IMA system impending failure prediction method based on multi-source data

Abstract

The embodiment of the application discloses a method for predicting imminent failure of an IMA system based on multi-source data, comprising the following steps: step 1, constructing an imminent failure propagation model of the IMA system according to components and structure of the IMA system; step 2, extracting information for predicting imminent failure in the imminent failure propagation model, and converting a graphical expression form into an array or matrix expression form; and step 3, performing forward reasoning prediction on imminent failure of the IMA system according to the information extracted in step 2. The embodiment of the application solves the problem that the IMA system has imminent failure, and the propagation of the imminent failure leads to cascade functional failure of the IMA system.

Description

Technical Field

[0001] This invention relates to, but is not limited to, the field of IMA system health management, and particularly to a method for predicting imminent failures in IMA systems based on multi-source data. Background Technology

[0002] Impending fault prediction is an important component of health management and system management. Unlike long-term failures or failures caused by long-term degradation and the effects of working stress, impending faults refer to functional failures caused by faults in nearby or surrounding systems and the time delay of their propagation. Effective prediction of this type of failure can help the system complete the corresponding system management (such as reconfiguration, restart, etc.) within a specified time, thereby mitigating the adverse effects of the fault and laying the foundation for improving the mission reliability of the aircraft.

[0003] Because the components of the Integrated Modular Avionics (IMA) system are closely interconnected, a near-failure in one component can cause malfunctions in other components, thus leading to the failure of the IMA system's cascading functionality. Summary of the Invention

[0004] The purpose of this invention is to provide a method for predicting imminent faults in an IMA system based on multi-source data, in order to solve the problem that imminent faults exist in the IMA system and the propagation of these imminent faults causes the cascading function of the IMA system to fail.

[0005] The technical solution of the present invention:

[0006] This invention provides a method for predicting imminent faults in an IMA system based on multi-source data, comprising:

[0007] Step 1: Based on the components and structure of the IMA system, construct a proximate fault propagation model for the IMA system;

[0008] Step 2: Extract the information used to predict the impending fault from the impending fault propagation model, and convert the graphical representation into an array or matrix representation.

[0009] Step 3: Based on the information extracted in Step 2, perform forward inference prediction of the impending fault of the IMA system.

[0010] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, step 1 includes:

[0011] Based on the failure modes, differences, and propagation relationships in the IMA system, we model the failure modes, BIT monitoring methods, and the impact of functional failures in the IMA system, and construct an imminent failure propagation model from failure modes to system functional failures.

[0012] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, step 1 specifically includes:

[0013] Step 11, construct the imminent failure meta-model, including: determining the hierarchical and interconnected relationships between the failure modes of each LRU according to the modeling elements and their attributes corresponding to the TFPG model specifications of each LRU in the IMA system, thereby establishing the imminent failure meta-model of the IMA system.

[0014] Step 12: Use the imminent fault meta-model template to build a TFPG-specific model, including: based on the modeling elements, hierarchical relationships, cross-linking relationships and related attributes defined in the imminent fault meta-model, construct a TFPG-specific model for a specific IMA system, clarify the specific propagation relationships of fault mode-difference point and difference point-difference point, and assign values ​​to the propagation time interval.

[0015] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above,

[0016] The imminent fault meta-model includes: the IMA system contains the fault mode type, path type, logical type of each LRU, and the attribute settings of the aforementioned elements;

[0017] The TFPG-specific model includes: specific fault modes, fault mode-difference points, and difference point-difference point paths for each LRU in a specific IMA system, and specific values ​​assigned to the attributes.

[0018] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, the information extracted in step 2 for predicting imminent failures includes:

[0019] Fault mode nodes, differences, adjacency matrix A, reachability matrix A*, minimum propagation time matrix t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max .

[0020] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, the information for predicting imminent failures extracted in step 2 specifically includes: m failure mode nodes and n difference points;

[0021] The n difference points include ordinary difference nodes and virtual nodes. The virtual nodes include "AND" logical nodes and "OR" logical nodes, and n1+n2+n3=n; where n1 is the number of ordinary difference nodes, n2 is the number of "OR" logical nodes, and n3 is the number of "AND" logical nodes.

[0022] The adjacency matrix A, reachability matrix A*, and minimum propagation time matrix t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max All are (n+m)×(n+m) dimensional matrices.

[0023] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, the information for predicting imminent failures extracted in step 2 includes...

[0024] Adjacency matrix A is used to reflect the connection relationships between nodes. ij =1 indicates node v i and node v j There are connections between them;

[0025] The reachability matrix A* reflects whether there is a path between two nodes. This indicates that there exists a path from v in the model. i to v j The path;

[0026] Minimum propagation time matrix t min Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... This indicates that the fault originates from the parent node v. i Propagate to child node v j Minimum propagation time required (A) ij =1); and A in the adjacency matrix ij =0, then the corresponding element in the minimum propagation time matrix

[0027] Maximum propagation time matrix t max Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... This indicates that the fault originates from the parent node v. i Propagate to child node v j Maximum propagation time required (A) ij =1); and A in the adjacency matrix ij =0, then the corresponding element in the maximum propagation time matrix

[0028] Minimum reachable time matrix A minEach element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node v. i Propagate to another node v j minimum time And reachable from the matrix Then the corresponding elements in the minimum reachable time matrix

[0029] Maximum reachable time matrix A max Each element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node v. i Propagate to another node v j Maximum time And reachable from the matrix Then the corresponding element in the maximum reachable time matrix

[0030] Optionally, in the IMA system imminent failure prediction method based on multi-source data as described above, step 3 includes: imminent failure propagation logic analysis and time range matching analysis;

[0031] The aforementioned obstacle propagation logic analysis includes: analyzing whether there are direct or indirect connections between nodes, and analyzing the requirements for activation of difference points;

[0032] The time range matching analysis includes: determining whether there is a contradiction between the actual difference point activation time and the fault propagation time constraint in the TFPG dedicated model.

[0033] Advantages of this invention: This invention provides a method for predicting impending failures in an IMA system based on multi-source data. Specifically, it is a short-term impending failure prediction method based on BIT data and functional difference data from multiple LRU / LRM (Large Number Replaceable Modules). In this method, on one hand, a proximate failure propagation model from fault mode to system functional failure is built using difference point information from multiple monitoring sources within the IMA system, abnormal event reporting information, and fault information propagation relationships. The model information is extracted, and a forward inference algorithm is used to predict the time range of the impending failure. This time range can be read by the system management function, thus providing a basis for optimizing further fault mitigation actions. On the other hand, by analyzing typical fault modes and failure propagation relationships of the IMA system, and based on the established fault prediction model, correlation analysis and time consistency analysis are performed according to actual or simulated abnormal event sequences to obtain possible fault modes and fault occurrence time estimates, thereby achieving impending failure prediction. The IMA system imminent fault prediction method based on multi-source data provided in this invention can predict system functional failures caused by primary LRU / LRM faults, and can provide the time range of functional failures based on different logical (AND, OR) relationships of fault propagation. Attached image description:

[0034] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of the present invention and do not constitute a limitation on the technical solutions of the present invention.

[0035] Figure 1 This is a schematic diagram of a TFPG model in an embodiment of the present invention;

[0036] Figure 2 A flowchart of a method for predicting imminent faults in an IMA system based on multi-source data, provided in an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the structure of the TFPG-specific model in an embodiment of the present invention;

[0038] Figure 4 This is a schematic diagram of a TFPG-specific model for the GPNCU platform provided in an embodiment of the present invention;

[0039] Figure 5 A schematic diagram illustrating the corresponding fault modes of each unit (LRU) in the IMA system provided for a specific embodiment of the present invention;

[0040] Figure 6 This is a schematic diagram of the TFPG model for each LRU in a specific embodiment of the present invention;

[0041] Figure 7 This is a schematic diagram of the meta-model of IMA system faults in a specific embodiment of the present invention;

[0042] Figure 8 For specific implementation of the present invention Figure 7 A schematic diagram of the top modeling element of the meta-model shown;

[0043] Figure 9 This is a schematic diagram of LRU-related constraints for IDU in a specific embodiment of the present invention;

[0044] Figure 10 This is a schematic diagram of a TFPG-specific model formed for GPNCU dual redundancy in a specific embodiment of the present invention;

[0045] Figure 11 This is a schematic diagram of the TFPG model obtained using simulation software in a specific embodiment of the present invention;

[0046] Figure 12 This is a schematic diagram of a single imminent fault prediction simulation under a single alarm sequence in a specific embodiment of the present invention;

[0047] Figure 13 This is a schematic diagram of the simulation of the imminent fault prediction of a normal alarm sequence under a single alarm sequence in a specific embodiment of the present invention. Detailed implementation method:

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

[0049] The steps illustrated in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases the steps shown or described may be performed in a different order than that presented here.

[0050] As explained in the background section above, due to the close interconnections among the components of the Integrated Modular Avionics (IMA) system, a near-failure in one component can cause malfunctions in other components, leading to the failure of the cascaded functions of the IMA system.

[0051] To address the issue of cascading failure in the aforementioned IMA system, this invention provides a method for predicting imminent failures in IMA systems based on multi-source data.

[0052] The present invention provides the following specific embodiments, which can be combined with each other. For the same or similar concepts or processes, they may not be described again in some embodiments.

[0053] The imminent fault prediction method for IMA systems based on multi-source data provided in this invention embodiment employs a time-based fault propagation graph. The following section first describes the imminent fault prediction method based on the time-based fault propagation graph.

[0054] A method for predicting impending faults based on the timed failure propagation graph (TFPG).

[0055] Due to the dynamic nature of the system, fault information propagation does not occur immediately. Instead, it requires some time to spread. For example, exceeding the high temperature in a damaged heater takes time to raise the temperature above the acceptable limit. Furthermore, in avionics systems, due to the extensive interconnections between related units / modules, a faulty unit or module can transmit abnormal information caused by the fault to other units / modules, leading to faults, errors / alarms, or other abnormal events. The property of fault information propagation time is evident due to the system's real-time requirements and the constraints of the primary time frame. To incorporate the dynamics into the fault model, each fault propagation is parameterized using a time interval called the propagation interval [tmin, tmax], which gives the minimum and maximum propagation time from a previous fault to a subsequent fault.

[0056] like Figure 1 The image shown is a schematic diagram of a TFPG model according to an embodiment of the present invention. Figure 1 The diagram shows an example of a TFPG model. In this diagram, rectangles represent failure modes. Circles represent differences with logical "OR" operations, and squares represent differences with logical "AND" operations. The edges between nodes represent the propagation relationships between failure modes and differences in the system, as well as between differences. The propagation characteristics of these relationships are parameterized by time intervals [tmin, tmax]. Differences that trigger alarms are marked with shaded areas in the diagram, and the time of the alarm is displayed above the corresponding difference.

[0057] The axiomatic form of the TFPG model is represented by a quintuple G = (F, D, E, t) min ,t max ),in:

[0058] F represents the set of failure modes;

[0059] D is the point of difference (e.g.) Figure 1 Set d) in the set, and satisfying

[0060] It is an edge set (or path set), representing the set of edges formed by a node in V and another node in V, where V = FUD;

[0061] t min ,t max Let E represent the minimum and maximum propagation times of any path e (e∈E), where E is the set of paths.

[0062] The path e = (v, v') ∈ E means that due to the propagation effect, the node will undergo a state change from v to v'. We use the symbols tmin and tmax to represent the minimum and maximum propagation time on the propagation path, respectively. Assume 0 ≤ e.tmin ≤ e.tmax.

[0063] like Figure 1 The TFPG model shown is used to describe the observed fault propagation among sensors, BITs, alarms, and other signals in a real-world system. In this case, the set of all observed deviations corresponds to the difference point d established in the TFPG model, and the path corresponds to the causal relationship of the system dynamics. Due to the dynamic nature of the system, the propagation of the effects of a fault between system components (this time typically depends on the system's time constraints and the temporal evolution of the underlying fault) takes a corresponding amount of time. In many cases, this time delay can be calculated by analyzing or simulating an accurate model (e.g., a real-time demand model).

[0064] Several assumptions are crucial. First, the TFPG model does not contain edges describing self-loop characteristics, and any failure mode is described as a root node, meaning it cannot be the destination of any edge. Furthermore, each divergence point d must be a successor to another divergence point or failure mode. Once a failure effect reaches node v', the state of node v' is permanently changed and unaffected by any future failure propagation. Therefore, the case of intermittent failures is not considered in this embodiment of the invention.

[0065] Figure 2 This is a flowchart illustrating a method for predicting imminent faults in an IMA system based on multi-source data, provided in an embodiment of the present invention. The method includes the following steps:

[0066] Step 1: Based on the components and structure of the IMA system, construct a proximate fault propagation model for the IMA system;

[0067] In this embodiment of the invention, the specific implementation of step 1 may include:

[0068] Based on the components of the IMA system (including failure modes and difference point information) and its structure (i.e., propagation relationships), the impact of failure modes, BIT (i.e., internal self-test) monitoring methods (referring to direct difference point information adjacent to failure modes) and functional failure (referring to all difference point information other than direct difference point information) on the IMA system is modeled, thereby constructing an imminent failure propagation model from failure mode to system functional failure.

[0069] Step 2: Extract the information used to predict the impending fault from the impending fault propagation model and convert the graphical representation into an array or matrix representation.

[0070] In this embodiment of the invention, considering that the output of the imminent failure prediction inference algorithm is the failure of related functions and its specific time range, it is necessary to extract the information used to predict the imminent failure from the imminent failure propagation model in advance.

[0071] Step 3: Based on the information extracted in Step 2, perform forward inference prediction of the impending fault of the IMA system.

[0072] In this embodiment of the invention, the specific implementation process of step 1 above may include:

[0073] Step 11, construct the imminent fault meta-model, constructed as follows:

[0074] Based on the modeling elements and their attributes corresponding to the TFPG model specifications of each LRU (Field Replaceable Unit) in the IMA system, the hierarchical and interconnection relationships between the failure modes of each LRU are determined, thereby establishing the imminent fault meta-model of the IMA system.

[0075] Step 12: Use the imminent fault meta-model template to build a TFPG-specific model. The construction method is as follows:

[0076] Based on the modeling elements, hierarchical relationships, cross-linking relationships, and related attributes defined in the near-fault meta-model, a TFPG-specific model is constructed for a specific IMA system to clarify the specific propagation relationships between fault modes and differences, and between differences and differences, and to assign values ​​to the propagation time intervals.

[0077] It should be noted that, in step 1 of this embodiment of the invention, the constructed near-fault meta-model includes: the IMA system contains the fault mode type, path type, logical type of each LRU, and attribute settings of the aforementioned elements. Additionally, the constructed TFPG-specific model includes: specific fault modes, fault mode-difference points, and difference point-difference point paths for each LRU in a specific IMA system, and specific values ​​assigned to the attributes.

[0078] In step 2 of this embodiment of the invention, the extracted information for predicting impending faults includes:

[0079] Fault mode nodes, differences, adjacency matrix A, reachability matrix A*, minimum propagation time matrix t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max .

[0080] In a specific implementation of this invention, the information on the predicted imminent fault extracted in step 2 specifically includes: m fault mode nodes and n difference points;

[0081] The n difference points include ordinary difference nodes and virtual nodes. The virtual nodes include "AND" logical nodes and "OR" logical nodes, and n1+n2+n3=n; where n1 is the number of ordinary difference nodes, n2 is the number of "OR" logical nodes, and n3 is the number of "AND" logical nodes.

[0082] And the adjacency matrix A, the reachability matrix A*, and the minimum propagation time matrix t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max All are (n+m)×(n+m) dimensional matrices.

[0083] Adjacency matrix A is used to reflect the connection relationships between nodes. ij =1 indicates node v i and node v j There are connections between them;

[0084] The reachability matrix A* reflects whether there is a path between two nodes. This indicates that there exists a path from v in the model. i to v j The path;

[0085] Minimum propagation time matrix t min Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... This indicates that the fault originates from the parent node v. i Propagate to child node v j Minimum propagation time required (A) ij =1); and A in the adjacency matrix ij =0, then the corresponding element in the minimum propagation time matrix

[0086] Maximum propagation time matrix t max Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... This indicates that the fault originates from the parent node v. i Propagate to child node v j Maximum propagation time required (A) ij =1); and A in the adjacency matrix ij =0, then the corresponding element in the maximum propagation time matrix

[0087] Minimum reachable time matrix A min Each element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node v. i Propagate to another node v j minimum time And reachable from the matrix Then the corresponding elements in the minimum reachable time matrix

[0088] Maximum reachable time matrix A max Each element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node v. i Propagate to another node v j Maximum time And reachable from the matrix Then the corresponding element in the maximum reachable time matrix

[0089] In step 3 of this embodiment of the invention, forward thrust prediction may include: proximity fault propagation logic analysis and time range matching analysis.

[0090] The implementation method of the above-mentioned obstacle propagation logic analysis is as follows: analyze whether there is a direct or indirect connection between nodes, and analyze the requirements for activation of difference points;

[0091] The above time range matching analysis is implemented by determining whether there is a contradiction between the actual activation time of the difference point and the fault propagation time constraint in the TFPG dedicated model.

[0092] The imminent failure prediction method for IMA systems based on multi-source data provided in this invention embodiment is specifically a short-term imminent failure prediction method based on BIT data and functional difference data of multiple LRU / LRM (Field Replaceable Modules). In this imminent failure prediction method, on the one hand, a proximate failure propagation model from failure mode to system functional failure is built using difference point information from multiple monitoring sources within the IMA system, abnormal event reporting information, and failure information propagation relationships. The model information is extracted, and a forward inference algorithm is used to predict the time range of the imminent failure. This time range can be read by the system management function, thus providing a basis for optimizing further failure mitigation actions. On the other hand, by analyzing typical failure modes and failure propagation relationships of the IMA system, based on the established failure prediction model, correlation analysis and time consistency analysis are performed according to actual or simulated abnormal event sequences to obtain possible failure modes and failure occurrence time estimates, thereby achieving imminent failure prediction. The IMA system imminent fault prediction method based on multi-source data provided in this invention can predict system functional failures caused by primary LRU / LRM faults, and can provide the time range of functional failures based on different logical (AND, OR) relationships of fault propagation.

[0093] The following detailed description of the implementation of the IMA system imminent fault prediction method based on multi-source data provided in this invention is illustrated through some specific embodiments.

[0094] The specific implementation of the IMA system approach fault prediction method based on multi-source data provided in this embodiment includes the following steps.

[0095] Step 1: Construct a proximate fault propagation model for the IMA system

[0096] 1) Constructing a meta-model of imminent faults

[0097] First, determine the modeling elements and their attributes (model, atoms, etc.) corresponding to different TFPG models; second, determine the hierarchical and cross-linking relationships (i.e., connection relationships) between different failure modes; finally, construct the imminent fault meta-model of the corresponding IMA system based on the aforementioned different elements, attributes, and connection relationships.

[0098] After establishing a complete metamodel, a corresponding XML file can be generated, and the compiler can generate a near-fault metamodel. Once a specific process is determined, a new metamodel project file can be created in GME. This includes... Figure 1 The main elements listed include, for example, failure modes, differences, and propagation paths. Failure modes and differences are categorized as "failures" and "alarms" (including BIT monitoring methods, functional failures, etc.).

[0099] 2) Construct a TFPG-specific model for the IMA system

[0100] The TFPG-specific model can be built directly using the near-fault meta-model template. For example... Figure 3 The diagram shown is a structural schematic of the TFPG-specific model in an embodiment of the present invention. In this TFPG-specific model, the Alarm icon represents a warning node (difference point), while the Fault icon represents a fault mode node. Alarms and faults are defined as modeling elements in the meta-model template, which also defines the connection relationships between elements.

[0101] Depend on Figure 3 As can be seen, the fault propagation path is directional, and the time interval [a, b] on each path represents the time constraint between two nodes, which is the setting configured in the element attributes during the above meta-model construction process. Based on the TFPG-specific model, an XML file can be compiled, thereby providing the model data foundation for subsequent TFPG prediction inference through software developed in MATLAB.

[0102] The following provides an example of how to build a TFPG-specific model.

[0103] Currently, advanced passenger aircraft such as the A380, B787, A350XWB, and C919 all use the IMA platform. Data is collected by the Remote Data Interface Unit (RDIU), then sent via the avionics Ethernet bus (A664 network) to the integrated processing cabinet for processing, and finally sent to the integrated display system and alarm system.

[0104] At the heart of the IMA system is a modular "core processing platform" (e.g., CCR). A core processing component cabinet is introduced into the physical architecture of the avionics system. The same standardized hardware modules are used for routine information processing and network communications. The processing modules within the cabinet are replaceable. In the IMA system, line-replaceable modules (LRMs) connected via a distributed network and bus replace traditional digital modules. This bus-based modular design centralizes computation and data processing tasks into shared modules on the platform.

[0105] Fault modeling based on TFPG can be applied to fault propagation modeling of RDIU, core processing platform and related avionics LRU.

[0106] The General Processing and Network Communication Unit (GPNCU) is a core processing platform responsible for running various managed applications based on ARINC653 and exchanging data based on A664. It primarily implements two sets of functions: resource functions and support functions. Resource functions provide two types of resources: data computation and data transmission. Support functions include: health management, configuration management, data loading, network management, fatigue indication, global time, status management, temperature monitoring, and fault recovery. For the aforementioned GPNCU platform, such as... Figure 4 The diagram shown is a schematic of a TFPG-specific model for the GPNCU platform provided in an embodiment of the present invention, specifically a TFPG-specific model based on a meta-model.

[0107] Figure 4 In the TFPG-specific model shown, the failure mode elements in the leftmost column, numbered "GPNCU_NSM", are defined by the meta-model, but no specific failure mode codes or names are specified. The two difference point elements on the right, identified by "ALARM", as well as the connections between failure modes and difference points and difference points and difference points, are also defined by the meta-model. The specific model, however, describes the specific failure modes (names or codes), difference points (names or codes), and connection relationships (source-destination relationships, propagation time attributes) based on the design information of the research object (such as FMEA).

[0108] Step 2: Extract information from the approaching fault propagation model used to predict approaching faults.

[0109] Extract the information required for the prediction algorithm from the established near-fault propagation model, and transform the graphical representation into a mathematical representation of arrays or matrices.

[0110] The extracted information and its mathematical expression will be explained one by one below:

[0111] (1) Fault mode node

[0112] Fault mode nodes are numbered starting from 1, and the total number of fault mode nodes is represented by m.

[0113] (2) Differences

[0114] The difference points are divided into ordinary difference nodes (i.e., single-input nodes) and virtual nodes (i.e., multi-input nodes). Virtual nodes include "AND" logic nodes and "OR" logic nodes. All difference points are numbered starting from 1. The total number of difference points is represented by n. The number of ordinary difference nodes is represented by n1, the number of "OR" logic nodes is represented by n2, and the number of "AND" logic nodes is represented by n3. n1 + n2 + n3 = n.

[0115] (3) Adjacency matrix A

[0116] The adjacency matrix A is an (n+m)×(n+m) dimensional adjacency matrix that reflects the connection relationships between nodes. If node v i and node v j There is a connection between A and B. ij =1; otherwise A ij =0.

[0117] (4) Reachability matrix A*

[0118] The reachability matrix A* is also an (n+m)×(n+m) dimensional matrix, reflecting whether there is a path between two nodes. If there exists a path from node v... i to node v j The path, otherwise

[0119] (5) Minimum propagation time matrix t min

[0120] Minimum propagation time matrix t min It is an (n+m)×(n+m) dimensional matrix, corresponding to the adjacency matrix A. Its elements... This indicates that the fault originates from the parent node v. i Propagate to child node v j Minimum propagation time required (A) ij =1). If A ij =0, then

[0121] (6) Maximum propagation time matrix t max

[0122] Maximum propagation time matrix t max It is an (n+m)×(n+m) dimensional matrix, corresponding to the adjacency matrix A. Its elements... This indicates that the fault originates from the parent node v. i Propagate to child node v j Maximum propagation time required (A) ij =1). If A ij =0, then

[0123] (7) Minimum reachable time matrix A min

[0124] Minimum reachable time matrix A min It is an (n+m)×(n+m) dimensional matrix, corresponding to the reachability matrix A*. Its elements... This indicates that the fault originates from a node v. i Propagate to another node v j minimum time if So

[0125] (8) Maximum reachable time matrix A max

[0126] Maximum reachable time matrix A max It is an (n+m)×(n+m) dimensional matrix, corresponding to the reachability matrix A*. The elements... This indicates that the fault originates from a node v. i Propagate to another node v j Maximum time if So

[0127] Step 3: Perform forward inference prediction for the impending fault of the IMA system.

[0128] Forward inference prediction includes: near fault propagation logic analysis and time range matching analysis;

[0129] Forward inference prediction relies on approach fault propagation logic analysis and time range matching analysis. Therefore, approach fault propagation logic analysis refers to the direct or indirect connections between nodes, as well as the activation requirements of difference points. Time range matching analysis determines whether there are any contradictions between the actual activation times of difference points and the fault propagation time constraints in the TFPG-specific model.

[0130] The forward operation algorithm, based on the two analysis results mentioned above, predicts the state of subsequent nodes of the activated differential node by forward expansion, playing a role in impending failure prediction in IMA system health management. The specific forward expansion process is described below:

[0131]

[0132] The following provides an implementation example illustrating the implementation method and effects of the embodiments of the present invention.

[0133] The loss of flight data on the co-pilot's side is one of the most serious functional failure events in avionics systems. This will be used as an example for implementation.

[0134] 1) Construct a proximate fault propagation model for the IMA system

[0135] Based on system interconnections and FMEA (Failure Mode and Effects Analysis), a fault tree can be used to describe functional failures in safety analysis where the passenger-side attitude information is lost, as follows: Figure 5 The diagram shown illustrates the corresponding fault modes of each unit (LRU) in the IMA system provided in a specific embodiment of the present invention.

[0136] superior Figure 5 Each basic event (such as) Figure 5 The fault modes (indicated by the circles in the middle) and their corresponding units are shown in the table below.

[0137]

[0138]

[0139] Based on the specific fault modes of the functional circuits in FMEA, more typical fault modes can be selected to refine the basic events, thereby obtaining TFPG models for each LRU, such as... Figure 6 The diagram shown is a schematic of the TFPG model for each LRU in a specific embodiment of the present invention, where the time interval for each path is represented by [tmin, tmax] in the second part. Figure 6 In the diagram, rectangles represent LRU failure modes according to FMEA. Circles indicate differences, including logical "OR", BIT, FDE, and system function failures. Squares indicate differences for "AND" logic types, which represent redundant structures.

[0140] To Figure 6 In this process, TFPG models are constructed for each LRU, resulting in a meta-model of the entire IMA system fault, as follows: Figure 7 The diagram shown is a schematic representation of the meta-model of IMA system faults in a specific embodiment of the present invention.

[0141] The aforementioned metamodel mainly includes IMA system functional failure events and the elements and attributes of faults and alarms for each LRU. The system fault alarm event element defines the name of the system event, the possible logic gates for the fault and alarm, and the connection constraints (propagation time interval [tmin, tmax]), such as... Figure 8 As shown, this is a specific implementation of the present invention targeting... Figure 7 A schematic diagram of the top modeling element of the shown meta-model.

[0142] As a class object representing "passenger-side attitude information failure," the fault and alarm elements of each LRU specify the possible logical combinations of faults and the connection constraints (propagation time interval [tmin, tmax]) for propagating "fault to alarm" and "alarm to alarm." Taking IDU as an example... Figure 9 The diagram shown is a schematic of LRU-related constraints for IDU in a specific embodiment of the present invention.

[0143] The fault and alarm information for specific functional circuits included in the LRU portion of the meta-model comes from the functional circuit fault modes in FMEA.

[0144] Based on the modeling paradigm of the TFPG-specific model obtained from the meta-model, TFPG-specific models for each LRU and subsystem were also constructed. Taking the dual redundancy of the GPNCU as an example, the corresponding TFPG-specific model is as follows: Figure 10 The diagram shown is a schematic of a TFPG-specific model formed for GPNCU dual redundancy in a specific embodiment of the present invention. Figure 10 In the naming of nodes, F represents the fault mode (FM), while D represents the difference. The time interval on the path indicates the range of time it takes for the fault to propagate along the path. Figure 10 The image only shows the GPNCU portion of the TFPG-specific model.

[0145] Figure 10 The specific information for each node is shown in the table below.

[0146] ID TYPE Failure Mode / Alarm / Logic GPNCU_F01 Failure Mode Switch malfunction GPNCU_F02 Failure Mode Switch malfunction GPNCU_F03 Failure Mode Loss of communication with RDIU GPNCU_F04 Failure Mode Switch malfunction GPNCU_F05 Failure Mode Switch malfunction GPNCU_F06 Failure Mode Loss of communication with RDIU GPNCU_D01 Differences NSM Fault / Continuous BIT GPNCU_D02 Differences NSM Fault / Continuous BIT GPNCU_D03 Differences NSM Fault / Continuous BIT GPNCU_D04 Differences NSM Fault / Continuous BIT GPNCU_D05 Differences GPNCU1 fault / FDE GPNCU_D06 Differences GPNCU2 fault / FDE GPNCU_D07 Differences GPNCU communication failure / FDE OR_01 Logic gates Logical "OR" OR_02 Logic gates Logical "OR" AND_01 Logic gates Logical AND

[0147] The structure and node information of other LRU components in the TFPG-specific model are similar to those of GPNCU. By integrating the TFPG models related to GPNCU, RDIU, IDU, and AHRS, a TFPG based on system functional faults can be obtained.

[0148] 2) Extract information from the approaching fault propagation model used to predict approaching faults.

[0149] Taking GPNCU as an example, the information used for prediction in the model is extracted.

[0150] (1) Fault mode node

[0151] The number of common failure modes (m) is 6.

[0152] (2) Differences

[0153] The number of ordinary difference points (n1) is 7, the number of "OR" logical nodes (n2) is 2, and the number of "AND" logical nodes (n3) is 1. The total number of difference points (n) is 10.

[0154] (3) Adjacency matrix A

[0155] In the GPNCU example, A is an (n+m)×(n+m) dimensional adjacency matrix, for example, a (16×16) dimensional adjacency matrix, with the following specific form.

[0156]

[0157] 4) Reachability matrix A*

[0158]

[0159] 5) Minimum propagation time matrix t min

[0160]

[0161] 6) Maximum propagation time matrix t max

[0162]

[0163] 7) Minimum reachable time matrix A min

[0164]

[0165] 8) Maximum reachable time matrix A max

[0166]

[0167] 3) Simulation of a case study on near-fault prediction based on TFPG

[0168] The main interface of the TFPG simulation software developed in MATLAB consists of three parts: a parameter configuration block, a TFPG graphical display panel, and a simulation text box, as shown in the figure. The parameter bar can be used to set the alarm time and each difference point. The graphical display of the TFPG model is mainly used to show the fault propagation sequence under given triggering conditions during simulation. The simulation results text box is used to display the predictive assumptions.

[0169] First, the TFPG model is exported as a standard XML file. Then, the model's structural parameters are imported into the MATLAB workspace using TFPG simulation software. TFPG models exported in this way are stored in GMEXML format. This file format is essentially the same as the standard XML format and can be considered a standard XML file. Figure 11 The diagram shown is a schematic of the TFPG model obtained using simulation software in a specific embodiment of the present invention.

[0170] Using GPNCU's TFPG as an example, the imminent fault prediction algorithm based on TFPG was verified and validated through simulation.

[0171] a) Simulation of single proximity fault prediction under a single alarm sequence

[0172] In this simulation, F2 is assumed to be the actual primary fault mode, and the actual fault duration is 10ms. This simulation is performed when the system structure is normal. The fault propagates according to the corresponding relationship, and the corresponding alarm occurs correctly. We set the corresponding alarm time (time 35, d2). Figure 12 The figure shown is a schematic diagram of a single approaching fault prediction simulation under a single alarm sequence in a specific embodiment of the present invention.

[0173] The predictive assumptions for the failure modes are displayed in the simulation results text box, as shown below.

[0174] *-*--*-*-*-*-*-*-*-*-*Time35 List of Hypothesis*-*-*-*-*-*-*-*-*-*-*

[0175] >>>>>>>Hypo 1<<<<<<<

[0176] Failure mode:FM2[0,15] SA:[2] FA:[] MA:[] IF D1:

[85] Plausibility:1.00

[0177] *-*-*-*-*-*-*-*-*-*-*Time35 List of Hypothesis*-*-*-*-*-*-*-*-*-*-*-*

[0178] In the above demonstration, the failure time of functional difference point D5 is t=85. This aligns with the results of fault logic analysis and time range matching analysis. The system should implement fault mitigation management as soon as possible at this point.

[0179] b) Simulation of impending fault prediction in normal alarm sequences

[0180] In this simulation, it is assumed that F2 and F4 are actual fault modes, and the actual fault times are time 3 (f2) and time 10 (f4). This simulation is performed when the system structure is normal. The fault propagates according to the correspondence, and the corresponding alarm occurs correctly. We set the corresponding alarm sequence (Time75, d5; Time 85, d5), as follows: Figure 13 The diagram shown is a simulation of the imminent fault prediction of a normal alarm sequence under a single alarm sequence in a specific embodiment of the present invention.

[0181] The prediction of the impending failure mode is displayed in the results text box, as shown below.

[0182] *-*-*-*-*-*-*-*-*-*-*-*Time 25List of Hypothesis*-*-*-*-*-*-*-*-*-*-*

[0183] >>>>>>>Hypo 1<<<<<<<

[0184] Failure mode:FM2[0,5] SA:[2] FA:[] MA:[] IF D5:

[75] Plausibility:1.00

[0185] *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

[0186] *-*-**-*-*-*-*-*-*-*Time 35List of Hypothesis*-**-*-*-*-*-*-*-*-*-*-*

[0187] >>>>>>>Hypo 1<<<<<<<

[0188] Failure mode:FM2[0,5] SA:[2] FA:[3] MA:[] IF D5:

[75] Plausibility:0.50

[0189] >>>>>>>Hypo 2<<<<<<<

[0190] Failure mode:FM4[0,15] SA:[3] FA:[2] MA:[] IF D6:

[85] D7:

[135] Plausibility:0.50

[0191] *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

[0192] In the above demonstration:

[0193] (1) The hypothesis generated when alarm d2 occurs at time 25 indicates that fault mode f2 has occurred. The imminent fault (D5 difference point) occurs at time 75, which is consistent with the fault logic and time matching.

[0194] (2) The hypothesis generated when alarm d3 occurs at time 35 indicates that fault mode f4 has occurred. The imminent fault (D6 difference point) occurs at time 85, which is consistent with the fault logic and time matching.

[0195] (3) Since f2 and f4 have both occurred, at this time (i.e. time 35), the predicted time for the imminent failure, i.e. the failure of the entire GPNCU (D7), is 135, which is consistent with the fault logic and time matching.

[0196] (4) The system should implement fault mitigation management as soon as possible to prevent unsafe events from occurring.

[0197] While the embodiments disclosed in this invention are as described above, the content is merely for the purpose of facilitating understanding of the invention and is not intended to limit the invention. Any person skilled in the art to which this invention pertains may make any modifications and changes to the form and details of the implementation without departing from the spirit and scope disclosed herein; however, the scope of patent protection of this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A method for predicting imminent faults in an IMA system based on multi-source data, characterized in that, include: Step 1: Based on the components and structure of the IMA system, construct the impending fault propagation model of the IMA system. The impending fault propagation model includes: the fault mode type, path type, logical type of each LRU included in the IMA system, and the attribute settings of the aforementioned elements. Step 2: Extract the information used to predict the impending fault from the impending fault propagation model, and convert the graphical representation into an array or matrix representation. Step 3: Based on the information extracted in Step 2, perform forward reasoning prediction of the imminent faults of the IMA system; The information extracted in step 2 for predicting impending faults includes: Fault mode nodes, differences, adjacency matrix A, reachability matrix A*, minimum propagation time matrix t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max ; The information on the predicted imminent fault extracted in step 2 specifically includes: m fault mode nodes and n difference points; Among them, the n difference points include ordinary difference nodes and virtual nodes. The virtual nodes include "AND" logical nodes and "OR" logical nodes, and ;in, n 1 represents the number of ordinary difference nodes. n 2 represents the number of "OR" logical nodes. n 3 represents the number of AND logical nodes; Its adjacency matrix A, reachability matrix A*, and minimum propagation time matrix are described. t min Maximum propagation time matrix t max Minimum reachable time matrix A min Maximum reachable time matrix A max All 3D matrix; The basis for forward reasoning prediction of the impending fault of the IMA system in step 3 includes: performing impending fault propagation logic analysis and time range matching analysis. The aforementioned obstacle propagation logic analysis includes: analyzing whether there are direct or indirect connections between nodes, and analyzing the requirements for activation of difference points; The time range matching analysis includes: determining whether there is a contradiction between the actual difference point activation time and the fault propagation time constraint in the TFPG dedicated model; The forward inference prediction is based on the above two analysis results, and predicts the state of subsequent nodes of the activated differential node by extending forward, so as to play a role in the imminent failure prediction in the health management of the IMA system. The information on the predicted imminent fault extracted in step 2, The adjacency matrix A is used to reflect the connection relationships between nodes. Represents a node v i and nodes v j There are connections between them; The reachability matrix A* reflects whether there is a path between two nodes. This indicates that there is a slave node in the model. v i To the node v j The path; Minimum propagation time matrix t min Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... Indicates that the fault originates from the parent node. v i propagate to child nodes v j Minimum propagation time required ; and in the adjacency matrix Then the corresponding element in the minimum propagation time matrix ; Maximum propagation time matrix t max Each element corresponds one-to-one with an element in the adjacency matrix A, where the elements are... Indicates that the fault originates from the parent node. v i propagate to child nodes v j Maximum propagation time required ; and in the adjacency matrix Then the corresponding element in the maximum propagation time matrix ; Minimum reachable time matrix A min Each element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node. v i propagate to another node v j minimum time And reachable from the matrix Then the corresponding element in the minimum reachable time matrix ; Maximum reachable time matrix A max Each element corresponds one-to-one with an element in the reachability matrix A*, where the element... This indicates that the fault originates from a node. v i propagate to another node v j Maximum time And reachable from the matrix Then the corresponding element in the maximum reachable time matrix .

2. The imminent fault prediction method for IMA systems based on multi-source data according to claim 1, characterized in that, Step 1 includes: Based on the failure modes, differences, and propagation relationships in the IMA system, we model the failure modes, BIT monitoring methods, and the impact of functional failures in the IMA system, and construct an imminent failure propagation model from failure modes to system functional failures.

3. The imminent fault prediction method for IMA systems based on multi-source data according to claim 2, characterized in that, Step 1 specifically includes: Step 11, construct the imminent failure meta-model, including: determining the hierarchical and interconnected relationships between the failure modes of each LRU according to the modeling elements and their attributes corresponding to the TFPG model specifications of each LRU in the IMA system, thereby establishing the imminent failure meta-model of the IMA system. Step 12: Use the imminent fault meta-model template to build a TFPG-specific model, including: based on the modeling elements, hierarchical relationships, cross-linking relationships and related attributes defined in the imminent fault meta-model, construct a TFPG-specific model for a specific IMA system, clarify the specific propagation relationships of fault mode-difference point and difference point-difference point, and assign values ​​to the propagation time interval.

4. The imminent fault prediction method for IMA systems based on multi-source data according to claim 3, characterized in that, The TFPG-specific model includes: specific fault modes, fault mode-difference points, and difference point-difference point paths for each LRU in a specific IMA system, and specific values ​​assigned to the attributes.

Images