CN114385256A

CN114385256A - Method and device for configuring system parameters

Info

Publication number: CN114385256A
Application number: CN202011136341.3A
Authority: CN
Inventors: 谢鹏程; 孙涛; 雷晓松
Original assignee: Huawei Cloud Computing Technologies Co Ltd
Current assignee: Huawei Cloud Computing Technologies Co Ltd
Priority date: 2020-10-22
Filing date: 2020-10-22
Publication date: 2022-04-22
Anticipated expiration: 2040-10-22
Also published as: CN114385256B

Abstract

The application provides a configuration method and a configuration device of system parameters, which combine a parameter configuration rule system and a machine learning system, so that the configuration parameter value range output by the parameter configuration rule system restrains the configuration parameter value range output by the machine learning system, and the machine learning system is helped to carry out optimization solution in a region with a more possible optimal solution, thereby obtaining a more accurate target value range of the configuration parameters.

Description

Method and device for configuring system parameters

Technical Field

The present application relates to the field of information technologies, and in particular, to a method and an apparatus for configuring system parameters.

Background

With the development of electronic information technology and artificial intelligence, the use of Information Technology (IT) facilities, such as computers, hard disks, video cloud transcoding systems or database systems, and the like, has become more and more common.

The IT facilities include hardware facilities and software facilities. Whether a hardware facility or a software facility is faced with different users, different operating environments and/or different traffic loads, the configuration parameters are often varied, because if the same configuration parameters are used for uniform configuration, the performance maximization of the IT facility cannot be realized.

According to the method for configuring the parameters of the IT facility to enable the performance of the IT facility to be better, the maximum value range of the configuration parameters of the IT facility and the performance requirements of the IT facility can be input into a machine learning system, then the machine learning system selects the value range of the configuration parameters which can enable the IT equipment to meet the performance requirements from the maximum value range of the configuration parameters based on a specified machine learning algorithm, and performs parameter configuration on the IT facility based on the value range of the configuration parameters selected by the machine learning system.

However, since the machine learning system is essentially a black box model, the value range of the configuration parameters selected based on the machine learning system may be inaccurate, which may cause the IT facility to be abnormal and have potential safety hazards.

Disclosure of Invention

The application provides a method for configuring system parameters, which realizes more accurate target value range of the configuration parameters and further increases the robustness of a machine learning system.

In a first aspect, the present application provides a method for configuring system parameters. The configuration method of the system parameters comprises the following steps: acquiring configuration information of a target system, wherein the configuration information is used for indicating the value range of at least one configuration parameter of the target system; outputting M parameter value ranges by using a parameter configuration rule system based on configuration information, wherein the M parameter value ranges correspond to M configuration parameters of a target system one by one, and M is a positive integer; outputting N first parameter value ranges by using a machine learning system based on configuration information, wherein the N first parameter value ranges correspond to N configuration parameters of a target system one by one, N is a positive integer greater than or equal to M, and the N configuration parameters comprise M configuration parameters; determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, wherein the target value range of the first configuration parameter in the N configuration parameters comprises a parameter value range corresponding to the first configuration parameter in the M parameter value ranges and all or part of the overlapped parameter value ranges corresponding to the first configuration parameter in the N first parameter value ranges; and configuring the target system according to the target value range of each configuration parameter in the N configuration parameters.

According to the method for configuring the system parameters, the parameter configuration rule system is combined with the machine learning system, so that the parameter value range output by the parameter configuration rule system is restricted by the parameter value range output by the machine learning system, the machine learning system is helped to carry out optimization solution in the area with the most probable optimal solution, and the more accurate target value range of the configuration parameters is obtained.

With reference to the first aspect, in a possible implementation manner, configuring a target system according to a target value range of each of N configuration parameters includes: outputting N second parameter value ranges based on the performance and configuration information of the target system by using a machine learning system; re-determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges, wherein the target value range of a first configuration parameter in the N configuration parameters comprises all or part of the parameter value ranges which are overlapped with the parameter value range corresponding to the first configuration parameter in the M parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N second parameter value ranges; and configuring the target system according to the target value range determined for each configuration parameter in the N configuration parameters.

The method for configuring the system parameters enables the machine learning system to continuously update the configuration parameter value range of the target system under the constraint of the parameter configuration rule system, so that a more accurate target value range is obtained; in addition, when the performance of the target system is larger than the requirement threshold, the target system is configured according to the default configuration parameters, and the robustness of the machine learning system is improved.

With reference to the first aspect, in a possible implementation manner, configuring a target system according to a target value range of each of N configuration parameters includes: and if the performance of the target system exceeds the performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

According to the configuration method of the system parameters, when the performance of the target system exceeds the performance threshold value, the target system is configured by using the default configuration parameters, and the robustness of the machine learning system is improved.

With reference to the first aspect, in a possible implementation manner, the target value range of the second configuration parameter in the N configuration parameters includes a third parameter value range in the parameter value range corresponding to the second configuration parameter in the N first parameter value ranges, and the third parameter value range includes a parameter value range corresponding to the second configuration parameter in the M parameter value ranges and all or part of the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, which are not overlapped with each other.

According to the configuration method of the system parameters, the target value range of the second configuration parameter of the target system can be not included in the value range output by the parameter configuration rule system, so that a more accurate target value range can be obtained under the condition that the value range output by the parameter configuration rule system is inaccurate.

With reference to the first aspect, in a possible implementation manner, the target value range of any one of the N configuration parameters, except for the first configuration parameter, is all or part of the value range, which is overlapped with the parameter value range corresponding to any one of the M parameter value ranges and the parameter value range corresponding to any one of the N first parameter value ranges.

According to the configuration method of the system parameters, the value ranges of the first parameters output by the machine learning system are required to be included in the value ranges of the output parameters of the system according to the parameter configuration rule, so that the safety performance of a target system is improved while a more accurate target value range is obtained.

With reference to the first aspect, in one possible implementation manner, the machine learning system is implemented based on a bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

In a second aspect, the present application provides an apparatus for configuring system parameters, including: the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring configuration information of a target system, and the configuration information is used for indicating the value range of at least one configuration parameter of the target system; a first parameter output module, configured to output M parameter value ranges based on configuration information using a parameter configuration rule system, where the M parameter value ranges correspond to M configuration parameters of the target system one to one, and M is a positive integer; a second parameter output module, configured to output, by using a machine learning system, N first parameter value ranges based on configuration information, where the N first parameter value ranges correspond to N configuration parameters of the target system one to one, N is a positive integer greater than or equal to M, and the N configuration parameters include the M configuration parameters; the processing module is used for determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, wherein the target value range of the first configuration parameter in the N configuration parameters comprises all or part of the parameter value ranges which are overlapped with the parameter value range corresponding to the first configuration parameter in the M parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N first parameter value ranges; and the configuration module is used for configuring the target system according to the target value range of each configuration parameter in the N configuration parameters.

With reference to the second aspect, in a possible implementation manner, the second parameter output module is further configured to: outputting N second parameter value ranges based on the performance and configuration information of the target system by using a machine learning system; the processing module is further configured to: re-determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges, wherein the target value range of a first configuration parameter in the N configuration parameters comprises all or part of the parameter value ranges which are overlapped with the parameter value range corresponding to the first configuration parameter in the M parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N second parameter value ranges; the configuration module is further to: and configuring the target system according to the target value range determined for each configuration parameter in the N configuration parameters.

With reference to the second aspect, in a possible implementation manner, the configuration module is further configured to: and if the performance of the target system exceeds the performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

With reference to the second aspect, in a possible implementation manner, the target value range of the second configuration parameter in the N configuration parameters includes a third parameter value range in the parameter value range corresponding to the second configuration parameter in the N first parameter value ranges, and the third parameter value range includes a parameter value range corresponding to the second configuration parameter in the M parameter value ranges and all or part of the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, which are not overlapped with each other.

With reference to the second aspect, in a possible implementation manner, the target value range of any one of the N configuration parameters, except for the first configuration parameter, is all or part of the value range, which is overlapped with the parameter value range corresponding to any one of the M parameter value ranges and the parameter value range corresponding to any one of the N first parameter value ranges.

With reference to the second aspect, in one possible implementation manner, the machine learning system is implemented based on a bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

In a third aspect, the present application provides a device for configuring system parameters, including: a memory and a processor; the memory is to store program instructions; the processor is configured to call the program instructions in the memory to execute the method for configuring the system parameters according to the first aspect or any one of the possible implementation manners.

In a fourth aspect, the present application provides a chip, which includes at least one processor and a communication interface, where the communication interface and the at least one processor are interconnected by a line, and the at least one processor is configured to execute a computer program or instructions to perform the method for configuring system parameters according to the first aspect or any one of the possible implementation manners.

In a fifth aspect, the present application provides a computer-readable medium storing program code for execution by a device, the program code comprising instructions for performing a method of configuring system parameters as described in the first aspect or any one of its possible implementations.

In a sixth aspect, the present application provides a computer program product containing instructions, which when run on a computer, causes the computer to perform the method for configuring system parameters according to the first aspect or any one of the possible implementations.

In a seventh aspect, the present application provides a server, including at least one processor and a communication interface, where the communication interface and the at least one processor are interconnected by a line, the communication interface is in communication with a target system, and the at least one processor is configured to execute a computer program or instructions to perform the method for configuring system parameters according to the first aspect or any one of the possible implementations.

Drawings

FIG. 1 is a schematic diagram of an IT system provided in accordance with an embodiment of the present application;

FIG. 2 is a schematic diagram of a parameter configuration system according to an embodiment of the present application;

FIG. 3 is a schematic flow chart diagram of a method for configuring system parameters according to an embodiment of the present application;

FIG. 4 is a schematic flow chart diagram of a method for configuring system parameters according to another embodiment of the present application;

fig. 5 is a schematic structural diagram of a configuration apparatus for system parameters according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a configuration device according to another embodiment of the present application.

Detailed Description

For understanding, the relevant terminology referred to in this application will be first described.

1. Rules engine

The rules engine evolved from the inference engine and is a component embedded in the application that separates business decisions from the application code and can write rules using predefined semantic modules.

It expresses the basic knowledge in a regular form and can be generally expressed in an if-then form, with a portion being a corresponding action if it contains given information or factors. For example, a rule may be a statement in the form "if this condition is satisfied, its action should be taken". For example, if an animal is a mammal and eats meat, such an animal is referred to as a carnivore.

Rules are a type of knowledge, and their typical use is to draw conclusions from a given set of rules through the actual situation. This conclusion may be a static result or a set of operations that need to be performed. The process of applying such rules is called reasoning. If the inference process is handled by a program, then the program is called an inference engine.

The control strategies employed by the inference engine based on the difference in knowledge representation are also different, common types include neural network-based, case-based and rule-based inference engines. Among them, the rule-based inference engine is also called "rule engine", which is easy to understand, easy to acquire, and easy to manage.

2. Machine learning

Machine learning is a general term for a class of algorithms that attempt to mine the rules implied therein from a large amount of historical data and use them for prediction or classification. More specifically, machine learning can be viewed as finding a function, with the input being sample data and the output being the desired result. In general, this function is relatively complex and not convenient for formalization.

It will be appreciated that the goal of machine learning is to adapt the learned function well to the "new sample", and not just to perform well on the training sample.

Generally, an algorithm in machine learning is used to find a function, which is mainly divided into the following three steps:

in the first step, a suitable model is selected, which usually depends on the actual problem, and an appropriate model is selected for different problems and tasks, and the model is a set of functions.

And secondly, judging the quality of a function, which needs to determine a measurement standard, namely a loss function, wherein the determination of the loss function also needs to be determined according to specific problems, for example, the regression problem generally adopts Euclidean distance, and the classification problem generally adopts a cross entropy cost function.

And thirdly, finding out the best function. The common methods are gradient descent algorithm or least square method.

After learning the "best" function, a test can be performed on the new sample to test whether the function is truly a "good" function.

3. Genetic algorithm

A great number of optimization and self-adaptation related problems exist in modern scientific theory research and practice, but people still have no power to the optimization and self-adaptation problems of large complex systems except some simple conditions. However, organisms in nature exhibit excellent ability in adaptation, and they can live and reproduce in the self-evolution rule of excellence and disadvantage, and the suitable ones survive, and gradually produce excellent species with high adaptability to their living environments. Genetic algorithms are search algorithms based on natural selection and population genetic mechanisms, which simulate the phenomena of reproduction, hybridization and mutation during natural selection and natural inheritance. When the problem is solved by using a genetic algorithm, each possible solution of the problem is coded into a chromosome, namely an individual, and a plurality of individuals form a population. At the start of the genetic algorithm, individuals are always generated randomly, each of which is evaluated according to a predetermined objective function, giving a fitness value. Based on the fitness value, some individuals are selected to generate the next generation, and the selection operation embodies the principle of survival of the fittest. The "good" individuals are used to generate the next generation. The "bad" individuals are eliminated. Then, the selected individuals are recombined through a crossover and mutation operator to generate a new generation, and the individuals of the generation inherit some excellent properties of the previous generation, so that the performance of the individuals of the generation is superior to that of the previous generation, and the individuals of the generation gradually evolve towards the direction of the optimal solution. Thus, genetic algorithms can be viewed as a process of population evolutionary initials consisting of feasible solutions.

When solving a problem by using a genetic algorithm, an objective function and a variable of the problem are determined firstly, and then the variable is coded, so that the problem solution is expressed by using a digital string in the genetic algorithm, and a genetic operator directly operates on the string. Encoding variables can be divided into binary encoding and real number encoding. And then determining the coding modes of the target function and the variable, and then carrying out genetic operation. From the perspective of optimization search, genetic operations can optimize the solution of the problem generation by generation, approaching the optimal solution, and the genetic operations can include the following three genetic operators: selection, crossover, and mutation. Wherein, selection and crossover basically complete most of the search functions of the genetic algorithm, and mutation increases the ability of the genetic algorithm to find the optimal solution. Selection, crossover and variation in genetic manipulation are explained below, respectively.

Selecting: selection refers to an operation of selecting good individuals from a population and eliminating poor individuals. It is based on fitness evaluation. The more suitable the individual is, the more likely it is to be selected, and the more "descendants" of the individual are in the next generation. The selection methods commonly used at present include a roulette method, an optimal individual retention method, an expectation method, a ranking selection method, a competition method, a linear standardization method, and the like.

And (3) crossing: crossover is an operation of generating a new individual by replacing and recombining partial structures of two parent individuals, and the purpose of crossover is to generate a new individual in the next generation, and the search ability of a genetic algorithm is dramatically improved by crossover operation. The crossing is an important means for acquiring excellent individuals by a genetic algorithm, two individuals are randomly selected from a matching library according to a certain crossing probability, the crossing positions are also random, and the crossing probability is generally large and ranges from 0.6 to 0.9.

Mutation: mutation is to randomly change the value of some genes of individuals in a population with a small mutation probability, and the basic process of mutation operation is as follows: generating a random number between 0 and 1, and if the random number is less than the mutation probability, performing mutation operation. The mutation operation is a local random search, and is combined with the selection operator and the crossover operator, so that the permanent loss of some information caused by the selection operator and the crossover operator can be avoided, the effectiveness of the genetic algorithm is ensured, the genetic algorithm has the local random search capability, and meanwhile, the diversity of the population can be kept for the genetic algorithm to prevent the occurrence of premature convergence.

4. Particle swarm algorithm

In the field of computer intelligence, and in particular, swarm intelligence optimization algorithms, one of the most commonly used algorithms at present is the Particle Swarm Optimization (PSO). PSO is an evolutionary computing technique, derived from behavioral studies on bird predation, and has proven to be a good optimization method.

The PSO algorithm is to initialize a group of particles in a feasible solution space, each particle represents a potential optimal solution of the extremum optimization problem, and the characteristics of the particles are represented by three indexes of position, speed and fitness value. The particles move in the solution space, and the individual positions are updated by tracking individual extrema and population extrema.

The individual extreme value refers to the optimal fitness position obtained by calculation in the positions experienced by the individual, and the group extreme value refers to the optimal fitness position searched by all particles in the population. And calculating a fitness value once every time the particles update the positions, and updating the positions of the individual extremum and the group extremum by comparing the fitness of the new particles with the positions of the individual extremum and the group extremum.

5. Bayesian algorithm

Bayesian methods, generally referred to as bayesian analysis methods, provide a method for calculating the probability of an assumption, and were developed by the mathematician of england by bayesian to describe the relationship between two conditional probabilities. By bayesian, it is meant that when the number of samples analyzed is large enough to approximate the population, the probability of occurrence of events in the samples in time will approximate the probability of occurrence of events in the population. In general, the probability of event a under the condition of event B occurrence is different from the probability of event B under the condition of event a; however, the two are in a definite relationship, and the Bayesian method is the description of the relationship. The method is a method for synthesizing the prior information about the unknown parameters and the sample information, obtaining posterior information according to a Bayesian formula, and inferring the unknown parameters according to the posterior information.

FIG. 1 is a schematic diagram of an IT system provided in accordance with an embodiment of the present application. As shown in FIG. 1, the IT system 100 of the present application may include a hardware facility 110 and a software facility 120. Hardware facilities 110 may include handheld devices, desktops, servers, routers, switches, hubs, notebooks, printers, Basic Input Output System (BIOS) systems, etc., and software facilities 120 may include software facilities such as databases, video cloud transcoders, operating systems, etc.

For the IT system shown in FIG. 1, there are a number of configuration parameters that affect ITs performance, both for hardware facility 110 and software facility 120.

The relationship between the configuration parameters and the performance of the IT system is described below by taking the example that the hardware facility 110 is a BIOS system.

The BIOS system is solidified on the mainboard of the computer, stores the most important basic input and output program, system setting information, self-checking program after power-on and system self-starting program, and has the main function of providing the bottommost and most direct hardware setting and control for the computer. In addition, the BIOS system also provides some system configuration parameters to the operating system. When the computer runs, the BIOS system is firstly entered, and whether the computer performance is superior or not depends on whether the BIOS system on the computer mainboard is advanced or not. For example, the BIOS system has a thread number parameter of a Central Processing Unit (CPU), and when the load number is high, the thread number parameter may be adjusted, or when the energy consumption is low, some configuration parameters of the BIOS system may be adjusted, so that the load runs faster or the energy consumption is low.

The relationship between the configuration parameters and performance of the IT system is described below by taking the example where the software facility 120 is a database system.

A database system (DBS) is an ideal data processing system developed to meet the needs of data processing, and is composed of a database and software thereof, wherein the software mainly includes an operating system, a utility program and a database management system. Often, the operating environment and traffic load faced when a database carries different users is also varied. For example, the same DBS may provide some data services to users, and some users may update the database (write data) frequently during interaction with the service system, for example, some users may have intensive reading operations. For another example, there are more duplicate data in some business data, and more columns in some business data. That is, different users have different requirements on the database, and therefore, the same DBS should have different configuration parameters for different users or different service loads. Or, the same configuration parameter in the same DBS may have different values when facing different users. For example, the DBS system has a connection duration parameter thereon, which can be adjusted when the load number is high or low, so as to adjust the time of the load connection.

The relationship between the configuration parameters and the performance of the IT system is described below by taking the software facility 120 as a video cloud transcoding system as an example.

The video cloud transcoding system is a system for converting a video into a video format suitable for being played by a mobile device through a server (cloud end). Generally, when downloading a movie, most of the movies are in Audio Video Interleaved (AVI) formats, and some mobile devices, such as mobile phones, do not support playing in these formats, at this time, a user needs to perform manual movie transcoding operation, and the mobile devices can play the transcoded movies. For video cloud transcoding, a set of configuration parameters such as GOP duration parameters in video encoding and decoding are included. There may be some effect on the black screen if the GOP duration is long. Different users have different requirements on video cloud transcoding and may also have different requirements in different time periods, for example, at night, the load of a video cloud transcoding system is high, and at this time, corresponding configuration parameters in video encoding and decoding need to be adjusted.

In the IT system 100 shown in fig. 1, no matter the hardware facility 110 or the software facility 120, at present, there are two main methods for configuring parameters of the IT system 100, one is rule engine-based configuration optimization, that is, a business expert extracts tuning logic according to experience, formulates an optimization rule for a specific system, and then forms a rule engine to configure and optimize system parameters; another is configuration optimization using a machine learning system, such as a machine learning system based on a genetic algorithm or a machine learning system based on a bayesian optimization algorithm, etc.

As an example, fig. 2 is a schematic diagram of a parameter configuration system to which the technical solution provided in the embodiment of the present application can be applied. As shown in fig. 2, the parameter configuration system may include a configuration device and a target system, the configuration device being an external device of the target system. The configuration device can receive initial configuration information of the target system and/or performance requirement information of the target system in a cloud service mode, and configures parameters of the target system based on the initial configuration information and/or the performance requirement information of the target system in the cloud service mode.

For example, when optimizing BIOS system parameter configuration using a configuration device including a machine learning system, the configuration device including the machine learning system may send the learned BIOS system parameters to the BIOS system. It should be noted that the configuration device of the parameter configuration system may be a configuration device including a rule engine algorithm, or may be a configuration device including a machine learning system. The target system may be any of the hardware facilities 110 or may also be any of the software facilities 120.

However, when facing massive parameters, due to limited experience of service experts, it is difficult to form a better rule engine and find an optimal solution; in the process of searching for the optimal parameter, because the machine learning system is a black box model essentially, the problem of poor learned system parameters may occur, thereby causing the abnormal condition of the system and potential safety hazards.

In order to solve the problems, the application provides a configuration method of system parameters, and the rule engine is used as prior information by combining configuration optimization based on the rule engine and configuration optimization of a machine learning system, so that the machine learning system is helped to perform optimal solution in a region with a more probable optimal solution, and more accurate parameters are obtained. In addition, when the performance of the target system exceeds a performance threshold, the target system is configured by using default configuration parameters, and the robustness of the machine learning system is increased.

Fig. 3 is a schematic flowchart of a configuration method of system parameters according to an embodiment of the present application. As shown in fig. 3, the method of the present embodiment may include S301, S302, S303, S304, and S305. The configuration method may be performed by the configuration apparatus shown in fig. 2.

S301, obtaining configuration information of the target system. The configuration information is used for indicating the value range of at least one attribute of the configuration parameter of the target system.

In this embodiment, the target system refers to a system to be tuned that includes configuration parameters, and the target system may be used to implement one or more functions. For example, the target system may be the IT system 100 shown in FIG. 1.

The configuration information in this embodiment is used to indicate a value range of at least one configuration parameter of the target system (i.e., the system to be tuned and optimized). A configuration parameter may include one or more attributes, and the value range of the configuration parameter may include the value range of each attribute of the configuration parameter. For example, the attribute of a configuration parameter may include two attributes of a numeric type and a numeric size, where the numeric type includes an enumeration type, an integer type, and a floating point type, and the numeric size includes an integer between 1 and 5.

It is understood that the configuration information may be different for different target systems, that is, the value range of the configuration parameter of the system to be tuned may be different. For example, the number of configuration parameters of the BIOS system, the value range of each configuration parameter, the type value range of each configuration parameter, and the number of configuration parameters of the DBS system, the value range of each configuration parameter, and the type value range of each configuration parameter may be different.

In one implementation, the configuration information of the target system may be recorded in a configuration file in JSON format, which is an abbreviation of JavaScript Object notification. In this way, the configuration means of the system parameters can read the configuration information from the configuration file.

S302, outputting M parameter value ranges by using a parameter configuration rule system based on the configuration information of the target system. The value ranges of the M parameters are in one-to-one correspondence with M configuration parameters of a target system, wherein M is a positive integer.

In this embodiment, the M configuration parameters may be part of configuration parameters in the target system, or may be all of the configuration parameters in the target system. For example, there are 10 configuration parameters of the target system, and based on the configuration information, the parameter configuration rule system is used, and the output M parameter value ranges may be the value ranges of 5 of the configuration parameters, or the value ranges of the 10 configuration parameters.

In this embodiment, the parameter configuration rule system is a system for optimizing configuration parameters of a target system, and the parameter configuration rule system is generally different for different target systems, and may be designed according to different functions of the target system. For example, for DBS, a parameter rule configuration system may be formulated according to the experience knowledge of experts in the DBS field; for BIOS, a parameter rule configuration system can be formulated according to experience knowledge of experts in the field of BIOS.

Optionally, the parameter configuration rule system in this embodiment may include multiple types of rules.

For example, a parameter configuration rule system may include multiple rules for a configuration parameter. As an example, assuming that the target system has only one configuration parameter a, the parameter configuration rule system is a system for optimizing the configuration parameter a of the target system, and includes two types of rules: under the condition 1, the value range of the parameter a is [1, 2 ]; in the condition 2, the value of the parameter a is a numerical value.

For another example, the parameter configuration rule system may include a plurality of rules for a plurality of configuration parameters. As an example, assuming that the target system has 2 configuration parameters w and z, in condition 1, the value range of w is output by the parameter configuration rule system, and in condition 2, the value range of z is output by the parameter configuration rule system.

In this embodiment, the input of the parameter configuration rule system is configuration information, and the output is M parameter value ranges. The M parameter value ranges are obtained through the configuration information, and the configuration information contains attribute information of the configuration parameters, so that the M parameter value ranges output through the parameter configuration rule system are located in the value ranges of the corresponding configuration information.

In one implementation, the parameter configuration rules system may provide a JSON formatted file that allows a user to enter a series of rules. Wherein the rules may comprise rules expressed if (if) statements.

And S303, outputting N first parameter value ranges by using the machine learning system based on the configuration information of the target system. The N first parameter value ranges correspond to N configuration parameters of the target system one to one, where N is a positive integer greater than or equal to M, and the N configuration parameters include M configuration parameters.

In this embodiment, the N configuration parameters may include all configuration parameters of the target system, or may include part of configuration parameters of the target system. In some implementations, the machine learning system may output value ranges of all configuration parameters of the target system, and in this implementation, the N configuration parameters may include the aforementioned M configuration parameters, that is, N is a positive integer greater than or equal to M.

Optionally, in other implementations, the machine learning system may output a value range of a part of the configuration parameters of the target system. For example, the target system has 10 configuration parameters, and assuming that 1 of the configuration parameters does not need to be optimized, the value ranges corresponding to the remaining 9 configuration parameters may be output by using the machine learning system.

In this embodiment, the machine learning system is a system capable of outputting a value range of a configuration parameter of a target system based on configuration information and performance requirements of the target system by using a machine learning algorithm.

The machine learning system in this embodiment may be a reinitialized system, or a system trained by historical data, that is, a machine learning system trained by historical data before the target system, which is not limited in this application.

The same machine learning algorithm may be used for different target systems, or different machine learning algorithms may be used. For example, for DBS and video cloud transcoding systems, the same machine learning system incorporating bayesian algorithms may be used; alternatively, DBS may use a machine learning system incorporating bayesian algorithms, while video cloud transcoding uses a machine learning system incorporating genetic algorithms.

Similarly, similarly to step S302, the input of the machine learning system is configuration information, and the output is the value ranges of the N first parameters. The N first parameter value ranges are also obtained through the configuration information, and the configuration information contains attribute information of the configuration parameters, so that the N first parameter value ranges output through the machine learning system are located in the value ranges of the corresponding configuration information.

It should be noted that, in the example, the specific implementation process of outputting the value ranges of the N first parameters by using the algorithm principle of the bayesian algorithm and the algorithm principle of the neural network may refer to the description in the related art, and details are not described here.

S304, determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, wherein the target value range of the first configuration parameter in the N configuration parameters comprises all or part of the parameter value ranges corresponding to the first configuration parameter in the M parameter value ranges and the parameter value ranges corresponding to the first configuration parameter in the N first parameter value ranges, which are overlapped.

In this embodiment, the target value range of the first configuration parameter includes all or part of the value ranges that overlap the parameter value range corresponding to the first configuration parameter in the M parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N first parameter value ranges, and may be understood as follows: the value range of the first configuration parameter can neither exceed the value range output by the parameter configuration rule system for the first configuration parameter nor exceed the value range output by the machine learning system for the first configuration parameter.

As an example, one configuration parameter of the target system is denoted by a, a value range of a value size of a output by the parameter configuration rule system includes a value from 0 to 2, and a value range of a value size of a output by the machine learning system is a value from 0 to 5, so that the target value range of a value size of a may include a portion where 0 to 2 and 0 to 5 completely overlap, that is, 0 to 2, and may also include a portion of a value range from 0 to 2, that is, 0 to 1.

As another example, a configuration parameter of the target system is denoted as b, a value range of the value size of b output by the parameter configuration rule system includes a value of 0.6, a value range of the value size of b output by the machine learning system includes a value from 0 to 1, and then the target value range of the value size of b may include a value of 0.6.

It can be understood that, since the target value range of each of the N configuration parameters is influenced by the M parameter value ranges output by the parameter configuration rule system and the N first parameter value ranges output by the machine learning system, the quality of the target value of each of the N configuration parameters is influenced by the parameter configuration rule system and the machine learning system.

S305, configuring the target system according to the target value range of each configuration parameter in the N configuration parameters.

As an example, after obtaining the target value range of each of the N configuration parameters, the final value of the configuration parameter may be selected from the target value range of each configuration parameter, and the value of the configuration parameter in the target system is configured as the final value. For example, a random selection method may be adopted to randomly select the final value of each configuration parameter from the target value range of the configuration parameter. It should be noted that, in this embodiment, the method for selecting the final value of each configuration parameter from the target value range of the configuration parameter is not limited.

It can be understood that, in this embodiment, configuring the target system according to the target value ranges of the N configuration parameters may include: and sending the target value ranges of the N configuration parameters to the target system so that the target system configures the values of the N configuration parameters into the values in the target value ranges.

According to the method for configuring the system parameters, the parameter configuration rule system and the machine learning system are combined, so that the parameter value range output by the parameter configuration rule system constrains the parameter value range output by the machine learning system, and the machine learning system is helped to perform optimal solution in the area with the more possible optimal solution, and therefore the more accurate target value range of the configuration parameters is obtained.

In a possible implementation manner of this embodiment, the first configuration parameter may be any one of the above M configuration parameters. That is to say, the target value range of any one of the N configuration parameters, except the first configuration parameter, is all or part of the value range overlapping with the parameter value range corresponding to any one of the M parameter value ranges and the parameter value range corresponding to any one of the N first parameter value ranges.

Or, it can be said that the target value range of each of the N configuration parameters is all or part of the value range in which the corresponding parameter value range in the M parameter value ranges overlaps with the corresponding parameter value range in the N first parameter value ranges. In other words, the target value range of each configuration parameter of the target system can neither exceed the parameter value range output by the parameter configuration rule system nor exceed the parameter value range output by the machine learning system.

As an example, assuming that the target system has 20 configuration parameters, which respectively correspond to the value ranges of the 20 configuration parameters output by the parameter configuration rule system and the value ranges of the 20 configuration parameters output by the machine learning system, for the target system, the target value ranges of all the configuration parameters should be in a range where the value ranges of the 20 configuration parameters output by the parameter configuration rule system and the value ranges of the 20 configuration parameters output by the machine learning system overlap, that is, the value ranges of the 20 configuration parameters output by the parameter configuration rule system cannot be violated.

According to the method for configuring the system parameters in the implementation mode, the value ranges of the first parameters output by the machine learning system are required to be included in the value ranges of the parameters output by the parameter configuration rule system, so that the safety performance of the target system is improved while a more accurate target value range is obtained.

In another possible implementation manner of this embodiment, the target value range of the second configuration parameter in the N configuration parameters includes a third parameter value range in the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, and the third parameter value range includes all or part of the parameter value ranges corresponding to the second configuration parameter in the M parameter value ranges and the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, which are not overlapped.

In this embodiment, the second configuration parameter refers to at least one configuration parameter of the N configuration parameters, and a third parameter value range of the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges indicates that the third parameter value range is a parameter value range corresponding to the output of the machine learning system.

In this embodiment, the third parameter value range includes all or part of the M parameter value ranges corresponding to the second configuration parameter and the N first parameter value ranges that do not overlap with the parameter value range corresponding to the second configuration parameter, which are configuration parameters of the target system, and may include that the target value range is not within the M parameter value ranges corresponding thereto, that is, may not satisfy the parameter value range output by the parameter configuration rule system, that is, may allow part of the parameter value ranges output by the parameter configuration rule system to violate.

As an example, assume that the target system has two configuration parameters c and d, where for the parameter d, the parameter d output by the parameter configuration rule system is a value range, such as 1 to 2. The parameter d output using the machine learning system is also a range of values, such as 1 to 5. Assuming that the type of the required parameter is an integer, the non-overlapping range of the value range of the parameter d output by the parameter configuration rule system and the value range of the parameter d output by the machine learning system is 2 to 5, and if the method of randomly determining the configuration parameter in 2 to 5 is adopted, the target value range of the determined configuration parameter d may be 3, that is, the value range of the configuration parameter d violates the value range output by the parameter configuration rule system.

As another example, assume two configuration parameters e and f of the target system, wherein for parameter e, the parameter e output by the parameter configuration rule system is a specific value, which is assumed to be 0.5. The parameter e output using the machine learning system is a range of values such as 0 to 1. The determined target value of the configuration parameter e may be 0.6, that is, the target value of the configuration parameter e violates the value range output by the parameter configuration rule system.

In the method for configuring the system parameter in the implementation manner, the target value range of the second configuration parameter of the target system may not be included in the value range output by the parameter configuration rule system, so that a more accurate target value range can be obtained under the condition that the value range output by the parameter configuration rule system is inaccurate.

Optionally, configuring the target system according to the target value range of each configuration parameter in the N configuration parameters includes: and if the performance of the target system exceeds the performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

In this embodiment, after the value ranges of the N first parameters are input into the target system, the performance of the corresponding target system is obtained. Since there may be a case where the performance of the target system is poor after the N first parameter value ranges are input to the target system and then the target system is input, a performance threshold and a default parameter may be specified for the performance of the target system. After configuring the target system based on the N first parameter value ranges, if the performance of the target system exceeds the performance threshold, the target system may configure the target system using default configuration parameters.

Or after configuring the target system based on the N first parameter value ranges, the target system may feed back the performance of the target system to the configuration device, and if the performance is greater than a preset performance threshold, the configuration device reconfigures the default configuration parameters for the target system, for example, re-sends the default configuration parameters to the target system; otherwise, the configuration device may adjust parameters in the machine learning system based on the performance, so that the machine learning system may output a better configuration parameter value range.

As an example, the CPU utilization rate is a performance of the target system, and when the value ranges of the N first parameters are input, the CPU utilization rate of the target system is 92%. However, when the processing speed of the CPU exceeds 90%, the processing speed is reduced. Therefore, it can be specified that a performance threshold of the CPU usage is 90%, and after the value ranges of the N first parameters are input, if the performance threshold of the CPU usage of the target system is greater than 90%, the default configuration parameter is adopted.

In one possible implementation, the performance threshold and the default configuration parameters may be obtained via a JSON-formatted file.

According to the configuration method of the system parameters, when the performance of the target system exceeds the performance threshold, the target system is configured by using the default configuration parameters, and the robustness of the machine learning system is improved.

Fig. 4 is a schematic flowchart of a method for configuring system parameters according to another embodiment of the present application. As shown in fig. 4, the method of the present embodiment may include S401, S402, S403, S404, S405, S406, and S407. The configuration method may be performed by a configuration device of system parameters.

S401, obtaining configuration information of the target system.

In this embodiment, the configuration information is used to indicate a value range of at least one configuration parameter of the target system, which may specifically refer to the relevant description in the foregoing embodiments, and details are not described here again.

S402, processing is carried out by a parameter configuration rule system.

Specifically, the processing of the parameter configuration rule system includes outputting M parameter value ranges based on the configuration information of the target system, where the M parameter value ranges correspond to M configuration parameters of the target system one to one. Based on the configuration information, the parameter configuration rule system outputs M parameter value ranges, which is described in relation to S302 in the embodiment shown in fig. 3 and is not described herein again.

And S403, processing by the machine learning system.

Specifically, the processing by the machine learning system includes outputting, by the machine learning system, N first parameter value ranges based on the configuration information of the target system, where the N first parameter value ranges are in one-to-one correspondence with the N configuration parameters of the target system. For the machine learning system to output the value ranges of the N first parameters based on the configuration information, reference may be made to the description related to S303 in the embodiment shown in fig. 3, which is not described herein again.

S404, determining a target value range of each configuration parameter.

Specifically, determining the target value range of each configuration parameter includes determining the target value range of each configuration parameter according to the M parameter value ranges and the N first parameter value ranges. The implementation process of this step can be described with reference to S304 in the embodiment shown in fig. 3, and is not described here again.

And S405, configuring the target system according to the target value range of each configuration parameter.

For this step, reference may be made to the related description in the above embodiments, and details are not repeated here.

And S406, judging whether the performance of the target system exceeds a performance threshold, if so, executing S403 again, and otherwise, executing S407.

And re-executing, namely feeding back the performance of the target system to the machine learning system, learning by the machine learning system, and re-inputting the value range of the configuration parameters of the target system.

As an example, feeding back the performance of the target system to the machine learning system, the machine learning system learning may include: and outputting N second parameter value ranges based on the performance and configuration information of the target system by using the machine learning system.

And re-determining the target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges. The target value range of a first configuration parameter in the N configuration parameters includes all or part of the value ranges of the M parameter value ranges, which are overlapped with the parameter value range corresponding to the first configuration parameter, and the parameter value ranges of the N second parameter value ranges, which are overlapped with the parameter value range corresponding to the first configuration parameter.

And configuring the target system according to the target value range determined for each configuration parameter in the N configuration parameters.

Wherein the machine learning system may adjust parameters of the machine learning system itself based on performance and configuration information of the target system. For example, a machine learning system including a bayesian algorithm may update the machine learning system by adjusting parameters of the bayesian algorithm based on performance and configuration information of the target system of round 20.

Optionally, the machine learning system is used, based on the performance and configuration information of the target system, the output N second parameter value ranges are different from the corresponding N first parameter value ranges, that is, the N first parameter value ranges are adjusted. For example, the target system contains two configuration parameters m and x, the machine learning system inputs the parameter value ranges of m and x obtained in the first round of learning into the target system to obtain the performance of the corresponding target system, wherein the parameter value range of m is 0 to 5, and the parameter value range of x is 2 to 6; and in 2 second parameter value ranges obtained by the machine learning system through second-round learning according to the performance and configuration information of the target system obtained through the first-round learning, the parameter value range of m is 0 to 4, and the parameter value range of x is 4 to 6.

In this embodiment, the performance of the target system may be considered as the output of the target system, for example, the performance of the target system may be CPU utilization, time, or throughput. The performance threshold refers to a threshold specified for the performance of the target system. For example, the target system's CPU utilization performance threshold may be specified as 90%, or the target system's temporal performance threshold may be specified as 0.5 seconds.

It should be noted that the target systems are different and their corresponding performance may be different.

In this embodiment, after the value ranges of the N first parameters are input into the target system, if the training is not finished, the N configuration parameters may be learned again based on the performance and configuration information of the target system. For example, the machine learning system may perform a second round of learning based on the performance of the target system obtained in the first round, or the machine learning system may perform a tenth round of learning based on the performance of the target system obtained in the ninth round.

It should be appreciated that there are a variety of ways to determine whether the training process is complete. As an example, the training process may be ended by setting the number of iterations of the training, for example, setting the number of iterations not to exceed 100 rounds, that is, when the number of iterations is greater than 100 rounds. As another example, whether the training process is finished may be determined by setting an error threshold, such as setting the error threshold to 0.01, that is, if the error is less than 0.01, the training process is finished.

Since the parameter configuration rule system is invariant, the first configuration parameter is invariant during each learning of system parameters based on the performance of the target system. The machine learning system is a model for continuous learning, so that the machine learning system can continuously learn the performance and configuration information of the target system after inputting the N first parameter value ranges into the target system, and obtain N second parameter value ranges again.

In a possible implementation manner, the N second parameter value ranges may be obtained by inputting the N first parameter value ranges and the performance of the target system into the machine learning system after the N first parameter value ranges are input into the target system, that is, the N second parameter value ranges and the performance of the target system after the N second parameter value ranges are input into the target system form a group of data that can update the machine learning system.

In this embodiment, the M parameter value ranges, that is, the M parameter value ranges described in S302, are unchanged because the parameter configuration rule system is unchanged, and thus the M parameter value ranges are unchanged. The target value range of the first configuration parameter includes all or part of the value ranges overlapping with the parameter value range corresponding to the first configuration parameter in the M parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N second parameter value ranges, and can be understood as follows: the value range of the first configuration parameter can neither exceed the value range output by the parameter configuration rule system for the first configuration parameter nor exceed the value range output by the machine learning system for the first configuration parameter.

In this embodiment, since the machine learning system performs learning output on the second value ranges of the N parameters, the target value range of each configuration parameter in the N configuration parameters needs to be determined again. The implementation process of re-determining the target value range of each configuration parameter in the N configuration parameters may refer to: and determining the relevant description in the target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, namely step S304. And will not be described in detail herein.

In this embodiment, after the target value range determined again by each of the N configuration parameters is obtained, the target system is configured.

It should be noted that the present embodiment only provides one update process of the configuration parameters of the target system. In actual application, the number of times of updating the system parameters may be specified artificially, for example, 100 times of updating or 200 times of updating may be specified. However, the specific implementation process of updating the system parameters each time is similar to steps S401 to S406, and is not described here again.

And S407, if the performance of the target system exceeds a performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

In this embodiment, the performance of the corresponding target system may be poor due to the fact that the value ranges of the N configuration parameters learned by the machine learning system are not good. Therefore, a performance threshold and a default parameter can be assigned to the performance of the target system, and when the determined target value range of each configuration parameter is input to the target system, if the performance of the target system exceeds the performance threshold, the target system is configured according to the default parameter. It is to be noted that specific meanings of the performance threshold values may be as described in the above embodiments.

In an implementation manner, the performance of the target system corresponding to the value ranges of the N configuration parameters obtained by the machine learning system in each round of learning is compared with a performance threshold, and if the performance of the target system exceeds the performance threshold, the target system is configured according to default parameters.

As an example, assuming that the machine learning system trains 100 rounds, in the 30 th round, the performance of the target system corresponding to the value ranges of the N configuration parameters learned by the machine learning system exceeds the performance threshold, at this time, the machine learning system ends learning, and the target system is configured by using the default configuration parameters.

The configuration method of the system parameters provided by this embodiment enables the machine learning system to continuously update the configuration parameter value range of the target system under the constraint of the parameter configuration rule system, thereby obtaining a more accurate target value range; in addition, when the performance of the target system is larger than the requirement threshold, the target system is configured according to the default configuration parameters, and the robustness of the machine learning system is improved.

Optionally, the machine learning system in any of the above illustrated embodiments may be implemented based on a bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

It should be noted that any of the above embodiments may be implemented alone, or at least two of the above embodiments may be implemented in any combination, which is not limited to this.

Fig. 5 is a schematic structural diagram of a configuration apparatus for system parameters according to an embodiment of the present application. The configuration apparatus of system parameters shown in fig. 5 may be used to perform the configuration method of system parameters described in any of the foregoing embodiments.

As shown in fig. 5, the apparatus 500 for configuring system parameters of the present embodiment includes: the system comprises an acquisition module 501, a first parameter output module 502, a second parameter output module 503, a processing module 504 and a configuration module 505.

The obtaining module 501 is configured to obtain configuration information of a target system.

The first parameter output module 502 is configured to output M parameter value ranges based on the configuration information using a parameter configuration rule system.

The second parameter output module 503 is configured to output N first parameter value ranges based on the configuration information by using a machine learning system.

The processing module 504 is configured to determine a target value range of each of the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, where the target value range of a first configuration parameter in the N configuration parameters includes all or part of the M parameter value ranges that overlap with the parameter value range corresponding to the first configuration parameter in the first parameter value ranges and the parameter value range corresponding to the first configuration parameter in the N first parameter value ranges.

The configuration module 505 is configured to configure the target system according to the target value range of each of the N configuration parameters.

As an example, the obtaining module 501 may be configured to execute the step of obtaining the configuration information of the target system in the method for configuring the system parameter described in any one of fig. 3 to fig. 4. For example, the obtaining module 501 is configured to execute S301.

As another example, the first parameter output module 503 may be configured to perform the step of outputting M parameter value ranges in the method for configuring a system parameter described in any one of fig. 3 to fig. 4. For example, the first parameter output module 503 is configured to execute S302 or S402.

As yet another example, the configuration module 505 may be configured to perform the step of configuring the target system in the method for configuring the system parameters described in any one of fig. 3 to 4. For example, the configuration module 505 is configured to perform S305 or S405.

In a possible implementation manner, the second parameter output module 503 is further configured to output, by using the machine learning system, N second parameter value ranges based on the performance and configuration information of the target system.

The processing module 504 is further configured to re-determine the target value range of each of the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges, where the target value range of a first configuration parameter of the N configuration parameters includes all or part of the parameter value ranges corresponding to the first configuration parameter of the M parameter value ranges and the parameter value ranges corresponding to the first configuration parameter of the N second parameter value ranges that overlap with each other.

The configuration module 505 is further configured to configure the target system according to the target value range determined again for each of the N configuration parameters.

In one possible implementation, the configuration module 505 is further configured to: and if the performance of the target system exceeds a performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

In a possible implementation manner, the target value range of the second configuration parameter in the N configuration parameters includes a third parameter value range in the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, and the third parameter value range includes all or part of the parameter value ranges corresponding to the second configuration parameter in the M parameter value ranges and the parameter value ranges corresponding to the second configuration parameter in the N first parameter value ranges, which are not overlapped.

In a possible implementation manner, the target value range of any one of the N configuration parameters, except for the first configuration parameter, is all or part of the value range, which is overlapped with the parameter value range corresponding to any one of the M parameter value ranges and the parameter value range corresponding to any one of the N first parameter value ranges.

In one possible implementation, the machine learning system is implemented based on a bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

Fig. 6 is a schematic structural diagram of a configuration device according to another embodiment of the present application. The apparatus shown in fig. 6 may be used to perform the method for configuring system parameters according to any of the foregoing embodiments.

As shown in fig. 6, the apparatus 600 of the present embodiment includes: memory 601, processor 602, communication interface 603, and bus 604. The memory 601, the processor 602, and the communication interface 603 are communicatively connected to each other via a bus 604.

The memory 601 may be a Read Only Memory (ROM), a static memory device, a dynamic memory device, or a Random Access Memory (RAM). The memory 601 may store programs and the processor 602 is configured to perform the steps of the method shown in fig. 5 when the programs stored in the memory 601 are executed by the processor 602.

The processor 602 may adopt a general Central Processing Unit (CPU), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits, for executing related programs to implement the lane inference method or the lane inference model training method according to the embodiment of the present application.

The processor 602 may also be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the method of planning an autonomous vehicle according to an embodiment of the present application may be performed by instructions in the form of hardware integrated logic circuits or software in the processor 602.

The processor 602 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, or discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 601, and the processor 602 reads the information in the memory 601, and performs the functions required to be performed by the units included in the thermometric apparatus according to the application in combination with the hardware thereof, for example, the steps/functions of the embodiments shown in fig. 3 or fig. 4 may be performed.

The communication interface 603 may enable communication between the apparatus 600 and other devices or communication networks using, but not limited to, transceiver means.

Bus 604 may include a pathway to transfer information between various components of apparatus 600 (e.g., memory 601, processor 602, communication interface 603).

It should be understood that the apparatus 600 shown in the embodiment of the present application may be an electronic device, or may also be a chip configured in the electronic device.

It should be understood that the processor in the embodiments of the present application may be a Central Processing Unit (CPU), and the processor may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

It will also be appreciated that the memory in the embodiments of the subject application can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. The non-volatile memory may be a read-only memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an electrically Erasable EPROM (EEPROM), or a flash memory. Volatile memory can be Random Access Memory (RAM), which acts as external cache memory. By way of example, but not limitation, many forms of Random Access Memory (RAM) are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct bus RAM (DR RAM).

The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. The procedures or functions according to the embodiments of the present application are wholly or partially generated when the computer instructions or the computer program are loaded or executed on a computer. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more collections of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a solid state disk.

It should be understood that the term "and/or" herein is merely one type of association relationship that describes an associated object, meaning that three relationships may exist, e.g., a and/or B may mean: a exists alone, A and B exist simultaneously, and B exists alone, wherein A and B can be singular or plural. In addition, the "/" in this document generally indicates that the former and latter associated objects are in an "or" relationship, but may also indicate an "and/or" relationship, which may be understood with particular reference to the former and latter text.

In the present application, "at least one" means one or more, "a plurality" means two or more. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, a-b, a-c, b-c, or a-b-c, wherein a, b, c may be single or multiple.

It should be understood that, in the various embodiments of the present application, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: u disk, removable hard disk, read only memory, random access memory, magnetic or optical disk, etc. for storing program codes.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for configuring system parameters, comprising:

acquiring configuration information of a target system, wherein the configuration information is used for indicating a value range of at least one configuration parameter of the target system;

outputting M parameter value ranges by using a parameter configuration rule system based on the configuration information, wherein the M parameter value ranges correspond to M configuration parameters of the target system one by one, and M is a positive integer;

outputting N first parameter value ranges by using a machine learning system based on the configuration information, wherein the N first parameter value ranges correspond to N configuration parameters of the target system one by one, N is a positive integer greater than or equal to M, and the N configuration parameters comprise the M configuration parameters;

determining a target value range of each configuration parameter in the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, wherein the target value range of the first configuration parameter in the N configuration parameters comprises all or part of the parameter value ranges corresponding to the first configuration parameter in the M parameter value ranges and the parameter value ranges corresponding to the first configuration parameter in the N first parameter value ranges;

and configuring the target system according to the target value range of each configuration parameter in the N configuration parameters.

2. The method of claim 1, wherein the configuring the target system according to the target value range of each of the N configuration parameters comprises:

outputting N second parameter value ranges based on the performance of the target system and the configuration information by using the machine learning system;

re-determining a target value range of each configuration parameter of the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges, where the target value range of the first configuration parameter of the N configuration parameters includes all or part of the parameter value ranges of the M parameter value ranges that are overlapped with the parameter value range corresponding to the first configuration parameter and the parameter value ranges of the N second parameter value ranges that are overlapped with the parameter value range corresponding to the first configuration parameter;

3. The method according to claim 1 or 2, wherein the configuring the target system according to the target value range of each of the N configuration parameters comprises:

and if the performance of the target system exceeds a performance threshold, configuring the target system according to the default value ranges of the N configuration parameters.

4. The method according to any one of claims 1 to 3, wherein the target value range of a second configuration parameter of the N configuration parameters includes a third parameter value range of the N first parameter value ranges corresponding to the second configuration parameter, and the third parameter value range includes all or part of the M parameter value ranges that do not overlap with the parameter value range corresponding to the second configuration parameter.

5. The method according to any one of claims 1 to 3, wherein the target value range of any one of the N configuration parameters, except the first configuration parameter, is all or part of the value ranges of the M parameter value ranges, which overlap with the parameter value range corresponding to the any one configuration parameter, and the parameter value range corresponding to the any one configuration parameter, among the N first parameter value ranges.

6. The method according to any one of claims 1 to 5, wherein the machine learning system is implemented based on a Bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

7. An apparatus for configuring system parameters, comprising:

the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring configuration information of a target system, and the configuration information is used for indicating the value range of at least one configuration parameter of the target system;

a first parameter output module, configured to output M parameter value ranges based on the configuration information using a parameter configuration rule system, where the M parameter value ranges correspond to M configuration parameters of the target system one to one, and M is a positive integer;

a second parameter output module, configured to output, by using a machine learning system, N first parameter value ranges based on the configuration information, where the N first parameter value ranges correspond to N configuration parameters of the target system one to one, N is a positive integer greater than or equal to M, and the N configuration parameters include the M configuration parameters;

a processing module, configured to determine a target value range of each of the N configuration parameters according to the M parameter value ranges and the N first parameter value ranges, where the target value range of a first configuration parameter of the N configuration parameters includes all or part of the M parameter value ranges that overlap with a parameter value range corresponding to the first configuration parameter and a parameter value range corresponding to the first configuration parameter of the N first parameter value ranges;

and the configuration module is used for configuring the target system according to the target value range of each configuration parameter in the N configuration parameters.

8. The apparatus of claim 7, wherein the second parameter output module is further configured to: outputting N second parameter value ranges based on the performance of the target system and the configuration information by using the machine learning system;

the processing module is further configured to: re-determining a target value range of each configuration parameter of the N configuration parameters according to the M parameter value ranges and the N second parameter value ranges, where the target value range of the first configuration parameter of the N configuration parameters includes all or part of the parameter value ranges of the M parameter value ranges that are overlapped with the parameter value range corresponding to the first configuration parameter and the parameter value ranges of the N second parameter value ranges that are overlapped with the parameter value range corresponding to the first configuration parameter;

the configuration module is further to: and configuring the target system according to the target value range determined for each configuration parameter in the N configuration parameters.

9. The apparatus of claim 7 or 8, wherein the configuration module is further configured to:

10. The apparatus according to any one of claims 7 to 9, wherein the target value range of a second configuration parameter of the N configuration parameters includes a third parameter value range of the N first parameter value ranges corresponding to the second configuration parameter, and the third parameter value range includes all or a part of the M parameter value ranges that do not overlap with the parameter value range corresponding to the second configuration parameter.

11. The apparatus according to any one of claims 7 to 9, wherein the target value range of any one of the N configuration parameters, except the first configuration parameter, is all or part of the value ranges of the M parameter value ranges, which overlap with the parameter value range corresponding to the any one configuration parameter, and the parameter value range corresponding to the any one configuration parameter, among the N first parameter value ranges.

12. The apparatus of any one of claims 7 to 11, wherein the machine learning system is implemented based on a bayesian algorithm, a genetic algorithm, or a particle swarm algorithm.

13. An apparatus for configuring system parameters, comprising: a memory and a processor;

the memory is to store program instructions;

the processor is configured to invoke program instructions in the memory to perform the method of any of claims 1 to 6.

14. A chip comprising at least one processor and a communication interface, the communication interface and the at least one processor interconnected by a line, the at least one processor being configured to execute a computer program or instructions to perform the method of any of claims 1 to 6.

15. A computer-readable medium, characterized in that the computer-readable medium stores program code for computer execution, the program code comprising instructions for performing the method of any of claims 1 to 6.