Multi-level intelligent distribution method and system for full-performance detection test tasks

Abstract

The invention discloses a multi-level intelligent distribution method and system for full-performance detection test tasks. The method comprises the following steps: acquiring a full-performance detection test task, analyzing the full-performance detection test task to obtain basic information, performing task feature mapping on the basic information to obtain a task feature vector, and performing adaptive clustering on the task according to the task feature vector to generate a task cluster; all available resource information is detected, resource capacity vectors are obtained based on the resource information, layering is carried out, different types of resource pools are obtained, the adaptation degree of task clusters and the resource pools is calculated, and the task clusters are distributed to the optimal resource pool based on the adaptation degree; and after all the task clusters are allocated to the resource pool, performing multi-level task allocation on the tasks in the task clusters based on task priorities, and optimizing a multi-level task allocation result through reinforcement learning to realize multi-level intelligent allocation of the full-performance detection test tasks. According to the scheme, the task scheduling efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of task scheduling, and specifically relates to a multi-level intelligent allocation method and system for full-performance testing tasks. Background Technology

[0002] With the continuous development of information technology and computing power, the demand for computing resources has increased dramatically, especially in fields such as cloud computing, big data, and artificial intelligence. Task scheduling and resource management have become key technologies for improving system performance. In traditional computing resource management, task scheduling algorithms mostly use static or preset rules to allocate tasks, ignoring the complex relationship between tasks and resources. This static allocation method often proves inadequate when facing diverse and dynamically changing task requirements, leading to uneven resource allocation, low resource utilization, and long system response times.

[0003] In existing technologies, scheduling algorithms focus on static resource management without fully considering the diversity and dynamic changes in task requirements. For example, simple round-robin scheduling algorithms, priority scheduling algorithms, and time-slice-based scheduling strategies, while effective in certain specific environments, cannot effectively handle complex situations such as uneven resource load and unclear inter-task relationships. Furthermore, traditional methods fail to consider historical task execution data and dynamic resource changes, thus affecting the accuracy and efficiency of system scheduling when handling complex computational tasks. In addition, existing technologies suffer from insufficient task information analysis leading to decreased allocation efficiency and poor task-resource matching resulting in low resource utilization. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a multi-level intelligent allocation method and system for full-performance testing tasks, thereby solving the technical problems of decreased allocation efficiency and low task and resource utilization.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0006] This invention first discloses a multi-level intelligent allocation method for full-performance testing tasks, which includes the following steps: Step S1: Obtain the full performance testing task, parse the full performance testing task to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters; Step S2: Detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. Calculate the fit between task clusters and resource pools, and allocate task clusters to the optimal resource pool based on the fit. Step S3: After allocating all task clusters to the resource pool, multi-level task allocation is performed on the tasks in the task clusters based on task priority, and the multi-level task allocation results are optimized through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks.

[0007] The present invention further includes the following preferred embodiments: The process of parsing the full-performance testing task to obtain basic information further includes: The acquired tasks are parsed using structured task parsing algorithms, unstructured task data parsing algorithms, rule-based task feature extraction, and task dependency parsing algorithms to obtain basic task information, including task complexity, priority, execution time, dependencies, environmental constraints, resource affinity, and historical task feedback.

[0008] The step of adaptively clustering tasks based on the task feature vectors to generate task clusters further includes: The task feature vector is normalized to obtain a normalized task feature vector representation. A nonlinear topological similarity metric calculation formula is then introduced.

[0009] in, It is a task and tasks The nonlinear topological similarity between tasks is used to measure how close two tasks are in the task feature space. It is the dimension of the task feature vector; It is a task feature index; It is the first Weighting coefficients for each task feature; It is a task In the Standardized features on dimensional features; It is a task In the Standardized features on dimensional features; It is the first Exponential scaling parameters for each task feature; It is the scaling factor for overall similarity calculation.

[0010] The step of adaptively clustering tasks based on the task feature vectors to generate task clusters further includes: By considering the similarity, dependency weights, and computational complexity correlations among comprehensive tasks, a comprehensive task similarity is defined, and initial cluster centers are determined by constructing a task association graph.

[0011] The resource pool is divided according to computing power, storage capacity, or sensors.

[0012] The formula for calculating the fit between the computing task cluster and the resource pool is as follows:

[0013] in, It is a task cluster With resource pool The compatibility between them; It is the first weighting coefficient, used to adjust the weighted similarity. Weights in fit calculation; It is the first The weights of each feature; It is the second weighting factor, used to adjust for the difference term. Weights in fit calculation; It represents the number of tasks in the task cluster; It is a task cluster The Middle The task in the first Task characteristics across multiple dimensions; It is the first in the resource pool One feature; Norm calculation; It is a constant.

[0014] The optimization of multi-level task allocation results through reinforcement learning further includes: First, the Q-table is initialized, and a reinforcement learning model is trained based on historical task execution data to predict the execution effect of different scheduling strategies. Then, during task scheduling, the Q value is calculated in real time, and the optimal action is selected to adjust the task scheduling strategy. After the task is completed, the Q value is updated, and the task scheduling decision for the next round is optimized. After multiple rounds of training, the optimal task scheduling strategy is learned.

[0015] This invention also discloses a multi-level intelligent allocation system for full-performance testing tasks, utilizing the aforementioned multi-level intelligent allocation method for full-performance testing tasks, comprising: The task cluster generation module is used to acquire full-performance testing test tasks, parse the full-performance testing test tasks to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters. The task cluster allocation module is used to detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. It calculates the fit between task clusters and resource pools, and allocates task clusters to the optimal resource pool based on the fit. The multi-level allocation module is used to allocate tasks in the task clusters in a multi-level manner based on task priority after all task clusters are allocated to the resource pool. It also optimizes the multi-level task allocation results through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks.

[0016] Accordingly, this application also discloses a terminal, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the multi-level intelligent allocation method based on the aforementioned full-performance testing task.

[0017] Accordingly, this application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned multi-level intelligent allocation method for the full-performance testing task.

[0018] The beneficial effects of this invention are as follows: Compared with the prior art, this invention provides a multi-level intelligent allocation method and system for full-performance testing tasks. Through deep analysis of task characteristics and adaptive clustering algorithms, tasks are allocated to appropriate resource pools more accurately. Comprehensive consideration of task priority, complexity, dependencies, and environmental adaptability makes task allocation more aligned with actual needs, improving overall task scheduling efficiency. An adaptive task cluster adjustment mechanism is introduced, dynamically adjusting the cluster to which a task belongs based on historical feedback or characteristic changes (such as changes in task execution success rate or computational requirements). This adaptive capability can cope with constantly changing task loads, improving robustness and flexibility. Utilizing a task clustering method based on nonlinear topological similarity measurement, the similarity between tasks can be measured more accurately, especially when task features are discretely distributed, maintaining high clustering accuracy and solving the problem that traditional methods cannot effectively handle nonlinear complex relationships. By introducing task-resource pool adaptability calculation, the adaptability between tasks and resources can be accurately evaluated in a real-time dynamic environment, ensuring timely and efficient task execution and avoiding task execution delays caused by resource mismatch. Attached Figure Description

[0019] Figure 1 This is a flowchart of the multi-level intelligent allocation method for full-performance testing tasks in this invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0021] The embodiments described in this application are merely some, not all, embodiments of the present invention. Based on the spirit of the present invention, other embodiments obtained by those skilled in the art without inventive effort are all within the protection scope of the present invention.

[0022] To address the shortcomings of existing technologies, this invention proposes a multi-level intelligent allocation method and system for full-performance testing tasks. The multi-level intelligent allocation method for full-performance testing tasks includes the following steps: Step S1: Obtain the full performance testing task, parse the full performance testing task to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters; Step S2: Detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. Calculate the fit between task clusters and resource pools, and allocate task clusters to the optimal resource pool based on the fit. Step S3: After allocating all task clusters to the resource pool, multi-level task allocation is performed on the tasks in the task clusters based on task priority, and the multi-level task allocation results are optimized through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks. Preferably, step S1 specifically includes: The system acquires full-performance testing tasks and uses various parsing techniques, including structured task parsing algorithms (such as JSON and XML parsing), unstructured task data parsing algorithms (natural language processing), rule-based task feature extraction (such as regular expressions and keyword mapping), and task dependency parsing (such as task dependency graphs), to parse the acquired tasks and obtain basic task information, such as task complexity, priority, execution time, dependencies, environmental constraints, resource affinity (whether it is compatible with specific devices), and historical task feedback. Then, it performs task feature mapping on the basic task information, quantifying the basic task information into feature vectors.

[0023] In a further preferred embodiment, the task feature vector is represented as:

[0024] in, It is a task The feature vector contains all the basic information about the task; Indicates the first One task; It is a task The task priority indicates the urgency and scheduling priority of the task. The larger the value, the more important the task is and the higher it should be scheduled. (1 = low, 2 = medium, 3 = high); It is a task Task complexity describes the amount of computational resources required for the task, including CPU / GPU / FPGA computing requirements, parallel computing requirements, etc., and is determined based on expert experience. This indicates low complexity, such as using only a single-core CPU; Indicates medium complexity, such as when parallel computation is required; This indicates high complexity, such as GPU computing and FPGA computing; It is a task Task dependencies, i.e., whether a task needs to wait for a preceding task to complete, or whether it has subsequent dependent tasks, are determined through a task dependency graph and execution logic. For example, if a task... If there is no dependency, then If the task If it depends on one task, then If the task If it depends on multiple tasks, then ( (Number of dependent tasks); It is a task Environmental constraint adaptability, i.e., whether the task requires a specific execution environment, such as temperature, humidity, electromagnetic interference, etc. Temperature required ~ If the current environment meets the requirements, then ;Task Temperature required However, in the current environment ,but ; It is a task Resource affinity, i.e., whether a task is suitable for a specific type of computing resource, such as high-precision sensors, GPU-accelerated computing, high-bandwidth storage, etc., such as task GPU computation is required; currently available GPUs are available. ,Task FPGA computation is required, but the current FPGA is unavailable. ; It is a task Historical execution feedback, namely, statistical data such as the success rate, failure rate, and average execution time of tasks in the past.

[0025] Preferably, step S1 specifically includes: Based on the task feature vector, a nonlinear dynamic task clustering algorithm is used to adaptively cluster tasks and generate task clusters. The nonlinear dynamic task clustering algorithm combines nonlinear topological similarity measurement, dynamic task feature transfer, and adaptive task cluster evolution to ensure the accuracy of task allocation and dynamic adaptability.

[0026] In the implementation of the nonlinear dynamic task clustering algorithm, to eliminate the influence of data scale and improve the stability of clustering, the task feature vectors are normalized to obtain a normalized task feature vector representation. After normalizing the task feature vectors, a nonlinear topological similarity metric is introduced to more accurately measure the similarity between tasks.

[0027]

[0028] in, It is a task and tasks The nonlinear topological similarity between tasks is used to measure how close two tasks are in the task feature space. It is the dimension of the task feature vector; It is a task feature index; It is the first The weighting coefficients of each task feature, which measure the contribution of the feature to the overall task similarity, are determined based on expert experience. It is a task In the Standardized features on dimensional features; It is a task In the Standardized features on dimensional features; It is the first The exponential scaling parameter for each task feature is used to adjust the nonlinear weights of different features and is determined based on expert experience. This is a scaling factor for overall similarity calculation, used to balance the similarity contributions between different features. It was obtained experimentally and can be set to 2. The aforementioned nonlinear topological similarity metric can more accurately describe the complex nonlinear relationships between tasks and maintains high computational accuracy even when the task feature distribution is relatively discrete.

[0029] Preferably, step S1 specifically includes: In the implementation of the nonlinear dynamic task clustering algorithm, the dependency weights and computational complexity correlations between tasks are calculated. A comprehensive task similarity is defined by combining the similarity, dependency weights, and computational complexity correlations between tasks. After calculating the task similarity, initial cluster centers are determined by constructing a task association graph.

[0030] First, calculate the dependency weights between tasks. :

[0031] in, and Representing tasks and Adaptability to the execution environment and This indicates the type of resources required by the task, and the dependency weights mentioned above are used to measure the degree of dependency between tasks. Simultaneously, the computational complexity correlation between tasks is calculated. :

[0032] in, and These represent the computational requirements of the tasks, and the introduction of the exponential function ensures that tasks with significantly different computational complexities are not grouped into the same cluster.

[0033] Finally, by considering the similarity, dependency weights, and computational complexity correlations between the tasks, a comprehensive task similarity is defined. :

[0034] After calculating the similarity of the comprehensive tasks, the task with the maximum connectivity is used as the initial cluster center. First, the task similarity is calculated. connectivity :

[0035] Select the one with the highest connectivity Each task is used as an initial cluster center to ensure that the initial centers can better represent the overall distribution of the task space.

[0036] After the initial cluster center selection is completed, the dynamic task clustering process begins. In this stage, each task is assigned to its most similar cluster center, and task calculations are performed. To the current cluster center correlation :

[0037] in, It is the current cluster center. Indicates task Similarity to cluster centers. If the task... Maximum correlation Less than the threshold set by expert experience If the optimal assignment is not found, the assignment will be recalculated to improve the robustness of task allocation.

[0038] Finally, after all tasks are categorized, multiple task clusters (TCTC) are generated, and the final assignment of a task is determined by the following formula:

[0039] in, Represents a task cluster, It is a cluster center.

[0040] Preferably, step S1 specifically includes: In the implementation of nonlinear dynamic task clustering algorithms, an adaptive adjustment mechanism for task clusters is introduced to cope with the dynamic changes in task characteristics. This mechanism considers the historical success rate of a particular task. A decrease of more than 10%, or calculation of demand If the magnitude of the change exceeds the threshold set based on expert experience, the task similarity needs to be recalculated and the task may need to be migrated to a new task cluster.

[0041] Preferably, step S2 specifically includes: The system detects resource information such as type, status, and specifications of all available resources, and standardizes the detected resource information, converting it into a unified format for subsequent processing.

[0042] The standardized resource information is used to obtain a resource capability vector, which is then transformed into a resource capability vector and defined as follows: Any one of these elements can be used It means that the first Resource capabilities; such as computing power, storage capacity, bandwidth, etc. This indicates the total number of resource types.

[0043] Preferably, step S2 specifically includes: Based on resource capability vectors, resources with different performance and characteristics are allocated to different resource pools according to expert experience, which facilitates efficient scheduling. Then, all resources are sorted according to their capability vectors and divided into multiple pools according to their capability size.

[0044] Pooling criteria include: computing power pooling (high-performance computing resource pool, medium-performance computing resource pool, low-performance computing resource pool); storage capacity pooling (high-performance storage pool (SSD), conventional storage pool (HDD), etc.); sensor pooling (high-precision sensor pool, low-precision sensor pool). The resource pools are as follows: computing resource pools sort all computing resources according to computing power vectors, allocating resources with stronger computing power to higher-level pools; storage resource pools allocate storage with fast read / write speeds to higher-level pools and slower storage to lower-level pools; sensor resource pools allocate high-precision sensors to upper-level pools and low-precision sensors to lower-level pools.

[0045] Preferably, step S2 specifically includes: A comprehensive fit calculation formula is introduced to calculate the fit between task clusters and resource pools. Based on the fit, a multi-objective optimization algorithm is used to allocate task clusters to the optimal resource pool under the premise of load balancing and optimal resource utilization.

[0046] The formula for calculating overall fit is as follows:

[0047] in, It refers to the adaptability, representing the task cluster. With resource pool The compatibility between them; It is the first weighting coefficient, used to adjust the weighted similarity. The weights in the fit calculation are determined based on expert experience. It is the first The weight of the i-th feature represents the weight of the i-th feature in the task feature vector. The importance of each feature; It is the second weighting factor, used to adjust for the difference term. The weights in the fit calculation are determined based on expert experience. It represents the number of tasks in a task cluster, indicating the task cluster. The number of tasks included; It is a task cluster The Middle The task in the first Task characteristics across multiple dimensions; It is the first in the resource pool One feature; Norm calculation; It is a small constant used to avoid division by zero errors in calculations and to ensure that the denominator in the calculation of difference terms is never zero.

[0048] Finally, based on the adaptability, a multi-objective optimization algorithm is used to allocate task clusters to the optimal resource pool under the premise of load balancing and optimal resource utilization.

[0049] Preferably, step S3 specifically includes: After selecting a resource pool for a task cluster, within the same task cluster, the task priorities are redefined based on expert experience and the original task priorities are replaced. The tasks in the task cluster are then sorted in descending order according to the redefined task priorities to obtain a preliminary task execution order list. Furthermore, a multi-level intelligent task allocation mechanism is introduced to intelligently allocate tasks from the initial task execution order list. Specifically, this includes a primary scheduling layer, an intermediate scheduling layer, and a high-level scheduling layer.

[0050] The primary scheduling layer is responsible for scheduling low-priority, low-complexity tasks. These tasks are directly assigned to a lightly loaded resource pool for execution. If the resource pool reaches a predetermined load threshold, the low-priority, low-complexity tasks will be postponed or wait in an idle resource pool. The primary scheduling algorithm employs a fast allocation strategy based on task priority, ensuring efficient execution of simple tasks.

[0051] The intermediate scheduling layer handles tasks with high priority and moderate computational complexity, making scheduling decisions to avoid resource waste. The intermediate scheduling layer uses a graph-based optimization task scheduling algorithm. Each task execution allocation in each resource pool is treated as a node in a graph. By solving the shortest path problem of the graph, the route of task allocation is optimized to minimize task execution time.

[0052] The high-level scheduling layer is responsible for scheduling high-priority tasks with extremely high computational complexity. It optimizes task execution strategies through reinforcement learning (such as deep Q-networks). By combining historical data from tasks and resource pools, it learns from negative factors such as task failures and delays to adjust scheduling strategies. The state space of reinforcement learning includes information such as the task queue, resource pool state, and task priorities; the action space includes operations such as task migration, priority adjustment, and resource reallocation; and the reward function is optimized based on task success rate, task latency, and resource utilization.

[0053] In particular, the low, high, and high priority of the task, as well as the low, moderate, and extremely high computational complexity, are determined based on expert experience combined with the specific scenario. By optimizing multi-level intelligent task allocation through reinforcement learning, a multi-level intelligent allocation of full-performance testing tasks is achieved. The reinforcement learning optimization task scheduling process is as follows: First, a Q-table is initialized, and a reinforcement learning model is trained based on historical task execution data to predict the execution effect of different scheduling strategies. Then, during task scheduling, the Q-value is calculated in real time, and the optimal action is selected to adjust the task scheduling strategy. After the task is completed, the Q-value is updated, and the next round of task scheduling decisions is optimized. After multiple rounds of training, the optimal task scheduling strategy can be learned, improving the task execution success rate and resource utilization.

[0054] Specifically, once a task enters the execution phase, its execution status is monitored in real time, including progress, resource usage, and anomaly detection. If a task fails, its priority is automatically adjusted or it is migrated to a new resource pool for re-execution. After a task is completed, execution data is collected, including success rate, completion time, and resource utilization efficiency. This data is then input into a reinforcement learning model to optimize future task scheduling strategies. Through continuous iteration and optimization, the intelligence of task scheduling is continuously improved, ensuring efficient task execution.

[0055] The beneficial effects of this invention are as follows: Compared with existing technologies, this invention provides a multi-level intelligent allocation method and system for full-performance testing tasks. Through deep analysis of task characteristics and adaptive clustering algorithms, tasks are allocated to appropriate resource pools more accurately. Comprehensive consideration of task priority, complexity, dependencies, and environmental adaptability makes task allocation more aligned with actual needs, improving overall task scheduling efficiency. An adaptive task cluster adjustment mechanism is introduced, dynamically adjusting the cluster to which a task belongs based on historical feedback or characteristic changes (such as changes in task execution success rate or computational requirements). This adaptive capability enables the handling of constantly changing task loads, improving robustness and flexibility. Utilizing a task clustering method based on nonlinear topological similarity measurement, the similarity between tasks can be measured more accurately, especially when task features are discretely distributed, maintaining high clustering accuracy and solving the problem that traditional methods cannot effectively handle nonlinear complex relationships. By introducing task-resource pool adaptability calculation, the adaptability between tasks and resources can be accurately evaluated in a real-time dynamic environment, ensuring timely and efficient task execution and avoiding task execution delays caused by resource mismatch.

[0056] This invention can be a system, method, and / or computer program product. This invention also discloses a multi-level intelligent allocation system for full-performance testing tasks based on the aforementioned multi-level intelligent allocation method for full-performance testing tasks, comprising: The task cluster generation module is used to acquire full-performance testing test tasks, parse the full-performance testing test tasks to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters. The task cluster allocation module is used to detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. It calculates the fit between task clusters and resource pools, and allocates task clusters to the optimal resource pool based on the fit. The multi-level allocation module is used to allocate tasks in the task clusters in a multi-level manner based on task priority after all task clusters are allocated to the resource pool. It also optimizes the multi-level task allocation results through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks.

[0057] Based on the spirit of this invention, those skilled in the art will readily conceive that a computer program product can be obtained based on the aforementioned multi-level intelligent allocation method for the full-performance testing task. The computer program product may include a computer-readable storage medium on which computer-readable program instructions are loaded to enable a processor to implement various aspects of this disclosure. That is, this application also includes a terminal comprising a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the aforementioned multi-level intelligent allocation method for the full-performance testing task.

[0058] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0059] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0060] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A multi-level intelligent allocation method for full-performance testing tasks, characterized in that, Includes the following steps: Step S1: Obtain the full performance testing task, parse the full performance testing task to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters; Step S2: Detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. Calculate the fit between task clusters and resource pools, and allocate task clusters to the optimal resource pool based on the fit. Step S3: After allocating all task clusters to the resource pool, multi-level task allocation is performed on the tasks in the task clusters based on task priority, and the multi-level task allocation results are optimized through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks.

2. The multi-level intelligent allocation method for full-performance testing tasks according to claim 1, characterized in that, The process of parsing the full-performance testing task to obtain basic information further includes: The acquired tasks are parsed using structured task parsing algorithms, unstructured task data parsing algorithms, rule-based task feature extraction, and task dependency parsing algorithms to obtain basic task information, including task complexity, priority, execution time, dependencies, environmental constraints, resource affinity, and historical task feedback.

3. The multi-level intelligent allocation method for full-performance testing tasks according to claim 2, characterized in that, The step of adaptively clustering tasks based on the task feature vectors to generate task clusters further includes: The task feature vector is normalized to obtain a normalized task feature vector representation. A nonlinear topological similarity metric calculation formula is then introduced. in, It is a task and tasks The nonlinear topological similarity between tasks is used to measure how close two tasks are in the task feature space. It is the dimension of the task feature vector; It is a task feature index; It is the first Weighting coefficients for each task feature; It is a task In the Standardized features on dimensional features; It is a task In the Standardized features on dimensional features; It is the first Exponential scaling parameters for each task feature; It is the scaling factor for overall similarity calculation.

4. The multi-level intelligent allocation method for full-performance testing tasks according to claim 3, characterized in that, The step of adaptively clustering tasks based on the task feature vectors to generate task clusters further includes: By considering the similarity, dependency weights, and computational complexity correlations among comprehensive tasks, a comprehensive task similarity is defined, and initial cluster centers are determined by constructing a task association graph.

5. The multi-level intelligent allocation method for full-performance testing tasks according to claim 4, characterized in that, The resource pool is divided according to computing power, storage capacity, or sensors.

6. The multi-level intelligent allocation method for full-performance testing tasks according to claim 5, characterized in that, The formula for calculating the fit between the computing task cluster and the resource pool is as follows: in, It is a task cluster With resource pool The compatibility between them; It is the first weighting coefficient, used to adjust the weighted similarity. Weights in fit calculation; It is the first The weights of each feature; It is the second weighting factor, used to adjust for the difference term. Weights in fit calculation; It represents the number of tasks in the task cluster; It is a task cluster The Middle The task in the first Task characteristics across multiple dimensions; It is the first in the resource pool One feature; Norm calculation; It is a constant.

7. The multi-level intelligent allocation method for full-performance testing tasks according to claim 6, characterized in that, The optimization of multi-level task allocation results through reinforcement learning further includes: First, the Q-table is initialized, and a reinforcement learning model is trained based on historical task execution data to predict the execution effect of different scheduling strategies. Then, during task scheduling, the Q value is calculated in real time, and the optimal action is selected to adjust the task scheduling strategy. After the task is completed, the Q value is updated, and the task scheduling decision for the next round is optimized. After multiple rounds of training, the optimal task scheduling strategy is learned.

8. A multi-level intelligent allocation system for full-performance testing tasks, characterized in that, include: The task cluster generation module is used to acquire full-performance testing test tasks, parse the full-performance testing test tasks to obtain basic information, perform task feature mapping on the basic information to obtain task feature vectors, and perform adaptive clustering of tasks based on the task feature vectors to generate task clusters. The task cluster allocation module is used to detect all available resource information, obtain resource capability vectors based on the resource information, and perform layering to obtain different types of resource pools. It calculates the fit between task clusters and resource pools, and allocates task clusters to the optimal resource pool based on the fit. The multi-level allocation module is used to allocate tasks in the task clusters in a multi-level manner based on task priority after all task clusters are allocated to the resource pool. It also optimizes the multi-level task allocation results through reinforcement learning to achieve multi-level intelligent allocation of full-performance testing tasks.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the multi-level intelligent allocation method for full-performance testing tasks according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the multi-level intelligent allocation method for the full-performance testing task as described in any one of claims 1-7.

Images