A static parameter adjustment method and device of a data processing task and a storage medium

Abstract

This invention relates to the field of data processing technology, and in particular to a method, device, and storage medium for adjusting static parameters of a data processing task. The method includes: acquiring the cluster resource utilization rate and basic information of the data processing task; obtaining the baseline values ​​of various static parameters required by the data processing task based on a static parameter template obtained from the corresponding computing engine; performing preliminary adaptation on the baseline values ​​of each static parameter based on several preset dimensional features affecting the adjustment of static parameters to obtain several preliminary adapted static parameters; determining each weighting factor and generating a weight adjustment coefficient through weighted calculation; adjusting each preliminary adapted static parameter according to the weight adjustment coefficient; and outputting the final static parameter configuration value. This invention realizes intelligent adjustment of static parameters from a general template to environmental adaptation, significantly improving resource utilization and task execution stability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, and storage medium for adjusting static parameters of a data processing task. Background Technology

[0002] In the field of big data processing, data processing tasks are usually processed through distributed cluster systems and computations are performed through computing frameworks. The proper configuration of static parameters, such as memory allocation, number of threads, and cache size, directly affects the execution efficiency of tasks and the utilization rate of cluster resources.

[0003] Currently, static parameter adjustments mainly rely on two methods: First, manual configuration based on human experience. Developers refer to the general recommended values ​​provided in the official documentation of the computing engine and combine them with their own experience to set parameters. However, the data scale and priority requirements vary greatly in different business scenarios, and the cluster load is constantly changing. Manual configuration is difficult to respond to environmental changes in real time, often resulting in overly conservative parameter settings that waste resources, or overly aggressive settings that cause task failures. Second, template-based configuration based on fixed rules. Some platforms have several built-in preset parameter templates, and users select the matching template according to the task type. However, these templates are all statically set and cannot be dynamically adjusted according to real-time running data. Therefore, existing parameter adjustment methods still suffer from low automation and inability to reasonably adapt to real-time operating environments. Resource utilization and stable task operation still need to be improved. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method, device, and storage medium for adjusting static parameters in data processing tasks. By introducing reinforcement learning and adaptive parameter weight adjustment, it achieves intelligent adjustment of static parameters from a general template to environmental adaptation, significantly improving resource utilization and task execution stability.

[0005] According to a first aspect of the present invention, a method for adjusting static parameters of a data processing task is provided, comprising the following steps: S1, obtain the target computation information corresponding to the data processing task; the target computation information includes cluster resource utilization and basic information of the data processing task; the basic information of the data processing task includes task data volume, estimated task processing time and task priority.

[0006] S2, based on the static parameter template obtained from the corresponding computing engine, obtain the baseline values ​​of each static parameter required for the data processing task; the static parameters include at least the memory allocation baseline value, the thread number configuration baseline value, and the cache configuration baseline value.

[0007] S3, based on several preset dimensional features that affect the adjustment of static parameters, performs preliminary adaptation on the baseline values ​​of each static parameter to obtain several preliminary adapted static parameters.

[0008] S4. Determine each weighting factor from the cluster resource utilization rate, basic information of data processing tasks, and preset task objectives, and generate weight adjustment coefficients through weighted calculation; wherein, the weights of each weighting factor used in the weighted calculation are obtained by adaptive iterative optimization based on real-time running feedback data through reinforcement learning algorithm.

[0009] S5, adjust each preliminary adaptation static parameter according to the weight adjustment coefficient, and output the final static parameter configuration value.

[0010] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, wherein the storage medium stores at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by a processor to perform the above-described data processing task, a static parameter adjustment method.

[0011] According to a third aspect of the present invention, an electronic device is provided, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0012] The present invention has at least the following beneficial effects: This invention provides a method for adjusting static parameters in data processing tasks. First, it obtains the cluster resource utilization and basic information of the data processing task. Based on the static parameter template obtained from the corresponding computing engine, it acquires the baseline values ​​of each static parameter required by the data processing task. Then, based on several preset dimensional features affecting static parameter adjustment, it performs preliminary adaptation on the baseline values ​​of each static parameter, obtaining several preliminary adapted static parameters. It determines the weighting factors affecting parameter adjustment and generates weight adjustment coefficients through weighted calculation. Based on these weight adjustment coefficients, it adjusts each preliminary adapted static parameter, outputting the final static parameter configuration value. As can be seen from the above, this invention solves the technical problem of static parameter configuration in data processing tasks relying on manual experience and being difficult to dynamically adapt to changes in the cluster environment. By introducing reinforcement learning and adaptive parameter weight adjustment, it achieves intelligent adjustment of static parameters from a general template to environmental adaptation, significantly improving resource utilization and task execution stability. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart of a static parameter adjustment method for a data processing task provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] This invention provides a method for adjusting static parameters of a data processing task, such as... Figure 1 As shown, the method includes the following steps: S1, obtain the target computation information corresponding to the data processing task; the target computation information includes cluster resource utilization and basic information of the data processing task; the basic information of the data processing task includes task data volume, estimated task processing time and task priority.

[0017] Specifically, basic information about the data processing task is obtained by parsing the task configuration file; the task priority ranges from 1 to 5, with 1 being the lowest priority and 5 being the highest priority.

[0018] Specifically, the cluster resource utilization rate conforms to the following steps: U = (S-S0) / S×50% + (C-C0) / C×50%, where U represents the cluster resource utilization rate, S is the total storage capacity of the cluster, S0 is the remaining storage capacity of the cluster, C is the total number of computing nodes in the cluster, and C0 is the number of idle computing nodes in the cluster.

[0019] The above-mentioned method, by collecting cluster resource utilization and analyzing the basic information of the task itself, establishes an objective basis for parameter adjustment, ensuring that subsequent parameter adjustments can be based on the actual load and task characteristics of the current environment, and provides real-time and accurate data support for subsequent parameter adjustments.

[0020] S2 obtains the baseline values ​​of each static parameter required for the data processing task based on the static parameter template obtained from the corresponding computing engine; for example, the computing engine can be the Spark computing engine or the Flink computing engine.

[0021] Specifically, the static parameters include at least memory allocation baseline values, thread count configuration baseline values, and cache configuration baseline values. Each baseline value is preset based on industry-standard or best practice settings.

[0022] S3, based on several preset dimensional features that affect the adjustment of static parameters, performs preliminary adaptation on the baseline values ​​of each static parameter to obtain several preliminary adapted static parameters.

[0023] Specifically, the preset dimension features include the number of nodes used by the cluster when executing data processing tasks, node memory size, task data volume, task priority mode, and task objectives.

[0024] Specifically, the task data volume can be categorized into three sizes: large, medium, and small. For example, when the task data volume D > 100GB, it is considered a large data volume; when 10GB ≤ D ≤ 100GB, it is considered a medium data volume; and when D < 10GB, it is considered a small data volume.

[0025] Specifically, task priority modes include high priority, normal priority, and low priority. For example, when the task priority P ≥ 4, it is high priority mode; when P ≤ 4, it is low priority mode; and when P = 3, it is normal priority mode.

[0026] Furthermore, the task objective is either prioritizing task stability or prioritizing task processing speed.

[0027] When the task objective prioritizes task stability, based on a preset stability coefficient, the memory allocation and cache configuration values ​​are increased, while the thread count configuration value is decreased. The preset stability coefficient σ ranges from 1.05 to 1.15. The increase in value is achieved by multiplying both the memory allocation and cache configuration values ​​by σ, while the decrease in value is achieved by multiplying the thread count configuration value by the reciprocal of σ.

[0028] When the task objective prioritizes task processing speed, based on a preset speed coefficient, the memory allocation value and thread count configuration value are increased, while the cache configuration value is decreased. The preset speed coefficient v ranges from 1.1 to 1.2. The increase in value is achieved by multiplying both the memory allocation value and thread count configuration value by v, while the decrease in value is achieved by multiplying the cache configuration value by the reciprocal of v.

[0029] In a specific embodiment, step S3 includes the following steps: S301, the memory allocation baseline value is adjusted according to the task priority mode, and optimized according to the ratio of node memory size to the corresponding memory allocation baseline value to obtain a preliminary adapted memory allocation value.

[0030] Specifically, the initial adapted memory allocation value M2 = M0 × (1 + a1 × (P - 3) + a2 × (M1 / M0 - 1)), where M0 is the memory allocation baseline value, M1 is the node memory size, and a1 and a2 are preset adjustment coefficients, with a1 > a2. For example, a1 = 0.1, a2 = 0.05. It should also be noted that when M1 / M0 exceeds 1.2 or falls below 0.8, it is truncated to prevent excessive adjustment. This formula integrates the logic of baseline, priority adjustment, and resource-aware adjustment, making the initial adapted memory allocation value more reliable.

[0031] S302, the parallelism of the baseline value for the number of threads is expanded according to the number of nodes used by the cluster when executing data processing tasks, and adjusted according to task priority to obtain a preliminary adapted value for the number of threads; wherein, the preliminary adapted value for the number of threads does not exceed a preset proportion of the number of idle computing nodes in the current cluster. For example, the preset proportion is 0.8.

[0032] Specifically, the initial thread count configuration value N2 = N0 × N1 × (1 + b × (P - 3)), where N0 is the baseline thread count configuration value, N1 is the number of nodes used by the cluster when executing data processing tasks, and b is a preset adjustment coefficient, such as 0.08. This formula directly multiplies the baseline thread count configuration value by the number of nodes, indicating that the thread count configuration value should increase linearly with the number of devices. It also introduces a priority adjustment term, which increases the number of threads to improve parallel processing capabilities for high-priority tasks and reduces the number of threads to give up the CPU for low-priority tasks. Hard constraints are also added to avoid cluster crashes caused by resource contention.

[0033] S303, based on the adjustment coefficient corresponding to the task data volume, the cache configuration baseline value is adapted to the data volume to obtain the initially adapted cache configuration value.

[0034] Specifically, the initial adapted cache configuration value K2 = K0 × (1 + c × θ), where K0 is the cache configuration baseline value, c is a preset adjustment coefficient, such as 0.1, and θ is the coefficient corresponding to the task data volume scale, such as 0.2 for large data volume scale, 0 for medium data volume scale, and -0.1 for small data volume scale. This formula introduces the task data volume scale on the basis of the cache configuration baseline value to achieve reasonable cache adaptation.

[0035] As described above, the baseline values ​​of each static parameter are specifically modified based on the preset dimensional features, so that the static parameters initially fit the hardware conditions and operating mode of the current task. Through formulaic objective calculation, the parameters are reasonably adapted, thus completing the first personalized adjustment of the static parameters from a general template to a specific task scenario.

[0036] S4. Determine each weighting factor from the cluster resource utilization rate, basic information of data processing tasks, and preset task objectives, and generate weight adjustment coefficients through weighted calculation; wherein, the weights of each weighting factor used in the weighted calculation are obtained by adaptive iterative optimization based on real-time running feedback data through reinforcement learning algorithm.

[0037] Specifically, step S4 includes the following steps: S401, Based on historical parameter adjustment cases, initialize the baseline weights of each weighting factor; this can be understood as: the baseline weights of each weighting factor are determined based on the weighting allocation ratio in the historical optimal parameter adjustment cases.

[0038] Specifically, the weighting factors are cluster resource utilization, task data volume, task priority, or task objective.

[0039] S402, Real-time acquisition of operational feedback data for the data processing task; the operational feedback data includes task processing latency, real-time cluster resource utilization, and current parameter adjustment error. The acquisition period is set to 10-30 seconds.

[0040] Specifically, task processing latency is the difference between the actual task completion time and the task start time.

[0041] Specifically, the current parameter adjustment error E = |M2 - M 优 | / M 优 +|N2-N 优 | / N 优 +|K2-K 优 | / K 优 , of which M 优 N 优 and K 优 The memory allocation value, thread count configuration value, and cache configuration value were adjusted in the case of adjusting the historical best parameters respectively.

[0042] S403, the running feedback data is used as the reward signal input to the reinforcement learning algorithm, so that the reinforcement learning algorithm iteratively corrects the baseline weights of each weighting factor according to the reward signal. The reinforcement learning algorithm employs the Q-learning algorithm.

[0043] Specifically, the reward signal meets the following conditions: R=1-E-|U t -α|-|(T-T0) / T0|, where R represents the reward value corresponding to the reward signal, E is the current parameter adjustment error, and U t The real-time cluster resource utilization rate is defined as T, the task processing latency is defined as T0, the estimated task processing time is defined as T0, and α is the preset ideal value for cluster resource utilization.

[0044] Furthermore, the iteratively corrected weight W t+1 =W t +η×(R-R0)×▽L(W t ), where t represents the number of iterations, η is the learning rate (ranging from 0.01 to 0.1), R0 is the historical evaluation reward signal, and ▽L(W t ) represents the gradient of the weight loss function; during the correction process, weighting factors that are positively correlated with the task processing effect, such as the cluster resource utilization weight and the task priority weight, are positively increased, while weighting factors that are negatively correlated with the task processing effect, such as the task data volume weight and the task target weight, are negatively decreased.

[0045] S404, when the parameter adjustment error after iterative correction meets the preset conditions, the current weight is taken as the final weight; otherwise, steps S402 to S403 are repeated until the parameter adjustment error after iterative correction meets the preset conditions or reaches a preset iteration number threshold; the preset conditions are that the parameter adjustment error after iterative correction is less than a preset error threshold and the cluster resource utilization is within a preset range. Those skilled in the art can set the preset error threshold according to actual needs, for example, 0.1. The preset range of cluster resource utilization can be set to 0.4 to 0.8 according to actual needs.

[0046] Furthermore, the weight adjustment coefficient λ = W1 × U + W2 × (D / D) 均 )+W3×(P / 3)+W4×G, where W1, W2, W3, and W4 are the iteratively corrected weights corresponding to cluster resource utilization, task data volume, task priority, and task objective, respectively, and W1+W2+W3+W4=1, U represents the cluster resource utilization in the basic information, and D 均 G represents the average data volume of historical data processing tasks. G is the task objective; when task stability is prioritized, G=0; when processing speed is prioritized, G=1. In exceptional cases, if a weight is less than 0 after an update, it is set to 0, and the remaining weights are redistributed proportionally.

[0047] As described above, by using reinforcement learning algorithms to adaptively and iteratively optimize the weights of the weighting factors, the weight allocation can dynamically match the real-time cluster status and task operation feedback, overcoming the shortcomings of traditional fixed weights that cannot flexibly respond to environmental changes. This provides a reliable basis for subsequent automated adjustment of static parameters and significantly improves the accuracy and automation of parameter adjustment.

[0048] S5, adjust each preliminary adaptation static parameter according to the weight adjustment coefficient, and output the final static parameter configuration value; this can be understood as: multiplying each preliminary adaptation parameter by the weight adjustment coefficient as the corresponding static parameter configuration value.

[0049] The above-mentioned weighted adjustment coefficients are applied to the initial adaptation parameters for final correction, generating static parameter configurations that conform to both task characteristics and current cluster status. Through fully automated parameter adjustment, dynamic optimal matching between parameter configurations and the operating environment is achieved.

[0050] Furthermore, the method also includes the following steps: S10, Collect historical parameter adjustment cases; the historical parameter adjustment cases include the values ​​of each weighted factor and the comprehensive evaluation index corresponding to the historical parameter adjustment cases; the comprehensive evaluation index is obtained based on the parameter adjustment error.

[0051] Specifically, the comprehensive evaluation index y=1-E means that the closer y is to 1, the better the adjustment effect.

[0052] S20, calculate the correlation factor between each weighted factor and the comprehensive evaluation index using the Pearson correlation coefficient; the correlation factor is used to quantify the strength of the correlation between the corresponding weighted factor and the adjustment effect. Those skilled in the art are familiar with the specific calculation method of the correlation factor using the Pearson correlation coefficient, and it will not be elaborated here. The correlation factor ranges from -1 to 1; a correlation factor greater than 0 indicates a positive correlation, and a correlation factor less than 0 indicates a negative correlation. The larger the absolute value of the correlation factor, the stronger the correlation.

[0053] S30, normalize the correlation factors corresponding to each weighted factor to obtain the normalized correlation factors. Preferably, the min-max normalization method can be used.

[0054] S40, the weights of each weighted factor after iterative optimization by the reinforcement learning algorithm are calibrated using the normalized correlation factor to obtain the final weight of each weighted factor; this can be understood as: the product of the normalized correlation factor and the weight of the iteratively optimized weighted factor is used as the calibrated weight.

[0055] The above-mentioned approach, by introducing a correlation factor to perform a secondary calibration of the mechanical energy of the iteratively corrected weights, achieves a dual integration of historical experience and real-time dynamic feedback. This ensures that the weight allocation not only conforms to the real-time operating status but also aligns with the core influence of each weighting factor, avoiding interference from irrelevant factors. This significantly improves the rationality and stability of the weight allocation, ensuring that the weighted calculation can more accurately focus on the core factors that truly affect the adjustment effect.

[0056] Furthermore, the method also includes a deployment environment adaptation step, specifically including: S100, identify the deployment environment type of the current data processing task; the deployment environment type is either a local deployment environment or a private deployment environment. Specifically, the deployment environment type is obtained by parsing the environment identifier field in the deployment configuration file.

[0057] For S200, in a localized deployment environment, the weight of the number of nodes used when executing data processing tasks is increased, and the calculation coefficients of the initial adaptation static parameters are adjusted. This adapts to the characteristics of localized deployment devices, which have fewer nodes and more concentrated resources.

[0058] Specifically, the value of a2 in the formula corresponding to the initial adapted memory allocation value M2 is adjusted from 0.05 to 0.1, and the value of b in the initial adapted thread number configuration value is adjusted from 0.08 to 0.1.

[0059] For S300, in a private deployment environment, a security factor is introduced to correct the final static parameters and adjust the baseline weight of task priorities. The security factor ranges from 1.05 to 1.1, and the product of the security factor and the final static parameters is used as the correction value. In addition, the baseline weight of task priorities is increased and adjusted.

[0060] S400 matches and verifies the static parameters after environment adaptation correction with the hard constraints of the corresponding deployment environment type, and outputs the final static parameter configuration that meets the environment constraints.

[0061] Specifically, the hard limit for localized deployment environments is that the static parameters after environment adaptation correction are no greater than 80% of the total available memory of all nodes, while the hard limit for private deployment environments is that the cache configuration value after environment adaptation correction is no less than 30% of the memory allocation value after environment adaptation correction.

[0062] As described above, by identifying the type of localized or private deployment environment, and adjusting the dimensional feature weights, adaptation coefficients, and security constraints accordingly, a deep coupling between static parameter configuration and the underlying hardware environment is achieved. This solves the problem of general parameter templates not being suitable for different deployment scenarios, ensuring that parameter adjustment strategies can be adapted to local conditions, and significantly improving the operational stability and resource utilization efficiency of tasks in specific environments.

[0063] Embodiments of the present invention also provide a non-transitory computer-readable storage medium, which can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a method in the method embodiment, wherein the at least one instruction or the at least one program is loaded and executed by the processor to implement the static parameter adjustment method for the data processing task provided in the above embodiment.

[0064] Embodiments of the present invention also provide an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0065] While specific embodiments of the invention have been described in detail by way of example, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. It should also be understood that various modifications can be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A method for static parameter adjustment of a data processing task, characterized in that The method includes the following steps: S1, Obtain the target computation information corresponding to the data processing task; the target computation information includes cluster resource utilization and basic information of the data processing task; the basic information of the data processing task includes task data volume, estimated task processing time and task priority. S2, based on the static parameter template obtained from the corresponding computing engine, obtain the baseline values ​​of each static parameter required for the data processing task; the static parameters include at least the memory allocation baseline value, the thread number configuration baseline value, and the cache configuration baseline value; S3, based on several preset dimensional features that affect the adjustment of static parameters, the baseline values ​​of each static parameter are initially adapted to obtain several initially adapted static parameters. S4, determine each weighting factor from cluster resource utilization, basic information of data processing tasks and preset task objectives, and generate weight adjustment coefficients through weighted calculation; wherein, the weights of each weighting factor used in the weighted calculation are obtained by adaptive iterative optimization based on real-time running feedback data through reinforcement learning algorithm; S5, adjust each preliminary adaptation static parameter according to the weight adjustment coefficient, and output the final static parameter configuration value.

2. The method of claim 1, wherein, The cluster resource utilization rate conforms to the following steps: U = (S-S0) / S×50% + (C-C0) / C×50%, where U represents the cluster resource utilization rate, S is the total storage capacity of the cluster, S0 is the remaining storage capacity of the cluster, C is the total number of computing nodes in the cluster, and C0 is the number of idle computing nodes in the cluster.

3. The method of claim 1, wherein, The preset dimension features include the number of nodes used by the cluster when executing data processing tasks, node memory size, task data volume, task priority mode, and task objectives.

4. The method of claim 1, wherein, The task objective is either prioritizing task stability or prioritizing task processing speed. When the task objective is to prioritize task stability, based on the preset stability coefficient, increase the memory allocation value and cache configuration value, and decrease the thread count configuration value; When the task objective is to prioritize task processing speed, based on a preset speed coefficient, increase the memory allocation value and thread count configuration value, and decrease the cache configuration value.

5. The method of claim 3, wherein, Step S3 includes the following steps: S301, the memory allocation baseline value is adjusted according to the task priority mode, and optimized according to the ratio of node memory size to the corresponding memory allocation baseline value to obtain a preliminary adapted memory allocation value. S302, the parallelism of the thread number configuration baseline value is expanded according to the number of nodes used by the cluster when executing data processing tasks, and adjusted according to task priority to obtain a preliminary adapted thread number configuration value; wherein, the preliminary adapted thread number configuration value does not exceed the preset proportion of the number of idle computing nodes in the current cluster. S303, based on the adjustment coefficient corresponding to the task data volume, the cache configuration baseline value is adapted to the data volume to obtain the initially adapted cache configuration value.

6. The method of claim 1, wherein, Step S4 includes the following steps: S401, Based on historical parameter adjustment cases, initialize the baseline weights of each weighting factor; the weighting factors are cluster resource utilization, task data volume, task priority, or task objective; S402, Real-time collection of the operation feedback data of the data processing task; the operation feedback data includes task processing latency, real-time cluster resource utilization, and current parameter adjustment error; S403, the running feedback data is used as the reward signal input to the reinforcement learning algorithm, so that the reinforcement learning algorithm iteratively corrects the baseline weights of each weighting factor according to the reward signal; S404, when the parameter adjustment error after iterative correction meets the preset conditions, the current weight is taken as the final weight; Conversely, repeat steps S402 to S403 until the parameter adjustment error after iterative correction meets the preset conditions or reaches the preset iteration number threshold. The preset conditions are that the parameter adjustment error after iterative correction is less than the preset error threshold and the cluster resource utilization rate is within the preset range.

7. The static parameter adjustment method for a data processing task according to claim 6, characterized in that, The reward signal meets the following conditions: R = 1 - |E|U t -α|-|(T-T0) / T0|, wherein R represents a reward value corresponding to a reward signal, E is a current parameter adjustment error, U t is a real-time cluster resource utilization rate, T is a task processing delay, T0 is a task estimated processing duration, and a is a preset ideal value of the cluster resource utilization rate.

8. The static parameter adjustment method for a data processing task according to claim 1, characterized in that, The method further includes the following steps: S10, Collect historical parameter adjustment cases; the historical parameter adjustment cases include the values ​​of each weighted factor and the comprehensive evaluation index corresponding to the historical parameter adjustment cases; the comprehensive evaluation index is obtained based on the parameter adjustment error; S20, calculate the correlation factor between each weighted factor and the comprehensive evaluation index using the Pearson correlation coefficient; the correlation factor is used to quantify the strength of the correlation between the corresponding weighted factor and the adjustment effect; S30, normalize the correlation factors corresponding to each weighted factor to obtain the normalized correlation factors; S40, the weights of each weighted factor after iterative optimization by the reinforcement learning algorithm are weighted and calibrated using the normalized correlation factor to obtain the final weights of each weighted factor.

9. A non-transitory computer-readable storage medium, wherein the storage medium stores at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the static parameter adjustment method for the data processing task as described in any one of claims 1-8.

10. An electronic device, characterized in that, Includes a processor and the non-transitory computer-readable storage medium as described in claim 9.

Images