GBIC-based multi-electric aircraft starting power generation system DBN model lightweight method

Abstract

The embodiment of the invention discloses a GBIC-based multi-electric aircraft starting power generation system DBN model lightweight method, and relates to the technical field of reliability design and modeling of an equipment complex system.The GBIC-based multi-electric aircraft starting power generation system DBN model lightweight method comprises the steps that a DBN model of a starting power generation system is established, nodes in the DBN model correspond to key components and working states of the starting power generation system, and the nodes in the DBN model correspond to the key components and the working states of the starting power generation system; directed edges in the DBN model are used for describing the mutual relation between the nodes, a redundant structure in the DBN model is recognized, lightweight processing is carried out, and the fault state of the starting power generation system is recognized through the DBN model subjected to the lightweight processing. Aiming at the problems of complex reasoning calculation and low efficiency in the use process of the dynamic Bayesian analysis method of the equipment complex polymorphic system, the grey system theory and the Bayesian information criterion are introduced, the lightweight of the dynamic Bayesian network of the complex polymorphic system is realized, the redundant structure is eliminated, and the accuracy of the result is ensured.

Description

Technical Field

[0001] This invention relates to the field of reliability design and modeling technology for starting and generating systems, specifically to transforming data related to starting and generating systems of multi-electric aircraft into a computer-processable format for subsequent processing. In particular, it relates to a lightweight method for a DBN model of a multi-electric aircraft starting and generating system based on the Grey-Bayesian Information Criterion (GBIC) required for computer processing. Background Technology

[0002] With the rapid development of electrification and intelligence in starting and generating systems, equipment systems are exhibiting characteristics of complexity, multiple failure states, and dynamic changes. On the one hand, equipment systems are subjected to diverse risk factors and failure modes in harsh working environments, resulting in diverse failure states that are difficult to describe simply as "operation / failure." On the other hand, the complex connections between components and the redundant design of functional structures make traditional static or simple model reliability analysis methods inadequate. Complex and multi-state systems with these two characteristics not only greatly increase the difficulty and cost of equipment maintenance but may also shorten the equipment's life cycle due to inaccurate reliability analysis methods. Therefore, reliability analysis methods for complex and multi-state equipment systems have become a hot topic in the field of reliability.

[0003] In recent years, various methods have been widely researched and applied to address the challenges of reliability analysis for complex multimorphic systems. Among them, the Bayesian Network (BN) method stands out due to its unique advantages. It can not only intuitively express complex probabilistic dependencies between systems (directed acyclic graph structure) and effectively handle multi-state changes, but also integrate static and dynamic information by introducing time slices to form a Dynamic Bayesian Network (DBN), enabling the modeling of the temporal evolution of system states. Crucially, the BN method possesses a unique bidirectional inference mechanism, capable of both forward predictive inference and backward diagnostic inference, making it highly suitable for reliability analysis of complex multimorphic systems. However, with the continuous increase in the complexity and redundancy of start-up and power generation systems, the corresponding model size has expanded dramatically, and the required dimension of the Conditional Probability Table (CPT) has also increased significantly. These factors have led to complex and inefficient inference calculations using traditional BN / DBN methods, not only consuming increasingly more computational resources but also making it difficult to improve the accuracy of the analysis results.

[0004] Therefore, how to reduce the complexity of the analysis model and save computing resources while maintaining the accuracy of calculations in the reliability analysis of starting and generating systems has become a research topic. Summary of the Invention

[0005] The embodiments of the present invention provide a lightweight method for DBN model of multi-electric aircraft starting and generating system based on GBIC, which can reduce the complexity of the analysis model and save computing resources in the reliability analysis of starting and generating system while maintaining the accuracy of calculation.

[0006] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0007] A lightweight method for DBN model of multi-electric aircraft starting and generating system based on GBIC, such as Figure 10 The following are included:

[0008] S1. Establish the DBN model of the starting and generating system.

[0009] In the DBN model, the nodes correspond to the key components and operating states of the starting and power generation system, and the directed edges in the DBN model are used to describe the relationships between the nodes.

[0010] In practical applications, there can be various types of starting and generating systems. In this embodiment, the starting and generating system is taken as an example. The key components include: power supply, signal feedback sensor, starting law calculation sensor, starting control signal generation component, computer control system, starting controller, exciter, starting generator, engine rotor, rotating rectifier, engine, voltage regulation circuit, rectifier bridge and starting and generating system.

[0011] Specifically, the operating states of the starting and generating system include: starting mode and generating mode. The normal operating state is the non-fault state, and the fault states are divided into abnormal operating state (Mistake), partial failure (Ploss), and complete failure (Tloss). The experiment verifies the accuracy and computational efficiency of the starting and generating system fault state calculation before and after the lightweight DBN model in the case of complete failure.

[0012] S2. Identify redundant structures in the DBN model and perform lightweight processing.

[0013] S3. Identify the fault status of the starting and generating system using a lightweight DBN model.

[0014] In this embodiment, S1 includes: reading the maintenance manual and fault isolation manual of the starting power generation system, and determining the CPT of the nodes containing the interval gray number in the DBN model.

[0015] The DBN model for the starting and generating system includes: , Let x represent the state transition probability of node x from time slice t to t+1. t Indicates the state of node t, x t+1 This represents the state of a node at time t+1, where t represents the time slice, i represents the node's corresponding number, and x represents the node's position. i t This represents the node state of the i-th node at time slice t, x i t+1 This indicates that the i-th node is in the state of t+1, and N represents the total number of nodes.

[0016] In this embodiment, S2 includes: obtaining the fit of the DBN model to the data and the parameter dimension of the nodes in the DBN model; calculating the GBIC value of the parent node combination in the DBN model, wherein the magnitude of the GBIC value is positively correlated with the structural redundancy of the DBN model.

[0017] Obtaining the goodness of fit of the DBN model to the data includes:

[0018] ,

[0019] ;

[0020] in, and They represent time slices t and t respectively. j The upper and lower bounds of the gray fuzzy likelihood value, where j represents the time slice number. and Indicates time slice t j The two initial nodes in the middle, and Indicates time slice t j The set of parent nodes corresponding to the two initial nodes at the time of the node. and They represent time slices t and t respectively. j The gray fuzzy conditional probability values ​​corresponding to the two initial nodes are given by p, which represents the total number of root nodes x, and q, which represents the total number of intermediate nodes y.

[0021] The parameter dimensions of nodes in the DBN model include: ,in, The parameter dimension of a node, and Representing node x i and y i The number of parameters, S represents the failure mode of the node, p represents the total number of root nodes x, and q represents the total number of intermediate nodes y.

[0022] Calculating the GBIC value of the parent node combination in the DBN model includes: calculating the uppermost and lowermost values ​​of the parent node combination in the DBN model to obtain the range of GBIC values, where: , ;

[0023] and Representing nodes respectively and The upper and lower extrema at task time T. and The time slices t and t represent the time time of the task time T, respectively. j Middle node and The maximum and minimum values ​​of GBIC corresponding to the combination of traversed parent nodes. and Representing nodes respectively and In time slice t j The GBIC value of the parent node combination at that time. and They represent the maximum likelihood values. and The upper and lower bounds, m represents the time slice t within the task time T. j The total number.

[0024] , ,

[0025] , ,in, Indicates time slice t j The conditional probability value that the internal node contains the gray number in the interval. Indicates correspondence The lower bound of the probability, Indicates correspondence The upper limit of probability, r represents the conditional probability value number containing the gray numbers in the interval, and n represents the time slice t. j The middle node contains the total number of conditional probability values ​​for the gray number interval.

[0026] This invention provides a lightweight DBN model for multi-electric aircraft starting and generating systems based on GBIC. Addressing the problem of rapidly expanding model size and significantly increased dimensionality of conditional probability tables in dynamic Bayesian analysis methods for complex polymorphic systems, leading to complex and inefficient inference calculations, this method introduces grey system theory and Bayesian information criteria to achieve lightweighting of the dynamic Bayesian network for complex polymorphic systems, eliminating redundant structures. Furthermore, it ensures improved accuracy and faster computation time for fault state calculations using the lightweight DBN. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a comparative schematic diagram of the model structure provided in the embodiments of the present invention, wherein, Figure 1 Part (a) in the diagram is a schematic of the BN structure. Figure 1 Part (b) in the diagram is a schematic of the DBN structure;

[0029] Figure 2 A schematic diagram of the logical architecture of the lightweight dynamic Bayesian network based on GBIC provided in an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram illustrating the operating mechanism and key component working relationships of a start-up generator (SG) system provided in an embodiment of the present invention, wherein... Figure 3 Part (a) in the diagram represents the SG system startup mode. Figure 3 Part (b) in the text refers to the SG system power generation mode; Figure 3 The English abbreviations in the text are as follows: BR represents rectifier bridge, CEC represents generator excitation contactor, EXG represents exciter, RR represents rotating rectifier, SG represents starter generator, SC represents starter contactor, CMSC represents general motor starter controller, GCB represents generator control circuit breaker, ATRU represents autotransformer rectifier, APB represents auxiliary power circuit breaker, EPC represents external power contactor, APU represents auxiliary power unit, GEN represents generator, and ATU represents autotransformer unit.

[0031] Figure 4 This is a schematic diagram illustrating the working mechanism and relationships between key components of the SG system provided in an embodiment of the present invention. Figure 4 The English abbreviations in the text are as follows: SFS represents signal feedback sensor, SLS represents starting law calculation sensor, SCSG represents starting control signal generation component, PWR represents power supply, DCU represents computer control system, SCU represents starting controller, APU represents auxiliary power unit, EXG represents exciter, SG represents starter generator, BR represents rectifier bridge, RR represents rotating rectifier, VR represents voltage regulating circuit, ENG represents engine, and OE represents airborne equipment.

[0032] Figure 5This is a schematic diagram of a DBN model for an SG system considering common-cause failure, provided in an embodiment of the present invention. Figure 5 Part (a) in the text describes the DBN model structure applied in the scenario of this embodiment. Figure 5 Part (b) in the text describes the dynamic process of the DBN model; Figure 5 The English abbreviations in the text are as follows: SFS represents signal feedback sensor, SLS represents starting law calculation sensor, SCSG represents starting control signal generation component, PWR represents power supply, DCU represents computer control system, SCU represents starting controller, EXG represents exciter, SG represents starter generator, ER represents engine rotor, RR represents rotating rectifier, BR represents rectifier bridge, VR represents voltage regulating circuit, ENG represents engine, SM represents starting mode, GM represents generator mode, and SGS represents starting generator system.

[0033] Figure 6 The node y within the task time T provided in the embodiments of the present invention 17 A schematic diagram of the GBIC result curve for the parent node combination;

[0034] Figure 7 The various time slices t provided in the embodiments of the present invention j Inner node y 17 A schematic diagram of the GBIC result curve for the parent node combination;

[0035] Figure 8 This is a schematic diagram of the LDBN structure of the SG system provided in an embodiment of the present invention; Figure 8 The numbers under the node symbols in the text mean: node x 1 , x 2 The two modes, Normal and Tloss, are represented by fuzzy numbers 0 and 1, respectively; nodes x 3 , x 4 , y 1, y The three modes of 2—Normal, Ploss (partial failure), and Tloss (complete failure)—are represented by fuzzy numbers 0, 0.5, and 1, respectively; nodes x 5, y i = 3, 4, …, 18 , Y The four modes of normal operation, mistake, partial failure (Ploss), and complete failure (Tloss) are represented by fuzzy numbers 0, 0.33, 0.66, and 1, respectively; the possible operating modes of a node can be represented by the numbers under its symbol.

[0036] Figure 9 A schematic diagram illustrating the result verification of the GBIC-based lightweight method provided in the embodiments of the present invention under a complete failure state;

[0037] Figure 10 This is a flowchart provided for an embodiment of the present invention. Detailed Implementation

[0038] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Embodiments of the present invention will be described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in the specification of the present invention means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or couplings. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0039] This embodiment requires the design of a lightweight DBN model for a multi-electric aircraft starting and generating system based on GBIC. The main design idea is to introduce grey system theory and integrate it with Bayesian information criteria to construct evaluation criteria for model fit and structural complexity. This allows for the elimination of redundant structures in complex polymorphic systems and simplification of connection relationships, resulting in a lightweight DBN (LDBN) model with significantly reduced computational complexity.

[0040] It should be noted that the dynamic Bayesian network mentioned in this embodiment refers to a probabilistic graphical model used to represent relationships between variables and to perform tasks such as prediction, diagnosis, and decision-making through probabilistic reasoning. It is an extension based on Bayes' theorem, consisting of nodes and directed edges. Nodes represent variables, and edges represent dependencies between variables. It is one of the most effective theoretical models in the field of uncertain knowledge representation and reasoning. The value of a node depends on the value of its parent node and its own conditional probability. It has the ability to describe the polymorphism and nondeterministic logical relationships of events. In the reasoning process of a Bayesian network, posterior probability, prediction, diagnosis, and decision-making can be used for analysis. Posterior probability refers to the probability of solving for other unknown events given some observed events; prediction refers to predicting future events given some observed events; diagnosis refers to inferring the cause of some abnormal events given certain occurrences; and decision-making refers to choosing the optimal decision based on some known actions and results. A Bayesian network (BN) is a directed acyclic graph consisting of nodes and directed edges. Nodes represent random variables, and directed edges represent conditional dependencies or causal relationships between these variables. Bayesian network analysis enables probabilistic reasoning by calculating the posterior probability of variables based on prior knowledge and observed evidence.

[0041] In Bayesian network analysis, a directed acyclic graph (DAG) is used to describe the conditional dependencies between variables. The basic BN model structure is as follows: Figure 1 As shown in part (a). The nodes x1, x2, ... x in the graph are shown. i Let y be random variables, x1, x2, ..., xn. i As the parent node of y, the joint distribution of the network is obtained by multiplying the probability distributions of the variables of each node in the network.

[0042] DBN (Dual Block Network) is an extension of BN (Breakthrough Normalization) in temporal logic, consisting of an initial BN and a transition network. The static BN model, by adding a time dimension, constitutes a probability distribution model capable of handling temporal information. DBN can be constructed by unfolding multiple time slices, where each slice corresponds to a static BN model. The structural dependencies between adjacent time slices are preserved, ensuring temporal consistency, such as... Figure 1 As shown in part (b).

[0043] In practical applications, the main construction methods for dynamic Bayesian networks can be divided into two categories: domain knowledge-based and data-driven. Domain knowledge-based methods include: expert knowledge methods, where experts determine the structure and parameters of the Bayesian network based on their domain knowledge and subjective judgment; and model structure learning methods, which automatically learn the structure of the Bayesian network from the perspective of model structure based on the characteristics of the domain model. Data-driven methods include: Bayesian learning, constraint satisfaction methods, minimum description length methods, and Gaussian graph model selection methods.

[0044] The construction of a Bayesian network can be divided into the following steps: 1) Determine the variables: First, it is necessary to determine the variables involved in the modeling, which can be continuous or discrete variables. 2) Construct the network structure: In a Bayesian network, nodes represent random variables, and edges represent the conditional dependencies between two variables. Therefore, it is necessary to determine the conditional dependencies between variables and construct the network structure. 3) Determine the probability distribution of the variables: After determining the network structure, it is necessary to determine the probability distribution of each variable. These probability distributions can be discrete or continuous. Methods for determining probability distributions include subjective assignment, maximum likelihood estimation, and Bayesian methods. 4) Model evaluation: After the model is built, it needs to be evaluated to determine its predictive performance. Commonly used methods include cross-validation, information criterion, and error assessment. 5) Model application: After the Bayesian network is built, it can be used to infer the conditional dependencies between variables and perform probabilistic inference, decision analysis, data mining, and other applications.

[0045] The Bayesian Information Criterion (BIC) is a model selection criterion derived from Bayesian theory. It uses Bayes' theorem to correct subjective probability estimates and, by simultaneously evaluating the model's likelihood and number of parameters, comprehensively considers the model's goodness of fit and complexity, thereby determining the most suitable model for a given dataset.

[0046] Grey system theory refers to the concept that a system is a whole composed of interrelated and constrained elements. Based on the degree of knowledge about these elements, systems are classified into three types: black systems, white systems, and grey systems. Black systems are complex, unknown systems whose internal characteristics are completely unknown and whose internal structure cannot be obtained. White systems are known systems whose internal characteristics and operating principles are fully understood, allowing for the establishment of precise mathematical models. Grey systems fall between black and white systems, with some internal characteristics known and others unknown. In theoretical analysis, grey systems can be described and understood by mining and analyzing known information, allowing for the acquisition of as much internal information as possible and the establishment of quantitative analysis models. However, in practical engineering applications, establishing an absolutely white system is often extremely difficult. Although mathematical statistics can be used to qualitatively or quantitatively analyze some factors influencing the system, it is difficult to exhaustively identify all factors related to the system, let alone determine the specific mapping relationships between these factors. In engineering practice, systems lacking definite mapping functions can be considered grey systems.

[0047] This embodiment is based on the above scheme and algorithm, with secondary design and significant improvements to the maintenance and inspection scheme applied to aviation equipment. This maintenance and inspection scheme is a lightweight version of the dynamic Bayesian network based on the Grey-Bayesian Information Criterion (GBIC). Specifically, to reduce redundancy in the DBN in complex multi-state systems, grey system theory is introduced and fused with the Bayesian Information Criterion to construct evaluation criteria for model fit and structural complexity. The redundancy between parent node combinations is compared, transforming the DBN model into a lightweight DBN (LDBN), thereby improving its computational efficiency.

[0048] The specific method is as follows:

[0049] 1. Calculation of maximum likelihood value for gray area: Calculate the maximum likelihood value for gray area. To evaluate the model's fit to the data. Here, N is the total number of nodes in the DBN; This represents the root node, and p is the number of root nodes; Let represent an intermediate node, q represent the number of intermediate nodes, and Y represent a leaf node. The task time T is divided into m time segments, each segment lasting t. j In this article, the subscript 'i' represents the node number, and 'j' represents the time slice number. For example, the i-th node x i The 'i' represents the node number; 't' jLet j represent the j-th time slice. Then, under the assumption of conditional independence, based on the conditional independence of the DBN nodes containing interval gray numbers, the joint probability is decomposed. Simultaneously, the interval gray number probability is determined by calculating the upper and lower bounds of the data separately, as follows:

[0050] ,

[0051] ;

[0052] in, and They represent time slices t and t respectively. j The upper and lower bounds of the gray fuzzy likelihood value, where j represents the time slice number. and Indicates time slice t j The two initial nodes in the middle, and Indicates time slice t j The set of parent nodes corresponding to the two initial nodes at the time of the node. and They represent time slices t and t respectively. j The gray fuzzy conditional probability values ​​corresponding to the two initial nodes are given by p, which represents the total number of root nodes x, and q, which represents the total number of intermediate nodes y.

[0053] 2. DBN Parameter Dimension Calculation: The parameter dimension R needs to be calculated from two levels: the parameter dimension of a node and the total parameter dimension of the network. The parameter dimension of a node is the number of parameters of a single node in the network. The parameter R is usually determined by the number of its parent nodes and its own number of states. The specific calculation method is as follows: , ,in, and Representing node x i and y i The number of parameters, Indicates the failure mode of the current node.

[0054] The total parameter dimension of a network is the sum of the parameters of all nodes in the entire Bayesian network, used to measure the overall complexity of the model. Therefore, the parameter dimension of a node in a DBN model can be obtained as follows:

[0055]

[0056] in, This represents the parameter dimension of a node. In GBIC calculation, this value is used to evaluate the complexity of the entire model and acts as a penalty term to influence the final model selection, preventing the model from becoming overly complex while maintaining a good fit.

[0057] 3. Grey Bayesian Information Criterion (GBIC) Calculation: The GBIC value is calculated by treating conditional probabilities as interval grey numbers, with its upper and lower bounds representing two extreme cases. If the conditional probability is an interval grey number, both the upper and lower bounds are used in the calculation. Otherwise, it is assumed that the probabilities of the upper and lower bounds are equal in the calculation. Node or The GBIC value is calculated as follows:

[0058]

[0059] in, and Representing nodes respectively and In time slice t j The GBIC value of the parent node combination at that time. and They represent the maximum likelihood values. and The upper and lower bounds.

[0060] According to the node and of and The upper and lower extrema of each parent node combination are taken as the node. and GBIC maximum value and ,in: , .

[0061] Based on this, considering the inherent uncertainty and interval characteristics of the interval gray number, a method capable of taking this uncertainty into account must be adopted to constrain the GBIC extrema. Therefore, constraints are introduced. Through this equation, the GBIC extrema can be reasonably defined. and The range of values ​​for is determined to ensure that it remains reasonable and effective while fully considering the uncertainty of the gray number within the interval. The constraints are as follows: , .

[0062] in, It is a time slice t j Middle node and The conditional probability value of the corresponding gray number in the interval; It corresponds to the gray number in the interval. The lower bound of the probability, It corresponds to the gray number in the interval. The upper limit of probability, and They are time slices t j Middle node and The maximum and minimum values ​​of the GBIC corresponding to the parent node traversal combination.

[0063] node and GBIC value at task time T and Will be obtained, represented as: , .

[0064] In this embodiment, the node conditional probability is represented by the gray number of intervals. This method uses the gray maximum likelihood value and parameter dimension to evaluate the model's fit and complexity. By integrating GBIC and parameter dimension, this method can comprehensively identify the optimal DBN model within task time T, eliminating redundant node relationships and thus achieving model lightweighting.

[0065] Furthermore, the lightweight logical architecture of the DBN model for multi-electric aircraft starting and generating systems based on GBIC is as follows: Figure 2 As shown, it mainly includes:

[0066] Step 1: Constructing a DBN model for complex polymorphic systems

[0067] First, based on the working mechanism and structural composition of the complex multimorphic system, the failures and characteristics of key components are identified, and the relationship between key components and system failure modes is clarified. Then, based on the working mechanism and failure modes of the complex multimorphic system, a network structure of a BN (Browser-Negative) model is constructed to accurately reflect the functional dependencies and failure propagation within the system. Finally, the BN model is transformed into a DBN (Devices-Based-Negative) model using time slices and task time, thus establishing the DBN model for the complex multimorphic system.

[0068] Step 2: Construct the Grey Conditional Probability Table (GCPT)

[0069] Based on the fault modes and working mechanisms of complex polymorphic systems, gray numbers from the gray system theory interval are introduced as the foundation of the DBN model. By combining this with maintenance manuals and fault isolation manuals for complex polymorphic systems, the interrelationships between nodes in the DBN model can be expressed. In particular, for the fuzzy and uncertain interrelationships existing in complex polymorphic systems, it can effectively characterize the structural relationships of the system content.

[0070] Step 3: Establish an LDBN model based on GBIC

[0071] A lightweight method for dynamic Bayesian networks based on GBIC incorporates grey system theory and integrates it with the BIC criterion. First, the grey maximum likelihood value is calculated based on the DBN model to measure the fit between the DBN model and the data. Second, the parameter dimension of the DBN is calculated as a measure of model complexity to prevent overfitting. Finally, based on the grey maximum likelihood value and parameter dimension, the GBIC of the parent node combinations corresponding to each node in the DBN model is obtained. Combined with the conditional constraint of the interval grey number, the results of each parent node combination are obtained, enabling the identification of redundant structures in the DBN model and improving model computational efficiency.

[0072] The effectiveness of the proposed GBIC-LDBN method will be verified through specific application examples below:

[0073] This study focuses on the starter-generator (SG) system of a multi-electric aircraft as a complex multi-state system. The SG system is a crucial component of the multi-electric aircraft's power system and is subject to various failure states caused by factors such as electrical faults, mechanical faults, environmental influences, and improper maintenance. Furthermore, the SG system is affected by load variations and environmental conditions, resulting in a dynamic operating state. Detailed schematic diagrams of its key components and working mechanisms are shown below. Figure 3 As shown. Its operation can be divided into two modes: start-up mode and power generation mode, as follows. Figure 3 part (a) and Figure 3 As shown in section (b). The failure of the starting and generating system in a multi-electric aircraft is mainly caused by the failure of key components. Based on the mechanism and main components of the multi-electric aircraft starting and generating system, the risk-causing factors can be identified as the failure of key components, including the main generator failure, exciter failure, excitation contactor failure, starting contactor failure, rectifier bridge failure, various sensor failures, power control unit failure, rotating rectifier failure, and voltage regulation circuit failure. Therefore, these key components are classified as follows: functional execution components responsible for directly controlling the starting / generating operation; information collection components responsible for collecting information such as the operating voltage and power of generators, engines, etc.; and information processing components responsible for summarizing and processing the collected information, such as... Figure 4 As shown in Table 1, based on the failure modes of different types of critical components, the failure states of critical components are classified into the following categories: Normal, Mistake, Partial Failure (Ploss), and Total Failure (Tloss). The classification of system states is based on the actual situation of the subsystems.

[0074] Table 1 Classification of Failure Modes of Starting and Generating Systems

[0075]

[0076] Through in-depth analysis of the working mechanism and failure modes of key components of the SG system, its DBN model was constructed, such as... Figure 5 The results are shown in parts (a) and (b). The mission time T is divided into three time slices, and the CPT of the DBN nodes containing the interval gray numbers is determined according to the aircraft maintenance manual and the fault isolation manual. The DBN model considers the common cause failure problem among critical components, characterizes the interaction between components and the fault propagation path within the system, and effectively reflects the dynamic characteristics of the system under different operating states. In the DBN model, nodes represent the critical components of the SG system and their operating states, and directed edges describe the relationships between nodes. Therefore, the DBN model of the SG system clearly shows the fault propagation process between components, as well as the impact of changes in the operating state of each component on system performance and the dynamics between related components. The names of the relevant critical components are shown in Table 2.

[0077] Table 2 Key Components and Operating Modes of the SG System

[0078]

[0079] Taking node SCU1 as an example, when node SCU1 fails, the DBN model can quickly identify two other components, EXG1 and SG1, that may be affected by the common cause failure, as well as the extent of the failure's impact on the entire system. This DBN-based modeling method provides strong support for fault diagnosis and maintenance decisions. Furthermore, the DBN model can describe the dynamic changes of the system, thereby improving its accuracy and reliability in assessing the operating status of the SG system.

[0080] The lightweight method for dynamic Bayesian networks based on GBIC aims to reduce the redundancy of DBNs in SG systems. For example... Figure 4 As shown in Table 2, the failure modes of each node in the DBN model were first clarified. The failure modes and working mechanism of the SG system were analyzed, and the power supplies (PWR1 and PWR2) were defined to have two failure modes, denoted as x. i = [0, 1], i = 1, 2. The Signal Feedback Sensor (SFS), Start-up Rule Calculation Sensor (SLS), and Computer Control Unit (DCU) have three fault modes, denoted as x. i y1, y2 = [0, 0.5, 1], i = 3, 4. The remaining nodes exhibit four failure modes, represented as x5, y6, y7, y8, y9, y1, y2 ... i Y = [0, 0.33, 0.66, 1], i = 3, 4, …, 18. Taking the CPT of nodes y1 and Y containing interval gray numbers as an example, as shown in Tables 3 and 4. Each row in the tables represents the time slice t. jWhen the parent node is in different fault states, the conditional probability of the child node failing.

[0081] Table 3. CPT of node y1 containing interval gray numbers at time slice t1

[0082]

[0083] Table 4. CPT of node Y containing interval gray numbers at time slice t1

[0084]

[0085] Due to the complex structure of the SG system and the existence of multiple fault states, it is necessary to eliminate redundant structures in the DBN to improve model accuracy. Therefore, for intermediate nodes composed of two or more parent nodes, the GBIC results of traversing and combining their parent nodes are calculated to identify redundant structures in the DBN model. The CPT of a node is the fundamental support for calculating the redundant structures of each intermediate node. By comparing and analyzing the GBIC results within task time T, it can be found that node y 17 There is redundant structure among the corresponding parent node combinations, as shown in Table 5.

[0086] Table 5 Node y 17 GBIC value of parent node combination

[0087]

[0088] like Figure 6 As shown, the GBIC result of the parent node combination of node y17. The overall trend is upward within the task time T. Among them, the GBIC results of combination 1 are... The lowest range [337.50, 370.77] indicates that this combination can more effectively characterize node y. 17 The state. Generally, when the difference in GBIC results exceeds 4, the model with the lower GBIC result has a better association. Therefore, among all parent node combinations, combination 1 (node ​​y) is the best. 13 and y 14 Combination 11 exhibits the best model fit. In contrast, other combinations show higher redundancy, especially combination 11, which has a GBIC result as high as [3650.45, 3689.26]. This value is significantly higher than other combinations, making it difficult to accurately represent the relationships between nodes.

[0089] To verify the effect of parent node combination on node y within task time T. 17 The impact was analyzed, and the GBIC changes of each parent node combination within the three time slices were examined, such as... Figure 7As shown in the figure. The results indicate that the GBIC variation trend of the parent node combination in each time slice is relatively consistent, and the overall trend is consistent with... Figure 5 Consistency indicates that each parent node combination affects node y at different task stages. 17 The influence of this is stable. It is worth noting that combination 1 (i.e., node y) 13 and y 14 The value remains at an extremely low level throughout the task time T. This indicates that the parent node combination can effectively characterize node y. 17 The state changes are reduced, and the complexity of the DBN model is reduced, resulting in a lightweight DBN model (LDBN).

[0090] In summary, a lightweight method based on GBIC is used to analyze the SG system model, and node y 17 The GBIC values ​​of the parent node combination show a consistent trend across different time slices. Specifically, node y... 13 and y 14 The combination of these elements consistently yields the lowest GBIC value, indicating that this parent node combination effectively characterizes node y. 17 The state changes. In contrast, the GBIC values ​​of other combinations are relatively high, especially combination 11, whose GBIC value is significantly higher than other combinations, indicating that there is a certain degree of redundancy in the association between nodes. Therefore, node y is selected. 13 and y 14 As node y 17 The parent node is more reasonable, reducing the complexity of the DBN model while improving its computational efficiency. The optimized LDBN model is as follows: Figure 8 As shown.

[0091] Finally, the effectiveness of the GBIC-based lightweight method was validated: To verify the effectiveness of the GBIC-based lightweight method, the most probable fault states (Tloss) of the SG system were studied, and a multi-dimensional comparative analysis was conducted, such as... Figure 9 As shown, on the one hand, the LDBN model using the lightweight method based on GBIC significantly improves computational efficiency compared to the DBN model, reducing execution time by 38.9%. On the other hand, the system's accuracy in identifying the most probable fault state remains unchanged. Throughout the task time, the accuracy of LDBN's gray fuzzy probability remains at 0.97, and the upper and lower limit deviation rates do not exceed 0.02 within the accuracy range. These results confirm that the lightweight method of dynamic Bayesian networks based on GBIC can significantly reduce model complexity while maintaining computational accuracy.

[0092] In summary, this embodiment addresses the problem of complex and inefficient inference calculations caused by the rapid expansion of the model size and a significant increase in the dimension of the conditional probability table during the application of dynamic Bayesian analysis methods for complex polymorphic systems. It introduces grey system theory and Bayesian information criteria to achieve lightweighting of dynamic Bayesian networks for complex polymorphic systems, eliminate redundant structures, and ensure the accuracy of the results.

[0093] A lightweight computational method for dynamic Bayesian networks based on GBIC is constructed. By organizing the parent node combinations of each node and judging the fit between time slices and task time, the optimal combination of parent node combinations is selected, and redundant node structures are eliminated, thus achieving lightweight dynamic Bayesian networks. Furthermore, by combining grey system theory with the DBN method, the problem of describing various fault states of complex polymorphic systems, caused by numerous interacting components, complex relationships, and diverse operating environments, is effectively solved. Interval grey numbers are used to describe the conditional probabilities in the DBN model, accurately describing the fault relationships between components in complex polymorphic systems.

[0094] Based on the specific experimental cases mentioned above, this embodiment uses the complex polymorphic system of a multi-electric aircraft starter-generator system as the research object. Through the failure faults and working mechanisms of the starter-generator system, a DBN model structure is constructed as the support for subsequent models. For this complex polymorphic system DBN model, the GBIC-LDBN method is used to remove redundant structures from the DBN model, constructing a lightweight DBN model (LDBN) for the starter-generator system. The advantages of LDBN and DBN models in terms of computational efficiency and accuracy are verified, demonstrating that LDBN can effectively improve computational efficiency while ensuring computational accuracy.

[0095] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The above descriptions are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A lightweight method for DBN model of multi-electric aircraft starting and generating system based on GBIC, characterized in that, include: S1. Establish a DBN model for the starting and generating system, wherein the nodes in the DBN model correspond to the key components and operating states of the starting and generating system; S2. Identify redundant structures in the DBN model and perform lightweight processing; S3. Using the lightweighted DBN model, identify the fault status of the starting and generating system.

2. The method according to claim 1, characterized in that, The operating states of the starting and power generation system include: starting mode and power generation mode; The fault states of the key components of the starting and power generation system include: normal operation, abnormal operation, partial failure, and complete failure.

3. The method according to claim 1, characterized in that, The DBN model of the starting and generating system includes: , Let x represent the state transition probability of node x from time slice t to time slice t+1. t Indicates the node state at time slice t, x t+1 This represents the state of a node at time t+1, where t represents the time slice, i represents the node's corresponding number, and x represents the node's position. i t This represents the node state of the i-th node at time slice t, x i t+1 This represents the state of the i-th node in time slice t+1, and N represents the total number of nodes.

4. The method according to claim 1 or 3, characterized in that, S2 include: Obtain the goodness of fit of the DBN model to the data and the parameter dimensions of the nodes in the DBN model; Calculate the GBIC value of the parent node combination in the DBN model, where the magnitude of the GBIC value is positively correlated with the structural redundancy of the DBN model.

5. The method according to claim 4, characterized in that, Obtaining the goodness of fit of the DBN model to the data includes: ， ； in, and Representing time slice t respectively j The upper and lower bounds of the gray fuzzy likelihood value, where j represents the time slice number. and Indicates time slice t j The two initial nodes in the middle, and Indicates time slice t j The set of parent nodes corresponding to the two initial nodes at the time of the node. and Representing time slice t respectively j The gray fuzzy conditional probability values ​​corresponding to the two initial nodes are given by p, which represents the total number of root nodes x, and q, which represents the total number of intermediate nodes y.

6. The method according to claim 5, characterized in that, Obtaining the parameter dimensions of nodes in the DBN model includes: ,in, The parameter dimension of a node, and Representing node x i and node y i The number of parameters.

7. The method according to claim 6, characterized in that, The calculation of the GBIC value of the parent node combination in the DBN model includes: Calculate the upper and lower extrema of the parent node combination in the DBN model to obtain the range of GBIC values, where: , ; and Representing nodes respectively and The upper and lower extrema at task time T. and The time slices t and t represent the time time of the task time T, respectively. j Nodes in and The maximum and minimum values ​​of GBIC corresponding to the combination of parent nodes traversed, where m represents the time slice t within the task time T. j The total number.

8. The method according to claim 7, characterized in that, ， ， , ,in, Indicates time slice t j The conditional probability value that the internal node contains the gray number in the interval. Indicates correspondence The lower bound of the probability, Indicates correspondence The upper limit of probability, r represents the conditional probability value number containing the gray numbers in the interval, and n represents the time slice t. j The middle node contains the total number of conditional probability values ​​for the gray number interval.

Images