Software network reliability determination method, device and server

Abstract

This specification provides a method, device and server for determining the reliability of a software network, which can be used in the field of artificial intelligence. Based on this method, before specific implementation, according to the preset training rules, the node ratio constraint and the network diameter constraint are jointly used to train to obtain a preset network reliability determination model suitable for relatively large and complex software networks. During specific implementation, after constructing the target software network, the target node network diagram of the target software network is first obtained; and the preset network reliability determination model is used to process the target node network diagram to determine whether the target software network meets the preset reliability requirements; and if it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted. Therefore, it can be better applied to relatively large and complex software networks, and by using the preset network reliability determination model, the reliability of the target software network can be automatically evaluated efficiently and accurately.

Description

Technical Field

[0001] This specification belongs to the field of artificial intelligence technology, and in particular relates to a method, device, and server for determining the reliability of a software network. Background Art

[0002] In financial business scenarios, trading platforms usually need to deploy and run corresponding software networks to implement specific transaction business data processing.

[0003] However, the processing of transaction data is complex and cumbersome, often involving multiple layers of approval, multiple business combinations, and multiple security verifications. This results in a large and complex software network. Existing methods often struggle to efficiently and accurately automatically evaluate the reliability of these software networks, posing a risk to the stability of subsequent transaction data processing.

[0004] To address the above issues, no effective solutions have been proposed so far. Summary of the Invention

[0005] This specification provides a method, device, and server for determining the reliability of a software network, which are well suited for relatively large and complex software networks. By utilizing a preset network reliability determination model, the reliability of the target software network can be automatically evaluated efficiently and accurately.

[0006] This specification provides a method for determining the reliability of a software network, including:

[0007] Obtain a target node network graph of a target software network;

[0008] The target node network graph is processed using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints;

[0009] Based on the target processing results, determine whether the target software network meets the preset reliability requirements;

[0010] When it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted.

[0011] In one embodiment, adjusting the network structure of the target software network includes:

[0012] Determine the matching target adjustment strategy based on the target processing results;

[0013] According to the target adjustment strategy, software nodes of related software in the target software network are added, and / or network connections between related software in the target software network are reduced to obtain an updated target software network.

[0014] In one embodiment, after obtaining the updated target software network, the method further includes:

[0015] obtaining an updated target node network graph regarding the updated target software network;

[0016] Using a preset network reliability determination model to process the updated target node network graph, determine whether the updated target software network meets the preset reliability requirements;

[0017] When it is determined that the updated target software network meets the preset reliability requirements, the updated target software network is run to perform corresponding transaction business data processing.

[0018] In one embodiment, the method further comprises:

[0019] Constructing multiple initial network reliability determination models according to a preset training rule; wherein different initial network reliability determination models contain different numbers of hidden layers;

[0020] Obtaining a sample node network diagram; and processing the sample node network diagram according to a preset processing rule to determine a reference value of the reliability of the sample node network diagram;

[0021] Using the sample node network graph and the reference value of the reliability of the sample node network graph, a plurality of initial network reliability determination models are trained to obtain a plurality of corresponding target network reliability determination models;

[0022] A preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

[0023] In one embodiment, processing the sample node network graph according to a preset processing rule to determine a reference value of the reliability of the sample node network graph includes:

[0024] Calling the target tool software to pre-process the sample node network diagram according to the preset processing rules to obtain the pre-processed sample node network diagram;

[0025] The node ratio constraint and the network diameter constraint are used together to determine the target spanning tree that meets the requirements in the preprocessed sample node network graph;

[0026] By calculating the non-intersecting sum of the union of target spanning trees that meet the requirements, a reference value of the reliability of the sample node network graph is determined.

[0027] In one embodiment, the sample node network graph is preprocessed according to a preset processing rule to obtain a preprocessed sample node network graph, including:

[0028] According to the preset processing rules, the connection relationship between the nodes in the sample node network diagram is determined;

[0029] According to the connection relationship between nodes in the sample node network graph, a corresponding relationship data value sequence is generated;

[0030] The relationship data value sequence is converted into the corresponding adjacency matrix as the preprocessed sample node network graph.

[0031] In one embodiment, the node ratio constraint and the network diameter constraint are jointly used to determine a target spanning tree that meets the requirements in the preprocessed sample node network graph, including:

[0032] According to the preset processing rules, the target tool software is used to jointly use the node ratio constraint and the network diameter constraint to perform multiple rounds of traversal search on the preprocessed sample node network graph to determine the target spanning tree that meets the requirements.

[0033] In one embodiment, according to preset processing rules, multiple rounds of traversal search are performed on the pre-processed sample node network graph according to the node ratio constraint, including:

[0034] The current round of traversal search is performed as follows:

[0035] Determine the number of valid nodes in the current round based on the node ratio constraint threshold, the total number of network nodes, and the number of rounds in the current round;

[0036] According to the number of valid nodes in the current round, the initial subgraph of the current round is determined in the preprocessed sample node network graph;

[0037] Detect and delete irrelevant subgraphs from the initial subgraph of the current round to obtain the remaining subgraph of the current round;

[0038] Determine the initial spanning tree of the current round based on the remaining subgraphs of the current round;

[0039] From the initial spanning trees of the current round, spanning trees with a network diameter greater than a network diameter constraint threshold are detected and deleted to obtain the target spanning tree of the current round.

[0040] In one embodiment, after obtaining the target spanning tree for the current round, the method further includes:

[0041] Check whether the number of valid nodes in the current round is equal to the total number of network nodes;

[0042] When it is determined that the number of valid nodes in the current round is equal to the total number of network nodes, the traversal search is stopped; and by summarizing the target spanning trees of multiple rounds, a target spanning tree that meets the requirements in the preprocessed sample node network graph is determined.

[0043] In one embodiment, a plurality of initial network reliability determination models are trained using a sample node network graph and a reference value of the reliability of the sample node network graph, including:

[0044] Model training is performed on a current initial network reliability determination model among the multiple initial network reliability determination models in the following manner:

[0045] Using the current initial network reliability determination model to process multiple sample node network graphs, multiple sample processing results are obtained;

[0046] Calculate the sample average error value based on the reference value of the reliability of multiple sample node network graphs and multiple sample processing results;

[0047] According to the sample average error value, the model parameters of the current initial network reliability determination model are adjusted.

[0048] In one embodiment, selecting a preset network reliability determination model that meets the requirements from a plurality of target network reliability determination models includes:

[0049] Processing the test node network diagram according to a preset processing rule to determine a reference value of the reliability of the test node network diagram;

[0050] Utilizing multiple target network reliability determination models to process the test node network diagrams respectively, and determining multiple corresponding test processing results;

[0051] Calculate the test average error value and test stability based on the reference value of the reliability of the test node network diagram and multiple test processing results;

[0052] According to the test average error value and the test stability, a preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

[0053] This specification also provides a device for determining the reliability of a software network, comprising:

[0054] An acquisition module, used for acquiring a target node network graph of a target software network;

[0055] a processing module, configured to process the target node network graph using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints;

[0056] A determination module is used to determine whether the target software network meets the preset reliability requirements based on the target processing result;

[0057] The adjustment module is used to adjust the network structure of the target software network when it is determined that the target software network does not meet the preset reliability requirements.

[0058] This specification also provides a server, including a processor and a memory for storing processor-executable instructions, wherein the processor implements the relevant steps of the method for determining the reliability of the software network when executing the instructions.

[0059] This specification also provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the relevant steps of the method for determining the reliability of a software network.

[0060] This specification also provides a computer program product, comprising a computer program, which implements the steps of the method for determining the reliability of a software network when the computer program is executed by a processor.

[0061] Based on the software network reliability determination method, device and server provided in this specification, before specific implementation, first, according to the preset training rules, the node ratio constraint and the network diameter constraint are jointly used to train a preset network reliability determination model suitable for a relatively large and complex software network. During specific implementation, after constructing the target software network, the target node network diagram of the target software network can be obtained first; and the target node network diagram can be processed using the preset network reliability determination model to obtain the corresponding target processing result; according to the target processing result, it is determined whether the target software network meets the preset reliability requirements; and if it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted. Therefore, it can be better applied to relatively large and complex software networks, and by using the preset network reliability determination model, the reliability of the target software network can be automatically evaluated efficiently and accurately, and the network structure of the target software network that does not meet the preset reliability can be adjusted in a timely, effective and accurate manner to obtain a relatively reliable target software network, and then the corresponding transaction business data processing can be carried out efficiently and stably based on the target software network. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of this specification, the following is a brief introduction to the drawings required for use in the embodiments. The drawings described below are only some of the embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0063] Figure 1 is a flowchart of a method for determining the reliability of a software network provided by an embodiment of this specification;

[0064] Figure 2 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0065] Figure 3 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0066] Figure 4 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0067] Figure 5 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0068] Figure 6 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0069] Figure 7 This is a schematic diagram of the structure of a server provided by an embodiment of this specification;

[0070] Figure 8 This is a schematic diagram of the structure of a software network reliability determination device provided by an embodiment of this specification;

[0071] Figure 9 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example;

[0072] Figure 10 This is a schematic diagram of an embodiment of a method for determining the reliability of a software network provided by an embodiment of this specification, in a scenario example. DETAILED DESCRIPTION

[0073] To help those skilled in the art better understand the technical solutions in this specification, the following will provide a clear and complete description of the technical solutions in the embodiments of this specification, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. All other embodiments derived by those skilled in the art based on the embodiments in this specification without creative effort shall fall within the scope of protection of this specification.

[0074] It should be noted that the user-related information and data involved in the embodiments of this specification are all information and data authorized by the user or fully authorized by relevant parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users or relevant parties to choose to authorize or refuse.

[0075] It should also be noted that in the embodiments of this specification, certain software, components, models and other existing solutions in the industry may be mentioned. They should be regarded as exemplary and their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0076] See Figure 1 As shown, the embodiment of this specification provides a method for determining the reliability of a software network. The method is specifically applied to the server side. When implemented, the method may include the following:

[0077] S101: Obtain a target node network graph of a target software network;

[0078] S102: Processing the target node network graph using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints;

[0079] S103: Determine whether the target software network meets the preset reliability requirements based on the target processing result;

[0080] S104: When it is determined that the target software network does not meet the preset reliability requirement, the network structure of the target software network is adjusted.

[0081] The target software network may be specifically understood as a software network deployed on a trading platform and responsible for processing specific transaction business data on the trading platform.

[0082] Specifically, the target software network may include multiple software nodes deployed on the same or different server nodes. Each software node corresponds to a piece of business software. Due to the relevance of different business software in the transaction data processing process, network connections related to transaction data processing may exist between different software nodes.

[0083] The above-mentioned preset network reliability determination model can be specifically understood as a neural network model that can automatically process the node network diagram of the software network of the input model, detect the reliability of the software network based on the node ratio constraint and network diameter constraint, and output the corresponding processing results.

[0084] Specifically, the above-mentioned preset reliability determination model can be a neural network model obtained by deep learning training based on preset processing rules and jointly using node ratio constraints and network diameter constraints.

[0085] Based on the above embodiment, a pre-trained preset network reliability determination model suitable for relatively large and complex software networks can be used to automatically evaluate the reliability of the target software network efficiently and accurately. When it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted in a targeted manner so that the target software network can have better performance, so that the target software network can be used to accurately and stably process specific transaction business data in the future.

[0086] In some embodiments, after obtaining the target node network graph of the target software network, the target node network graph may be preprocessed first; and then the preprocessed target node network graph may be used in subsequent data processing.

[0087] In some embodiments, the above software network reliability determination method can be specifically applied to a server side. The above server can also be connected to a test terminal.

[0088] The aforementioned server may specifically include a backend data processing server, which is applied to the trading platform and is capable of performing functions such as data transmission and data processing. Specifically, the server may be, for example, an electronic device with data computing, storage, and network interaction capabilities. Alternatively, the server may be a software program running on the electronic device that provides support for data processing, storage, and network interaction. In this embodiment, the number of servers is not specifically limited. The server may be a single server, several servers, or a server cluster formed by several servers.

[0089] The test terminal may specifically include a front-end, used by the tester, that enables data collection, data transmission, and other functions. Specifically, the test terminal may be, for example, an electronic device such as a desktop computer, tablet computer, or laptop computer. Alternatively, the test terminal may be a software application that can run on the aforementioned electronic devices.

[0090] In specific implementation, after the target software network for processing target transaction business data (for example, transfer transaction business data) is built in the background of the trading platform, testers are required to first conduct reliability testing on the target software network to determine whether the target software network meets the preset reliability requirements.

[0091] At this time, the tester can initiate a target test request for the target software network through the test terminal, wherein the target test request at least carries the network identifier of the target software network.

[0092] Correspondingly, the server receives the target test request and determines the target software network to be tested based on the network identifier; then obtains and converts the target software network into a corresponding target node network diagram based on the connection relationship data of the target software network; and then uses the preset network reliability determination model to process the target node network diagram to determine whether the target software network meets the preset reliability requirements; and feeds back to the test terminal.

[0093] Specifically, when the test terminal determines that the target software network meets the preset reliability requirements, the target software network test can be determined to have passed; and then the target software network can be officially started to process specific transaction business data of the trading platform.

[0094] On the contrary, when the test terminal determines that the target software network does not meet the preset reliability requirements, it can adjust the target software network in a targeted manner according to the feedback data to improve the reliability of the target software network so as to meet the preset reliability requirements.

[0095] In some embodiments, during specific implementation, the target processing result at least includes the reliability of the target software network.

[0096] Accordingly, determining whether the target software network meets the preset reliability requirements based on the target processing result may include, in specific implementation, comparing the reliability of the target software network with a preset reliability evaluation threshold to obtain a corresponding reliability comparison result; and determining, based on the reliability comparison result, that the target software network meets the preset reliability requirements when the reliability of the target software network is determined to be greater than or equal to the preset reliability evaluation threshold. Conversely, determining that the target software network does not meet the preset reliability requirements when the reliability of the target software network is determined to be less than the preset reliability evaluation threshold.

[0097] The preset reliability evaluation threshold is determined as follows: a large number of node network diagrams of real software networks with known good and poor reliability are collected as first sample node network diagrams; at the same time, based on the first standard node network diagram and combined with expert experience, virtual node network diagrams of software networks with good and poor reliability are simulated and constructed as second sample node network diagrams; the first sample node network diagram and the second sample node network diagram are mixed to obtain a mixed sample node network diagram; the mixed sample node network diagram is processed using a preset network reliability determination model to obtain multiple sample reliabilities; the multiple sample reliabilities are clustered to obtain a first reliability sample data group and a second reliability sample data group; wherein the first reliability sample data group corresponds to the mixed sample node network with good reliability, and the second reliability sample data group corresponds to the mixed sample node network with poor reliability. Statistical operations are performed based on the sample reliabilities contained in the first reliability sample data group; and the preset reliability evaluation threshold is determined based on the corresponding statistical distribution parameters.

[0098] In some embodiments, the target processing result may further include auxiliary parameters associated with the reliability of the target software network. These auxiliary parameters can reflect the local reliability of each sub-network within the target software network. These auxiliary parameters are intermediate data generated by a preset network reliability determination model when processing the target node network graph.

[0099] Accordingly, during specific implementation, when determining whether the target software network meets the preset reliability requirements based on the target processing results, the unreliable areas in the target software network can be determined based on the auxiliary parameters in the target processing results; and then the network structure of the above-mentioned reliable areas can be adjusted in a targeted manner to obtain a target software network whose reliability meets the preset reliability requirements.

[0100] In some embodiments, the above-mentioned adjustment of the network structure of the target software network may include the following when implemented:

[0101] S1: Determine the matching target adjustment strategy based on the target processing results;

[0102] S2: According to the target adjustment strategy, software nodes of related software in the target software network are added, and / or network connections between related software in the target software network are reduced to obtain an updated target software network.

[0103] Based on the above embodiment, a matching target adjustment strategy can be determined and utilized according to the target processing result to perform targeted adjustments on the network structure of the target software network to improve the reliability of the target software network.

[0104] In a specific implementation, based on the target processing result, a preset adjustment strategy library can be searched using a combination of reliability and auxiliary parameters to determine a matching preset adjustment strategy as the target adjustment strategy. The preset adjustment strategy library stores multiple preset adjustment strategies, each of which corresponds to at least one combination of reliability and auxiliary parameters.

[0105] Before specific implementation, a large number of adjustment records for software networks can be collected; and a large number of adjustment records for software networks can be clustered to obtain multiple preset adjustment strategies; and then multiple preset adjustment strategies can be combined to build a preset adjustment strategy library.

[0106] In some embodiments, see Figure 2 As shown, after obtaining the updated target software network, the method may further include the following contents when implemented:

[0107] S1: Obtain an updated target node network graph of the updated target software network;

[0108] S2: using a preset network reliability determination model to process the updated target node network graph to determine whether the updated target software network meets the preset reliability requirements;

[0109] S3: When it is determined that the updated target software network meets the preset reliability requirements, the updated target software network is run to perform corresponding transaction business data processing.

[0110] In specific implementation, if it is determined that the updated target software network meets the preset reliability requirements, the adjustment can be stopped and the current updated target software network can be determined as the final target software network. Then, the trading platform can process the corresponding transaction business data by running the target software network.

[0111] On the contrary, when it is determined that the updated target software network does not meet the preset reliability requirements, a matching target adjustment strategy can be re-determined based on the processing results of the current updated target software network; and then, based on the target adjustment strategy, further adjustments can be made on the basis of the current updated target software network until a software network that meets the preset reliability requirements is obtained.

[0112] Based on the above embodiment, the software network can be adjusted multiple times and the reliability of the software network can be evaluated multiple times to ultimately obtain a software network that meets preset reliability requirements.

[0113] In some embodiments, see Figure 3 As shown, when the above method is implemented, it may also include the following contents:

[0114] S1: constructing multiple initial network reliability determination models according to a preset training rule; wherein different initial network reliability determination models contain different numbers of hidden layers;

[0115] S2: Obtain a sample node network diagram; and process the sample node network diagram according to a preset processing rule to determine a reference value of the reliability of the sample node network diagram;

[0116] S3: using the sample node network graph and the reference value of the reliability of the sample node network graph, performing model training on multiple initial network reliability determination models to obtain corresponding multiple target network reliability determination models;

[0117] S4: Filtering out a preset network reliability determination model that meets the requirements from a plurality of target network reliability determination models.

[0118] The number of hidden layers included in the initial network reliability determination model may be 1, 2, or any other number of hidden layers.

[0119] In specific implementation, the sample node network graph can be processed according to preset processing rules, using the node ratio constraint and the network diameter constraint in combination, to determine a reference value of the reliability of the sample node network graph.

[0120] Then, according to the preset training rules, the reference value of the reliability of the above-mentioned sample node network diagram is used to perform deep learning-based model training on multiple initial network reliability determination models, so that node proportion constraints and network diameter constraints can be introduced and used in the model training process to obtain a preset network reliability determination model that is suitable for larger and more complex software networks and has better model effects and higher model accuracy.

[0121] Based on the above embodiments, according to preset training rules and preset processing rules, by jointly using node ratio constraints and network diameter constraints, a preset network reliability determination model that is suitable for the software network and has good model effect can be efficiently trained.

[0122] In some embodiments, see Figure 4 As shown, the above processing of the sample node network diagram according to the preset processing rules to determine the reference value of the reliability of the sample node network diagram may include the following contents during specific implementation:

[0123] S1: calling the target tool software to pre-process the sample node network diagram according to the preset processing rules to obtain the pre-processed sample node network diagram;

[0124] S2: Jointly use the node ratio constraint and the network diameter constraint to determine the target spanning tree that meets the requirements in the preprocessed sample node network graph;

[0125] S3: Determine a reference value of the reliability of the sample node network graph by calculating the non-intersecting sum of the union of target spanning trees that meet the requirements.

[0126] The sample node network diagram may be obtained based on a sample software network of known reliability.

[0127] Based on the above embodiment, the node ratio constraint and the network diameter constraint can be effectively used simultaneously according to the preset processing rules to obtain a reference value of the reliability of the sample node network graph that is suitable for model training and has good results.

[0128] During specific implementation, the system tools in the target tool software can be called to efficiently process the sample node network diagram and determine the reference value of the reliability of the sample node network diagram.

[0129] Specifically, the target tool software may be MATLAB, and the system tool may be NNTool.

[0130] Specifically, the aforementioned MATLAB refers to a standard numerical computing software that is widely used in algorithm development, data analysis, and calculation, and has excellent and professional algorithms. The aforementioned NNTool refers to a powerful neural network system tool that provides a simple and convenient GUI interface.

[0131] In some embodiments, the above-mentioned preprocessing of the sample node network graph according to the preset processing rules to obtain the preprocessed sample node network graph may include the following contents during specific implementation:

[0132] S1: Determine the connection relationship between nodes in the sample node network diagram according to the preset processing rules;

[0133] S2: Generate corresponding relationship data value sequences based on the connection relationships between nodes in the sample node network graph;

[0134] S3: Convert the relationship data value sequence into the corresponding adjacency matrix as the preprocessed sample node network graph.

[0135] The adjacency matrix may specifically refer to a data structure that uses a two-dimensional array to store relationships (edges or arcs) between vertices.

[0136] Based on the connection relationships between nodes in the sample node network graph, a corresponding sequence of relationship data values ​​is generated. In specific implementations, the connection relationships between nodes in the sample node network graph can be used to determine whether a valid connection exists between two nodes. If a valid connection exists, the corresponding relationship data value is set to 1; otherwise, it is set to 0. The relationship data values ​​are then arranged and combined in sequence to obtain a corresponding {0, 1} sequence, i.e., a relationship data value sequence. The relationship data value sequence is then converted into a corresponding adjacency matrix to obtain data that is convenient for subsequent computational processing, i.e., a preprocessed sample node network graph. The preprocessed sample node network graph can then be used in place of the original sample node network graph for subsequent data processing, thereby improving the efficiency and accuracy of subsequent data processing.

[0137] Based on the above embodiments, the sample node network graph can be accurately converted into a preprocessed sample node network graph suitable for subsequent operation processing through preprocessing according to preset processing rules.

[0138] In specific implementation, before determining the connection relationship between nodes in the sample node network diagram according to preset processing rules, the sample node network diagram may be first cleaned to eliminate or repair obviously erroneous data in advance.

[0139] When specifically determining the connection relationship between nodes in the sample node network diagram, for nodes where it is unclear whether there is a valid connection, the probability value of the connection being in a connected state can be obtained and calculated based on the operation test records of the relevant sample software network; then the probability value is compared with the preset connectivity probability threshold; when it is determined that the probability value is greater than the preset connectivity probability threshold, the corresponding connection relationship data value is determined to be 1, otherwise it is determined to be 0.

[0140] In some embodiments, the node ratio constraint and the network diameter constraint are used together to determine a target spanning tree that meets the requirements in the preprocessed sample node network graph. The specific implementation may include the following:

[0141] According to the preset processing rules, the target tool software is used to jointly use the node ratio constraint and the network diameter constraint to perform multiple rounds of traversal search on the preprocessed sample node network graph to determine the target spanning tree that meets the requirements.

[0142] The node ratio constraint mentioned above at least includes a node ratio constraint threshold value, which can be expressed as λ. Based on the node ratio constraint mentioned above (for example, it can be expressed as n is the total number of nodes), the reliability of the software network performance can be limited from a global dimension by limiting the number of nodes that are in a connected state when the software network is operating normally.

[0143] The network diameter constraint includes at least a network diameter constraint threshold, which can be denoted as D. Specifically, the network diameter refers to the number of edges used to connect nodes. Based on this network diameter constraint, the reliability of the software network's performance can be locally determined by limiting the number of edges connecting nodes.

[0144] The target spanning tree may correspond to a minimum path set or a minimum cut set related to the target node network graph.

[0145] The minimum path set may specifically refer to a set of minimum node edges that enables the network to work properly, that is, each edge is indispensable, and if any one of the edges is reduced, the network will not work properly.

[0146] The above-mentioned minimum cut set can specifically refer to a subset sequence of points and edges of a network. In this subset sequence, if any one of the edges is removed, the node network graph corresponding to the network will change from a connected graph to a disconnected graph.

[0147] Based on the above embodiment, by jointly using the node ratio constraint and the network diameter constraint as constraints, a target spanning tree that meets the requirements can be accurately found based on the global dimension and the local dimension.

[0148] In some embodiments, see Figure 5 As shown, the above-mentioned method performs multiple rounds of traversal search on the pre-processed sample node network graph according to the preset processing rules and the node ratio constraint. When specifically implemented, it may include: performing the current round of traversal search in the following manner:

[0149] S1: Determine the number of valid nodes in the current round based on the node ratio constraint threshold, the total number of network nodes, and the number of rounds in the current round;

[0150] S2: According to the number of valid nodes in the current round, determine the initial subgraph of the current round in the preprocessed sample node network graph;

[0151] S3: Detect and delete irrelevant subgraphs from the initial subgraph of the current round to obtain the remaining subgraph of the current round;

[0152] S4: Determine the initial spanning tree of the current round based on the remaining subgraphs of the current round;

[0153] S5: Detect and delete spanning trees whose network diameters are larger than a network diameter constraint threshold from the initial spanning trees of the current round, and obtain a target spanning tree of the current round.

[0154] In a specific implementation, the total number of network nodes is n, the node ratio constraint threshold is λ, and the current round number is M, i.e., the Mth round. Accordingly, the number of valid nodes p in the current round can be: λ·n+M-1. Next, based on the number of valid nodes p in the current round, the preprocessed sample node network graph can be traversed to find a subgraph with p node combinations, which serves as the initial subgraph for the current round.

[0155] Next, the initial subgraph of the current round can be traversed and searched to find the subgraph that cannot connect the initial node and the end node, which is recorded as an irrelevant subgraph; and the irrelevant subgraph is removed from the initial subgraph of the current round to obtain the remaining subgraph of the current round.

[0156] Then, the corresponding tree generation algorithm can be used to generate the corresponding spanning tree according to the remaining subgraphs of the current round to obtain the initial spanning tree of the current round; then the network diameter of the initial spanning tree of the current round is calculated, and the spanning trees with network diameters that are greater than the network diameter constraint threshold are searched and deleted to obtain the target spanning tree of the current round.

[0157] Based on the above embodiment, the target spanning tree of the current round can be accurately obtained by performing a traversal search of the current round according to the preset processing rules. Furthermore, multiple rounds of traversal searches can be performed in the above manner to obtain target spanning trees of multiple rounds.

[0158] In some embodiments, after obtaining the target spanning tree for the current round, the method may further include the following steps when implemented:

[0159] S1: Check whether the number of valid nodes in the current round is equal to the total number of network nodes;

[0160] S2: When it is determined that the number of valid nodes in the current round is equal to the total number of network nodes, the traversal search is stopped; and by summarizing the target spanning trees of multiple rounds, a target spanning tree that meets the requirements in the preprocessed sample node network graph is determined.

[0161] On the contrary, when it is determined that the number of valid nodes in the current round is less than the total number of network nodes, the next round of traversal search can be performed in the above manner to determine the target spanning tree for the next round until the number of valid nodes is equal to the total number of network nodes.

[0162] Based on the above embodiment, multiple rounds of traversal search can be performed to comprehensively and completely find a target spanning tree that meets the requirements in the preprocessed sample node network graph.

[0163] In some embodiments, during specific implementation, a non-intersecting product sum algorithm can be used to first calculate the union of target spanning trees that meet the requirements, and then calculate the non-intersecting sum of the union to obtain a reference value of the reliability of the sample node network graph, which can be recorded as: (G, λ, D).

[0164] Specifically, the disjoint sum of the union of target spanning trees that meet the requirements can be calculated as follows:

[0165]

[0166] Among them, E1, E2...E n They represent respectively: the set of target spanning trees of the first round, the set of target spanning trees of the second round...the set of target spanning trees of the nth round.

[0167] In some embodiments, the above-mentioned sample node network diagram and the reference value of the reliability of the sample node network diagram are used to train multiple initial network reliability determination models. For specific implementation, please refer to Figure 6 As shown, the model training of the current initial network reliability determination model among the multiple initial network reliability determination models can be performed in the following manner:

[0168] S1: Use the current initial network reliability determination model to process multiple sample node network graphs to obtain multiple sample processing results;

[0169] S2: Calculate the sample average error value based on the reliability reference values ​​of multiple sample node network graphs and multiple sample processing results;

[0170] S3: According to the sample average error value, adjust the model parameters of the current initial network reliability determination model.

[0171] Based on the above embodiment, the above method can be repeated, and the reference value of the reliability of the sample node network diagram can be used to perform multiple rounds of training and adjustment on multiple initial network reliability determination models until the sample average error value is less than or equal to the preset sample average error threshold, thereby obtaining multiple target network reliability determination models that meet the requirements.

[0172] In some embodiments, the above-mentioned screening out of a preset network reliability determination model that meets the requirements from a plurality of target network reliability determination models may include the following when implemented:

[0173] S1: Processing the test node network diagram according to a preset processing rule to determine a reference value of the reliability of the test node network diagram;

[0174] S2: using multiple target network reliability determination models to process the test node network graphs respectively and determine multiple corresponding test processing results;

[0175] S3: Calculate the test average error value and test stability based on the reference value of the reliability of the test node network diagram and multiple test processing results;

[0176] S4: According to the test average error value and the test stability, a preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

[0177] Based on the above embodiments, a preset network reliability determination model that is suitable for the software network and has better effects can be accurately determined from multiple target reliability determination models.

[0178] During specific implementation, the test node network diagram can be randomly divided into multiple groups; and the target network reliability determination model can be tested multiple times using multiple groups of test node network diagrams to obtain multiple test average error values; then, the corresponding average value can be calculated based on the multiple test average error values ​​of the same target network reliability determination model, as the test average error of the target network reliability determination model; at the same time, the corresponding test stability can be obtained by performing variance calculation based on the multiple test average error values ​​of the same target network reliability determination model.

[0179] Furthermore, a weighted calculation can be performed based on the test average error value and the test stability to obtain a corresponding model evaluation value; and then based on the model evaluation value, a preset network reliability determination model that meets the requirements is screened out.

[0180] As can be seen from the above, based on the reliability determination method of the software network provided in the embodiment of this specification, before specific implementation, the preset network reliability determination model applicable to a relatively large and complex software network is obtained by first using the node ratio constraint and the network diameter constraint training in combination according to the preset training rules. During specific implementation, after constructing the target software network, the target node network diagram of the target software network can be obtained first; and the target node network diagram can be processed using the preset network reliability determination model to obtain the corresponding target processing result; based on the target processing result, it is determined whether the target software network meets the preset reliability requirements; and if it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted. Thus, it can be better applied to relatively large and complex software networks, and the preset network reliability determination model can be used to automatically evaluate the reliability of the target software network efficiently and accurately, and the network structure of the target software network that does not meet the preset reliability can be adjusted in a timely, effective and accurate manner to obtain a relatively reliable target software network, and then the corresponding transaction business data processing can be carried out efficiently and stably based on the target software network.

[0181] This specification embodiment provides a server, see Figure 7 The server includes a network communication port 701, a processor 702, and a memory 703, and the above structures are connected through internal cables so that each structure can perform specific data interaction.

[0182] The network communication port 701 may be used to obtain a target node network diagram of a target software network;

[0183] The processor 702 can be specifically used to process the target node network graph using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules, combined with node ratio constraints and network diameter constraints; based on the target processing result, determine whether the target software network meets the preset reliability requirements; if it is determined that the target software network does not meet the preset reliability requirements, adjust the network structure of the target software network.

[0184] The memory 703 can be specifically used to store corresponding instruction programs, as well as target node network diagrams, target processing results and other data.

[0185] Based on the above method, the relevant structural performance of the server can be effectively utilized, the data processing speed of the electronic device can be improved, and the relevant data processing for determining the reliability of the software network can be efficiently realized.

[0186] In this embodiment, the network communication port 701 can be a virtual port that is bound to different communication protocols, thereby being capable of sending or receiving different data. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for email data communication. Furthermore, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM or CDMA; it can also be a Wi-Fi chip; or it can be a Bluetooth chip.

[0187] In this embodiment, the processor 702 may be implemented in any suitable manner. For example, the processor may take the form of a microprocessor or a processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, an embedded microcontroller, etc. This specification is not intended to limit this.

[0188] In this embodiment, the memory 703 may include multiple levels. In a digital system, anything that can store binary data can be a memory. In an integrated circuit, a circuit with a storage function that does not have a physical form is also called a memory, such as RAM, FIFO, etc. In a system, a storage device with a physical form is also called a memory, such as a memory stick, TF card, etc.

[0189] An embodiment of the present specification also provides a computer-readable storage medium based on the above-mentioned software network reliability determination method, wherein the computer-readable storage medium stores computer program instructions, which, when executed, implement the following: obtaining a target node network diagram of the target software network; processing the target node network diagram using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules, combined with node ratio constraints and network diameter constraints; determining whether the target software network meets preset reliability requirements based on the target processing result; and adjusting the network structure of the target software network if it is determined that the target software network does not meet the preset reliability requirements.

[0190] In this embodiment, the storage medium includes, but is not limited to, random access memory (RAM), read-only memory (ROM), cache, hard disk drive (HDD), or memory card. The memory can be used to store computer program instructions. The network communication unit can be an interface configured in accordance with the standards specified by the communication protocol for network connection communication.

[0191] In this embodiment, the functions and effects specifically implemented by the program instructions stored in the computer-readable storage medium can be explained in comparison with other implementations and will not be repeated here.

[0192] An embodiment of the present specification also provides a computer program product, comprising at least a computer program, which, when executed by a processor, implements the following method steps: obtaining a target node network diagram of a target software network; processing the target node network diagram using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules, combined with node ratio constraints and network diameter constraints; determining whether the target software network meets preset reliability requirements based on the target processing result; and adjusting the network structure of the target software network if it is determined that the target software network does not meet the preset reliability requirements.

[0193] See Figure 8 As shown, the embodiment of this specification also provides a device for determining the reliability of a software network, which may specifically include the following structural modules:

[0194] The acquisition module 801 may be specifically used to acquire a target node network diagram of a target software network;

[0195] The processing module 802 may be specifically configured to process the target node network graph using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints;

[0196] The determination module 803 may be specifically configured to determine whether the target software network meets the preset reliability requirements based on the target processing result;

[0197] The adjustment module 804 may be specifically configured to adjust the network structure of the target software network when it is determined that the target software network does not meet the preset reliability requirement.

[0198] In some embodiments, when the above-mentioned adjustment module 804 is specifically implemented, the network structure of the target software network can be adjusted in the following manner: according to the target processing result, a matching target adjustment strategy is determined; according to the target adjustment strategy, software nodes of related software in the target software network are added, and / or, network connections between related software in the target software network are reduced to obtain an updated target software network.

[0199] In some embodiments, after obtaining the updated target software network, the device, when implemented, can also be used to: obtain an updated target node network diagram of the updated target software network; use a preset network reliability determination model to process the updated target node network diagram to determine whether the updated target software network meets the preset reliability requirements; if it is determined that the updated target software network meets the preset reliability requirements, run the updated target software network to perform corresponding transaction business data processing.

[0200] In some embodiments, when the device is implemented, it can also be used to: construct multiple initial network reliability determination models according to preset training rules; wherein different initial network reliability determination models contain different numbers of hidden layers; obtain a sample node network diagram; and process the sample node network diagram according to preset processing rules to determine a reference value of the reliability of the sample node network diagram; use the sample node network diagram and the reference value of the reliability of the sample node network diagram to perform model training on multiple initial network reliability determination models to obtain corresponding multiple target network reliability determination models; and screen out a preset network reliability determination model that meets the requirements from the multiple target network reliability determination models.

[0201] In some embodiments, when the device is implemented, the sample node network diagram can be processed according to preset processing rules in the following manner to determine a reference value for the reliability of the sample node network diagram: calling the target tool software to preprocess the sample node network diagram according to the preset processing rules to obtain a preprocessed sample node network diagram; jointly using the node proportion constraint and the network diameter constraint to determine a target spanning tree that meets the requirements in the preprocessed sample node network diagram; and determining a reference value for the reliability of the sample node network diagram by calculating the non-intersecting sum of the target spanning trees that meet the requirements.

[0202] In some embodiments, when the device is implemented, the sample node network diagram can be preprocessed according to preset processing rules in the following manner to obtain a preprocessed sample node network diagram: according to the preset processing rules, the connection relationship between the nodes in the sample node network diagram is determined; according to the connection relationship between the nodes in the sample node network diagram, a corresponding relationship data value sequence is generated; the relationship data value sequence is converted into a corresponding adjacency matrix as the preprocessed sample node network diagram.

[0203] In some embodiments, when the device is implemented, the node ratio constraint and the network diameter constraint can be used in combination in the following manner to determine a target spanning tree that meets the requirements in the preprocessed sample node network diagram: according to preset processing rules, using target tool software, the node ratio constraint and the network diameter constraint are used in combination to perform multiple rounds of traversal searches on the preprocessed sample node network diagram to determine a target spanning tree that meets the requirements.

[0204] In some embodiments, when the device is implemented, multiple rounds of traversal search can be performed on the preprocessed sample node network graph according to the node ratio constraint according to preset processing rules in the following manner: determine the number of valid nodes in the current round according to the node ratio constraint threshold, the total number of network nodes, and the number of rounds of the current round; determine the initial subgraph of the current round in the preprocessed sample node network graph according to the number of valid nodes in the current round; detect and delete irrelevant subgraphs from the initial subgraph of the current round to obtain the remaining subgraph of the current round; determine the initial spanning tree of the current round according to the remaining subgraph of the current round; detect and delete spanning trees with network diameters greater than the network diameter constraint threshold from the initial spanning tree of the current round to obtain the target spanning tree of the current round.

[0205] In some embodiments, after obtaining the target spanning tree of the current round, the device can also be used, when implemented, to: detect whether the number of valid nodes in the current round is equal to the total number of network nodes; stop the traversal search when it is determined that the number of valid nodes in the current round is equal to the total number of network nodes; and determine the target spanning tree that meets the requirements in the preprocessed sample node network graph by summarizing the target spanning trees of multiple rounds.

[0206] In some embodiments, when the above-mentioned device is implemented, the sample node network diagram and the reference value of the reliability of the sample node network diagram can be used to perform model training on multiple initial network reliability determination models in the following manner: use the current initial network reliability determination model to process multiple sample node network diagrams to obtain multiple sample processing results; calculate the sample average error value based on the reference value of the reliability of the multiple sample node network diagrams and the multiple sample processing results; adjust the model parameters of the current initial network reliability determination model based on the sample average error value.

[0207] In some embodiments, when the above-mentioned device is implemented, a preset network reliability determination model that meets the requirements can be screened out from multiple target network reliability determination models in the following manner: the test node network diagram is processed according to a preset processing rule to determine a reference value of the reliability of the test node network diagram; the test node network diagram is processed separately using multiple target network reliability determination models to determine a corresponding multiple test processing results; based on the reference value of the reliability of the test node network diagram and multiple test processing results, the test average error value and the test stability are calculated; based on the test average error value and the test stability, a preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

[0208] It should be noted that the units, devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in terms of functions and are divided into various modules and described separately. Of course, when implementing this specification, the functions of each module can be implemented in the same or multiple software and / or hardware, or the module that implements the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0209] As can be seen from the above, the software network reliability determination device provided in the embodiments of this specification can be well applied to relatively large and complex software networks. It uses a preset network reliability determination model to automatically evaluate the reliability of the target software network efficiently and accurately, and can timely and effectively and accurately adjust the network structure of the target software network that does not meet the preset reliability to obtain a more reliable target software network, and then can subsequently perform corresponding transaction business data processing efficiently and stably based on the target software network.

[0210] In a specific scenario example, the software network reliability determination method provided in this specification can be applied to measure network reliability through deep learning of neural networks. The specific implementation process can be found in the following content.

[0211] In this scenario, a neural network specifically refers to an abstraction of the human brain's neural network from an information processing perspective, building a simple model that can be connected to form different networks. It is a computational model composed of a large number of interconnected nodes, each representing a specific output function. Network reliability specifically refers to a metric that measures network performance. Networks are a special case of interconnected systems, and network reliability primarily studies network reliability algorithms and their complexity.

[0212] In this scenario, existing methods, primarily applicable to small networks, mostly employ state enumeration and full probability decomposition methods to calculate network reliability. However, since calculating network reliability is an NP-hard problem, the computational complexity increases rapidly as the network structure expands, compromising overall processing efficiency and accuracy. This problem becomes particularly acute for large, multi-state networks, such as software networks.

[0213] In response to the above problems, the applicant considered combining neural network deep learning with network reliability calculation. The network reliability (for example, reliability) is determined by combining diameter and node ratio constraints (for example, using node ratio constraints and network diameter constraints in combination), and the true value of the reliability obtained by solving a simplified algorithm based on detecting and deleting irrelevant subgraphs is used to train the neural network model. When calculating the exact solution, Matlab (for example, a target tool software) is introduced and used to write a corresponding network reliability calculation program, and applied to an example of network reliability calculation to calculate the exact value of the example. At the same time, using Matlab's approximate algorithm and deep learning technology for the neural network reliability module (i.e., expanding the hidden layer of the neural network to multiple layers), the neural network is trained through a training set to obtain an approximate value of the exact solution; the approximate value is then compared with the above-mentioned exact value to obtain the error size of the reduction algorithm and the neural network deep learning technology in network reliability assessment, and to verify whether the neural network deep learning technology has computational efficiency and reduces the time required for calculation.

[0214] In this scenario, a network (e.g., the target software network) can be abstracted as a graph G(V,E) (e.g., the target node network graph), where V{v1,v2,…,vn) represents the set of nodes in the network, and E{E1,E2,…,Em} represents the set of operational edges in the network, where n is the number of nodes and m is the number of edges. To better highlight the research focus, the reliability graph G(V,E) can be simplified, and the following two assumptions are proposed:

[0215] Assumption 1: The node will not fail, that is, the node is always in normal working state.

[0216] Assumption 2: Network links, or edges in a graph, are not completely reliable due to environmental factors, channel interference, and other factors. Each edge is assigned a working probability p. Furthermore, each edge is bidirectional and independent, and can be in either a working or faulty state.

[0217] Based on the above assumptions, when there is a group of nodes that can communicate with each other When a subgraph of G has these k nodes, the probability that G is in working state at a certain moment is called k-end reliability; when k = V, it is terminal reliability; when k = 2, it is two-end reliability.

[0218] At the same time, to facilitate research and explanation, I use the node ratio constraint and the network diameter constraint together to define network reliability as follows:

[0219] Definition 1: The probability that all nodes in a network G communicate with each other is called the network reliability of G, denoted by R(G).

[0220] Accordingly, according to Definition 1, the reliability model can be improved by controlling the percentage nodes.

[0221] Definition 2: In G(V,E), the reliability of λ is at least The probability that a node is connected in a group of n nodes is denoted by R(G,λ) (e.g., node ratio constraint).

[0222] Accordingly, according to Definition 2, the larger λ is, the higher the requirement is. For example, λ = 0.8 means that at least 80% of the nodes in the network are connected to each other without functional node information.

[0223] On the other hand, if two nodes in the graph are connected through too many edges, the reliability of the network will be affected. Setting Definition 3.

[0224] Definition 3: In G(V,E), the reliability of D is that there exists a path consisting of edges between every pair of nodes, and the number of edges is bounded by a given integer D, denoted by R(G,D) (e.g., the network diameter constraint).

[0225] Accordingly, Definition 3 reflects the local performance and reliability of some nodes, that is, the diameter constraint of G is less than or equal to D. For example, when D = 3, the minimum diameter connecting any two points in the graph G(V,E) cannot be greater than 3. If it is greater than 3, the two points are considered disconnected.

[0226] In addition, compared with Definition 2 and Definition 3, the node ratio constraint and diameter constraint are not considered at the same time. Therefore, the reliability model of diameter constraint and node ratio constraint is introduced, namely:

[0227] In G(V,E), at least The probability that k nodes communicate in a group of n nodes, and the diameter of the subgraph composed of these k nodes is no greater than a given integer D, λ, and the reliability of D, denoted as R(G,λ,D) (for example, using a joint node ratio constraint and a network diameter constraint).

[0228] We also consider reliability algorithms, which are generally categorized as exact and approximate. Exact algorithms are primarily suitable for small, simple networks. Reliability calculations for large, complex networks often face a "combinatorial explosion" problem, where the computational effort increases exponentially with the number of nodes. This often results in low computational efficiency and high complexity. In this scenario, we consider calculating the reliability of a small network and comparing it with the results of training a small network using deep neural network learning techniques. This study examines the feasibility of applying deep neural network learning to large, complex networks in terms of computational efficiency and accuracy. Specifically, we consider using the disjoint product sum algorithm.

[0229] The premise for calculating reliability using the aforementioned disjoint sum algorithm is to find all minimal path sets and minimal cut sets in the network. A minimal path set is defined as a path set whose edge sequence, when removed from any one edge, no longer constitutes a path set. A minimal path set represents the minimum set of nodes and edges required for a network to function properly. In other words, every edge is essential; removing any one of them would render the network inoperable. A minimal cut set is a sequence of edge subsets in a network graph where, if any one edge is removed from the subset, the graph changes from a connected graph to a disconnected graph. The smallest subset that satisfies this condition is called a minimal cut set.

[0230] Accordingly, based on the above-mentioned disjoint product algorithm, the reliability of the network can be calculated by finding the union of the minimum path set or the minimum cut set and then solving the disjoint sum of the union.

[0231] The neural network model used may include an input layer, hidden layers, and an output layer. A single circle represents a neuron, and arrows indicate the direction of transmission. The number of neurons in the input layer corresponds to the dimension of our input matrix. The number of neurons in the hidden layer is generally related to the input and output layers, with the output layer being related to the dimension of the target output data.

[0232] During specific implementation, the neural network simulation tool NNTOOL can be used.

[0233] Specifically, if you want to establish a good neural network algorithm, simulation program and output analysis of the results, it takes a long time of exploration and experimentation and has high technical requirements for the user. At the same time, changes in parameters such as network type, structure, and excitation transfer function often cause major changes to the algorithm, which is inefficient and has poor versatility, causing many inconveniences in training and simulation. Therefore, if there is an auxiliary tool that integrates neural network system training, parameter adjustment, simulation and analysis, it will greatly facilitate research work. Among them, the above-mentioned MATLAB has always been an internationally recognized standard numerical calculation software, widely used in algorithm development, data analysis and calculation, and has excellent and professional algorithms. Among them, the above-mentioned NNTool is a powerful neural network system tool that provides a simple and convenient GUI interface. Using it, researchers can easily study and solve a considerable number of neural network problems.

[0234] In this scenario, taking into account the above considerations, NNTool can be used to combine multiple constraints to accurately calculate reliability (e.g., reliability reference value). This calculated reliability is then used to train a neural network model, resulting in a model that can automatically evaluate reliability (the pre-set network reliability determination model). This model can then be used to evaluate network reliability.

[0235] In the specific implementation, in order to calculate R(G, λ, D), we consider The network reliability is calculated by using the subgraphs and their corresponding spanning trees. Please refer to Table 1 and the network reliability R(G,λ,D) is calculated by the corresponding algorithm.

[0236] Table 1

[0237]

[0238]

[0239] When implementing this algorithm using Matlab, the network reliability results derived from the algorithm can be compared with simulation results from deep neural network learning. However, the number of samples required for neural network training should be critical; otherwise, the simulation results will exhibit significant errors and the comparison will be unsatisfactory. Manually drawing dozens or even hundreds of network diagrams would be prohibitively time-consuming, so we used a program to automatically generate a 01 sequence for the network diagram. If an edge exists between two nodes, a 1 is used, and if not, a 0 is used. For details, see Table 2.

[0240] Table 2

[0241]

[0242] Convert the {0,1} sequence into an adjacency matrix. A node-to-node adjacency matrix is ​​more convenient for finding subgraphs and spanning trees later in the algorithm, so we convert the {0,1} sequence into an adjacency matrix. Here, we assume the number of nodes is a. See Table 3 for the specific conversion.

[0243] Table 3

[0244]

[0245]

[0246] In specific implementation, the spanning tree algorithm can be used to find all spanning trees. Delete the spanning trees with a diameter greater than D and calculate the reliability, where A(T e ) indicates T e Please refer to Table 4 for the diameter.

[0247] Table 4

[0248]

[0249] The results show that, compared with the exact reliability value, the error of the neural network approximation is less than 0.03 regardless of the number of hidden layers, showing a good approximate fitting effect. The error under deep learning approximation is even smaller. However, the more hidden layers there are, the better the approximation effect is not necessarily the case. For example, the error of 20 hidden layers is smaller than that of 2 hidden layers. This shows that the hidden layer has nothing to do with the approximation effect. When training a neural network, you can train several networks to find the relatively optimal approximation value. For details, please refer to Table 5 and Figure 9 and Figure 10 .

[0250] Table 5

[0251]

[0252] The simulation results and analysis above demonstrate that using Matlab neural network deep learning technology to approximate network reliability is feasible. Limited by the memory and time required to calculate the exact reliability value, the present invention trains the neural network with relatively few samples. With a sufficient number of samples, the error between the approximate value and the true value decreases. Once the neural network is trained, simply inputting a sequence of network graphs can yield a reliable approximation of reliability with minimal computational effort and computational time.

[0253] It is necessary to perform repeated and complex calculations on network reliability, such as comparing the advantages and disadvantages of network models and optimizing the design of network systems. The use of neural network deep learning methods can greatly save computing time and computing power.

[0254] Through the above scenario examples, it is verified that the software network reliability determination method provided in this manual can indeed be well applied to relatively large and complex software networks, and the corresponding neural network model can be trained and used to automatically evaluate the reliability of the target software network efficiently and accurately.

[0255] Although this specification provides the method operation steps as described in the embodiments or flow charts, more or fewer operation steps may be included based on conventional or non-creative means. The order of steps listed in the embodiments is only one way of executing the order of many steps and does not represent the only execution order. When the device or client product in practice is executed, it can be executed in sequence or in parallel according to the method shown in the embodiments or the drawings (for example, a parallel processor or a multi-threaded processing environment, or even a distributed data processing environment). The term "comprise", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, product or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, product or device. In the absence of more restrictions, it is not excluded that there are other identical or equivalent elements in the process, method, product or device including the elements. Words such as first and second are used to represent names and do not represent any particular order.

[0256] Those skilled in the art will also appreciate that, in addition to implementing the controller in pure computer-readable program code, it is entirely possible to implement the same functionality by logically programming the method steps in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, and the like. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be considered structures within the hardware component. Alternatively, the devices for implementing various functions can be considered both software modules implementing the method and structures within the hardware component.

[0257] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, classes, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in local and remote computer-readable storage media, including storage devices.

[0258] Through the description of the above embodiments, it can be seen that those skilled in the art can clearly understand that this specification can be implemented by means of software plus the necessary general hardware platform. Based on this understanding, the technical solution of this specification can essentially be embodied in the form of a software product. This computer software product can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of this specification.

[0259] The various embodiments in this specification are described in a progressive manner. References to the common or similar parts of the various embodiments are sufficient. Each embodiment focuses on the differences from the other embodiments. This specification can be used in a variety of general-purpose or specialized computer system environments or configurations. For example, personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments that include any of the above systems or devices.

[0260] Although the present specification has been described through embodiments, those skilled in the art will appreciate that there are many modifications and variations to the present specification without departing from the spirit of the present specification. It is intended that the appended claims include these modifications and variations without departing from the spirit of the present specification.

Claims

1. A method for determining the reliability of a software network, characterized in that: include: Obtain a target node network graph of a target software network; The target node network graph is processed using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints; Based on the target processing results, determine whether the target software network meets the preset reliability requirements; When it is determined that the target software network does not meet the preset reliability requirements, the network structure of the target software network is adjusted.

2. The method according to claim 1, characterized in that Adjust the network structure of the target software network, including: Determine the matching target adjustment strategy based on the target processing results; According to the target adjustment strategy, software nodes of related software in the target software network are added, and / or network connections between related software in the target software network are reduced to obtain an updated target software network.

3. The method according to claim 2, characterized in that After obtaining the updated target software network, the method further includes: obtaining an updated target node network graph regarding the updated target software network; Using a preset network reliability determination model to process the updated target node network graph, determine whether the updated target software network meets the preset reliability requirements; When it is determined that the updated target software network meets the preset reliability requirements, the updated target software network is run to perform corresponding transaction business data processing.

4. The method according to claim 1, wherein The method further comprises: Constructing multiple initial network reliability determination models according to a preset training rule; wherein different initial network reliability determination models contain different numbers of hidden layers; Obtaining a sample node network diagram; and processing the sample node network diagram according to a preset processing rule to determine a reference value of the reliability of the sample node network diagram; Using the sample node network graph and the reference value of the reliability of the sample node network graph, a plurality of initial network reliability determination models are trained to obtain a plurality of corresponding target network reliability determination models; A preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

5. The method according to claim 4, characterized in that The sample node network diagram is processed according to a preset processing rule to determine a reference value of the reliability of the sample node network diagram, including: Calling the target tool software to pre-process the sample node network diagram according to the preset processing rules to obtain the pre-processed sample node network diagram; The node ratio constraint and the network diameter constraint are used together to determine the target spanning tree that meets the requirements in the preprocessed sample node network graph; By calculating the non-intersecting sum of the union of target spanning trees that meet the requirements, a reference value of the reliability of the sample node network graph is determined.

6. The method according to claim 5, characterized in that According to the preset processing rules, the sample node network diagram is preprocessed to obtain the preprocessed sample node network diagram, including: According to the preset processing rules, the connection relationship between the nodes in the sample node network diagram is determined; According to the connection relationship between nodes in the sample node network graph, a corresponding relationship data value sequence is generated; The relationship data value sequence is converted into the corresponding adjacency matrix as the preprocessed sample node network graph.

7. The method according to claim 5, characterized in that The node ratio constraint and the network diameter constraint are used together to determine the target spanning tree that meets the requirements in the preprocessed sample node network graph, including: According to the preset processing rules, the target tool software is used to jointly use the node ratio constraint and the network diameter constraint to perform multiple rounds of traversal search on the preprocessed sample node network graph to determine the target spanning tree that meets the requirements.

8. The method according to claim 7, characterized in that According to the preset processing rules, multiple rounds of traversal search are performed on the preprocessed sample node network graph based on the node ratio constraint, including: The current round of traversal search is performed as follows: Determine the number of valid nodes in the current round based on the node ratio constraint threshold, the total number of network nodes, and the number of rounds in the current round; According to the number of valid nodes in the current round, the initial subgraph of the current round is determined in the preprocessed sample node network graph; Detect and delete irrelevant subgraphs from the initial subgraph of the current round to obtain the remaining subgraph of the current round; Determine the initial spanning tree of the current round based on the remaining subgraphs of the current round; From the initial spanning trees of the current round, spanning trees with a network diameter greater than a network diameter constraint threshold are detected and deleted to obtain the target spanning tree of the current round.

9. The method according to claim 8, characterized in that After obtaining the target spanning tree of the current round, the method further includes: Check whether the number of valid nodes in the current round is equal to the total number of network nodes; When it is determined that the number of valid nodes in the current round is equal to the total number of network nodes, the traversal search is stopped; and by summarizing the target spanning trees of multiple rounds, a target spanning tree that meets the requirements in the preprocessed sample node network graph is determined.

10. The method according to claim 4, characterized in that Using the sample node network graph and the reference value of the reliability of the sample node network graph, multiple initial network reliability determination models are trained, including: Model training is performed on a current initial network reliability determination model among the multiple initial network reliability determination models in the following manner: Using the current initial network reliability determination model to process multiple sample node network graphs, multiple sample processing results are obtained; Calculate the sample average error value based on the reference value of the reliability of multiple sample node network graphs and multiple sample processing results; According to the sample average error value, the model parameters of the current initial network reliability determination model are adjusted.

11. The method according to claim 4, characterized in that A preset network reliability determination model that meets the requirements is selected from multiple target network reliability determination models, including: Processing the test node network diagram according to a preset processing rule to determine a reference value of the reliability of the test node network diagram; Utilizing multiple target network reliability determination models to process the test node network diagrams respectively, and determining multiple corresponding test processing results; Calculate the test average error value and test stability based on the reference value of the reliability of the test node network diagram and multiple test processing results; According to the test average error value and the test stability, a preset network reliability determination model that meets the requirements is screened out from multiple target network reliability determination models.

12. A device for determining the reliability of a software network, characterized in that: include: An acquisition module, used for acquiring a target node network graph of a target software network; a processing module, configured to process the target node network graph using a preset network reliability determination model to obtain a corresponding target processing result; wherein the preset network reliability determination model is a neural network model trained according to preset training rules in conjunction with node ratio constraints and network diameter constraints; A determination module is used to determine whether the target software network meets the preset reliability requirements based on the target processing result; The adjustment module is used to adjust the network structure of the target software network when it is determined that the target software network does not meet the preset reliability requirements.

13. A server, characterized in that: The method comprises a processor and a memory for storing processor-executable instructions, wherein the processor implements the steps of the method according to any one of claims 1 to 11 when executing the instructions.

14. A computer-readable storage medium, characterized in that Computer instructions are stored thereon, and when the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 11 are implemented.

15. A computer program product, characterized in that The invention comprises a computer program, which implements the steps of the method according to any one of claims 1 to 11 when the computer program is executed by a processor.

Images