CN114611990A

CN114611990A - Method and device for evaluating contribution rate of element system of network information system

Info

Publication number: CN114611990A
Application number: CN202210311890.2A
Authority: CN
Inventors: 周宇; 龙真真; 周文
Original assignee: Zhongke Jingrui Changsha Technology Co ltd
Current assignee: Zhongke Jingrui Changsha Technology Co ltd
Priority date: 2022-03-28
Filing date: 2022-03-28
Publication date: 2022-06-10

Abstract

The invention discloses a method and a device for evaluating contribution rate of a network information system element system, wherein the method comprises the following steps: extracting key elements of the network information system contribution rate based on the existing data sample, and constructing a system contribution rate model by adopting a system contribution rate evaluation method of reinforcement learning; automatically constructing a reinforcement learning model based on the performance indexes of the network information system elements; constructing a system contribution rate calculation model based on the enhanced training model parameters and the model parameters of the small samples, training, extracting model parameters, calculating the weight of each performance index of the system element, designing an accumulative weight calculation method, calculating the final system element contribution rate by calculating the accumulative weight, and finishing the construction of the system contribution rate calculation model; by adopting the evaluation method, the system contribution rate is evaluated on the degree of influence of each element of the network information system on the overall performance and capacity, and the evaluation is used as an index for measuring the economic benefit of the network information system.

Description

Method and device for evaluating contribution rate of element system of network information system

Technical Field

The invention relates to the field of decision support of a network information system, in particular to a method and a device for evaluating contribution rate of a network information system element system.

Background

According to the research conditions at home and abroad, the existing assessment on the system contribution rate, even the assessment problems in other fields, adopts an analytical calculation method, and the parameters of the model are usually determined by experts based on experience, so that the credibility and the authority are difficult to ensure.

For example, a conventional method for calculating the contribution rate of a relatively authoritative system is shown as follows:

the calculation method is an analytical model, the greatest difficulty of the formula is to calculate the capacity values of the system before and after the addition of the equipment A, and the usually adopted method is also an analytical method, namely, the system capacity is hierarchically decomposed from top to be decomposed into measurable performance indexes, and then comprehensive weighting is carried out from bottom to top, but the weighting weight cannot be determined, only an expert evaluation method can be used at present, and the method is difficult to obtain an authoritative and credible evaluation value.

The role played and the multiplication effect of each network information system element in the generating process are different, so how to evaluate the contribution rate of the network information system element in the system is a problem to be solved at present. As long as the data samples are authentic, the evaluation conclusion is drawn to be equally trustworthy.

Disclosure of Invention

In order to solve the problems that model parameters are difficult to determine and an evaluation result is not authoritative in the evaluation based on an analytic model in the prior art, the invention provides a system contribution rate evaluation method based on reinforcement learning, which comprises the following steps:

extracting key elements of the network information system contribution rate based on a preset data sample, and constructing a system contribution rate model by adopting a system contribution rate evaluation method of reinforcement learning;

automatically constructing a reinforcement learning model based on the performance indexes of the network information system elements;

the method comprises the steps of building a system contribution rate calculation model based on small sample enhanced training model parameters and model parameters, training, extracting model parameters, calculating the weight of each performance index of system elements, designing an accumulative weight calculation method, calculating the final system element system contribution rate by calculating the accumulative weight, completing building of the system contribution rate calculation model, and completing evaluation of the network information system contribution rate through a teacher model and a student model.

Preferably, the key elements of the contribution rate of the network information system include:

training, logistics, personnel, information, concepts and ordinances, equipment, infrastructure, organizations;

wherein the key elements have interoperability between: the system capability is finally generated by the capability of providing services among the key elements, receiving services from other models and using the services to realize mutual effective cooperative work.

Preferably, the system contribution rate evaluation method for reinforcement learning includes:

building a reinforcement learning model based on the contribution rate evaluation, wherein the learning model is built for modeling environment, action, state and reward;

training system elements based on a reinforcement learning model in a simulation environment: placing an intelligent agent in a simulation environment to carry out reinforcement learning training in a systematic confrontation mode;

constructing a system contribution rate calculation model based on the extracted reinforcement learning model parameters: the system contribution rate evaluation capability learned by the reinforcement learning model is converted into output, and a system contribution rate calculation model is established in a mode of extracting model parameters and is used for evaluating the system contribution rate;

the intelligent agent refers to an individual which can generate different responses to different stimuli according to a certain rule and is used for abstracting the template.

Preferably, the automatic construction of the reinforcement learning model by the performance indexes of the network information system elements includes:

constructing based on a network information system knowledge graph and combining a knowledge-assisted reinforcement learning model;

extracting knowledge related to the knowledge map based on a network information system: wherein the knowledge graph consists of data from different sources and with different structures; the different structures include: the body meta-model is a person body, and comprises age and organization attributes, and the persons can be divided into information persons, commanders, support persons and various contextual information of roles, targets, tasks and knowledge requirements of users;

constructing an auxiliary reinforcement learning model based on the extracted knowledge; the auxiliary reinforcement learning comprises introducing prior knowledge in a knowledge graph, and the intelligent agent is enabled to explore an unknown state space according to a condition learning strategy of the current state, so that a better strategy is ensured to be learned.

Preferably, the method for constructing the system contribution rate calculation model based on the enhanced training model parameters of the small samples and the model parameters includes:

analyzing a change mode of characteristic migration between system element entities based on different scene tasks;

based on a reinforcement learning model self-adaptive migration mechanism under a multi-scenario task;

preferably, the reinforcement learning model adaptive migration mechanism based on the multi-scenario task includes:

aiming at different tasks, combining a neuron response mode, and analyzing the weight characteristics of an original teacher model; then, an original frame teacher model is learned through a student model with a simple structure, the original weight information is subjected to compression and elongation operations, efficient training of the student model is achieved, and finally the goal of strengthening the learning model based on the contribution rate of a small sample rapid training system is achieved.

Preferably, the teacher model refers to existing knowledge, and the student model refers to new knowledge; when a new problem occurs, training is not required to be restarted, adjustment is performed on the existing model, new data is used for training, and training can be completed by using a small amount of samples.

Based on the same inventive concept, the invention also provides a system for evaluating the contribution rate of the element system of the network information system, which comprises the following steps:

the system contribution rate model construction module is used for constructing a reinforcement learning model for contribution rate evaluation; training a reinforcement learning model in a simulation environment; extracting parameters of a reinforcement learning model to construct a system contribution rate calculation model;

the system element automatic construction module is used for constructing a network information system knowledge graph, extracting the related knowledge of the network information system knowledge graph and completing the construction of a knowledge auxiliary reinforcement learning model;

the small sample data rapid training module is used for analyzing the change pattern of characteristic migration between system element entities under different scene tasks, realizing a reinforcement learning model self-adaptive migration mechanism under multi-scene tasks, and extracting the reinforcement learning model parameters and calculating the system contribution rate based on the migration strategy and oriented to the small sample data;

and the debugging and verifying module is used for performing statistical analysis on the network characteristics of a complex network-based combat system structure based on a combat system topology model formed by a network dynamic organization.

According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor. The memory stores instructions executable by the at least one processor to enable the at least one processor to perform a method according to an embodiment of the disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions for causing the computer to perform a method according to an embodiment of the present disclosure.

Compared with the prior art, the invention has the beneficial effects that:

1. from the theoretical perspective of system engineering, the degree of influence of the system contribution rate on the overall performance and capacity of each element of a network information system is evaluated;

2. an intelligent agent is trained by utilizing reinforcement learning, so that the intelligent agent has the capability of evaluating the contribution rate of a network information system element system, the evaluation result is more authoritative than that of the conventional evaluation method of the artificial design evaluation standard, and meanwhile, a model parameter can be extracted to construct a system contribution rate calculation model;

3. combining the knowledge graph with reinforcement learning, and assisting automatic construction of a reinforcement learning model by using information such as equipment, personnel, performance indexes and the like provided by the constructed network information system knowledge graph;

4. based on transfer learning, the existing prior is introduced into the current task learning, so that the cross-domain small sample-oriented fast learning is realized;

5. the evaluation of the contribution rate of the element system of the network information system can promote the process of modern war systematization, but the problems of difficult determination of model parameters, uncertain evaluation results and the like exist in the current evaluation process;

6. the scheme is based on reinforcement learning, so that an intelligent agent has the capability of evaluating the system contribution rate through training in a simulation environment, and model parameters can be extracted to construct a system contribution rate calculation model; establishing a knowledge graph for network information system elements, and designing a set of method capable of automatically constructing a reinforcement learning model;

7. the transfer learning can enable the model to learn the model parameters only through a small data sample.

Drawings

FIG. 1 is a sequence diagram illustrating steps of a method for evaluating information system contribution rate according to the present invention;

FIG. 2 is a schematic diagram of the overall content structure of the present invention;

FIG. 3 is a schematic diagram of key elements of the present invention;

FIG. 4 is a schematic diagram of the deep learning-based system contribution rate evaluation process of the present invention;

FIG. 5 is a schematic diagram of a construction route of a knowledge-graph-aided reinforcement learning model according to the present invention;

FIG. 6 is a schematic diagram of a network information system knowledge graph construction relationship of the present invention;

FIG. 7 is a schematic view of a service model and visualization of a knowledge-graph-based network information system of the present invention;

FIG. 8 is a schematic diagram of a small sample data migration strategy-based parameter and system contribution rate reinforcement learning model of the present invention;

FIG. 9 is a schematic diagram of model adaptive migration according to the present invention;

FIG. 10 is a schematic structural diagram of a computer device according to a third embodiment of the present invention

Detailed Description

For a better understanding of the present invention, reference is made to the following description taken in conjunction with the accompanying drawings and examples.

Example one

The invention provides a method for evaluating contribution rate of a network information system element system, as shown in figure 1, which mainly comprises the following steps:

s1, extracting key elements of the network information system contribution rate based on the existing data sample, and constructing a system contribution rate model by adopting a system contribution rate evaluation method of reinforcement learning;

s2, automatically constructing a reinforcement learning model based on the performance indexes of the network information system elements;

s3, constructing a system contribution rate calculation model based on the enhanced training model parameters of the small samples and the model parameters, training, extracting model parameters, calculating the weight of each performance index of the system elements, designing an accumulative weight calculation method, calculating the final system element system contribution rate by calculating the accumulative weight, and completing construction of the system contribution rate calculation model based on the above basis.

The overall content structure of each step is shown in fig. 2: the method mainly comprises the following steps: the system contribution rate evaluation stage and method based on reinforcement learning, the automatic construction of a reinforcement learning model based on the performance indexes of network information system elements, and the reinforcement learning model based on the parameters of the migration strategy and the system contribution rate thereof facing small sample data.

1. The concept connotation of the contribution rate of the network information system is as follows:

the interoperability and coordinated development of the elements are the key to the capability of generating the system, as shown in fig. 3:

2. system contribution rate evaluation stage and method based on reinforcement learning

Human subjective experience causes the evaluation result of the contribution rate of the element system of the network information system to be not authoritative, and the intelligent agent can make far more times of attempts than can be made in the real world through the self-learning process of deep reinforcement learning in the simulation environment, so that more reasonable answers are provided when new problems are confronted. The experience and knowledge of the intelligent agent are stored in the parameters of the deep reinforcement learning model, so that the required information can be extracted from the parameters to construct a system contribution rate calculation model for guiding the decision making of the intelligent agent. As shown in fig. 4:

(1) construction of reinforcement learning model for contribution rate evaluation

Reinforcement learning models build on modeling environment, actions, states, and rewards. Training the intelligent agent to maximize the efficiency (reward) under the condition of the same resources can enable the intelligent agent to master the capability of configuring network information system elements and also imply the capability of evaluating the contribution rate of the system.

(2) Reinforcement learning model training process in simulation environment

And placing the intelligent agent in a simulation environment to carry out reinforcement learning training in a systematic confrontation mode. The contribution rate of each network information system element in the training process is finally embodied in the efficiency of the intelligent agent under a specific environment. Meanwhile, human priori knowledge in the knowledge graph is introduced to accelerate training speed and enable the intelligent agent to obtain good network information system element configuration capacity.

(3) Extraction reinforcement learning model parameter construction system contribution rate calculation model

A trained agent can give the appropriate configuration of network information architecture elements in a given environment, but cannot explicitly give the architecture contribution rate. This is because the system contribution rate evaluation ability learned by the reinforcement learning model is implicit in the model parameters, and the model can convert the ability into output but cannot tell us about its evaluation process. Therefore, a system contribution rate calculation model can be established in a mode of extracting model parameters, so that the system contribution rate is intuitively given for decision making.

3. Automatic construction of reinforcement learning model based on network information system element performance indexes

Aiming at the problems that model parameters are difficult to determine, an evaluation result is not authoritative and the like in the evaluation of the contribution rate of an element system of the existing network information system, a unified framework is established for representing and storing behaviors, personnel, events, cases and internal relations thereof, and a knowledge graph is used as the most effective expression of things and relations. The reward function is designed using a priori knowledge provided in the knowledge-graph. This approach, also known as heuristic reward function design. After introducing the designed reward function, the reward earned by the agent becomes the sum of the instant reward value and the additional reward value of the environment.

The knowledge graph is used as a means for assisting reinforcement learning, and a matching result is returned in an intelligent and efficient mode by combining the use situation, so that quick and accurate knowledge information is provided for a user. The following three aspects are realized to automatically construct a model for reinforcement learning based on performance indexes of network information system elements, as shown in fig. 5:

(1) construction of knowledge graph based on network information system

The concept of Knowledge Graph (knowledgegraph) was first a new concept proposed by Google. A knowledge graph is essentially a knowledge base of Semantic networks (Semantic networks). We can also simply understand the knowledge Graph as a Multi-relationship Graph (Multi-relationship Graph), which refers to the Graph with multiple node types and edges between multiple nodes, and can also be called Heterogeneous Network (Heterogeneous Network). Nodes in a knowledge graph are generally referred to as entities (entities), such as certain events, people, etc., and edges are referred to as relationships (relationships), representing the relationship between two connected entities. Knowledge maps provide the ability to analyze problems from a relational perspective.

The network information system element system knowledge graph mainly aims at supporting various intelligent services, constructing an association graph among various elements and covering various military activities, personnel, equipment, performance indexes and the like. The knowledge graph of the network information system needs to be based on the construction of the current business database, extract an ontology and a relation between the ontology from the existing structured, semi-structured and unstructured databases by ontology mapping and using a machine learning method, and further extract metadata of the ontology knowledge base to form a voxel model, as shown in fig. 6. And the service data, the ontology knowledge and the ontology meta-model are sequentially promoted according to an abstract level. In the business database, the data is mostly entity relationship type, such as Zhang III of intelligence personnel, age 23 years old, army is army. After the ontology is mapped, the ontology knowledge is the ontology of the information personnel, including attributes such as age and military species. After the metadata is extracted, the body meta-model is a person body, and the person body comprises attributes such as age, organization and the like, and can be divided into information personnel, commanders, support personnel and the like.

The network information system service model based on the knowledge graph is shown in fig. 4 (left), and aims to solve user requirements, obtain various aspects of context information such as user roles, targets, tasks, knowledge requirements and the like by performing semantic analysis and knowledge context extraction based on a service database, a body knowledge base, a body meta-model base and the like, call a knowledge graph generation service, form service entities, bodies and relations thereof related to service requirement organization to construct the knowledge graph, perform knowledge retrieval, reasoning and service planning based on the graph, and push results of the user requirements. Meanwhile, abstracting and evaluating the entity relationship which is frequently called, and adding, deleting or modifying the relationship of the knowledge ontology base according to the evaluation result to update the knowledge ontology. According to the design rule of the knowledge graph, the ontology/entity and the relationship are simultaneously stored in the database, and various service calls and reasoning analysis are performed, as shown in fig. 7 (right).

(2) Extraction of knowledge related to network information system knowledge graph

The knowledge graph is composed of data of different sources and different structures, wherein a lot of knowledge is contained, different knowledge can play different roles in model training, and in order to use the auxiliary knowledge more efficiently, relevant knowledge in the knowledge graph of the network information system needs to be extracted, so that necessary auxiliary information is provided for reinforcement learning. Firstly, determining related entities, then extracting the first-order, second-order and other h-order entities of the entities as seed sets, and capturing local information and global information of the entities. The network information system knowledge graph can provide attribute information of a certain device and related information of other related devices or persons, can express the attribute and related relation of the entity from the perspective of local and global, and can construct a small network information system knowledge graph by the extracted related entity and related information, thereby providing strong knowledge support for reinforcement learning.

Each entity in the knowledge Graph represents equipment, a person or a certain performance index, the knowledge Graph can be regarded as a Graph Network (Graph Network), the Graph Network has strong relationship induction bias, a direct interface is provided for manipulating structured knowledge and generating structured behaviors, then the Graph Network is expanded on the operation of a neural Network, a reinforcement learning model is built, and the reinforcement learning model automatically built based on the performance indexes of Network system elements is completed.

(3) Construction of knowledge-aided reinforcement learning model

The reinforcement learning currently faces the dilemma of dimension disaster of state-action space, sparse reward, delay, contradiction between exploration and utilization under limited resources, and the like, and the introduction of human knowledge into the reinforcement learning is helpful for solving the problems. The mode of introducing human knowledge into reinforcement learning is various, and a feasible scheme is realized by using a knowledge graph to assist in designing a reward function and a heuristic exploration strategy.

In the training process of the simulation environment, the reward signals may have sparseness and delay, namely, the intelligent agent can obtain a valuable reward signal after executing a series of actions, which is also called as the credibility allocation problem of reinforcement learning. The sparsity of the reward signal not only results in slow convergence of reinforcement learning, but also may result in difficulty in learning the optimal strategy. Thus, the reward function may be designed using a priori knowledge provided in the knowledge-graph. This approach is also referred to as heuristic reward function design. After introducing the designed reward function, the reward earned by the agent becomes the sum of the instant reward value and the additional reward value of the environment.

Because the simulation environment is too complex, the agent may not learn the optimal strategy with sufficient exploration of all states. In order to overcome the contradiction between exploration and utilization under limited resources, the prior knowledge in the knowledge graph can be introduced, so that an intelligent agent learns the strategy according to the condition of the current state on one hand, and an unknown state space is explored on the other hand, and the possibility of learning a better strategy is kept.

4. Small sample-based enhanced training model parameter and model parameter-based system contribution rate calculation model constructed

In order to enable the model to have the rapid learning capability under a small sample, the scheme introduces the knowledge learned on other scene tasks into the reinforcement learning model training on the small sample data by using the migration strategy, and the problem of training the reinforcement learning model facing the small sample data based on the migration strategy under the multi-scene task needs to be solved. The part mainly analyzes from three aspects, namely migration of the same entity characteristic under different scene tasks, migration of knowledge among models under different scene tasks, reinforcement learning model training and construction of a system contribution rate calculation model based on model parameters. As shown in fig. 8.

(1) Variation pattern analysis of feature migration between system element entities under different scene tasks

Under the condition of a small sample, the model is easy to be over-fitted on the same system element data, and the system contribution rate evaluation under a new scene task is difficult to process. Therefore, in order to construct reasonable feature migration, the following ideas are adopted to describe the feature changes of system element entities under different scene tasks:

firstly, consider that the characterization of the same element in different task scenarios relies mainly on the same intrinsic features. Performing feature analysis on the small sample data set aiming at the same system element, and measuring invariant features in the network learning process;

and then, the obtained invariant features are migrated to a data sample under another scene task for use, and the problem of poor entity characterization generalization capability of a small sample under different scene tasks is finally solved by combining a feature expression model.

(2) Self-adaptive migration mechanism of reinforcement learning model under multi-scenario task

The structural design of the deep reinforcement learning model is very complex and skillful. The design of the model network structure mostly depends on the requirements required by the initial task, and the network framework is too heavy or too simple for the new scene task. However, designing different network architectures for each data not only consumes time and labor, but also wastes a lot of computing resources. The ability to extend the initial network learning "knowledge" can be considered and overcome the burdensome nature of redesigning the network. Firstly, aiming at different tasks, combining a neuron response mode and analyzing the weight characteristics of an original teacher model. Then, an original frame teacher model is learned through a student model with a simple structure, and original weight information is subjected to compression and elongation operations, as shown in fig. 9, so that training of the student model is realized.

Combining the step (1), finally achieving a small sample-based contribution rate reinforcement learning model of a rapid training system, wherein the teacher model refers to existing knowledge, and the student model refers to new knowledge; when a new problem is met, the training does not need to be restarted, the adjustment is carried out on the existing model, the new data is used for training, and the training is completed by using a small amount of samples.

(3) The method comprises the steps of (1) constructing a small sample-based system contribution rate reinforcement learning model by combining (1) and (2), training and extracting model parameters, calculating the weight of each performance index of system elements, designing an accumulative weight calculation method, calculating the final system element system contribution rate by calculating the accumulative weight, and completing the construction of the system contribution rate calculation model based on the basis.

Example two

Based on the same inventive concept, the invention also provides a system for evaluating the contribution rate of the network information system element system, which comprises the following steps:

the system element automatic construction module is used for constructing a knowledge graph of a network information system, extracting relevant knowledge of the knowledge graph of the network information system and completing construction of a knowledge auxiliary reinforcement learning model;

and the debugging and verifying module is used for performing statistical analysis on the network characteristics of a complex network-based combat system structure based on a combat system topological model formed by network dynamic organization.

EXAMPLE III

Fig. 10 is a schematic structural diagram of an electronic device (or computer device) according to a third embodiment of the present invention. The computer device shown in fig. 10 is only an example, and should not bring any limitation to the function and the scope of use of the embodiments of the present invention.

As shown in fig. 10, the computer device is in the form of a general purpose computing device. The components of computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.

Bus 18 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. By way of example, such architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, micro-channel architecture (MAC) bus, enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

Computer device 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by computer device 12 and includes both volatile and nonvolatile media, removable and non-removable media.

The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)30 and/or cache memory 32. Computer device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 10, and commonly referred to as a "hard drive"). Although not shown in FIG. 10, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk, and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk such as a CD-ROM, DVD-ROM, or other optical media may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.

A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of embodiments of the invention as described.

Computer device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), with one or more devices that enable a user to interact with computer device 12, and/or with any devices (e.g., network card, modem, etc.) that enable computer device 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface 22. Also, computer device 12 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the Internet) via network adapter 20. As shown, network adapter 20 communicates with the other modules of computer device 12 via bus 18. It should be appreciated that although not shown in FIG. 10, other hardware and/or software modules may be used in conjunction with computer device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.

The processing unit 16 executes various functional applications and data processing by running the program stored in the system memory 28, for example, to implement a method for evaluating contribution rate of network information system element system provided by the embodiment of the present invention.

Example four

A fourth embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program implements each process of the above-mentioned method for detecting an obstacle, and can achieve the same technical effect, and in order to avoid repetition, details are not repeated here.

A computer readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. A computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The foregoing is considered as illustrative of the preferred embodiments of the invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for evaluating contribution rate of a network information system element system is characterized by comprising the following steps:

2. The method according to claim 1, wherein the key elements of the network information system contribution rate comprise:

the key elements have interoperability among the following key elements: the system capability is finally generated by the capability of providing services among the key elements, receiving services from other models and using the services to realize mutual effective cooperative work.

3. The method for evaluating the contribution rate of the network information system element system according to claim 1, wherein the method for evaluating the contribution rate of the system based on reinforcement learning comprises:

training system elements based on a reinforcement learning model in a simulation environment: the method comprises the steps that an intelligent agent is placed in a simulation environment to carry out reinforcement learning training in a systematic confrontation mode;

4. The method for evaluating the contribution rate of the network information system element system according to claim 1, wherein the automatic construction of the reinforcement learning model by the performance index of the network information system element comprises:

constructing based on a knowledge graph of a network information system and in combination with a knowledge-assisted reinforcement learning model;

extracting knowledge related to the knowledge graph based on a network information system: wherein the knowledge graph consists of data of different sources and different structures; the different structures include: the body meta-model is a person body, and comprises age and organization attributes, and the persons can be divided into information persons, commanders, support persons and various contextual information of roles, targets, tasks and knowledge requirements of users;

5. The method according to claim 1, wherein the method for evaluating the contribution rate of the network information system element system comprises the steps of:

analyzing the change pattern of characteristic migration between system element entities based on different scene tasks;

based on a reinforcement learning model self-adaptive migration mechanism under a multi-scenario task.

6. The method according to claim 5, wherein the reinforcement learning model adaptive migration mechanism based on multi-scenario task comprises:

aiming at different tasks, combining a neuron response mode, and analyzing the weight characteristics of an original teacher model; then, an original frame teacher model is learned through a student model, original weight information is subjected to compression and elongation operations, student model training is achieved, and finally the goal of strengthening the learning model based on the contribution rate of a small sample training system is achieved.

7. The method for evaluating contribution rate of network information system element system according to claim 6, wherein the teacher model refers to existing knowledge, and the student model refers to new knowledge; when a new problem occurs, training is not required to be restarted, adjustment is performed on the existing model, new data is used for training, and training can be completed by using a small amount of samples.

8. An apparatus for evaluating a contribution rate of a network information system element system, comprising:

the small sample data rapid training module is used for analyzing the change pattern of characteristic migration between system element entities under different scene tasks, performing a reinforcement learning model self-adaptive migration mechanism under multi-scene tasks, and performing a reinforcement learning model parameter extraction and system contribution rate calculation model based on a migration strategy and oriented to small sample data;

and the debugging and verifying module is used for constructing a combat system topological model formed by network dynamic organization and carrying out statistical analysis on network characteristics of a complex network-based combat system structure.

9. An electronic device, comprising:

at least one processor; and a memory communicatively coupled to the at least one processor, wherein:

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of network information system element hierarchy contribution ratio assessment of any of claims 1-7.

10. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the network information hierarchy element hierarchy contribution rate evaluation method of any one of claims 1 to 7.