Server performance prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses a server performance prediction method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining the performance operation state data of a server, carrying out the normalization processing of the performance operation state data, obtaining the normalized data, and carrying out the prediction of the performance of the server. Then the normalized data are input into a preset performance prediction model, a performance prediction result is obtained, model parameters of the preset performance prediction model are adjusted by using a preset parameter adjustment algorithm according to the direction in which the fitness function value of the initial model parameters of the preset performance prediction model is increased, and the performance prediction result is obtained. The obtained parameter and fitness function value can reflect a globally optimal solution until the parameter adjustment frequency of the initial model parameter is greater than a preset frequency threshold value, and falling into a locally optimal solution is avoided. The global optimal solution is found in the parameter space, so that the accuracy of server performance prediction is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a performance prediction method and device for a server, an electronic device, and a storage medium. Background Art

[0002] With the widespread application of network services, network services put more and more pressure on the operation of servers. In order to ensure that the server can operate normally in the face of high-concurrency network service requests, it is necessary to predict the performance of the server.

[0003] In the related technologies of server performance prediction, the support vector machine (SVM) model is usually used to predict the performance of the server. Since the SVM model will fall into a local optimal solution during the process of optimizing the model parameters, the effect of server performance prediction is poor. Summary of the invention

[0004] The present application provides a method and device for predicting server performance, an electronic device and a storage medium, so as to at least solve the problem that the performance prediction effect of the server in the related art is poor.

[0005] The present application provides a method for predicting performance of a server, including: Get the performance status data of the server; Normalizing the performance operation status data to obtain normalized data; The normalized data is input into the preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjusted model parameters of the preset performance prediction model are obtained by using a preset parameter adjustment algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters of the preset performance prediction model until the number of parameter adjustments of the initial model parameters is greater than a preset number threshold.

[0006] The present application also provides a performance prediction device for a server, comprising: An acquisition unit, used to acquire performance operation status data of the server; A processing unit, used for normalizing the performance operation status data to obtain normalized data; The prediction unit is used to input the normalized data into a preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjustment model parameters of the preset performance prediction model are to use a preset parameter adjustment algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters of the preset performance prediction model until the number of parameter adjustments of the initial model parameters is greater than a preset number threshold, thereby obtaining the parameters.

[0007] The present application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of any of the above-mentioned server performance prediction methods when executing the computer program.

[0008] The present application also provides a computer-readable storage medium, in which a computer program is stored, wherein when the computer program is executed by a processor, the steps of any of the above-mentioned server performance prediction methods are implemented.

[0009] The present application also provides a computer program product, including a computer program, which implements the steps of any of the above-mentioned server performance prediction methods when the computer program is executed by a processor.

[0010] Through this application, by obtaining the performance operation status data of the server and normalizing the performance operation status data, normalized data is obtained to unify the data scale between different performance operation status data and reduce the data differences between different performance operation status data, and then the normalized data is input into the preset performance prediction model to obtain the performance prediction result, wherein the adjustment model parameters of the preset performance prediction model are adjusted by using the preset parameter adjustment algorithm, in the direction of increasing the fitness function value of the initial model parameters of the preset performance prediction model, the initial model parameters are adjusted until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, the obtained parameters, the fitness function value can reflect the global optimal solution, avoid falling into the local optimal solution, and realize the search for the global optimal solution in the parameter space, thereby improving the accuracy of server performance prediction. Therefore, the technical problem of poor performance prediction effect of the server can be solved, and the technical effect of improving the accuracy of server performance prediction can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A flow chart of a method for predicting server performance provided by an embodiment of the present application; Figure 2 A schematic diagram of the classification of wolf pack levels provided in an embodiment of the present application; Figure 3 A schematic diagram of the operation flow of a differential evolution algorithm provided in an embodiment of the present application; Figure 4 A schematic flow chart of the entire process of performance prediction of a server provided in an embodiment of the present application; Figure 5 A schematic diagram of the structure of a performance prediction device for a server provided in an embodiment of the present application; Figure 6 A schematic diagram of the structure of another server performance prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0014] It should be noted that, in the description of this application, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence.

[0015] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0016] In conjunction with the specific application environment architecture or specific hardware architecture on which the execution of the performance prediction method of the server depends, the specific application environment architecture or the specific hardware architecture is described herein.

[0017] An embodiment of the present application provides a performance prediction method for a server, and the method is described in detail in conjunction with the execution flow of the performance prediction method for the server.

[0018] Figure 1 A flow chart of a method for predicting server performance provided in an embodiment of the present application.

[0019] like Figure 1 As shown, the method comprises the following steps: Step 101, obtaining performance operation status data of the server.

[0020] Performance operation status data refers to a collection of relevant data indicators that reflect the performance of the server during operation. Performance operation status data can reflect the server's computing resource usage, workload, response speed and other performance characteristics. Performance status data includes but is not limited to the server's used memory, the server's central processing unit, the server's free memory, the server application's used memory, the server application's free memory, the server application's received or sent byte stream, the server application's input / output read and write byte stream, and the server application's hypertext transfer protocol response time.

[0021] You can collect the server's performance status data by using the performance monitoring tool that comes with the server operating system. By obtaining the server's performance status data, you can achieve comprehensive monitoring of the server's comprehensive performance and help users discover potential problems with the server in a timely manner.

[0022] Step 102, normalize the performance operation status data to obtain normalized data.

[0023] Normalization is a data preprocessing technique that scales the data so that it falls into a specific interval, such as [0,1] or [-1,1]. Normalization can eliminate the impact of dimensions and dimension ranges between different eigenvalues, make data comparable, and improve the convergence speed and prediction accuracy of the preset performance prediction model. Normalized data refers to performance operating status data after normalization, and its value is mapped to a specified interval. Normalized data enables data of different dimensions and dimension ranges to be compared and analyzed on the same scale.

[0024] Since there may be burstiness and instability between the different performance operation status data collected from the servers, a large difference in the order of magnitude between different performance operation status data will have an adverse effect on the performance prediction of the preset performance prediction model. In addition, the preset performance prediction model is more sensitive to data between [0,1]. In order to improve efficiency, the performance operation status data is normalized so that its range is between [0,1].

[0025] By normalizing the performance status data to the same scale, the preset performance prediction model can more accurately learn the features and patterns in the data and avoid deviations caused by different data dimensions, thereby improving the accuracy of server performance prediction.

[0026] Step 103, input the normalized data into the preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjustment model parameters of the preset performance prediction model are to use the preset parameter adjustment algorithm to adjust the initial model parameters in the direction of increasing the fitness function value of the initial model parameters of the preset performance prediction model, until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, and the parameters are obtained.

[0027] The preset performance prediction model refers to a model that is pre-set and trained in the server performance prediction task. The model can predict the performance of the server in the future based on the input server performance operation status data. It is usually trained based on a large amount of historical performance data and has the ability to estimate and predict server performance trends. The preset performance prediction model can be a support vector machine model, but it should be clear that this statement is not intended to limit the preset performance prediction model to only a support vector machine model, but can also be other models.

[0028] The preset parameter adjustment algorithm is an algorithm used to adjust the initial model parameters of the preset performance prediction model. Its goal is to increase the fitness function value of the model parameters. The fitness function value is a standard for measuring the quality of model parameters. It is usually designed based on indicators such as prediction error and model complexity. The parameter adjustment algorithm gradually improves the model parameters through an iterative optimization process to improve the prediction performance of the model.

[0029] The initial model parameters are the original parameter settings of the preset performance prediction model before parameter adjustment. These parameters are the starting point for model training and subsequent optimization. The setting of their initial values ​​may affect the convergence speed of model training and the final prediction effect.

[0030] The fitness function value is calculated by the fitness function and is used to quantitatively evaluate the performance of model parameters in server performance prediction. The higher the fitness function value, the better the model parameters are, which can make the model's prediction of server performance more accurate and reliable.

[0031] Adjusting model parameters refers to optimizing and adjusting the initial model parameters within the preset performance prediction model to improve the performance prediction accuracy of the preset performance prediction model. The preset times threshold is a preset upper limit of the number of parameter adjustments. When the number of model parameter adjustments reaches or exceeds the preset times threshold, the parameter adjustment process stops.

[0032] By optimizing the initial model parameters through a preset parameter adjustment algorithm, the fitness function value of the model can be increased, that is, the prediction error can be reduced, thereby improving the accuracy of server performance prediction.

[0033] Through this application, by obtaining the performance operation status data of the server and normalizing the performance operation status data, normalized data is obtained to unify the data scale between different performance operation status data and reduce the data differences between different performance operation status data, and then the normalized data is input into the preset performance prediction model to obtain the performance prediction result, wherein the adjustment model parameters of the preset performance prediction model are adjusted by using the preset parameter adjustment algorithm, in the direction of increasing the fitness function value of the initial model parameters of the preset performance prediction model, the initial model parameters are adjusted until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, the obtained parameters, the fitness function value can reflect the global optimal solution, avoid falling into the local optimal solution, and realize the search for the global optimal solution in the parameter space, thereby improving the accuracy of server performance prediction. Therefore, the technical problem of poor performance prediction effect of the server can be solved, and the technical effect of improving the accuracy of server performance prediction can be achieved.

[0034] In practical applications, in order to improve the accuracy of performance prediction of the preset performance prediction model on the performance operation status data, it is necessary to train the preset performance prediction model, which can be achieved by but not limited to the following methods, including: obtaining the performance operation status data for training; using the performance operation status data for training to train the initial performance prediction model to obtain the preset performance prediction model and initial model parameters; using the preset parameter adjustment algorithm to adjust the initial model parameters in the direction of increasing the fitness function value of the initial model parameters, until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, thereby obtaining the adjusted model parameters.

[0035] Training performance operation status data refers to server performance operation status data specifically used to train performance prediction models. This data is usually collected from the actual operation of the server and covers various performance indicators of the server under different loads and at different time points.

[0036] The training performance operation status data is divided into a training set and a validation set. The training set is used to train the initial performance prediction model, and the gradient descent algorithm is used to update the parameters of the initial performance prediction model. During the training process, the error between the predicted output of the initial performance prediction model and the true value is calculated, and the parameters of the model are adjusted according to the error back propagation to gradually reduce the error. At the same time, the validation set is used to monitor the performance of the initial performance prediction model to prevent overfitting.

[0037] The preset parameter adjustment algorithm (such as differential evolution algorithm, gray wolf parameter search algorithm, etc.) is used to optimize and adjust the initial model parameters. Taking the differential evolution algorithm as an example, the specific steps are as follows: Population initialization: Randomly generate a certain number of model parameter sets as the initial population, each parameter set contains all the parameters to be optimized of the model (such as C and γ of the support vector machine). Fitness calculation: For each parameter set in the population, the model is trained using the normalized training data, and then the prediction performance of the model is calculated using the test data to obtain the corresponding fitness function value. Mutation operation: According to the mutation strategy of differential evolution, the parameter set in the population is mutated to generate a new candidate parameter set. Crossover operation: The mutated candidate parameter set is cross-mixed with the original parameter set to generate a new test parameter set. Selection operation: The fitness function value of the test parameter set is calculated and compared with the corresponding original parameter set, and the parameter set with higher fitness is retained to enter the next generation population. Iteration termination condition judgment: Repeat the above mutation, crossover, and selection operations until the number of parameter adjustments reaches the preset threshold, and the optimized model parameters are obtained, that is, the model parameters are adjusted.

[0038] For the gray wolf parameter search algorithm, it simulates the group hunting behavior of gray wolves and searches for the optimal model parameters by updating the position of gray wolves. The specific steps include: Initialize the gray wolf population: randomly initialize a certain number of gray wolf positions, and each gray wolf position represents a set of model parameters. Fitness evaluation: calculate the fitness function value of the model parameters corresponding to each gray wolf position, and determine the current optimal solution (that is, the parameters with the highest fitness). Gray wolf position update: according to the gray wolf's level and hunting strategy, update the gray wolf's position, simulate the movement of the gray wolf in the search space, and gradually approach the optimal solution. Iteration termination condition judgment: when the number of parameter adjustments reaches the preset number threshold, stop the iteration and obtain the final optimized model parameters, that is, adjust the model parameters.

[0039] During the optimization process, the preset parameter adjustment algorithm comprehensively considers factors such as the model's predictive performance and complexity, and avoids overfitting of the model to the training data by reasonably adjusting the parameters.

[0040] As a refinement of the above embodiment, the preset parameter adjustment algorithm includes a preset differential evolution algorithm and a preset gray wolf parameter search algorithm. When executing the preset parameter adjustment algorithm, the initial model parameters are adjusted in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, and the adjusted model parameters are obtained, it can be implemented in but not limited to the following manner, including: using the preset differential evolution algorithm, the initial model parameters are adjusted in a direction that increases the fitness function value of the initial model parameters, until the number of parameter adjustments of the initial model parameters is greater than the first preset number threshold, and the first adjusted model parameters are obtained; using the preset gray wolf parameter search algorithm, the first adjusted model parameters are adjusted in a direction that increases the fitness function value of the first adjusted model parameters, until the number of parameter adjustments of the first adjusted model parameters is greater than the second preset number threshold, and the second adjusted model parameters are obtained; the preset number threshold includes the first preset number threshold and the second preset number threshold; and the second adjusted model parameter is determined as the adjusted model parameter.

[0041] The preset gray wolf parameter search algorithm can be a gray wolf optimization algorithm, which simulates the hierarchy and respective responsibilities of wolves in the hunting process. The gray wolf optimization algorithm has the advantages of simple principle, few model parameters to be adjusted, easy implementation, and strong global search capability. When the gray wolf optimization algorithm is used to optimize a continuous function, the number of wolves is set to N, and the search space dimension is set to ,exist In the dimensional space The position of the alpha wolf is defined as , in order to better understand the wolf pack level in the gray wolf optimization algorithm, such as Figure 2 As shown, Figure 2 A schematic diagram of the classification of wolf pack levels provided in the embodiment of the present application is shown in FIG. Wolf, Wolf, The wolves are defined as the first, second, and third levels, and the remaining wolves are set as Wolf, Wolves are the lowest level. Then the group hunting process begins, and the lower levels must obey the instructions and leadership of the upper levels.

[0042] The gray wolf optimization algorithm optimizes the optimal solution vector position update process. The optimization process of the gray wolf optimization algorithm is to randomly generate a group of gray wolves in the search space. During the evolution process, the position of the prey is Wolf, Wolf, Wolf Assessment and Location, Wolves calculate their distance to their prey, then approach, surround, and attack the prey until it is captured.

[0043] The preset differential evolution algorithm can be a differential evolution algorithm. The differential evolution algorithm is a meta-heuristic search algorithm that simulates the biological evolution mechanism. Through repeated iterative operations, it retains those excellent individuals that adapt to the environment. Compared with other evolutionary algorithms, the differential evolution algorithm retains the best population individuals in the global search. It adopts a one-to-one competition survival mode, so it can reduce the complexity of genetic operations. At the same time, because the differential evolution algorithm has a special memory ability, it can be used to dynamically track the current search state to adjust the search plan, so it has better global convergence ability and good robustness, and does not need to rely on the characteristic information of the problem, so it can be used to solve many optimization problems in complex environments that cannot be solved by mathematical methods.

[0044] The main control parameters of the differential evolution algorithm include three: population size (NP), crossover probability (CR) and scaling factor (F). NP refers to the amount of information in the initial population of the algorithm. The larger the value of NP, the richer the information of the population. The trouble that comes with it is that the amount of calculation will increase and the solution will be difficult. On the contrary, if the value of NP is small, the diversity of the population will be limited, which is not conducive to the algorithm to obtain the global optimal solution. The worst result is that the search will stagnate directly. CR is the crossover probability, which mainly refers to the amount of information exchanged between the offspring and the parent generation and the intermediate variants during the crossover process. As the value of CR continues to increase, the degree of information exchange will also increase. On the contrary, if the value of CR is too small, the diversity of the population will decrease rapidly, resulting in difficulties in global optimization. Compared with CR, the scaling factor F is more likely to affect the performance of the algorithm. The smaller the value of F, the better the algorithm's ability to search locally; on the contrary, the larger the value of F, the easier it is for the algorithm to jump out of the local minimum, but the convergence speed will slow down. At the same time, the value of F will also affect the diversity of the population. The differential evolution algorithm process is generally used to solve the global optimization problem of continuous variables, and its main working steps are based on three operations, namely mutation, crossover and selection.

[0045] In order to better understand the operation process of the differential evolution algorithm, Figure 3 As shown, Figure 3A schematic diagram of the operation flow of a differential evolution algorithm provided in an embodiment of the present application, the basic principle of the algorithm is to randomly generate an initial population, then randomly select two individuals from the population, and then use the difference vector between the two individuals to obtain the random change source of the third individual, and then sum the difference vector with the third individual according to certain rules after weighting to obtain a new individual, that is, a mutant individual. Then, the mutant individual and a target individual set previously are parameter-mixed to obtain a new individual. This operation process is called crossover. In the specific experimental process, if the fitness of a certain individual is greater than that of the target individual, then the experimental individual is replaced by the target individual, otherwise, the target individual is retained. This process is called selection. In each iteration of the experiment, each body vector needs to be used as a target individual once, and then the algorithm needs to continuously iterate and calculate. During the calculation process, high-quality individuals are retained and inferior individuals are removed. Through continuous searching, the optimal solution is obtained.

[0046] The preset parameter adjustment algorithm takes into account factors such as the model's prediction performance and complexity during the optimization process, and avoids overfitting of the model to the training data by adjusting the parameters reasonably. This enables the model to better balance the degree of fit and generalization ability when predicting server performance, reduces the problem of poor prediction results in actual applications due to overfitting, and improves the stability and robustness of the model.

[0047] As a refinement of the above embodiment, when executing a preset differential evolution algorithm, adjusting the initial model parameters in a direction that increases the fitness function value of the initial model parameters, until the number of parameter adjustments of the initial model parameters is greater than a first preset number threshold, and the first adjusted model parameters are obtained, it can be implemented in but not limited to the following manner, including: randomly dividing all model parameters in the initial model parameters into at least two model parameter sets; performing fitness function calculations on the at least two model parameter sets respectively, and obtaining fitness function values ​​corresponding to the at least two model parameter sets; and adjusting the model parameters of the at least two model parameter sets in a direction that increases the fitness function value of the model parameter sets according to the fitness function values ​​corresponding to the at least two model parameter sets, until the number of parameter adjustments of the model parameter sets is greater than the first preset number threshold, and the first adjusted model parameters are obtained.

[0048] The first preset number threshold is the upper limit of the number of parameter adjustments in the preset differential evolution algorithm. When the number of model parameter adjustments reaches or exceeds this threshold, the differential evolution algorithm stops and the first adjusted model parameters are obtained. The setting of this threshold is to control the computing resource consumption and time cost of parameter optimization, while avoiding problems such as overfitting caused by excessive optimization. The first adjusted model parameters are the model parameters obtained after optimization and adjustment by the preset differential evolution algorithm.

[0049] Before starting the differential evolution algorithm optimization, you first need to prepare the initial model parameters. The initial model parameters can be randomly generated or preset values ​​obtained from previous training. The initial model parameters are randomly divided into at least two model parameter sets so that they can be optimized independently. For each divided model parameter set, calculate its fitness function value. The calculation of the fitness function usually involves measuring the performance of each parameter set through the training or prediction process of the model. For example, the mean square error is used to measure the difference between the model prediction value and the actual value. The differential evolution algorithm optimizes each model parameter set through mutation, crossover and selection operations. The specific steps are as follows: Mutation operation: Generate a new solution based on the differences between individuals in the current population. Generate a new candidate parameter set through differential mutation. Crossover operation: Exchange information between different parameter sets according to certain crossover strategies. Selection operation: Determine which parameter sets can survive and continue to be optimized based on the fitness function value. The parameter set with a larger fitness function value is retained and continues to be updated. By setting a first preset number threshold, the differential evolution algorithm will stop optimizing when the number of adjustments exceeds the threshold. At this time, the obtained parameter set is the first adjustment model parameter, that is, the optimized parameter set. After the differential evolution algorithm is completed, the optimized "first adjustment model parameter" set is output. Each parameter set in this set has been optimized, and the fitness function value is relatively high, which can be used for further training or prediction of the model.

[0050] The initial model parameters are divided into multiple parameter sets for independent optimization, so that each parameter set can independently explore the solution space and better adapt to different optimization needs. Especially when dealing with complex models or large-scale data, the optimization effect is more significant.

[0051] As a refinement of the above embodiment, when performing parameter adjustment on the model parameters of at least two model parameter sets according to the fitness function values ​​corresponding to each of the at least two model parameter sets in a direction that increases the fitness function values ​​of the model parameter sets, until the number of parameter adjustments of the model parameter sets is greater than a first preset number threshold, and the first adjusted model parameters are obtained, it can be implemented in but not limited to the following manner, including: screening out a target model parameter set whose fitness function value is greater than the first preset fitness threshold from the at least two model parameter sets; adjusting the remaining model parameter sets in a direction that increases the fitness function values ​​of the model parameters in the remaining model parameter sets; The model parameters in the target model parameter set are adjusted to obtain the adjusted remaining model parameter set; the remaining model parameter set is any model parameter set other than the target model parameter set in at least two model parameter sets; a fitness function is calculated on the adjusted remaining model parameter set to obtain the fitness function value of the adjusted remaining model parameter set; a model parameter set having a fitness function value greater than a first preset fitness threshold is screened out from the remaining model parameter set until the number of parameter adjustments of the model parameters in the remaining model parameter set is greater than the first preset number threshold; the model parameters in the target model parameter set and the screened remaining model parameter set are determined as the first adjusted model parameters.

[0052] The first preset number threshold is the upper limit of the number of parameter adjustments in the preset differential evolution algorithm. When the number of model parameter adjustments reaches or exceeds this threshold, the differential evolution algorithm stops and the first adjusted model parameters are obtained. The threshold is set to control the computing resource consumption and time cost of parameter optimization, while avoiding problems such as overfitting caused by excessive optimization. The first adjusted model parameters are model parameters obtained after optimization and adjustment by the preset differential evolution algorithm. These parameters can initially improve the prediction performance of the server performance prediction model and are the basis for further optimization and adjustment.

[0053] By screening out target model parameter sets with higher fitness function values ​​and making targeted parameter adjustments and optimizations to the remaining model parameter sets, we can more effectively explore the parameter space, avoid the influence of local optimal solutions, and enable model parameters to be better configured globally, thereby significantly improving the accuracy of server performance predictions.

[0054] As a refinement of the above embodiment, when executing the preset gray wolf parameter search algorithm, adjusting the first adjustment model parameters in a direction that increases the fitness function value of the first adjustment model parameters until the number of parameter adjustments of the first adjustment model parameters is greater than a second preset number threshold, and obtaining the second adjustment model parameters, it can be implemented in but not limited to the following manner, including: performing fitness function calculation on all model parameters in the first adjustment model parameters to obtain the fitness function values ​​corresponding to each of all model parameters in the first adjustment model parameters; adjusting all model parameters in the first adjustment model parameters in a direction that increases the fitness function values ​​of all model parameters in the first adjustment model parameters according to the fitness function values ​​corresponding to each of all model parameters in the first adjustment model parameters, until the number of parameter adjustments of all model parameters in the first adjustment model parameters is greater than the second preset number threshold, and obtaining the second adjustment model parameters.

[0055] The preset gray wolf parameter search algorithm is an optimization algorithm that simulates the hunting behavior of gray wolf groups. It regards the optimal solution in the search space as prey, and gradually approaches the optimal solution by simulating the hierarchy and hunting strategy of gray wolves, thereby achieving fine adjustment of parameters. The second preset number threshold is the stopping condition for parameter adjustment when the first adjustment model parameters are adjusted using the preset gray wolf parameter search algorithm. When the number of parameter adjustments of all model parameters in the first adjustment model parameters is greater than the threshold, the parameter adjustment process is stopped. The second adjustment model parameters are model parameters after the first adjustment model parameters are further adjusted by the preset gray wolf parameter search algorithm, and these parameters are more finely optimized on the basis of local optimization.

[0056] Initialize the gray wolf group: Generate multiple gray wolf individuals for each model parameter, each of which represents a set of model parameters. Simulate the hunting behavior of gray wolves: In each round of iteration, the gray wolf adjusts the model parameters by tracking the optimal solution (i.e., the direction that minimizes the fitness function value). Update model parameters: Update model parameters according to the behavior pattern of the gray wolf group until the second preset number threshold is met. Determine whether the second preset number threshold is reached: Each time a parameter adjustment is made, the number of adjustments is recorded. When the number of parameter adjustments of the first adjustment model parameters exceeds the second preset number threshold, it means that the optimization has reached the predetermined number of times and the adjustment can be stopped. Obtain the second adjustment model parameters: After multiple iterative adjustments, the model parameters obtained are the second adjustment model parameters. These parameters are finely optimized by the gray wolf parameter search algorithm, have a better fitness function value, and can effectively improve the model performance.

[0057] By using the Gray Wolf Parameter Search Algorithm to fine-tune the model parameters, the performance of the model can be effectively improved. The Gray Wolf Parameter Search Algorithm can avoid falling into the local optimal solution and provide global optimization capabilities, thereby finding a better set of model parameters.

[0058] As a refinement of the above embodiment, when executing the fitness function values ​​corresponding to all model parameters in the first adjustment model parameters, parameter adjustment is performed on all model parameters in the first adjustment model parameters in a direction of increasing the fitness function values ​​of all model parameters in the first adjustment model parameters until the number of parameter adjustments of all model parameters in the first adjustment model parameters is greater than a second preset number threshold, and the second adjustment model parameters are obtained, it can be implemented in but not limited to the following manner, including: screening out model parameters from a preset number of second adjustment model parameters in order of fitness function values ​​from large to small; and determining the screened out model parameters from the preset number of second adjustment model parameters as the second adjustment model parameters.

[0059] By presetting the Gray Wolf Parameter Searching Algorithm to further optimize and adjust the first adjustment model parameters, and selecting the model parameters with the highest fitness function value as the second adjustment model parameters, the global search capability of the Gray Wolf Parameter Searching Algorithm can be fully utilized, effectively avoiding local optimal solutions, and enabling model parameters to be better configured globally, thereby significantly improving the accuracy of server performance prediction.

[0060] As a refinement of the above embodiment, when performing parameter adjustment on all model parameters in the first adjustment model parameters in a direction that increases the fitness function values ​​of all model parameters in the first adjustment model parameters according to the fitness function values ​​corresponding to each of the model parameters in the first adjustment model parameters, until the number of parameter adjustment of all model parameters in the first adjustment model parameters is greater than a second preset number threshold, and the second adjustment model parameters are obtained, it can also be implemented in but not limited to the following manner, including: performing parameter adjustment on the remaining model parameters in a direction that increases the fitness function values ​​of the remaining model parameters in the first adjustment model parameters to obtain adjusted remaining model parameters; performing fitness function calculation on the adjusted remaining model parameters to obtain adjusted fitness function values ​​of the remaining model parameters; screening out a preset number of adjusted remaining model parameters in order from large to small of the fitness function values ​​of the adjusted remaining model parameters, until the number of parameter adjustment of the adjusted remaining model parameters is greater than the second preset number threshold; and determining the screened preset number of adjusted remaining model parameters as the second adjustment model parameters.

[0061] By repeatedly screening parameters in descending order according to the fitness function value, we can gradually focus on parameters with better performance and remove those parameters that contribute less to the improvement of model performance, thereby obtaining better quality second adjustment model parameters.

[0062] As a refinement of the above embodiment, when using the training performance operation status data to train the initial performance prediction model to obtain the preset performance prediction model and initial model parameters, it can be implemented in but not limited to the following ways, including: normalizing the training performance operation status data to obtain normalized data for training; inputting the normalized data for training into the initial performance prediction model for model training to obtain the preset performance prediction model and initial model parameters.

[0063] Normalization eliminates the dimensional differences between different features, allowing the model to treat each feature more fairly and avoiding excessive impact of certain features on model training due to excessive numerical range, thereby improving the model training effect and prediction accuracy.

[0064] In practical applications, after the normalized data is input into the preset performance prediction model for performance prediction processing and the performance prediction result is obtained, the performance prediction result needs to be restored, which can be achieved by but is not limited to the following methods, including: denormalizing the performance prediction result to obtain the true value of the performance prediction result.

[0065] Denormalization refers to the process of restoring normalized data to the original data scale. During model training, the data is normalized to eliminate the dimensional differences between different features; after the prediction results are obtained, they need to be denormalized to obtain a true value with practical significance. The true value of the performance prediction result: the performance prediction value with actual physical meaning and measurement unit obtained after denormalization. For example, the above CPU utilization normalization value 0.6 is denormalized to obtain the true CPU utilization percentage, such as 60%.

[0066] The denormalization process converts the normalized performance prediction results into real values ​​with actual physical meaning and measurement units, so that users can intuitively understand the prediction results and facilitate decision-making and analysis.

[0067] As a refinement of step 101, when obtaining the performance operation status data of the server, it can be implemented in but not limited to the following ways, including: obtaining first status data of used computing resources of the server, and obtaining second status data of idle computing resources of the server; the performance operation status data includes first status data and second status data.

[0068] The first state data of used computing resources refers to the data related to the computing resources that have been occupied and used by the server during operation, such as the number of used CPU cores, used memory capacity, used disk space, number of established network connections, etc. These data reflect the current load and resource usage of the server. The second state data of idle computing resources refers to the data related to the computing resources that have not been occupied and are available for use by the server, such as the number of idle CPU cores, idle memory capacity, available disk space, number of established network connections, etc. These data reflect the current resource surplus and potential processing capacity of the server.

[0069] By obtaining the first state data of used computing resources and the second state data of idle computing resources, it is possible to fully understand the resource usage and remaining capacity of the server. This helps system administrators to promptly discover server load problems, reasonably arrange resources, and avoid resource waste or shortage.

[0070] In order to better understand the whole process of server performance prediction, Figure 4 As shown, Figure 4 A flow chart of the entire process of performance prediction of a server provided in an embodiment of the present application, wherein the differential evolution algorithm and the gray wolf optimization algorithm are used to find the optimal parameter value; after obtaining the optimal parameter, the support vector machine model is combined to solve the prediction problem. When the training set is trained by the support vector machine using the differential evolution algorithm and the gray wolf optimization algorithm, two optimal parameters C and γ can be obtained, and then the values ​​of these two parameters of the support vector machine model can be determined, and then the input value of the data set to be predicted is input to obtain the predicted output value.

[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0072] The embodiment of the present application also provides a performance prediction device for a server, Figure 5 A schematic diagram of the structure of a performance prediction device for a server provided in an embodiment of the present application is shown in FIG. Figure 5 As shown, including: An acquisition unit 21 is used to acquire the performance operation status data of the server; The processing unit 22 is used to normalize the performance operation status data to obtain normalized data; The prediction unit 23 is used to input the normalized data into the preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjustment model parameters of the preset performance prediction model are to use the preset parameter adjustment algorithm to adjust the initial model parameters in the direction of increasing the fitness function value of the initial model parameters of the preset performance prediction model until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, thereby obtaining the parameters.

[0073] Through this application, by obtaining the performance operation status data of the server and normalizing the performance operation status data, normalized data is obtained to unify the data scale between different performance operation status data and reduce the data differences between different performance operation status data, and then the normalized data is input into the preset performance prediction model to obtain the performance prediction result, wherein the adjustment model parameters of the preset performance prediction model are adjusted by using the preset parameter adjustment algorithm, in the direction of increasing the fitness function value of the initial model parameters of the preset performance prediction model, the initial model parameters are adjusted until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, the obtained parameters, the fitness function value can reflect the global optimal solution, avoid falling into the local optimal solution, and realize the search for the global optimal solution in the parameter space, thereby improving the accuracy of server performance prediction. Therefore, the technical problem of poor performance prediction effect of the server can be solved, and the technical effect of improving the accuracy of server performance prediction can be achieved.

[0074] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the device also includes: The acquisition unit 21 is also used to acquire the performance running status data for training; The training unit 24 is used to train the initial performance prediction model using the training performance operation status data to obtain a preset performance prediction model and initial model parameters; The adjustment unit 25 is used to use a preset parameter adjustment algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters exceeds a preset number threshold, thereby obtaining the adjusted model parameters.

[0075] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the preset parameter adjustment algorithm includes a preset differential evolution algorithm and a preset gray wolf parameter search algorithm, and the adjustment unit 25 includes: A first adjustment module 251 is used to use a preset differential evolution algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters is greater than a first preset number threshold, thereby obtaining first adjusted model parameters; The second adjustment module 252 is used to use a preset gray wolf parameter search algorithm to adjust the first adjustment model parameter in a direction that increases the fitness function value of the first adjustment model parameter, until the number of parameter adjustments of the first adjustment model parameter is greater than a second preset number threshold, thereby obtaining the second adjustment model parameter; the preset number threshold includes a first preset number threshold and a second preset number threshold; The determination module 253 is configured to determine the second adjustment model parameter as the adjustment model parameter.

[0076] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the first adjustment module 251 is also used for: randomly divide all model parameters in the initial model parameters into at least two model parameter sets; Performing fitness function calculations on at least two model parameter sets respectively to obtain fitness function values ​​corresponding to the at least two model parameter sets; According to the fitness function values ​​corresponding to at least two model parameter sets, the model parameters of at least two model parameter sets are adjusted in a direction that increases the fitness function values ​​of the model parameter sets, until the number of parameter adjustments of the model parameter sets is greater than a first preset number threshold, thereby obtaining first adjusted model parameters.

[0077] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the first adjustment module 251 is also used for: Screening out a target model parameter set having a fitness function value greater than a first preset fitness threshold value from at least two model parameter sets; Adjust the model parameters in the remaining model parameter set in a direction that increases the fitness function value of the model parameters in the remaining model parameter set to obtain an adjusted remaining model parameter set; the remaining model parameter set is any model parameter set in at least two model parameter sets except the target model parameter set; Performing fitness function calculation on the adjusted remaining model parameter set to obtain the fitness function value of the adjusted remaining model parameter set; Filter out the model parameter sets whose fitness function values ​​are greater than the first preset fitness threshold value from the remaining model parameter sets, until the number of parameter adjustments of the model parameters in the remaining model parameter sets is greater than the first preset number threshold value; The model parameters in the target model parameter set and the screened remaining model parameter sets are determined as first adjustment model parameters.

[0078] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the second adjustment module 252 is also used for: Perform fitness function calculation on all model parameters in the first adjustment model parameters to obtain fitness function values ​​corresponding to all model parameters in the first adjustment model parameters; According to the fitness function values ​​corresponding to all model parameters in the first adjustment model parameters, parameter adjustment is performed on all model parameters in the first adjustment model parameters in a direction that increases the fitness function values ​​of all model parameters in the first adjustment model parameters until the number of parameter adjustments of all model parameters in the first adjustment model parameters is greater than a second preset number threshold, thereby obtaining the second adjustment model parameters.

[0079] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the second adjustment module 252 is also used for: Filter out the model parameters from a preset number of second adjustment model parameters in the order of fitness function values ​​from large to small; The model parameters among the screened preset number of second adjustment model parameters are determined as the second adjustment model parameters.

[0080] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the second adjustment module 252 is also used for: Adjust the remaining model parameters in a direction that increases the fitness function value of the remaining model parameters in the first adjustment model parameters to obtain adjusted remaining model parameters; Performing fitness function calculation on the adjusted remaining model parameters to obtain fitness function values ​​of the adjusted remaining model parameters; Screening out a preset number of adjusted remaining model parameters according to the fitness function values ​​of the adjusted remaining model parameters from large to small, until the number of parameter adjustments of the adjusted remaining model parameters is greater than a second preset number threshold; The preset number of adjusted remaining model parameters that are screened out are determined as second adjusted model parameters.

[0081] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the training unit 24 is also used for: Normalizing the training performance running status data to obtain normalized training data; The normalized training data is input into the initial performance prediction model for model training to obtain a preset performance prediction model and initial model parameters.

[0082] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the device also includes: The processing unit 22 is also used to, after inputting the normalized data into a preset performance prediction model for performance prediction processing and obtaining a performance prediction result, perform denormalization processing on the performance prediction result to obtain a true value of the performance prediction result.

[0083] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the acquisition unit 21 is also used for: The first status data of the used computing resources of the server is obtained, and the second status data of the idle computing resources of the server is obtained; the performance operation status data includes the first status data and the second status data.

[0084] For the description of the features in the embodiment corresponding to the performance prediction device of the server, reference can be made to the relevant description of the embodiment corresponding to the performance prediction method of the server, which will not be repeated here.

[0085] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any of the above-mentioned server performance prediction method embodiments.

[0086] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above-mentioned server performance prediction method embodiments when running.

[0087] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0088] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any of the above-mentioned server performance prediction method embodiments are implemented.

[0089] An embodiment of the present application also provides another computer program product, including a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in any of the above-mentioned server performance prediction method embodiments are implemented.

[0090] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0091] The above is a detailed introduction to a performance prediction method and device for a server, an electronic device, and a storage medium provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A method for predicting server performance, characterized in that: include: Get the performance status data of the server; Normalizing the performance operation status data to obtain normalized data; The normalized data is input into a preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjusted model parameters of the preset performance prediction model are obtained by using a preset parameter adjustment algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters of the preset performance prediction model until the number of parameter adjustments of the initial model parameters is greater than a preset number threshold.

2. The method for predicting server performance according to claim 1, characterized in that: The method further comprises: Obtaining performance running status data for training; Using the training performance operation status data, the initial performance prediction model is trained to obtain the preset performance prediction model and the initial model parameters; The preset parameter adjustment algorithm is used to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters is greater than the preset number threshold, thereby obtaining the adjusted model parameters.

3. The method for predicting server performance according to claim 2, characterized in that: The preset parameter adjustment algorithm includes a preset differential evolution algorithm and a preset gray wolf parameter search algorithm. The preset parameter adjustment algorithm is used to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters exceeds the preset number threshold, and the adjusted model parameters are obtained including: Using the preset differential evolution algorithm, adjusting the initial model parameters in a direction that increases the fitness function value of the initial model parameters, until the number of parameter adjustments of the initial model parameters is greater than a first preset number threshold, thereby obtaining first adjusted model parameters; Using the preset gray wolf parameter search algorithm, adjusting the first adjustment model parameter in a direction that increases the fitness function value of the first adjustment model parameter until the number of parameter adjustments of the first adjustment model parameter is greater than a second preset number threshold, thereby obtaining a second adjustment model parameter; the preset number threshold includes the first preset number threshold and the second preset number threshold; The second adjustment model parameter is determined as the adjustment model parameter.

4. The method for predicting server performance according to claim 3, characterized in that: The using the preset differential evolution algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters until the number of parameter adjustments of the initial model parameters is greater than a first preset number threshold, and obtaining the first adjusted model parameters includes: Randomly dividing all model parameters in the initial model parameters into at least two model parameter sets; Performing fitness function calculations on the at least two model parameter sets respectively to obtain fitness function values ​​corresponding to the at least two model parameter sets respectively; According to the fitness function values ​​corresponding to each of the at least two model parameter sets, the model parameters of the at least two model parameter sets are adjusted in a direction that increases the fitness function values ​​of the model parameter sets, until the number of parameter adjustments of the model parameter sets is greater than the first preset number threshold, thereby obtaining the first adjusted model parameters.

5. The method for predicting server performance according to claim 4, characterized in that: The step of adjusting the model parameters of the at least two model parameter sets according to the fitness function values ​​corresponding to the at least two model parameter sets in a direction that increases the fitness function values ​​of the model parameter sets until the number of parameter adjustments of the model parameter sets is greater than the first preset number threshold, and obtaining the first adjusted model parameters includes: Filter out a target model parameter set having a fitness function value greater than a first preset fitness threshold value from the at least two model parameter sets; Adjust the model parameters in the remaining model parameter set in a direction that increases the fitness function value of the model parameters in the remaining model parameter set to obtain the adjusted remaining model parameter set; the remaining model parameter set is any model parameter set in the at least two model parameter sets except the target model parameter set; Performing fitness function calculation on the adjusted remaining model parameter set to obtain a fitness function value of the adjusted remaining model parameter set; Filter out the model parameter sets whose fitness function values ​​are greater than the first preset fitness threshold value from the remaining model parameter sets, until the number of parameter adjustments of the model parameters in the remaining model parameter sets is greater than the first preset number threshold value; The target model parameter set and the filtered model parameters in the remaining model parameter set are determined as the first adjustment model parameters.

6. The method for predicting server performance according to claim 3, characterized in that: The step of using the preset gray wolf parameter search algorithm to adjust the first adjustment model parameter in a direction that increases the fitness function value of the first adjustment model parameter until the number of parameter adjustments of the first adjustment model parameter is greater than a second preset number threshold, and obtaining the second adjustment model parameter includes: Performing fitness function calculation on all model parameters in the first adjustment model parameters to obtain fitness function values ​​corresponding to all model parameters in the first adjustment model parameters; According to the fitness function values ​​corresponding to all model parameters in the first adjustment model parameters, parameter adjustment is performed on all model parameters in the first adjustment model parameters in a direction that increases the fitness function values ​​of all model parameters in the first adjustment model parameters until the number of parameter adjustments of all model parameters in the first adjustment model parameters is greater than the second preset number threshold, so as to obtain the second adjustment model parameters.

7. The method for predicting server performance according to claim 6, characterized in that: According to the fitness function values ​​corresponding to all the model parameters in the first adjustment model parameters, in a direction that increases the fitness function values ​​of all the model parameters in the first adjustment model parameters, adjusting all the model parameters in the first adjustment model parameters until the number of parameter adjustments of all the model parameters in the first adjustment model parameters is greater than the second preset number threshold, and obtaining the second adjustment model parameters includes: Filtering out model parameters from a preset number of the second adjustment model parameters in descending order of the fitness function values; The model parameters among the preset number of the second adjustment model parameters screened out are determined as the second adjustment model parameters.

8. The method for predicting server performance according to claim 7, characterized in that: The step of adjusting all the model parameters in the first adjusting model parameters according to the fitness function values ​​corresponding to all the model parameters in the first adjusting model parameters in a direction that increases the fitness function values ​​of all the model parameters in the first adjusting model parameters until the number of parameter adjustments of all the model parameters in the first adjusting model parameters exceeds the second preset number threshold, and obtaining the second adjusting model parameters further includes: Adjust the remaining model parameters in a direction that increases the fitness function value of the remaining model parameters in the first adjustment model parameters to obtain the adjusted remaining model parameters; Performing fitness function calculation on the adjusted remaining model parameters to obtain fitness function values ​​of the adjusted remaining model parameters; Filtering the preset number of the adjusted remaining model parameters in descending order of the fitness function values ​​of the adjusted remaining model parameters until the number of parameter adjustments of the adjusted remaining model parameters is greater than the second preset number threshold; The preset number of the adjusted remaining model parameters that are screened out are determined as the second adjusted model parameters.

9. The method for predicting server performance according to claim 2, characterized in that: The using the training performance operation status data to train the initial performance prediction model to obtain the preset performance prediction model and the initial model parameters includes: Normalizing the training performance running state data to obtain normalized training data; The normalized training data is input into the initial performance prediction model for model training to obtain the preset performance prediction model and the initial model parameters.

10. The method for predicting server performance according to claim 1, characterized in that: After inputting the normalized data into a preset performance prediction model for performance prediction processing to obtain a performance prediction result, the method further includes: The performance prediction result is denormalized to obtain a true value of the performance prediction result.

11. The method for predicting server performance according to claim 1, characterized in that: The obtaining of the performance operation status data of the server includes: The first status data of the used computing resources of the server is obtained, and the second status data of the idle computing resources of the server is obtained; the performance operation status data includes the first status data and the second status data.

12. A performance prediction device for a server, characterized in that: include: An acquisition unit, used to acquire performance operation status data of the server; A processing unit, used for normalizing the performance operation status data to obtain normalized data; A prediction unit is used to input the normalized data into a preset performance prediction model for performance prediction processing to obtain a performance prediction result; the adjustment model parameters of the preset performance prediction model are obtained by using a preset parameter adjustment algorithm to adjust the initial model parameters in a direction that increases the fitness function value of the initial model parameters of the preset performance prediction model until the number of parameter adjustments of the initial model parameters is greater than a preset number threshold.

13. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the method for predicting server performance as claimed in any one of claims 1 to 11 when executing the computer program.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the performance prediction method for a server as claimed in any one of claims 1 to 11.

15. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the performance prediction method of the server as claimed in any one of claims 1 to 11 are implemented.

Images