CN117406689A - Data driving and knowledge guiding fault diagnosis method and system - Google Patents
Data driving and knowledge guiding fault diagnosis method and system Download PDFInfo
- Publication number
- CN117406689A CN117406689A CN202311334125.3A CN202311334125A CN117406689A CN 117406689 A CN117406689 A CN 117406689A CN 202311334125 A CN202311334125 A CN 202311334125A CN 117406689 A CN117406689 A CN 117406689A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- knowledge
- equipment
- mode
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 76
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000002159 abnormal effect Effects 0.000 claims abstract description 30
- 238000001514 detection method Methods 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims abstract description 9
- 239000000203 mixture Substances 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 9
- 230000000007 visual effect Effects 0.000 claims description 9
- 230000005856 abnormality Effects 0.000 claims description 8
- 230000001149 cognitive effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000001364 causal effect Effects 0.000 claims description 4
- 238000005065 mining Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 238000007500 overflow downdraw method Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 8
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 230000010485 coping Effects 0.000 abstract description 2
- 230000006872 improvement Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 8
- 230000035945 sensitivity Effects 0.000 description 5
- 239000008186 active pharmaceutical agent Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 239000000470 constituent Substances 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000013024 troubleshooting Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/0227—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions
- G05B23/0229—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions knowledge based, e.g. expert systems; genetic algorithms
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/041—Abduction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Automation & Control Theory (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The equipment fault diagnosis technology is to judge whether the current state is normal or not according to the related information recorded in the running process of the equipment, identify the nature and the position of the fault, locate the cause of the fault and give out corresponding coping strategies. The invention discloses a data driving and knowledge guiding fault diagnosis method and system, which are used for detecting abnormal states of equipment by collecting and analyzing operation state data of industrial equipment in real time and adopting a big data abnormal detection method, realizing rapid positioning of fault sources based on constructed multi-mode fault knowledge maps, and developing prediction and early warning of potential fault risks, thereby being beneficial to timely detecting, rapid and accurate diagnosing and early warning of equipment faults, improving equipment robustness and use efficiency and realizing cost reduction and efficiency improvement of enterprises.
Description
Technical Field
The invention belongs to the technical field of equipment fault diagnosis, and particularly relates to a fault diagnosis method and system for data driving and knowledge guiding.
Background
With the continuous development of social informatization and continuous progress of scientific technology, the production efficiency and the automation degree of industrial equipment are higher and higher, and the working strength is increased continuously. Meanwhile, the complexity and the refinement degree of the equipment are higher and higher, and the components are more closely related, so that a fault chain reaction is often generated, even potential hidden danger or tiny faults can possibly cause the whole equipment to be faulty and even generate larger catastrophic damage, such as environmental pollution and the like, and the consequences are extremely serious. Therefore, the function of the equipment fault diagnosis technology is more and more obvious, the existing or to-be-faulty equipment can be discovered at an early stage, the fault cause can be rapidly positioned, a processing scheme is given, and the fault development trend is predicted, so that the fault discovery maintenance time is greatly shortened, the maintenance quality is improved, the maintenance cost is saved, and the equipment availability and the robustness are further improved.
The equipment fault diagnosis technology is to judge whether the current state is normal or not according to the related information recorded in the running process of the equipment, identify the nature and the position of the fault, locate the cause of the fault and give out corresponding coping strategies. With the increasing size and complexity of modern industrial equipment and systems, the problems of reliability, availability, maintainability and safety are also highlighted, so that the research of fault diagnosis mechanisms and fault diagnosis technologies is promoted. The current fault diagnosis method mainly comprises an expert system, a genetic algorithm, data mining, an artificial neural network, a fuzzy set theory and the like, but any single fault diagnosis technology is difficult to adapt to the requirements of modern complex system fault diagnosis along with the continuous expansion of equipment scale and the increase of diversity of fault types. Therefore, a new intelligent fault diagnosis technology is explored on the basis of the existing fault diagnosis method, so that the fault discovery, quick diagnosis and troubleshooting capability are improved, potential fault risks can be predicted in advance, and the current new requirements on the fault diagnosis method and system are met.
In view of the above, there is a need for a data driving and knowledge guiding fault diagnosis method and system based on a new generation artificial intelligence technology, which can realize reliable and interpretable rapid detection, accurate diagnosis and early warning of equipment faults, and improve the robustness and the use efficiency of the equipment.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a data driving and knowledge guiding fault diagnosis method and system, and aims to solve the problems that the prior art cannot fully utilize equipment-related multi-source multi-mode running state data, is difficult to quickly and accurately locate a fault source in the face of a large amount of fault information, and is difficult to realize reliability and interpretability of fault diagnosis and prediction results.
The invention relates to a fault diagnosis method for data driving and knowledge guiding, which comprises the following steps:
s10, continuously acquiring equipment operation state data, and recording equipment multi-source multi-mode operation state information in real time.
S20, analyzing, processing, collecting and recording equipment operation state data in real time, mining and identifying abnormal state information, and synchronously extracting fault related knowledge from the data.
S30, performing fault diagnosis and fault prediction based on a multi-mode fault knowledge graph according to abnormal state information, analyzing and reasoning that the fault is a fault, hidden danger or risk, and realizing accurate positioning of a fault source.
S40, fault diagnosis and prediction results are displayed based on the visual interface, and diagnosis basis is displayed.
Further, the step S10 specifically includes: the equipment operation state comes from a plurality of data sources, including sensors, system logs and manual input, and the data formats are images, audio, video, texts and structured records; aiming at multi-source multi-mode data, based on a multi-mode data fusion method, information fusion association of the multi-source data on a semantic level is realized, and real-time running states of equipment are recorded from multiple dimensions.
Further, the step S20 specifically includes: through real-time monitoring of multi-source multi-mode time sequence data, abnormal detection for the time sequence data is achieved by using a transducer, and potential abnormal information in the multi-source multi-mode data is mined.
Further, body modeling is conducted on equipment composition and fault modes, meanwhile, dependence or inclusion relation among all composition modules in the equipment, corresponding relation with the fault modes, port input-output relation and mutual influence relation among all fault modes are built, sub-graph matching association fusion is conducted on the port input-output relation and the mutual influence relation with a pre-built fault knowledge graph, and knowledge updating operation is conducted on the current existing knowledge graph according to requirements.
Further, the step S30 specifically includes: aiming at the detected abnormal information, carrying out fault diagnosis and tracing based on a pre-constructed fault knowledge graph, realizing the rapid positioning of a fault source and providing a solution; and meanwhile, fault prediction reasoning is carried out, so that potential risks or early warning of faults of the equipment are realized.
Further, based on a pre-constructed multi-mode fault knowledge graph, aiming at a plurality of fault point information, the rapid tracing for the multi-point fault is realized by a graph calculation method, the real fault source is accurately positioned, the diagnosis reasoning basis is recorded, and the reliability and the interpretability of the diagnosis result are ensured. The constructed multi-mode fault knowledge graph is a large-scale directed graph, and the equipment fault mode codes are entity attributes. For N fault mode codes, firstly, respectively searching paths upwards in a fault knowledge graph to generate N corresponding paths P 1 ,P 2 ,......,P N The method comprises the steps of carrying out a first treatment on the surface of the Then, sub-graph matching is carried out on the N paths to obtain M (M is less than or equal to N) sub-graphs; and finally, respectively searching root nodes in the M subgraphs, wherein the obtained K (K is less than or equal to M) nodes are source fault points corresponding to the current N faults.
Further, fault prediction reasoning is carried out through a cognitive reasoning network, based on a pre-constructed multi-mode fault knowledge graph, potential hidden danger or risk abnormality information is predicted through a causal reasoning method according to occurrence probability of each fault history aiming at a plurality of pieces of fault point information, fault points possibly generated in the future are output, analysis reasoning basis is generated, and reliability and interpretability of reasoning results are ensured. The cognitive reasoning network is a directed acyclic graph (Directed Acyclic Graph, DAG) and is marked as a triplet G= { V, E, F }, wherein a vertex set V is a non-empty set of nodes in the graph; edge setEach edge is represented by a node pair as (x, y), which is called x as a starting point and y as an ending point; f is a set of relationships, each relationship F (x, y) corresponding to a relationship between a node pair (x, y).
Further, the step S40 specifically includes: the method is characterized in that equipment fault phenomena, reasons and processing methods are displayed through a visual interface, potential hidden danger or risk abnormality information deduced by fault prediction is displayed, possible fault points and fault modes in the future are displayed, and prediction basis is visually displayed through an inference chain mode.
Further, based on the fault knowledge characteristics of the field of the equipment, a fault ontology model is built, and an automatic extraction of fault knowledge SPO (main Predicate-project-Object) triples is realized by combining top-down and bottom-up, so that a fault knowledge map is continuously built. The fault knowledge graph is expressed in the form of (entity, relation, entity) or (entity, relation, attribute) triples, wherein the multi-mode information such as pictures, audios, videos and the like can be used as a single entity or can be used as an entity attribute. Based on the ontology model, the association between the equipment composition and the hierarchy structure and the fault modes and the causal influence relation between the fault modes are established, meanwhile, a certain fault history occurrence probability statistical attribute (comprising fault occurrence times and occurrence probability two attributes) is provided, and the integrity of the semantic description of the fault modes is realized through the complementation of multi-mode information such as texts, images, audios and videos. For each equipment failure relationship represented by a triplet, there are four basic attributes, namely co-occurrence (co-occurrence), probability of co-occurrence (probability), specificity (specificity), reliability (reliability).
The invention relates to a data driving and knowledge guiding fault diagnosis system, which can realize a data driving and knowledge guiding fault diagnosis method, and comprises the following parts:
and the data acquisition module is used for continuously acquiring multi-source and multi-mode data related to the running state of the recording equipment.
And the fault detection module is used for analyzing, processing and collecting recorded equipment running state data in real time and identifying abnormal state information.
And the knowledge extraction module is used for extracting fault related knowledge from the data after the association processing and continuously updating the constructed multi-mode fault knowledge graph.
And the fault diagnosis module is used for carrying out fault diagnosis, fault tracing and fault prediction based on the multi-mode fault knowledge graph according to the abnormal state, and analyzing and reasoning that the fault is a fault, hidden danger or risk.
The fault display module displays diagnosis basis of fault diagnosis, tracing, prediction results and interpretability based on the visual interface.
The invention has the beneficial effects that
1. The fault mode knowledge of the equipment can be analyzed and mined from massive equipment running state data, and continuous construction and knowledge updating of a fault knowledge graph can be realized; meanwhile, the occurrence probability of faults of each component module of the equipment is counted based on historical data, so that the fault diagnosis and prediction accuracy is further improved.
2. Based on the multi-mode knowledge graph, equipment fault diagnosis and prediction are carried out, the related knowledge of fault events can be obtained from multiple dimensions, and unified characterization is realized.
3. The multi-mode knowledge graph based on entity and event fusion is used for carrying out equipment fault cognition reasoning, so that the timeliness and accuracy of fault diagnosis and prediction can be remarkably improved, and the reliability and the interpretability of diagnosis and prediction results are further improved.
Drawings
Fig. 1 is a flowchart of a fault diagnosis method of the present invention.
Fig. 2 is a diagram showing the components of the fault diagnosis system according to the present invention.
Fig. 3 is a diagram illustrating an ontology model of a fault knowledge graph according to the present invention.
FIG. 4 is a diagram illustrating a chain of evidence for failure prediction according to the present invention.
Detailed Description
The following description of the technical solutions in the embodiments of the present invention will be clear and complete, and it is obvious that the described embodiments are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the data driving and knowledge guiding fault diagnosis method of the present invention includes the steps of:
s10, continuously acquiring equipment operation state data, and recording equipment multi-source multi-mode operation state information in real time.
S20, analyzing, processing and collecting recorded equipment operation state data in real time, and mining and identifying abnormal state information.
S30, performing fault diagnosis and fault prediction based on a multi-mode fault knowledge graph according to abnormal state information, analyzing and reasoning that the fault is a fault, hidden danger or risk, and realizing rapid and accurate positioning of a fault source.
S40, fault diagnosis, prediction results and diagnosis basis are displayed based on the visual interface, and reliability and interpretability of the results are achieved.
As shown in fig. 2, the data driving and knowledge guiding fault diagnosis system of the present invention comprises the following parts:
and the data acquisition module is used for continuously acquiring multi-source and multi-mode data related to the running state of the recording equipment.
And the fault detection module is used for analyzing, processing and collecting recorded equipment running state data in real time and identifying abnormal state information.
And the knowledge extraction module is used for extracting fault related knowledge from the data after the association processing and continuously updating the constructed fault knowledge graph.
And the fault diagnosis module is used for carrying out fault diagnosis, fault tracing and fault prediction based on a fault knowledge graph according to the abnormal state, and analyzing and reasoning that the fault is a fault, hidden danger or risk.
The fault display module displays diagnosis basis of fault diagnosis, tracing, prediction results and interpretability based on the visual interface.
Specifically, the step S10 is executed in the data acquisition module, and the detailed implementation process is as follows:
and recording the running state information of the multi-source multi-mode equipment in real time. The device operating state may come from a number of data sources including sensors, system logs, manual inputs, etc., and the data formats may be images, audio, video, text, structured recordings, etc. Aiming at multi-source multi-mode data, the data acquisition module realizes information fusion and association of the multi-source data on a semantic level based on a multi-mode data fusion method, and records real-time running states of equipment from multiple dimensions. Under the condition of fully utilizing the complementarity of multi-source multi-mode data, the multi-dimensional three-dimensional representation of the equipment failure phenomenon is realized.
Each type of device and its constituent parts have a unique ID number to illustrate the distinction, for the variety of device types and the complexity of the structural composition. When the data acquisition module acquires and analyzes the multi-source multi-mode running state data of the equipment in real time, the running state information belonging to the same equipment or a composition structure is subjected to association processing through the ID, and then the associated multi-mode running state data is sent to the fault detection and knowledge extraction module.
Specifically, the step S20 is executed in the fault detection and knowledge extraction module, and the detailed implementation process is as follows:
and analyzing and processing the running state data of the equipment in real time. By monitoring multi-source multi-mode time sequence data in real time, abnormal detection for the time sequence data is realized by adopting a transducer, and potential abnormal information in the multi-source multi-mode data is mined. The abnormal state of the equipment at a certain moment is described by a plurality of modal data, the unification of the same abnormality and different modalities on the semantic level is realized through multi-modal representation learning and fusion, the complementation description is carried out on the same abnormality representation, the real-time extraction capability of the abnormal event is formed, and the abnormal event is converted into a fault mode.
And extracting the composition structure and the port input-output relation from the equipment operation state data to realize knowledge updating of the knowledge graph. By analyzing the data, identifying the component structure names and the upper and lower relationships related to the equipment from the data, and forming a plurality of triples (module A, dependence, module B) or (module A, including module B); meanwhile, the port input-output relation inside the equipment or among the equipment is analyzed from the data to form a port input-output relation triplet (port A, input and port B). Aiming at the triples extracted from the equipment operation state data, carrying out sub-graph matching association fusion on the triples and constructed fault knowledge graphs, and carrying out knowledge updating operation on the current existing knowledge graphs according to the requirement.
Specifically, the knowledge extraction operation in step S20 is performed based on an ontology model designed in advance, and the detailed implementation process of the ontology model is as follows:
and combining the field characteristics of the equipment, carrying out body modeling aiming at equipment composition and fault modes, and simultaneously establishing dependence or inclusion relation among all composition modules in the equipment, corresponding relation with the fault modes, port input-output relation and mutual influence relation among all fault modes. An example of such an ontology model is shown in fig. 3. Specifically, two attributes of the occurrence number and occurrence probability of the fault are increased for each type of fault, and are defined as follows:
the number of occurrences: n (F) i ) Representing a fault F i Is the number of occurrences of (a).
Probability of occurrence:where F represents the set of faults, |f| represents the total number of faults.
For each equipment failure relationship represented by a triplet, there are four basic attributes, namely co-occurrence (co-occurrence), probability of co-occurrence (probability), specificity (specificity), reliability (reliability). Number of co-occurrence times N (S) i ,O ij ) Representing subject entities S given a relationship R i And object entity O ij The number of co-occurring faults.
The co-occurrence probability is used for measuring the occurrence probability of object entities under the premise of a given subject entity, and the formula is defined as follows:
specificity means object entity O ij With subject entity S i Co-occurrence probability at O ij The ratio in co-occurrence probability with all subject entities S is:
and is also provided with
Where i=1, 2, …, |s|; j=1, 2, …, J i And J i Given a relationship R, with S i Number of related object entities.
The specificity uses the relative value of the probability value with respect to the original co-occurrence probability, enabling capturing of the importance of the object entity with respect to the subject entity.
Confidence is used to measure S i And O ij The degree of confidence of the relationship is defined as:
wherein,is the minimum number of co-occurrences (e.g., value 10), and C is the base trust value (e.g., value 1). The definition indicates the number of co-occurrences N (S i ,O ij ) The higher the confidence is, the higher, however, the confidence should not differ significantly when the number of co-occurrences of two different relationships is large.
One failure may have multiple phenomena and will also generally correspond to multiple solutions. Thus, to better locate fault problems, the subject related entities may be ranked.
Typically using TF/IDF as the given relation R (S i ,O ij ) Is the ranking function of (1), namely:
TFIDF(S i ,O ij )＝TF(S i ,O ij )×IDF(O ij )，
wherein the method comprises the steps ofFor all i and j, i.e. S ij Is sum of O ij A collection of related subject entities.
In the field of fault diagnosis, co-occurrence probability and specificity will be considered simultaneously when referring to the strength of the relationship between two entities. Since both values are statistically based, the reliability of these values also needs to be considered, and a new scoring function PSR, specifically defined as:
PSR(S i ,O ij )＝P(S i ,O ij )×Speci(S i ,O ij )×reliability(S i ,O ij )
the PSR function is used to rank and extract relevant object entities given the subject entity.
For the relation between abnormal examination results and faults of the examination items, the PSR function needs to be optimized as follows:
PSR(S i ,O ij )＝P(S i ,O ij )×RAR(S i ,O ij ) Wherein
N abn (S i ,O ij ) Represents S i And O ij
Co-occurrence and O ij And (5) checking the fault times of abnormal results. Care should be taken that:
(1)AR(S i ,O ij ) Representing an investigation item O ij With respect to fault S i Is an abnormality rate of (a).
(2)Representing an investigation item O ij With respect to fault S i Abnormal rate of all other faults except those.
(3)RAR(S i ,O ij ) Representing the ratio of the two values, the higher the ratio is, the investigation item O ij With respect to fault S i The more important the diagnosis of (c).
Wherein,or->Possibly 0.
The occurrence times and occurrence probability values of each fault are obtained by statistics from the accumulated historical data by adopting a data driving mode.
Specifically, the step S30 is executed in the fault diagnosis module, and the detailed implementation process is as follows:
and performing fault tracing and prediction reasoning according to the abnormal state information of the equipment. Aiming at the detected abnormal information, carrying out fault diagnosis and tracing based on a pre-constructed multi-mode fault knowledge graph, realizing the rapid positioning of a fault source and providing a solution; and meanwhile, fault prediction reasoning is carried out, so that potential risks or early warning of faults of the equipment are realized.
And realizing fault tracing in a graph calculation mode. The constructed fault knowledge graph is a large directed graph, and the equipment fault mode codes are entity attributes. For N fault mode codes, firstly, respectively searching paths upwards in a fault knowledge graph to generate N corresponding paths P 1 ,P 2 ,......,P N The method comprises the steps of carrying out a first treatment on the surface of the Then, sub-graph matching is carried out on the N paths to obtain M (M is less than or equal to N) sub-graphs; and finally, respectively searching root nodes in the M subgraphs, wherein the obtained K (K is less than or equal to M) nodes are source fault points corresponding to the current N faults.
And carrying out fault prediction reasoning through a cognitive reasoning network. The cognitive inference network is a directed acyclic graph (Directed Acyclic Graph,DAG), denoted as triplet g= { V, E, F }, where the set of vertices V is a non-empty set of nodes in the graph; edge setEach edge is represented by a node pair as (x, y), which is called x as a starting point and y as an ending point; f is a set of relationships, each relationship F (x, y) corresponding to a relationship between a node pair (x, y).
The nodes of the inference network comprise inference results R r Sub-feature classifier result R 1 ～R n 、R 1 ～R n Related parameters of rule bases K and K, which are constituent elements of set V. Wherein, the result R of the sub-feature classifier i Including a sub-feature classifier C i Observe R i Probability value P of (2) i Its sensitivity (sensitivity) M on the verification set i And specificity (specificity) Y i The method comprises the steps of carrying out a first treatment on the surface of the The relevant parameters of the rule base K are the artificial evaluation value K of the reliability of K r Its value interval is 0,1]. The cognitive reasoning network G is represented by a knowledge graph KG, and the knowledge graph is created by adopting a top-down method:
(1) Will reason the result R r And sub-feature classifier result R 1 ～R n Respectively establish triples<R r ,F(R r ,R i ),R i >,i＝1,...,n。
(2) R is R r Establishing triples with a rule base K based on<R r ,F(R i ,Y i ),Y i >。
(3) Classifier result R of sub-features i And establishing triples respectively between the triples and related parameters thereof, namely:<R i ,F(R i ,P i ),P i >，<R i ,F(R i ,M i ),M i >，<R i ,F(R i ,Y i ),Y i >。
(4) The rule base K and its reliability evaluation value K r Creating triples therebetween<K,F(K,K r ),K r >。
The knowledge graph KG structure showing the equipment failure prediction evidence chain G is shown in figure 4, wherein the reasoning R is stored r Some important parameter values in the process.
Specifically, the step S40 is executed in the fault presenting module, and the detailed implementation process is as follows:
and visually displaying fault diagnosis and prediction results. Displaying information such as equipment fault phenomena, reasons, processing methods and the like through a visual display interface, and visually displaying diagnosis reasoning basis through a reasoning chain mode; and displaying possible future fault points and fault modes according to potential hidden danger or risk abnormality information deduced by fault prediction, and visually displaying prediction basis in an inference chain mode.
Calculating inference result confidence T using DS evidence theory Rr . DS evidence theory is an imprecise reasoning theory, and is widely applied to the aspect of evidence (data) synthesis. The DS evidence theory firstly sets an identification frame theta which comprises all assumptions; then, probability is allocated to each hypothesis, and the allocation function is called a Mass function; and finally, fusing the results based on the Dempster rule, namely:
wherein,is a normalized coefficient.
First calculate the sensitivity M of each result i And specificity Y i Parameter values. Assuming that the number of true positives is TP, the number of false positives is FP, the number of true negatives is TN, the number of false negatives is FN, and the calculation formulas of the sensitivity M and the specificity Y are as follows:that is, the sensitivity is the probability of correctly judging the positive, and the specificity is the probability of correctly judging the negative. Then, for->The calculations were performed as follows:
(1) Defining a mapping function to represent each R i Mapping relation between the parameter and the related parameter, namely: p (P) i ＝m 1 (R i )，M i ＝m 2 (R i )，Y i ＝m 3 (R i )。
(2) Solving a normalization system S:wherein n is R i Is a number of (3).
(3) Fusion sub-feature classification result R 1 ～R n Calculating the reliability T of the machine learning part e ：
Wherein,is R i Corresponding sub-feature f i Weights of (2);
(4) Fusing reliability T of machine learning part e Evaluation value K of rule base K r Calculating the credibility
The present invention is not limited to the above-described specific embodiments, and various modifications and variations are possible. Any modification, equivalent replacement, improvement, etc. of the above embodiments according to the technical substance of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A data driving and knowledge guiding fault diagnosis method is characterized in that: the method comprises the following steps:
s10, continuously acquiring equipment operation state data, and recording equipment multi-source multi-mode operation state information in real time;
s20, analyzing, processing, collecting and recording equipment operation state data in real time, mining and identifying abnormal state information, and synchronously extracting fault related knowledge from the data;
s30, performing fault diagnosis and fault prediction based on a multi-mode fault knowledge graph according to abnormal state information, analyzing and reasoning that the fault is a fault, hidden danger or risk, and realizing accurate positioning of a fault source;
s40, fault diagnosis and prediction results are displayed based on the visual interface, and diagnosis basis is displayed.
2. The data-driven and knowledge-guided fault diagnosis method according to claim 1, wherein: the step S10 specifically includes: the equipment operation state comes from a plurality of data sources, including sensors, system logs and manual input, and the data formats are images, audio, video, texts and structured records; aiming at multi-source multi-mode data, based on a multi-mode data fusion method, information fusion association of the multi-source data on a semantic level is realized, and real-time running states of equipment are recorded from multiple dimensions.
3. The data-driven and knowledge-guided fault diagnosis method according to claim 1, wherein: the step S20 specifically includes: through real-time monitoring of multi-source multi-mode time sequence data, abnormal detection for the time sequence data is achieved by using a transducer, and potential abnormal information in the multi-source multi-mode data is mined.
4. A data-driven and knowledge-guided fault diagnosis method according to claim 3, wherein: and carrying out body modeling on equipment composition and fault modes, simultaneously establishing dependence or inclusion relation among all composition modules in the equipment, corresponding relation with the fault modes, port input-output relation and mutual influence relation among all fault modes, carrying out sub-graph matching association fusion on the port input-output relation and the mutual influence relation with a pre-constructed fault knowledge graph, and carrying out knowledge updating operation on the current existing knowledge graph according to requirements.
5. The data-driven and knowledge-guided fault diagnosis method according to claim 1, wherein: the step S30 specifically includes: aiming at the detected abnormal information, carrying out fault diagnosis and tracing based on a pre-constructed fault knowledge graph, realizing the rapid positioning of a fault source and providing a solution; and meanwhile, fault prediction reasoning is carried out, so that potential risks or early warning of faults of the equipment are realized.
6. The data-driven and knowledge-guided fault diagnosis method of claim 5, wherein: based on a pre-constructed multi-mode fault knowledge graph, aiming at a plurality of fault point information, the rapid tracing to the multi-point fault is realized by a graph calculation method, the real fault source is accurately positioned, the diagnosis reasoning basis is recorded, and the reliability and the interpretability of the diagnosis result are ensured.
7. The data-driven and knowledge-guided fault diagnosis method of claim 5, wherein: performing fault prediction reasoning through a cognitive reasoning network, combining occurrence probabilities of various fault histories according to a plurality of fault point information which are constructed in advance based on a multi-mode fault knowledge graph, predicting potential hidden danger or risk abnormality information through a causal reasoning method, outputting fault points which are possibly generated in the future, generating analysis reasoning basis, and ensuring reliability and interpretability of reasoning results.
8. The data-driven and knowledge-guided fault diagnosis method according to claim 1, wherein: the step S40 specifically includes: the method is characterized in that equipment fault phenomena, reasons and processing methods are displayed through a visual interface, potential hidden danger or risk abnormality information deduced by fault prediction is displayed, possible fault points and fault modes in the future are displayed, and prediction basis is visually displayed through an inference chain mode.
9. The data-driven and knowledge-guided fault diagnosis method according to claim 1, wherein: the multi-mode fault knowledge graph is constructed specifically as follows: based on the fault knowledge characteristics of the field of equipment, a fault ontology model is built, automatic extraction of fault knowledge SPO triples is realized by combining top-down and bottom-up, a fault knowledge graph is continuously built, and the fault knowledge graph is in the form of (entity, relation, entity) or (entity, relation, attribute) triples; based on the fault ontology model, establishing association between equipment composition and hierarchy structure and fault modes, and causal influence relation between the fault modes, providing a certain fault history occurrence probability statistical attribute, and realizing the integrity of the semantic description of the fault modes through mutual complementation of multi-mode information.
10. A data driven and knowledge guided fault diagnosis system, characterized by: a method of performing the data driven and knowledge guided fault diagnosis of any one of claims 1 to 9, comprising the steps of:
the data acquisition module is used for continuously acquiring multi-source and multi-mode data related to the running state of the recording equipment;
the fault detection module analyzes, processes and collects recorded equipment operation state data in real time and identifies abnormal state information;
the knowledge extraction module extracts fault related knowledge from the data after the association processing and continuously updates the constructed multi-mode fault knowledge graph;
the fault diagnosis module is used for carrying out fault diagnosis, fault tracing and fault prediction based on a multi-mode fault knowledge graph according to the abnormal state, and analyzing and reasoning that the fault is a fault, hidden danger or risk;
the fault display module displays diagnosis basis of fault diagnosis, tracing, prediction results and interpretability based on the visual interface.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311334125.3A CN117406689A (en) | 2023-10-13 | 2023-10-13 | Data driving and knowledge guiding fault diagnosis method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311334125.3A CN117406689A (en) | 2023-10-13 | 2023-10-13 | Data driving and knowledge guiding fault diagnosis method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117406689A true CN117406689A (en) | 2024-01-16 |
Family
ID=89493623
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311334125.3A Pending CN117406689A (en) | 2023-10-13 | 2023-10-13 | Data driving and knowledge guiding fault diagnosis method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117406689A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118014564A (en) * | 2024-04-10 | 2024-05-10 | 山东和兑智能科技有限公司 | Power equipment fault diagnosis system and method based on data driving |
| CN118011990A (en) * | 2024-04-10 | 2024-05-10 | 中国标准化研究院 | Industrial data quality monitoring and improving system based on artificial intelligence |
| CN118316197A (en) * | 2024-04-12 | 2024-07-09 | 国网内蒙古东部电力有限公司 | Automatic control system and method for power system signal monitoring |
| CN118503794A (en) * | 2024-07-18 | 2024-08-16 | 武汉深捷科技股份有限公司 | Transformer substation equipment abnormality detection system and method based on multi-mode data |
| CN118568653A (en) * | 2024-08-05 | 2024-08-30 | 山东大学 | Multi-characteristic parameter-based combined electrical appliance switching equipment state sensing and fault diagnosis method |
| CN118625720A (en) * | 2024-08-09 | 2024-09-10 | 浙江浙能电力股份有限公司萧山发电厂 | Industrial control method, electronic equipment and computer readable storage medium |
-
2023
- 2023-10-13 CN CN202311334125.3A patent/CN117406689A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118014564A (en) * | 2024-04-10 | 2024-05-10 | 山东和兑智能科技有限公司 | Power equipment fault diagnosis system and method based on data driving |
| CN118011990A (en) * | 2024-04-10 | 2024-05-10 | 中国标准化研究院 | Industrial data quality monitoring and improving system based on artificial intelligence |
| CN118316197A (en) * | 2024-04-12 | 2024-07-09 | 国网内蒙古东部电力有限公司 | Automatic control system and method for power system signal monitoring |
| CN118503794A (en) * | 2024-07-18 | 2024-08-16 | 武汉深捷科技股份有限公司 | Transformer substation equipment abnormality detection system and method based on multi-mode data |
| CN118503794B (en) * | 2024-07-18 | 2024-10-11 | 武汉深捷科技股份有限公司 | Transformer substation equipment abnormality detection system and method based on multi-mode data |
| CN118568653A (en) * | 2024-08-05 | 2024-08-30 | 山东大学 | Multi-characteristic parameter-based combined electrical appliance switching equipment state sensing and fault diagnosis method |
| CN118625720A (en) * | 2024-08-09 | 2024-09-10 | 浙江浙能电力股份有限公司萧山发电厂 | Industrial control method, electronic equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114579875B (en) | Equipment fault diagnosis and maintenance knowledge recommendation system based on knowledge graph | |
| CN117406689A (en) | Data driving and knowledge guiding fault diagnosis method and system | |
| CN113723632B (en) | Industrial equipment fault diagnosis method based on knowledge graph | |
| Ladj et al. | A knowledge-based Digital Shadow for machining industry in a Digital Twin perspective | |
| US20220137612A1 (en) | Transformer fault diagnosis and positioning system based on digital twin | |
| de Andrade Melani et al. | A framework to automate fault detection and diagnosis based on moving window principal component analysis and Bayesian network | |
| Wang et al. | Crowdsourced reliable labeling of safety-rule violations on images of complex construction scenes for advanced vision-based workplace safety | |
| CN113688169B (en) | Mine potential safety hazard identification and early warning system based on big data analysis | |
| JP2018180759A (en) | System analysis system and system analysis method | |
| CN116771576B (en) | Comprehensive fault diagnosis method for hydroelectric generating set | |
| CN117393076B (en) | Intelligent monitoring method and system for heat-resistant epoxy resin production process | |
| Navinchandran et al. | Studies to predict maintenance time duration and important factors from maintenance workorder data | |
| Lyu et al. | A data-driven approach for identifying possible manufacturing processes and production parameters that cause product defects: A thin-film filter company case study | |
| CN113393084B (en) | Job ticket flow management system | |
| Zhang et al. | Labelvizier: Interactive validation and relabeling for technical text annotations | |
| Schemmer et al. | Towards meaningful anomaly detection: The effect of counterfactual explanations on the investigation of anomalies in multivariate time series | |
| Yang et al. | Industrial expert systems review: A comprehensive analysis of typical applications | |
| US20200393799A1 (en) | Information processing apparatus, information processing method, and non-transitory computer readable medium | |
| Bai et al. | Data-driven approaches: Use of digitized operational data in process safety | |
| Junhuai et al. | Fault detection method based on adversarial reinforcement learning | |
| CN112699927A (en) | Pipeline fault diagnosis method and system | |
| CN118411811B (en) | ChatGPT-based global perception early warning method, chatGPT-based global perception early warning system, medium and electronic equipment | |
| WO2024209450A1 (en) | Automated fault detection algorithm reporting and resolving issues in any type of industrial production lines based on artificial intelligence | |
| CN117150439B (en) | Automobile manufacturing parameter detection method and system based on multi-source heterogeneous data fusion | |
| Petri et al. | Information‐Enabled Decision‐Making in Big Data Scenarios |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |