Upgrade task scheduling method and device, electronic equipment and storage medium

Abstract

The invention provides an upgrade task scheduling method and device, electronic equipment and a storage medium. The method comprises the steps that real-time state data and upgrade task attribute data uploaded by multiple vehicles are received; generating multi-dimensional dynamic feature information based on the real-time state data and the upgrading task attribute data; based on the multi-dimensional dynamic feature information and an evaluation strategy, determining a priority score of each vehicle upgrading task, the evaluation strategy being dynamically adjusted according to a real-time system load; generating a task distribution plan based on the priority score and the real-time system load; and issuing the task distribution plan to the corresponding vehicle. Through the scheme, the scheduling decision can be closely coordinated with the dynamically changing system operation state, so that the fundamental conversion from static and offline planning to dynamic and online scheduling is realized, and finally, the success rate of the upgrading task and the overall utilization efficiency of system resources are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of automotive electronics and software technology, and in particular to an upgrade task scheduling method, apparatus, electronic device and storage medium. Background Technology

[0002] With the rapid development of intelligent connected vehicles, in-vehicle software systems are becoming increasingly complex. Over-the-Air (OTA) technology for remotely updating vehicle firmware and software has become a crucial means of ensuring vehicle safety and functionality. An efficient OTA upgrade system is of great significance for automakers in reducing recall costs and improving customer satisfaction.

[0003] In existing technologies, OTA upgrade systems typically employ a centralized task scheduling architecture. The cloud server, based on a preset upgrade strategy (e.g., grouping vehicles by static attributes such as vehicle model, region, or software version), generates and distributes upgrade tasks in batches during fixed time windows (e.g., off-peak network periods). Upon receiving the upgrade command, the vehicle performs the download and installation operations and reports the final result to the cloud. This scheduling method, based on fixed rules and offline planning, achieves basic automated management of upgrade tasks.

[0004] However, when faced with a large-scale, highly dynamic vehicle operating environment, the scheduling decisions of the above methods are out of sync with the actual system operating status and cannot be adaptively adjusted in real time, making it difficult to effectively guarantee the success rate of upgrade tasks and the efficiency of system resource utilization. Summary of the Invention

[0005] This application provides an upgraded task scheduling method, apparatus, electronic device, and storage medium to solve the problem in the prior art that when facing a large-scale, highly dynamic vehicle operating environment, the scheduling decision is out of sync with the actual system operating state and cannot be adaptively adjusted in real time.

[0006] Firstly, this application provides an upgrade task scheduling method, including: Receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; Based on the real-time status data and the upgrade task attribute data, multi-dimensional dynamic feature information is generated; Based on the multidimensional dynamic feature information and evaluation strategy, the priority score of each vehicle upgrade task is determined, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. Based on the priority score and the real-time system load, a task distribution plan is generated; The task distribution plan is then sent to the corresponding vehicles.

[0007] In one possible implementation, generating multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data includes: Based on the real-time status data, a vehicle group status matrix is ​​constructed; Based on the upgrade task attribute data, the upgrade urgency of each vehicle is determined; The upgrade urgency is fused with the vehicle group state matrix to obtain a digital model describing the state of the vehicle group, and the multi-dimensional dynamic feature information is extracted from the digital model.

[0008] In one possible implementation, the method further includes: During the process of generating a task distribution plan based on the priority score and the real-time system load, the constraints related to the real-time system load and the multi-dimensional dynamic feature information are continuously verified. If any vehicle fails to match a task due to not meeting the constraints, the task matching process is backtracked, and subsequent tasks in the task distribution plan are matched based on the backtracked process status, the current priority score, and the real-time system load.

[0009] In one possible implementation, the method further includes: Obtain the current weight coefficients of the multiple feature dimensions that constitute the multidimensional dynamic feature information; Based on the changes in the real-time system load, the current weight coefficients of each feature dimension are adjusted to determine the priority score of each vehicle upgrade task using the adjusted weights.

[0010] In one possible implementation, generating a task distribution plan based on the priority score and the real-time system load includes: The initial length of the rolling time window is determined based on the total number of tasks to be upgraded in the real-time system load and the real-time network bandwidth. Based on the real-time network health status in the real-time system load, the initial length is dynamically adjusted to obtain the actual length of the rolling time window; Within a rolling time window of the actual length, batch scheduling tasks are generated based on the priority score, and the scheduling tasks generated for each batch are organized into the task distribution plan.

[0011] In one possible implementation, the method further includes: During the process of generating batch scheduling tasks based on the priority score, the system bandwidth utilization and regional task density in the real-time system load are continuously monitored. If the system bandwidth utilization or the regional task density is detected to exceed a preset threshold, a dynamic traffic splitting mechanism is triggered, marking at least one of the lowest priority tasks in the current task queue as to be delayed for execution.

[0012] In one possible implementation, the method further includes: Receive task execution status data fed back by the vehicle after executing the task distribution plan; Based on the task execution status data, the evaluation strategy is optimized, and the optimized evaluation strategy is applied to the priority scoring and determination process of subsequent upgrade tasks.

[0013] Secondly, this application provides an upgraded task scheduling device, comprising: The receiving module is used to receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; The first generation module is used to generate multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data; The determination module is used to determine the priority score of each vehicle upgrade task based on the multi-dimensional dynamic feature information and the evaluation strategy, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. The second generation module is used to generate a task distribution plan based on the priority score and the real-time system load; The task distribution module is used to distribute the task distribution plan to the corresponding vehicles.

[0014] In one possible implementation, the first generation module is specifically used for: Based on the real-time status data, a vehicle group status matrix is ​​constructed; Based on the upgrade task attribute data, the upgrade urgency of each vehicle is determined; The upgrade urgency is fused with the vehicle group state matrix to obtain a digital model describing the state of the vehicle group, and the multi-dimensional dynamic feature information is extracted from the digital model.

[0015] In one possible implementation, the device further includes a backtracking module for: During the process of generating a task distribution plan based on the priority score and the real-time system load, the constraints related to the real-time system load and the multi-dimensional dynamic feature information are continuously verified. If any vehicle fails to match a task due to not meeting the constraints, the task matching process is backtracked, and subsequent tasks in the task distribution plan are matched based on the backtracked process status, the current priority score, and the real-time system load.

[0016] In one possible implementation, the device further includes an adjustment module for: Obtain the current weight coefficients of the multiple feature dimensions that constitute the multidimensional dynamic feature information; Based on the changes in the real-time system load, the current weight coefficients of each feature dimension are adjusted to determine the priority score of each vehicle upgrade task using the adjusted weights.

[0017] In one possible implementation, the second generation module is specifically used for: The initial length of the rolling time window is determined based on the total number of tasks to be upgraded in the real-time system load and the real-time network bandwidth. Based on the real-time network health status in the real-time system load, the initial length is dynamically adjusted to obtain the actual length of the rolling time window; Within a rolling time window of the actual length, batch scheduling tasks are generated based on the priority score, and the scheduling tasks generated for each batch are organized into the task distribution plan.

[0018] In one possible implementation, the apparatus further includes an execution module for: During the process of generating batch scheduling tasks based on the priority score, the system bandwidth utilization and regional task density in the real-time system load are continuously monitored. If the system bandwidth utilization or the regional task density is detected to exceed a preset threshold, a dynamic traffic splitting mechanism is triggered, marking at least one of the lowest priority tasks in the current task queue as to be delayed for execution.

[0019] In one possible implementation, the device further includes an optimization module for: Receive task execution status data fed back by the vehicle after executing the task distribution plan; Based on the task execution status data, the evaluation strategy is optimized, and the optimized evaluation strategy is applied to the priority scoring and determination process of subsequent upgrade tasks.

[0020] Thirdly, this application provides an apparatus comprising: a processor and a memory, wherein the processor is configured to execute an upgrade task scheduler stored in the memory to implement the upgrade task scheduling method described in any one of the first aspects.

[0021] Fourthly, this application provides a storage medium storing one or more programs that can be executed by one or more processors to implement the upgrade task scheduling method described in any one aspect.

[0022] Compared with the prior art, the technical solution provided in this application has the following advantages: The method provided in this application first receives and integrates real-time vehicle status and upgrade task attributes, providing a comprehensive dynamic data foundation for decision-making; then, based on this dynamic information, it generates feature information reflecting the real-time state of the system, achieving precise quantitative perception of the "system operating status"; subsequently, it introduces an evaluation strategy that can be dynamically adjusted according to the real-time system load to determine task priorities, enabling scheduling decisions to adapt to environmental changes and break free from the constraints of fixed rules; finally, it generates and issues a task distribution plan based on the dynamic priority and the real-time system load, ensuring real-time matching between scheduling decisions and the actual carrying capacity of the current system. Through this solution, scheduling decisions can be closely coordinated with the dynamically changing system operating status, thereby achieving a fundamental shift from static, offline planning to dynamic, online scheduling, ultimately significantly improving the success rate of upgrade tasks and the overall utilization efficiency of system resources. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0026] Figure 1 A flowchart illustrating an embodiment of an upgrade task scheduling method provided in this application; Figure 2 A flowchart illustrating another embodiment of the upgrade task scheduling method provided in this application; Figure 3 A flowchart illustrating another embodiment of the upgrade task scheduling method provided in this application; Figure 4 A block diagram illustrating an embodiment of an upgrade task scheduling device provided in this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0029] To address the technical problem that existing technologies fail to adapt to real-time adaptive adjustments when dealing with large-scale, highly dynamic vehicle operating environments, as scheduling decisions are disconnected from the actual system operating state. This application provides an upgrade task scheduling method that enables scheduling decisions to closely coordinate with the dynamically changing system operating state, thereby achieving a fundamental shift from static, offline planning to dynamic, online scheduling. Ultimately, this significantly improves the success rate of upgrade tasks and the overall utilization efficiency of system resources.

[0030] Figure 1 This is a flowchart illustrating an embodiment of an upgrade task scheduling method provided in this application. Figure 1 As shown, the method includes the following steps: Step 101: Receive real-time status data and upgrade task attribute data uploaded from multiple vehicles.

[0031] Real-time status data refers to the set of vehicle operating parameters collected in real time, including but not limited to network signal strength, geographical location, remaining battery power, CPU (Central Processing Unit) load, and memory usage.

[0032] Upgrade task attribute data: refers to the set of attribute parameters related to the software update task, including the security vulnerability level, functional difference coefficient, and vendor mandatory upgrade strategy weight based on the CVE (Common Vulnerabilities and Exposures) standard.

[0033] In this embodiment, the vehicle continuously uploads real-time status data to the cloud via a secure communication channel. Simultaneously, the cloud retrieves upgrade task attribute data for each vehicle from the software version management database, forming a complete data input set. All data is asynchronously received and buffered via a distributed message queue, ensuring data integrity and processing efficiency in high-concurrency scenarios.

[0034] Step 102: Generate multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data.

[0035] Multidimensional dynamic feature information: refers to the standardized feature vector obtained after feature engineering, including but not limited to dimensions such as network quality score, vehicle availability score, and task urgency score.

[0036] In this embodiment, the cloud first cleans and normalizes the received real-time status data to construct a dynamically updated vehicle group status matrix (a set of vehicle group status data organized in matrix form, where rows correspond to individual vehicles and columns correspond to different status indicators). Simultaneously, based on upgrade task attribute data, a lightweight inference model is used to calculate the upgrade urgency of each vehicle (a derived feature value calculated by the lightweight inference model, derived from the security vulnerability level of the vehicle's current software version, functional difference coefficients, and the manufacturer's forced upgrade strategy). Finally, the upgrade urgency of each vehicle is used as a derived feature value to form a new feature column, which is then concatenated into the original vehicle group status matrix. This constitutes a digital model describing the vehicle group status that simultaneously covers both real-time vehicle status and business upgrade requirements. Standardized multi-dimensional dynamic feature information is extracted from this model for subsequent priority evaluation.

[0037] Step 103: Based on the multi-dimensional dynamic feature information and evaluation strategy, determine the priority score of each vehicle upgrade task, wherein the evaluation strategy is dynamically adjusted according to the real-time system load.

[0038] Evaluation strategy: refers to the calculation strategy that defines the weight coefficients of each feature dimension and their dynamic adjustment rules, and is implemented using a weight-based multi-objective optimization algorithm mechanism.

[0039] Priority score: A quantitative value of task priority calculated using a weighted summation model.

[0040] Real-time system load: refers to the current processing pressure and resource usage of the cloud system, including indicators such as network bandwidth utilization and the size of the task queue to be scheduled.

[0041] In this embodiment, during system initialization, initial weight coefficients are assigned to each feature dimension constituting the multidimensional dynamic feature information. During score calculation, the online rule engine continuously monitors real-time system load metrics. When the overall system load exceeds a preset threshold, the weight coefficients of network quality-related features are automatically increased. Finally, based on the dynamically adjusted weights, a weighted calculation is performed on the multidimensional dynamic features of each vehicle to generate a priority score matrix. Specifically, each vehicle corresponds to a multidimensional feature vector. The system uses the latest weight coefficients of each dimension as the weighted vector and calculates the vehicle's score value through dot product operations (i.e., summing the corresponding feature values ​​multiplied by their corresponding weights). This weighted calculation is performed in parallel on all vehicles in the vehicle group to obtain the score value for each vehicle. Then, the score values ​​of all vehicles are aggregated to form a priority score matrix with vehicles as rows and score values ​​as columns. This matrix presents the global priority ranking of all vehicles to be upgraded, providing a direct basis for subsequent batch scheduling.

[0042] Step 104: Generate a task distribution plan based on the priority score and the real-time system load.

[0043] Task distribution plan: refers to the final set of dispatch instructions generated based on task matching results. It specifies the matching relationship between upgrade tasks and vehicles, as well as the corresponding execution sequence (such as batches, time windows) and resource allocation.

[0044] In this embodiment, the scheduling task manager first calculates the initial length of the elastic rolling time window (i.e., the task execution time period) based on the total number of tasks to be upgraded and the real-time network bandwidth. Subsequently, the system dynamically adjusts this initial length based on the real-time network health status, with the adjustment range controlled within ±30% of the initial length. During task matching, the system matches target vehicles for each upgrade task in descending order of priority score. During each matching, multi-dimensional constraints related to the target vehicle are verified in real-time, including its network connection status, remaining battery power, and online status. When a constraint failure is detected, causing the current task matching to fail, the system automatically triggers a backtracking mechanism, reverting to the most recent successful matching state, and retrying to match the current and subsequent tasks based on the latest vehicle status, priority score, and system load.

[0045] Meanwhile, the system continuously monitors global load metrics. When bandwidth utilization exceeds a threshold or regional task density exceeds its corresponding threshold, a dynamic traffic splitting strategy is automatically triggered, postponing the K lowest-priority tasks to the next time window. The entire process employs an adaptive scheduling mechanism with state backtracking capabilities. When changes in constraints or system overload are detected, the system can automatically adjust matching paths or task queues, thereby generating a robust task distribution plan.

[0046] Step 105: Distribute the task distribution plan to the corresponding vehicles.

[0047] In this embodiment, the cloud distributes the generated task distribution plan to each target vehicle via a TLS (Transport Layer Security) encrypted channel. After receiving the plan, the vehicle performs the download and installation operations according to the time window and upgrade parameters specified in the plan. During execution, the vehicle transmits download progress, execution status, and error information back to the cloud in real time, providing a data foundation for subsequent strategy optimization.

[0048] The technical solution provided in this application first receives and integrates real-time vehicle status and upgrade task attributes, providing a comprehensive dynamic data foundation for decision-making. Then, based on this dynamic information, it generates feature information reflecting the real-time state of the system, achieving precise quantitative perception of the "system operating status." Subsequently, it introduces an evaluation strategy that can dynamically adjust according to real-time system load to determine task priorities, enabling scheduling decisions to adapt to environmental changes and break free from the constraints of fixed rules. Finally, it generates and distributes task distribution plans based on this dynamic priority and real-time system load, ensuring real-time matching between scheduling decisions and the actual carrying capacity of the current system. Through this solution, scheduling decisions are closely coordinated with the dynamically changing system operating status, thereby achieving a fundamental shift from static, offline planning to dynamic, online scheduling, ultimately significantly improving the success rate of upgrade tasks and the overall utilization efficiency of system resources.

[0049] Figure 2 A flowchart illustrating another embodiment of the upgrade task scheduling method provided in this application. Figure 2 The process shown is in Figure 1 Based on the illustrated process, the following steps are included: Step 201: Construct a vehicle group status matrix based on the real-time status data.

[0050] Vehicle group status matrix: refers to a set of vehicle group status data organized in matrix form, where rows of the matrix correspond to individual vehicles and columns correspond to different real-time status indicators, including network signal strength, vehicle location, remaining battery power, and system resource load.

[0051] In this embodiment, after receiving real-time status data from multiple vehicles, the cloud scheduling engine performs data aggregation and feature extraction based on a distributed message queue, transforming the scattered vehicle information into a unified and computable digital model. Specifically, the multi-source status information is aggregated and cleaned through a stream processing framework to generate a dynamically updated matrix with vehicles as rows and status indicators as columns. The construction of this matrix provides a digital foundation for subsequent global optimization, realizing a unified representation and real-time tracking of the status of the vehicle group.

[0052] Step 202: Based on the upgrade task attribute data, determine the upgrade urgency of each vehicle.

[0053] Upgrade Urgency: This refers to a derived feature value generated by a lightweight inference model. It is calculated by combining the security vulnerability level of the current software version of the vehicle, the functional difference coefficient, and the manufacturer's mandatory upgrade policy. The calculation formula is: Urgency = f(CVE_Level, Function_Gap, Policy_Weight), where: Urgency: refers to the upgrade urgency, used to characterize the urgency of performing a software upgrade update for a specific vehicle; f(): refers to the specific calculation function or inference model, which can be a weighted summation, logistic regression, or other lightweight machine learning model; CVE_Level: refers to the security vulnerability level, which is an assessment value of the severity of security vulnerabilities existing in the current software version of the vehicle based on the CVE standard; Function_Gap: refers to the functional difference coefficient, which is a value used to quantify the degree of difference in functionality and performance between the target upgrade version and the current software version of the vehicle; Policy_Weight: refers to the policy weight, which is a mandatory priority coefficient set by the manufacturer for the upgrade task based on business strategy, regulatory requirements, or operational needs.

[0054] In this embodiment, a lightweight inference model is used to process the upgrade task attribute data. The security vulnerability level is evaluated based on the CVE standard, the functional difference coefficient quantifies the degree of functional difference between the old and new versions, and the vendor's forced upgrade strategy reflects business priority. The system obtains a numerical upgrade urgency score through weighted calculation. This score accurately describes the urgency of the software update and provides a key decision basis for subsequent priority assessment.

[0055] Step 203: Fuse the upgrade urgency with the vehicle group state matrix to obtain a digital model describing the state of the vehicle group, and extract the multi-dimensional dynamic feature information from the digital model.

[0056] Digital Model: This refers to an enhanced structured data matrix formed by concatenating the upgrade urgency as a new feature column with the vehicle group status matrix. The rows of this matrix correspond to vehicles, and the columns simultaneously contain their real-time operational status indicators and core business attributes.

[0057] Multidimensional dynamic feature information: refers to the standardized feature vector extracted from the digital model, which includes dimensions such as network quality, vehicle availability, and task urgency.

[0058] In this embodiment, the calculated upgrade urgency is dynamically updated to the vehicle group state matrix as a new feature column, forming a digital vehicle group model that simultaneously includes real-time operating status and business attributes. The standardized multi-dimensional dynamic feature vector is obtained from this model through a feature extraction engine. These features include both original state indicators and derived business features, providing a comprehensive and quantitative decision-making basis for subsequent dynamic scheduling.

[0059] Figure 2 The process shown demonstrates how a digital model integrating a vehicle group status matrix with upgrade urgency is constructed, transforming raw data into intelligent features. This process preserves the real-time operating status of vehicles while incorporating business logic understanding, enabling subsequent scheduling decisions to possess both environmental and business awareness capabilities. This lays a solid foundation for achieving precise dynamic scheduling.

[0060] In another embodiment of this application, the method may further include the following steps: during the process of generating a task distribution plan based on the priority score and the real-time system load, continuously verifying the constraints related to the real-time system load and the multi-dimensional dynamic feature information; if any vehicle fails to match a task due to not meeting the constraints, then backtrack the task matching process and match subsequent tasks in the task distribution plan based on the backtracked process status, the current priority score, and the real-time system load.

[0061] Constraints: These refer to a set of preset thresholds or status requirements that the target vehicle must meet during the task matching process. If any information such as the vehicle's current network connection status, remaining battery power, system load, or geographical location does not meet the preset constraints, the task matching will directly fail.

[0062] Task matching failed: This means that the current task cannot be successfully matched with a suitable vehicle because the constraints are not met.

[0063] Backtracking the task matching process: refers to a self-scheduling mechanism based on a nondeterministic state machine that rolls back the task matching state to the position of the most recent successful match.

[0064] Process status: refers to the snapshot of the scheduling status saved during the task matching process, including information such as the list of matched vehicles and the queue of tasks to be processed.

[0065] This embodiment introduces a self-scheduling mechanism based on a nondeterministic state machine during the task matching phase. When any vehicle fails to match a task due to not meeting the constraints, the system automatically backtracks to the stable state of the most recent successful match, instead of recalculating globally. By combining the process state of the backtracking point, the latest priority score, and the real-time system load, only the affected suffix tasks are locally rescheduled. This mechanism significantly reduces computational overhead and task latency while ensuring scheduling accuracy.

[0066] To facilitate understanding, the task matching process is illustrated below with a specific example: Step 1: Sequence matching and state recording Vehicles are matched to tasks in order of priority. After successfully matching vehicle A to task T1 and vehicle B to task T2, a "success checkpoint" is created to fully save the scheduling snapshot at this moment (including the confirmed assignments T1->A and T2->B).

[0067] Step 2: Real-time verification failure triggers backtracking When matching vehicle C for task T3 in sequence, real-time verification finds that vehicle C fails to meet the constraints (such as the vehicle's battery level being lower than the safety threshold), and task T3 matching is determined to have failed.

[0068] Step 3: Intelligent Backtracking and Local Replanning Backtracking: The system automatically rolls back to the previous "success checkpoint," that is, the stable state after the allocations T1->A and T2->B are completed. These two allocation results are locked and will not be changed in subsequent replanning.

[0069] Replanning: Based on the latest system status at this time (including the latest data of other available vehicles, the latest priority score, and real-time system load), the system re-initiates the matching process for all subsequent unassigned tasks starting from T3 (such as T3, T4, T5).

[0070] This solution introduces an intelligent backtracking mechanism, which enables rapid fault tolerance and local self-healing during task scheduling. It effectively solves the problem of overall scheduling plan failure caused by dynamic environmental changes, and significantly improves the robustness and response efficiency of the system while ensuring scheduling quality.

[0071] In another embodiment, the method may further include the following steps: obtaining the current weight coefficients of multiple feature dimensions constituting the multidimensional dynamic feature information; adjusting the current weight coefficients of each feature dimension according to the changes in the real-time system load, so as to determine the priority score of each vehicle upgrade task using the adjusted weights.

[0072] Feature dimension: refers to the specific feature components that constitute multidimensional dynamic feature information, including network quality score, vehicle availability score, task urgency score, system load score, and geographical distance score.

[0073] Current weight coefficients: refer to the weight parameters assigned to each feature dimension in the evaluation model, including network quality weight α, vehicle availability weight β, task urgency weight γ, system load weight δ, and geographical distance weight ε, etc.

[0074] Real-time system load changes: refers to the dynamic changes in the overall system load rate, especially the status changes when the load rate exceeds a preset threshold (such as 75%).

[0075] This embodiment uses an online rule engine to dynamically monitor the overall system load rate. When the load rate exceeds a preset threshold, it automatically increases the weight coefficient of the network quality feature dimension (e.g., increases the network quality weight α by 15%), and adjusts the weights of other feature dimensions accordingly. This dynamic weight adjustment mechanism ensures that the system prioritizes network stability when the load is high, and maintains a balanced consideration of each dimension when the load is normal, thus achieving adaptive optimization of the evaluation strategy for the system's operating status.

[0076] This embodiment establishes a linkage mechanism between weighting coefficients and system load, enabling the priority evaluation strategy to have environmental adaptability. It can prioritize the stability of the upgrade process under high load and maintain the balance of evaluation dimensions under normal load, thereby maintaining the optimal scheduling decision quality under different system states.

[0077] Figure 3 A flowchart illustrating another embodiment of the upgrade task scheduling method provided in this application. Figure 3 The process shown is in Figure 1 Based on the illustrated process, the following steps are included: Step 301: Determine the initial length of the rolling time window based on the total number of tasks to be upgraded in the real-time system load and the real-time network bandwidth.

[0078] Total number of upgrade tasks: refers to the number of upgrade tasks currently waiting to be scheduled in the system.

[0079] Real-time network bandwidth: refers to the available bandwidth capacity of the communication link between the cloud and the vehicle.

[0080] Rolling time window: refers to a dynamically adjustable time interval used to execute upgrade tasks in batches.

[0081] This embodiment determines the baseline scale of the time window through quantitative calculation. Specifically, the initial length is calculated using the formula L_base = Task_Volume / (Avg_Bandwidth * 0.8), where L_base is the initial length of the rolling time window; Task_Volume is the total number of tasks to be upgraded; Avg_Bandwidth is the average real-time network bandwidth; and 0.8 is a safety factor for reserving bandwidth buffers. This calculation method ensures that the initial time window setting matches the actual processing capacity of the system, establishing an accurate baseline for subsequent dynamic adjustments.

[0082] Step 302: Based on the real-time network health status in the real-time system load, dynamically adjust the initial length to obtain the actual length of the rolling time window.

[0083] Real-time network health status: refers to the stability and quality indicators of network links, including parameters such as transmission delay, packet loss rate, and signal fluctuation.

[0084] Actual length: refers to the final executable time window length after dynamic adjustment.

[0085] This embodiment continuously monitors network quality indicators and makes flexible adjustments within a range of ±30% based on the initial length. When the network condition is good, the window is appropriately expanded to improve throughput, and when the network fluctuates, the window is contracted to ensure the stability of task execution, thus achieving real-time adaptation of the time window size to the network condition.

[0086] Step 303: Within the rolling time window of the actual length, generate batch scheduling tasks according to the priority score, and organize the scheduling tasks generated for each batch into the task distribution plan.

[0087] Batch scheduling tasks: refers to a set of upgrade tasks that are executed in parallel within the same time window, and their generation follows the priority ordering principle.

[0088] In this embodiment, under a defined time window constraint, upgrade tasks are sorted and selected based on priority scores, and batch tasks to be executed within the current window are generated in descending order. The system ensures that task execution matches the system's current processing capacity through a dynamic resource evaluation mechanism, and automatically activates task scheduling optimization strategies when resource constraints are identified. Finally, the generated batch scheduling tasks are organized and arranged according to the time window sequence to form a complete task distribution plan.

[0089] Figure 3The process shown combines dynamic time window management with priority scheduling to achieve orderly batch execution and systematic organization of upgrade tasks. This ensures timely processing of high-priority tasks and guarantees system stability through a resource awareness mechanism. The resulting task distribution plan has better executability and resource adaptability, effectively improving the overall efficiency and reliability of upgrade scheduling.

[0090] In another embodiment of this application, the method may further include the following steps: during the process of generating batch scheduling tasks based on the priority score, continuously monitoring the system bandwidth utilization rate and regional task density in the real-time system load; if the system bandwidth utilization rate or the regional task density is detected to exceed a preset threshold, a dynamic diversion mechanism is triggered to mark at least one task with the lowest priority in the current task queue to be scheduled as delayed execution.

[0091] System bandwidth utilization: refers to the proportion of cloud egress bandwidth currently in use, reflecting the degree of network link congestion.

[0092] Regional task density: refers to the concentration of upgrade tasks to be performed within a unit geographical area, used to identify local network hotspots.

[0093] Dynamic traffic distribution mechanism: refers to a traffic control strategy that automatically adjusts the task execution order based on system load.

[0094] Delayed execution: refers to the operation state of removing a specific task from the current execution queue and rescheduling it to a later time window for execution.

[0095] In this embodiment, during the task matching phase, the system monitors the system bandwidth utilization rate and the regional task density based on geographic location analysis in real time. When any indicator exceeds a preset threshold, a dynamic traffic splitting mechanism is immediately triggered. The system automatically calculates the number of tasks that need to be delayed according to the formula K = ceil(Overflow_Rate * Queue_Length, where K is the number of tasks that need to be delayed; Overflow_Rate is the proportion of bandwidth utilization exceeding the threshold; Queue_Length is the length of the current queue of tasks to be scheduled; and ceil is a rounding function). It then selects the K lowest priority tasks from the current queue of tasks to be scheduled and marks them as delayed tasks, and reintegrates them into the scheduling queue of the subsequent rolling time window. This mechanism effectively prevents the spread of system overload and regional network congestion.

[0096] This embodiment establishes a dynamic traffic splitting mechanism with dual thresholds, which enables intelligent peak shaving of system load peaks and timely dispersal of regional hotspots. This not only ensures the smooth execution of high-priority tasks, but also maintains the overall stability of the system through task delay strategies, significantly improving the service reliability of the upgrade system under sudden high load scenarios.

[0097] In another embodiment of this application, the method may further include the following steps: receiving task execution status data fed back by the vehicle after executing the task distribution plan; optimizing the evaluation strategy based on the task execution status data; and applying the optimized evaluation strategy to the priority scoring determination process of subsequent upgrade tasks.

[0098] Task execution status data: refers to the process and result data generated by the vehicle during the execution of the upgrade task, including task success rate, download latency, bandwidth utilization, and network load fluctuations.

[0099] Evaluation strategy optimization: refers to the process of iteratively updating the weight parameters in the priority evaluation model based on the reinforcement learning policy engine.

[0100] This embodiment establishes a vehicle-cloud closed-loop feedback channel to continuously collect task execution status data from vehicles, including indicators such as task success rate and download latency. Based on a reinforcement learning policy engine, using upgrade success rate and network load balancing as reward signals, a gradient descent algorithm is used to fine-tune and incrementally update the weight parameters in the evaluation strategy online. The optimized weight parameters are verified and saved to the weight model, and then applied to the priority evaluation process in subsequent scheduling cycles, forming a continuously optimized "learning-execution-feedback" closed loop.

[0101] This embodiment introduces an online optimization mechanism based on reinforcement learning, enabling the system to learn autonomously from historical execution experience and continuously improve scheduling strategies. This achieves dynamic evolution and adaptive adjustment of evaluation model parameters, effectively improving the system's adaptability and decision-making accuracy under different operating environments, and ultimately promoting a continuous improvement in upgrade success rate and resource utilization efficiency.

[0102] Figure 4 This is a block diagram illustrating an embodiment of an upgrade task scheduling device provided in this application. Figure 4 As shown, the device includes: The receiving module 41 is used to receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; The first generation module 42 is used to generate multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data; The determination module 43 is used to determine the priority score of each vehicle upgrade task based on the multi-dimensional dynamic feature information and the evaluation strategy, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. The second generation module 44 is used to generate a task distribution plan based on the priority score and the real-time system load; The task distribution module 45 is used to distribute the task distribution plan to the corresponding vehicles.

[0103] In one possible implementation, the first generation module is specifically used for: Based on the real-time status data, a vehicle group status matrix is ​​constructed; Based on the upgrade task attribute data, the upgrade urgency of each vehicle is determined; The upgrade urgency is fused with the vehicle group state matrix to obtain a digital model describing the state of the vehicle group, and the multi-dimensional dynamic feature information is extracted from the digital model.

[0104] In one possible implementation, the device further includes a backtracking module for: During the process of generating a task distribution plan based on the priority score and the real-time system load, the constraints related to the real-time system load and the multi-dimensional dynamic feature information are continuously verified. If any vehicle fails to match a task due to not meeting the constraints, the task matching process is backtracked, and subsequent tasks in the task distribution plan are matched based on the backtracked process status, the current priority score, and the real-time system load.

[0105] In one possible implementation, the device further includes an adjustment module for: Obtain the current weight coefficients of the multiple feature dimensions that constitute the multidimensional dynamic feature information; Based on the changes in the real-time system load, the current weight coefficients of each feature dimension are adjusted to determine the priority score of each vehicle upgrade task using the adjusted weights.

[0106] In one possible implementation, the second generation module is specifically used for: The initial length of the rolling time window is determined based on the total number of tasks to be upgraded in the real-time system load and the real-time network bandwidth. Based on the real-time network health status in the real-time system load, the initial length is dynamically adjusted to obtain the actual length of the rolling time window; Within a rolling time window of the actual length, batch scheduling tasks are generated based on the priority score, and the scheduling tasks generated for each batch are organized into the task distribution plan.

[0107] In one possible implementation, the apparatus further includes an execution module for: During the process of generating batch scheduling tasks based on the priority score, the system bandwidth utilization and regional task density in the real-time system load are continuously monitored. If the system bandwidth utilization or the regional task density is detected to exceed a preset threshold, a dynamic traffic splitting mechanism is triggered, marking at least one of the lowest priority tasks in the current task queue as to be delayed for execution.

[0108] In one possible implementation, the device further includes an optimization module for: Receive task execution status data fed back by the vehicle after executing the task distribution plan; Based on the task execution status data, the evaluation strategy is optimized, and the optimized evaluation strategy is applied to the priority scoring and determination process of subsequent upgrade tasks.

[0109] like Figure 5 As shown in the figure, this application provides a device including a processor 111, a communication interface 112, a memory 113, and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114. Memory 113 is used to store computer programs; In one embodiment of this application, when the processor 111 executes a program stored in the memory 113, it implements the upgrade task scheduling method provided in any of the foregoing method embodiments, including: Receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; Based on the real-time status data and the upgrade task attribute data, multi-dimensional dynamic feature information is generated; Based on the multidimensional dynamic feature information and evaluation strategy, the priority score of each vehicle upgrade task is determined, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. Based on the priority score and the real-time system load, a task distribution plan is generated; The task distribution plan is then sent to the corresponding vehicles.

[0110] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the upgrade task scheduling method provided in any of the foregoing method embodiments.

[0111] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0112] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0113] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0114] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An upgrade task scheduling method, characterized in that, The method includes: Receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; Based on the real-time status data and the upgrade task attribute data, multi-dimensional dynamic feature information is generated; Based on the multidimensional dynamic feature information and evaluation strategy, the priority score of each vehicle upgrade task is determined, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. Based on the priority score and the real-time system load, a task distribution plan is generated; The task distribution plan is then sent to the corresponding vehicles.

2. The method according to claim 1, characterized in that, The generation of multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data includes: Based on the real-time status data, a vehicle group status matrix is ​​constructed; Based on the upgrade task attribute data, the upgrade urgency of each vehicle is determined; The upgrade urgency is fused with the vehicle group state matrix to obtain a digital model describing the state of the vehicle group, and the multi-dimensional dynamic feature information is extracted from the digital model.

3. The method according to claim 1, characterized in that, The method further includes: During the process of generating a task distribution plan based on the priority score and the real-time system load, the constraints related to the real-time system load and the multi-dimensional dynamic feature information are continuously verified. If any vehicle fails to match a task due to not meeting the constraints, the task matching process is backtracked, and subsequent tasks in the task distribution plan are matched based on the backtracked process status, the current priority score, and the real-time system load.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the current weight coefficients of the multiple feature dimensions that constitute the multidimensional dynamic feature information; Based on the changes in the real-time system load, the current weight coefficients of each feature dimension are adjusted to determine the priority score of each vehicle upgrade task using the adjusted weights.

5. The method according to claim 1, characterized in that, The step of generating a task distribution plan based on the priority score and the real-time system load includes: The initial length of the rolling time window is determined based on the total number of tasks to be upgraded in the real-time system load and the real-time network bandwidth. Based on the real-time network health status in the real-time system load, the initial length is dynamically adjusted to obtain the actual length of the rolling time window; Within a rolling time window of the actual length, batch scheduling tasks are generated based on the priority score, and the scheduling tasks generated for each batch are organized into the task distribution plan.

6. The method according to claim 5, characterized in that, The method further includes: During the process of generating batch scheduling tasks based on the priority score, the system bandwidth utilization and regional task density in the real-time system load are continuously monitored. If the system bandwidth utilization or the regional task density is detected to exceed a preset threshold, a dynamic traffic splitting mechanism is triggered, marking at least one of the lowest priority tasks in the current task queue as to be delayed for execution.

7. The method according to claim 1, characterized in that, The method further includes: Receive task execution status data fed back by the vehicle after executing the task distribution plan; Based on the task execution status data, the evaluation strategy is optimized, and the optimized evaluation strategy is applied to the priority scoring and determination process of subsequent upgrade tasks.

8. An upgraded task scheduling device, characterized in that, The device includes: The receiving module is used to receive real-time status data and upgrade task attribute data uploaded from multiple vehicles; The first generation module is used to generate multi-dimensional dynamic feature information based on the real-time status data and the upgrade task attribute data; The determination module is used to determine the priority score of each vehicle upgrade task based on the multi-dimensional dynamic feature information and the evaluation strategy, wherein the evaluation strategy is dynamically adjusted according to the real-time system load. The second generation module is used to generate a task distribution plan based on the priority score and the real-time system load; The task distribution module is used to distribute the task distribution plan to the corresponding vehicles.

9. An electronic device, characterized in that, include: A processor and a memory, the processor being configured to execute an upgrade task scheduler stored in the memory to implement the upgrade task scheduling method according to any one of claims 1-7.

10. A storage medium, characterized in that, The storage medium stores one or more programs, which can be executed by one or more processors to implement the upgrade task scheduling method according to any one of claims 1-7.

Images