Resource efficient scheduling method for intelligent network school data intelligent service architecture design

Abstract

The invention relates to the field of computer program control, in particular to an efficient resource scheduling method for intelligent network school data intelligent service architecture design. The method comprises the following steps: acquiring task state data in the intelligent network school data intelligent service architecture, and integrating the task state data into a feedback signal; calculating a resource competition coefficient between tasks based on the feedback signal, and estimating a task resource demand trend at a future moment to obtain a predicted load at the future moment; and performing resource scheduling based on the predicted load at the future moment to obtain an allocation result, introducing a feedback mechanism, dynamically optimizing a scheduling strategy, and forming a closed-loop scheduling process. The problems that a traditional resource scheduling method cannot dynamically capture resource demand changes of tasks, response to sudden high-concurrency requests is slow, and the resource use efficiency is low are solved; quantitative analysis of resource competition relations among tasks is lacked, and resource distribution is not uniform. The problem that the resource reservation is insufficient or the allocation is excessive due to the fact that the dynamic competition characteristic between the tasks cannot be captured and the prediction result is inaccurate is solved.

Description

Technical Field

[0001] The present invention relates to the field of computer program control, and in particular to a resource efficient scheduling method designed for a digital intelligence service architecture of a smart online school. Background Art

[0002] With the development of smart education, smart online schools, as an important form of digital transformation of education, have gradually become an important carrier of teaching activities. Relying on cloud computing, big data and artificial intelligence technologies, smart online schools provide students and teachers with online learning, interactive teaching and intelligent education services anytime and anywhere. However, the service requirements of smart online schools are highly dynamic and complex. For example, the rapid change in the number of users in real-time classrooms, the centralized computing requirements of large-scale online testing tasks, and the high bandwidth occupation of multimedia teaching resources make the underlying resource scheduling face great challenges. In the digital intelligence service architecture of smart online schools, limited resources such as computing resources, storage resources and network bandwidth need to provide support for multiple types of tasks at the same time, including real-time interaction, data processing and storage services. Due to the diversity of task types, significant differences in priorities and frequent changes in resource requirements, traditional static resource allocation strategies are difficult to adapt to this dynamic and complex environment.

[0003] In addition, the task load of smart online schools has obvious timing and volatility. For example, students' sudden high concurrent access behavior may cause resources to be consumed quickly in a short period of time, while resources may be idle during non-peak hours. Therefore, how to achieve efficient allocation and utilization of resources under dynamic load environments has become an urgent problem to be solved in the operation of smart online schools. At the same time, tasks in smart online schools often have different priorities. For example, tasks related to real-time interaction require low latency support, while batch tasks have relatively low requirements for the timeliness of resource allocation. This difference puts higher requirements on the refinement of resource scheduling.

[0004] However, traditional resource scheduling methods have the following technical problems: they are unable to dynamically capture changes in task resource requirements, resulting in slow responses to sudden high-concurrency requests and inefficient resource utilization; they lack quantitative analysis of resource competition relationships between tasks, resulting in uneven resource allocation; they are unable to capture the dynamic competition characteristics between tasks, resulting in inaccurate prediction results, which in turn leads to insufficient resource reservation or over-allocation. Summary of the invention

[0005] The present invention provides a resource efficient scheduling method designed for the digital service architecture of smart online schools, so as to solve the problems that traditional resource scheduling methods cannot dynamically capture changes in resource requirements of tasks, resulting in slow response to sudden high-concurrency requests and inefficient resource utilization; lack of quantitative analysis of the resource competition relationship between tasks, resulting in uneven resource allocation; and cannot capture the dynamic competition characteristics between tasks, resulting in inaccurate prediction results, which in turn leads to the problem of insufficient resource reservation or excessive allocation.

[0006] The present invention provides a resource efficient scheduling method for the design of a smart online school digital intelligence service architecture, which specifically includes the following technical solutions:

[0007] A resource efficient scheduling method designed for a smart online school digital intelligence service architecture includes the following steps:

[0008] S1. Collect task status data in the smart online school digital service architecture and integrate it into feedback signals; based on the feedback signals, calculate the resource competition coefficient between tasks;

[0009] S2. Based on the feedback signal and the resource competition coefficient between tasks, the future load change trend is quantitatively estimated to obtain the predicted load at the future moment; based on the predicted load at the future moment, resource scheduling is performed to obtain the allocation result; based on the allocation result, a feedback mechanism is introduced to dynamically optimize the scheduling strategy to form a closed-loop scheduling process.

[0010] Preferably, the S1 specifically includes:

[0011] The task status data is encapsulated in the form of a time series during the collection process, specifically including the current resource demand, actual resource consumption, task priority, request frequency, resource load and remaining resource amount of each task.

[0012] Preferably, the S1 specifically includes:

[0013] Based on the request frequency in the task status data, the change in resource demand is evaluated; the resource load, change in resource demand and remaining resource amount of the task in the task status data are weighted and integrated to generate a feedback signal. The calculation formula of the feedback signal is as follows:

[0014] F i (t) = α·L i (t-1)+β·D i (t)+γ·R i (t)

[0015] Among them, F i (t) represents the feedback signal of the i-th task at time t; L i (t-1) is the resource load of the ith task at the previous moment; D i(t) is the change in resource demand of the i-th task at the current moment; R i (t) represents the current remaining resources of the i-th task; α is the weight coefficient, which is used to control the influence of the resource load at the previous moment on the current feedback signal; β is the weight coefficient, which is used to control the influence of the change in resource demand at the current moment on the feedback signal; γ is the weight coefficient, which is used to control the influence of the current remaining resources on the feedback signal.

[0016] Preferably, the S1 specifically includes:

[0017] Based on the feedback signal and the resource load in the task status data, the resource competition coefficient between tasks is calculated.

[0018] Preferably, the S2 specifically includes:

[0019] By accumulating historical data of feedback signals and resource competition coefficients between tasks, dynamic load forecasting is performed, and future load change trends are quantitatively estimated to obtain the predicted load at future moments. The specific formula is as follows:

[0020]

[0021] in, represents the predicted load of the i-th task at the future time i+ΔT; ΔT is the predicted time interval; F i (tk) is the feedback signal of the i-th task at time tk, indicating the historical feedback signal; C ij (tk) is the resource competition coefficient between the i-th task and the j-th task at time tk, indicating the historical resource competition coefficient between tasks; k is the index of the time step; m is the time window length, indicating the range of time steps used for prediction; λ k is a weight parameter used to control the influence of historical feedback signals on the predicted load at future moments; μ k It is a weight parameter used to control the influence of the historical resource competition coefficient between tasks on the predicted load at future moments.

[0022] Preferably, the S2 specifically includes:

[0023] Based on the predicted load at the future moment, combined with the feedback signal and the resource competition coefficient between tasks, resources are allocated to obtain the allocation result.

[0024] Preferably, the S2 specifically includes:

[0025] The scheduling strategy parameters are obtained by fitting the feedback signal, the resource competition coefficient between tasks and the predicted load at future moments.

[0026] Preferably, the S2 specifically includes:

[0027] The feedback mechanism compares the actual resource consumption with the allocation results, calculates the scheduling deviation, and introduces a feedback gain coefficient to adjust the scheduling strategy parameters, dynamically optimize the scheduling strategy, and form a closed-loop scheduling process.

[0028] The beneficial effects of the technical solution of the present invention are:

[0029] 1. Through the embedded real-time monitoring module of the program control device, the task status data is comprehensively collected, and the feedback signal is generated by weighted integration of historical resource load, resource demand change and remaining resource. The feedback signal can accurately reflect the current resource demand and resource load of the task, while taking into account the impact of historical load, providing accurate quantitative basis for resource allocation.

[0030] 2. By calculating the resource competition coefficient between tasks, the intensity of resource competition between tasks can be effectively quantified, reflecting the resource competition between high-demand tasks. By analyzing the competition intensity, an adjustment basis is provided for resource allocation, thereby avoiding delays in high-priority tasks due to insufficient resource allocation, and reducing the problem of excessive resource occupation by low-priority tasks.

[0031] 3. Combining the feedback signal and the historical data of the resource competition coefficient between tasks, the future resource demand is accurately estimated. By capturing the dynamic characteristics of task requirements, sufficient computing and storage resources are reserved for resource allocation to prevent sudden task processing failures due to insufficient resources. The resource allocation formula realizes dynamic adjustment of resource allocation by combining feedback signals, predicted loads at future times and the resource competition coefficient between tasks. It can meet current resource needs and future trends at the same time, effectively avoid allocation bottlenecks caused by excessive resource concentration, and ensure efficient use of resources and fair allocation.

[0032] 4. Through the feedback mechanism, the actual resource consumption and allocation results are compared and analyzed, the scheduling strategy parameters for the next round are corrected, a closed-loop scheduling process is formed, the resource allocation effect is continuously optimized, and the overall performance of the resource scheduling method is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of a resource efficient scheduling method designed for the digital service architecture of a smart online school as described in the present invention. DETAILED DESCRIPTION

[0034] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0036] The following is a detailed description of a specific solution of a resource efficient scheduling method designed for a digital service architecture of a smart online school provided by the present invention in conjunction with the accompanying drawings.

[0037] Refer to the attached Figure 1 , which shows a flow chart of a resource efficient scheduling method designed for a smart online school digital intelligence service architecture provided by an embodiment of the present invention, the method comprising the following steps:

[0038] S1. Collect task status data in the smart online school digital service architecture and integrate it into feedback signals; based on the feedback signals, calculate the resource competition coefficient between tasks;

[0039] The service requests of smart online schools are dynamic, such as sudden high concurrent access by students or volatility of real-time computing. Through the embedded real-time monitoring module of the program control device, the task status data in the digital service architecture of smart online schools is comprehensively collected. The collected task status data includes the current resource demand of each task (such as the resource demand of computing resources, storage resources and network bandwidth), actual resource consumption (reflecting the specific consumption of computing resources, storage resources and network bandwidth actually used by the task), task priority (based on the identification weight of the task's importance level (such as critical tasks or tasks with high user frequency)), request frequency (evaluating the load change trend by counting the number of accesses to the task per unit time), resource load (reflecting the overall situation of the actual resources currently occupied by the task) and remaining resources (describing the resource reserves that have been allocated but not yet consumed by the task). All task status data are encapsulated in the form of time series during the collection process to ensure that the subsequent feedback signal calculation can accurately capture the state change history of the task.

[0040] The collected task status data is integrated into feedback signals as a quantitative expression of the task resource requirements at the current moment, which comprehensively measures the resource requirements and resource load of the task at the current moment, so as to provide it for subsequent resource competition modeling and allocation decision-making. The calculation formula of the feedback signal is as follows:

[0041] F i (t) = α·L i (t-1)+β·D i (t)+γ·R i (t)

[0042] Among them, F i (t) represents the feedback signal of the i-th task at time t, indicating the intensity of the task’s demand for resources at the current moment; L i (t-1) is the resource load of the i-th task at the previous moment, indicating the residual impact of the historical resource load on the feedback signal; D i (t) is the change in resource demand of the i-th task at the current moment, which is obtained based on the current request frequency evaluation and reflects the direct impact of the current request on scheduling; R i (t) represents the current remaining resources of the i-th task, which is used to measure whether the task has occupied sufficient resources; α is the weight coefficient, which is used to control the influence of the resource load at the previous moment on the current feedback signal, and is obtained through regression analysis optimization; β is the weight coefficient, which is used to control the influence of the change in resource demand at the current moment on the feedback signal, and is obtained through regression analysis optimization; γ is the weight coefficient, which is used to control the influence of the current remaining resources on the feedback signal, and is calculated by the gradient descent optimization algorithm that minimizes the scheduling error.

[0043] Multiple tasks of the smart online school need to share a limited resource pool (such as computing nodes or storage servers), and resource competition will form between different tasks. Resource competition is the main reason for inefficient resource scheduling; resource competition may cause high-priority tasks to be delayed due to insufficient resource allocation, and may also cause low-priority tasks to occupy too many resources; therefore, the resource competition coefficient between tasks is calculated through the logic modeling module of the program control device to quantify the mutual impact of tasks on resources. The specific calculation formula of the resource competition coefficient is as follows:

[0044]

[0045] Among them, C ij (t) represents the resource competition coefficient between the i-th task and the j-th task at time t, which is used to measure the competition intensity between the two tasks in the shared resource pool; F i (t) and F j (t) is the feedback signal of the i-th task and the j-th task at time t, which respectively represent the resource demand intensity of each task; L i (t) and L j (t) are the resource loads of the i-th task and the j-th task at time t, respectively, describing the resource usage of the tasks.

[0046] S2. Based on the feedback signal and the resource competition coefficient between tasks, the future load change trend is quantitatively estimated to obtain the predicted load at the future moment; based on the predicted load at the future moment, resource scheduling is performed to obtain the allocation result; based on the allocation result, a feedback mechanism is introduced to dynamically optimize the scheduling strategy to form a closed-loop scheduling process.

[0047] After the competition relationship between tasks is clear, dynamic load prediction is performed. The program control device predicts the resource demand trend of tasks at future moments through feedback signals and resource competition coefficients between tasks, and obtains the predicted load at future moments to quantitatively estimate the changes in resource demand in the future period, thereby reserving necessary computing and storage resources for resource allocation; by integrating feedback signals at multiple moments and competition relationships between tasks, the dynamic characteristics of task requirements can be captured. Dynamic load prediction uses feedback signals and resource competition coefficients between tasks as inputs, and the specific formula is as follows:

[0048]

[0049] in, represents the predicted load of the i-th task at the future time t+ΔT, which is used to guide resource allocation decisions; ΔT is the predicted time interval, that is, the delay of the future time point concerned in the prediction relative to the current time t; F i (tk) is the feedback signal of the i-th task at time tk, indicating the historical feedback signal; C ij (tk) is the resource competition coefficient between the i-th task and the j-th task at time tk, indicating the historical resource competition coefficient between tasks; k is the index of the time step; m is the time window length, indicating the range of time steps used for prediction; λ k is a weight parameter used to control the influence of historical feedback signals on the predicted load at future moments, obtained through experimental training; μ k is a weight parameter used to control the influence of the historical resource competition coefficient between tasks on the predicted load at future moments, which is obtained through experimental training. By accumulating historical feedback signals F i (tk) and the historical resource competition coefficient C between tasks ij (tk), quantitatively estimate the future load change trend.

[0050] The predicted load at a future moment is used to guide resource scheduling at the current moment. The core goal of resource allocation is to balance the current resource demand with the predicted load at a future moment, while fully considering the intensity of competition between tasks, so as to achieve efficient and fair resource allocation. The feedback signal provides the intensity of the resource demand of the task at the current moment, and the predicted load at a future moment estimates the future resource demand of the task. The combination of the two ensures that resource allocation not only meets current needs, but also prepares for possible changes in the future. At the same time, the resource competition coefficient between tasks is used as a regulating factor, which reduces the risk of excessive concentration of resources on certain high-demand tasks, thereby avoiding resource bottlenecks. The resource allocation formula is as follows:

[0051]

[0052] Among them, R i (t) is the amount of resources allocated to the i-th task at time t, which represents the allocation result. Its calculation process closely combines the feedback signal and the predicted load at future moments to ensure the rationality of resource allocation. The sum of the resource competition coefficients between tasks in the denominator is used to limit the interference of resource competition between tasks on the allocation result, ensuring the fairness and stability of the overall resource allocation.

[0053] The allocation results will produce certain errors in actual execution, so the next round of scheduling strategy needs to be corrected through the feedback mechanism; the scheduling feedback module calculates the scheduling deviation by comparing the actual resource consumption with the allocation results; based on the scheduling deviation, the next round of scheduling strategy parameters are adjusted through the feedback optimization formula. The feedback optimization formula is:

[0054]

[0055] Among them, S(t+1) is the scheduling policy parameter at time t+1, which represents the scheduling policy parameter after adjustment; S(t) is the scheduling policy parameter at time t, which represents the scheduling policy parameter before adjustment. It is comprehensively generated by the feedback signal of the current round, the resource competition coefficient between tasks, and the predicted load at the future time. The linear expression is obtained by fitting the above three types of data: It is the weight parameter that obtains the best fitting result during the fitting process; is the actual resource consumption of the i-th task at time t, and the current allocation result R i The difference of (t) reflects the scheduling deviation; δ is the feedback gain coefficient, which is used to control the adjustment amplitude. Its value is determined by the actual resource demand and obtained through experiments; N is the total number of tasks.

[0056] In summary, a resource-efficient scheduling method designed for the digital service architecture of smart online schools has been completed.

[0057] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0058] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0059] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A resource efficient scheduling method designed for the digital intelligence service architecture of a smart online school, characterized in that: The following steps are involved: S1. Collect task status data in the smart online school digital service architecture and integrate it into feedback signals; based on the feedback signals, calculate the resource competition coefficient between tasks; S2. Based on the feedback signal and the resource competition coefficient between tasks, the future load change trend is quantitatively estimated to obtain the predicted load at the future time; based on the predicted load at the future time, resource scheduling is performed to obtain the allocation result; Based on the allocation results, a feedback mechanism is introduced to dynamically optimize the scheduling strategy and form a closed-loop scheduling process.

2. According to claim 1, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S1 specifically includes: The task status data is encapsulated in the form of a time series during the collection process, specifically including the current resource demand, actual resource consumption, task priority, request frequency, resource load and remaining resource amount of each task.

3. According to claim 2, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S1 specifically includes: Based on the request frequency in the task status data, the change in resource demand is evaluated; the resource load, change in resource demand and remaining resource amount of the task in the task status data are weighted and integrated to generate a feedback signal. The calculation formula of the feedback signal is as follows: F i (t)=α·L i (t-1)+β·D i (t)+γ·R i (t) Among them, F i (t) represents the feedback signal of the i-th task at time t; L i (t-1) is the resource load of the ith task at the previous moment; D i (t) is the change in resource demand of the i-th task at the current moment; R i (t) represents the current remaining resources of the i-th task; α is the weight coefficient, which is used to control the influence of the resource load at the previous moment on the current feedback signal; β is the weight coefficient, which is used to control the influence of the change in resource demand at the current moment on the feedback signal; γ is the weight coefficient, which is used to control the influence of the current remaining resources on the feedback signal.

4. According to claim 3, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S1 specifically includes: Based on the feedback signal and the resource load in the task status data, the resource competition coefficient between tasks is calculated.

5. According to claim 1, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S2 specifically includes: By accumulating historical data of feedback signals and resource competition coefficients between tasks, dynamic load forecasting is performed, and future load change trends are quantitatively estimated to obtain the predicted load at future moments. The specific formula is as follows: in, represents the predicted load of the i-th task at the future time t+ΔT; ΔT is the predicted time interval; F i (tk) is the feedback signal of the i-th task at time tk, indicating the historical feedback signal; C ij (tk) is the resource competition coefficient between the i-th task and the j-th task at time tk, indicating the historical resource competition coefficient between tasks; k is the index of the time step; m is the time window length, indicating the range of time steps used for prediction; λ k is a weight parameter used to control the influence of historical feedback signals on the predicted load at future moments; μ k It is a weight parameter used to control the influence of the historical resource competition coefficient between tasks on the predicted load at future moments.

6. According to claim 5, a resource efficient scheduling method for the design of a digital intelligence service architecture for a smart online school is characterized in that: The S2 specifically includes: Based on the predicted load at the future moment, combined with the feedback signal and the resource competition coefficient between tasks, resources are allocated to obtain the allocation result.

7. According to claim 6, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S2 specifically includes: The scheduling strategy parameters are obtained by fitting the feedback signal, the resource competition coefficient between tasks and the predicted load at future moments.

8. According to claim 7, a resource efficient scheduling method designed for a smart online school digital intelligence service architecture is characterized in that: The S2 specifically includes: The feedback mechanism compares the actual resource consumption with the allocation results, calculates the scheduling deviation, and introduces a feedback gain coefficient to adjust the scheduling strategy parameters, dynamically optimize the scheduling strategy, and form a closed-loop scheduling process.

Images