicts occur, demonstrating the mechanism's effectiveness. Simultaneously, by setting priority ranking rules for candidate task steps based on atomic tasks and integration steps, this application ensures that critical processes continuously obtain the necessary resources once started. When handling urgent tasks, the order of subsequent task steps can be automatically adjusted. No task timeout failures occurred during testing, effectively protecting the execution integrity of critical task steps with strict time constraints.

[0066] Furthermore, this application designs a dual-strategy backup mechanism combining a primary strategy and a backup strategy. If the primary strategy fails to complete the simulation within the preset time limit due to computation timeout or extreme resource congestion, the system automatically switches to a degraded scheduling strategy based on task urgency, ensuring the system can maintain basic operation even in extreme situations and avoiding downtime. Combined with a consistency verification mechanism after scheduling results are output, the system can verify the consistency between the resource state snapshot and the real-time state before execution, further guaranteeing the effectiveness of the final execution plan.

[0067] Combination Figure 2 As shown, Figure 2 A schematic block diagram of a resource scheduling device based on a virtual time axis provided in an embodiment of the present invention. The resource scheduling device 200 based on a virtual time axis includes: The data acquisition unit 201 is used to collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and the corresponding dependencies. Model building unit 202 is used to build virtual timelines for various target resources based on the global state snapshot, and abstract resource occupancy into time occupancy segments to obtain a multi-dimensional virtual timeline model; The step sorting unit 203 is used to filter candidate task steps that meet the execution conditions from the task step set according to the dependency relationship, and sort them according to a preset priority rule. The conflict detection unit 204 is used to perform resource allocation for the current candidate task steps using the multi-dimensional virtual time axis model, and at the same time to perform conflict detection on the time-occupied segment to obtain the conflict detection result. The loop calculation unit 205 is used to verify the feasibility of the task steps based on the conflict detection results using recursive deduction. If the verification fails, it rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. The loop continues until the verification is successful and the resource scheduling result is output.

[0068] In this embodiment, the data acquisition unit 201 collects real-time status information of each target resource to generate a global status snapshot, and loads the set of task steps to be executed and their corresponding dependencies; the model building unit 202 constructs a virtual timeline for each type of target resource based on the global status snapshot, and abstracts resource occupation into time occupation segments to obtain a multi-dimensional virtual timeline model; the step sorting unit 203 filters candidate task steps with execution conditions from the set of task steps according to the dependencies, and sorts them according to a preset priority rule; the conflict detection unit 204 uses the multi-dimensional virtual timeline model to perform resource trial allocation for the current candidate task steps, and performs conflict detection on the time occupation segments to obtain conflict detection results; the loop calculation unit 205 verifies the feasibility of the task steps based on the conflict detection results using recursive deduction, and rolls back to the resource trial allocation stage when the verification fails, and selects the next candidate task step to re-perform the resource trial allocation, looping until the verification is successful to output the resource scheduling result.

[0069] In one embodiment, the data acquisition unit 201 is specifically used for: At the start of each scheduling cycle, the occupancy status, current position and remaining time of each target resource are collected and integrated to obtain the real-time status information corresponding to each target resource. The real-time status information corresponding to each of the target resources is aggregated and encapsulated to generate the global status snapshot. Read all task steps to be executed, parse the prerequisite dependencies between each task step to establish a task dependency topology, and obtain the set of task steps and their corresponding dependencies.

[0070] In one embodiment, the model building unit 202 is specifically used for: Read the global state snapshot to identify the resource categories of various key resources in the global state snapshot; An independent virtual timeline is constructed for each of the resource categories, and each virtual timeline corresponds to the time occupancy record of each resource category. Each instance of the occupation of the target resources is abstracted into a time occupation segment; By integrating the virtual timelines corresponding to each resource category and the associated time occupancy segments, a multi-dimensional virtual timeline model is obtained.

[0071] In one embodiment, the step sorting unit 203 is specifically used for: Traverse each task step in the set of task steps, and determine whether the prerequisites of each task step have been completed according to the dependency relationship. Select the task steps whose prerequisites have been completed as the candidate task steps. Identify the task steps marked as atomic tasks or integration steps among the candidate task steps, and set the scheduling priority of the atomic task or integration step to be higher than that of the other candidate task steps; All candidate task steps are sorted according to the scheduling priority to obtain a sorted sequence of candidate task steps.

[0072] In one embodiment, the collision detection unit 204 is specifically used for: Select the current candidate task step from the candidate task step sequence, and attempt to allocate the future time occupancy segment to the current candidate task step through the multi-dimensional virtual time axis model to obtain the trial allocation time occupancy segment; The trial-allocated time segment is compared one by one with the preset time segment in the multi-dimensional virtual time axis model to determine whether there is any overlap in time intervals; If there is no overlap in the time intervals, it is further determined whether the next task step can obtain a usable resource time interval on the multidimensional virtual time axis model after the current candidate task step is completed, and a conflict detection result is generated.

[0073] In one embodiment, the loop calculation unit 205 is specifically used for: When the conflict detection result is passed, the trial allocation time segment is written into the multidimensional virtual time axis model to update the virtual system state; Based on the updated virtual system state, the recursive call scheduling deduction continues to perform resource trial allocation and conflict detection on the next candidate task step of the current candidate task step, and determines whether each subsequent candidate task step can obtain a feasible solution. If the next candidate task step can obtain a feasible solution, then the resource allocation of the current candidate task step is confirmed to be valid, and all the candidate task steps that pass the verification and their corresponding time occupancy segments are output as the resource scheduling result. If the next candidate task step has a deadlock or an unsolvable conflict in the recursive deduction, then the trial allocation time segment corresponding to the current candidate task step is cancelled, and the multidimensional virtual time axis model is restored to the state before the resource trial allocation. Select new candidate task steps from the candidate task step sequence and re-execute the resource allocation, conflict detection, and recursive deduction. Repeat this process until successful verification is achieved and the resource scheduling result is output.

[0074] In one embodiment, the resource scheduling device 200 based on a virtual timeline is further configured to: When the resource scheduling result cannot be output within the preset time limit, switch to the backup scheduling strategy; Based on the current time window and the urgency of each candidate task step, the most urgent candidate task step is selected for downgrade scheduling in order to output the resource scheduling result. After outputting the resource scheduling result, the consistency of the global state snapshot with the current real-time state information of each target resource is checked. If the check is consistent, the resource scheduling result is executed; if the check is inconsistent, the resource scheduling result is discarded and scheduling is retried.

[0075] Since the embodiments of the apparatus and the embodiments of the method correspond to each other, please refer to the description of the embodiments of the method for the embodiments of the apparatus, which will not be repeated here.

[0076] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed, can perform the steps provided in the above embodiments. The storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0077] This invention also provides a computer device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the computer device may also include various network interfaces, a power supply, a graphics card, etc., to utilize the graphics card's performance to operate the model, such as for inference and training.

[0078] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

[0079] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A resource scheduling method based on a virtual time axis, characterized in that, include: Collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and their corresponding dependencies; Based on the global state snapshot, a virtual timeline is constructed for each type of target resource, and resource occupation is abstracted into time occupation segments to obtain a multi-dimensional virtual timeline model. Based on the dependencies, candidate task steps that meet the execution conditions are selected from the set of task steps and sorted according to a preset priority rule; The multidimensional virtual timeline model is used to perform resource allocation for the current candidate task steps, and conflict detection is performed on the time-occupied segments to obtain conflict detection results. Based on the conflict detection results, the feasibility of the task steps is verified by recursive deduction. If the verification fails, the process rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. This process is repeated until the verification is successful and the resource scheduling result is output.

2. The resource scheduling method based on a virtual time axis according to claim 1, characterized in that, Also includes: When the resource scheduling result cannot be output within the preset time limit, switch to the backup scheduling strategy; Based on the current time window and the urgency of each candidate task step, the most urgent candidate task step is selected for downgrade scheduling in order to output the resource scheduling result. After outputting the resource scheduling result, the consistency of the global state snapshot with the current real-time state information of each target resource is checked. If the check is consistent, the resource scheduling result is executed; if the check is inconsistent, the resource scheduling result is discarded and scheduling is retried.

3. The resource scheduling method based on a virtual time axis according to claim 1, characterized in that, The process of collecting real-time status information of each target resource to generate a global status snapshot and loading the set of task steps to be executed and their corresponding dependencies includes: At the start of each scheduling cycle, the occupancy status, current position and remaining time of each target resource are collected and integrated to obtain the real-time status information corresponding to each target resource. The real-time status information corresponding to each of the target resources is aggregated and encapsulated to generate the global status snapshot. Read all task steps to be executed, parse the prerequisite dependencies between each task step to establish a task dependency topology, and obtain the set of task steps and their corresponding dependencies.

4. The resource scheduling method based on a virtual time axis according to claim 1, characterized in that, The process involves constructing virtual timelines for various target resources based on the global state snapshot, and abstracting resource usage into time-occupancy segments to obtain a multi-dimensional virtual timeline model, including: Read the global state snapshot to identify the resource categories of various key resources in the global state snapshot; An independent virtual timeline is constructed for each of the resource categories, and each virtual timeline corresponds to the time occupancy record of each resource category. Each instance of the occupation of the target resources is abstracted into a time occupation segment; By integrating the virtual timelines corresponding to each resource category and the associated time occupancy segments, a multi-dimensional virtual timeline model is obtained.

5. The resource scheduling method based on a virtual time axis according to claim 1, characterized in that, The step of filtering candidate task steps that meet the execution conditions from the set of task steps based on the dependency relationship, and sorting them according to a preset priority rule, includes: Traverse each task step in the set of task steps, and determine whether the prerequisites of each task step have been completed according to the dependency relationship. Select the task steps whose prerequisites have been completed as the candidate task steps. Identify the task steps marked as atomic tasks or integration steps among the candidate task steps, and set the scheduling priority of the atomic task or integration step to be higher than that of the other candidate task steps; All candidate task steps are sorted according to the scheduling priority to obtain a sorted sequence of candidate task steps.

6. The resource scheduling method based on a virtual time axis according to claim 5, characterized in that, The process of using the multi-dimensional virtual timeline model to perform resource allocation for the current candidate task steps, and simultaneously performing conflict detection on the time-occupied segments to obtain conflict detection results, includes: Select the current candidate task step from the candidate task step sequence, and attempt to allocate the future time occupancy segment to the current candidate task step through the multi-dimensional virtual time axis model to obtain the trial allocation time occupancy segment; The trial-allocated time segment is compared one by one with the preset time segment in the multi-dimensional virtual time axis model to determine whether there is any overlap in time intervals; If there is no overlap in the time intervals, it is further determined whether the next task step can obtain a usable resource time interval on the multidimensional virtual time axis model after the current candidate task step is completed, and a conflict detection result is generated.

7. The resource scheduling method based on a virtual time axis according to claim 6, characterized in that, The feasibility of the task steps is verified using recursive deduction based on the conflict detection results. If the verification fails, the process rolls back to the resource trial allocation stage, selects the next candidate task step, and re-performs the resource trial allocation. This process is repeated until the verification is successful to output the resource scheduling result, including: When the conflict detection result is passed, the trial allocation time segment is written into the multidimensional virtual time axis model to update the virtual system state; Based on the updated virtual system state, the recursive call scheduling deduction continues to perform resource trial allocation and conflict detection on the next candidate task step of the current candidate task step, and determines whether each subsequent candidate task step can obtain a feasible solution. If the next candidate task step can obtain a feasible solution, then the resource allocation of the current candidate task step is confirmed to be valid, and all the candidate task steps that pass the verification and their corresponding time occupancy segments are output as the resource scheduling result. If the next candidate task step has a deadlock or an unsolvable conflict in the recursive deduction, then the trial allocation time segment corresponding to the current candidate task step is cancelled, and the multidimensional virtual time axis model is restored to the state before the resource trial allocation. Select new candidate task steps from the candidate task step sequence and re-execute the resource allocation, conflict detection, and recursive deduction. Repeat this process until successful verification is achieved and the resource scheduling result is output.

8. A resource scheduling device based on a virtual time axis, characterized in that, include: The data acquisition unit is used to collect real-time status information of each target resource to generate a global status snapshot, and load the set of task steps to be executed and their corresponding dependencies. The model building unit is used to construct virtual timelines for various target resources based on the global state snapshot, and abstract resource occupancy into time occupancy segments to obtain a multi-dimensional virtual timeline model. The step sorting unit is used to filter candidate task steps that meet the execution conditions from the task step set according to the dependency relationship, and sort them according to a preset priority rule. The conflict detection unit is used to perform resource allocation for the current candidate task steps using the multi-dimensional virtual time axis model, and at the same time to perform conflict detection on the time-occupied segments to obtain conflict detection results. The loop calculation unit is used to verify the feasibility of the task steps based on the conflict detection results using recursive deduction. If the verification fails, it rolls back to the resource trial allocation stage and selects the next candidate task step to re-perform the resource trial allocation. The loop continues until the verification is successful and the resource scheduling result is output.

9. A computer device, characterized in that, The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the resource scheduling method based on a virtual time axis as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the resource scheduling method based on a virtual time axis as described in any one of claims 1 to 7.

Images