CN116010828A - Method and system for diagnosing fault cause of unit - Google Patents
Method and system for diagnosing fault cause of unit Download PDFInfo
- Publication number
- CN116010828A CN116010828A CN202211621819.0A CN202211621819A CN116010828A CN 116010828 A CN116010828 A CN 116010828A CN 202211621819 A CN202211621819 A CN 202211621819A CN 116010828 A CN116010828 A CN 116010828A
- Authority
- CN
- China
- Prior art keywords
- data
- fault
- clustering
- failure
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012545 processing Methods 0.000 claims abstract description 47
- 238000004458 analytical method Methods 0.000 claims abstract description 45
- 238000003745 diagnosis Methods 0.000 claims abstract description 18
- 238000010977 unit operation Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims description 34
- 230000008859 change Effects 0.000 claims description 26
- 238000012795 verification Methods 0.000 claims description 15
- 238000013459 approach Methods 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000012417 linear regression Methods 0.000 claims description 4
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 6
- 238000012423 maintenance Methods 0.000 abstract description 6
- 238000007689 inspection Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 16
- 230000008569 process Effects 0.000 description 7
- 238000004220 aggregation Methods 0.000 description 6
- 230000002776 aggregation Effects 0.000 description 6
- 238000009833 condensation Methods 0.000 description 5
- 230000005494 condensation Effects 0.000 description 5
- 238000005520 cutting process Methods 0.000 description 5
- 238000003860 storage Methods 0.000 description 4
- 238000003491 array Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000001502 supplementing effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Landscapes
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The application relates to the technical field of unit fault cause diagnosis, in particular to a unit fault cause diagnosis method and system, wherein the method comprises the following steps: acquiring unit operation state data; determining a fault node in the operation state data; carrying out prior processing on the data of the previous stage of the fault node, and judging the fault reason of the fault node according to the prior processing; if the failure of obtaining the failure reason of the failure node according to the prior processing, inputting the data of the previous stage of the failure node into a pre-trained failure analysis model, and outputting the failure reason of the failure node. In the application, to the diagnosis of unit fault reason, at first carry out the analysis through prior processing, when the failure of prior inspection processing obtained the fault reason, rethread fault analysis model carries out the analysis to more accurate analysis and traceback to the fault reason, the technical scheme in this application compares in the manual detection unit trouble, can not receive the influence of artifical subjective judgement, and does not rely on maintenance experience, and the rate of accuracy is higher.
Description
5 technical field
The application relates to the technical field of unit fault cause diagnosis, in particular to a unit fault cause diagnosis method and system.
Background
0 during operation of the Cold chain Unit, failsafe is one of the most common Unit faults, often
The method has the characteristics of high occurrence frequency, rapid fault occurrence speed and difficult cause tracing. When the unit breaks down, the fault cause of the unit is detected by a plurality of methods through manual detection from a plurality of angles, the manual detection result is more influenced by manual subjective judgment, and the accuracy of the detection result is lower due to the fact that the maintenance experience is relied on, so that the fault cause is difficult to trace.
Disclosure of Invention
Therefore, the invention aims to provide a method and a system for diagnosing the failure cause of a unit, so as to solve the problems that the accuracy of manually detecting the failure cause of the unit is low and the failure cause is difficult to trace in the prior art.
0 according to a first aspect of the embodiment of the present invention, there is provided a unit fault cause diagnosis method, including:
acquiring unit operation state data;
determining a fault node in the operating state data;
performing prior processing on the data of the previous stage of the fault node, and judging the fault according to the prior processing
The failure cause of the node;
5 if the failure reason of the failure node is not obtained according to the prior processing, the failure node is the former one
The data of the stage is input into a pre-trained fault analysis model, and the fault reason of the fault node is output.
Preferably, the method further comprises:
acquiring historical running state data of a plurality of units as training data;
and training the fault analysis model according to the training data.
Preferably, training the fault analysis model according to the training data comprises:
determining a fault node in the training data;
carrying out prior processing on the data of the previous stage of the fault node, and screening out the data of the fault reason of the fault node, which is not obtained according to prior processing, from the training data as cluster data;
performing primary clustering on the clustered data to obtain an initial clustering result;
and carrying out secondary clustering on the initial clustering result to screen out a target clustering result in the initial clustering result.
Preferably, the first clustering is performed on the clustered data to obtain an initial clustering result, including:
determining a cluster K value of the cluster data;
clustering the clustering data according to the clustering K value to obtain clustering clusters and clustering center point data;
the upper limit and the lower limit of each clustering cluster approach to the clustering center point based on a linear regression method to obtain an approach result;
and the approach result is used as standard time data to be transmitted back to the clustering cluster set to cut the data in the clustering cluster set, so that an initial clustering result is obtained.
Preferably, when cutting the data in the clustering cluster, the principle of multiple cutting and fewer supplementing is followed, so as to ensure that the jitter time intervals of the data in the clustering cluster are consistent.
Preferably, the second clustering is performed on the initial clustering result to screen out a target clustering result in the initial clustering result, including:
dividing the cut data into a plurality of dimensions, and observing the data change characteristics of parameters of each dimension in standard time data;
setting a loss function as a comparison standard, comparing all data in the clustering cluster in pairs, and acquiring the data change characteristic with the lowest loss function under each parameter as an alternative data change characteristic through negative sampling;
transversely pairing the alternative data change characteristics of all parameters in the clustering cluster to obtain target data change characteristics corresponding to the clustering cluster as a target clustering result;
and recording a fault reason corresponding to the change characteristic of the target data as a model output result.
Preferably, the method further comprises:
obtaining verification data, and dividing the verification data into a plurality of groups for cross verification on the trained fault analysis model;
obtaining a cross verification result; the cross-validation results include at least: accuracy, recall and model evaluation parameters;
and when the accuracy is higher than a first preset value, the recall rate is higher than a second preset value, and the model evaluation parameter is higher than a third preset value, determining that the fault analysis model is qualified in verification.
Preferably, the a priori processing includes:
judging whether the failure cause of the failure node is an external factor according to the data of the previous stage of the failure node;
if the external factors are external factors, outputting the fault reasons of the fault nodes;
if the external factors are not the external factors, outputting the failure reason that the failure node cannot be obtained.
Preferably, the method further comprises: acquiring updated training data;
and updating and iterating the fault analysis model according to the updated training data.
According to a second aspect of the embodiment of the present invention, there is provided a unit fault cause diagnosis system, including:
the acquisition module is used for acquiring the running state data of the unit;
a determining module, configured to determine a failure node in the operation state data;
the first fault judging module is used for carrying out prior processing on the data of the previous stage of the fault node and judging the fault reason of the fault node according to the prior processing;
and the second fault judging module is used for inputting the data of the previous stage of the fault node into a pre-trained fault analysis model and outputting the fault reason of the fault node if the fault reason of the fault node cannot be obtained according to the prior processing.
The technical scheme that this application provided can include following beneficial effect: the application provides a method and a system for diagnosing a fault cause of a unit, wherein the method comprises the following steps: acquiring unit operation state data; determining a fault node in the operation state data; carrying out prior processing on the data of the previous stage of the fault node, and judging the fault reason of the fault node according to the prior processing; if the failure of obtaining the failure reason of the failure node according to the prior processing, inputting the data of the previous stage of the failure node into a pre-trained failure analysis model, and outputting the failure reason of the failure node. In the application, to the diagnosis of unit fault reason, at first carry out the analysis through prior processing, when the failure of prior inspection processing obtained the fault reason, rethread fault analysis model carries out the analysis to more accurate analysis and traceback to the fault reason, the technical scheme in this application compares in the manual detection unit fault reason, can not receive the influence of artifical subjective judgement, and does not rely on maintenance experience, and the rate of accuracy is higher.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application.
FIG. 1 is a flow chart illustrating a method of diagnosing a cause of a unit fault according to an exemplary embodiment;
FIG. 2 is a flowchart illustrating a first clustering in a unit fault cause diagnosis method, according to an exemplary embodiment;
FIG. 3 is a flowchart illustrating a second clustering in a unit fault cause diagnosis method, according to an exemplary embodiment;
FIG. 4 is a schematic block diagram illustrating a unit fault cause diagnostic system according to an exemplary embodiment.
Reference numerals: an acquisition module-41; a determining module-42; a first failure determination module-43; and a second fault determination module-44.
Detailed Description
Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the accompanying claims.
Example 1
Fig. 1 is a flowchart illustrating a method for diagnosing a cause of a unit fault according to an exemplary embodiment, and referring to fig. 1, a method for diagnosing a cause of a unit fault includes:
s11, acquiring unit operation state data;
step S12, determining a fault node in the running state data;
step S13, carrying out prior processing on the data of the previous stage of the fault node, and judging the fault reason of the fault node according to the prior processing;
and S14, if the failure causes of the failure nodes are not obtained according to prior processing, inputting the data of the previous stage of the failure nodes into a pre-trained failure analysis model, and outputting the failure causes of the failure nodes.
It should be noted that, the technical solution in this embodiment is applied to the fault cause diagnosis process of the unit, and is specifically applied to fault protection diagnosis of the cold chain unit. In the running process of the cold chain, the fault protection is one of the most common faults, and the fault protection is characterized by high occurrence frequency, rapid fault occurrence speed and most difficult cause tracing.
In this embodiment, the unit operation state data within one month may be obtained to perform the fault cause analysis.
In specific practice, after the unit operation state data is obtained, the unit operation state data is firstly required to be subjected to data cleaning, whether null data exist in the unit operation state data or not is judged, whether the data types are correct or not is judged, and whether the data corresponding relations are consistent or not is judged.
And then cutting the running state data of the unit by taking the fault node as a reference, and reserving the fault node and the data of the previous stage of the fault node for each piece of cut data.
It should be noted that the previous stage may be a previous time stage or a previous operation stage, which is not limited herein.
It can be understood that the unit fault cause diagnosis method in this embodiment includes: acquiring unit operation state data; determining a fault node in the operation state data; carrying out prior processing on the data of the previous stage of the fault node, and judging the fault reason of the fault node according to the prior processing; if the failure of obtaining the failure reason of the failure node according to the prior processing, inputting the data of the previous stage of the failure node into a pre-trained failure analysis model, and outputting the failure reason of the failure node. In this embodiment, for the diagnosis of the failure cause of the unit, the failure cause is firstly analyzed through the prior processing, and when the failure cause cannot be obtained through the prior inspection processing, the failure cause is analyzed through the failure analysis model, so that the failure cause is more accurately analyzed and traced.
Example two
Note that the a priori processing includes:
judging whether the failure cause of the failure node is an external factor according to the data of the previous stage of the failure node;
if the external factors are external factors, outputting the fault reasons of the fault nodes;
if the external factors are not the external factors, outputting the failure reason that the failure node cannot be obtained.
It will be appreciated that the failure cause of the output failure node may be affected by external factors, and taking the high voltage protection failure as an example, the occurrence of high voltage protection may be affected by the following external factors:
1) The high-voltage protection switch fault causes the high-voltage protection fault, that is, the actual condensation temperature of the unit does not reach the condensation temperature set by the unit, and the reason for the high-voltage protection is the high-voltage protection switch fault.
In the specific implementation, the condensation temperature data in the data of the previous stage of the fault node is observed, and if the condensation temperature of the previous stage of the fault node does not exceed the condensation temperature set by the unit, the fault cause of the high-voltage protection fault is judged to be the high-voltage protection switch fault.
2) The high-voltage protection early warning is caused by the excessively high temperature of the external environment, and the reason for the fault is uncontrollable.
In the concrete implementation, external environment temperature data in the data of the previous stage of the fault node is processed
And observing, if the external environment temperature of the previous stage of the fault node exceeds a set external temperature value, judging that the fault reason of the high 5-voltage protection fault is that the external environment temperature is too high.
Preferably, the set ambient temperature value may be, but is not limited to, 50 ℃.
If the failure cause of the high-voltage protection failure is not the external factor, outputting the failure cause of failure node, and then continuing analysis by the failure analysis model.
Example III
0, the method further comprises:
acquiring historical running state data of a plurality of units as training data;
and training a fault analysis model according to the training data.
Specifically, training the fault analysis model according to the training data includes:
determining a fault node in the training data;
5 carrying out prior processing on the data of the previous stage of the fault node, and screening out failure according to the training data
The prior processing is used for obtaining the data of the fault reason of the fault node as cluster data;
performing first clustering on the clustered data to obtain an initial clustering result;
and carrying out secondary clustering on the initial clustering result to screen out a target clustering result in the initial clustering result.
0, referring to fig. 2, clustering data is first clustered to obtain initial cluster nodes
The fruit comprises:
s21, determining a clustering K value of the clustered data;
s22, clustering the clustering data according to the clustering K value to obtain a clustering cluster and a clustering center
Point data;
step S23, clustering the upper limit and the lower limit of each cluster to the cluster center based on a linear regression method
The points approach to obtain an approach result;
and S24, returning the approach result as standard time data to a clustering cluster set to cut the data in the clustering cluster set, so as to obtain an initial clustering result.
It will be appreciated that after the influence of external factors is removed, the rest of the clustered data first needs to be clustered, and in specific practice, the clustered K value of the clustered data may be determined using an elbow method. The core idea of the elbow method is: as K increases, the sample division becomes finer, the degree of aggregation of each cluster increases gradually, and then the square error and the clustering error of all samples naturally become smaller gradually. And when K is smaller than the real cluster number, the aggregation degree of each cluster can be greatly increased due to the increase of K, the aggregation degree return obtained by increasing K again can be rapidly reduced when K reaches the real cluster number, the aggregation degree return obtained by increasing K is rapidly reduced, and then the aggregation degree return is gradually flattened along with the continuous increase of the K value, namely the relation diagram of the aggregation error and K is in the shape of an elbow, and the K value corresponding to the elbow is the real cluster number of the data.
After determining the clustering K value of the clustering data, clustering the clustering data according to the clustering K value to obtain clustering clusters and clustering center point data. In specific practice, the clustering data may be clustered using a K-means++ algorithm. The K-means++ algorithm is an optimization of the method of randomly initializing the centroid for the K-Means. The K-Means algorithm, also known as the K-average or K-Means algorithm, is a widely used clustering algorithm. The K-Means algorithm is an unsupervised algorithm focused on similarity, and uses distance as a criterion for measuring similarity between data objects, namely, the smaller the distance between data objects is, the higher the similarity between the data objects is, and the more likely it is that the data objects are in the same class cluster. K-Means is so called because it can find K different clusters, and the center of each cluster is calculated using the average of the values contained in the cluster. The location selection of K initialized centroids has a large impact on both the final clustering result and the run time, so the selection of the appropriate K centroids is required. If it is simply a completely random choice, it may result in slow algorithm convergence, and the K-means++ algorithm is an optimization of the method for K-Means random initialization of the centroid. The most essential difference between the K-means++ algorithm and the K-Means algorithm is the initialization process at the K cluster centers. The basic principle of the K-means++ algorithm in the process of initializing the cluster centers is to make the mutual distance between the initial cluster centers as far as possible, so that the problems can be avoided.
And clustering the clustering data based on a K-means++ algorithm according to the clustering K value to obtain K clustering results and clustering center point data, approaching the upper limit and the lower limit of each clustering result to the clustering center point by a linear regression method to obtain an approaching result, and returning the approaching result as standard time data to a clustering cluster set to cut the data in the clustering cluster set to obtain an initial clustering result.
In specific practice, when cutting the data in the clustering cluster, the principle of more cutting and less supplementing is followed to ensure that the jitter time intervals of the data in the clustering cluster are consistent.
It should be noted that the K-means++ algorithm in this embodiment is only exemplary, and other clustering methods such as DBSCAN or K-Means may be used in specific practice.
It should be noted that, the essence of the second clustering is to perform the second clustering on the data in the single cluster according to the similarity distribution of the data change characteristics, so as to obtain the effective data change characteristics.
It should be noted that referring to fig. 3, performing a second clustering on the initial clustering result to screen out a target clustering result in the initial clustering result includes:
s31, dividing the cut data into a plurality of dimensions, and observing the data change characteristics of parameters of each dimension in standard time data;
step S32, setting a loss function as a comparison standard, comparing all data in the clustering cluster in pairs, and acquiring a data change characteristic with the lowest loss function under each parameter as an alternative data change characteristic through negative sampling;
step S33, carrying out transverse pairing on the alternative data change characteristics of all parameters in the clustering cluster to obtain target data change characteristics corresponding to the clustering cluster as a target clustering result;
and step S34, recording a fault reason corresponding to the target data change characteristic as a model output result.
It will be appreciated that a plurality of cut data are obtained from the first clustering, each of which may be divided into six to seven dimensions, each dimension having a corresponding parameter. In specific practice, it is generally divided into seven dimensions, suction pressure, discharge pressure, suction temperature, suction superheat, condensing temperature, discharge temperature, and ambient temperature. In this case, it is necessary to observe the continuous variation of each parameter in the standard time data, in order to compare the data variation characteristics of different parameters in the previous stage of the failed node.
In the specific practice of the present invention, a Loss of Function Loss Function of Loss can be set as an alignment criterion, loss of Loss Function in this embodiment the Function uses the setting of (a×2+b×2). The standard condition for setting the Loss Function is data of uniform time period and continuous change. In order to compare the data change characteristics of parameters of each dimension, the Loss Function is needed to be used for carrying out pairwise comparison on all data in the clustering cluster, and if the obtained Loss Function is smaller, the data change characteristics are more representative, and the availability of the fault data is higher. There may be many settings of the Loss Function, and (a+b+2) set in this embodiment is a parameter standard commonly used for the Loss Function.
In specific practice, after the target data change characteristics corresponding to the clustering cluster are obtained, fault reasons corresponding to the target data change characteristics can be defined and recorded according to historical experience or maintenance specialists.
Preferably, a maintenance method corresponding to the fault cause is also determined and recorded.
At this time, training of training data corresponding to one unit is completed, and because the training data includes historical running state data of a plurality of units, training of the historical running state data of all units is completed before training of the fault analysis model is completed.
After training is completed, the fault analysis model also needs to be verified.
Based on this, the method further comprises:
acquiring verification data, and dividing the verification data into a plurality of groups for cross verification on the trained fault analysis model;
obtaining a cross verification result; the cross-validation results include at least: accuracy, recall and model evaluation parameters;
and when the accuracy is higher than a first preset value, the recall rate is higher than a second preset value, and the model evaluation parameter is higher than a third preset value, determining that the fault analysis model is qualified in verification.
In specific practice, the trained failure analysis model may be cross-validated based on, but not limited to, a sliding window algorithm.
It should be noted that, the model evaluation parameter is an F1-measure value, the F1-measure value is a harmonic mean of an accurate value and a recall rate, the F1-measure value is a statistical measure, and is mainly used for evaluating the quality of the model, and the model is qualified and effective when the F1-measure value is higher.
It should be noted that the method further includes:
acquiring updated training data;
and updating and iterating the fault analysis model according to the updated training data.
It can be appreciated that the fault analysis model has the capability of updating iterative learning, and is more trained on the data
After the method is new, the fault analysis model can be updated and iterated according to the updated training data, sustainable updated fault knowledge storage is provided for the fault analysis 5 model, and more data change characteristics can be collected along with the filling of the updated data to perfect the fault analysis model.
Example IV
FIG. 4 is a schematic illustration of a unit fault cause diagnostic system according to an exemplary embodiment
Referring to fig. 4, a unit fault cause diagnosis system includes: the 0 acquisition module 41 is used for acquiring the running state data of the unit;
a determination module 42 for determining a failed node in the operational status data;
the first fault judging module 43 is configured to perform prior processing on data of a previous stage of the fault node, and judge a fault cause of the fault node according to the prior processing;
a second fault judging module 44 for inputting the data of the previous stage of the fault node into a pre-trained fault analysis model and outputting the fault if the fault source 5 factor of the fault node cannot be obtained according to the prior process
The failure cause of the node.
It can be understood that the unit fault cause diagnosis system in this embodiment includes: an acquisition module 41, configured to acquire unit operation state data; a determination module 42 for determining a failed node in the operational status data; the first fault judging module 43 is configured to perform a priori 0 processing on data of a previous stage of the fault node, and judge a fault cause of the fault node according to the priori processing; a second failure determination module 44 for
If the failure of obtaining the failure reason of the failure node according to the prior processing, inputting the data of the previous stage of the failure node into a pre-trained failure analysis model, and outputting the failure reason of the failure node. In this embodiment, for the diagnosis of the cause of the unit fault, the unit fault is first analyzed through a priori processing, and the prior empirical processing fails to obtain
When the fault reasons are detected, the fault analysis model is used for analyzing, so that the fault reasons are more accurately analyzed and traced 5, and compared with the manual detection of the fault reasons of the unit, the technical scheme in the embodiment can not be used for detecting the fault reasons of the unit
The method is influenced by manual subjective judgment, does not depend on maintenance experience, and has higher accuracy.
It is to be understood that the same or similar parts in the above embodiments may be referred to each other, and that in some embodiments, the same or similar parts in other embodiments may be referred to.
It should be noted that in the description of the present application, the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. Furthermore, in the description of the present application, unless otherwise indicated, the meaning of "plurality" means at least two.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and further implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present application.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Those of ordinary skill in the art will appreciate that all or a portion of the steps carried out in the method of the above-described embodiments may be implemented by a program to instruct related hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, includes one or a combination of the steps of the method embodiments.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing module, or each unit may exist alone physically, or two or more units may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product.
The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Although embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the application, and that variations, modifications, alternatives, and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the application.
Claims (10)
1. The unit fault cause diagnosis method is characterized by comprising the following steps of:
acquiring unit operation state data;
determining a fault node in the operating state data;
carrying out prior processing on the data of the previous stage of the fault node, and judging the fault reason of the fault node according to the prior processing;
if the failure of obtaining the failure reason of the failure node according to the prior processing, inputting the data of the previous stage of the failure node into a pre-trained failure analysis model, and outputting the failure reason of the failure node.
2. The method according to claim 1, wherein the method further comprises:
acquiring historical running state data of a plurality of units as training data;
and training the fault analysis model according to the training data.
3. The method of claim 1, wherein training the failure analysis model based on the training data comprises:
determining a fault node in the training data;
carrying out prior processing on the data of the previous stage of the fault node, and screening out the data of the fault reason of the fault node, which is not obtained according to prior processing, from the training data as cluster data;
performing primary clustering on the clustered data to obtain an initial clustering result;
and carrying out secondary clustering on the initial clustering result to screen out a target clustering result in the initial clustering result.
4. A method according to claim 3, wherein the clustering of the clustered data for the first time results in an initial clustering result, comprising:
determining a cluster K value of the cluster data;
clustering the clustering data according to the clustering K value to obtain clustering clusters and clustering center point data;
the upper limit and the lower limit of each clustering cluster approach to the clustering center point based on a linear regression method to obtain an approach result;
and the approach result is used as standard time data to be transmitted back to the clustering cluster set to cut the data in the clustering cluster set, so that an initial clustering result is obtained.
5. The method of claim 4, wherein the data within the cluster clusters is cut to follow a more-cut less-complement principle to ensure consistent jitter time intervals of the data within the cluster clusters.
6. The method of claim 4, wherein performing a second clustering on the initial cluster results to screen out target cluster results in the initial cluster results comprises:
dividing the cut data into a plurality of dimensions, and observing the data change characteristics of parameters of each dimension in standard time data;
setting a loss function as a comparison standard, comparing all data in the clustering cluster in pairs, and acquiring the data change characteristic with the lowest loss function under each parameter as an alternative data change characteristic through negative sampling;
transversely pairing the alternative data change characteristics of all parameters in the clustering cluster to obtain target data change characteristics corresponding to the clustering cluster as a target clustering result;
and recording a fault reason corresponding to the change characteristic of the target data as a model output result.
7. The method according to claim 2, wherein the method further comprises:
obtaining verification data, and dividing the verification data into a plurality of groups for cross verification on the trained fault analysis model;
obtaining a cross verification result; the cross-validation results include at least: accuracy, recall and model evaluation parameters;
and when the accuracy is higher than a first preset value, the recall rate is higher than a second preset value, and the model evaluation parameter is higher than a third preset value, determining that the fault analysis model is qualified in verification.
8. The method of claim 1, wherein the prior processing comprises:
judging whether the failure cause of the failure node is an external factor according to the data of the previous stage of the failure node;
if the external factors are external factors, outputting the fault reasons of the fault nodes;
if the external factors are not the external factors, outputting the failure reason that the failure node cannot be obtained.
9. The method according to claim 2, wherein the method further comprises:
acquiring updated training data;
and updating and iterating the fault analysis model according to the updated training data.
10. A system for diagnosing a cause of a unit fault, comprising:
the acquisition module is used for acquiring the running state data of the unit;
a determining module, configured to determine a failure node in the operation state data;
the first fault judging module is used for carrying out prior processing on the data of the previous stage of the fault node and judging the fault reason of the fault node according to the prior processing;
and the second fault judging module is used for inputting the data of the previous stage of the fault node into a pre-trained fault analysis model and outputting the fault reason of the fault node if the fault reason of the fault node cannot be obtained according to the prior processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211621819.0A CN116010828A (en) | 2022-12-16 | 2022-12-16 | Method and system for diagnosing fault cause of unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211621819.0A CN116010828A (en) | 2022-12-16 | 2022-12-16 | Method and system for diagnosing fault cause of unit |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116010828A true CN116010828A (en) | 2023-04-25 |
Family
ID=86037045
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211621819.0A Pending CN116010828A (en) | 2022-12-16 | 2022-12-16 | Method and system for diagnosing fault cause of unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116010828A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842431A (en) * | 2023-08-31 | 2023-10-03 | 中国船舶集团国际工程有限公司 | Steel structure health monitoring and evaluating method |
-
2022
- 2022-12-16 CN CN202211621819.0A patent/CN116010828A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842431A (en) * | 2023-08-31 | 2023-10-03 | 中国船舶集团国际工程有限公司 | Steel structure health monitoring and evaluating method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN107677472B (en) | The bearing state noise diagnostics algorithm that network-oriented Variable Selection is merged with Characteristic Entropy | |
| US20150219530A1 (en) | Systems and methods for event detection and diagnosis | |
| CN110703214B (en) | Weather radar state evaluation and fault monitoring method | |
| CN109033499A (en) | A kind of aero-engine method for predicting residual useful life of multistage consistency check | |
| KR101843365B1 (en) | Integrated Diagnostic System and Database based on rules and cases | |
| CN115982602B (en) | Photovoltaic transformer electrical fault detection method | |
| CN116010828A (en) | Method and system for diagnosing fault cause of unit | |
| CN116678552B (en) | Abnormality monitoring method for optical fiber stress sensor in variable temperature environment | |
| CN113420061B (en) | Steady state working condition analysis method, optimization debugging method and system of oil refining and chemical production device | |
| CN116450399A (en) | Fault diagnosis and root cause positioning method for micro service system | |
| CN113298162A (en) | Bridge health monitoring method and system based on K-means algorithm | |
| CN112149750A (en) | Water supply network pipe burst identification data driving method | |
| CN115277464A (en) | Cloud network change flow anomaly detection method based on multi-dimensional time series analysis | |
| CN116295948A (en) | Abnormality detection method, system and storage medium of industrial temperature sensor in large temperature difference environment | |
| CN112434755B (en) | Data anomaly sensing method based on heterogeneous system | |
| CN110057588B (en) | Bearing early fault detection and diagnosis method and system based on fusion of singular value and graph theory characteristics | |
| CN116910677A (en) | Industrial instrument fault diagnosis method and system | |
| CN107798149A (en) | A kind of aircraft maintainability appraisal procedure | |
| CN110956112A (en) | Novel high-reliability slewing bearing life evaluation method | |
| CN109389313A (en) | A kind of failure modes diagnostic method based on weighting neighbour's decision | |
| CN115766513A (en) | Anomaly detection method and device | |
| CN113586554A (en) | Hydraulic motor detection system and method | |
| Tao et al. | A novel model of one-class bearing fault detection using SVDD and genetic algorithm | |
| CN112699609A (en) | Diesel engine reliability model construction method based on vibration data |
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 |