Criticality analysis method and system for failure modes of numerical control equipment

Abstract

The present invention relates to the technical field of numerical control equipment fault analysis, and discloses a method and system for analyzing the criticality of a numerical control equipment fault mode, wherein the method comprises: obtaining operation data of the numerical control equipment, analyzing the operation data, and obtaining a set of numerical control equipment fault-related subsystems; constructing a strong link subsystem based on the set of numerical control equipment fault-related subsystems; analyzing the strong link subsystem to determine a fault mode level-related model of the strong link subsystem; determining the fault mode influence of the strong link subsystem based on the fault mode level-related model of the strong link subsystem; determining the fault probability by using a Bayesian network based on the fault mode level-related model of the strong link subsystem; and calculating the criticality of the fault mode of the numerical control equipment based on the fault mode influence and the fault probability. The present invention can obtain an accurate fault mode criticality.

Description

Technical Field

[0001] The present invention relates to the technical field of numerical control equipment failure analysis, and in particular to a method and system for analyzing the criticality of a numerical control equipment failure mode. Background Art

[0002] In the process of ranking failure modes using traditional FMECA (Failure Mode, Effects and Criticality Analysis), the following problems are inevitable: (1) When ranking the criticality, only the severity of the failure mode itself is considered, and the relationship between the failure modes is not considered, making the obtained failure mode impact not accurate enough. (2) In traditional FMECA, the criticality of the failure mode of CNC equipment is determined based on three factors: the failure mode frequency ratio, the failure mode impact, and the failure rate of the component. The failure mode frequency ratio is used as a priori probability, and it is not accurate enough to substitute it into the calculation of criticality. Summary of the invention

[0003] In order to solve the shortcomings of the prior art, the present invention provides a criticality analysis method and system for numerical control equipment failure modes; a DEMATEL-ISM (Decision-Making and Trial Evaluation Laboratory; Interpretative Structural Modelling) method is used to obtain a failure mode-level fault correlation model, which is converted into an ANP (Analytic Network Process) network model, and the ANP method is used to obtain the failure mode influence; the DEMATEL-ISM network model is converted into a Bayesian model, and the Bayesian method is used to obtain the failure mode frequency ratio (failure probability).

[0004] On the one hand, a method for analyzing the criticality of a failure mode of numerical control equipment is provided, comprising: obtaining operation data of the numerical control equipment, analyzing the operation data, and obtaining a set of numerical control equipment failure-related subsystems; constructing a strongly linked subsystem based on the set of numerical control equipment failure-related subsystems; wherein, constructing a strongly linked subsystem based on the set of numerical control equipment failure-related subsystems comprises: constructing a direct impact matrix between the failure subsystems ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , and obtain the hierarchical structure of the fault subsystem; based on the influence causal graph between each fault node, a strongly linked subsystem is obtained; the strong link means that the centrality between the fault nodes is greater than the set threshold; the strongly linked subsystem is analyzed to determine the fault mode level correlation model of the strongly linked subsystem; based on the fault mode level correlation model of the strongly linked subsystem, the fault mode influence of the strongly linked subsystem is determined; based on the fault mode level correlation model of the strongly linked subsystem, the Bayesian network is used to determine the fault probability; based on the fault mode influence and the fault probability, the criticality of the fault mode of the CNC equipment is calculated.

[0005] On the other hand, a criticality analysis system for numerical control equipment failure modes is provided, comprising: an analysis module, which is configured to: obtain operation data of numerical control equipment, analyze the operation data, and obtain a set of numerical control equipment failure-related subsystems; a construction module, which is configured to: construct a strongly linked subsystem based on the set of numerical control equipment failure-related subsystems; wherein, the construction of the strongly linked subsystem based on the set of numerical control equipment failure-related subsystems includes: constructing a direct impact matrix between the failure subsystems ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , obtain the hierarchical structure of the fault subsystem; based on the influence causal diagram between each fault node, obtain the strongly linked subsystem; the strong link means that the centrality between the fault nodes is greater than the set threshold; the first determination module is configured to: analyze the strongly linked subsystem and determine the fault mode level correlation model of the strongly linked subsystem; the second determination module is configured to: determine the fault mode influence of the strongly linked subsystem based on the fault mode level correlation model of the strongly linked subsystem; the third determination module is configured to: based on the fault mode level correlation model of the strongly linked subsystem, use the Bayesian network to determine the fault probability; the calculation module is configured to: calculate the criticality of the fault mode of the CNC equipment based on the fault mode influence and the fault probability.

[0006] The above technical solution has the following advantages or beneficial effects: when ranking the criticality, not only the severity of the failure mode itself is considered, but also the relationship between the failure modes is considered, so that the obtained failure mode influence is sufficiently accurate. In traditional FMECA, the criticality of the failure mode of CNC equipment is determined based on three factors: the failure mode frequency ratio, the failure mode influence, and the failure rate of the component. The failure mode frequency ratio is calculated as the posterior probability by the Bayesian method, and it is more accurate to substitute it into the calculation of the criticality. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0008] Figure 1 This is a flow chart of the method of embodiment 1. DETAILED DESCRIPTION

[0009] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0010] Failure mode criticality analysis is an important part of quantifying the consequences of failure modes based on FMEA (Failure Mode and Effect Analysis). Failure mode criticality analysis is conducted based on the probability of occurrence of the failure mode and the comprehensive impact of the failure mode on the degree of fatal damage to the system. The criticality of the failure mode reflects the degree of harm to the functionality, effectiveness and operator safety of the entire machine when the fault subsystem fails. By conducting a failure mode criticality analysis on CNC equipment, we can obtain the key components that affect its reliability and find out the weak links that affect the reliability of CNC equipment, so as to improve reliability in a targeted manner.

[0011] In recent years, with the rapid development of FMECA (Failure Mode, Effects and Criticality Analysis) technology, FMECA has been widely used in the reliability research of CNC equipment. However, CNC equipment is a complex mechatronics product. There is coupling between the fault nodes of CNC equipment, which makes the system failure modes not isolated, but mutually influential. One failure mode may be the cause of the next failure mode. Simply analyzing and evaluating the failure mode will often make the results deviate a lot from the actual situation. These complex relationships have a great impact on CNC equipment, so they should be fully considered.

[0012] As systematic methods for analyzing complex problems, the Decision Experiment and Evaluation Laboratory (DEMATEL) and the Interpretive Structural Model (ISM) have a considerable degree of complementarity: the non-zero elements in the overall influence matrix from the DEMATEL method and the non-zero elements in the reachability matrix from the ISM method both indicate that there is a relationship between the system factors, and conversely, zero elements indicate that there is no relationship between the system factors. In terms of the amount of information contained, the overall influence matrix of the DEMATEL method is more than the reachability matrix of the ISM method. Therefore, the reachability matrix can be calculated and obtained using the DEMATEL method.

[0013] However, the integrated DEMATEL / ISM (Decision-Making and Trial Evaluation Laboratory); ISM (Interpretative Structural Modelling) method can only determine the causal relationship order, direction and hierarchy of system factors, and cannot obtain priority weights. The importance and influence of factors must be taken into account in a complete causal relationship. Therefore, in addition to integrating DEMATEL and ISM methods, it is also necessary to combine the network analytic hierarchy process (ANP) to obtain the importance of factors. ANP is an extended method of the analytic hierarchy process (AHP), which is quite applicable to complex system decisions with dependency and feedback relationships between system factors. This method is not complete in dealing with the relationship between internal factors, and the impact can only be evaluated after the network structure relationship of the factors is clarified. However, ANP is lacking in the method and process of establishing the network structure relationship, which can be compensated by DEMATEL and ISM methods. Therefore, the DEMATEL / ISM / ANP method can be used to obtain the influence degree, and the visualization model obtained by DEMATEL / ISM can be converted into a Bayesian network model. The Bayesian method is used to obtain the posterior fault probability and perform fault hazard analysis on CNC equipment.

[0014] Embodiment 1: Figure 1 As shown, this embodiment provides a method for analyzing the criticality of a failure mode of numerical control equipment, including: S101: acquiring operation data of numerical control equipment, analyzing the operation data, and obtaining a set of fault-related subsystems of the numerical control equipment; S102: constructing a strong link subsystem based on the set of fault-related subsystems of the numerical control equipment; S103: analyzing the strong link subsystem, and determining a fault mode level-related model of the strong link subsystem; S104: determining the fault mode influence of the strong link subsystem based on the fault mode level-related model of the strong link subsystem; S105: determining the fault probability by using a Bayesian network based on the fault mode level-related model of the strong link subsystem; S106: calculating the criticality of the failure mode of the numerical control equipment based on the fault mode influence and the fault probability.

[0015] Furthermore, the S101: obtaining operation data of numerical control equipment, wherein the numerical control equipment refers to equipment that uses numerical control technology to control the working process of a working machine, such as a machining center. The operation data refers to fault data that occurs during the operation of the numerical control equipment.

[0016] Further, the S101: analyzing the operation data to obtain a set of fault-related subsystems of the numerical control equipment, including: when there is a related fault between one subsystem and another subsystem, the two subsystems are considered to be related subsystems; when there is no related fault between one subsystem and another subsystem, the two subsystems are considered not to be related subsystems; determining the set of fault-related subsystems constituting the numerical control equipment , Representative A subsystem that has related faults with other subsystems. Indicates the number of faulty subsystems where related faults exist.

[0017] For example, a subsystem is a part of a system, which also includes two or more combined devices, but it cannot independently complete the specific functions of the system. For example, the representative product of CNC equipment, the machining center, contains many subsystems, such as pneumatic system, electrical system, CNC system, etc.

[0018] Example of related subsystems: For example, if a fault in the electrical system causes a fault in the pneumatic system, then the two are related subsystems. If the faults in the two systems do not have an impact on the affected system, then they are unrelated subsystems.

[0019] It should be understood that by analyzing and sorting the collected fault data of a series of CNC equipment, the main fault subsystems can be determined. The fault subsystem factors have a complex mutual influence relationship and can form a multi-level complex network structure, so they can be studied based on a systematic method.

[0020] Furthermore, the S102: constructing a strong link subsystem based on a set of CNC equipment fault-related subsystems includes: S102-1: constructing a direct impact matrix between fault subsystems ; S102-2: Direct impact matrix Perform normalization to obtain the normalized matrix ; S102-3: Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; S102-4: based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; S102-5: Based on the overall impact matrix , determine the reachability matrix ; S102-6: For the reachable matrix Perform regional decomposition, and classify each decomposed region to obtain a set of levels; S102-7: Based on the class division result and the reachability matrix , and obtain the hierarchical structure of the fault subsystem; S102-8: Based on the impact causal graph between each fault node, obtain the strongly linked subsystem.

[0021] Furthermore, the S102-1: constructing a direct impact matrix between fault subsystems ,include: ;in, For subsystem Impact Subsystem The number of impacts, hour, , Indicates the number of faulty subsystems with correlated faults.

[0022] Furthermore, the S102-2: Direct Impact Matrix Perform normalization to obtain the normalized matrix ,include: ;in, For the Subsystem Impact Subsystem The number of impacts, Indicates the number of faulty subsystems with correlated faults.

[0023] Furthermore, the S102-3: based on the normalized matrix , calculate the comprehensive impact relationship matrix , based on the comprehensive impact relationship matrix , determine the impact causal graph between each fault node, including: ;in, is the identity matrix, is the standardized matrix, Representation Subsystem Subsystem of Indirect impact of the stage, Indicates that there is a fault related to other subsystems. subsystems, Indicates that there is a fault related to other subsystems. subsystems, , , .

[0024] Assume comprehensive influence relationship matrix The influence of behavior and , comprehensive impact relationship matrix The sum of the columns is the degree of influence ; Using the centrality as the horizontal axis and the cause degree as the vertical axis, draw the causal effect diagram between each fault node of the CNC equipment; ; ;in, Indicates the degree of influence, Indicates the degree of influence.

[0025] According to the impact and the degree of influence , determine the centrality and causality: ; ;in, represents the centrality, Indicates the degree of cause.

[0026] Centrality indicates the degree of connection between elements in the system. The larger the centrality value, the closer the connection between elements.

[0027] The causal degree determines the position of each factor in the system. If the causal degree value is greater than 0, it means that the influence of this factor on other factors is greater than the influence of other factors on itself, and it is called a causal factor; if the causal degree is less than 0, it means that the influence of this factor on other factors is less than the influence of other factors on itself, and it is called a result factor.

[0028] The closer to the right side of the horizontal axis of the causal diagram, the higher the centrality, indicating that the fault node is in the center of the entire CNC equipment, affecting other fault nodes and being affected by other fault nodes; the farther away from the positive direction of the vertical axis of the causal diagram, the stronger the degree of influence of the fault node on other fault nodes and the weaker the degree of influence by other fault nodes; the farther away from the negative direction of the vertical axis, the weaker the degree of influence of the fault node on other fault nodes and the stronger the degree of influence by other fault nodes.

[0029] Furthermore, the S102-4: based on the comprehensive impact relationship matrix , calculate the overall impact matrix ,include: ;in, is the identity matrix, Indicates that after considering the impact of the faulty subsystem on itself, the subsystem Subsystem The extent of direct and indirect impacts Indicates the number of faulty subsystems with correlated faults.

[0030] It should be understood that the hierarchical structure diagram is the visualization result of the DEMATEL-ISM model. In order to correctly evaluate the relationship between the fault factors and obtain a reasonable structure diagram, it is necessary to set a threshold to simplify the influence relationship in the matrix T. After setting the threshold, only in the matrix Only factors with an impact greater than the threshold will be selected and displayed in the impact relationship diagram. If the threshold is too low, the relationship diagram will be too complex and unable to display the necessary information for decision making; if the threshold is too high, many factors will be presented as independent factors and unable to show their relationship with other factors. Therefore, in order to correctly evaluate the relationship between faults and obtain an ideal hierarchical structure diagram, it is necessary to set a reasonable threshold to simplify the comprehensive impact relationship matrix. Comprehensive impact matrix It can only clearly reflect the mutual influence relationship and degree between different fault subsystems, but does not consider the influence of the fault subsystem on itself. Therefore, it is necessary to calculate the overall influence relationship of the fault subsystem.

[0031] Furthermore, S102-5: Based on the overall impact matrix , determine the reachability matrix ,include: ;in, The value range of is 1~n; The value range of is 1~n; Value: .

[0032] Furthermore, the S102-6: for the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level set, including: define components The set of affected components is components The reachable set , affecting components The component collection of The predecessor set , by calculating the reachable set and antecedent set ,right Carry out regional decomposition and level division.

[0033] Regional decomposition: Defined as a common set, the common set The elements are defined as the starting set ,for In and ,like ,but and If they are not in the same area, For the same area, thus realizing the decomposition of the region.

[0034] The result of regional decomposition is denoted as ;in, For the A collection of components in relatively independent areas. . Level division: For the same area , and get satisfaction in turn Components, find out the set of levels, expressed as: ; ;…… ;in, It is the set of levels obtained by level division.

[0035] Furthermore, the step S102-7: according to the level division result and the reachable matrix , and obtain the hierarchical structure of the fault subsystem, including: according to the level division result and the reachability matrix , for the reachability matrix The skipped binary relations are removed, and the binary relations that reach themselves are removed to obtain the skeleton matrix, which is used as the hierarchical structure model of the fault subsystem.

[0036] It should be understood that a multi-level hierarchical structure model is obtained based on the above stratification results and reachable matrix. Considering that there are many failure modes of CNC equipment and many nodes involved, the subsequent process of establishing the ANP evaluation structure model of all nodes is complicated, and the nodes with unclear influence relationships have little influence on the importance of nodes with strong connections. By dividing the above causal graph and setting the coordinate axis threshold, subsystem nodes with strong connections are screened out to further determine the influence relationship.

[0037] Furthermore, the S102-8: based on the impact causal graph between each fault node, a strongly linked subsystem is obtained, including: dividing the causal graph by setting a coordinate axis threshold, deleting nodes with a centrality less than the set threshold, and using the deleted causal graph as a strongly linked subsystem.

[0038] For example, the influence causal diagram between the various fault nodes of the machining center is drawn by taking the centrality as the horizontal axis and the cause degree as the vertical axis. The closer to the right side of the horizontal axis, the higher the centrality, indicating that the fault node is in the center of the entire machining center, affecting other fault nodes and being affected by other fault nodes; the farther away from the positive direction of the vertical axis, the stronger the degree of influence of the fault node on other fault nodes and the weaker the degree of influence by other fault nodes; the farther away from the negative direction of the vertical axis, the weaker the degree of influence of the fault node on other fault nodes and the stronger the degree of influence by other fault nodes.

[0039] It should be understood that both DEMATEL and ISM are systematic methods that can be used to analyze complex problems. The non-zero elements in the overall impact matrix of the DEMATEL method and the reachable matrix of the ISM method both indicate that there is a mutual influence relationship between system factors. Conversely, the zero element indicates that there is no mutual influence relationship between system factors. It can be seen that the two methods are complementary to a certain extent. In terms of the amount of information contained, the overall impact matrix of the DEMATEL method is more than the reachable matrix of the ISM method.

[0040] Therefore, the DEMATEL method can be used to calculate and obtain the reachable matrix. However, although the integration of DEMATEL and ISM methods can obtain the order, direction and hierarchy of the factor influence relationship, it is impossible to obtain the factor priority weight and still cannot form a complete correlation influence relationship. Therefore, it is necessary to combine the ANP method to obtain the importance of factors in order to obtain a complete correlation relationship. ANP is an extended method of the Analytic Hierarchy Process (AHP), which is suitable for complex system decision-making with dependencies and feedback relationships between system factors.

[0041] Before applying this method, the network structure relationship of the factors must be determined before the weight evaluation can be performed. However, ANP is lacking in the method and process of establishing the network structure relationship, which can be compensated by DEMATEL and ISM methods. In this way, the fault correlation model and the influence of the fault mode can be obtained. Based on the above fault correlation model, the Bayesian method is used to obtain the accurate fault probability.

[0042] The purpose of dividing the fault impact levels of the fault nodes of CNC equipment is to determine the strength of the mutual influence between the fault nodes. The quantitative table of the degree of influence designed by this technology is obtained based on the collected historical fault data. According to the correlation between the node failures, the number of mutually affecting node failures is used as the judgment standard. The evaluation indicators are assigned values ​​based on the expert's judgment on the intensity of influence between each fault node. Since the expert evaluation method is highly subjective, the fault transmission mechanism between the fault nodes is considered on the basis of the expert scoring, and the frequency of fault occurrence of the fault node is used as an auxiliary factor of the impact degree value.

[0043] Further, S103: Analyze the strong link subsystem to determine the fault mode level correlation model of the strong link subsystem, including: S103-1: When there is a related fault between one fault mode and another fault mode, the two fault modes are considered to be related fault modes; S103-2: When there is no related fault mode between one fault mode and another fault mode, the two fault modes are considered not to be related fault modes; S103-3: The set of related fault modes determined in this way , Representative a failure mode that has related failures with other failure modes, Represents the number of failure modes with related failures. S103-4: Based on the related failure mode set, construct a failure mode hierarchical model; the failure mode hierarchical model is a failure mode level correlation model of strongly linked subsystems.

[0044] Further, the S103-4: constructing a fault mode hierarchical model based on a set of related fault modes includes: S103-41: constructing a direct impact matrix between fault modes ; S103-42: Direct impact matrix Perform normalization to obtain the normalized matrix ; S103-43: Based on normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix ; S103-44: Based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; S103-45: Based on the overall impact matrix , determine the reachability matrix ; S103-46: for reachable matrix Perform regional decomposition, and classify each decomposed region to obtain a set of levels; S103-47: Based on the class division result and the reachability matrix , and obtain the fault mode hierarchical structure.

[0045] Furthermore, the S103-41: constructing a direct impact matrix between failure modes ,include: ;in, Failure Mode Impact failure mode The number of impacts, hour, , Indicates the number of fault sub-patterns that have correlated faults.

[0046] Further, S103-42: Direct Impact Matrix Perform normalization to obtain the normalized matrix ,include: ;in, For the Failure Mode Impact Failure Mode The number of impacts, Indicates the number of failure modes with related failures. Further, S103-43: Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix ,include: ;in, is the identity matrix, is the standardized matrix, Indicates failure mode Failure Mode of Indirect impact of the stage, Indicates that there is a fault related to other failure modes. Failure modes, Indicates that there is a fault related to other failure modes. Failure modes, , , .

[0047] Further, the S103-44: based on the comprehensive impact relationship matrix , calculate the overall impact matrix ,include: ;in, is the identity matrix, Indicates that after considering the impact of the failure mode on itself, the failure mode Failure Mode The extent of direct and indirect impacts Indicates the number of failure modes that have dependent failures.

[0048] It should be understood that the hierarchical structure diagram is the visualization result of the DEMATEL-ISM model. In order to correctly evaluate the relationship between the fault factors and obtain a reasonable structure diagram, it is necessary to set a threshold to simplify the influence relationship in the matrix T. After setting the threshold, only in the matrix Only factors with an impact greater than the threshold will be selected and displayed in the impact relationship diagram. If the threshold is too low, the relationship diagram will be too complex and unable to display the necessary information for decision making; if the threshold is too high, many factors will be presented as independent factors and unable to show their relationship with other factors. Therefore, in order to correctly evaluate the relationship between faults and obtain an ideal hierarchical structure diagram, it is necessary to set a reasonable threshold to simplify the comprehensive impact relationship matrix. Comprehensive impact matrix It can only clearly reflect the mutual influence relationship and degree between different failure modes, but does not consider the impact of the failure mode on itself. Therefore, it is necessary to calculate the overall influence relationship of the failure mode.

[0049] Further, S103-45: Based on the overall impact matrix , determine the reachability matrix ,include: ;in, The value range of is 1~n; The value range of is 1~n; Value: .

[0050] Furthermore, the S103-46: for the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain each level set, including: define the failure mode The set of affected failure modes is failure mode The reachable set , affecting components The failure mode set is the failure mode The predecessor set , by calculating the reachable set and antecedent set ,right Carry out regional decomposition and level division.

[0051] Regional decomposition: Defined as a common set, the common set The elements are defined as the starting set ,for In and ,like ,but and If they are not in the same area, For the same area, thus realizing the decomposition of the region.

[0052] The result of regional decomposition is denoted as ;in, For the A collection of components in relatively independent areas. .

[0053] Level division: for the same area , and get satisfaction in turn Failure mode, find out the set of each level, expressed as: ; ;…… ;in, It is the set of levels obtained by level division.

[0054] Further, the S103-47: according to the level division result and the reachable matrix , and obtain the fault mode hierarchical structure, including: The skipped binary relations are removed, and the binary relations that reach themselves are removed to obtain the skeleton matrix, which is used as the fault mode recursive hierarchical structure.

[0055] Furthermore, the S104: based on the failure mode level correlation model of the strongly linked subsystem, determining the failure mode influence of the strongly linked subsystem, includes: through the Assess / Compare function of the Super Decision software, inputting the comparison data formed by the expert evaluation, and obtaining the pairwise comparison matrix between each failure mode; the same failure mode has different degrees of influence in different nodes. After the failure cause mechanism analysis is completed, the 1-9 scaling method is used to score the pairwise comparison matrix between each failure mode to quantify the judgment matrix.

[0056] On the basis of the strong link subsystem fault mode level correlation model, the fault impact of the CNC equipment fault mode is used as the target decision factor in the control layer, and the fault mode node is used as the element set in the element layer. An ANP network structure model between the fault modes of the strong link nodes is established.

[0057] In order to calculate the failure impact of the failure mode nodes of the strong link subsystem of CNC equipment, it is necessary to establish the impact matrix between each failure mode.

[0058] Through the method of expert evaluation, the mutual influence relationship between various failure modes is obtained.

[0059] Engineers with many years of work experience can be selected to score the impact of each node. After sorting out the scoring results, an impact matrix for pairwise comparison is formed.

[0060] Furthermore, the S105: based on the failure mode level correlation model of the strongly linked subsystem, using the Bayesian network to determine the failure probability includes: S105-1: constructing the Bayesian network; S105-2: determining the posterior probability of the Bayesian network.

[0061] Furthermore, the S105-1: constructing a Bayesian network includes: mapping the fault modes in the fault mode level correlation model of the strongly linked subsystems into Bayesian network nodes, mapping the causal relationship between the fault modes in the fault mode level correlation model of the strongly linked subsystems into directed edges in the Bayesian network, and mapping the hierarchical structure in the fault mode level correlation model of the strongly linked subsystems into a structure in the Bayesian network.

[0062] The fault mode hierarchical structure constructed by DEMATEL-ISM method and Bayesian network have great similarities in describing the structure of the system. According to the above conversion rules, the obtained fault mode factor hierarchical structure is converted into the initial Bayesian network.

[0063] The prior probability of the root node is determined by the occurrence degree of the improved FMECA failure mode to determine the conditional probability. The conditional probability table is used to describe the causal relationship between nodes in the Bayesian network. According to the direct influence relationship matrix constructed based on the previous failure mode, the causal relationship between factors is described. Therefore, based on the direct influence relationship matrix and the reliability database, the Bayesian network conditional probability table is obtained.

[0064] Further, the S105-2: determining the posterior probability of the Bayesian network includes: S105-21: inputting the prior probability and CPT (conditional probability table) of the root node into the Bayesian network calculation software GeNIe (state0 represents that the event does not occur, and state1 represents that the event occurs), constructing a complete BN model, obtaining causal reasoning results, and then obtaining the posterior failure probability of the fault mode; S105-22: analyzing the constructed BN model through the GeNIe software (i.e. setting the root node state to "state1=100%" to obtain the diagnostic reasoning result. Then set the intermediate node to the abnormal occurrence state, that is, its state=100%. Through Bayesian network simulation, the posterior probability of each evidence node can be inferred, that is, the risk cause chain, thereby diagnosing the failure path of the strong link subsystem of the CNC equipment.

[0065] It should be understood that Beyes basic theory: In the field of uncertain knowledge analysis and evaluation, Bayesian network is one of the most effective theoretical models. BN is a directed acyclic graph, the main components of which include directed edges connecting different nodes and nodes representing variables. Nodes can be described by random variables, the connection relationship between nodes is represented by directed edges between nodes, and the weight value of the relationship between nodes is represented by conditional probability. The root node is usually described based on the prior probability. Node variables can represent various abstract expressions such as expert experience. Under certain conditions, BN can infer and deduce various uncertain and fuzzy data and knowledge, and make decisions based on multiple prior conditions. It is widely used in describing and solving problems such as uncertainty and probability. Today, Bayesian networks have been widely used in the work of state assessment, fault diagnosis and reliability assessment and analysis of high-end equipment.

[0066] In a Bayesian network, the relationship between different variables can be described qualitatively through a directed acyclic graph, and the degree of association between nodes can be quantitatively described using conditional probability. The purpose of building a model is to perform scientific probability analysis and calculation on the weight values ​​of each node in the Bayesian network model.

[0067] set up represents a Bayesian network, then Representatives have A directed acyclic graph of variable nodes, nodes represents a random variable, When a node leads to a directed edge, the node is another node The parent node of is the node pointed to by is the child node, that is The directed edge between parent and child nodes is expressed as a conditional probability function, that is, ,use To indicate the degree of association between the parent node and the child node.

[0068] Prior probability: Based on the existing basic data and theoretical methods, the failure probability of nodes in the Bayesian network can be inferred. However, the value obtained by this calculation method is a result that has not been verified by experiments and is a subjective judgment. In the inference calculation process after this study, the prior probability refers to the node probability after the basic data is integrated based on expert experience evaluation and adjusted by belief. In probability theory, it is assumed that Represents the sample space The basic event in The prior probability can be used To express.

[0069] Posterior probability: refers to the node adjusting the probability value based on the prior probability to make it more in line with the actual situation. It mainly uses the Bayesian network reasoning method. That is, as the sample data continues to change, the probability is obtained after the information is re-corrected. The previous posterior probability will become the next prior probability, and the new probability value can be obtained in the next calculation, which is called the posterior probability. In probability theory, let the sample space The root node in is , then the root node If a fault has occurred, the event The probability of failure can be calculated by the prior probability And the known information data is calculated, denoted as .

[0070] Conditional probability: Bayesian networks can evaluate and represent uncertainty probability because of their conditional independence. Therefore, conditional probability is particularly important. Let the sample space middle is the basic event, then the event In the event that has already occurred The conditional probability of occurrence is expressed as: .

[0071] Total probability formula: Assuming the sample space The basic events are , and are mutually incompatible, then any event The marginal probability can be obtained through the total probability formula, which is expressed as follows:

[0072] .

[0073] Bayesian formula: Assume sample space The basic event in , the events have no intersection, and the Bayesian formula is expressed as: Substituting the total probability formula into the Bayesian formula yields: ; Bayesian network reasoning is a method based on probability theory to infer and derive the values ​​of each node in the Bayesian network. In the calculation process, the required variable probability parameters are calculated through the network structure and probability parameters, and the variables with known states are called causes. There are three ways of Bayesian network inference. There are supporting reasoning, diagnostic reasoning and causal reasoning. Among them, causal reasoning has the characteristics of top-down, and the probability value of the variable is calculated through the known causes; bottom-up reasoning is the characteristic of diagnostic reasoning and causal reasoning, and the specific reasons for the conclusion are deduced from the conclusion; finally, supporting reasoning provides a scientific explanation for the phenomenon that has occurred.

[0074] Causal reasoning refers to the link process of calculating the posterior probability of a node in a Bayesian network, which is top-down. It is calculated based on the prior probability of the root node and the conditional probability values ​​between each node. At this time, some nodes with known fault states in the Bayesian network are special conditions. Assuming that the child node has two fault states 0 and 1, then under the condition that the child node is in the fault state 1, the conditional probability is a given value in the Bayesian network inference and calculation process. In the comprehensive CNC equipment reliability evaluation and analysis process, the Bayesian network consists of a series of fault events such as data interruption, data duplication, data fixed offset, data zero, continuous data growth, data jump, data jitter and outliers. Among them, the failure probability of the root node is the probability of the occurrence of the basic event, and the failure probability values ​​of the sub-nodes such as software failure, communication interruption, power failure, sensor abnormality, external environment, device failure, equipment status change, severe working conditions, environmental interference, software defects, parameter errors, poor communication quality, module damage, complete failure, fixed offset, sensitivity decrease and drift can be inferred and calculated, that is, the corresponding conditional probability value when the root node fails or is in a normal state.

[0075] If there are two types of node failure states in the Bayesian network, namely normal state and failure state, they can be represented by the numbers 1 and 0 respectively. 1 indicates that the node is in an unassessable state or fails to operate normally, and 0 indicates the state of the node when it is operating normally. When calculating the known situation of leaf node failure, the probability of the node in the Bayesian network in the failure state is . That is, the probability value of the root node in the failure state is calculated by the evidence set of the leaf node.

[0076] Assume that the child node or parent node in the Bayesian network is To express, for The conditional probability value of , then the product of the probability distributions of all variable nodes is the distribution of the joint probability, that is: ; By knowing the fault status of the node, calculating the probability of the remaining child nodes in the network is the forward reasoning process of the Bayesian network. Assume that the leaf node and the root node The connection contains Node variables , The probability of being in state 1 is:

[0077] .

[0078] According to the viewpoint of system theory, accidents are caused by the failure of system functions. In order to further quantify the strength of the influence relationship between the failure modes of CNC equipment and determine the main path leading to the dangerous state of the system, a top-level event system failure is added on the basis of the hierarchical structure constructed by integrating the DEMATEL-ISM method, and it is mapped to the Bayesian network using the GeNIe network to determine the conditional probability distribution and prior probability of the root node, and a Bayesian network is established to determine the posterior probability.

[0079] Furthermore, the S106: calculating the criticality of the failure mode of the numerical control equipment based on the failure mode influence and the failure probability includes: ;in, Representing components i Hazard degree, Representing components Failure Mode The occurrence rate of the failure mode that causes the component to fail, Representing components Failure Mode The probability of a failure causing damage to the component, Representing components basic failure rate.

[0080] It should be understood that the components i Failure Mode j The severity of the failure that causes the entire system to fail , then: ; Parts i The degree of damage to the subsystem is: ; The larger the value, the greater the severity of the failure mode. Representing components The number of types of failure modes that occur; Representing components Failure Mode The occurrence rate of the failure mode that causes the component to fail; Representing components i The basic failure rate can be calculated by modeling using the historical failure data of the parts. Since the analysis result of the criticality is to compare the criticality values ​​of various parts, the purpose is to select key parts according to the size of the criticality value, so the average failure rate can be used. , approximately Values: ;in, Representing components The total number of failures within a specified time period, Representing components Accumulate working time within the specified time. is the conditional probability of the failure mode leading to a certain failure consequence, that is, the probability that the component fails in the failure mode and causes damage to the component; it is a subjective estimate, and its value is shown in Table 1. It is a rough estimate given subjectively by staff in the prior art.

[0081] Table 1 Value basis

[0082]

[0083] Aiming at the current situation that the failure mechanisms of CNC equipment are numerous and complex, the failure data are few and correlated, and the failure mechanism analysis is mostly qualitative, this paper proposes a quantitative analysis method for the criticality of the failure mechanism. First, according to the occurrence of relevant failures of CNC equipment, the strongly linked subsystem is obtained; for the strongly linked subsystem, DEMATEL / ISM is used again to obtain the fault mode level fault correlation model, and the influence of the strongly linked subsystem failure mode is obtained by matrix calculation in combination with the ANP method; the Bayes analysis method is used to calculate the probability of occurrence of the failure mode of the strongly linked subsystem, thereby realizing the quantitative analysis of the criticality of the failure mode of CNC equipment. Finally, taking a certain machining center as an example, it is verified that the proposed method can improve the accuracy and timeliness of traditional FMECA.

[0084] The present invention firstly analyzes the fault subsystem and fault mode of CNC equipment based on the principle of CNC equipment and the FMECA analysis method based on system theory. Then, based on the analysis results, the influence relationship between factors is quantified, the direct influence matrix X is constructed, the matrix X is normalized, and the comprehensive influence relationship matrix T of factors is further obtained. , determine the impact causal graph between each fault node; based on the impact causal graph between each fault node, obtain the strongly linked subsystem; use DEMATEL-ISM to analyze the correlation of fault modes of strongly linked subsystems, and combine ANP to obtain the impact of fault modes of strongly linked subsystems; map the DEMATEL-ISM related fault model to the Bayesian network according to the conversion rules, determine the prior probability of the nodes in the Bayesian network based on the actual reliability data, construct a conditional probability table, and determine the failure probability of the system; use the hazard analysis formula to obtain the hazard degree of CNC equipment failure.

[0085] Embodiment 2: This embodiment provides a criticality analysis system for numerical control equipment failure modes, including: an analysis module, which is configured to: obtain the operation data of the numerical control equipment, analyze the operation data, and obtain a set of numerical control equipment failure-related subsystems; a construction module, which is configured to: construct a strongly linked subsystem based on the set of numerical control equipment failure-related subsystems; wherein, the construction of the strongly linked subsystem based on the set of numerical control equipment failure-related subsystems includes: constructing a direct impact matrix between the failure subsystems ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , obtain the hierarchical structure of the fault subsystem; based on the influence causal diagram between each fault node, obtain the strongly linked subsystem; the strong link means that the centrality between the fault nodes is greater than the set threshold; the first determination module is configured to: analyze the strongly linked subsystem and determine the fault mode level correlation model of the strongly linked subsystem; the second determination module is configured to: determine the fault mode influence of the strongly linked subsystem based on the fault mode level correlation model of the strongly linked subsystem; the third determination module is configured to: based on the fault mode level correlation model of the strongly linked subsystem, use the Bayesian network to determine the fault probability; the calculation module is configured to: calculate the criticality of the fault mode of the CNC equipment based on the fault mode influence and the fault probability.

[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The criticality analysis method of CNC equipment failure mode is characterized by: include: Acquire operation data of numerical control equipment, analyze the operation data, and obtain a set of subsystems related to faults of the numerical control equipment; Based on the set of numerical control equipment fault-related subsystems, a strongly linked subsystem is constructed; wherein, based on the set of numerical control equipment fault-related subsystems, a strongly linked subsystem is constructed, including: constructing a direct impact matrix between fault subsystems ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , and obtain the hierarchical structure of the fault subsystem; based on the causal graph between each fault node, a strong link subsystem is obtained; the strong link means that the centrality between the fault nodes is greater than the set threshold; Analyze the strongly linked subsystems and determine the failure mode level related models of the strongly linked subsystems; Based on the failure mode level correlation model of the strongly linked subsystem, determine the failure mode impact of the strongly linked subsystem; Based on the failure mode level correlation model of strongly linked subsystems, a Bayesian network is used to determine the failure probability; Based on the failure mode impact and failure probability, the criticality of the failure mode of CNC equipment is calculated; Reachability Matrix Perform regional decomposition, divide each decomposition area into levels, and obtain the level sets, including: Defining Components The set of affected components is components The reachable set , affecting components The component collection is a component The predecessor set , by calculating the reachable set and antecedent set ,right Carry out regional decomposition and level division; Regional decomposition: Defined as a common set, the common set The elements are defined as the starting set ,for In and ,like ,but and If they are not in the same area, For the same area, thus realizing the decomposition of the area; The result of regional decomposition is denoted as ; in, For the A collection of components in relatively independent areas. ; Level division: for the same area , and get satisfaction in turn Components, find out the set of each level, expressed as: ； ； …… ； in, is the set of levels obtained by level division; According to the level division results and reachable matrix , and obtain the hierarchical structure of the fault subsystem, including: According to the level division results and reachable matrix , for the reachability matrix Remove the skipped binary relations and the binary relations that reach themselves, and obtain the skeleton matrix, which is used as the hierarchical structure model of the fault subsystem; Based on the causal influence graph between each fault node, a strongly linked subsystem is obtained, including: dividing the causal graph by setting a coordinate axis threshold, deleting nodes with a centrality less than the set threshold, and using the deleted causal graph as a strongly linked subsystem; Analyze the strong link subsystem and determine the failure mode level related models of the strong link subsystem, including: When there is a correlated failure between one failure mode and another failure mode, the two failure modes are considered to be correlated failure modes; When there is no related failure mode between one failure mode and another failure mode, the two failure modes are considered not to be related failure modes; The set of related failure modes determined thereby , Representative a failure mode that has related failures with other failure modes, represents the number of failure modes with related failures; Based on the relevant failure mode set, a failure mode hierarchical model is constructed; the failure mode hierarchical model is a failure mode level correlation model of strongly linked subsystems; Based on the relevant failure mode set, a failure mode hierarchical model is constructed, including: Constructing a direct impact matrix between failure modes ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , and obtain the fault mode hierarchical structure; Based on the failure mode level correlation model of the strongly linked subsystem, a Bayesian network is used to determine the failure probability, including: constructing a Bayesian network; constructing a Bayesian network, including: mapping the failure modes in the failure mode level correlation model of the strongly linked subsystem into Bayesian network nodes, mapping the causal relationship between the failure modes in the failure mode level correlation model of the strongly linked subsystem into directed edges in the Bayesian network, and mapping the recursive hierarchical structure in the failure mode level correlation model of the strongly linked subsystem into a structure in the Bayesian network; determining the posterior probability of the Bayesian network.

2. The criticality analysis method of numerical control equipment failure mode according to claim 1, characterized in that: The operation data is analyzed to obtain a set of fault-related subsystems of the numerical control equipment, including: when there is a related fault between one subsystem and another subsystem, the two subsystems are considered to be related subsystems; when there is no related fault between one subsystem and another subsystem, the two subsystems are considered not to be related subsystems; determine the set of fault-related subsystems constituting the numerical control equipment , Representative A subsystem that has related faults with other subsystems. Indicates the number of faulty subsystems where related faults exist.

3. The criticality analysis method of numerical control equipment failure mode according to claim 2, characterized in that: Construct direct impact matrix between faulty subsystems ,include: ； in, For subsystem Impact Subsystem The number of impacts, hour, , Indicates the number of faulty subsystems with correlated faults; Based on the normalized matrix , calculate the comprehensive impact relationship matrix , based on the comprehensive impact relationship matrix , determine the impact causal graph between each fault node, including: ； in, is the identity matrix, is the standardized matrix, Representation Subsystem Subsystem of Indirect impact of the stage, Indicates that there is a fault related to other subsystems. subsystems, Indicates that there is a fault related to other subsystems. subsystems, , , ; Assume comprehensive influence relationship matrix The influence of behavior and , comprehensive impact relationship matrix The sum of the columns is the degree of influence ; Using the centrality as the horizontal axis and the cause degree as the vertical axis, draw the influence cause and effect diagram between each fault node of the CNC equipment.

4. The criticality analysis method of numerical control equipment failure mode according to claim 3 is characterized in that: Based on the comprehensive impact relationship matrix , calculate the overall impact matrix ,include: ； in, is the identity matrix, Indicates that after considering the impact of the faulty subsystem on itself, the subsystem Subsystem The extent of direct and indirect impacts Indicates the number of faulty subsystems with correlated faults; Based on the overall impact matrix , determine the reachability matrix ,include: ; in, The value range of is 1~n; The value range of is 1~n; Value: .

5. CNC equipment failure mode criticality analysis system, characterized by: include: An analysis module is configured to: obtain operation data of the numerical control equipment, analyze the operation data, and obtain a set of subsystems related to the fault of the numerical control equipment; A construction module is configured to: construct a strongly linked subsystem based on a set of numerical control equipment fault-related subsystems; wherein, the construction of a strongly linked subsystem based on a set of numerical control equipment fault-related subsystems includes: constructing a direct impact matrix between fault subsystems ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , determine the impact causal diagram between each fault node; based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , and obtain the hierarchical structure of the fault subsystem; based on the causal graph between each fault node, a strong link subsystem is obtained; the strong link means that the centrality between the fault nodes is greater than the set threshold; A first determination module is configured to: analyze the strong link system and determine a failure mode level correlation model of the strong link subsystem; A second determination module is configured to: determine the failure mode impact of the strong link subsystem based on the failure mode level correlation model of the strong link subsystem; A third determination module is configured to: determine the failure probability using a Bayesian network based on a failure mode level correlation model of the strongly linked subsystems; A calculation module is configured to: calculate the criticality of the failure mode of the numerical control equipment based on the failure mode influence and the failure probability; Reachability Matrix Perform regional decomposition, divide each decomposition area into levels, and obtain the level sets, including: Defining Components The set of affected components is components The reachable set , affecting components The component collection is a component The predecessor set , by calculating the reachable set and antecedent set ,right Carry out regional decomposition and level division; Regional decomposition: Defined as a common set, the common set The elements are defined as the starting set ,for In and ,like ,but and If they are not in the same area, For the same area, thus realizing the decomposition of the area; The result of regional decomposition is denoted as ; in, For the A collection of components in relatively independent areas. ; Level division: for the same area , and get satisfaction in turn Components, find out the set of each level, expressed as: ； ； …… ； in, is the set of levels obtained by level division; According to the level division results and reachable matrix , and obtain the hierarchical structure of the fault subsystem, including: According to the level division results and reachable matrix , for the reachability matrix Remove the skipped binary relations and the binary relations that reach themselves, and obtain the skeleton matrix, which is used as the hierarchical structure model of the fault subsystem; Based on the causal influence graph between each fault node, a strongly linked subsystem is obtained, including: dividing the causal graph by setting a coordinate axis threshold, deleting nodes with a centrality less than the set threshold, and using the deleted causal graph as a strongly linked subsystem; Analyze the strong link subsystem and determine the failure mode level related models of the strong link subsystem, including: When there is a correlated failure between one failure mode and another failure mode, the two failure modes are considered to be correlated failure modes; When there is no related failure mode between one failure mode and another failure mode, the two failure modes are considered not to be related failure modes; The set of related failure modes determined thereby , Representative a failure mode that has related failures with other failure modes, represents the number of failure modes with related failures; Based on the relevant failure mode set, a failure mode hierarchical model is constructed; the failure mode hierarchical model is a failure mode level correlation model of strongly linked subsystems; Based on the relevant failure mode set, a failure mode hierarchical model is constructed, including: Constructing a direct impact matrix between failure modes ; Direct impact matrix Perform normalization to obtain the normalized matrix ; Based on the normalized matrix , calculate the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix ; Based on the comprehensive impact relationship matrix , calculate the overall impact matrix ; Based on the overall impact matrix , determine the reachability matrix ; For the reachable matrix Perform regional decomposition, divide each decomposed region into levels, and obtain the level sets; according to the level division results and the reachability matrix , and obtain the fault mode hierarchical structure; Based on the failure mode level correlation model of the strongly linked subsystem, a Bayesian network is used to determine the failure probability, including: constructing a Bayesian network; constructing a Bayesian network, including: mapping the failure modes in the failure mode level correlation model of the strongly linked subsystem into Bayesian network nodes, mapping the causal relationship between the failure modes in the failure mode level correlation model of the strongly linked subsystem into directed edges in the Bayesian network, and mapping the recursive hierarchical structure in the failure mode level correlation model of the strongly linked subsystem into a structure in the Bayesian network; determining the posterior probability of the Bayesian network.

Images