Fault propagation path identification method for ship electronic information system based on public computing environment

Abstract

The present invention relates to a method for identifying fault propagation paths of ship electronic information systems based on a public computing environment, and belongs to the technical field of fault propagation analysis of ship electronic information systems. The method solves the problem of fault propagation path identification in existing ship electronic information systems. The present invention first extracts the resource hardware and functional software of the public computing environment connected to the ship electronic information system to be identified as hardware nodes and software nodes based on the working mechanism of the ship electronic information system, establishes a public resource topology structure model, calculates the hardware node failure probability based on the hardware node entity fault characterization data, calculates the software node failure probability based on the model structure and the hardware node failure probability, and calculates the fault propagation capability of all nodes in the public resource topology structure model; and then uses a particle swarm algorithm to calculate the path with the maximum fault propagation probability. The present invention is suitable for identifying fault propagation paths of ship electronic information systems based on a public computing environment.

Description

Technical Field

[0001] The invention belongs to the technical field of fault propagation analysis of ship electronic information systems. Background Art

[0002] With the continuous improvement of ship informationization, the integration of ship electronic information systems has been significantly improved in both depth and breadth. Horizontally, it involves all functional systems on the ship, and vertically, it covers the entire information chain, including acquisition and control, network transmission, computing and storage. The system architecture is simpler and more integrated. However, while this new flat, resource-sharing architecture significantly reduces the size and burden of electronic information systems and improves the first-time processing rate of information, it also brings difficulties to the fault propagation analysis of ship electronic information systems.

[0003] The public computing environment (CCE) of a ship's electronic information system is an integrated resource hardware and functional software system designed to process, store, analyze, and transmit various electronic information data generated during ship operations. This ship-wide CCE emphasizes resource sharing, providing basic services to support the normal operation of all functional systems on board. It directly participates in the closed-loop operation of all functional systems on board. This computing environment potentially tightly integrates these systems, and failures in some nodes could trigger large-scale fault propagation risks, leading to reduced or even incomplete functionality in other systems. Therefore, an accurate and effective method for identifying fault propagation paths is of great practical significance for the intelligent operation and maintenance of ship CCE-based electronic information systems.

[0004] Fault propagation path identification in shipboard electronic information systems involves evaluating the possible fault propagation paths and the scope of the fault's impact by identifying and analyzing fault information. Domestic and international scholars have conducted extensive research on rotating machinery fault diagnosis methods. In the early days, these methods relied on empirical judgment and manual analysis. With technological advancements, fault diagnosis methods have gradually incorporated more scientific methods and advanced technologies to improve the accuracy and efficiency of fault identification. Currently, model-driven methods have become the mainstream research approach for system fault propagation path identification. First, the system structure and fault information are analyzed to construct a system topology model. Fault propagation analysis is then performed on the topology model to determine the fault propagation path. However, low analysis efficiency and poor accuracy in fault propagation path identification remain. Summary of the Invention

[0005] The present invention aims to solve the problem of fault propagation path identification in existing ship electronic information systems, and provides a method for identifying fault propagation paths in ship electronic information systems based on a public computing environment.

[0006] The method for identifying fault propagation paths of a ship electronic information system based on a public computing environment of the present invention comprises:

[0007] Step 1: Based on the working mechanism of the ship's electronic information system, extract the resource hardware and functional software of the public computing environment connected to the ship's electronic information system to be identified by the fault propagation path, take the resource hardware as the hardware node, and establish a resource hardware subnet based on the actual connection relationship between the resource hardware; take the functional software as the software node, and establish a functional software subnet based on the data transmission relationship between the functional software;

[0008] Then, based on the deployment relationship between resource hardware and functional software, add edges between hardware nodes and software nodes to obtain a public resource topology model;

[0009] Step 2: Collect hardware node entity fault characterization data, and use the range of the characterization data of the fault characterization data when the hardware node entity is operating normally as the normal indicator range of the hardware entity; calculate the failure probability of all hardware nodes by calculating the deviation between the hardware node entity fault characterization data and the normal indicator range;

[0010] Calculate the dependency between software nodes and hardware nodes and the dependency between software nodes based on the connection relationship between software nodes and hardware nodes and the connection relationship between software nodes in the public resource topology model;

[0011] By using the dependencies between software nodes and between software nodes and hardware nodes, the failure probability of software nodes affected by hardware failure in the public resource topology model is calculated.

[0012] Step 3: Using the failure probability of hardware nodes, the failure probability of software nodes affected by the hardware failure state, and the failure propagation characteristics of the nodes, a recursive method is used to calculate the failure propagation capability of all nodes in the public resource topology model.

[0013] By utilizing the fault propagation capabilities of all nodes, the failure probability of hardware nodes, the dependencies between software nodes and hardware nodes, and the dependencies between software nodes, a particle swarm algorithm is used to optimize all fault paths of the faulty nodes and obtain the path with the highest fault propagation probability.

[0014] Furthermore, in the present invention, in step 1, the method for establishing the resource hardware subnet is:

[0015] First, the resource hardware is abstracted as nodes in the network, and the computing speed, storage resources, and network bandwidth resources of the resource hardware are added as attributes of the hardware nodes. According to the actual connection relationship of the resource hardware, edges are added to the hardware nodes to establish a resource hardware subnet.

[0016] Furthermore, in the present invention, in step 1, the method for establishing the functional software subnet is:

[0017] First, the functional software is abstracted into software nodes in the network. The average execution time, required computing resources, storage resources, bandwidth resources and auxiliary data flow diagram (DFD) of the functional software when performing specific tasks are added as attributes of the software nodes. The data flow direction between functional software is abstracted as the connection edge of the network node to establish a functional software subnet.

[0018] Furthermore, in the present invention, in step 2, the failure probability calculation method of the hardware node is:

[0019] Determine the fault type of the hardware node and calculate the fault probability based on the hardware node fault type. Hardware node faults include sudden faults and evolving faults. The probability of sudden faults is 1 or 0. When the fault probability is 1, it indicates that the node is faulty. When the hardware node physical fault characterization data is within the corresponding normal indicator range and the fault probability is 0, it indicates that the node is not faulty.

[0020] The failure probability ε of an evolving fault node is expressed as the deviation between the hardware node entity fault characterization data XP and the normal indicator range XT of the node:

[0021]

[0022] Wherein, XT = [XT-, XT+], XT- is the lower limit value of the node status characterization parameter when the node is normal, and XT+ is the upper limit value of the node status characterization parameter when the node is normal.

[0023] Furthermore, in the present invention, in step 2, the method for calculating the dependency between software nodes is:

[0024] Based on the node calling and called relationships, the dependencies between software nodes are calculated. The dependencies between software nodes are obtained by calculating the weights of the directed network:

[0025]

[0026] Among them, RS ij Indicates the dependency between software nodes, Indicates the probability of calling between software nodes, NC ij is the number of times a software node calls a software node that has a connection relationship with it, ∑NC ij Indicates the number of times all software nodes are called during the entire execution process ∑NC ij ; Indicates the degree of function propagation between software nodes, BNC ij is the number of times a node is called by other nodes. Is the total number of calls during the entire execution process.

[0027] Furthermore, in the present invention, in step 2, the method for calculating the dependency between the software node and the hardware node is:

[0028] The pairwise comparison matrix A is established based on resource C, storage resource M, and network bandwidth N:

[0029]

[0030] Among them, a kl represents the relative importance of criterion k relative to criterion l, where k and l are not equal, and both k and l are equal to 1, 2, or 3; calculate the eigenvector W of the pairwise comparison matrix A, and use the eigenvector W as the dependency between the software node and the hardware node:

[0031] A×W=λ max ×W

[0032] Where λ max is the maximum eigenvalue of the pairwise comparison matrix A.

[0033] Furthermore, in the present invention, in step 2, the method for calculating the failure probability of the software nodes affected by the hardware failure state in the public resource topology model is:

[0034] First, according to the topology structure of the public resource topology structure model, a fault propagation mode is determined, wherein the fault propagation mode includes a point-to-point fault propagation mode and a multi-point fault propagation mode;

[0035] In the point-to-point fault propagation mode, the probability that the faulty node A propagates the fault state to node B is:

[0036] ε B =ε A RS AB PC A

[0037] ε A is the failure probability of faulty node A, RS AB represents the failure probability ε A The actual dependence between nodes A and B and PC A Indicates the fault propagation capability of faulty node A;

[0038] The multi-point fault propagation modes include: one-transmission-multiple-path mode and multiple-transmission-one-path mode;

[0039] In the one-to-multi-path mode, the probability of a one-to-three fault propagation is explained as follows:

[0040] The failure of hardware node A propagates to nodes B, C, and D, causing the status of nodes B, C, and D to fail. The failure probabilities of nodes B, C, and D are:

[0041] ε B =ε A RS AB PC A

[0042] ε C =ε A RS AC PC A

[0043] ε D =ε A RS AD PC A

[0044] Among them, T B 、T C 、T D Represent the tolerance of node B, node C and node D respectively; RS AB Indicates the dependency of node A on node B, RS AC Indicates the dependency of node A on node C, RS AD Indicates the dependency of node A on node D, PC A Indicates the fault propagation capability of node A;

[0045] The multi-transmission-one-path mode also includes the AND-related fault propagation mode and the OR-related fault propagation mode. The three-transmission-one-path mode is used to illustrate the following:

[0046] When the fault propagation relationship is established: the failure probability of node A is:

[0047] ε A =ε B RS BA PC B +ε C RS CA PC C +ε D RS DA PC D

[0048] Or when the relationship fault propagates: the failure probability of node A is:

[0049] ε A =max{ε B-A ,ε C-A ,ε D-A}

[0050] =max{ε B RS BA PC B +ε C RS CA PCC +ε D RS DA PC D}

[0051] RS BA 、RS CA and RS DA They represent the dependency of node B on node A, the dependency of node C on node A, and the dependency of node D on node A, respectively. B Indicates the fault propagation capability of node B, PC C represents the fault propagation capability of node C, PC D represents the fault propagation capability of node D, ε B-A represents the probability that node B will propagate the fault to node A, ε C-A represents the failure probability of node C propagating the failure to node A, ε D-A represents the failure probability of node D propagating the failure to node A.

[0052] Furthermore, in the present invention, in step 3, the method of calculating the fault propagation capability of all nodes in the public resource topology model includes: calculating the fault propagation capability of hardware nodes and the fault propagation capability of software nodes;

[0053] The calculation method of the hardware node's failure resistance adopts the PageRank algorithm. The specific formula is:

[0054]

[0055] Among them, PC(v yi ) represents the hardware node v yi The fault propagation capability attribute value, M(v yi ) indicates pointing to node v yi The node set, L(v yj ) represents node v yj The number of directed edges connected;

[0056] The calculation formula for the fault propagation capability of a software node is:

[0057] PC(u)=aEIC(u)+bIC(u)

[0058] Where a represents the weight coefficient of the node's external call capability, b represents the weight coefficient of the node's call capability, and the call capability of software node u is IC(u):

[0059]

[0060] in, is the in-degree of node u, is the maximum in-degree value in the software sub-network, m is the number of nodes in the software sub-network, Represents the calling capability of the target node u itself, ν ri1 is the i1th node in the direct in-degree neighbor set DINS of the objective function u, ν rj1 is the j1th node in the direct in-degree neighbor set DINS of the target node u, Nin represents the number of nodes in the direct in-degree neighbor set of the target node u, Indicates the probability that the target node in the software sub-network is called by the adjacent node; Weight (ν ri1 ,u) is the software node ν ri1 The number of calls to the target node u, Weight(ν ri1 ,ν rj1 ) is the software node ν ri1 For software node ν rj1 The number of calls;

[0061] External call capability EIC(u) of node u:

[0062]

[0063] Among them, α represents the external call capability of the target node itself (0≤α≤1), v ri2 is a node in the direct out-degree set DONS of the target node u, Nout represents the number of nodes in the direct out-degree set of the target node u, Indicates the probability that the target node calls other neighbor nodes in the weighted software execution network, Weight(u,v ri2 ) represents the target node u to the software node v ri2 The number of calls; Weight(v rj2 ,v ri2 ) represents the software node v rj2 For software node v ri2 The number of calls.

[0064] Furthermore, in the present invention, in step 3, the method for obtaining the path with the highest probability of fault propagation is:

[0065] Step 3: 1. Use the faulty hardware node as the initial position of the particle swarm. The path of each particle represents a possible fault propagation path, and the particle speed represents the adjustment direction and step size of the fault propagation path. Randomly initialize the particle swarm algorithm.

[0066] Step 32: Calculate the fitness value of each particle;

[0067] Step 33: Determine whether the fitness value of the currently obtained particle is greater than the optimal fitness value of the current particle at the historical moment. If so, use the fitness value at the current moment as the optimal fitness value, and update the position and speed of the particle corresponding to the optimal fitness value. Otherwise, return to step 32 and continue to calculate the fitness value of the current particle until the number of iterations reaches the threshold; output the path of the particle with the optimal fitness value.

[0068] Furthermore, in the present invention, in step 32, the method for calculating the fitness value of each particle is:

[0069]

[0070] weight i,i+1 =ε i RS i,i+1 PC i

[0071] Among them: fitness is the fitness value of the particle, n is the number of nodes in the path, weight i,i+1 is the weight of the edge between node i and node i+1, ε i is the failure probability of node i, RS i,i+1 is the actual degree of dependence between node i and node i+1 under the failure probability, PC i The fault propagation capability of node A.

[0072] Furthermore, in the present invention, in step 33, the formula for iteratively updating the position and velocity of the particle corresponding to the optimal fitness value is:

[0073]

[0074]

[0075] Where: is the velocity of the particle after the next particle iteration; w is the inertia weight; is the current particle speed; is the position of the particle after the next iteration; is the current position of the particle; c1 and c2 are the learning factors of the particle, c1 is the speed after the next iteration affected by its own optimal speed, c2 is the speed after the next iteration affected by the global optimal position, and both c1 and c2 are numbers greater than 0; ξ and η are random numbers, and their values ​​are in [0,1]; is the optimal position coordinate of the current particle; is the optimal position coordinate of the particle after iteration.

[0076] This paper uses the deployment topology and execution call chain information of a ship's electronic information system during runtime to construct a topological model of the system. Based on this, it fully utilizes system fault information to analyze fault propagation and uses a heuristic algorithm to search for fault propagation paths. This provides powerful support for the maintenance and optimization of shipboard electronic information systems in a functional computing environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 A flow chart of the method of the present invention;

[0078] Figure 2 Build processes for topology models;

[0079] Figure 3 This is the fault propagation mechanism diagram from node A to node B;

[0080] Figure 4 Flowchart for node failure probability calculation;

[0081] Figure 5 This is the relationship diagram between the failure probability and node tolerance of evolving failures;

[0082] Figure 6 This is the relationship between the failure probability of sudden failure and node tolerance;

[0083] Figure 7 Schematic diagram of the multi-point fault propagation process;

[0084] Figure 8 It is a multi-transmission to multi-point fault propagation process. DETAILED DESCRIPTION

[0085] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other in the absence of conflict.

[0086] Specific embodiment 1: This embodiment is described in detail with reference to the drawings. The method for identifying fault propagation paths of a ship electronic information system based on a public computing environment described in this embodiment includes:

[0087] Step 1: Based on the working mechanism of the ship's electronic information system, extract the resource hardware and functional software of the public computing environment connected to the ship's electronic information system to be identified by the fault propagation path, take the resource hardware as the hardware node, and establish a resource hardware subnet based on the actual connection relationship between the resource hardware; take the functional software as the software node, and establish a functional software subnet based on the data transmission relationship between the functional software;

[0088] Then, based on the deployment relationship between resource hardware and functional software, add edges between hardware nodes and software nodes to obtain a public resource topology model;

[0089] Step 2: Collect hardware node entity fault characterization data, and use the range of the characterization data of the fault characterization data when the hardware node entity is operating normally as the normal indicator range of the hardware entity; calculate the failure probability of all hardware nodes by calculating the deviation between the hardware node entity fault characterization data and the normal indicator range;

[0090] Calculate the dependency between software nodes and hardware nodes and the dependency between software nodes based on the connection relationship between software nodes and hardware nodes and the connection relationship between software nodes in the public resource topology model;

[0091] By using the dependencies between software nodes and between software nodes and hardware nodes, the failure probability of software nodes affected by hardware failure in the public resource topology model is calculated.

[0092] Step 3: Using the failure probability of hardware nodes, the failure probability of software nodes affected by the hardware failure state, and the failure propagation characteristics of the nodes, a recursive method is used to calculate the failure propagation capability of all nodes in the public resource topology model.

[0093] By utilizing the fault propagation capabilities of all nodes, the failure probability of hardware nodes, the dependencies between software nodes and hardware nodes, and the dependencies between software nodes, a particle swarm algorithm is used to optimize all fault paths of the faulty nodes and obtain the path with the highest fault propagation probability.

[0094] Furthermore, in the present invention, in step 1, the method for establishing the resource hardware subnet is:

[0095] First, the resource hardware is abstracted as nodes in the network, and the computing speed, storage resources, and network bandwidth resources of the resource hardware are added as attributes of the hardware nodes. According to the actual connection relationship of the resource hardware, edges are added to the hardware nodes to establish a resource hardware subnet.

[0096] Furthermore, in the present invention, in step 1, the method for establishing the functional software subnet is:

[0097] First, the functional software is abstracted into software nodes in the network. The average execution time, required computing resources, storage resources, bandwidth resources and auxiliary data flow diagram (DFD) of the functional software when performing specific tasks are added as attributes of the software nodes. The data flow direction between functional software is abstracted as the connection edge of the network node to establish a functional software subnet.

[0098] Furthermore, in the present invention, in step 2, the failure probability calculation method of the hardware node is:

[0099] Determine the fault type of the hardware node. Hardware node faults include sudden faults and evolving faults. The probability of sudden faults is 1 or 0. When the fault probability is 1, it indicates that the node is faulty. When the hardware node physical fault characterization data is within the corresponding normal indicator range and the fault probability is 0, it indicates that the node is not faulty.

[0100] The failure probability ε of an evolving fault node is expressed as the deviation between the hardware node entity fault characterization data XP and the normal indicator range XT of the node:

[0101]

[0102] Wherein, XT = [XT-, XT+], XT- is the lower limit value of the node status characterization parameter when the node is normal, and XT+ is the upper limit value of the node status characterization parameter when the node is normal.

[0103] Furthermore, in the present invention, in step 2, the method for calculating the dependency between software nodes is:

[0104] Based on the node calling and called relationships, the dependencies between software nodes are calculated. The dependencies between software nodes are obtained by calculating the weights of the directed network:

[0105]

[0106] Among them, RS ij Indicates the dependency between software nodes, Indicates the probability of calling between software nodes, NC ij is the number of times a software node calls a software node that has a connection relationship with it, ∑NC ij Indicates the number of times all software nodes are called during the entire execution process ∑NC ij ; Indicates the degree of function propagation between software nodes, BNC ij is the number of times a node is called by other nodes. Is the total number of calls during the entire execution process.

[0107] Furthermore, in the present invention, in step 2, the method for calculating the dependency between the software node and the hardware node is:

[0108] The pairwise comparison matrix A is established based on resource C, storage resource M, and network bandwidth N:

[0109]

[0110] Among them, a kl represents the relative importance of criterion k relative to criterion l, where k and l are not equal, and both k and l are equal to 1, 2, or 3; calculate the eigenvector W of the pairwise comparison matrix A, and use the eigenvector W as the dependency between the software node and the hardware node:

[0111] A×W=λ max ×W

[0112] Where λ max is the maximum eigenvalue of the pairwise comparison matrix A.

[0113] In this embodiment, the consistency index of the pairwise comparison matrix A is calculated. Finally, the consistency index CI (Consistency Index) is calculated. Where n is the number of criteria. Calculate the consistency ratio CR (Consistency Ratio): Where RI is the random consistency index, which depends on the size of the pairwise comparison matrix. For a 3x3 matrix, RI = 0.58. If CR < 0.1, the consistency of the pairwise comparison matrix is ​​considered acceptable.

[0114] Furthermore, in the present invention, in step 2, the method for calculating the failure probability of the software nodes affected by the hardware failure state in the public resource topology model is:

[0115] First, according to the topology structure of the public resource topology structure model, a fault propagation mode is determined, wherein the fault propagation mode includes a point-to-point fault propagation mode and a multi-point fault propagation mode;

[0116] In the point-to-point fault propagation mode, the probability that the faulty node A propagates the fault state to node B is:

[0117] ε B =ε A RS AB PC A

[0118] ε A is the failure probability of faulty node A, RS AB represents the failure probability ε A The actual dependence between nodes A and B and PCA Indicates the fault propagation capability of faulty node A;

[0119] The multi-point fault propagation mode includes two modes: one-transmission-multiple-path mode and multiple-transmission-one-path mode;

[0120] In the one-to-multi-path mode, the probability of a one-to-three fault propagation is explained as follows:

[0121] The failure of hardware node A propagates to nodes B, C, and D, causing the status of nodes B, C, and D to fail. The failure probabilities of nodes B, C, and D are:

[0122] ε B =ε A RS AB PC A

[0123] ε C =ε A RS AC PC A

[0124] ε D =ε A RS AD PC A

[0125] Among them, T B 、T C 、T D Represent the tolerance of node B, node C and node D respectively; RS AB Indicates the dependency of node A on node B, RS AC Indicates the dependency of node A on node C, RS AD Indicates the dependency of node A on node D, PC A Indicates the fault propagation capability of node A;

[0126] The multi-transmission-one-path mode also includes the AND-related fault propagation mode and the OR-related fault propagation mode. The one-to-three fault propagation mode is used as an example to illustrate:

[0127] When the fault propagation relationship is established: the failure probability of node A is:

[0128] ε A =ε B RS BA PC B +ε C RS CA PC C +ε D RS DA PC D

[0129] Or when the relationship fault propagates: the failure probability of node A is:

[0130] ε A =max{ε B-A ,ε C-A ,ε D-A}

[0131] =max{ε B RS BA PC B +ε C RS CA PC C +ε D RS DA PC D}

[0132] RS BA 、RS CA and RS DA They represent the dependency of node B on node A, the dependency of node C on node A, and the dependency of node D on node A, respectively. B Indicates the fault propagation capability of node B, PC C represents the fault propagation capability of node C, PC D represents the fault propagation capability of node D, ε B-A represents the probability that node B will propagate the fault to node A, ε C-A represents the failure probability of node C propagating the failure to node A, ε D-A represents the failure probability of node D propagating the failure to node A.

[0133] Furthermore, in the present invention, in step 3, the method of calculating the fault propagation capability of all nodes in the public resource topology model includes: calculating the fault propagation capability of hardware nodes and the fault propagation capability of software nodes;

[0134] The calculation method of the hardware node's failure resistance adopts the PageRank algorithm. The specific formula is:

[0135]

[0136] Among them, PC(v yi ) represents the hardware node v yi The fault propagation capability attribute value, M(v yi ) indicates pointing to node v yi The node set, L(v yj ) represents node v yj The number of directed edges connected;

[0137] The calculation formula for the fault propagation capability of a software node is:

[0138] PC(u)=aEIC(u)+bIC(u)

[0139] Where a represents the weight coefficient of the node's external call capability, b represents the weight coefficient of the node's call capability, and the call capability of software node u is IC(u):

[0140]

[0141] in, is the in-degree of node u, is the maximum in-degree value in the software sub-network, m is the number of nodes in the software sub-network, Represents the calling capability of the target node u itself, ν ri1 is the i1th node in the direct in-degree neighbor set DINS of the objective function u, ν rj1 is the j1th node in the direct in-degree neighbor set DINS of the target node u, Nin represents the number of nodes in the direct in-degree neighbor set of the target node u, Indicates the probability that the target node in the software sub-network is called by the adjacent node; Weight (ν ri1 ,u) is the software node ν rj1 The number of calls to the target node u, Weight(ν ri1 ,ν rj1 ) is the software node ν ri1 For software node ν rj1 The number of calls;

[0142] External call capability EIC(u) of node u:

[0143]

[0144] Among them, α represents the external call capability of the target node itself (0≤α≤1), v ri2 is a node in the direct out-degree set DONS of the target node u, Nout represents the number of nodes in the direct out-degree set of the target node u, Indicates the probability that the target node calls other neighbor nodes in the weighted software execution network, Weight(u,v ri2 ) represents the target node u to the software node v ri2 The number of calls; Weight(v rj2 ,v ri2 ) represents the software node v rj2 For software node v ri2 The number of calls.

[0145] Furthermore, in the present invention, in step 3, the method for obtaining the path with the highest probability of fault propagation is:

[0146] Step 3: 1. Use the faulty hardware node as the initial position of the particle swarm. The path of each particle represents a possible fault propagation path, and the particle speed represents the adjustment direction and step size of the fault propagation path. Randomly initialize the particle swarm algorithm.

[0147] Step 32: Calculate the fitness value of each particle;

[0148] Step 33: Determine whether the fitness value of the currently obtained particle is greater than the optimal fitness value of the current particle at the historical moment. If so, use the fitness value at the current moment as the optimal fitness value, and update the position and speed of the particle corresponding to the optimal fitness value. Otherwise, return to step 32 and continue to calculate the fitness value of the current particle until the number of iterations reaches the threshold; output the path of the particle with the optimal fitness value.

[0149] Furthermore, in the present invention, in step 32, the method for calculating the fitness value of each particle is:

[0150]

[0151] weight i,i+1 =ε i RS i,i+1 PC i

[0152] Where: fitness is the fitness value of the particle, n is the number of nodes in the path. weight i,i+1 is the weight of the edge between node i and node i+1, ε i is the failure probability of node i, RS i,i+1 is the actual degree of dependence between node i and node i+1 under the failure probability, PC i The fault propagation capability of node A.

[0153] Furthermore, in the present invention, in step 33, the formula for iteratively updating the position and velocity of the particle corresponding to the optimal fitness value is:

[0154]

[0155] Where: is the velocity of the particle after the next particle iteration; w is the inertia weight; is the current particle speed; is the position of the particle after the next iteration; is the current position of the particle; c1 and c2 are the learning factors of the particle, c1 is the speed after the next iteration affected by its own optimal speed, c2 is the speed after the next iteration affected by the global optimal position, and both c1 and c2 are numbers greater than 0; ξ and η are random numbers, and their values ​​are in [0,1]; is the optimal position coordinate of the current particle; is the optimal position coordinate of the particle after iteration.

[0156] The specific process of the present invention in practical application is as follows:

[0157] 1. Topology Model Construction

[0158] The process of constructing the topology model of the ship electronic information system in the public computing environment is as follows: Figure 1 As shown, it mainly includes four steps: extraction of public computing environment resource hardware and resource software, construction of resource hardware subnet, construction of functional software subnet, and construction of topology model.

[0159] Each link specifically includes the following implementation:

[0160] (1) Extraction of public computing environment resource hardware and resource software.

[0161] The public computing environment is an integrated resource hardware and functional software system. Before building the topology model, it is necessary to first extract the resource hardware and resource software in the system from top to bottom according to the working mechanism of the ship's electronic information system to facilitate subsequent modeling.

[0162] (2) Resource hardware subnet construction:

[0163] The specific steps are as follows: first, abstract the resource hardware into nodes in the network, add attributes to the nodes based on the resource hardware's computing speed, storage resources, network bandwidth and other resources, such as C1 = <54TFLOPS, 42GB, 36Gbps>, and add edges to the nodes based on the actual connection relationship of the resource hardware, thereby completing the construction of the resource hardware subnet.

[0164] (3) Functional software node construction:

[0165] The specific steps are as follows: first, abstract the functional software into nodes in the network, and add attributes such as average execution time, required computing resources, storage resources, and bandwidth resources to the functional software nodes, such as P1 = <100ms, 10TFLOPS, 5GB, 36Gbps>. Analyze the data flow diagram (DFD) of the functional software when the system performs specific tasks, abstract the data flow direction between the functional software into the edges of the network nodes, and then complete the construction of the functional software subnet.

[0166] (4) Topology model construction:

[0167] Based on the previous steps, the functional software subnet and resource hardware subnet of the ship's electronic information system public computing environment have been constructed. Next, we need to analyze the actual deployment relationship between the functional software and resource hardware and add edges between the functional software nodes and the resource hardware nodes.

[0168] After adding the dependency edge between the resource node and the functional software node, the construction of the topology model of the ship electronic information system can be completed. The model construction process is as follows: Figure 2 As shown, the dotted line represents the dependency relationship between resource hardware and functional software.

[0169] 2. Analysis of fault propagation process:

[0170] After constructing the topology model of the ship's electronic information system, the next step is to analyze the fault propagation theory of the functional software nodes and resource hardware nodes within the topology model, providing a theoretical basis for applying the fault propagation path identification solution to the topology model. This specifically consists of two parts: point-to-point fault propagation process analysis and multi-point fault propagation process analysis. The fault propagation analysis specifically includes the following implementation steps:

[0171] (1) Analysis of point-to-point fault propagation process:

[0172] When studying the quantitative relationship of the point-to-point fault propagation process, the following can be used: Figure 3 The algorithm flow shown is used to analyze whether the fault state at node A is propagated to node B. Whether the fault state at node A can be propagated to node B is related to three factors: the fault probability ε of node A A , the actual dependence degree RS between the two nodes under the failure probability AB and the fault propagation capability PC of node A A Then the failure probability of node B is ε B =ε A RS AB PC A .

[0173] The following describes the calculation process of three factors: node failure probability, actual dependence between nodes, and node fault propagation capability.

[0174] (1-1) Node Failure Probability Calculation Process

[0175] The node failure probability calculation process is as follows: Figure 4As shown, the first step is to select the node's characterization parameters. The node status characterization parameters select the relevant signal parameters of the output terminal of the device entity corresponding to the node. If the node's status can be determined based on these signal parameters, they are called node status characterization parameters. Next, determine the node's normality indicator, the node normality indicator XT: The node normality indicator refers to the indicator that indicates the node is within the normal state range under ideal operating conditions. This can be evaluated using the node status characterization parameters. The node normality indicator is expressed as XT = [XT-, XT+], where XT- is the lower limit of the node status characterization parameter when the node is normal, and XT+ is the upper limit of the node status characterization parameter when the node is normal.

[0176] The faults in the ship's electronic information system include evolutionary faults and sudden faults. The relationship between these two types of faults and node fault tolerance is as follows: Figure 5 and Figure 6 As shown, the node tolerance refers to the node's tolerance for fault intensity. According to the calculation process of the node failure probability, the node tolerance also refers to the degree to which the node's actual state characterization parameter value deviates from the node's normal index that the node can tolerate. Among them, the failure probability of a sudden failure is 1 when it exceeds the fault tolerance, and the failure probability is 0 when it is less than the tolerance. For the failure probability ε of an evolving fault node: the failure probability of a node refers to the deviation between the actual node state characterization parameter value XP of the node and the node's normal index XT when the node is working under actual working conditions.

[0177]

[0178] (1-2) Node Dependency

[0179] Node dependency includes two parts: the calculation of node dependency between functional software and the calculation of dependency between functional software and resource hardware.

[0180] (1-2-1) Node dependency calculation between functional software nodes

[0181] The association relationship between nodes is analyzed simultaneously from the perspective of node calling and being called, and the degree of node dependence is measured as the weight of the directed network.

[0182]

[0183] The formula consists of two parts. The first part represents the probability of calling between nodes, which is the number of times a node calls other nodes NC ij The number of calls to all nodes during the entire execution process ∑NC ij The second part represents the degree of function propagation between nodes, which is the number of times the node is called by other nodes. ijThe total number of calls during the entire execution process ∑BNC ij ratio.

[0184] (1-2-2) Dependency between functional software nodes and resource nodes

[0185] In a public computing environment, the computing resources, storage resources, and network bandwidth resources occupied by functional software will affect the software's execution efficiency and execution time. The Analytic Hierarchy Process (AHP) is used to evaluate the impact of computing resources, storage resources, network bandwidth, and other resources on software functionality. At the criteria level, the comparison matrix A of computing resources C, storage resources M, and network bandwidth N is shown below:

[0186]

[0187] Among them, a ij Indicates the relative importance of criterion i relative to criterion j.

[0188] First, calculate the eigenvector W and eigenvalue λ of the pairwise comparison matrix A max is the largest eigenvalue of the matrix A. The weight vector W needs to satisfy the consistency condition, that is, A×W=λ max × W. Whether the feature vector or the weight vector is a vector;

[0189] In actual application, the present invention also includes the calculation of the consistency index and consistency ratio of the pairwise comparison matrix A. When the consistency index and consistency ratio meet the consistency requirements, continue to execute the following steps. Otherwise, it is necessary to re-evaluate the impact of computing resources, storage resources, network bandwidth and other resources on the software function.

[0190] The specific calculation method of consistency index and consistency ratio is:

[0191] Calculate the consistency index CI (Consistency Index) Where n is the number of criteria.

[0192] Calculate the consistency ratio CR (Consistency Ratio): Where RI is the random consistency index, which depends on the size of the pairwise comparison matrix. For a 3x3 matrix, RI = 0.58. If CR < 0.1, the consistency of the pairwise comparison matrix is ​​considered acceptable.

[0193] (1-3) Calculation of node propagation capability

[0194] First, calculate the fault probability of a node FP: Based on the fault accumulation characteristics of the node, use the recursive method to define the fault probability FP of node u as shown in the following formula:

[0195] External call capability EIC(u) of node u:

[0196]

[0197] Among them, α represents the external call capability of the target node itself (0≤α≤1), v ri2 is a node in the direct out-degree set DONS of the target node u, Nout represents the number of nodes in the direct out-degree set of the target node u, Indicates the probability that the target node calls other neighbor nodes in the weighted software execution network, Weight(u,v ri2 ) represents the target node u to the software node v ri2 The number of calls; Weight(v rj2 ,v ri2 ) represents the software node v rj2 For software node v ri2 The number of calls.

[0198] According to the fault propagation characteristics of the node, using the recursive method, the software node u calls the capability IC(u):

[0199]

[0200] in, is the in-degree of node u, is the maximum in-degree value in the software sub-network, m is the number of nodes in the software sub-network, Represents the calling capability of the target node u itself, ν ri1 is the i1th node in the direct in-degree neighbor set DINS of the objective function u, ν rj1 is the j1th node in the direct in-degree neighbor set DINS of the target node u, Nin represents the number of nodes in the direct in-degree neighbor set of the target node u, Indicates the probability that the target node in the software sub-network is called by the adjacent node; Weight (ν ri1 ,u) is the software node ν ri1 The number of calls to the target node u, Weight(ν ri1 ,ν rj1 ) is the software node ν ri1 For software node ν rj1 The number of calls;

[0201] From the perspective of the active caller, if a node calls a faulty node, it will be affected by the fault. From the perspective of the called node, if a fault exists in the node, it will propagate the fault to the calling node, resulting in a "fault accumulation" effect and a "fault ripple" effect. For each information domain software / hardware node, the fault propagation capability of the software node is constructed by comprehensively considering the fault probability (FP) and the fault propagation capability (FPC).

[0202] PC(u)=aEIC(u)+bIC(u)

[0203] Wherein, a represents the weight coefficient of the fault possibility, b represents the weight coefficient of the fault propagation capability, and a / b is preferably 0.5.

[0204] Taking into account the evaluation indicators of resource hardware node fault propagation capability, the PageRank algorithm is used to evaluate and analyze the fault propagation capability of hardware nodes;

[0205]

[0206] Among them, PC(v yi ) represents the hardware node v yi The fault propagation capability attribute value, M(v yi ) indicates pointing to node v yi The node set, L(v yj ) represents node v yj The number of directed edges connected;

[0207] (2) Analysis of multi-point fault propagation process

[0208] The multi-point fault propagation path mode is divided into "one-to-multiple-path mode" and "multiple-to-one-path mode". The three-to-one and one-to-three modes are used as examples. Figure 7 and Figure 8 shown.

[0209] In the one-pass multi-path mode, the failure probabilities of nodes B, C, and D are: B =ε A RS AB PC A , ε C =ε A RS AC PC A , ε D =ε A RS AD PC A By judging ε B , ε C , ε D With T B 、T C 、TD The relationship between them can be used to determine the propagation of the fault of node A to nodes B, C, and D.

[0210] In the multi-transmission-one-path model, there are two types depending on the actual situation: 1. "AND relationship". This means that only when nodes B, C, and D all fail can the fault state be propagated to node A, causing node A to fail. This "AND relationship" can be described as A = B ∩ C ∩ D. 2. "OR relationship". This means that when any node among B, C, and D fails, the fault state can be propagated to node A, causing node A to fail. This "OR relationship" can be described as A = B ∪ C ∪ D. In the "AND relationship", the failure probability of node A is:

[0211] ε A =ε B RS BA PC B +ε C RS CA PC C +ε D RS DA PC D

[0212] In the “OR relationship”, the failure probability of node A is:

[0213] ε A =max{ε B-A ,ε C-A ,ε D-A}

[0214] =max{ε B ? RS BA PC B +ε C RS CA PC C +ε D RS DA PC D}

[0215] 3. Fault Propagation Path Identification Algorithm

[0216] After completing the construction of the topological model of the ship's electronic information system in the previous steps and analyzing the fault propagation mode between point-to-point and multi-point in the topological model, the strength value of each fault propagation path can be calculated according to the process. In order to accelerate the search for the fault propagation path with the highest probability, the present invention adopts a heuristic algorithm to solve this problem.

[0217] The particle swarm algorithm searches for the optimal value in space based on the properties of the particles themselves (speed and position), and uses the optimal value found as the individual extreme value. Then, the optimal values ​​of all particles in this space are compared, and the optimal individual extreme value is the optimal solution. The main steps are as follows:

[0218] (1) Initialization settings, set basic parameters, determine the size and dimension of the particle swarm, and set the total number of operation steps. In addition, during the continuous iteration of particles, the position of a particle is constantly changing. After the iteration, the coordinates of the optimal position of a particle are The optimal position coordinates of all particles in the particle swarm after iteration are In addition, the particle velocity ν id To be restricted, that is, ν id ∈[ν min ,ν max ](ν min is the lower limit of particle velocity, ν max is the upper limit of particle velocity).

[0219] (2) Randomly generate an initial particle swarm, where each particle represents a possible fault propagation path. The particle position represents the path, and the velocity represents the adjustment direction and step size of the path. Set the particle swarm in the particle swarm algorithm, and randomly set the positions of N particles in the particle swarm, with their values ​​randomly set to 0 or 1. The smallest individual extreme value is used as the initial global extreme value. Generally, the particle velocity value v i The initial value range is set to [-4,4].

[0220] (3) Fitness calculation: Calculate the fitness value of each particle, which is determined by the objective function.

[0221]

[0222] weight i,i+1 =ε i RS i,i+1 PC i

[0223] Where: n is the number of nodes in the path. weight i,i+1 is the weight of the edge between node i and node i+1, ε i is the failure probability of node i, RS i,i+1 is the actual degree of dependence between node i and node i+1 under the failure probability, PC i The fault propagation capability of node A.

[0224] (4) In step (3), the fitness value of each particle is obtained and compared with its own historical optimal value. If the new value is better than the historical optimal value, the new fitness value will replace the previous value and become the new optimal value of the particle. The original optimal position of the particle is replaced by the position of the particle.

[0225] (5) Iterate according to the following formula to update the position and velocity of the particle.

[0226]

[0227] Where: is the velocity of the particle after the next particle iteration; w is the inertia weight; is the current particle speed; is the position of the particle after the next iteration; is the current position of the particle; c1 and c2 are the learning factors of the particle, c1 is the speed after the next iteration affected by its own optimal speed, c2 is the speed after the next iteration affected by the global optimal position, and both c1 and c2 are numbers greater than 0; ξ and η are random numbers, and their values ​​are in [0,1]; is the optimal position coordinate of the current particle; is the optimal position coordinate of the particle after iteration.

[0228] (6) Until the preset maximum number of iterations is reached, the final optimal path is output and the algorithm ends.

[0229] The present invention has the following advantages:

[0230] (1) A hierarchical network is used to construct a topological model of the public computing environment of the ship's electronic information system, which is an integrated resource hardware and functional software system. This hierarchical model not only makes the relationship and dependency between each functional module and hardware resources more intuitive and clear, but also provides an analytical basis for subsequent fault propagation analysis.

[0231] (2) When analyzing the fault propagation process, starting from the three parameters of the dependency between nodes, the probability of node failure, and the fault propagation intensity of nodes in the topological structure, the fault information of the system can be effectively utilized. By combining these three parameters, a refined fault propagation model can be established, which improves the accuracy of fault propagation path identification.

[0232] (3) The use of heuristic algorithms to search for fault propagation paths greatly improves the efficiency of searching for fault propagation paths in shipboard electronic information systems. By utilizing heuristic functions and empirical rules, heuristic algorithms can quickly find potential fault propagation paths in complex systems, significantly reducing the search space and computing time. This method, combined with the topological model of the hierarchical network, can more efficiently locate the source of the fault and the propagation path, and quickly identify the affected functional modules and resource hardware.

[0233] (4) A method for identifying fault propagation paths in ship electronic information systems based on a public computing environment is proposed. This method greatly improves the speed and accuracy of fault detection and response, enabling the system to take appropriate isolation and repair measures in the shortest time and reduce the impact of faults on system operation.

[0234] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A method for identifying fault propagation paths of ship electronic information systems based on a public computing environment, characterized in that: include: Step 1: Based on the working mechanism of the ship's electronic information system, extract the resource hardware and functional software of the public computing environment connected to the ship's electronic information system whose fault propagation path is to be identified, use the resource hardware as a hardware node, and establish a resource hardware subnet based on the actual connection relationship between the resource hardware; Take the functional software as the software node and establish the functional software subnet according to the data transmission relationship between the functional software; Then, based on the deployment relationship between resource hardware and functional software, add edges between hardware nodes and software nodes to obtain a public resource topology model; Step 2: Collect hardware node entity fault characterization data, and use the range of the characterization data of the fault characterization data when the hardware node entity is operating normally as the normal indicator range of the hardware node entity; calculate the failure probability of all hardware nodes by calculating the deviation between the hardware node entity fault characterization data and the normal indicator range; Calculate the dependency between software nodes and hardware nodes and the dependency between software nodes based on the connection relationship between software nodes and hardware nodes and the connection relationship between software nodes in the public resource topology model; By using the dependencies between software nodes and between software nodes and hardware nodes, the failure probability of software nodes affected by hardware failure in the public resource topology model is calculated. Step 3: Using the failure probability of hardware nodes, the failure probability of software nodes affected by the hardware failure state, and the failure propagation characteristics of the nodes, a recursive method is used to calculate the failure propagation capability of all nodes in the public resource topology model. By utilizing the fault propagation capabilities of all nodes, the failure probability of hardware nodes, the dependencies between software nodes and hardware nodes, and the dependencies between software nodes, a particle swarm algorithm is used to optimize all fault paths of the faulty nodes and obtain the path with the highest fault propagation probability.

2. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 1, characterized in that: In step 1, the method for establishing the resource hardware subnet is: First, the resource hardware is abstracted as nodes in the network, and the computing speed, storage resources, and network bandwidth resources of the resource hardware are added as attributes of the hardware nodes. According to the actual connection relationship of the resource hardware, edges are added to the hardware nodes to establish a resource hardware subnet.

3. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 1 or 2, characterized in that: In step 1, the method for establishing the functional software subnet is: First, the functional software is abstracted into software nodes in the network. The average execution time, required computing resources, storage resources, bandwidth resources and auxiliary data flow diagram (DFD) of the functional software when performing specific tasks are added as attributes of the software nodes. The data flow direction between functional software is abstracted as the connection edge of the network node to establish a functional software subnet.

4. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 3, characterized in that: In step 2, the failure probability calculation method of the hardware node is: Determine the fault type of the hardware node and calculate the fault probability based on the hardware node fault type. Hardware node faults include sudden faults and evolving faults. The probability of sudden faults is 1 or 0. When the fault probability is 1, it indicates that the node is faulty. When the hardware node physical fault characterization data is within the corresponding normal indicator range and the fault probability is 0, it indicates that the node is not faulty. The failure probability ε of an evolving faulty node is expressed as the deviation between the hardware node entity fault characterization data XP and the normal indicator range XT of the hardware node entity: Wherein, XT = [XT-, XT+], XT- is the lower limit value of the node status characterization parameter when the hardware node is normal, and XT+ is the upper limit value of the node status characterization parameter when the hardware node is normal.

5. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 4, characterized in that: In step 2, the method for calculating the dependency between software nodes is: Based on the calling and called relationships of software nodes, the dependencies between software nodes are calculated. The dependencies between software nodes are obtained by calculating the weights of the directed network: Among them, RS ij Indicates the dependency between software nodes, Indicates the probability of calling between software nodes, NC ij is the number of times a software node calls a software node that has a connection relationship with it, ∑NC ij Indicates the number of times all software nodes are called during the entire execution process ∑NC ij ; Indicates the degree of function propagation between software nodes, BNC ij is the number of times a software node is called by other nodes, ∑BNC ij Is the total number of calls during the entire execution process.

6. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 4, characterized in that: In step 2, the method for calculating the dependency between software nodes and hardware nodes is: Create a pairwise comparison matrix A: Among them, a kl represents the relative importance of criterion k relative to criterion l, where k and l are not equal, and both k and l are equal to 1, 2, or 3; calculate the eigenvector W of the pairwise comparison matrix A, and use the eigenvector W as the dependency between the software node and the hardware node: A×W=λ max ×W Where λ max is the maximum eigenvalue of the pairwise comparison matrix A.

7. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 5 or 6, characterized in that: In step 2, the method for calculating the failure probability of the software nodes affected by the hardware failure state in the public resource topology model is: First, according to the topology structure of the public resource topology structure model, a fault propagation mode is determined, wherein the fault propagation mode includes a point-to-point fault propagation mode and a multi-point fault propagation mode; In the point-to-point fault propagation mode, the probability that the faulty hardware node A propagates the fault status to the software node B is: e B =e A ·RS AB ·PC A ε A is the failure probability of faulty node A, RS AB represents the failure probability ε A The dependency of node A on node B, PC A Indicates the fault propagation capability of faulty node A; The multi-point fault propagation mode includes two modes: one-transmission-multiple-path mode and multiple-transmission-one-path mode; In the one-pass multi-path mode, The failure of hardware node A propagates to software nodes B, C, and D, causing state failures of software nodes B, C, and D. The failure probabilities of software nodes B, C, and D are: e B =e A ·RS AB ·PC A e C =e A ·RS AC ·PC A e D =e A ·RS AD ·PC A Among them, ε A , ε B , ε C and ε D RS represents the failure probability of nodes A, B, C, and D respectively AB Indicates the dependency of node A on node B, RS AC Indicates the dependency of node A on node C, RS AD Indicates the dependency of node A on node D, PC A Indicates the fault propagation capability of node A; The multi-transmission one-path mode also includes: and relationship fault propagation mode and or relationship fault propagation mode, When the fault propagation relationship is established: the failure probability of software node A′ is: e A′ =e B′ ·RS B′A′ ·PC B′ +e C′ ·RS C′A′ ·PC C′ +e D′ ·RS D′A′ ·PC D′ Or when the relationship fault propagates: the failure probability of software node A′ is: e A′ =max{e B′-A′ ,he C′-A′ ,he D′-A′ } =max{e B′ ·RS B′A′ ·PC B′ ,he C′ ·RS C′A′ ·PC C′ ,he D′ ·RS D′A′ ·PC D′ } RS B′A′ 、RS C′A′ and RS D′A′ They represent the dependency of hardware node B′ on software node A′, the dependency of hardware node C′ on software node A′, and the dependency of hardware node D′ on software node A′, respectively. B′ represents the fault propagation capability of hardware node B′, PC C′ represents the fault propagation capability of hardware node C′, PC D′ represents the fault propagation capability of hardware node D′, ε B′-A′ represents the failure probability of node B′ propagating the failure to node A′, ε C′-A′ represents the probability that node C′ will propagate the fault to node A′, ε D′-A′ represents the failure probability of node D′ propagating the failure to node A′.

8. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 7, characterized in that: In step 3, the method for calculating the fault propagation capability of all nodes in the public resource topology model includes calculating the fault propagation capability of hardware nodes and the fault propagation capability of software nodes; The specific formula for calculating the fault propagation capability of a hardware node is: Among them, PC(v yi ) represents the hardware node v yi The fault propagation capability attribute value, M(v yi ) indicates pointing to the hardware node v yi The node set, L(v yj ) represents the hardware node v yj The number of directed edges connected; the calculation formula for the fault propagation capability of a software node is: PC(u)=aEIC(u)+bIC(u) Where a represents the weight coefficient of the node's external call capability, b represents the weight coefficient of the node's call capability, and the call capability of software node u is IC(u): in, is the in-degree of node u, is the maximum in-degree value in the software sub-network, m is the number of nodes in the software sub-network, Represents the calling capability of the target node u itself, ν ri1 is the i1th node in the direct in-degree neighbor set DINS of the target node u, ν rj1 is the j1th node in the direct in-degree neighbor set DINS of the target node u, Nin represents the number of nodes in the direct in-degree neighbor set of the target node u, Indicates the probability that the target node in the software sub-network is called by the adjacent node; Weight (ν ri1 ,u) is the software node ν ri1 The number of calls to the target node u, Weight(ν ri1 ,ν rj1 ) is the software node ν ri1 For software node ν rj1 The number of calls; External call capability EIC(u) of node u: Among them, α represents the external call capability of the target node itself (0≤α≤1), v ri2 is a node in the direct out-degree set DONS of the target node u, Nout represents the number of nodes in the direct out-degree set of the target node u, Indicates the probability that the target node calls other neighbor nodes in the weighted software execution network, Weight(u,v ri2 ) represents the target node u to the software node v ri2 The number of calls; Weight(v rj2 ,v ri2 ) represents the software node v rj2 For software node v ri2 The number of calls.

9. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 4, characterized in that: In step 3, the method for obtaining the path with the highest probability of fault propagation is: Step 3:

1. Use the faulty hardware node as the initial position of the particle swarm. The path of each particle represents a possible fault propagation path, and the particle speed represents the adjustment direction and step size of the fault propagation path. Randomly initialize the particle swarm algorithm. Step 32: Calculate the fitness value of each particle; Step 33: Determine whether the fitness value of the currently obtained particle is greater than the optimal fitness value of the current particle at the historical moment. If so, use the fitness value at the current moment as the optimal fitness value, and update the position and speed of the particle corresponding to the optimal fitness value. Otherwise, return to step 32 and continue to calculate the fitness value of the current particle until the number of iterations reaches the threshold; output the path of the particle with the optimal fitness value.

10. The method for identifying fault propagation paths of ship electronic information systems based on a public computing environment according to claim 9, characterized in that: In step 3.2, the method for calculating the fitness value of each particle is: weight i,i+1 =ε i ·RS i,i+1 ·PC i Among them: fitness is the fitness value of the particle, n is the number of nodes in the path, weight i,i+1 is the weight of the edge between node i and node i+1, ε i is the failure probability of node i, RS i,i+1 is the failure probability ε i The dependency of node i on node i+1, PC i is the fault propagation capability of node i.

Images