A failure analysis standard process adjustment method and system

Abstract

The application relates to a failure analysis standard process adjustment method and system, and belongs to the technical field of failure analysis. The method solves the problem that the existing failure analysis standard process cannot be dynamically adjusted and generated. The yield of a fault point is obtained based on historical maintenance records, the yield of the fault point is analyzed through clustering, normal points in the fault point are obtained, and the fault occurrence probability of the normal points is counted. Based on a failure analysis knowledge graph, the fault occurrence probability of the normal points is taken as the fault occurrence probability of a corresponding detection step entity. According to the detection step entity and the relationship between the entities, detection steps with sequences under each detection process are obtained as an initial step set. According to the fault occurrence probability of the detection step entity, the order of the detection steps in the initial step set is adjusted, and the adjusted failure analysis detection process under each detection process is obtained according to the adjusted initial step set. The dynamic adjustment of the failure analysis standard process is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of failure analysis, in particular to a failure analysis standard process adjustment method and system. BACKGROUND

[0002] Failure analysis is generally based on failure mode and phenomenon, through analysis and verification, simulates and reproduces the phenomenon of failure, finds out the cause of failure, and excavates the mechanism of failure. Failure analysis has strong practical significance in improving product quality, technical development, improvement, product repair and arbitration of failure accidents. With the increase of equipment, the storage capacity of industrial failure analysis data will increase exponentially.

[0003] In the prior art, for a defective component, the failure phenomenon can be obtained according to manual diagnosis, and then based on the experience of engineers, the failure analysis process is determined through the failure phenomenon, and the defective component is detected step by step to obtain the final fault point.

[0004] The existing failure analysis process is not unified, and the detection steps and detection order are related to the industry knowledge and experience of engineers, and the analysis efficiency is low. SUMMARY

[0005] In view of the above analysis, the embodiments of the present application aim to provide a failure analysis standard process adjustment method and system to solve the problem that the existing failure analysis standard process cannot be dynamically adjusted and generated.

[0006] In one aspect, the embodiments of the present application provide a failure analysis standard process adjustment method, comprising the following steps:

[0007] Based on the historical maintenance record, the yield of the fault point is obtained, the yield of the fault point is analyzed by clustering analysis, and the normal point in the fault point is obtained; based on the historical failure analysis record, the failure occurrence probability of the normal point is counted;

[0008] Based on the failure analysis knowledge graph, the part entity name associated with the normal point name and each detection step entity under each detection process is matched, the failure occurrence probability of the normal point is taken as the failure occurrence probability of the corresponding detection step entity; according to the detection step entity and the relationship between entities, the detection steps with sequence under each detection process are obtained as an initial step set;

[0009] According to the failure occurrence probability of the detection step entity, the detection step sequence in each initial step set is adjusted, and according to the adjusted each initial step set, the adjusted failure analysis detection process is obtained.

[0010] Based on the further improvement of the above method, the failure analysis knowledge graph includes: physical structure class entities, fault phenomenon class entities, detection process class entities, and the relationships between entities; wherein, the physical structure class entities include part entities and circuit entities, the fault phenomenon class entities include detection item entities and test command entities; the detection process class entities include detection step entities and detection result entities.

[0011] Based on the further improvement of the above method, based on historical maintenance records, the yield of the fault point is obtained, and the normal points in the fault points are obtained by clustering the yield of the fault points, including:

[0012] Based on historical maintenance records, the yield of the fault points in the period is counted;

[0013] Remove the fault points whose yield does not conform to the sigma principle to obtain the fault points to be clustered;

[0014] Using a density clustering algorithm, the fault points to be clustered are clustered according to the yield of the fault points in the same period and the preset neighborhood radius, and the clustered categories are obtained; the fault points in the category whose number is greater than or equal to the number threshold are taken as normal points.

[0015] Based on the further improvement of the above method, according to the detection step entities and the relationships between entities, the detection steps with order under each detection process are obtained as the initial step set, including:

[0016] Based on the failure analysis knowledge graph, the detection step entities under each detection process are obtained;

[0017] For each detection process, the detection step entity with the preset first step operation step code is taken as the first step operation, the detection step entities corresponding to the next operation relationship of the first step operation associated with detection normal and abnormal are recursively obtained respectively until the current detection step entity does not exist the next operation relationship of detection normal and abnormal, and the initial step set of the current detection process is obtained.

[0018] Based on the further improvement of the above method, according to the detection step entities and the relationships between entities, the detection steps with order under each detection process are obtained as the initial step set, including:

[0019] Based on the failure analysis knowledge graph, the detection step entities associated with the part entities under each detection process are obtained, according to the circuit entities associated with the part entities, the detection steps corresponding to the part entities belonging to the same circuit are put into each circuit set according to the order attribute value of the part entities;

[0020] For each detection process, the total probability of each line set is summarized according to the failure occurrence probability of the detection step entity as the first probability, and each line set is sorted in descending order according to the first probability; merging each line set, adjusting the relationship between adjacent detection step entities in the line set, and obtaining the initial step set of the current detection process.

[0021] Based on the further improvement of the above method, the detection step sequence in each initial step set is adjusted according to the failure occurrence probability of the detection step entity, including:

[0022] For each detection process, identify whether there is a line set, if there is, take each line set as a step set to be moved; otherwise, according to the next step operation of the detection step entity in the initial step set, get the step set to be moved;

[0023] Identify whether there is a strong correlation mark in each step set to be moved, if not, move the detection step entity in descending order according to the failure occurrence probability of the detection step entity; otherwise, the detection step entity with strong correlation mark is taken as a linkage step, the failure occurrence probability of each linkage step is obtained, the failure occurrence probability of each linkage step in the same line set is compared with the failure occurrence probability of the detection step entity without strong correlation mark, and each is put into the corresponding line set in descending order, wherein the linkage step is moved together;

[0024] Update the relationship between detection step entities in the step set to be moved.

[0025] Based on the further improvement of the above method, the step set to be moved is obtained according to the next step operation of the detection step entity in the initial step set, including:

[0026] Get the detection step entity with different next step operations of detection normal and detection abnormal in the initial step set as a branch node;

[0027] In turn, each branch node is respectively according to the next step operation on the two branches of its detection normal and detection abnormal, using recursive algorithm, sequentially obtaining the detection step entity and putting it into the corresponding step set to be moved, until there is no next step operation;

[0028] If the first detection step entity in the initial step set is not a branch node, the detection step entity from the first detection step entity to the first branch node is put into a step set to be moved.

[0029] Based on the further improvement of the above method, the detection step entity with strong correlation mark is obtained by the following steps:

[0030] Based on all detection step entities and the corresponding part entities belonging to the circuit, multiple detection step entities belonging to the same circuit are considered as transactions, and each detection step entity is considered as a project. Using the generalized sequence pattern algorithm, multiple frequent sequence sets are obtained according to the preset support and confidence. The detection step entities corresponding to each frequent sequence set are marked with a strong correlation label; different frequent sequence sets correspond to a unique strong correlation label.

[0031] Based on further improvements to the above method, and according to the adjusted set of initial steps, the adjusted failure analysis and detection process is obtained, including:

[0032] For each adjusted initial step set, extract the entities of each detection step in sequence and merge them as the base steps;

[0033] Based on the operation type attribute value of each detection step entity, obtain the corresponding operation step description template, fill the operation step description template with the physical structure class entity name or test command attribute of the test command entity associated with the detection step entity, and add it to the basic steps.

[0034] Based on the next step operation relationship between the normal and / or abnormal detection of each detection step entity, obtain the corresponding detection step entity, take the step number attribute value as the corresponding next step operation, and add it to the basic steps.

[0035] Based on the detection result entity associated with the detection step entity, the name of the physical structure entity associated with the detection result entity is used as the detection fault point and added to the basic steps;

[0036] Output the basic steps to obtain the adjusted standard failure analysis process.

[0037] On the other hand, embodiments of the present invention provide a failure analysis standard process adjustment system, including:

[0038] The fault occurrence probability calculation module is used to obtain the yield rate of fault points based on historical maintenance records, obtain the normal points among the fault points through cluster analysis of the yield rate of fault points, and calculate the fault occurrence probability of normal points based on historical failure analysis records.

[0039] The detection step acquisition module is used to match the normal point name with the part entity name associated with each detection step entity under each detection process based on the failure analysis knowledge graph, and use the failure probability of the normal point as the failure probability of the corresponding detection step entity; according to the detection step entity and the relationship between entities, the sequential detection steps under each detection process are obtained as the initial step set.

[0040] The detection step adjustment module is used for adjusting the sequence of the detection steps in each initial step set according to the failure occurrence probability of the detection step entity, and obtaining an adjusted failure analysis detection process according to each adjusted initial step set.

[0041] Compared with the prior art, the application can realize at least one of the following beneficial effects: the historical data is fully utilized, the relationship of failure analysis is mined, the failure analysis knowledge graph is constructed, the related knowledge query of the failure analysis standard process is assisted to be realized, and the modification of the failure analysis standard process by an artificial is assisted; the detection steps are divided and the strong correlation between the detection steps is obtained by using the line principle and the physical structure of the parts, so that the accuracy of the detection step adjustment is improved; the failure occurrence probability of the parts is analyzed by analyzing the historical maintenance records, and the failure occurrence probability is used as the basis for the detection step adjustment, so that the detection step of the part with high failure occurrence probability is moved forward, the efficiency of the failure analysis is improved, and the adjusted failure analysis standard process is dynamically generated, and the convenience of use of the user is increased.

[0042] In the application, the above technical solutions can be combined with each other to realize more preferred combination solutions. Other features and advantages of the application will be described in the subsequent description, and some advantages will become apparent from the description, or will be understood by implementing the application. The purposes and other advantages of the application can be realized and obtained from the contents specifically pointed out in the description and the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0043] The accompanying drawings are included to provide a further understanding of the application and are incorporated herein and constitute a part of the application. The drawings illustrate embodiments of the application and, together with the description, serve to explain the principles of the application.

[0044] Figure 1 A failure analysis standard process adjustment method flowchart in embodiment 1 of the application;

[0045] Figure 2 A tree diagram of each detection step in the detection process in embodiment 1 of the application;

[0046] Figure 3 A tree diagram of the detection step adjusted based on the line set in embodiment 1 of the application;

[0047] Figure 4 A tree diagram of the detection step adjusted without the line set in embodiment 1 of the application. DETAILED DESCRIPTION

[0048] The preferred embodiments of the application are specifically described below with reference to the drawings, wherein the drawings constitute a part of the application and are used to explain the principles of the embodiments of the application, and are not used to limit the scope of the application.

[0049] Embodiment 1

[0050] One specific embodiment of the present application discloses a failure analysis standard process adjustment method, as shown in the figure, comprising the following steps: Figure 1

[0051] S11: Based on the historical maintenance record, the yield of the fault point is obtained, the yield of the fault point is analyzed by clustering, and the normal point in the fault point is obtained; based on the historical failure analysis record, the failure probability of the normal point is counted;

[0052] S12: Based on the failure analysis knowledge graph, the normal point name and the part entity name associated with each detection step entity under each detection process are matched, and the failure probability of the normal point is taken as the failure probability of the corresponding detection step entity; according to the detection step entity and the relationship between entities, the detection steps with sequence under each detection process are obtained as the initial step set;

[0053] S13: According to the failure probability of the detection step entity, the detection step sequence in each initial step set is adjusted, and according to the adjusted each initial step set, the adjusted failure analysis detection process is obtained.

[0054] In implementation, the relevant failure analysis standard process is queried through the fault phenomenon, the fault parts are detected according to the failure analysis standard process, and the fault point is located. In order to improve the analysis efficiency, the maintenance record and the failure analysis record are used to optimize the detection step in the failure analysis standard process, and the failure analysis knowledge graph is updated.

[0055] It should be noted that the process of failure analysis of the defective assembly by the maintenance engineer according to the failure analysis standard process is recorded in the historical failure analysis record, which details the failure analysis standard process code, the defective assembly code, and the final fault point (i.e. the part). In step S11, the historical maintenance record is the record of the maintenance engineer according to the fault point in the historical failure analysis record, which records whether the fault point is repaired and the repair time, etc. According to whether the fault point is repaired, the yield of the fault point, i.e. the probability of the fault point being repaired, is counted according to the period.

[0056] It should be noted that the period can be a month or a week, and the statistical range is determined according to the actual maintenance situation, such as the number of historical maintenance records and the frequency of maintenance. Exemplarily, the yield of the fault point in the last 6 months is counted.

[0057] ​After obtaining the yield of all failure points, remove the failure points whose yield does not conform to the sigma principle to obtain the failure points to be clustered; adopt the density clustering algorithm, and according to the yield of the failure points to be clustered in the same period and the preset neighborhood radius, the failure points to be clustered are clustered to obtain each category of the clustering; the failure points in the category whose number is greater than or equal to the number threshold are taken as normal points.

[0058] Exemplarily, the three-sigma principle based on the normal distribution is used, when the yield of the failure point is less than the average value-3 sigma, it belongs to the abnormal point, and needs to be removed from the failure point. Using the DBSCAN algorithm, according to the yield of the remaining failure points in the last month, the remaining failure points are clustered. Based on the clustering result, if the number of failure points in the category is less than the number threshold, the failure points in the category are outliers, otherwise, the failure points in the category are normal points.

[0059] Based on the historical failure analysis record, the failure occurrence probability of the normal point is counted according to the period, that is, the probability of the normal point being judged as a failure point in all failure analysis records of the assembly to which it belongs. It should be noted that the period can also be a month or a week, and the time period range can be the same as or different from the time period range for statistical yield in the clustering analysis.

[0060] It should be noted that step S11 determines the parts with high failure occurrence probability based on the analysis of the historical maintenance records and the historical failure analysis records, and when performing failure analysis, the parts with high failure occurrence probability are detected first, which facilitates faster positioning of the failure point.

[0061] In step S12, the detection steps of the parts, the lines and each detection process are stored in the failure analysis knowledge graph as knowledge entities. The failure analysis knowledge graph is constructed by utilizing and mining historical data of failure analysis, and includes physical structure class entities, failure phenomenon class entities, detection process class entities, and relationships between entities; wherein, the physical structure class entities include part entities and line entities, the failure phenomenon class entities include detection item entities and test command entities; the detection process class entities include detection step entities and detection result entities.

[0062] Specifically, the construction of the failure analysis knowledge graph includes steps S101-S106, as follows:

[0063] S101: According to the bill of materials and the netlist, the physical structure class entities and their attributes and relationships are established, and the physical structure class entities include part entities and line entities.

[0064] It should be noted that the bill of materials is also called product structure table, material list, and the raw materials, spare parts, assemblies of the product are disassembled, and each item of material is recorded according to the order of manufacturing process according to material code, name, specification, unit, quantity, loss, etc., arranged into a list, used to obtain assembly information in the field of failure analysis; the net list is used to describe the connection relationship between parts in the assembly.

[0065] The embodiment obtains the parts and lines and their related attributes by analyzing the bill of materials, and instantiates them as part entities and line entities in the knowledge graph; by analyzing the net list, the association relationship between the parts and the lines, and the order of each part on the same line are obtained, and the triple relationship between the part entities and the line entities is established, and the order attribute value is added in the part entity, so that the maintenance engineer can detect the parts according to the order and reduce repeated operations.

[0066] Exemplarily, Component, Description, VendorName and VendorPartNo in the bill of materials BOM are respectively mapped to the attribute values of part name, detailed description of the part, supplier name and supplier code in the part entity. If there is a case that an assembly includes multiple parts, the assembly code is added in the attribute of the part entity.

[0067] It should be noted that each entity has a unique entity identifier, which is generated according to the preset rules, exemplarily, by splicing multiple attribute values, so as to facilitate subsequent direct acquisition of related attribute values according to the entity identifier.

[0068] S102: According to the test file and the test log, the fault phenomenon class entity and its attributes and relationships are established, and the fault phenomenon class entity includes detection item entity and test command entity.

[0069] It should be noted that the test file is a collection of detection items for each fault phenomenon of the assembly, and is used to obtain the detection items; the test log is a log file recorded by the test machine when testing the assembly according to the test file, and is used to obtain the test commands, test standard return values and other related information required during testing.

[0070] In this embodiment, when the test file and the test log are parsed, the respective to-be-parsed contents are intercepted from the original files according to the file format and content identifier; based on the TF-IDF algorithm, the to-be-parsed content of the test file content is extracted for keywords, and a keyword is obtained for each detection item. The keyword is used as the detection item name, and the specific test command and the standard return value when the test is passed corresponding to each detection item are extracted from the to-be-parsed content of the test log, and are respectively instantiated as the detection item entity and the test command entity; and the association relationship between the detection item entity and the test command entity is established according to the keyword.

[0071] S103: According to the historical failure analysis standard process, the detection process class entity and its attributes and relationships are established, and the detection process class entity includes detection step entity and detection result entity.

[0072] It should be noted that the historical failure analysis standard process is generally a manually written detection process, which is used to obtain each detection step and its determination result under each detection process, and the determination result indicates the faulty parts.

[0073] In this embodiment, by analyzing the historical failure analysis standard process, the operation step name, operation type, operation step description, operation step return value type, next step operation name of detection normal and abnormal are extracted; according to the preset rule, the operation step number, the number and name of the detection process to which the operation step belongs are generated, and are instantiated as detection step entity; since the next step operation of detection normal and abnormal is also instantiated as detection step entity, the next step operation relationship of detection normal and / or abnormal between detection step entities is established by using the next step operation; the determination result and the related part name are extracted, instantiated as detection result entity, and the corresponding physical structure class entity is obtained according to the related part name, and the association relationship between the detection result entity and the detection step entity and the physical structure class entity is established.

[0074] Specifically, the operation type of the detection step includes visual inspection, measurement, disassembly and sending command, etc.

[0075] S104: According to the relationship between the historical failure analysis standard process and the test file, the relationship between the detection project entity and the detection step entity is established.

[0076] It should be noted that based on the domain business knowledge, the detection file and the historical failure analysis standard process used for each assembly fault phenomenon are known, one test file extracts multiple detection projects, corresponds to one detection process, and one detection process has multiple detection steps. In this embodiment, by establishing the association relationship between the detection project name and the detection process name, the detection step is obtained according to the detection project name, that is, the relationship between the detection project entity and the detection step entity is established.

[0077] S105: Based on the physical structure class entity, the operation step description attribute value in the detection step entity is processed to obtain the step segmentation result, and the relationship between the physical structure class entity and the detection step entity is established according to the step segmentation result.

[0078] It should be noted that the operation step description attribute value in the detection step entity is a specific operation step written by a person, and usually an operation step involves a detection object and a detection method. The detection object is a physical structure class entity that has been established in the knowledge graph. In order to improve the accuracy of word segmentation, a corpus is constructed based on corpus data in the industrial analysis field, and the established physical structure class entity name is used as an auxiliary library. The word segmentation result is obtained by using jieba word segmentation.

[0079] The word segmentation result of the operation step description is used to extract the relationship between entities, including:

[0080] The step word segmentation result is compared with the entity name of the physical structure class entity. If they are the same, the entity corresponding to the entity name is taken as the tail entity, and the detection step entity corresponding to the step word segmentation result is taken as the head entity. The operation type attribute value in the head entity is taken as the operation relationship, and a triple of <head entity-operation relationship-tail entity> is established.

[0081] Exemplarily, the operation step description of the detection step entity G004 is: "measure whether the power size of C2000 is between (10dBm~30dBm), please input the measurement result", and the word segmentation result obtained by jieba word segmentation is: "measure whether the power size is between (B~C) between, please input the measurement result", where A=C2000, B=10dBm, and C=30dBm. A, B, and C are compared with the entity name of the physical structure class entity. A corresponds to the part entity with the entity name C2604, and a triple of <G004-measure-C2000> is established, that is, the detection step entity G004 is associated with the part entity C2000.

[0082] Considering some special cases, such as non-standard operation step description, and the operation type is sending a command, the operation step description involves a test command. For these special cases, the operation step description that cannot obtain the entity relationship is further analyzed by step S105.

[0083] S106: Based on the failure analysis file writing specification, extract the part-of-speech sequence template of each operation type. After performing part-of-speech tagging on the step word segmentation result, compare the step word segmentation result with the part-of-speech sequence template. For the consistent step word segmentation result, establish the relationship between the physical structure class entity and the detection step entity, or the relationship between the detection step entity and the test command entity according to the part-of-speech sequence template.

[0084] Exemplarily, the part-of-speech sequence template is: v n x n, the operation step of the detection step entity K002 is described as: "check pre-charge resistor, pre-charge relay", and after part-of-speech tagging, the following is obtained: check / v pre-charge resistor / n, / x pre-charge relay / n, which is consistent with the part-of-speech sequence template, if the pre-charge resistor and the pre-charge relay corresponding entity can be queried in the knowledge graph, then check is taken as the relationship of the triple, and the relationship of the detection step entity and the battery pack, the pre-charge resistor and the pre-charge relay entity is established respectively.

[0085] Through the construction of the knowledge graph, the utilization rate of historical data is improved. When the assembly fails, the failure analysis detection process is generated by searching the fault phenomenon keywords and the detection process entities and their attributes and relationships under the detection process obtained by searching.

[0086] In step S12, first, the normal point name and the part entity name associated with each detection step entity under each detection process in step S11 are matched, and the fault occurrence probability of the normal point is taken as the fault occurrence probability of the corresponding detection step entity.

[0087] It should be noted that some detection step entities associated with parts entities are not fault points, and the fault occurrence probability of the detection step entity cannot be matched with the normal point name, and the fault occurrence probability of the detection step entity is 0.

[0088] Before adjusting the detection steps under each detection process, the detection steps with sequence under each detection process are obtained as an initial step set according to the detection step entities and the relationships between the entities, and the embodiment provides two methods:

[0089] Method 1: based on the failure analysis knowledge graph, the detection step entities under each detection process are obtained;

[0090] For each detection process, the detection step entity with the preset first step operation step code is taken as the first step operation, the next step operation corresponding detection step entity is recursively obtained according to the next step operation relationship of the first step operation associated with the detection normal and abnormal, until the current detection step entity does not exist the next step operation relationship of the detection normal and abnormal, and the initial step set of the current detection process is obtained;

[0091] Exemplarily, there are 10 detection steps under one detection process, in order to be easier to explain, the relationship between the detection steps is presented in a tree structure as follows: Figure 2 wherein the numbers 1-10 correspond to the 10 detection step numbers, Y represents detection normal, N represents detection abnormal, the operation step code of step 1 is the preset first step operation step code A001, and the initial step set is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10} obtained by recursively detecting the normal branch first and then recursively detecting the abnormal branch.

[0092] Method 2: Based on the failure analysis knowledge graph, the part entities associated with each detection step entity under each detection process are obtained, according to the line entities associated with the part entities, and according to the order attribute value of the part entities, the detection steps corresponding to the parts belonging to the same line are put into each line set;

[0093] For each detection process, the total probability of each line set is summarized according to the failure occurrence probability of the detection step entity as the first probability, and each line set is sorted in descending order according to the first probability; merge each line set, adjust the relationship between adjacent detection step entities in the line set, and get the initial step set of the current detection process.

[0094] It should be noted that adjusting the relationship between adjacent detection step entities in the line set includes:

[0095] Identify whether the first detection step entity in the previous line set in the adjacent two line sets has a next operation relationship of detection normal and abnormal at the same time, if it exists, then the next operation relationship of detection step entity in the current line set which is not associated is associated to the first detection step entity in the next line set; if not, according to the detection result entity associated with the last detection step entity in the previous line set, according to the determination result of abnormal or normal in the detection result entity, the corresponding detection normal or abnormal next operation relationship between the first detection step entity in the next line set is established.

[0096] Exemplarily, Figure 2 The probability value in is the failure occurrence probability of the detection step, and the parts of steps 1, 3 and 7 have not failed, so the normal point name cannot be matched, and the failure occurrence probability is 0. If {1, 2}, {3, 4, 5, 6} and {7, 8, 9, 10} belong to 3 line sets respectively, the detection steps in each line set have been arranged in line order, according to the failure occurrence probability corresponding to each detection step, the total probability of {1, 2} is 30%, the total probability of {3, 4, 5, 6} is 20%, and the total probability of {7, 8, 9, 10} is 50%, then according to the order of {7, 8, 9, 10} {1, 2} {3, 4, 5, 6}, if the fault point in the determination result of step 10 has the determination normal, the next operation relationship of detection abnormal is established, and step 1 is associated, and the initial step set is {7, 8, 9, 10, 1, 2, 3, 4, 5, 6}.

[0097] It should be noted that the above method 1 is suitable for detecting a relatively simple process, and the detection steps involved in the detection process are not many. Method 2 considers the probability of fault occurrence, combines the principles of the line and the physical structure of the parts, and places the detection steps belonging to the same line together, preferentially detects the steps with high overall fault occurrence probability, and sorts the detection steps according to the order of the parts in the line to reduce the actions of the maintenance engineers and improve work efficiency. The two methods can be dynamically selected according to the actual situation, such as: marking the detection steps corresponding to the same line part entity with a line mark, if there is no line mark in the detection process, select method 1, otherwise select method 2.

[0098] In step S13, the order of the detection steps in the initial step set is adjusted according to the fault occurrence probability of the detection step entity, including:

[0099] ① For each detection process, identify whether there is a line set, if there is, then take each line set as the step set to be moved; otherwise, according to the next step operation of the detection step entity in the initial step set when the detection is normal and abnormal, obtain the step set to be moved.

[0100] It should be noted that since the detection step usually needs to select the next step operation according to the actual detection situation, such as: visual inspection C9400, normal input "1", abnormal input "0", when the next step operation corresponding to the detection normal and the detection abnormal is different, branch operation occurs. Therefore, when adjusting the steps according to the fault occurrence probability, the steps need to be adjusted within the range of each step that can be moved, and cannot be moved randomly in the entire detection process or across the line set.

[0101] Therefore, when method 2 is used in step S12 and the line set has been divided, each line set is taken as the step set to be moved, and the movement of the detection step can only be in the current line set; when method 1 is used in step S12, while maintaining the branch nodes in the initial detection steps, the detection step entities that are continuous and not branch nodes on each branch are put into a step set to be moved. Specifically, it includes:

[0102] Obtain the detection step entities in the initial step set that have both detection normal and abnormal next step operations and different next step operations as branch nodes;

[0103] In turn, for each branch node, according to the next step operation on the two branches of its detection normal and detection abnormal, respectively, use the recursive algorithm to sequentially obtain the detection step entities and put them into the respective step set to be moved until there is no next step operation;

[0104] If the first detection step entity in the initial step set is not a branch node, the first detection step entity to the first detection step entity of the branch node is put into a to-be-moved step set.

[0105] Exemplarily, in the step 1 and 3 are branch nodes, and three to-be-moved step sets are obtained: {2}, {4, 5, 6} and {7, 8, 9, 10}. Figure 2

[0106] It should be noted that, when obtaining the to-be-moved step set, if the detection step entity corresponding to the next operation already exists in any to-be-moved step set, the new to-be-moved step set is not put in, but the relationship between the two is recorded additionally, and the relationship is added after the step adjustment. This situation is mainly for some important parts in a detection process, which are detected multiple times. At this time, the next operation of a certain detection step is the step that has been detected before, so that the relationship between the two is temporarily disconnected. Figure 2 For example, the next operation of the step 6 can be the step 2, at this time, the relationship between the step 6 and the step 2 is temporarily disconnected, and the step 2 is not put in the to-be-moved step set {4, 5, 6}, but the relationship between the step 6 and the step 2 is recorded additionally.

[0107] ②Identify whether there is a strong correlation mark in each to-be-moved step set, if not, sort the detection step entities according to the fault occurrence probability from large to small; otherwise, the detection step entity with the strong correlation mark is taken as a linkage step, the fault occurrence probability of each linkage step is obtained, the fault occurrence probability of each linkage step in the current same line set is compared with the fault occurrence probability of the detection step entity without the strong correlation mark, and the detection step entities are put into the corresponding line set in the order from large to small, wherein the linkage steps are moved together.

[0108] It should be noted that, in the embodiment, some detection steps need to be executed continuously, therefore, the detection step entities of the linkage operation are marked with a strong correlation mark, the probabilities are summarized, and the detection efficiency is improved while ensuring the accuracy of the detection step adjustment.

[0109] Specifically, the strong correlation mark of the detection step entity is obtained by the following steps:

[0110] Based on all detection step entities and the lines to which the parts corresponding to the detection step entities belong, a plurality of detection step entities belonging to the same line are taken as a transaction, each detection step entity is taken as an item, a generalized sequence pattern algorithm GSP algorithm is used, a plurality of frequent sequence sets are obtained according to the preset support and confidence, the detection step entities corresponding to each frequent sequence set are marked with a strong correlation mark, and different frequent sequence sets correspond to unique strong correlation marks. ​

[0111] Exemplarily, in the case shown in Figure 2 , step 1 and step 2 have the same strong correlation mark, the probability obtained by aggregation is 30%, step 3 and step 4 have the same strong correlation mark, the probability obtained by aggregation is 4%, step 7 and step 8 have the same strong correlation mark, the probability obtained by aggregation is 20%, and step 9 and step 10 have the same strong correlation mark, the probability obtained by aggregation is 30%. In the case where the line set exists, the line set {1, 2} does not need to be adjusted, the line set {7, 8, 9, 10} is adjusted to {9, 10, 7, 8}, and the line set {3, 4, 5, 6} is adjusted to {6, 5, 3, 4}. The final adjusted detection steps are as shown in Figure 3 . In the case where the line set does not exist, the to-be-moved step set {2} does not need to be adjusted. Since step 4 is strongly related to step 3, and 3 does not belong to the to-be-moved set {4, 5, 6}, the to-be-moved set {4, 5, 6} is adjusted to {4, 6, 5}, the to-be-moved set {7, 8, 9, 10} is adjusted to {9, 10, 7, 8}, and the final adjusted detection steps are as shown in Figure 4 .

[0112] ③ Update the relationship between the detection step entities in the to-be-moved step set.

[0113] Specifically, it includes:

[0114] Take a group of adjacent detection step entities in each to-be-moved step set in turn, judge whether there is a next step operation relationship of normal detection and / or abnormal detection between the group of detection step entities, if not, obtain the detection result entity associated with the previous detection step entity, and establish the corresponding next step operation relationship of normal detection or abnormal detection between the next detection step entity according to the determination result of abnormal or normal in the detection result entity, and take the next group of adjacent detection step entities until the end of traversal.

[0115] It should be noted that, according to the determination result of abnormal or normal in the detection result entity, the corresponding next step operation relationship of normal detection or abnormal detection between the next detection step entity is established, which includes:

[0116] When the determination result in the detection result entity is normal and the corresponding fault point is not empty, the previous detection step entity and the next detection step entity establish a next operation relationship of detecting abnormal; when the determination result in the detection result entity is abnormal and the corresponding fault point is not empty, the previous detection step entity and the next detection step entity establish a next operation relationship of detecting normal; when the determination result in the detection result entity is empty, the previous detection step entity and the next detection step entity establish a next operation relationship of detecting normal simultaneously; when the previous detection step entity has both normal and abnormal determination results and the fault points are not empty, the fault point corresponding to the determination result of normal is emptied, and the previous detection step and the next detection step establish a next operation relationship of detecting normal.

[0117] Exemplarily, the next operation relationship between step 10 and step 7, step 3 and step 6, step 6 and step 5, and step 5 and step 4 is updated; if the determination result of step 10 only has the fault point of determining normal, the next operation relationship of detecting abnormal is established with step 7. In addition, if the relationship between steps is additionally recorded when the set of moving steps is obtained, whether the relationship still exists between the entities after moving is judged, and if not, the relationship is established again. Figure 3

[0118] Further, if the operation step code of the first detection step entity in the adjusted initial step set is not the preset first step operation step code, the operation step code of the first detection step entity in the initial step set before adjustment is changed to a new operation step code, and the relationship with the previous detection step entity is updated. Exemplarily, in the initial step set before adjustment, the operation step code of step 1 is regenerated, and the operation step code of step 7 is set to the preset A001. Figure 3

[0119] According to the adjusted initial step set, an adjusted failure analysis detection flow under each detection flow is obtained, including:

[0120] Each detection step entity in the adjusted initial step set is taken out in sequence and merged as a basic step;

[0121] According to the operation type attribute value of each detection step entity, a corresponding operation step description template is obtained, and the test command attribute of the physical structure class entity name or the test command entity associated with the detection step entity is filled into the operation step description template to obtain an operation step description, which is added to the basic step;

[0122] According to the next operation relationship of detecting normal and / or abnormal of each detection step entity, a corresponding detection step entity is obtained, and the step number attribute value is taken as a corresponding next operation, which is added to the basic step;​​

[0123] According to the detection result entity associated with the detection step entity, the physical structure class entity name associated with the detection result entity is added to the basic step as a detection fault point.

[0124] The output basic step is obtained to obtain the adjusted failure analysis standard process.

[0125] Exemplarily, Table 1 is a generated failure analysis standard process. The step number and operation type in Table 1 are directly derived from the attribute value of the detection step entity, the operation step description is filled into the corresponding operation step description template according to the associated part entity, and the next step is to obtain the step number of the detection step entity according to the next operation corresponding to the detection normal (0) and / or detection abnormal (1). The fault point is obtained according to the detection result corresponding to the fault part when the detection is normal (0) and / or the detection result (1) when the detection is abnormal.

[0126] Table 1 Failure analysis standard process example

[0127]

[0128] Compared with the prior art, the failure analysis standard process adjustment method of the embodiment fully utilizes historical data, mines the relationship of failure analysis, constructs a failure analysis knowledge graph, assists in realizing related knowledge query of the failure analysis standard process, and assists in realizing manual modification of the failure analysis standard process; utilizes line principles and part physical structures to divide detection steps and obtain strong correlations between detection steps, thereby improving the accuracy of detection step adjustment; analyzes historical maintenance records to analyze the fault occurrence probability of the part, which is used as a detection step adjustment basis, and moves the detection step of the part with high fault occurrence probability, thereby improving the efficiency of failure analysis, and dynamically generating an adjusted failure analysis standard process, thereby increasing the convenience of user use.

[0129] Embodiment 2

[0130] Another embodiment of the application discloses a failure analysis standard process adjustment system, so as to realize the failure analysis standard process adjustment method in embodiment 1. The specific implementation mode of each module is referred to the corresponding description in embodiment 1. The system comprises:

[0131] The fault occurrence probability calculation module is configured to obtain the yield of the fault point based on the historical maintenance record, analyze the yield of the fault point through clustering, obtain the normal point in the fault point, and count the fault occurrence probability of the normal point.

[0132] The detection step acquisition module is configured to match the part entity name associated with the normal point name and each detection step entity under each detection process based on the failure analysis knowledge graph, and take the failure occurrence probability of the normal point as the failure occurrence probability of the corresponding detection step entity; and acquire the detection steps with sequence under each detection process as the initial step set according to the detection step entity and the relationship between the entities.

[0133] The detection step adjustment module is configured to adjust the detection step sequence in each initial step set according to the failure occurrence probability of the detection step entity, and obtain the adjusted failure analysis detection process according to each adjusted initial step set.

[0134] It should be noted that there are usually many detection steps in the failure analysis standard process, and all the detection step information can be seen in Table 1 after derivation. When used, the step execution needs to be checked. Preferably, the system of the embodiment is used to query the corresponding detection process according to the failure phenomenon of the faulty assembly, and automatically display the first step operation. After the maintenance engineer performs the operation step, the actual detection value is input, the system automatically judges, and the corresponding fault point is displayed, or the next step operation is displayed, which is convenient for the maintenance engineer to use.

[0135] Exemplarily, in Table 1, step A001 is associated with part C9400, the associated detection result is that the fault point is C9400 when the detection is not normal, and the next step operation relationship of the detection normal is established with step A002. Therefore, when performing failure analysis, the maintenance engineer first displays the operation type and operation step description of step A001. When the detected part C9400 has a problem, the actual detection value 1 is input, the fault point C9400 is obtained, and one failure analysis is completed. When the detected part C9400 has no problem, the actual detection value 0 is input, step A002 is executed, and the operation type and operation step description of step A002 are displayed. When the detected part R3300 has no problem, the actual detection value 0 is input, and step A003 is executed, until the fault point is obtained, and one failure analysis is completed.

[0136] Since the failure analysis standard process adjustment system of the embodiment is related to the foregoing generation method, they can be mutually referred to, and this is repeated description, which will not be described here. Since the system embodiment and the method embodiment have the same principle, the system embodiment also has the corresponding technical effects of the method embodiment.

[0137] Those skilled in the art can understand that all or part of the processes of the above-mentioned embodiments can be completed by a computer program instructing related hardware. The program can be stored in a computer readable storage medium, such as a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0138] The above description is only the preferred embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A method for adjusting a standard failure analysis process, characterized in that, Includes the following steps: Based on historical maintenance records, the yield rate of fault points is obtained, and the normal points among the fault points are obtained through cluster analysis of the yield rate of fault points; based on historical failure analysis records, the probability of failure occurrence of normal points is statistically analyzed. Based on the failure analysis knowledge graph, the names of normal points are matched with the names of the parts entities associated with the entities of each inspection step under each inspection process. The failure probability of the normal point is used as the failure probability of the corresponding inspection step entity. Based on the inspection step entities and the relationships between entities, the sequential inspection steps under each inspection process are obtained as the initial step set. Based on the failure probability of the detection step entity, the order of the detection steps in each initial step set is adjusted, and the adjusted failure analysis detection process is obtained based on the adjusted initial step sets. The step of adjusting the order of detection steps in each initial step set according to the failure probability of the detection step entity includes: For each inspection process, identify whether a set of lines exists, where the set of lines includes the inspection steps corresponding to the part entities belonging to the same line; if it exists, use each set of lines as the set of steps to be moved; otherwise, obtain the set of steps to be moved based on the next operation of the inspection step entities in the initial set of steps, whether the inspection is normal or abnormal. Identify whether there is a detection step entity with a strong correlation mark in each set of steps to be moved; if not, move the detection step entities from largest to smallest according to their failure probability; otherwise, the detection step entities with strong correlation marks are treated as linked steps that need to be executed together continuously, obtain the failure probability of each linked step, compare the failure probability of each linked step in the current same line set with the failure probability of the detection step entities without strong correlation marks, and put them into the corresponding line set in descending order, where linked steps move together; Update the relationships between the entities in the detection steps set to be moved.

2. The failure analysis standard process adjustment method according to claim 1, characterized in that, The failure analysis knowledge graph includes: physical structure entities, fault phenomenon entities, detection process entities, and the relationships between entities; among them, physical structure entities include component entities and circuit entities, fault phenomenon entities include detection item entities and test command entities, and detection process entities include detection step entities and detection result entities.

3. The failure analysis standard process adjustment method according to claim 2, characterized in that, The process of obtaining the yield rate of fault points based on historical maintenance records, and then using cluster analysis to analyze the yield rate of fault points to obtain normal points among the fault points, includes: Based on historical maintenance records, the yield rate of fault points is calculated on a periodic basis. Remove fault points whose yield does not conform to the sigma principle to obtain the fault points to be clustered; A density clustering algorithm is used to cluster the fault points to be clustered based on the yield rate of the fault points in the same period and the preset neighborhood radius, so as to obtain the cluster categories; the fault points in the categories with a number of fault points greater than or equal to the number threshold are regarded as normal points.

4. The failure analysis standard process adjustment method according to claim 3, characterized in that, The step involves obtaining sequential detection steps within each detection process based on the entities involved in the detection steps and the relationships between these entities, forming an initial set of steps, including: Based on the failure analysis knowledge graph, obtain the entities of each detection step under each detection process; For each detection process, the detection step entity with the preset first step operation code is taken as the first step operation. Based on the next step operation relationship of normal and abnormal detection associated with the first step operation, the detection step entity corresponding to the next step operation is recursively obtained until the current detection step entity has no next step operation relationship of normal and abnormal detection, thus obtaining the initial step set of the current detection process.

5. The failure analysis standard process adjustment method according to claim 3, characterized in that, The step involves obtaining sequential detection steps within each detection process based on the entities involved in the detection steps and the relationships between these entities, forming an initial set of steps, including: Based on the failure analysis knowledge graph, the part entities associated with each detection step under each detection process are obtained. According to the circuit entities associated with the part entities, the detection steps corresponding to the part entities belonging to the same circuit are put into each circuit set according to the sequential attribute values ​​of the part entities. For each detection process, based on the failure probability of the detection step entity, the total probability of each line set is summarized as the first probability, and the line sets are sorted from largest to smallest according to the first probability; the line sets are merged, and the relationship between adjacent detection step entities in the line sets is adjusted to obtain the initial step set of the current detection process.

6. The method for adjusting the standard failure analysis process according to claim 1, characterized in that, The step of obtaining the set of steps to be moved based on the normal and abnormal detection of the detection step entities in the initial set of steps includes: Obtain the detection step entities in the initial step set that have both normal and abnormal detection next operations, and whose next operations are different, as branch nodes; For each branch node, based on the next operation on its two branches, which are detected as normal and those as abnormal, a recursive algorithm is used to sequentially obtain the detection step entities and put them into their respective sets of steps to be moved, until there is no next operation; If the first detection step entity in the initial step set is not a branch node, then the detection step entity from the first detection step entity to the first branch node is placed into a set of steps to be moved.

7. The failure analysis standard process adjustment method according to claim 1, characterized in that, The detection step entity with strong correlation markers is obtained through the following steps: Based on all detection step entities and the corresponding part entities belonging to the circuit, multiple detection step entities belonging to the same circuit are considered as transactions, and each detection step entity is considered as a project. Using the generalized sequence pattern algorithm, multiple frequent sequence sets are obtained according to the preset support and confidence. The detection step entities corresponding to each frequent sequence set are marked with a strong correlation label; different frequent sequence sets correspond to a unique strong correlation label.

8. The method for adjusting the standard failure analysis process according to claim 1, characterized in that, The adjusted failure analysis and detection process is obtained based on the adjusted set of initial steps, including: For each adjusted initial step set, extract the entities of each detection step in sequence and merge them as the base steps; Based on the operation type attribute value of each detection step entity, obtain the corresponding operation step description template, fill the operation step description template with the physical structure class entity name or test command attribute of the test command entity associated with the detection step entity, and add it to the basic steps. Based on the next step operation relationship between the normal and / or abnormal detection of each detection step entity, obtain the corresponding detection step entity, take the step number attribute value as the corresponding next step operation, and add it to the basic steps. Based on the detection result entity associated with the detection step entity, the name of the physical structure entity associated with the detection result entity is used as the detection fault point and added to the basic steps; Output the basic steps to obtain the adjusted standard failure analysis process.

9. A failure analysis standard process adjustment system, characterized in that, include: The fault occurrence probability calculation module is used to obtain the yield rate of fault points based on historical maintenance records, and to obtain the normal points among the fault points through cluster analysis of the yield rate of fault points; Based on historical failure analysis records, the probability of failure occurring at normal points is statistically analyzed. The detection step acquisition module is used to match the normal point name with the part entity name associated with each detection step entity under each detection process based on the failure analysis knowledge graph, and use the failure probability of the normal point as the failure probability of the corresponding detection step entity. Based on the entities in the detection steps and the relationships between entities, obtain the sequential detection steps under each detection process, and use them as the initial set of steps; The detection step adjustment module is used to adjust the order of detection steps in each initial step set according to the failure probability of the detection step entity, and obtain the adjusted failure analysis detection process according to each initial step set after adjustment. The process by which the detection step adjustment module adjusts the order of detection steps in each initial step set based on the failure probability of the detection step entity includes: For each inspection process, identify whether a set of lines exists, where the set of lines includes the inspection steps corresponding to the part entities belonging to the same line; if it exists, use each set of lines as the set of steps to be moved; otherwise, obtain the set of steps to be moved based on the next operation of the inspection step entities in the initial set of steps, whether the inspection is normal or abnormal. Identify whether there is a detection step entity with a strong correlation mark in each set of steps to be moved; if not, move the detection step entities from largest to smallest according to their failure probability; otherwise, the detection step entities with strong correlation marks are treated as linked steps that need to be executed together continuously, obtain the failure probability of each linked step, compare the failure probability of each linked step in the current same line set with the failure probability of the detection step entities without strong correlation marks, and put them into the corresponding line set in descending order, where linked steps move together; Update the relationships between the entities in the detection steps set to be moved.

Images