CN117892114B - Crane fault prediction method and system based on Internet of Things - Google Patents
Crane fault prediction method and system based on Internet of Things Download PDFInfo
- Publication number
- CN117892114B CN117892114B CN202410288075.8A CN202410288075A CN117892114B CN 117892114 B CN117892114 B CN 117892114B CN 202410288075 A CN202410288075 A CN 202410288075A CN 117892114 B CN117892114 B CN 117892114B
- Authority
- CN
- China
- Prior art keywords
- component
- fault
- twin
- matrix
- abnormal
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000004458 analytical method Methods 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 102
- 230000002159 abnormal effect Effects 0.000 claims description 95
- 230000015654 memory Effects 0.000 claims description 34
- 238000010276 construction Methods 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 13
- 238000001514 detection method Methods 0.000 claims description 11
- 230000002452 interceptive effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000004806 packaging method and process Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 7
- 230000000694 effects Effects 0.000 abstract description 5
- 230000005856 abnormality Effects 0.000 description 6
- 238000007689 inspection Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 239000013598 vector Substances 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012536 packaging technology Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a crane fault prediction method and a crane fault prediction system based on the Internet of things, which relate to the technical field of data processing. The technical problems that in the prior art, the crane performs fault early warning based on single component fault monitoring, and the fault false alarm or the early warning is inaccurate due to lack of comprehensive analysis are solved. The technical effect of accurately judging the crane faults based on comprehensive analysis and avoiding early warning redundant errors is achieved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a crane fault prediction method and system based on the Internet of things.
Background
In crane equipment, the early warning system has the problem of redundancy error of fault early warning. When a certain component of the crane breaks down, the current early warning system can give an alarm instantly, but sometimes the fault can be caused by other components working cooperatively with the crane, which can lead to the occurrence of false alarms or inaccurate early warning of the early warning system.
In this case, the early warning system cannot determine the root cause of the fault and it is difficult to make an accurate alarm. Such redundancy errors can lead to damage and unnecessary maintenance costs of the crane and can cause confusion and danger to personnel using the crane.
In summary, in the prior art, the crane performs fault early warning based on single component fault monitoring, and the technical problems of fault false alarm or inaccurate early warning are caused due to lack of comprehensive analysis.
Disclosure of Invention
The application provides a crane fault prediction method and system based on the Internet of things, which are used for solving the technical problems that in the prior art, a crane performs fault early warning based on single component fault monitoring, and the fault is misreported or the early warning is inaccurate due to lack of comprehensive analysis.
In view of the problems, the application provides a crane fault prediction method and a crane fault prediction system based on the Internet of things.
According to a first aspect of the application, a crane fault prediction method based on the Internet of things is provided, and the method comprises the following steps: collecting operation data sets respectively corresponding to all parts of the target crane during operation; extracting features from the operation data set of each component to obtain a feature state real-time matrix of each component, wherein the feature state real-time matrix is an N multiplied by M matrix of M real-time variable features recorded at t_j time sequence nodes, j= {0, 1..N }, and N is the number of the time sequence nodes; constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is an N multiplied by M matrix of M sample variable characteristics recorded at t_j time sequence nodes, j= {0, 1..N }, and N is the number of the time sequence nodes; analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a plurality of fault characteristics corresponding to the abnormal component; analyzing the operation connection relation among the abnormal parts to obtain the connection relation of the operation of the abnormal parts; and establishing Siam Net twin recognition networks of the plurality of fault characteristics based on the connection relation of the abnormal component operation, and predicting according to twin recognition results corresponding to the Siam Net twin recognition networks to obtain a fault prediction result.
In a second aspect of the present application, there is provided a crane fault prediction system based on the internet of things, the system comprising: the operation data acquisition unit is used for acquiring operation data sets corresponding to all parts of the target crane respectively when the target crane operates; the feature extraction execution unit is used for carrying out feature extraction on the operation data set of each component to obtain a feature state real-time matrix of each component, wherein the feature state real-time matrix is an N multiplied by M matrix of M real-time variable features recorded at t_j time sequence nodes, j= {0, 1..N }, and N is the number of the time sequence nodes; a memory matrix construction unit, configured to construct a feature state memory matrix of each component in the target crane according to a running sample data set of the target crane, where the feature state memory matrix is an nxm matrix of M sample variable features recorded at t_j time sequence nodes, j= {0, 1..n }, where N is the number of time sequence nodes; the fault characteristic acquisition unit is used for analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component and acquiring a plurality of fault characteristics corresponding to the abnormal component; the connection relation analysis unit is used for analyzing the operation connection relation among the abnormal parts and acquiring the connection relation of the operation of the abnormal parts; the identification network construction unit is used for establishing Siam Net twin identification networks of the plurality of fault characteristics based on the connection relation of the abnormal component operation, and predicting according to twin identification results corresponding to the Siam Net twin identification networks to obtain a fault prediction result.
One or more technical schemes provided by the application have at least the following technical effects or advantages:
The method provided by the embodiment of the application is characterized in that the operation data sets respectively corresponding to all the components of the target crane are collected when the target crane works; extracting features from the operation data set of each component to obtain a feature state real-time matrix of each component, wherein the feature state real-time matrix is an N multiplied by M matrix of M real-time variable features recorded at t_j time sequence nodes, j= {0, 1..N }, and N is the number of the time sequence nodes; constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is an N multiplied by M matrix of M sample variable characteristics recorded at t_j time sequence nodes, j= {0, 1..N }, and N is the number of the time sequence nodes; analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a plurality of fault characteristics corresponding to the abnormal component; analyzing the operation connection relation among the abnormal parts to obtain the connection relation of the operation of the abnormal parts; and establishing Siam Net twin recognition networks of the plurality of fault characteristics based on the connection relation of the abnormal component operation, and predicting according to twin recognition results corresponding to the Siam Net twin recognition networks to obtain a fault prediction result. The crane fault accurate judgment based on comprehensive analysis is achieved, the early warning redundancy error is avoided, and the crane early warning accuracy and reliability are improved.
Drawings
Fig. 1 is a flow chart of a crane fault prediction method based on the internet of things.
Fig. 2 is a schematic flow chart of acquiring part fault characteristics in the crane fault prediction method based on the internet of things.
Fig. 3 is a schematic structural diagram of a crane fault prediction system based on the internet of things.
Reference numerals illustrate: the system comprises an operation data acquisition unit 1, a feature extraction execution unit 2, a memory matrix construction unit 3, a fault feature acquisition unit 4, a connection relation analysis unit 5 and an identification network construction unit 6.
Detailed Description
The application provides a crane fault prediction method and system based on the Internet of things, which are used for solving the technical problems that in the prior art, a crane performs fault early warning based on single component fault monitoring, and the fault is misreported or the early warning is inaccurate due to lack of comprehensive analysis. The crane fault accurate judgment based on comprehensive analysis is achieved, the early warning redundancy error is avoided, and the crane early warning accuracy and reliability are improved.
The technical scheme of the invention accords with related regulations on data acquisition, storage, use, processing and the like.
In the following, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings, and it should be understood that the described embodiments are only some embodiments of the present invention, but not all embodiments of the present invention, and that the present invention is not limited by the exemplary embodiments described herein. 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. It should be further noted that, for convenience of description, only some, but not all of the drawings related to the present invention are shown.
Example 1
As shown in fig. 1, the application provides a crane fault prediction method based on the internet of things, which comprises the following steps:
a100, collecting operation data sets respectively corresponding to all parts of the target crane when the target crane works;
In one embodiment, after collecting the operation data sets corresponding to each component of the target crane during operation, the method step a100 provided by the present application further includes:
A110, packaging a denoising processing module by using wavelet transform;
And A120, inputting the operation data sets corresponding to the components into the denoising processing module to perform denoising processing to obtain the operation data set processed by the components, and performing feature extraction on the operation data set processed by the components.
Specifically, in this embodiment, the target crane is an engineering mechanical device for lifting and carrying heavy objects, and the components of the target crane at least comprise a supporting leg, a crane arm, a crane hook, a driving system and a control system.
According to the embodiment, operation data corresponding to each component of the target crane during operation are collected, and the obtained operation data set of each component is used for illustrating that the operation data set of the lifting hook component at least comprises hook load data, lifting speed data and hook position data, and the operation data set of the lifting arm component at least comprises operation arm length data, lifting height data and inclination angle data.
It should be appreciated that the operational data set obtained by direct acquisition based on operational monitoring of the present embodiment often contains various noise and interference that can have a significant impact on the quality and accuracy of the data in the operational data set.
Based on this, in this embodiment, a denoising processing module is pre-built by using a wavelet transformation packaging technology to perform data denoising processing on the operation data set, where the construction process of the denoising processing module is as follows, and a suitable wavelet basis function is selected according to the actual operation condition and the fault recognition precision requirement of the target crane, and the specific selection of the wavelet basis function is not limited in this embodiment, so that the wavelet basis function and the denoising algorithm are packaged together to form a reusable denoising processing module.
And inputting the operation data sets corresponding to the components into the denoising processing module to perform denoising processing to obtain operation data sets processed by the components, and performing feature extraction on the operation data sets processed.
According to the embodiment, the denoising pretreatment of the operation data set is carried out by adopting the wavelet transformation packaged denoising treatment module, so that the quality and accuracy of the operation data are improved, and a more reliable basis is provided for subsequent operation data analysis and fault decision.
A200, extracting features of the operation data set of each component to obtain a feature state real-time matrix of each component, wherein the feature state real-time matrix is in the presence of/>, Recorded by each time sequence node, on M real-time variable characteristicsMatrix of/>，/>Is the number of timing nodes;
A300, constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is in the following state />, Recorded by each time sequence node, on M sample variable characteristicsMatrix of/>，/>Is the number of timing nodes;
Specifically, in the present embodiment, each component of the target crane has M operation monitoring indexes, and the operation data set of each corresponding component includes M sets of real-time variable data corresponding to the M operation monitoring indexes.
Based on the above, N time sequence nodes are preset, the interval time between adjacent time sequence nodes is consistent, and feature extraction is performed on the M groups of real-time variable data of each component based on the N time sequence nodes, so as to obtain a feature state real-time matrix of each component, where the feature state real-time matrix is in the presence of/>, Recorded by each time sequence node, on M real-time variable characteristicsIs a matrix of the (c) in the matrix,，/>Is the number of timing nodes.
And interacting the operation log of the target crane, and calling and obtaining the operation sample data set of the target crane operated in a normal state without any faults, wherein the operation sample data set comprises part operation sample data sets of all parts.
The same method for extracting the characteristics of the operation data sets of all the components to obtain the characteristic state real-time matrix of each component is adopted, the characteristic extraction is carried out on the component operation sample data sets of all the components, and the characteristic state memory matrix of each component in the target crane is constructed, wherein the characteristic state memory matrix is that/>, Recorded by each time sequence node, on M sample variable characteristicsMatrix of/>，/>Is the number of timing nodes.
A400, analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a plurality of fault characteristics corresponding to the abnormal component;
in one embodiment, as shown in fig. 2, the feature state real-time matrix and the feature state memory matrix of each component are analyzed to obtain a plurality of fault features corresponding to each component, and the method step a400 provided by the present application further includes:
a410, analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a characteristic state deviation matrix, wherein the characteristic state deviation matrix is the deviation degree of M real-time variable characteristics;
a420, obtaining characteristic deviation indexes of all the components by carrying out deviation calculation on the characteristic state deviation matrix;
And A430, screening the components with the deviation indexes greater than or equal to the preset deviation indexes from the characteristic deviation indexes of the components to be identified as abnormal components.
Specifically, in this embodiment, the real-time matrix of the characteristic state and the memory matrix of the characteristic state of each component are overlapped according to the same structure, so as to ensure that the number of rows and the number of columns of the two matrices are equal, and the data of the corresponding positions are aligned.
And starting from the first element of the overlapped matrix, comparing the data of the corresponding positions in the two matrices one by one, calculating the deviation percentage, and replacing the two data of the corresponding element positions in the overlapped matrix by adopting the deviation percentage to obtain the characteristic state deviation matrix of each component, wherein the characteristic state deviation matrix is understood to be the deviation degree of M real-time variable characteristics.
And based on the matrix form of the characteristic state deviation matrix, performing matrix solving calculation by adopting a solving linear equation set or a matrix equation to obtain characteristic deviation indexes of all the components, wherein the characteristic deviation indexes integrally reflect the degree of deviation of the operating parameters of the components from normal values.
And setting a plurality of preset deviation indexes of each component in abnormal operation conditions based on the judgment history experience of the abnormal operation of the component and the calculated numerical conditions of the current characteristic deviation indexes.
Traversing the corresponding preset deviation indexes by adopting the characteristic deviation indexes of each component so as to screen one or more components which are larger than or equal to the preset deviation indexes and mark the components as abnormal components. The embodiment achieves the technical effect of accurately analyzing and judging the fault condition of the component based on matrix analysis.
A500, analyzing the operation connection relation among the abnormal parts to obtain the connection relation of the operation of the abnormal parts;
Specifically, in the present embodiment, the job connection relationship between each component and other components in the target crane is obtained interactively, where the job connection relationship includes a functional relationship in which there is coordination and cooperation between one component and other components to complete a specific task.
And specifically, obtaining a plurality of groups of operation connection relations of the plurality of components in a total way, taking each component as a topological node of the topological structure, and connecting the topological nodes by referring to the plurality of groups of operation connection relations to obtain the component topological structure of the target crane.
When the number of the abnormal parts obtained in the step a400 is multiple (two or more), positioning the abnormal parts in the part topology, deleting the topology nodes corresponding to the non-abnormal parts, and obtaining the abnormal part topology.
The connection relation of the abnormal part operation, which characterizes the functional relation (connection relation) between each abnormal part and other abnormal parts, can be intuitively known based on the abnormal part topology structure. The two abnormal parts are respectively a big arm and a small arm, and the two abnormal parts are in operation connection based on the characteristic that the big arm and the small arm realize lifting and moving of a lifting object through cooperative movement.
A600, establishing Siam Net twin recognition networks of the plurality of fault characteristics based on the connection relation of the operation of the abnormal components, and predicting according to twin recognition results corresponding to the Siam Net twin recognition networks to obtain a fault prediction result.
In one embodiment, the Siam Net twin identification network for the plurality of fault signatures is established, and the method step a600 provided by the present application further includes:
A610, analyzing the connection relation of the abnormal part operation to obtain k twin part groups, wherein each twin part group in the k twin part groups comprises two parts, and the two parts are the connection relation of the interactive operation;
And A620, calling and inputting fault characteristics corresponding to the corresponding twin component groups according to the k twin component groups to obtain k Siam Net twin identification networks, wherein each Siam Net twin identification network is used for carrying out fault characteristic similarity accompanying detection.
In one embodiment, the predicting is performed according to a twinning recognition result corresponding to the Siam Net twinning recognition network, and the calculating step a600 of the twinning recognition result further includes:
a630, carrying out the concomitant detection of the fault characteristic similarity on the k twin component groups by utilizing the k Siam Net twin recognition networks to obtain k concomitant check operators;
a640, judging the k accompanying check operators, identifying the twin component group which is in a preset check threshold, outputting the identified twin component group, and outputting the identified twin component group as the twin recognition result.
In one embodiment, the predicting is performed according to the twinning recognition result corresponding to the Siam Net twinning recognition network, so as to obtain a fault prediction result, and the method step a600 provided by the present application further includes:
a650, obtaining the twin recognition result, wherein the twin recognition result comprises identification of a twin component group;
A660, combining the identification twin component groups to generate fault prediction groups, wherein each fault prediction group comprises two identification twin component groups with coincident components;
A670, performing prediction traversal according to the fault prediction group to obtain a fault prediction result for identifying each component state, wherein the component states comprise fault states or fault accompanying states;
and A680, positioning the fault component of the target crane according to the fault prediction result of the state of each component.
It will be appreciated that there is typically an interaction between two components in operative connection, and if one component is in a fault condition it may result in a false faulty operating condition of the other component not in the fault condition. Specifically, assuming that there are two components a and B in operative connection with each other, i.e., the operation states of the component a and the component B are affected with each other, if the component a is in a failure state, it may not provide a correct output or feedback to the component B, resulting in that the component B may not work properly, in which case, although the component B itself does not fail, it may not work properly due to the failure of the component a, resulting in a false failure operation state.
Based on this, the present embodiment, in combination with the abnormal component topology, can intuitively learn the characteristics of the connection relationship of the abnormal component job that characterizes the functional relationship (connection relationship) between each abnormal component and other abnormal components.
And taking two abnormal parts with operation connection relations (connection relations of interaction operations) as a twin part group, and analyzing the connection relations of the operation of the abnormal parts to obtain k twin part groups.
And after k twin component groups are determined based on the job connection relationship between the abnormal components, further verifying to determine whether the abnormality of both components in the twin component groups is a true abnormality or one is a true abnormality and one is a false fault abnormality (concomitant false abnormality).
The specific component abnormality verification implementation method is as follows:
k Siam Net twin recognition networks for performing fault feature similarity accompanying detection are pre-constructed, and based on the similarity of the construction processes of the k Siam Net twin recognition networks, the construction process of the first Siam Net twin recognition network for performing fault feature similarity accompanying detection on the first twin component group is taken as an example, and a Siam Net twin recognition network construction method is elaborated.
Specifically, the Siam Net twin recognition network is composed of two sub-networks with the same structure, and is used for processing the fault characteristics of the first association component and the second association component of the input first twin component group respectively and carrying out the accompanying detection of the similarity based on the fault characteristics. The two sub-networks share parameters in order to learn the similarity between the twinning components. Typically the network structure of the subnetwork in the Siam Net twin recognition network comprises a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN), etc.
And in the synchronous use process of the first association component and the second association component in the first twin component group, fault characteristic data of the first association component in an abnormal working state and the second association component in a normal working state are obtained interactively, fault characteristic data of the second association component in the abnormal working state and the first association component in the normal working state are obtained interactively, and fault characteristic data of the first association component and the second association component in the abnormal working state are obtained interactively.
In this embodiment, three sets of fault feature data, such as the above three types, are used as training data, and a conventional output precision optimization training method for performing Convolutional Neural Network (CNN) or cyclic neural network (RNN) is used to train the first Siam Net twin recognition network, and in the training process, model parameters are optimized by minimizing a loss function, so that the output feature can accurately represent the fault similarity between twin components.
The present embodiment obtains a plurality of fault characteristics of a plurality of abnormal parts in step a400, and obtains a first fault characteristic and a second fault characteristic of a first associated part and a second associated part in a first twin part group in a current step mapping call.
And respectively inputting the first fault feature and the second fault feature into a first sub-network and a second sub-network of the first Siam Net twin recognition network to extract feature vectors to obtain a first feature vector and a second feature vector, further calculating the similarity of the first feature vector and the second feature vector based on Euclidean distance in the first Siam Net twin recognition network, and outputting the similarity as a first accompanying check operator to finish the accompanying detection of the fault feature similarity of the first twin component group based on the first Siam Net twin recognition network.
It will be appreciated that the greater the value of the first accompanying check operator, the greater the similarity characterizing the first and second fault signatures, i.e., the greater the likelihood that one of the first and second associated components is in a false fault condition.
In the embodiment, the same method of constructing the first Siam Net twin recognition network is adopted, k Siam Net twin recognition networks are constructed and obtained, and k companion inspection operators are obtained by performing companion detection of fault feature similarity on the k twin component groups through the k Siam Net twin recognition networks.
In this embodiment, an inspection threshold value (unspecific similarity data) for judging that one associated component is necessarily in a twinning component group in a false fault state is preset, the k accompanying inspection operators are judged based on the inspection threshold value, a twinning component group (a twinning component group that one associated component is necessarily in a false fault state) in the preset inspection threshold value is identified, a plurality of identification twinning component groups are output, and a plurality of identification twinning component groups are output as the twinning recognition result.
Combining the identified twinned component groups based on component coincidence to generate failure prediction groups, each failure prediction group comprising two identified twinned component groups with coincident components, wherein, illustratively, the combination of the two identified twinned component groups of a certain failure prediction group is component A and component B, component B and component C respectively, then component B is a coincident component.
It should be appreciated that within a failure prediction set, if component B is a failed component and component a and component C are normal components that are accompanied by an affected failure, then there is similarity in the failure characteristics of component a and component C.
Based on this, the embodiment adopts the same method of constructing the first Siam Net twin recognition network, constructs Siam Net twin recognition networks of the component a and the component C, inputs fault characteristics of the component a and the component C, obtains an a-C companion check operator, and determines that the component B is in a fault state as a fault prediction result and that the fault prediction results of the component a and the component C are in a fault companion state when the a-C companion check operator satisfies a preset check threshold set in the early stage.
It should be appreciated that there is also a component overlap between a plurality of the failure prediction groups, and thus the plurality of abnormal components obtained in step a400 are all subjected to failure state prediction as overlapping components within the failure prediction groups.
Therefore, the embodiment performs prediction traversal on a plurality of the failure prediction groups based on the same failure prediction method of the component B to obtain failure prediction results for identifying the states of the components, and positions the failed component of the target crane according to the failure prediction results for identifying the states of the components.
According to the embodiment, accurate component maintenance management is performed on the basis of a plurality of fault components obtained through positioning, so that the technical effects of accurately judging crane faults based on comprehensive analysis, avoiding early warning redundancy errors and improving crane early warning accuracy and reliability are achieved.
In one embodiment, the predicting is performed according to the twinning recognition result corresponding to the Siam Net twinning recognition network, so as to obtain a fault prediction result, and the method step a610 provided by the present application further includes:
A611, analyzing the connection relation of the abnormal component operation, and judging whether the abnormal component comprises an independent abnormal component, wherein the independent abnormal component is a component without an interactive connection operation relation;
a612, if the abnormal component comprises an independent abnormal component, carrying out independent prediction according to the fault characteristics corresponding to the independent abnormal component to obtain an independent fault prediction result;
and A613, adding the independent fault prediction result to the fault prediction result.
Specifically, in this embodiment, based on the topology of the abnormal component, the characteristics of the connection relationship of the operation of the abnormal component, which characterizes the functional relationship (connection relationship) between each abnormal component and other abnormal components, may be intuitively known, an individual topology node is directly located and obtained in the topology of the abnormal component, and a component corresponding to the individual topology node is used as an independent abnormal component, where the independent abnormal component is a component without an interactive connection operation relationship.
If the abnormal component comprises an independent abnormal component, directly marking the fault prediction result of the independent abnormal component as a fault state, synchronously storing the fault characteristic corresponding to the independent abnormal component and the fault state as the independent fault prediction result of the corresponding abnormal component, and adding the independent fault prediction result into the fault prediction result.
According to the method and the device, the operation connection relation between the components which are judged to be abnormal is intuitively and accurately known through constructing the topological structure of the abnormal components, so that whether the components are in a fault state or not is accurately predicted from two dimensions of component single body and component connection, the technical effects of avoiding early warning redundancy errors and improving early warning accuracy and reliability of a crane are achieved.
Example two
Based on the same inventive concept as the crane fault prediction method based on the internet of things in the foregoing embodiment, as shown in fig. 3, the present application provides a crane fault prediction system based on the internet of things, where the system includes:
The operation data acquisition unit 1 is used for acquiring operation data sets corresponding to all parts of the target crane respectively when the target crane works;
a feature extraction execution unit 2, configured to perform feature extraction on an operation data set of each component to obtain a feature state real-time matrix of each component, where the feature state real-time matrix is in the following state />, Recorded by each time sequence node, on M real-time variable characteristicsMatrix of/>，/>Is the number of timing nodes;
a memory matrix construction unit 3 for constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is that />, Recorded by each time sequence node, on M sample variable characteristicsMatrix of/>，/>Is the number of timing nodes;
A fault feature obtaining unit 4, configured to analyze the feature state real-time matrix and the feature state memory matrix of each component, and obtain a plurality of fault features corresponding to the abnormal component;
a connection relation analysis unit 5 for analyzing the operation connection relation between the abnormal parts and obtaining the connection relation of the operation of the abnormal parts;
The identifying network constructing unit 6 is configured to establish Siam Net twin identifying networks of the plurality of fault features based on the connection relationship of the abnormal component operation, and predict according to twin identifying results corresponding to the Siam Net twin identifying networks, so as to obtain a fault predicting result.
In one embodiment, the fault signature acquisition unit 4 further comprises:
Analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a characteristic state deviation matrix, wherein the characteristic state deviation matrix is the deviation degree of M real-time variable characteristics;
Obtaining characteristic deviation indexes of all the components by carrying out deviation calculation on the characteristic state deviation matrix;
And screening the components with the deviation index greater than or equal to the preset deviation index from the characteristic deviation indexes of each component, and marking the components as abnormal components.
In one embodiment, the identification network construction unit 6 further comprises:
analyzing the connection relation of the abnormal part operation to obtain k twin part groups, wherein each twin part group in the k twin part groups comprises two parts, and the two parts are the connection relation of the interactive operation;
And calling and inputting fault characteristics corresponding to the corresponding twin component groups according to the k twin component groups to obtain k Siam Net twin identification networks, wherein each Siam Net twin identification network is used for carrying out fault characteristic similarity accompanying detection.
In one embodiment, the identification network construction unit 6 further comprises:
Carrying out the concomitant detection of the fault feature similarity on the k twin component groups by using the k Siam Net twin recognition networks to obtain k concomitant check operators;
Judging the k accompanying check operators, identifying the twin component group at a preset check threshold, outputting an identified twin component group, and outputting the identified twin component group as the twin recognition result.
In one embodiment, the identification network construction unit 6 further comprises:
Obtaining the twin recognition result, wherein the twin recognition result comprises identification of a twin component group;
combining according to the identification twin component groups to generate fault prediction groups, wherein each fault prediction group comprises two identification twin component groups with coincident components;
Performing prediction traversal according to the fault prediction group to obtain a fault prediction result for identifying each component state, wherein the component states comprise fault states or fault accompanying states;
to identify a failure prediction of the status of each component, and to locate a failed component of the target crane.
In one embodiment, the identification network construction unit 6 further comprises:
analyzing the connection relation of the abnormal part operation, and judging whether the abnormal part comprises an independent abnormal part or not, wherein the independent abnormal part is a part without an interactive connection operation relation;
if the abnormal component comprises an independent abnormal component, carrying out independent prediction according to the fault characteristics corresponding to the independent abnormal component to obtain an independent fault prediction result;
And adding the independent fault prediction result to the fault prediction result.
In one embodiment, the operation data acquisition unit 1 further comprises:
Packaging a denoising processing module by using wavelet transformation;
and inputting the operation data sets corresponding to the components into the denoising processing module to perform denoising processing to obtain operation data sets processed by the components, and performing feature extraction on the operation data sets processed.
Any of the methods or steps described above may be stored as computer instructions or programs in various non-limiting types of computer memories, and identified by various non-limiting types of computer processors, thereby implementing any of the methods or steps described above.
Based on the above-mentioned embodiments of the present invention, any improvements and modifications to the present invention without departing from the principles of the present invention should fall within the scope of the present invention.
Claims (7)
1. The crane fault prediction method based on the Internet of things is characterized by comprising the following steps of:
Collecting operation data sets respectively corresponding to all parts of the target crane during operation;
Extracting features from the operation data set of each component to obtain a real-time matrix of the feature state of each component, wherein the real-time matrix of the feature state is that />, Recorded by each time sequence node, on M real-time variable characteristicsIs a matrix of the (c) in the matrix,，/>Is the number of timing nodes;
Constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is in the following state M sample variable features recorded by time sequence nodesMatrix of/>，/>Is the number of timing nodes;
analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a plurality of fault characteristics corresponding to the abnormal component;
Analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a characteristic state deviation matrix, wherein the characteristic state deviation matrix is the deviation degree of M real-time variable characteristics;
Obtaining characteristic deviation indexes of all the components by carrying out deviation calculation on the characteristic state deviation matrix;
screening the components with the deviation index greater than or equal to the preset deviation index from the characteristic deviation index of each component to be used as abnormal components;
Analyzing the operation connection relation among the abnormal parts to obtain the connection relation of the operation of the abnormal parts;
And establishing Siam Net twin recognition networks of the plurality of fault characteristics based on the connection relation of the abnormal component operation, and predicting according to twin recognition results corresponding to the Siam Net twin recognition networks to obtain a fault prediction result.
2. The method of claim 1, wherein establishing a Siam Net twinning identification network of the plurality of fault signatures comprises:
analyzing the connection relation of the abnormal part operation to obtain k twin part groups, wherein each twin part group in the k twin part groups comprises two parts, and the two parts are the connection relation of the interactive operation;
And calling and inputting fault characteristics corresponding to the corresponding twin component groups according to the k twin component groups to obtain k Siam Net twin identification networks, wherein each Siam Net twin identification network is used for carrying out fault characteristic similarity accompanying detection.
3. The method of claim 2, wherein predicting is based on a twinning recognition result corresponding to the Siam Net twinning recognition network, the calculating of the twinning recognition result comprising:
Carrying out the concomitant detection of the fault feature similarity on the k twin component groups by using the k Siam Net twin recognition networks to obtain k concomitant check operators;
Judging the k accompanying check operators, identifying the twin component group at a preset check threshold, outputting an identified twin component group, and outputting the identified twin component group as the twin recognition result.
4. The method of claim 3, wherein predicting according to the twinning identification result corresponding to the Siam Net twinning identification network, obtaining a fault prediction result comprises:
Obtaining the twin recognition result, wherein the twin recognition result comprises identification of a twin component group;
combining according to the identification twin component groups to generate fault prediction groups, wherein each fault prediction group comprises two identification twin component groups with coincident components;
Performing prediction traversal according to the fault prediction group to obtain a fault prediction result for identifying each component state, wherein the component states comprise fault states or fault accompanying states;
to identify a failure prediction of the status of each component, and to locate a failed component of the target crane.
5. The method of claim 3, wherein predicting according to the twinning identification result corresponding to the Siam Net twinning identification network, obtaining a fault prediction result, further comprises:
analyzing the connection relation of the abnormal part operation, and judging whether the abnormal part comprises an independent abnormal part or not, wherein the independent abnormal part is a part without an interactive connection operation relation;
if the abnormal component comprises an independent abnormal component, carrying out independent prediction according to the fault characteristics corresponding to the independent abnormal component to obtain an independent fault prediction result;
And adding the independent fault prediction result to the fault prediction result.
6. The method of claim 1, further comprising, after collecting the operation data sets corresponding to the respective components of the target crane when performing the operation:
Packaging a denoising processing module by using wavelet transformation;
and inputting the operation data sets corresponding to the components into the denoising processing module to perform denoising processing to obtain operation data sets processed by the components, and performing feature extraction on the operation data sets processed.
7. Crane fault prediction system based on the internet of things, which is characterized in that the system comprises:
The operation data acquisition unit is used for acquiring operation data sets corresponding to all parts of the target crane respectively when the target crane operates;
the feature extraction execution unit is used for carrying out feature extraction on the operation data set of each component to obtain a feature state real-time matrix of each component, wherein the feature state real-time matrix is that />, Recorded by each time sequence node, on M real-time variable characteristicsMatrix of/>，/>Is the number of timing nodes;
A memory matrix construction unit for constructing a characteristic state memory matrix of each component in the target crane according to the operation sample data set of the target crane, wherein the characteristic state memory matrix is in the following state />, Recorded by each time sequence node, on M sample variable characteristicsMatrix of/>，/>Is the number of timing nodes;
The fault characteristic acquisition unit is used for analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component and acquiring a plurality of fault characteristics corresponding to the abnormal component;
The connection relation analysis unit is used for analyzing the operation connection relation among the abnormal parts and acquiring the connection relation of the operation of the abnormal parts;
The identification network construction unit is used for establishing Siam Net twin identification networks of the plurality of fault characteristics based on the connection relation of the abnormal component operation, and predicting according to twin identification results corresponding to the Siam Net twin identification networks to obtain a fault prediction result;
The failure feature acquisition unit further includes:
Analyzing the characteristic state real-time matrix and the characteristic state memory matrix of each component to obtain a characteristic state deviation matrix, wherein the characteristic state deviation matrix is the deviation degree of M real-time variable characteristics; obtaining characteristic deviation indexes of all the components by carrying out deviation calculation on the characteristic state deviation matrix; and screening the components with the deviation index greater than or equal to the preset deviation index from the characteristic deviation indexes of each component, and marking the components as abnormal components.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410288075.8A CN117892114B (en) | 2024-03-14 | 2024-03-14 | Crane fault prediction method and system based on Internet of Things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410288075.8A CN117892114B (en) | 2024-03-14 | 2024-03-14 | Crane fault prediction method and system based on Internet of Things |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117892114A CN117892114A (en) | 2024-04-16 |
| CN117892114B true CN117892114B (en) | 2024-05-24 |
Family
ID=90649096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410288075.8A Active CN117892114B (en) | 2024-03-14 | 2024-03-14 | Crane fault prediction method and system based on Internet of Things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117892114B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114399066A (en) * | 2022-01-15 | 2022-04-26 | 中国矿业大学(北京) | Mechanical equipment predictability maintenance system and maintenance method based on weak supervision learning |
| KR20230036776A (en) * | 2021-09-08 | 2023-03-15 | 울산대학교 산학협력단 | System and method for fault diagnosis of fuel cell energy management system based on digital twin |
| CN117189713A (en) * | 2023-07-27 | 2023-12-08 | 中国人民解放军海军潜艇学院 | Hydraulic system fault diagnosis method based on digital twin driving |
| CN117490760A (en) * | 2023-10-24 | 2024-02-02 | 山东省科学院自动化研究所 | Calender monitoring and fault early warning method and system based on digital twin technology |
-
2024
- 2024-03-14 CN CN202410288075.8A patent/CN117892114B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20230036776A (en) * | 2021-09-08 | 2023-03-15 | 울산대학교 산학협력단 | System and method for fault diagnosis of fuel cell energy management system based on digital twin |
| CN114399066A (en) * | 2022-01-15 | 2022-04-26 | 中国矿业大学(北京) | Mechanical equipment predictability maintenance system and maintenance method based on weak supervision learning |
| CN117189713A (en) * | 2023-07-27 | 2023-12-08 | 中国人民解放军海军潜艇学院 | Hydraulic system fault diagnosis method based on digital twin driving |
| CN117490760A (en) * | 2023-10-24 | 2024-02-02 | 山东省科学院自动化研究所 | Calender monitoring and fault early warning method and system based on digital twin technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117892114A (en) | 2024-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3098681B1 (en) | Artificial intelligence based health management of host system | |
| CN109800127A (en) | A kind of system fault diagnosis intelligence O&M method and system based on machine learning | |
| CN109840157A (en) | Method, apparatus, electronic equipment and the storage medium of fault diagnosis | |
| KR20170125265A (en) | Plant system, and fault detecting method thereof | |
| JP2008536219A (en) | Diagnosis and prediction method and system | |
| CN111190804A (en) | Multi-level deep learning log fault detection method for cloud native system | |
| CN111897695B (en) | Method and device for acquiring KPI abnormal data sample and computer equipment | |
| JPS6014303A (en) | Knowledge-based diagnosis system | |
| CN113760670A (en) | Cable joint abnormity early warning method and device, electronic equipment and storage medium | |
| CN112817785A (en) | Anomaly detection method and device for micro-service system | |
| CN116304909A (en) | Abnormality detection model training method, fault scene positioning method and device | |
| CN117892114B (en) | Crane fault prediction method and system based on Internet of Things | |
| CN109990803B (en) | Method and device for detecting system abnormity and method and device for sensor processing | |
| CN111813872B (en) | Method, device and equipment for generating fault troubleshooting model | |
| CN113093695A (en) | Data-driven SDN controller fault diagnosis system | |
| KR102108975B1 (en) | Apparatus and method for condition based maintenance support of naval ship equipment | |
| CN109213119B (en) | Complex industry key component fault prediction method and system based on online learning | |
| CN115219883A (en) | Automatic detection system and detection method for electric automation equipment | |
| CN109886292B (en) | Abnormal reason diagnosis method based on abnormal association graph | |
| CN109558258B (en) | Method and device for positioning root fault of distributed system | |
| CN116381419B (en) | Transmission line fault processing method, device, computer equipment and storage medium | |
| WO2023044907A1 (en) | Method and device for monitoring equipment health and computer readable storage medium | |
| Wang et al. | Complex equipment diagnostic reasoning based on neural network algorithm | |
| CN118132451B (en) | Automatic test and error diagnosis system and method for computer operating system | |
| CN118051657B (en) | Method and system for testing case library for fault location of data private line |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |