CN112700016A - Predictive maintenance method and apparatus - Google Patents

Predictive maintenance method and apparatus Download PDF

Info

Publication number
CN112700016A
CN112700016A CN202011577019.4A CN202011577019A CN112700016A CN 112700016 A CN112700016 A CN 112700016A CN 202011577019 A CN202011577019 A CN 202011577019A CN 112700016 A CN112700016 A CN 112700016A
Authority
CN
China
Prior art keywords
feature vector
feature
vector
fusion
new
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.)
Granted
Application number
CN202011577019.4A
Other languages
Chinese (zh)
Other versions
CN112700016B (en
Inventor
李锐
王建华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Inspur Scientific Research Institute Co Ltd
Original Assignee
Jinan Inspur Hi Tech Investment and Development Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jinan Inspur Hi Tech Investment and Development Co Ltd filed Critical Jinan Inspur Hi Tech Investment and Development Co Ltd
Priority to CN202011577019.4A priority Critical patent/CN112700016B/en
Publication of CN112700016A publication Critical patent/CN112700016A/en
Application granted granted Critical
Publication of CN112700016B publication Critical patent/CN112700016B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • General Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Artificial Intelligence (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Game Theory and Decision Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Development Economics (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a predictive maintenance method and a device, wherein the method comprises the following steps: acquiring operation information of target equipment needing predictive maintenance; inputting the operation information into a pre-constructed knowledge graph to output a first feature vector by the knowledge graph; inputting the operation information into a pre-constructed machine learning model so as to output a second feature vector by the machine learning model; performing fusion processing on the first feature vector and the second feature vector to obtain fusion features; the fused features are input to a pre-constructed predictive model to output predictive maintenance information from the predictive model. The scheme can improve the reliability of the predictive maintenance result.

Description

Predictive maintenance method and apparatus
Technical Field
The invention relates to the technical field of computers, in particular to a predictive maintenance method and a predictive maintenance device.
Background
The predictive maintenance means that the future operation state of the equipment is predicted according to the historical operation data of the currently operated equipment, and a corresponding maintenance plan is made according to the predicted operation state. Therefore, predictive maintenance is one of the core applications in the field of smart manufacturing, and is a necessary device health management means in the digital manufacturing process.
At present, the conventional predictive maintenance method usually depends on experienced engineers to perform prediction judgment on the operation data of the current operating equipment according to historical experience, but the obtained result of predictive maintenance is low in accuracy due to the fact that the mode of manual prediction judgment is high in subjectivity.
Disclosure of Invention
The invention provides a predictive maintenance method and a predictive maintenance device, which can improve the reliability of a predictive maintenance result.
In a first aspect, embodiments of the present invention provide a predictive maintenance method,
acquiring operation information of target equipment needing predictive maintenance;
inputting the operation information into a pre-constructed knowledge graph to output a first feature vector by the knowledge graph;
inputting the running information into a pre-constructed machine learning model to output a second feature vector by the machine learning model;
performing fusion processing on the first feature vector and the second feature vector to obtain fusion features;
inputting the fused features into a pre-constructed predictive model to output predictive maintenance information from the predictive model.
In one possible design, when the operational information includes an analog signal,
after the obtaining of the operation information of the target device requiring the predictive maintenance and before the inputting of the operation information into the pre-constructed knowledge graph to output the first feature vector by the knowledge graph, further comprising:
carrying out noise reduction processing on the analog signal to obtain a noise-reduced analog signal;
and performing feature extraction on the denoised analog signal according to a plurality of preset feature dimensions to obtain feature information, wherein the feature dimensions comprise: at least one of time domain, frequency domain, and time-frequency domain;
the inputting the operation information into a pre-constructed knowledge graph to output a first feature vector by the knowledge graph comprises:
inputting the feature information as the operation information to the knowledge-graph to output a first feature vector by the knowledge-graph.
In one possible design, when the feature dimension includes the time domain, the feature information includes: at least one of a mean, variance, standard deviation, root mean square, crest factor, and form factor;
when the feature dimension includes the frequency domain, the feature information includes: at least one of a root mean square frequency, an average frequency, a center of gravity frequency, and a frequency standard deviation;
when the feature dimension includes the time-frequency domain, the feature information includes: wavelet energy characteristics.
In one possible design, the fusing the first feature vector and the second feature vector to obtain a fused feature includes:
acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure BDA0002863493870000031
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijThe preset weight value is used for representing the jth sub-vector in the ith feature vector, and m represents the number of sub-vectors in the ith feature vector;
respectively calculating the first feature vector and the second feature vector by using the weight value of each sub-vector;
normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain the fusion features.
In a possible design, the fusing the new first feature vector and the new second feature vector to obtain the fused feature includes:
calculating the similarity between the new first feature vector and the new second feature vector according to a second equation;
the second equation includes:
Figure BDA0002863493870000032
wherein, S (. mu.) isn,μm) For characterizing a feature vector munAnd a feature vector mumThe similarity of the vectors is represented by the similarity of the vectors, i is used for representing the ith vector in each feature vector, and q is used for representing the number of vectors in each feature vector;
acquiring a fusion method matched with the similarity from a preset mapping relation, wherein the mapping relation is used for representing the relations between different similarities and different fusion methods;
and carrying out fusion processing on the new first feature vector and the new second feature vector according to the fusion method to obtain the fusion features.
In one possible design, the predictive maintenance information includes at least one of: the maintenance system comprises an operation state health value, a maintenance necessity degree value, a residual service life value, at least one maintenance time and a maintenance parameter corresponding to each maintenance time;
in one possible design, the knowledge graph is implemented by a neural network and professional experience, wherein the professional experience is obtained by the operation information of a plurality of devices and the predictive maintenance result corresponding to the operation information of the plurality of devices.
In a second aspect, an embodiment of the present invention further provides a predictive maintenance apparatus, including: the system comprises an acquisition module, a first input module, a fusion module and a second input module;
the acquisition module is used for acquiring the operation information of the target equipment needing predictive maintenance;
the first input module is used for inputting the operation information acquired by the acquisition module into a pre-constructed knowledge graph so as to output a first feature vector by the knowledge graph; and
the operation information acquired by the acquisition module is input into a pre-constructed machine learning model, so that a second feature vector is output by the machine learning model;
the fusion module is used for performing fusion processing on the first feature vector and the second feature vector to obtain fusion features;
and the second input module is used for inputting the fusion characteristics obtained by the fusion module into a pre-constructed prediction model so as to output predictive maintenance information by the prediction model.
In one possible design, the fusion module is specifically configured to perform the following processing:
acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure BDA0002863493870000041
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijFor characterizing the ith feature vectorM represents the number of the subvectors in the ith feature vector;
respectively calculating the first feature vector and the second feature vector by using the weight value of each sub-vector;
normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain the fusion features.
In a third aspect, the present invention further provides an intelligent device, including: at least one memory and at least one processor;
the at least one memory to store a machine readable program;
the at least one processor is configured to invoke the machine readable program to perform the predictive maintenance method of any of the first and second aspects.
In a fourth aspect, the present invention also provides a computer-readable medium,
the computer readable medium has stored thereon computer instructions that, when executed by a processor, cause the processor to perform the predictive maintenance method of any of the first and second aspects.
According to the technical scheme, after the operation information of the target equipment needing predictive maintenance is obtained, the first feature vector is output by using the pre-constructed knowledge map, the second feature vector is output by using the pre-constructed machine learning model, then the first feature vector and the second feature vector are subjected to fusion processing to obtain fusion features, and finally the predictive maintenance information corresponding to the operation information is obtained based on the pre-constructed prediction model. Therefore, in the process of determining the predictive maintenance information based on the operation information of the target equipment, an engineer with abundant experience does not need to perform prediction judgment, and the organic combination of the knowledge graph and the machine learning model can avoid the condition of low accuracy of the prediction result caused by the subjectivity of manual predictive maintenance, so that the reliability of the predictive maintenance information is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a predictive maintenance method provided by an embodiment of the present invention;
FIG. 2 is a schematic diagram of an apparatus for predictive maintenance according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a predictive maintenance device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer and more complete, the technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention, and based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without creative efforts belong to the scope of the present invention.
As shown in fig. 1, an embodiment of the present invention provides a predictive maintenance method, which may include the steps of:
step 101: operation information of a target device requiring predictive maintenance is acquired.
The state data is physical state data such as vibration, temperature, pressure, current and voltage acquired by a sensor in real time
In this step, in an embodiment of the present invention, the operation information of the target device is physical state data such as vibration, temperature, pressure, current, and voltage collected by the sensor. Specifically, the equipment needing predictive maintenance is provided with a plurality of sensors, and each sensor can collect the operation information and the real-time state information of the equipment. The operating information thus obtained may be, for example, information collected from sensors, i.e., analog signals. Specifically, when the operation information acquired in step 101 includes an analog signal, after step 101, the acquired analog signal may be further processed, which specifically includes the following steps:
step S1: carrying out noise reduction processing on the analog signal to obtain a noise-reduced analog signal;
step S2: carrying out feature extraction on the denoised analog signal according to a plurality of preset feature dimensions to obtain feature information, wherein the feature dimensions comprise: at least one of time domain, frequency domain, and time-frequency domain.
In this step, because the analog signal generally includes many interference signals such as environmental noise signals, it is necessary to perform noise reduction processing on the acquired analog signal, and then perform feature extraction on the noise-reduced analog signal based on a preset feature dimension to obtain feature information. Therefore, the noise reduction processing can remove the interference signals in the analog signals, namely, the interference signals in the analog signals are filtered, so that the negative influence of the interference signals on the feature extraction is reduced, the feature extraction of the analog signals is easier, on one hand, the accuracy of the feature extraction can be improved, on the other hand, the efficiency of the feature extraction can be improved, and therefore, the accuracy of the predictive maintenance information is improved, and meanwhile, the efficiency of the predictive maintenance can be improved.
In the embodiment of the present invention, the feature extraction of the analog signal after noise reduction refers to a process of extracting useful information from the analog signal, that is, acquiring feature information.
It can be understood that, in this step, the more sufficient the feature information obtained by performing feature extraction on the noise-reduced analog signal is, the more accurate and comprehensive the feature vector obtained by using the obtained feature information is.
Based on this, in order to obtain the feature information of the noise-reduced simulation information as much as possible to sufficiently express the physical features of the device, different feature dimensions may be set in advance, and feature extraction may be performed through the different feature dimensions. In one embodiment of the present invention, the first and second electrodes are,
when the feature dimension includes a time domain, the feature information includes: at least one of a mean, variance, standard deviation, root mean square, crest factor, and form factor;
when the feature dimension includes a frequency domain, the feature information includes: at least one of a root mean square frequency, an average frequency, a center of gravity frequency, and a frequency standard deviation;
when the feature dimension includes a time-frequency domain, the feature information includes: wavelet energy characteristics.
It should be understood that, for the same analog signal, feature extraction is performed by using different feature dimensions, and the obtained feature information is different. Therefore, the analog signals are subjected to feature extraction by utilizing various different feature dimensions, so that corresponding feature information can be obtained under different reference systems, more feature information can be extracted, and the physical features of the equipment can be fully expressed, thereby being beneficial to subsequent processing and analysis.
Step 102: the run information is input to a pre-constructed knowledge graph to output a first feature vector from the knowledge graph.
In this step, in an embodiment of the present invention, the construction process of the knowledge graph specifically includes: firstly, constructing a main body structure of a knowledge graph under the guidance of an industry expert, acquiring knowledge through a plurality of data sources and storing the knowledge in a database; then, labeling and extracting the acquired knowledge to obtain a large number of triples, and fusing the triples to construct the association between data, including: entity alignment, attribute alignment, conflict resolution, normalization and the like; and finally, constructing a knowledge graph and verifying the knowledge graph, wherein the method comprises the following steps: completion, error correction, outer link, update, etc.
In an embodiment of the invention, the construction of the knowledge graph can be realized based on a neural network and professional experience, wherein the professional experience is obtained through the operation information of a plurality of devices and the predictive maintenance result corresponding to the operation information of the plurality of devices. Namely, constructing a knowledge graph through a neural network, and verifying the knowledge graph by using professional experience, comprising: completion, error correction, outer link, update, etc. Because the neural network is a second mode for simulating the thinking of the human brain by using the characteristics of the algorithm, data can be rapidly processed by the synergistic effect of a plurality of neurons, and the construction efficiency of the knowledge graph can be improved. Meanwhile, the construction of the knowledge graph can not leave professional experience, and the knowledge graph can be well verified by using the professional experience, so that the accuracy of the feature vector output by the knowledge graph is higher, and the accuracy and the reliability of predictive maintenance information can be improved.
Step 103: the operation information is input to a pre-constructed machine learning model to output a second feature vector by the machine learning model.
In this step, the machine learning model may also directly process the analog signals in the operation information and output corresponding feature vectors, and in an embodiment of the present invention, the process of processing the analog signals may refer to the steps S1 to S2, which are not described in detail herein.
Of course, the machine learning model may be constructed by processing collected several historical data in a machine learning manner, where the historical data includes a repair and maintenance record, an expert experience, a record of a field operator, and the like of each device.
Step 104: and performing fusion processing on the first feature vector and the second feature vector to obtain fusion features.
In the step, the first feature vector and the second feature vector are fused in consideration of the difference and the sameness between the first feature vector output by the knowledge graph and the second feature vector output by the machine learning model, so that fusion features which tend to take complementarity and difference between the feature vectors into consideration can be obtained, and the accuracy of the predictive maintenance information obtained by fusing the features is higher.
It should be understood that each feature vector is composed of one or more components that characterize the physical properties of the device, i.e., different component vectors may constitute different feature vectors.
It can be understood that each component vector in the same feature vector of different devices affects the devices to a different extent due to differences between the different devices. For example, the operating temperature has a greater effect on equipment a and a lesser effect on equipment B. Based on the above problem, in an embodiment of the present invention, the step 104 may specifically include the following steps:
step 1041: acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
step 1042: calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure BDA0002863493870000091
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijThe preset weight value is used for representing the jth sub-vector in the ith feature vector, and m represents the number of sub-vectors in the ith feature vector;
step 1043: respectively calculating a first feature vector and a second feature vector by using the weight value of each sub-vector;
step 1044: normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
step 1045: and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain fusion features.
In this step, each component in each feature vector corresponds to a preset weight value, and the weight value of each component in the feature vector can be obtained by the first equation, that is, the greater the influence degree of the component with the largest weight value in each feature vector in the feature vector is, and in the same way, the smaller the weight value of the component is, the smaller the influence degree of the component in the feature vector is. The first characteristic vector and the second characteristic vector are respectively calculated by utilizing the weight value of each component vector, so that the calculated first characteristic vector and the calculated second characteristic vector have pertinence, namely, the characteristic vector calculated based on the weight value can effectively solve the problem of focus differentiation of different component vectors, so that the accuracy degree of the fusion characteristics obtained after subsequent normalization processing and fusion processing is higher, and the reliability of predictive maintenance information is improved.
In the embodiment of the present invention, taking the first feature vector as an example, the first feature vector is α ═ (α)1,α2,Λαn) Wherein, the component amount α1Weighted value of beta1Component of alpha2Weighted value of beta2…, component αnWeighted value of betanThen, the calculated first feature vector is α' ═ α (α)1β1,α2β2,Λαnβn)。
It should be understood that, because the attributes of the above-mentioned several vectors in the calculated first feature vector and the calculated second feature vector are different, the dimensional difference is relatively large, so before the fusion processing, the normalization processing is performed on the calculated first feature vector and the calculated second feature vector respectively to obtain a new first feature vector and a new second feature vector, so that each feature vector can be under the same quantity, the stability and representativeness of the feature vectors can be improved, and the processing quantity of the subsequent fusion processing can not be too large, that is, the efficiency of the subsequent fusion processing can be ensured.
Further, for the new first feature vector and the new second feature vector, the fusion features obtained when different fusion methods are used for fusion processing may be different. Based on this, in an embodiment of the present invention, after step 1044 and before step 1045, the following steps may be further included:
calculating the similarity between the new first feature vector and the new second feature vector according to a second equation;
the second equation includes:
Figure BDA0002863493870000101
wherein, S (. mu.) isn,μm) For characterizing a feature vector munAnd a feature vector mumThe similarity of the vectors is represented by the similarity of the vectors, i is used for representing the ith vector in each feature vector, and q is used for representing the number of vectors in each feature vector;
acquiring a fusion method matched with the similarity from a preset mapping relation, wherein the mapping relation is used for representing the relations between different similarities and different fusion methods;
and carrying out fusion processing on the new first feature vector and the new second feature vector according to a fusion method to obtain fusion features.
In the embodiment of the invention, a mapping relation used for representing the similarity and the fusion method is preset, the similarity between the new first feature vector and the new second feature vector is obtained through calculation according to the second formula, the fusion method matched with the similarity is determined from the mapping relation, and then the fusion method is used for fusion processing to obtain the feature vector. Therefore, the similarity between the feature vectors can be subjected to the capacity blending processing by adopting a fusion method matched with the similarity, so that the obtained fusion features are more accurate, and the reliability of the predictive maintenance information can be improved.
It should be understood that the fusion method may be, for example, a fusion function, or may also be a fusion algorithm, and the fusion method is not specifically limited in the embodiment of the present invention.
Specifically, the similarity between the new first feature vector and the new second feature vector is used to measure the difference between the two feature vectors if S (μ:)n,μm) The closer to 0, the more dissimilar the two eigenvectors are; if S (mu)n,μm) The closer to 1 or-1, the more two feature vectors are indicatedSimilarly.
Step 105: the fused features are input to a pre-constructed predictive model to output predictive maintenance information from the predictive model.
In this step, the predictive maintenance information includes at least one of the following information: the system comprises an operation state health value, a maintenance necessity degree value, a residual service life value and at least one maintenance time and a maintenance parameter corresponding to each maintenance time.
It is understood that, in this step, corresponding predictive maintenance information can be obtained according to the fusion features and the prediction model, and can be stored and displayed, so that relevant personnel can know the predictive maintenance information of the target device according to the displayed content, and accordingly, a maintenance plan can be made as required.
It should be understood that, in the embodiment of the present invention, the operation information of the target device may be obtained according to a preset period to perform predictive maintenance, and the current state of the target device may also be continuously monitored in real time, so that the state of the device may be known in real time according to the predictive maintenance information, thereby prolonging the maintenance period of the device to the maximum extent and reducing the life-span maintenance cost of the device.
In the embodiment of the present invention, after obtaining the operation information of the target device that needs to perform predictive maintenance, the predictive maintenance method outputs the first feature vector by using the pre-constructed knowledge graph, outputs the second feature vector by using the pre-constructed machine learning model, then performs fusion processing on the first feature vector and the second feature vector to obtain a fusion feature, and finally obtains the predictive maintenance information corresponding to the operation information based on the pre-constructed prediction model. Therefore, in the process of determining the predictive maintenance information based on the operation information of the target equipment, an engineer with abundant experience does not need to perform prediction judgment, and the organic combination of the knowledge graph and the machine learning model can avoid the condition of low accuracy of the prediction result caused by the subjectivity of manual predictive maintenance, so that the reliability of the predictive maintenance information is improved.
As shown in fig. 2 and 3, the embodiment of the present invention provides a predictive maintenance device. The predictive maintenance device embodiments may be implemented by software, or by hardware, or by a combination of hardware and software. From a hardware level, as shown in fig. 2, a hardware structure diagram of a device in which the predictive maintenance apparatus provided in the embodiment of the present invention is located is shown, in addition to the processor, the memory, the network interface, and the nonvolatile memory shown in fig. 2, the device in the embodiment may also generally include other hardware, such as a forwarding chip responsible for processing a packet, and the like. Taking a software implementation as an example, as shown in fig. 3, as a logical apparatus, the apparatus is formed by reading, by a CPU of a device in which the apparatus is located, corresponding computer program instructions in a non-volatile memory into a memory for execution.
As shown in fig. 3, the predictive maintenance apparatus provided in this embodiment includes: an acquisition module 301, a first input module 302, a fusion module 303 and a second input module 304;
an obtaining module 301, configured to obtain operation information of a target device that needs to perform predictive maintenance;
a first input module 302, configured to input the operation information acquired by the acquisition module 301 into a pre-constructed knowledge graph, so that a first feature vector is output by the knowledge graph; and
the system is used for inputting the running information acquired by the acquisition module 301 into a pre-constructed machine learning model so as to output a second feature vector by the machine learning model;
a fusion module 303, configured to perform fusion processing on the first feature vector and the second feature vector to obtain a fusion feature;
and a second input module 304, configured to input the fusion characteristics obtained by the fusion module 303 to a pre-constructed prediction model, so as to output the predictive maintenance information from the prediction model.
In an embodiment of the present invention, based on the predictive maintenance apparatus shown in fig. 3, the predictive maintenance apparatus further includes: a feature extraction module;
a feature extraction module, configured to perform the following processing on the analog signal in the operation information acquired by the acquisition module 301:
carrying out noise reduction processing on the analog signal to obtain a noise-reduced analog signal;
carrying out feature extraction on the denoised analog signal according to a plurality of preset feature dimensions to obtain feature information, wherein the feature dimensions comprise: at least one of time domain, frequency domain, and time-frequency domain;
the first input module 302 is further configured to input the feature information obtained by the feature extraction module as operation information to the knowledge graph, so that the knowledge graph outputs the first feature vector.
Further, in one embodiment of the present invention,
when the feature dimension includes a time domain, the feature information includes: at least one of a mean, variance, standard deviation, root mean square, crest factor, and form factor;
when the feature dimension includes a frequency domain, the feature information includes: at least one of a root mean square frequency, an average frequency, a center of gravity frequency, and a frequency standard deviation;
when the feature dimension includes a time-frequency domain, the feature information includes: wavelet energy characteristics.
In one embodiment of the present invention, the first and second electrodes are,
the fusion module 303 is specifically configured to perform the following processing:
acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure BDA0002863493870000131
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijThe preset weight value is used for representing the jth sub-vector in the ith feature vector, and m represents the number of sub-vectors in the ith feature vector;
respectively calculating a first feature vector and a second feature vector by using the weight value of each sub-vector;
normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain fusion features.
Further, in one embodiment of the present invention,
the fusion module 303 is specifically configured to perform the following processing:
calculating the similarity between the new first feature vector and the new second feature vector according to a second equation;
the second equation includes:
Figure BDA0002863493870000141
wherein, S (. mu.) isn,μm) For characterizing a feature vector munAnd a feature vector mumThe similarity of the vectors is represented by the similarity of the vectors, i is used for representing the ith vector in each feature vector, and q is used for representing the number of vectors in each feature vector;
acquiring a fusion method matched with the similarity from a preset mapping relation, wherein the mapping relation is used for representing the relations between different similarities and different fusion methods;
and carrying out fusion processing on the new first feature vector and the new second feature vector according to a fusion method to obtain fusion features.
In one embodiment of the present invention, the first and second electrodes are,
the predictive maintenance information includes at least one of: the maintenance system comprises an operation state health value, a maintenance necessity degree value, a residual service life value, at least one maintenance time and a maintenance parameter corresponding to each maintenance time;
in one embodiment of the present invention, the first and second electrodes are,
the knowledge graph is realized through a neural network and professional experience, wherein the professional experience is obtained through the operation information of a plurality of devices and the predictive maintenance results corresponding to the operation information of the plurality of devices.
The configuration illustrated in the embodiment of the present invention is not intended to specifically limit the predictive maintenance device. In other embodiments of the invention, the predictive maintenance device may include more or fewer components than shown, or some components may be combined, some components may be split, or a different arrangement of components. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.
Because the content of information interaction, execution process, and the like among the modules in the device is based on the same concept as the method embodiment of the present invention, specific content can be referred to the description in the method embodiment of the present invention, and is not described herein again.
An embodiment of the present invention further provides an intelligent device, including: at least one memory and at least one processor;
the at least one memory to store a machine readable program;
the at least one processor is configured to invoke the machine readable program to perform the predictive maintenance method in any embodiment of the invention.
Embodiments of the present invention also provide a computer-readable medium storing instructions for causing a computer to perform a predictive maintenance method as described herein. Specifically, a method or an apparatus equipped with a storage medium on which a software program code that realizes the functions of any of the above-described embodiments is stored may be provided, and a computer (or a CPU or MPU) of the method or the apparatus is caused to read out and execute the program code stored in the storage medium.
In this case, the program code itself read from the storage medium can realize the functions of any of the above-described embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
Examples of the storage medium for supplying the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (e.g., CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD + RW), a magnetic tape, a nonvolatile memory card, and a ROM. Alternatively, the program code may be downloaded from a server computer via a communications network.
Further, it should be clear that the functions of any one of the above-described embodiments may be implemented not only by executing the program code read out by the computer, but also by causing an operating system or the like operating on the computer to perform a part or all of the actual operations based on instructions of the program code.
Further, it is to be understood that the program code read out from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then causes a CPU or the like mounted on the expansion board or the expansion module to perform part or all of the actual operations based on instructions of the program code, thereby realizing the functions of any of the above-described embodiments.
It should be noted that not all steps and modules in the above flows and system structure diagrams are necessary, and some steps or modules may be omitted according to actual needs. The execution order of the steps is not fixed and can be adjusted as required. The system structure described in the above embodiments may be a physical structure or a logical structure, that is, some modules may be implemented by the same physical entity, or some modules may be implemented by a plurality of physical entities, or some components in a plurality of independent devices may be implemented together.
In the above embodiments, the hardware module may be implemented mechanically or electrically. For example, a hardware module may comprise permanently dedicated circuitry or logic (such as a dedicated processor, FPGA or ASIC) to perform the corresponding operations. A hardware module may also include programmable logic or circuitry (e.g., a general-purpose processor or other programmable processor) that may be temporarily configured by software to perform the corresponding operations. The specific implementation (mechanical, or dedicated permanent, or temporarily set) may be determined based on cost and time considerations.
While the invention has been shown and described in detail in the drawings and in the preferred embodiments, it is not intended to limit the invention to the embodiments disclosed, and it will be apparent to those skilled in the art that various combinations of the code auditing means in the various embodiments described above may be used to obtain further embodiments of the invention, which are also within the scope of the invention.

Claims (10)

1. A predictive maintenance method, comprising:
acquiring operation information of target equipment needing predictive maintenance;
inputting the operation information into a pre-constructed knowledge graph to output a first feature vector by the knowledge graph;
inputting the running information into a pre-constructed machine learning model to output a second feature vector by the machine learning model;
performing fusion processing on the first feature vector and the second feature vector to obtain fusion features;
inputting the fused features into a pre-constructed predictive model to output predictive maintenance information from the predictive model.
2. The method of claim 1, wherein when the operational information comprises an analog signal,
after the obtaining of the operation information of the target device requiring the predictive maintenance and before the inputting of the operation information into the pre-constructed knowledge graph to output the first feature vector by the knowledge graph, further comprising:
carrying out noise reduction processing on the analog signal to obtain a noise-reduced analog signal;
and performing feature extraction on the denoised analog signal according to a plurality of preset feature dimensions to obtain feature information, wherein the feature dimensions comprise: at least one of time domain, frequency domain, and time-frequency domain;
the inputting the operation information into a pre-constructed knowledge graph to output a first feature vector by the knowledge graph comprises:
inputting the feature information as the operation information to the knowledge-graph to output a first feature vector by the knowledge-graph.
3. The method of claim 2,
when the feature dimension includes the time domain, the feature information includes: at least one of a mean, variance, standard deviation, root mean square, crest factor, and form factor;
when the feature dimension includes the frequency domain, the feature information includes: at least one of a root mean square frequency, an average frequency, a center of gravity frequency, and a frequency standard deviation;
when the feature dimension includes the time-frequency domain, the feature information includes: wavelet energy characteristics.
4. The method according to claim 1, wherein the fusing the first feature vector and the second feature vector to obtain a fused feature comprises:
acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure FDA0002863493860000021
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijThe preset weight value is used for representing the jth sub-vector in the ith feature vector, and m represents the number of sub-vectors in the ith feature vector;
respectively calculating the first feature vector and the second feature vector by using the weight value of each sub-vector;
normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain the fusion features.
5. The method according to claim 4, wherein the fusing the new first feature vector and the new second feature vector to obtain the fused feature comprises:
calculating the similarity between the new first feature vector and the new second feature vector according to a second equation;
the second equation includes:
Figure FDA0002863493860000031
wherein, S (. mu.) isn,μm) For characterizing a feature vector munAnd a feature vector mumThe similarity of the vectors is represented by the similarity of the vectors, i is used for representing the ith vector in each feature vector, and q is used for representing the number of vectors in each feature vector;
acquiring a fusion method matched with the similarity from a preset mapping relation, wherein the mapping relation is used for representing the relations between different similarities and different fusion methods;
and carrying out fusion processing on the new first feature vector and the new second feature vector according to the fusion method to obtain the fusion features.
6. The method according to any one of claims 1 to 5,
the predictive maintenance information includes at least one of: the maintenance system comprises an operation state health value, a maintenance necessity degree value, a residual service life value, at least one maintenance time and a maintenance parameter corresponding to each maintenance time;
and/or the presence of a gas in the gas,
the knowledge graph is realized through a neural network and professional experience, wherein the professional experience is obtained through operation information of a plurality of devices and predictive maintenance results corresponding to the operation information of the plurality of devices.
7. A predictive maintenance device, comprising: the system comprises an acquisition module, a first input module, a fusion module and a second input module;
the acquisition module is used for acquiring the operation information of the target equipment needing predictive maintenance;
the first input module is used for inputting the operation information acquired by the acquisition module into a pre-constructed knowledge graph so as to output a first feature vector by the knowledge graph; and
the operation information acquired by the acquisition module is input into a pre-constructed machine learning model, so that a second feature vector is output by the machine learning model;
the fusion module is used for performing fusion processing on the first feature vector and the second feature vector to obtain fusion features;
and the second input module is used for inputting the fusion characteristics obtained by the fusion module into a pre-constructed prediction model so as to output predictive maintenance information by the prediction model.
8. The apparatus of claim 7,
the fusion module is specifically configured to perform the following processing:
acquiring a preset weight value corresponding to each component in the first feature vector and the second feature vector;
calculating to obtain a weight value corresponding to each component according to each preset weight value and the following first formula;
the first type includes:
Figure FDA0002863493860000041
wherein, KijWeight value, k, for characterizing the jth component vector in the ith feature vectorijThe preset weight value is used for representing the jth sub-vector in the ith feature vector, and m represents the number of sub-vectors in the ith feature vector;
respectively calculating the first feature vector and the second feature vector by using the weight value of each sub-vector;
normalizing the calculated first feature vector and the second feature vector to respectively obtain a new first feature vector and a new second feature vector;
and carrying out fusion processing on the new first feature vector and the new second feature vector to obtain the fusion features.
9. Smart device, characterized in that it comprises: at least one memory and at least one processor;
the at least one memory to store a machine readable program;
the at least one processor, configured to invoke the machine readable program to perform the predictive maintenance method of any of claims 1-6.
10. A computer readable medium having computer instructions stored thereon, which when executed by a processor, cause the processor to perform the predictive maintenance method of any of claims 1-6.
CN202011577019.4A 2020-12-28 2020-12-28 Predictive maintenance method and apparatus Active CN112700016B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011577019.4A CN112700016B (en) 2020-12-28 2020-12-28 Predictive maintenance method and apparatus

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011577019.4A CN112700016B (en) 2020-12-28 2020-12-28 Predictive maintenance method and apparatus

Publications (2)

Publication Number Publication Date
CN112700016A true CN112700016A (en) 2021-04-23
CN112700016B CN112700016B (en) 2022-08-02

Family

ID=75512415

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011577019.4A Active CN112700016B (en) 2020-12-28 2020-12-28 Predictive maintenance method and apparatus

Country Status (1)

Country Link
CN (1) CN112700016B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113792944A (en) * 2021-11-16 2021-12-14 深圳普菲特信息科技股份有限公司 Predictive maintenance method and system

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110223751A (en) * 2019-05-16 2019-09-10 平安科技（深圳）有限公司 Prescription evaluation method, system and computer equipment based on medical knowledge map
CN110263324A (en) * 2019-05-16 2019-09-20 华为技术有限公司 Text handling method, model training method and device
CN110727779A (en) * 2019-10-16 2020-01-24 信雅达系统工程股份有限公司 Question-answering method and system based on multi-model fusion
CN110737763A (en) * 2019-10-18 2020-01-31 成都华律网络服务有限公司 Chinese intelligent question-answering system and method integrating knowledge map and deep learning
CN111563606A (en) * 2020-04-30 2020-08-21 新智数字科技有限公司 Equipment predictive maintenance method and device
CN111984513A (en) * 2020-08-25 2020-11-24 浙江天垂科技有限公司 Predictive maintenance method, device, equipment and storage medium

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110223751A (en) * 2019-05-16 2019-09-10 平安科技（深圳）有限公司 Prescription evaluation method, system and computer equipment based on medical knowledge map
CN110263324A (en) * 2019-05-16 2019-09-20 华为技术有限公司 Text handling method, model training method and device
CN110727779A (en) * 2019-10-16 2020-01-24 信雅达系统工程股份有限公司 Question-answering method and system based on multi-model fusion
CN110737763A (en) * 2019-10-18 2020-01-31 成都华律网络服务有限公司 Chinese intelligent question-answering system and method integrating knowledge map and deep learning
CN111563606A (en) * 2020-04-30 2020-08-21 新智数字科技有限公司 Equipment predictive maintenance method and device
CN111984513A (en) * 2020-08-25 2020-11-24 浙江天垂科技有限公司 Predictive maintenance method, device, equipment and storage medium

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113792944A (en) * 2021-11-16 2021-12-14 深圳普菲特信息科技股份有限公司 Predictive maintenance method and system
CN113792944B (en) * 2021-11-16 2022-03-11 深圳普菲特信息科技股份有限公司 Predictive maintenance method and system

Also Published As

Publication number Publication date
CN112700016B (en) 2022-08-02

Similar Documents

Publication Publication Date Title
CN111504676B (en) Equipment fault diagnosis method, device and system based on multi-source monitoring data fusion
US10831577B2 (en) Abnormality detection system, abnormality detection method, abnormality detection program, and method for generating learned model
CN105488539B (en) The predictor method and device of the generation method and device of disaggregated model, power system capacity
CN113884961B (en) SOC calibration method, modeling device, computer equipment and medium
US7577548B1 (en) Integrated framework for diagnosis and prognosis of components
CN112326280B (en) Fault detection method and device for train bogie and computer readable medium
CN105675038A (en) Device for predicting faults of instruments
US20220004163A1 (en) Apparatus for predicting equipment damage
CN114580263A (en) Knowledge graph-based information system fault prediction method and related equipment
CN116453438B (en) Display screen parameter detection method, device, equipment and storage medium
CN116522153B (en) Lithium battery capacity prediction method, lithium battery capacity prediction device, computer equipment and storage medium
CN114548199A (en) Multi-sensor data fusion method based on deep migration network
CN112700016B (en) Predictive maintenance method and apparatus
JP2020101948A (en) Action recognition system and action recognition method
CN116257663A (en) Abnormality detection and association analysis method and related equipment for unmanned ground vehicle
CN113487223B (en) Risk assessment method and system based on information fusion
CN112308148A (en) Defect category identification and twin neural network training method, device and storage medium
CN111338876A (en) Fault mode and influence analysis decision method, system and storage medium
CN117036732B (en) Electromechanical equipment detection system, method and equipment based on fusion model
CN110490132B (en) Data processing method and device
CN117170915A (en) Data center equipment fault prediction method and device and computer equipment
KR20210091591A (en) An electronic device including evaluation operation of originated technology
US20200349495A1 (en) Analytical model training method for customer experience estimation
CN114416686A (en) Vehicle equipment fingerprint CARID identification system and identification method
CN114169398A (en) Photovoltaic direct-current arc fault identification method and device based on random forest algorithm

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20220713

Address after: 250100 building S02, No. 1036, Langchao Road, high tech Zone, Jinan City, Shandong Province

Applicant after: Shandong Inspur Scientific Research Institute Co.,Ltd.

Address before: 250100 First Floor of R&D Building 2877 Kehang Road, Sun Village Town, Jinan High-tech Zone, Shandong Province

Applicant before: JINAN INSPUR HIGH-TECH TECHNOLOGY DEVELOPMENT Co.,Ltd.

GR01 Patent grant
GR01 Patent grant