Fault Diagnosis Method and Device for Air Conditioning Systems Based on Parameter Correlation Bayesian Networks

Abstract

This invention proposes a fault diagnosis method and device for air conditioning systems based on parameter-correlated Bayesian networks. Utilizing the correlation between material parameters of components, it quickly determines the real-time values ​​of key performance parameters of the components and compares them with the calibration values ​​at the time of component delivery to determine the real-time failure rate of various component faults. Combining the Bayesian network model of the air conditioning system, it sequentially calculates the real-time failure rate and real-time posterior probability of each level of sub-nodes, and determines whether a node is faulty by comparing the difference between the statistical posterior probability and the real-time posterior probability. This method is convenient for real-time data measurement and processing, and highly operable. The use of Bayesian networks and posterior probability for fault determination makes the diagnostic process more flexible and the fault location more accurate. With slight adjustments, it can be applied to the fault diagnosis of various models of air conditioning systems, demonstrating strong practical generalization.

Description

Technical Field

[0001] This invention relates to a method and apparatus for fault diagnosis of air conditioning systems based on parameter-correlated Bayesian networks, belonging to the field of reliability analysis technology. Background Technology

[0002] Due to the harsh working conditions of construction machinery, air conditioning systems experience complex and variable external stresses, leading to various failure modes. However, the complex structure and numerous components of air conditioning systems make fault location difficult during diagnosis, delaying repairs, impacting project schedules, and even posing safety risks. Therefore, research into fault detection and diagnosis methods for air conditioning systems is of great significance.

[0003] Different types of construction machinery operate under varying conditions, resulting in significant differences in the external stresses experienced by their air conditioning systems. This leads to diverse failure modes, and fault diagnosis models suffer from poor portability across different vehicle models, hindering their widespread application. Furthermore, the correlation between component performance parameters and the random nature of component performance degradation mean that system failures are often caused by the combined effects of multiple components, making it difficult to accurately pinpoint the weakest link. Currently, there is a lack of fault diagnosis models that consider random degradation processes and parameter correlations, are readily portable for different air conditioning systems, and can provide accurate fault diagnosis results.

[0004] The main existing technical solutions are as follows:

[0005] 1. Troubleshooting method:

[0006] Troubleshooting is a method of gradually eliminating factors that may cause system failure, narrowing down the scope of the problem, and ultimately finding and fixing it. Its basic process involves identifying the problem and collecting relevant information, then conducting data analysis and troubleshooting, and finally verifying the problem through experiments.

[0007] However, troubleshooting methods require a high level of experience from the troubleshooters, and the process, from information collection and analysis to troubleshooting and experimental verification, is quite lengthy, resulting in low troubleshooting efficiency and high maintenance costs. For troubleshooting and locating more complex systems, the feasibility is poor.

[0008] 2. Artificial intelligence technology:

[0009] Artificial intelligence technology is applied to system fault diagnosis and localization, with machine learning and data mining being commonly used techniques. By learning from large amounts of fault data and identifying fault patterns, and by analyzing and modeling system fault data, intelligent diagnosis of potential faults and prediction of equipment lifespan can be achieved.

[0010] However, the effectiveness of artificial intelligence technology in fault diagnosis is greatly affected by the amount of training sample data, thus requiring a large amount of data. For sample data with limited quantity and high randomness, it is difficult to obtain effective fault diagnosis results.

[0011] Real-time monitoring technology

[0012] By monitoring the system's operating status in real time and issuing timely alarms when abnormal data is detected in the equipment operation according to the system's preset rules or thresholds, the system analyzes and diagnoses the real-time collected data, identifies possible causes of equipment failures, and helps technicians take timely countermeasures.

[0013] However, real-time monitoring technology is suitable for equipment status monitoring and fault diagnosis in fixed scenarios. When applied to outdoor equipment with variable operating conditions and harsh environments, the reliability of the monitoring system is difficult to guarantee, the application cost is high, and false status reports may also occur. Summary of the Invention

[0014] This invention belongs to the field of reliability analysis technology. Specifically, it addresses the problem that the complex and variable system structure and random degradation of components in the fault diagnosis and analysis of air conditioning systems for engineering machinery make the fault reasoning and analysis process complicated and difficult to locate faults. This invention proposes a fault diagnosis method and device for air conditioning systems based on parameter-dependent Bayesian networks to obtain good fault diagnosis results, and it can be quickly transferred and applied to air conditioning systems with different structures.

[0015] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0016] In a first aspect, the present invention provides a fault diagnosis method for parameter-dependent Bayesian networks, comprising the following steps:

[0017] Step 1: Determine the fault type of the air conditioning system of the construction machinery. The same fault type corresponds to the same fault event in the fault tree. Use the fault event as the base event of the fault tree model to build the fault tree model of the air conditioning system.

[0018] Step 2: Map the fault tree model to a Bayesian network;

[0019] Step 3: Combine historical fault data to calculate the probability of occurrence of the bottom event in the fault tree model, and use it as the prior probability of the leaf node of the Bayesian network corresponding to the bottom event of the fault tree model, so as to obtain the prior probability of each leaf node of the Bayesian network.

[0020] Step 4: Calculate the posterior probability of each level of child nodes in the Bayesian network based on the Bayesian network mapped by the fault tree model and the prior probabilities of each leaf node in the Bayesian network.

[0021] Step 5: Obtain the regression model between the performance-related parameters of the components; the input of the regression model between the performance-related parameters of the components are easily measurable performance-related parameters of the components, and the output is the real-time key performance parameters of the components;

[0022] Step 6: Measure the real-time data of the performance-related parameters of the components, and combine the regression model between the performance-related parameters of the components to obtain the real-time data of the key performance parameters of the components.

[0023] Step 7: Obtain a real-time failure rate assessment model for multiple faults of components. The input variables of the real-time failure rate assessment model are the factory calibration values ​​of key performance parameters of components, the real-time predicted values ​​of key performance parameters of components output by the relevant parameter regression model, and the market statistical failure rate of components. The output variable is the real-time failure rate of components under the corresponding failure mode.

[0024] Step 8: Obtain the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components. Combine the real-time data of key performance parameters of components output by the regression model to calculate the real-time failure rate of each level of sub-nodes in the Bayesian network. Use the real-time failure rate of each level of sub-nodes as the real-time prior probability to calculate the real-time posterior probability of each level of sub-nodes in the Bayesian network.

[0025] Step 9: Compare the real-time posterior probability and posterior probability of each level of child nodes in the Bayesian network to diagnose the fault status of each level of child nodes and obtain the fault diagnosis results.

[0026] Step 10: Output the fault diagnosis results.

[0027] Furthermore, the engineering machinery can be any one of mining trucks, loaders, excavators, or cranes.

[0028] Further, step 1: Determine the fault type of the air conditioning system of the construction machinery. The same fault type corresponds to the same fault event in the fault tree. Use the fault event as the base event of the fault tree model to build the fault tree model of the air conditioning system, including:

[0029] In the fault tree model, the fault of the air conditioning system is regarded as the top event, the faults of each subsystem are regarded as first-level intermediate events, the faults of the components are regarded as second-level intermediate events, the faults of the parts are regarded as third-level intermediate events, and all fault events constitute the bottom event of the fault tree model.

[0030] Further, step 2: mapping the fault tree model to a Bayesian network includes:

[0031] When a fault tree model is converted into a Bayesian network, the structure of the fault tree model corresponds to that of the Bayesian network. Each level of fault event in the fault tree model corresponds to a node in the Bayesian network: the bottom events of the fault tree model correspond to the leaf nodes of the Bayesian network, and the first-level intermediate events, second-level intermediate events, and third-level intermediate events in the fault tree model correspond to the first-level child nodes, second-level child nodes, and third-level child nodes of the Bayesian network, respectively.

[0032] The connections between fault events in the fault tree model correspond to the edges in the Bayesian network. The logic gates that connect events in the fault tree model are replaced with the conditional probability tables of the Bayesian network. The prior probabilities of the leaf nodes in the Bayesian network correspond to the occurrence probabilities of the bottom events in the fault tree model.

[0033] Further, step 3: Combining historical fault data, calculate the occurrence probability of the bottom event in the fault tree model, and use it as the prior probability of the corresponding leaf node of the Bayesian network for the bottom event in the fault tree model, thus obtaining the prior probability of each leaf node of the Bayesian network, including:

[0034] By combining historical fault data, the probability of occurrence of the bottom event in the fault tree model is calculated, and this probability is used as the prior probability of the corresponding leaf node in the Bayesian network.

[0035] The historical fault data includes all fault types in the air conditioning system and the frequency of each fault type.

[0036] The probability of occurrence of the underlying events in the fault tree model is the probability of occurrence of various fault types obtained after statistical analysis of the fault data, and it is used as the prior probability of the corresponding leaf node of the Bayesian network.

[0037] Further, step 4: Based on the Bayesian network mapped by the fault tree model and the prior probabilities of each leaf node in the Bayesian network, calculate the posterior probabilities of each level of child nodes in the Bayesian network, including:

[0038] In Bayesian networks, conditional probabilities are used to express the logical relationships between nodes. The logic gates in the fault tree model correspond to the conditional probability tables in the Bayesian network.

[0039] If node Connect n child nodes , by express The probability of a node failing is expressed as:

[0040] (1)

[0041] Then node In the event that occurs, the i-th child node The formula for calculating the posterior probability of occurrence is as follows:

[0042] (2)

[0043] in, express The i-th child node connected to the node The prior probability, , where n is the total number of child nodes of node B; Represents child nodes In the event of a node The conditional probability of occurrence Indicates the condition that node B occurs. The conditional probability of occurrence, i.e., the i-th child node The posterior probability of occurrence;

[0044] n child nodes and The nodes correspond to leaf nodes and third-level child nodes, respectively; , where n is the total number of child nodes of node B; calculate the posterior probability of all leaf nodes and the probability of a third-level child node occurring;

[0045] third-level child nodes As a higher-level second-level child node The bottom event, if the second-level child node Connects m third-level child nodes ,by Represents a node The probability of a failure can be obtained from the above steps. In the event that occurs, the j-th child node The posterior probability of occurrence, where m child nodes and nodes These correspond to third-level child nodes and second-level child nodes, respectively. The posterior probability of all third-level child nodes and the probability of a second-level child node occurring are calculated.

[0046] By treating the second-level child node C as the base event of the higher-level child node, the posterior probabilities of all leaf nodes, third-level child nodes, second-level child nodes, and first-level child nodes of the compressor system are obtained sequentially through the above steps.

[0047] Further, step 5: Obtain the regression model between the performance-related parameters of the components, including:

[0048] Obtain test data on the performance parameters of components and establish a performance parameter database;

[0049] Based on the performance parameter database, the parameter pairs that have a correlation are analyzed to obtain the correspondence between easily measurable component performance-related parameters and key performance parameters of component materials, thus obtaining the correlation parameter pairs consisting of easily measurable component performance-related parameters and key performance parameters of component materials.

[0050] Based on the detection data of parameter pairs consisting of easily measurable component performance parameters and key component material performance parameters, a regression model between the relevant parameters is established using support vector machine regression analysis:

[0051]

[0052] Among them, variables Easily measurable component performance-related parameters are used as model input variables. Key performance parameters of component materials, which correspond to easily measurable performance parameters of the components, are used as output variables of the model.

[0053] Furthermore, the relevant parameter pairs consisting of easily measurable component performance-related parameters and key component material performance parameters include:

[0054] For the same material, there are usually multiple pairs of relevant parameters. Based on the actual problem requirements, degradation data of the relevant parameter pairs in the performance parameter database are used to establish corresponding regression models. The regression models established for different pairs of relevant parameters for the same material or different materials are independent of each other.

[0055] The performance-related parameters of commonly used materials for components are shown in Table 4 below:

[0056] Table 4: Summary Table of Material-Related Parameters for Compressor Components

[0057] Material Relevant parameters Metals (steel, copper, aluminum) Hardness - tensile strength, toughness - shear modulus, temperature - electrical conductivity Plastics, rubber Shear modulus-toughness oilseeds Viscosity - specific heat capacity, density - thermal conductivity

[0058] Further, step 6: Measure real-time data of component performance-related parameters, and combine this data with the regression model among these parameters to obtain real-time data of the component's key performance parameters, including:

[0059] For key performance parameters of components that are difficult to measure, regression models between the corresponding parameters can be obtained by using the correspondence between easily measurable performance parameters of components and key performance parameters of component materials.

[0060] By monitoring the easily measurable material parameters of the components in real time, real-time data on the performance-related parameters of the components can be obtained.

[0061] Based on real-time data of component performance-related parameters, a regression model of relevant parameters is used to calculate and obtain real-time data of key component performance parameters.

[0062] Further, step 7: Obtain a real-time failure rate assessment model for various component faults, including:

[0063] The prior probabilities of various failure modes of components are determined by the statistical results of historical failure data, covering various operating conditions of all products throughout their entire life cycle, and are the expected values ​​of failure rates.

[0064] By utilizing real-time data of key performance parameters of components and combining them with their factory calibration values, a real-time failure rate assessment model is established to calculate the real-time failure rate of various faults in the components.

[0065] If the factory-calibrated value and the real-time predicted value under actual operating conditions of the key performance parameter corresponding to a certain failure mode of a component are respectively... and The market statistics for this failure mode show a failure rate of [missing information]. The real-time failure rate The evaluation model is represented as follows:

[0066] (3)

[0067] The input variables of the real-time failure rate assessment model are the factory calibration values ​​of key performance parameters of components, the real-time predicted values ​​of key performance parameters of components output by the relevant parameter regression model, and the market statistical failure rate of components. The output variable is the real-time failure rate of components under the corresponding failure mode.

[0068] Further, step 8: Obtain the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components. Combine this with the real-time data of real-time key performance parameters of components output by the regression model to calculate the real-time failure rate of each level of the Bayesian network's sub-nodes. Using the real-time failure rate of each level of the sub-nodes as the real-time prior probability, calculate the real-time posterior probability of each level of the sub-nodes in the Bayesian network, including:

[0069] Obtain the factory calibration values ​​of key performance parameters of components and the market statistics failure rate of components;

[0070] Based on the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components, combined with the real-time data of key performance parameters of components output by the regression model, the real-time failure rate of various failures of components is calculated based on the real-time failure rate evaluation model, that is, the real-time failure rate of leaf nodes in the Bayesian network.

[0071] Real-time failure rate using one type of failure mode of a component Based on the logical relationships between the nodes of the Bayesian network, the real-time failure rate of the components is determined. :

[0072] by Indicates the real-time failure rate of components , These represent the real-time failure rates of the k-th failure mode, respectively. ,but:

[0073] (4)

[0074] in This indicates the component ( ) when the k-th failure mode occurs in real time. The probability of a failure occurring in real time.

[0075] By combining the various faults of components and their real-time failure rates, and using the posterior probability calculation formula, the real-time posterior probabilities of various fault modes of components are calculated, i.e., the real-time posterior probabilities of leaf nodes are obtained. Among l fault modes, the real-time posterior probability of the k-th fault mode is... The calculation formula is:

[0076] (5)

[0077] in, Indicates components ( Given that a fault occurs in real time, the probability of the kth fault mode occurring in real time among l fault modes;

[0078] Based on the real-time failure rate of components By understanding the logical relationships between the nodes of the Bayesian network, the real-time failure rate of this component can be determined. And the real-time posterior probability of the component, the real-time posterior probability of the component corresponds to the real-time posterior probability of the third-level child node.

[0079] By analogy, the real-time failure rate and real-time posterior probability of the second-level sub-nodes and the real-time failure rate and real-time posterior probability of the first-level sub-nodes are calculated to obtain the real-time failure rate and real-time posterior probability of each level of sub-nodes, which can be applied to the system's fault diagnosis and location.

[0080] Further, step 9: Compare the real-time posterior probabilities and posterior probabilities of each level of child nodes in the Bayesian network to diagnose the fault status of each level of child nodes and obtain fault diagnosis results, including:

[0081] Based on the obtained real-time posterior probability of failure at each level of child nodes. By comparing real-time posterior probabilities at each level Posterior probabilities of child nodes at each level in a Bayesian network The relative value is used to calculate the degree of abnormality in the failure rate, δ.

[0082] (6)

[0083] Based on the degree of failure rate anomaly δ and the failure rate anomaly threshold at each level of node To determine the node where the fault occurred, This represents the abnormal threshold of the failure rate for the j-th type of fault in the i-th level child node.

[0084] Furthermore, based on the degree of failure rate anomaly δ and the failure rate anomaly threshold of each node... To determine the node where the fault occurred, including:

[0085] When the system fails When an incident occurs, begin troubleshooting with the next highest-level event, starting with the system components that have the highest failure rate. The investigation begins by comparing the real-time posterior failure rate of system components with the relative difference between the statistical posterior failure rate and any abnormalities. If the abnormality of the failure rate of a sub-node exceeds the abnormal failure rate threshold, then the sub-node is determined to be a faulty node.

[0086] If this system component If a node is not faulty, then low-fault nodes of the same level are checked in descending order of real-time failure rate.

[0087] If this system component is a faulty node, then its child nodes and the next level child nodes of its child nodes should be checked sequentially according to the node level of the Bayesian network.

[0088] Further, step 10: Output the fault diagnosis results, including:

[0089] Obtain the fault diagnosis and analysis results for each level of sub-nodes in the system. The output results are presented in the form of statistical tables or logical structure diagrams, including the historical fault rate, real-time fault rate, fault rate anomaly threshold reference value, and fault diagnosis result information for each level of node in the system, and generate and output an output report.

[0090] Secondly, the present invention provides a fault diagnosis device based on parameter correlation Bayesian networks, comprising:

[0091] Memory, used to store computer programs / instructions;

[0092] A processor for executing the computer program / instructions to implement the steps of the method described in the first aspect.

[0093] Thirdly, the present invention provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0094] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0095] (1) Compared with other troubleshooting methods, this invention uses Bayesian networks for fault diagnosis, which is highly operable and efficient. By using the statistical analysis results of historical fault data as the prior probabilities of each sub-node of the system, it is easy to obtain preliminary information on relatively weak components and nodes. It does not require much experience from maintenance personnel and is highly operable. Combined with fault diagnosis rules, through parameter correlation regression models and component status assessment models, and combined with Bayesian posterior probability formulas, the fault location can be quickly determined, which is highly efficient.

[0096] (2) Compared with artificial intelligence technology, this invention uses historical fault data as a benchmark value and real-time measurement data of key performance parameters of components as the application value for fault diagnosis. It introduces the concept of real-time failure rate and combines it with the critical importance of components to determine the fault, which has better flexibility. It has good generalization ability in dealing with system fault diagnosis problems with strong randomness in the degradation process and harsh and changeable working environment.

[0097] (3) Compared with real-time monitoring technology, this invention is based on parameter correlation to quickly predict and apply the key performance parameters of components, achieving good equipment status monitoring effect with low monitoring cost, and has good operability in harsh outdoor environments. Attached Figure Description

[0098] Figure 1 This is a flowchart of the technical solution;

[0099] Figure 2 This is a Bayesian network topology diagram of an air conditioning system.

[0100] Figure 3 This is a fault tree model diagram of an air conditioning compressor system. Detailed Implementation

[0101] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0102] Example 1:

[0103] This embodiment proposes a fault diagnosis method for air conditioning systems based on parameter-correlation Bayesian networks. Taking the fault diagnosis of a mining machine air conditioning system as an example, the following technical solution steps are adopted to implement it, as follows: Figure 1 As shown:

[0104] (1) By utilizing the correlation between the material parameters of the components, the real-time values ​​of the key performance parameters of the components can be quickly determined, and compared with the calibration values ​​of the components at the time of delivery, combined with the real-time performance status evaluation model, the real-time failure rate of various faults of the components can be determined.

[0105] (2) Combine the real-time failure rates of various component failures with the Bayesian network model of the air conditioning system, and calculate the real-time failure rates of each level of sub-nodes in turn.

[0106] (3) Calculate the real-time posterior probability based on the real-time failure rate of each sub-node, and determine whether it is a fault node by comparing the difference between the statistical posterior probability and the real-time posterior probability and using the anomaly threshold.

[0107] Specifically, the method in this embodiment includes the following steps:

[0108] Step 1: Determine the fault type of the mine truck air conditioning system and build a fault tree model;

[0109] Specifically, the failure of the air conditioning system is taken as the top event, the failures of each subsystem are taken as first-level intermediate events, the failures of components are taken as second-level intermediate events, the failures of parts are taken as third-level intermediate events, and all the above failure events constitute the bottom event of the model.

[0110] The corresponding faults in the subsystems are: cracking, damage, and leakage of the refrigerant pipe assembly; damage, breakage, blockage, and jamming of the indoor unit assembly; and damage, jamming, and breakage of the compressor.

[0111] The corresponding faults of the components are: hot water pipe assembly falling off or breaking; condenser fan damage; thermostat assembly burning out or open circuit, etc.

[0112] Component failures include bearing wear and breakage, loose tension wheel, clutch damage, broken fan blades, core wear, loose or detached ferrules, broken pipes, loose joints, and aging seals.

[0113] Step 2: Map the fault tree model to a Bayesian network;

[0114] Specifically, when converting a fault tree model to a Bayesian network, the structure of the fault tree model corresponds to that of the Bayesian network, and the fault events at each level in the fault tree model correspond to the nodes at each level in the Bayesian network: the base events in the fault tree model correspond to the leaf nodes in the Bayesian network, and the first, second, and third intermediate events in the fault tree model correspond to the first, second, and third child nodes in the Bayesian network, respectively. The connections between fault events in the fault tree model correspond to the edges in the Bayesian network, and the logic gates connecting events in the fault tree model are replaced with the conditional probability tables of the Bayesian network. The prior probabilities of the leaf nodes in the Bayesian network correspond to the occurrence probabilities of the base events in the fault tree model.

[0115] Figure 2 This is a Bayesian network topology diagram of an air conditioning system.

[0116] Step 3: Combine historical fault data to calculate the probability of occurrence of the bottom event in the fault tree model, and use it as the prior probability of the corresponding leaf node of the Bayesian network;

[0117] Specifically, by combining historical fault data, the occurrence probability of the bottom event in the fault tree model is calculated and used as the prior probability of the corresponding leaf node in the Bayesian network. The occurrence probability of the bottom event in the fault tree model is the probability of occurrence of various faults obtained after statistical analysis of historical monitoring data of the air conditioning system. Since this invention has many subsystems, the occurrence probability of the bottom event in the air conditioning compressor system is taken as an example, as shown in Table 1; the conditional probabilities of other subsystems are not listed.

[0118] Table 1: Statistical Table of Probability of Basic Events in Air Conditioning Compressor System

[0119] code Bottom Event Probability of occurrence (times / 1000h) code Bottom Event Probability of occurrence (times / 1000h) A1 Bearing wear 0.117 A8 Compressor internal leakage 0.014 A2 bearing breakage 0.019 A9 Controller burnt out 0.017 A3 Sealing aging 0.012 A10 Loose wire harness 0.043 A4 Fasteners detached 0.009 A11 Cover plate detached 0.008 A5 Clutch jamming 0.007 A12 belt breakage 0.005 A6 The tension wheel is stuck. 0.031 A13 Cracked support 0.005 A7 Tensioner wheel shift 0.023

[0120] The fault codes for the compressor system's basic events are A1-A13, the fault codes for the third-level intermediate nodes are B1-B7, and so on. The fault codes for other second-level intermediate nodes in the air conditioning compressor system are C1-C4, and the fault code for the first-level intermediate node compressor system is D1. Faults at each level of the compressor system's sub-nodes are shown in Table 2, and the fault tree model diagram is shown below. Figure 3 .

[0121] Table 2: Summary Table of Faults at Various Sub-nodes of the Compressor System

[0122]

[0123] Step 4: Based on Step 2 and Step 3, calculate the posterior probability of each level of child nodes;

[0124] Specifically, based on steps 2 and 3, the posterior probabilities of each level of child nodes are calculated. In Bayesian networks, conditional probabilities are used to express the logical relationships between nodes; the logic gates in the fault tree model correspond to the conditional probability tables in the Bayesian network.

[0125] Taking an air conditioning compressor system as an example, its Bayesian network has leaf nodes A1-A13, and its parent nodes B1-B7 are third-level child nodes. Among these, the third-level child node compressor bearing B1 connects to two root node bearings that are worn. and bearing breakage The probability of a failure in the corresponding third-level child node B1 can be expressed as:

[0126]

[0127] leaf nodes and The posterior probabilities of occurrence are expressed as follows:

[0128]

[0129] Similarly, the third-level child nodes... As a higher-level second-level child node The bottom event, if the second-level child node Connects m third-level child nodes ,by Represents a node The probability of a failure can be obtained from the above steps. In the event that occurs, the j-th child node The posterior probability of occurrence, where m child nodes and nodes These correspond to the third-level child nodes and second-level child nodes in step 2, respectively.

[0130] Specifically, the five third-level child nodes B1-B5 are designated as higher-level second-level child nodes. The bottom event, with Represents a node The probability of a failure can be obtained from the above steps. In the event of an event, the posterior probability of each of the five child nodes occurring.

[0131] By analogy, the posterior probabilities of all leaf nodes, third-level child nodes, second-level child nodes, and first-level child nodes of the compressor system can be obtained in sequence.

[0132] The calculation results are shown in Table 3 below:

[0133] Table 3: Posterior probability table of all child nodes of the compressor system

[0134]

[0135] Step 5: Utilize a database of component performance parameters to establish a regression model among the relevant component performance parameters;

[0136] Specifically, a regression model is established among the performance parameters of components using a database of relevant performance parameters. The physicochemical parameters of component materials are intrinsic properties of the materials, and some parameters are strongly correlated. Moreover, the measurement of some parameters is relatively simple and convenient. Therefore, this correlation can be used to calculate the key performance parameters of components under different operating conditions through regression analysis, and to evaluate the real-time failure rate of components.

[0137] During the maintenance and repair of existing market products, a performance parameter database is established using the test data of component performance parameters. Correlated parameter pairs are analyzed, and support vector machine regression analysis is used to establish regression models between these parameters. , where variables For easy measurement of component material performance parameters, These are the key performance parameters of the component materials.

[0138] Taking compressor systems as an example, the performance-related parameters of commonly used materials for components are shown in Table 4 below:

[0139] Table 4: Summary Table of Material-Related Parameters for Compressor Components

[0140] Material Relevant parameters Metals (steel, copper, aluminum) Hardness - tensile strength, toughness - shear modulus, temperature - electrical conductivity Plastics, rubber Shear modulus-toughness oilseeds Viscosity - specific heat capacity, density - thermal conductivity

[0141] The regression model is calculated based on the correlation between the performance parameters of component materials, using easily measurable measured data and relevant regression models. The regression model takes easily measurable measured parameter data as input and outputs key performance parameters of components that are difficult to measure or whose measurement involves damage. Model types include, but are not limited to, support vector machine regression models and multinomial regression models.

[0142] Step 6: Measure the real-time data of the component performance parameters, and analyze the real-time key performance parameters of the component by combining the regression model between relevant parameters;

[0143] Specifically, real-time data of component performance parameters is measured, and regression models between relevant parameters are used to obtain real-time data of key component performance parameters. For key performance parameters of components that are difficult to measure, real-time monitoring of easily measurable component material parameters is performed, and regression models of relevant parameters are used to quickly calculate and obtain real-time data of key component performance parameters.

[0144] Taking the compressor bearing in a compressor system as an example, its tensile strength and shear modulus are respectively and The historical failure rates for wear and fracture are denoted as follows: and In actual working conditions, the measured hardness and toughness were as follows: and Using regression models The predicted real-time tensile strength and shear modulus are respectively and ,Right now:

[0145]

[0146] Step 7: Establish a real-time failure rate assessment model for various component faults and analyze the real-time functional status of the components;

[0147] Specifically, by combining the calibrated values ​​of multiple performance parameters of the components at the time of manufacture, a real-time failure rate assessment model for various faults of the components is established to analyze the real-time functional status of the components.

[0148] The prior probability of a component is a statistical result of historical failure data, covering various operating conditions throughout the entire life cycle of all products, and represents the expected failure rate. However, for individual components of a product, factors such as operating environment and performance parameter degradation, coupled with the random nature of degradation, make historical failure rates insufficient to describe the real-time failure rate of a single product's component at any given time. Therefore, it is necessary to utilize real-time data of the component's key performance parameters, combined with its factory calibration values, to establish a real-time failure rate assessment model and calculate the real-time failure rate of various component failures.

[0149] If the factory-calibrated value of a certain key performance parameter of a component and the value measured under actual operating conditions are respectively... and Market statistics show a failure rate of The real-time failure rate The evaluation model is as follows:

[0150]

[0151] Taking the aforementioned compressor bearing as an example, the real-time failure rate of bearing wear and breakage. and , can be represented as:

[0152]

[0153] The mentioned real-time performance status assessment model for components uses the factory-calibrated values ​​and real-time predicted values ​​of key performance parameters of components as input variables, including but not limited to exponential models, logarithmic models, linear models, etc., and outputs the real-time failure rate of various faults of components.

[0154] Step 8: Combining Step 3 and Step 7, calculate the real-time failure rate of each level of sub-nodes; the calculated real-time failure rates of various component failures are then used to obtain the results by combining the Bayesian network model of the air conditioning system, which is used to describe the current failure rate of each level of sub-nodes in the air conditioning system.

[0155] Specifically, in step 7, the real-time failure rate of various component faults is the real-time failure rate of the leaf nodes in the Bayesian network. Combining this with step 4, the real-time failure rate of the component is determined using the conditional probability of the Bayesian network, and the real-time posterior probability of the component is calculated.

[0156] Real-time failure rate of component with one type of failure mode Based on the logical relationships between the nodes of the Bayesian network, the real-time failure rate of the components is determined. .by Indicates the real-time failure rate of components , These represent the real-time failure rates for each of the l failure modes. ,but:

[0157]

[0158] in This indicates the component ( ) when the k-th failure mode occurs in real time. The probability of a failure occurring in real time.

[0159] By combining the various faults of the component and its real-time failure rate, and using the posterior probability calculation formula in step 4, the real-time posterior probabilities of various fault modes of the component can be calculated. For example, among l fault modes, the real-time posterior probability of the k-th fault mode. The calculation formula is:

[0160]

[0161] Right now Indicates components ( Given a real-time failure, the probability of the k-th failure mode occurring in real time among l failure modes is given by: , , The meaning is the same as above, so I will not repeat it.

[0162] Similarly, if a certain subsystem is composed of The real-time failure rate is Based on the component composition, the real-time failure rate of the subsystem can be determined according to the logical relationships between the nodes of the Bayesian network. ,as well as The real-time posterior probability of each component can be obtained. By analogy, the real-time failure rate and real-time posterior probability of each level of sub-node can be obtained, which can be applied to the system's fault diagnosis and location.

[0163] Taking compressor systems as an example, bearing wear and breakage failures and The real-time failure rate is and Then the bearing Real-time failure rate for:

[0164]

[0165] Then the faulty node and The real-time posterior probability formulas for bearing wear and breakage are as follows:

[0166]

[0167] Similarly, compressor failure C1 consists of failures in the five components B1-B5 listed in Table 2, with real-time failure rates of respectively. Then, the real-time failure rate of the compressor can be determined according to the above Bayesian network conditional probability formula. And the real-time posterior probabilities of the five components.

[0168] By analogy, the real-time failure rate and real-time posterior probability of each sub-node of the compressor system can be obtained.

[0169] Step 9: Compare the posterior probabilities of each level of child nodes to diagnose the fault status of each level of child nodes;

[0170] Specifically, step 8 obtains the real-time posterior probability of failure for each level of child node. By comparing real-time posterior probabilities at each level And statistical data posterior probability Calculate the degree of failure rate anomaly based on the failure rate anomaly thresholds for each level of node. To determine the node where the fault occurred. This includes the fault rate anomaly threshold. The threshold for critical components is determined based on their importance and historical experience. For example, the threshold for critical components is relatively low, while the threshold for redundant components can be relatively high. The principle for checking each level of nodes step by step is "higher-level first, then lower-level; high-level faults first, then low-level faults".

[0171] Based on the obtained real-time posterior probability of failure at each level of child nodes. By comparing real-time posterior probabilities at each level Posterior probabilities of child nodes at each level in a Bayesian network The relative value is used to calculate the degree of abnormality in the failure rate, δ.

[0172]

[0173] Based on the degree of failure rate anomaly δ and the failure rate anomaly threshold at each level of node To determine the node where the fault occurred, if the abnormality δ of the fault rate at each level of the node is greater than the corresponding abnormality threshold for the fault rate. If so, then this node is determined to be a faulty node. This represents the abnormal threshold of the failure rate for the j-th type of fault in the i-th level child node.

[0174] Taking a compressor system as an example, when the system malfunctions... When an incident occurs, begin troubleshooting with the next highest-level event, starting with the compressor component that has the highest failure rate. Troubleshooting begins. This is achieved by comparing the real-time post-hoc failure rates of the compressor components. and statistical posterior failure rate The relative difference is used to determine whether a node is a faulty node, and the abnormal threshold of its relative difference is used. The criteria for determining a faulty node are based on experience:

[0175]

[0176] If the compressor assembly is a non-faulty node, that is Real-time posterior failure rate of nodes Compared to statistical posterior failure rate It did not exceed the threshold Then, following the principle of "high-level faults first, low-level faults later," the low-level fault nodes of the same level are checked sequentially, such as tensioner wheel faults. The second highest failure rate, with real-time posterior probability and anomaly threshold being respectively... and Further investigation was conducted into the faulty points of the tensioner wheel;

[0177] If the compressor assembly is the faulty node, that is Real-time posterior failure rate of nodes Compared to statistical posterior failure rate Exceeding the threshold Then, following the principle of "higher-level first, then lower-level," the three levels of sub-nodes are checked sequentially, such as bearing failure. As a third-level child node, the real-time posterior probability and the anomaly threshold are respectively and .

[0178] The real-time failure rates of each level of sub-nodes are obtained as prior probabilities and are calculated layer by layer using a Bayesian algorithm. The statistical posterior probabilities of each level of sub-nodes in the system are obtained from the statistical results of historical failure data and Bayesian network calculation and analysis. The abnormal threshold of the posterior probability is the upper limit of the relative difference between the real-time posterior probability and the statistical posterior probability when there are no faults in each level of sub-nodes in the system. It is used to determine whether the current sub-node is a faulty node. The abnormal threshold includes, but is not limited to, relative values, absolute differences, dynamic differences depending on the situation, etc.

[0179] Step 10: Output the fault diagnosis results.

[0180] Step 9 yields the fault determination and analysis results for each level of sub-nodes in the system. The output results are presented in the form of statistical tables or logical structure diagrams, including information such as the historical fault rate, real-time fault rate, fault rate anomaly threshold reference value, and fault determination results for each level of node in the system. An output report is generated and handed over to the technical operation and maintenance personnel for processing.

[0181] The method in this embodiment is convenient and highly operable in real-time data measurement and processing. By using Bayesian networks and posterior probability to determine faults, the diagnostic process becomes more flexible and fault location becomes more accurate. With slight adjustments, it can be applied to fault diagnosis of various types of air conditioning systems, demonstrating strong practical generalization.

[0182] Additionally, it should be noted that:

[0183] In step 1, the air conditioning system uses a four-level fault tree model to construct fault modes. It can be replaced by a three-level, five-level, or even other-level fault tree model to achieve the same purpose.

[0184] In step 5, the regression analysis model between the performance-related parameters of the components can be established using methods such as logistic regression, linear regression, and nonlinear multinomial regression. The appropriate method can be selected based on the actual calculation accuracy requirements and the amount of data, and the same purpose can be achieved.

[0185] In step 7, the real-time fault assessment model established based on key performance parameters is not limited to the exponential model selected in this invention. Other models, such as linear, exponential, logarithmic, power functions, and their variants, used to describe the relationship between statistical fault rates and real-time fault rates can achieve the same purpose.

[0186] In step 9, the abnormal threshold of the posterior probability of each child node is determined by the statistical posterior probability and the real-time posterior probability. It is not limited to the relative value used in this invention, but can be replaced by the absolute difference or the dynamic difference depending on the situation, without affecting the purpose of this invention.

[0187] Example 2:

[0188] This embodiment provides a fault diagnosis device based on parameter correlation Bayesian networks, including:

[0189] Memory, used to store computer programs / instructions;

[0190] A processor for executing the computer program / instructions to implement the steps of the method described in Embodiment 1.

[0191] Example 3:

[0192] This embodiment provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0193] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0194] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0195] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0196] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0197] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fault diagnosis method based on parameter correlation Bayesian networks, characterized in that, Includes the following steps: Step 1: Determine the fault type of the air conditioning system of the construction machinery. The same fault type corresponds to the same fault event in the fault tree. Use the fault event as the base event of the fault tree model to build the fault tree model of the air conditioning system. Step 2: Map the fault tree model to a Bayesian network; Step 3: Combine historical fault data to calculate the probability of occurrence of the bottom event in the fault tree model, and use it as the prior probability of the leaf node of the Bayesian network corresponding to the bottom event of the fault tree model, so as to obtain the prior probability of each leaf node of the Bayesian network. Step 4: Calculate the posterior probability of each level of child nodes in the Bayesian network based on the Bayesian network mapped by the fault tree model and the prior probabilities of each leaf node in the Bayesian network. Step 5: Obtain the regression model between the performance-related parameters of the components; the input of the regression model between the performance-related parameters of the components are easily measurable performance-related parameters of the components, and the output is the real-time key performance parameters of the components; Step 6: Measure the real-time data of the performance-related parameters of the components, and combine the regression model between the performance-related parameters of the components to obtain the real-time data of the key performance parameters of the components. Step 7: Obtain a real-time failure rate assessment model for multiple faults of components. The input variables of the real-time failure rate assessment model are the factory calibration values ​​of key performance parameters of components, the real-time predicted values ​​of key performance parameters of components output by the relevant parameter regression model, and the market statistical failure rate of components. The output variable is the real-time failure rate of components under the corresponding failure mode. Step 8: Obtain the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components. Combine the real-time data of key performance parameters of components output by the regression model to calculate the real-time failure rate of each level of sub-nodes in the Bayesian network. Use the real-time failure rate of each level of sub-nodes as the real-time prior probability to calculate the real-time posterior probability of each level of sub-nodes in the Bayesian network. Step 9: Compare the real-time posterior probability and posterior probability of each level of child nodes in the Bayesian network to diagnose the fault status of each level of child nodes and obtain the fault diagnosis results. Step 10: Output the fault diagnosis results; Step 1: Determine the fault type of the air conditioning system of the construction machinery. The same fault type corresponds to the same fault event in the fault tree. Use the fault event as the base event in the fault tree model to build the fault tree model of the air conditioning system, including: In the fault tree model, the fault of the air conditioning system is regarded as the top event, the faults of each subsystem are regarded as first-level intermediate events, the faults of the components are regarded as second-level intermediate events, the faults of the parts are regarded as third-level intermediate events, and all fault events constitute the bottom event of the fault tree model.

2. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 2: Map the fault tree model to a Bayesian network, including: When a fault tree model is converted into a Bayesian network, the structure of the fault tree model corresponds to that of the Bayesian network. Each level of fault event in the fault tree model corresponds to a node in the Bayesian network: the bottom events of the fault tree model correspond to the leaf nodes of the Bayesian network, and the first-level intermediate events, second-level intermediate events, and third-level intermediate events in the fault tree model correspond to the first-level child nodes, second-level child nodes, and third-level child nodes of the Bayesian network, respectively. The connections between fault events in the fault tree model correspond to the edges in the Bayesian network. The logic gates that connect events in the fault tree model are replaced with the conditional probability tables of the Bayesian network. The prior probabilities of the leaf nodes in the Bayesian network correspond to the occurrence probabilities of the bottom events in the fault tree model.

3. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 2, characterized in that, Step 3: Combining historical fault data, calculate the occurrence probability of the bottom event in the fault tree model, and use it as the prior probability of the corresponding leaf node in the Bayesian network for the bottom event in the fault tree model. This yields the prior probabilities of each leaf node in the Bayesian network, including: By combining historical fault data, the probability of occurrence of the bottom event in the fault tree model is calculated, and this probability is used as the prior probability of the corresponding leaf node in the Bayesian network. The historical fault data includes all fault types in the air conditioning system and the frequency of each fault type. The probability of occurrence of the underlying events in the fault tree model is the probability of occurrence of various fault types obtained after statistical analysis of the fault data, and it is used as the prior probability of the corresponding leaf node of the Bayesian network.

4. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 4: Based on the Bayesian network mapped by the fault tree model and the prior probabilities of each leaf node in the Bayesian network, calculate the posterior probabilities of each level of child nodes in the Bayesian network, including: In Bayesian networks, conditional probabilities are used to express the logical relationships between nodes. The logic gates in the fault tree model correspond to the conditional probability tables in the Bayesian network. If node Connect n child nodes , by express The probability of a node failing is expressed as: （1） Then node In the event that occurs, the i-th child node The formula for calculating the posterior probability of occurrence is as follows: （2） in, express The i-th child node connected to the node The prior probability, , where n is the total number of child nodes of node B; Represents child nodes In the event of a node The conditional probability of occurrence Indicates the condition that node B occurs. The conditional probability of occurrence, i.e., the i-th child node The posterior probability of occurrence; n child nodes and The nodes correspond to leaf nodes and third-level child nodes, respectively; , where n is the total number of child nodes of node B; calculate the posterior probability of all leaf nodes and the probability of a third-level child node occurring; third-level child nodes As a higher-level second-level child node The bottom event, if the second-level child node Connects m third-level child nodes ,by Represents a node The probability of a failure can be obtained from the above steps. In the event that occurs, the j-th child node The posterior probability of occurrence, where m child nodes and nodes These correspond to third-level child nodes and second-level child nodes, respectively; the posterior probability of all third-level child nodes and the probability of second-level child nodes occurring are calculated. By treating the second-level child node C as the base event of the higher-level child node, the posterior probabilities of all leaf nodes, third-level child nodes, second-level child nodes, and first-level child nodes of the compressor system are obtained sequentially through the above steps.

5. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 5: Obtain the regression model between the performance-related parameters of the components, including: Obtain test data on the performance parameters of components and establish a performance parameter database; Based on the performance parameter database, the parameter pairs that have a correlation are analyzed to obtain the correspondence between easily measurable component performance-related parameters and key performance parameters of component materials, thus obtaining the correlation parameter pairs consisting of easily measurable component performance-related parameters and key performance parameters of component materials. Based on the detection data of parameter pairs consisting of easily measurable component performance parameters and key component material performance parameters, a regression model between the relevant parameters is established using support vector machine regression analysis: ； Among them, variables Easily measurable component performance-related parameters are used as model input variables. Key performance parameters of component materials, which correspond to easily measurable performance parameters of the components, are used as output variables of the model.

6. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 6: Measure real-time data of component performance-related parameters, and combine this data with the regression model among these parameters to obtain real-time data of the component's key performance parameters, including: For key performance parameters of components that are difficult to measure, regression models between the corresponding parameters can be obtained by using the correspondence between easily measurable performance parameters of components and key performance parameters of component materials. By monitoring the easily measurable material parameters of the components in real time, real-time data on the performance-related parameters of the components can be obtained. Based on real-time data of component performance-related parameters, a regression model of relevant parameters is used to calculate and obtain real-time data of key component performance parameters.

7. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 7: Obtain a real-time failure rate assessment model for various component faults, including: The prior probabilities of various failure modes of components are determined by the statistical results of historical failure data, covering various operating conditions of all products throughout their entire life cycle, and are the expected values ​​of failure rates. By utilizing real-time data of key performance parameters of components and combining them with their factory calibration values, a real-time failure rate assessment model is established to calculate the real-time failure rate of various faults in components. If the factory-calibrated value and the real-time predicted value under actual operating conditions of the key performance parameter corresponding to a certain failure mode of a component are respectively... and The market statistics for this failure mode show a failure rate of [missing information]. The real-time failure rate The evaluation model is represented as follows: （3） The input variables of the real-time failure rate assessment model are the factory calibration values ​​of key performance parameters of components, the real-time predicted values ​​of key performance parameters of components output by the relevant parameter regression model, and the market statistical failure rate of components. The output variable is the real-time failure rate of components under the corresponding failure mode.

8. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 8: Obtain the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components. Combine this with the real-time data of key performance parameters of components output by the regression model to calculate the real-time failure rate of each level of the Bayesian network. Use the real-time failure rate of each level of the sub-nodes as the real-time prior probability to calculate the real-time posterior probability of each level of the sub-nodes in the Bayesian network, including: Obtain the factory calibration values ​​of key performance parameters of components and the market statistics failure rate of components; Based on the factory calibration values ​​of key performance parameters of components and the market statistical failure rate of components, combined with the real-time data of key performance parameters of components output by the regression model, the real-time failure rate of various failures of components is calculated based on the real-time failure rate evaluation model, that is, the real-time failure rate of leaf nodes in Bayesian network. Real-time failure rate of a component using one type of failure mode Based on the logical relationships between the nodes of the Bayesian network, the real-time failure rate of the components is determined. : by Indicates the real-time failure rate of components , These represent the real-time failure rates of the k-th failure mode, respectively. ,but: （4） in This indicates the component ( ) when the k-th failure mode occurs in real time. The probability of a failure occurring in real time; By combining the various faults of components and their real-time failure rates, and using the posterior probability calculation formula, the real-time posterior probabilities of various fault modes of components are calculated, i.e., the real-time posterior probabilities of leaf nodes are obtained. Among l fault modes, the real-time posterior probability of the k-th fault mode is... The calculation formula is: （5） in, Indicates components ( Given that a fault occurs in real time, the probability of the kth fault mode occurring in real time among l fault modes; Based on the real-time failure rate of components By understanding the logical relationships between the nodes of the Bayesian network, the real-time failure rate of the component corresponding to that part can be determined. And the real-time posterior probability of the component, the real-time posterior probability of the component corresponds to the real-time posterior probability of the third-level child node; By analogy, the real-time failure rate and real-time posterior probability of the second-level sub-nodes and the real-time failure rate and real-time posterior probability of the first-level sub-nodes are calculated to obtain the real-time failure rate and real-time posterior probability of each level of sub-nodes, which can be applied to the system's fault diagnosis and location.

9. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 9: Compare the real-time posterior probability and posterior probability of each child node in the Bayesian network to diagnose the fault state of each child node and obtain the fault diagnosis results, including: Based on the obtained real-time posterior probability of failure at each level of child nodes. By comparing real-time posterior probabilities at each level Posterior probabilities of child nodes at each level in a Bayesian network The relative value is used to calculate the degree of abnormality in the failure rate, δ. （6） Based on the degree of failure rate anomaly δ and the failure rate anomaly threshold at each level of node To determine the node where the fault occurred, This represents the abnormal threshold of the failure rate for the j-th type of fault in the i-th level child node.

10. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 9, characterized in that, Based on the degree of failure rate anomaly δ and the failure rate anomaly threshold at each level of node To determine the node where the fault occurred, including: When the system fails When an incident occurs, begin troubleshooting with the next highest-level event, starting with the system components that have the highest failure rate. The investigation begins by comparing the real-time posterior failure rate of system components with the relative difference between the statistical posterior failure rate and any abnormalities. If the abnormality of the failure rate of a sub-node exceeds the abnormal failure rate threshold, then the sub-node is determined to be a faulty node. If this system component If a node is not faulty, then low-fault nodes of the same level are checked in descending order of real-time failure rate. If this system component is a faulty node, then its child nodes and the next level child nodes of its child nodes should be checked sequentially according to the node level of the Bayesian network.

11. The fault diagnosis method based on parameter correlation Bayesian networks according to claim 1, characterized in that, Step 10: Output the fault diagnosis results, including: Obtain the fault determination and analysis results of each level of sub-nodes in the system; the output results are presented in the form of statistical tables or logical structure diagrams, including the historical fault rate, real-time fault rate, fault rate abnormal threshold reference value, and fault determination result information of each level of node in the system, and generate an output report and output it.

12. A fault diagnosis device based on parameter correlation Bayesian network, comprising: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the method according to any one of claims 1-11.

Images