Metaintelligent model in intelligent information network and implementation method

Abstract

The invention provides a meta-intelligent model in an intelligent information network and an implementation method, and belongs to the technical field of data processing. The meta-intelligent model comprises a multi-domain sensing module, a learning analysis module, a network knowledge module and an optimization decision module; the multi-domain sensing module senses a multi-domain environment inside and outside the network to obtain multi-domain characteristic parameters, and pre-processes the multi-domain characteristic parameters; the learning analysis module performs data analysis and fusion on the preprocessed multi-domain feature parameters by adopting a machine learning mode, so that a trained multi-domain environment prediction learning model is obtained, and the multi-domain feature information and the trained multi-domain environment prediction learning model are stored in the network knowledge module; the network knowledge module stores network multi-dimensional polymorphic data information with the same representation form, including multi-domain situation information and a trained multi-domain environment prediction learning model; and the optimization decision-making module is used for matching a comprehensive decision-making scheme matched with the current network multi-domain environment from the network knowledge module.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a meta-intelligence model in an intelligent information network and an implementation method thereof. Background Art

[0002] Intelligent information network is one of the key technologies to realize network intelligence. Its main purpose is to build a network that meets the requirements of voice, data, and video integration, the requirements of various business applications, and the requirements of reliability, security, efficiency, easy expansion, and easy management. The biggest feature of meta-intelligence is to improve the network's autonomous cognitive ability, that is, to improve the network's data perception and analysis and processing capabilities. At present, the industry has also conducted research on the use of cognitive methods to build intelligent information networks, and this has gradually become the mainstream direction of research, such as cognitive wireless networks and cognitive networks with cognitive methods as the core, and networks that partially use cognitive methods to improve the level of intelligence. Summary of the invention

[0003] The purpose of the present invention is to provide a meta-intelligence model implementation scheme in an intelligent information network, so that the network has self-optimization, self-evolution and self-learning functions.

[0004] The first aspect of the present invention proposes a method for realizing a meta-intelligence model in an intelligent information network, wherein the meta-intelligence model constructed by the method includes a multi-domain perception module, a learning analysis module, a network knowledge module and an optimization decision module; wherein: the multi-domain perception module is called to perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters and pre-process the multi-domain feature parameters; the learning analysis module is called to perform data analysis and fusion on the pre-processed multi-domain feature parameters in a machine learning manner, thereby training a multi-domain environment prediction learning model, and storing the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; the network knowledge module is called to store The network multi-dimensional and multi-state data information includes the multi-domain situation information and the trained multi-domain environment prediction learning model; the optimization decision module is called to match the comprehensive decision-making scheme matching the current network multi-domain environment from the network knowledge module; wherein, when the comprehensive decision-making scheme matching the current network multi-domain environment does not exist in the network knowledge module, the current network multi-domain environment is subjected to learning and analysis, the trained multi-domain environment prediction learning model in the network knowledge module is called, the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making scheme.

[0005] According to the method of the first aspect of the present invention, the communication nodes of the network are deployed in a hierarchical distributed manner, the nodes are divided into X levels, the total number is Y, and the multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where:

[0006]

[0007] Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

[0008]

[0009] According to the method of the first aspect of the present invention, the method for performing data analysis and fusion on the preprocessed multi-domain feature parameters by means of machine learning, thereby the trained multi-domain environment prediction learning model specifically includes:

[0010] Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

[0011]

[0012] According to the method of the first aspect of the present invention, calling the trained multi-domain environment prediction learning model Φ in the network knowledge modulep , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1:

[0013]

[0014] Using the multi-domain situation prediction information Generate comprehensive decision solutions:

[0015]

[0016] Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node of the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method.

[0017] The second aspect of the present invention proposes a meta-intelligence model in an intelligent information network. The meta-intelligence model includes a multi-domain perception module, a learning analysis module, a network knowledge module, and an optimization decision module; wherein: the multi-domain perception module is configured to: perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters, and pre-process the multi-domain feature parameters; the learning analysis module is configured to: use machine learning to perform data analysis and fusion on the pre-processed multi-domain feature parameters, so as to train a multi-domain environment prediction learning model, and store the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; the network knowledge module is configured to: store network multi-dimensional and multi-state data information with the same representation form, including the multi-domain state The optimization decision module is configured to: match a comprehensive decision-making scheme matching the current network multi-domain environment from the network knowledge module; wherein, when the comprehensive decision-making scheme matching the current network multi-domain environment does not exist in the network knowledge module, the current network multi-domain environment is subjected to learning and analysis, the trained multi-domain environment prediction learning model in the network knowledge module is called, the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making scheme.

[0018] According to the system of the second aspect of the present invention, the communication nodes of the network are deployed in a hierarchical distributed manner, the nodes are divided into X levels, the total number is Y, and the multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where:

[0019]

[0020] Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

[0021]

[0022] According to the second aspect of the present invention, the system is used for performing data analysis and fusion on the preprocessed multi-domain feature parameters in a machine learning manner, so as to train the multi-domain environment prediction learning model, specifically comprising:

[0023] Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

[0024]

[0025] According to the system of the second aspect of the present invention, the trained multi-domain environment prediction learning model Φ in the network knowledge module is called p , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1:

[0026]

[0027] Using the multi-domain situation prediction information Generate comprehensive decision solutions:

[0028]

[0029] Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node of the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method.

[0030] The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method for implementing a meta-intelligence model in an intelligent information network described in the first aspect of the present disclosure is implemented.

[0031] The fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for realizing a meta-intelligence model in an intelligent information network described in the first aspect of the present disclosure is implemented.

[0032] In summary, compared with the prior art, the technical solution proposed in the present invention has the following significant advantages: (1) It has comprehensive multi-domain environment perception function. The multi-domain perception module in the model can fully perceive the internal and external environment of the network, such as the electromagnetic environment, the network status, and the user behavior, and obtain the multi-dimensional characteristic parameters of the network, providing a solid foundation for the network to adapt to environmental changes; (2) It has efficient learning and analysis function. The learning and analysis module uses a flexible learning model training method to select a suitable trained learning model through environmental matching to achieve efficient learning and analysis functions and provide technical support for the network adaptive environment; (3) It has flexible data information storage function. The network knowledge module can store various types of network data information. Each module can call the existing data information at any time to realize its own function and provide knowledge reserve for the network adaptive environment; (4) It has complete optimization decision-making function. Combining real-time and historical parameter data, each node in the network selects the best optimization method to generate a decision plan. The decision plans between the nodes can interact through broadcasting and other methods, and finally integrate a comprehensive decision plan to provide intelligent decision-making for the network adaptive environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 Schematic diagram of a model according to an embodiment of the present invention.

[0035] Figure 2 The figure is a structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0037] The first aspect of the present invention provides a method for implementing a meta-intelligence model in an intelligent information network, such as Figure 1 As shown, the meta-intelligence model constructed by the method includes a multi-domain perception module, a learning analysis module, a network knowledge module and an optimization decision module; wherein: the multi-domain perception module is called to perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters and pre-process the multi-domain feature parameters; the learning analysis module is called to use machine learning to perform data analysis and fusion on the pre-processed multi-domain feature parameters, thereby training a multi-domain environment prediction learning model, and storing the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; the network knowledge module is called to store network multi-dimensional and multi-state data information having the same representation form, including The multi-domain situation information and the trained multi-domain environment prediction learning model; calling the optimization decision module to match a comprehensive decision-making solution that matches the current network multi-domain environment from the network knowledge module; wherein, when the comprehensive decision-making solution that matches the current network multi-domain environment does not exist in the network knowledge module, the current network multi-domain environment is subjected to learning and analysis, and the trained multi-domain environment prediction learning model in the network knowledge module is called, and the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making solution.

[0038] In some embodiments, the communication nodes of the network are deployed in a hierarchical distributed manner, the nodes are divided into X levels, the total number is Y, and the multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where:

[0039]

[0040] Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

[0041]

[0042] In some embodiments, the method for performing data analysis and fusion on the preprocessed multi-domain feature parameters by means of machine learning, thereby obtaining the trained multi-domain environment prediction learning model, specifically includes:

[0043] Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

[0044]

[0045] In some embodiments, the trained multi-domain environment prediction learning model Φ in the network knowledge module is calledp , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1:

[0046]

[0047] Using the multi-domain situation prediction information Generate comprehensive decision solutions:

[0048]

[0049] Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node of the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method. Specific embodiments

[0051] The method comprises the following steps: Step 1), multi-domain perception: responsible for obtaining characteristic information such as channel characteristics, network topology, user behavior, etc. in multiple domains such as external environment domain, network status domain, user behavior domain, etc. Step 2), learning analysis: responsible for learning the characteristic parameters obtained by multi-domain perception to form a mapping relationship between multi-domain characteristic parameters. Step 3), network knowledge: complete the unified representation, storage, sharing and other functions of network knowledge, and support other modules to realize corresponding functions. Step 4), optimized decision: use the characteristic information, network knowledge and learning analysis methods of multi-domain perception to realize comprehensive decision-making, so that the network can adapt to internal and external changes.

[0052] 1. Multi-domain perception

[0053] Multi-domain environment perception refers to the process of fully perceiving the multi-domain environment inside and outside the network, acquiring multi-domain characteristic parameter data and performing preliminary processing.

[0054] The multi-domain environment refers to the internal and external environment of the network, including the external environment domain, network status domain and user behavior domain. The electromagnetic environment domain includes electromagnetic spectrum description information such as electromagnetic wave space domain, time domain, frequency domain, energy density, etc. The network status domain includes network topology, connectivity, available resources, protocol changes, service proportion, transmission rate, network management status and other network status description information. The user behavior domain includes user type, service preference, user distribution, priority, access mode, channel category, service volume distribution, service type, access process and other user behavior feature description information. The specific implementation plan is as follows:

[0055] The intelligent network communication nodes are deployed in a hierarchical distributed manner. The nodes are divided into X levels and the total number is Y. Then the multi-domain perception module of the i-th communication node at the n-th level (n∈X, i∈Y) obtains the multi-domain feature parameters of the external environment domain, network status domain, and user behavior domain at time k.

[0056]

[0057] In the formula, Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay External environment domain characteristic parameters, Network topology Network congestion level Packet loss rate etc. network state domain characteristic parameters, User business type User Priority User Location The user behavior domain characteristic parameters are expressed as follows:

[0058]

[0059] 2. Learning Analytics

[0060] Learning analysis refers to the process of acquiring multi-domain situation information by using learning methods such as machine learning to analyze and fuse feature parameters obtained from multi-domain perception. Multi-domain situation information is stored in the network knowledge module and transmitted to the optimization decision module.

[0061] Depending on whether it is real-time learning, learning analysis is divided into online learning and offline learning. Online learning refers to the process of training a learning model once each time a set of multi-domain feature parameters is obtained through multi-domain perception technology, and the model is continuously updated iteratively based on real-time data. Its purpose is to minimize the error of the entire learning analysis process, and the learning model can adapt to environmental changes in a timely manner, but abnormal data has a greater impact on the model. Offline learning refers to the use of a large number of multi-domain feature parameters obtained through multiple perceptions to uniformly train the learning model. Its purpose is to obtain a learning model that minimizes the true error based on a large number of feature parameter samples. The learning effect is better but takes longer. According to different requirements such as real-time performance and learning effect, the appropriate learning method can be flexibly selected. For example, when the real-time requirements are high, online learning can be selected, and when the prediction accuracy requirements are high, offline learning can be selected.

[0062] Untrained learning model from network knowledge Select a prediction learning model that matches the current multi-domain environment The multi-domain feature parameters from the starting time to the k time are used for training.

[0063]

[0064] 3. Network knowledge

[0065] Network knowledge refers to multi-dimensional and multi-state data information of the network with a unified representation form, which can not only be directly generated by network cognition, but also can generate new knowledge autonomously through learning and other methods. Network knowledge can specifically include node interaction information, expert experience information, basic network parameters, multi-domain environmental characteristic parameters, multi-domain situation information, learning models, decision-making plans and related logical correspondences.

[0066] Unlike the construction of traditional knowledge bases, network knowledge is closely related to the internal and external environment of the network. The volume and internal coupling degree of network knowledge are proportional to factors such as network operation time and environmental complexity. In order to ensure network operation efficiency and resource utilization, network knowledge construction and application must have the following functions:

[0067] 1) It can realize the function of rapid update of massive knowledge. When the internal and external environment of the network changes rapidly and dynamically, the online knowledge acquired through perception needs to be continuously updated and iterated. When the internal and external environment of the network becomes increasingly complex, the volume of network knowledge may surge. In order to ensure that network knowledge is real-time and effective, it is necessary to have the function of rapid update of massive knowledge.

[0068] 2) It can realize the autonomous generation of offline network knowledge. In the offline state, it is necessary to have the function of generating new knowledge through learning existing knowledge, such as predicting multi-domain situation information or decision-making plans based on historical multi-domain feature parameters, etc., to reduce the pressure of real-time perception and learning of the network.

[0069] 3) It can realize the autonomous representation function of different types of network knowledge. The amount of network knowledge acquired and generated autonomously in real time is huge, and the degree of coupling between each type of knowledge is high. In order to facilitate storage, calling and processing, a unified autonomous representation function is required.

[0070] Multi-domain feature parameters and trained learning models are uniformly represented Then store it in network knowledge.

[0071] 4. Optimize decision making

[0072] Optimization decision refers to solving the problem of overall optimization of network resources under multiple attributes and constraints such as user demand, service quality, network resources and utilization efficiency, and then generating an optimization method for decision-making solutions. This method can maximize the user's requirements for service quality such as information capacity, reliable transmission, and timeliness, while maximizing the utilization efficiency of network resources.

[0073] Determine whether the network needs to make optimization decisions. If the current network performance is good, continue to execute multi-domain perception in a loop. If the network performance is affected by the multi-domain environment, optimization decisions need to be made:

[0074] First, check whether there is an anti-interference decision-making scheme matching the current multi-domain environment in the network knowledge. If there is a similar decision-making scheme, further call the learning model and other methods in the network knowledge to fine-tune the existing similar decision-making scheme to generate a comprehensive decision-making scheme that is highly matched with the current internal and external environment of the network; if there is no similar decision-making scheme, perform learning analysis and online decision-making. First, perform learning analysis on the network multi-domain environment, that is, call the multi-domain environment prediction learning model Φ that has been trained in the network knowledge p , input the feature parameters obtained by multi-domain perception, and output the multi-domain environment prediction information at time k+1, that is

[0075]

[0076] Leveraging multi-domain environment prediction information Make online decisions and select appropriate decision methods, such as selecting end-to-end learning models for end-to-end decisions and multi-objective optimization methods for non-end-to-end decisions. At the same time, input real-time knowledge, historical knowledge, and other node decision plans. Under the comprehensive network demand conditions, generate the final comprehensive decision plan, such as switching channels that meet the transmission quality and bandwidth requirements, adjusting the transmission signal power and modulation method, reconstructing the transmission route, adjusting user priority and user location, etc. The comprehensive decision plan is uniformly represented and stored in the network knowledge. The generation of the comprehensive decision plan can be expressed as

[0077]

[0078] Among them, CPlann,i To generate a comprehensive decision plan for the i-th node at level n, CPlan n,l is the comprehensive decision-making scheme for other different nodes, D n,i is the comprehensive network demand, and π is the online decision-making method.

[0079] The second aspect of the present invention proposes a meta-intelligence model in an intelligent information network. The meta-intelligence model includes a multi-domain perception module, a learning analysis module, a network knowledge module, and an optimization decision module; wherein: the multi-domain perception module is configured to: perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters, and pre-process the multi-domain feature parameters; the learning analysis module is configured to: use machine learning to perform data analysis and fusion on the pre-processed multi-domain feature parameters, so as to train a multi-domain environment prediction learning model, and store the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; the network knowledge module is configured to: store network multi-dimensional and multi-state data information with the same representation form, including the multi-domain state The optimization decision module is configured to: match a comprehensive decision-making scheme matching the current network multi-domain environment from the network knowledge module; wherein, when the comprehensive decision-making scheme matching the current network multi-domain environment does not exist in the network knowledge module, the current network multi-domain environment is subjected to learning and analysis, the trained multi-domain environment prediction learning model in the network knowledge module is called, the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making scheme.

[0080] In some embodiments, the communication nodes of the network are deployed in a hierarchical distributed manner, the nodes are divided into X levels, the total number is Y, and the multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where:

[0081]

[0082] Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

[0083]

[0084] In some embodiments, the method for performing data analysis and fusion on the preprocessed multi-domain feature parameters by means of machine learning, thereby obtaining the trained multi-domain environment prediction learning model, specifically includes:

[0085] Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

[0086]

[0087] In some embodiments, the trained multi-domain environment prediction learning model Φ in the network knowledge module is called p , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1:

[0088]

[0089] Using the multi-domain situation prediction information Generate comprehensive decision solutions:

[0090]

[0091] Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node of the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method.

[0092] The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method for implementing a meta-intelligence model in an intelligent information network described in the first aspect of the present disclosure is implemented.

[0093] Figure 2 is a structural diagram of an electronic device according to an embodiment of the present invention, such as Figure 2 As shown, the electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, near field communication (NFC) or other technologies. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covered on the display screen, or a button, a trackball or a touch pad set on the housing of the electronic device, or an external keyboard, touch pad or mouse, etc.

[0094] Those skilled in the art will understand that Figure 2 The structure shown in the figure is only a structural diagram of the part related to the technical solution of the present disclosure, and does not constitute a limitation on the electronic device to which the technical solution of the present application is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0095] The fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for realizing a meta-intelligence model in an intelligent information network described in the first aspect of the present disclosure is implemented.

[0096] In summary, compared with the prior art, the technical solution proposed in the present invention has the following significant advantages: (1) It has comprehensive multi-domain environment perception function. The multi-domain perception module in the model can fully perceive the internal and external environment of the network, such as the electromagnetic environment, the network status, and the user behavior, and obtain the multi-dimensional characteristic parameters of the network, providing a solid foundation for the network to adapt to environmental changes; (2) It has efficient learning and analysis function. The learning and analysis module uses a flexible learning model training method to select a suitable trained learning model through environmental matching to achieve efficient learning and analysis functions and provide technical support for the network adaptive environment; (3) It has flexible data information storage function. The network knowledge module can store various types of network data information. Each module can call the existing data information at any time to realize its own function and provide knowledge reserve for the network adaptive environment; (4) It has complete optimization decision-making function. Combining real-time and historical parameter data, each node in the network selects the best optimization method to generate a decision plan. The decision plans between the nodes can interact through broadcasting and other methods, and finally integrate a comprehensive decision plan to provide intelligent decision-making for the network adaptive environment.

[0097] Please note that the technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. The above-mentioned embodiments only express several implementation methods of the present application, and their descriptions are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present application, several variations and improvements can be made, which all belong to the scope of protection of the present application. Therefore, the scope of protection of the patent in this application shall be based on the attached claims.

Claims

1. A method for implementing a meta-intelligence model in an intelligent information network, characterized in that: The meta-intelligence model constructed by the method includes a multi-domain perception module, a learning analysis module, a network knowledge module and an optimization decision module; wherein: Calling the multi-domain perception module to perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters, and preprocessing the multi-domain feature parameters; Calling the learning analysis module to perform data analysis and fusion on the preprocessed multi-domain feature parameters in a machine learning manner, thereby training a multi-domain environment prediction learning model, and storing the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; Calling the network knowledge module to store network multi-dimensional and multi-state data information having the same representation form, including the multi-domain situation information and the trained multi-domain environment prediction learning model; Calling the optimization decision module to match a comprehensive decision solution matching the current network multi-domain environment from the network knowledge module; Among them, when there is no comprehensive decision-making solution matching the current network multi-domain environment in the network knowledge module, the current network multi-domain environment is subjected to learning analysis, and the trained multi-domain environment prediction learning model in the network knowledge module is called, and the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making solution.

2. The method for implementing a meta-intelligence model in an intelligent information network according to claim 1, characterized in that: The communication nodes of the network are deployed in a hierarchical distributed manner. The nodes are divided into X levels, with a total of Y. The multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where: Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

3. The method for implementing a meta-intelligence model in an intelligent information network according to claim 2, characterized in that: The method for performing data analysis and fusion on the preprocessed multi-domain feature parameters by means of machine learning, thereby training the multi-domain environment prediction learning model, specifically includes: Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

4. The method for implementing a meta-intelligence model in an intelligent information network according to claim 3, characterized in that: Call the trained multi-domain environment prediction learning model Φ in the network knowledge module p , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1: Using the multi-domain situation prediction information Generate comprehensive decision solutions: Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node at the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method.

5. A meta-intelligence model in an intelligent information network, characterized in that: The meta-intelligence model includes a multi-domain perception module, a learning analysis module, a network knowledge module and an optimization decision module; wherein: The multi-domain perception module is configured to: perceive the multi-domain environment inside and outside the network to obtain multi-domain feature parameters, and pre-process the multi-domain feature parameters; The learning analysis module is configured to: perform data analysis and fusion on the preprocessed multi-domain feature parameters in a machine learning manner, thereby training a multi-domain environment prediction learning model, and store the multi-domain feature information and the trained multi-domain environment prediction learning model in the network knowledge module; The network knowledge module is configured to: store network multi-dimensional multi-state data information having the same representation form, including the multi-domain situation information and the trained multi-domain environment prediction learning model; The optimization decision module is configured to: match a comprehensive decision solution matching the current network multi-domain environment from the network knowledge module; Among them, when there is no comprehensive decision-making solution matching the current network multi-domain environment in the network knowledge module, the current network multi-domain environment is subjected to learning analysis, and the trained multi-domain environment prediction learning model in the network knowledge module is called, and the current multi-domain feature parameters extracted from the current network multi-domain environment are used as the input of the trained multi-domain environment prediction learning model, and the multi-domain situation prediction information of the current network multi-domain environment is output to generate the comprehensive decision-making solution.

6. The meta-intelligence model in an intelligent information network according to claim 5, characterized in that: The communication nodes of the network are deployed in a hierarchical distributed manner. The nodes are divided into X levels, with a total of Y. The multi-domain perception module of the i-th communication node of the n-th level obtains the multi-domain feature parameters at time k n∈X,i∈Y, where: Ea k is the external environment domain perception input, Na k is the network state domain perception input, Ua k is the user behavior domain perception input, Ψ e (Ea k ) is the electromagnetic environment domain perception method, Ψ n (Na k ) is the network state domain perception method, Ψ u (Ua k ) is the user behavior domain awareness method, Channel Quality Interference intensity Transmission signal delay The characteristic parameters of the external environment domain constituted by Network topology Network congestion level Packet loss rate The network state domain characteristic parameters constituted by User business type User Priority User Location The characteristic parameters of user behavior domain are:

7. The meta-intelligence model in an intelligent information network according to claim 6, characterized in that: The method for performing data analysis and fusion on the preprocessed multi-domain feature parameters by means of machine learning, thereby training the multi-domain environment prediction learning model, specifically includes: Untrained learning model from the network knowledge module Select a prediction learning model that matches the current network multi-domain environment The multi-domain feature parameters from the start time to the k time are used for training to obtain the trained multi-domain environment prediction learning model:

8. The meta-intelligence model in an intelligent information network according to claim 7, characterized in that: Call the trained multi-domain environment prediction learning model Φ in the network knowledge module p , using the current multi-domain feature parameters extracted from the current network multi-domain environment as the input of the trained multi-domain environment prediction learning model, and outputting the multi-domain situation prediction information at time k+1: Using the multi-domain situation prediction information Generate comprehensive decision solutions: Among them, CPlan n,i is the comprehensive decision-making plan for the i-th node at the n-th level, CPlan n,l is the comprehensive decision-making scheme of other nodes different from the i-th node of the n-th level, D n,i is the comprehensive network demand, and π is the online decision-making method.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the method for implementing a meta-intelligence model in an intelligent information network as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for implementing a meta-intelligence model in an intelligent information network as described in any one of claims 1 to 4 is implemented.

Images