CN109754110A - A kind of method for early warning and system of traction converter failure - Google Patents
A kind of method for early warning and system of traction converter failure Download PDFInfo
- Publication number
- CN109754110A CN109754110A CN201711094900.7A CN201711094900A CN109754110A CN 109754110 A CN109754110 A CN 109754110A CN 201711094900 A CN201711094900 A CN 201711094900A CN 109754110 A CN109754110 A CN 109754110A
- Authority
- CN
- China
- Prior art keywords
- data
- train
- fault
- early warning
- random forest
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/72—Electric energy management in electromobility
Abstract
This application discloses a kind of method for early warning of traction converter failure, comprising: extracts the operation data of commutation inversion module and network from historical data, and obtains data characteristics and fault message using commutation inversion module operation data building EOVW nergy Index；Data characteristics and network operation data are associated by vehicle, license number and time, data after association are executed into clustering operation by vehicle and fault type, obtain the time range of data exception before failure, and to abnormal data additional fault early warning label；All data introducing random forest grader is iterated training, until the classification accuracy of all random forest graders reaches threshold value, is classified using classifier after training to actual operating data, not by then by sending warning information.It improves the service life of traction convertor, reduce operation expense.The application further simultaneously discloses a kind of early warning system of traction converter failure, has above-mentioned beneficial effect.
Description
Technical field
This application involves rail traffic fault pre-alarming technical field, in particular to the pre- police of a kind of traction converter failure
Method and system.
Background technique
Train is one of modern main means of transport, and great effect has been played in terms of passenger and cargo transport, peace
Full problem is constantly subjected to paying close attention to for people from all walks of life, and wherein the traction convertor in train traction main circuit is important as train
Component part, its safety problem are even more the most important thing.
Wherein, current transformer commutation inversion module is the core component of traction convertor, in the prior art, is become to monitor
The working condition of device is flowed, various sensors (such as voltage, current sensor) is installed in current transformer commutation inversion module, works as mould
When block breaks down, fault-signal can be sent to drivers' cab by current transformer traction control unit, and driver is reminded to take reset, block
Failure current transformer, the spare current transformer of starting, the even measures such as parking, the safety of train is ensured with this.
But the prior art is only the condition monitoring using sensor to commutation inversion module, and is detecting failure letter
Number when alarm, i.e., only can just detect fault-signal after failure occurs for a certainty and be sent to drivers' cab.And one
Current transformer catastrophe failure (the fried damage of such as thyristor or diode) occurs for denier, repairs excessively high with update cost cost.
So how to provide a kind of can be analyzed according to data for traction convertor, failure hair is found in advance
The fault pre-alarming mechanism of raw omen is those skilled in the art's urgent problem to be solved.
Summary of the invention
The purpose of the application is to provide the method for early warning and system of a kind of traction converter failure, drives by using data
Dynamic model pre-warning method, on the one hand accuracy with higher, on the other hand can excavate from mass data and be not easy to examine
The stealth of the failure and device felt contacts, and improves the service life of traction convertor, reduces operation expense, and is right
Stability, safety, the reliability for promoting train play facilitation.
In order to solve the above technical problems, the application provides a kind of method for early warning of traction converter failure, the method for early warning
Include:
Extracted respectively from the traction convertor historical data and network history data of train obtain the train rectification it is inverse
Become module operation data and network operation data, and is obtained using commutation inversion module operation data building EOVW nergy Index
To the data characteristics and fault message of commutation inversion module；Wherein, the fault message includes fault type；
The data characteristics and the network operation data are closed by the different automobile types of train, license number and time
Connection, data after being associated with；
Data after the association are executed into clustering operation by different vehicles and the fault type, before obtaining failure
The time range of data exception, and to the abnormal data additional fault early warning label in the time range；
The abnormal data for being attached with the fault pre-alarming label and normal data are introduced random forest grader to change
Generation training, until the classification accuracy of all random forest graders reaches threshold value, random forest grader after being trained；
Wherein, the random forest grader constructs to obtain according to failure mode and dichotomy；
Train actual operating data is analyzed using random forest grader after the training, if not passing through the instruction
Random forest grader after white silk, then send warning information by preset path.
Optionally, commutation inversion module is obtained using commutation inversion module operation data building EOVW nergy Index
Data characteristics and fault message, comprising:
The commutation inversion module operation data is passed through into wavelet transformation in seconds and extracts each sensor information
EOVW nergy Index；
Using the EOVW nergy Index as the data characteristics of the commutation inversion module of current second, and obtain the failure letter
Breath.
Optionally, data after the association are executed into clustering operation by different vehicles and the fault type, obtained
The time range of data exception before to failure, comprising:
Data after the association are subjected to hierarchical clustering processing by different vehicles and different fault types, obtain level
Data after clustering processing；Wherein, data include cluster information and center position information after the hierarchical clustering processing；
K-means clustering processing is carried out using the cluster information and the center position information, determines data before failure
Abnormal time range.
Optionally, after sending warning information by preset path, further includes:
Whether judgement repeatedly receives identical warning information within a preset time；
If so, it is abnormal to determine that the train exists, and be there is into abnormal judgement result in the train and be sent to train
Driver.
Optionally, which further includes；
All warning information received are recorded and saved, warning information log are generated, so as to according to described pre-
Alert information log carries out subsequent analysis and traces.
Present invention also provides a kind of early warning system of traction converter failure, which includes:
Operation data obtains and analytical unit, for dividing from the traction convertor historical data and network history data of train
The commutation inversion module operation data and network operation data for arriving the train are indescribably obtained, and utilizes the commutation inversion module
Operation data building EOVW nergy Index obtains the data characteristics and fault message of commutation inversion module；Wherein, the failure letter
Breath includes fault type；
Data associating unit, for the different automobile types by the data characteristics and the network operation data by train, vehicle
Number and the time be associated, data after being associated with；
Clustering operation execution unit, for holding data after the association by different vehicles and the fault type
The operation of row clustering obtains the time range of data exception before failure, and additional to the abnormal data in the time range
Fault pre-alarming label；
Classifier training unit, for will be attached with the fault pre-alarming label abnormal data and normal data introduce with
Machine forest classified device is iterated training, until the classification accuracy of all random forest graders reaches threshold value, is instructed
Random forest grader after white silk；Wherein, the random forest grader constructs to obtain according to failure mode and dichotomy；
Physical fault predicting unit, for using random forest grader after the training to train actual operating data into
Row analysis, if not sending warning information by preset path by random forest grader after the training.
Optionally, the operation data obtains and analytical unit includes:
Wavelet transformation subelement, for proposing the commutation inversion module operation data by wavelet transformation in seconds
Take out the EOVW nergy Index of each sensor information；
Subelement is handled, for using the EOVW nergy Index as the data characteristics of the commutation inversion module of current second,
And obtain the fault message.
Optionally, the clustering operation execution unit includes:
Hierarchical clustering handles subelement, for by data after the association by different vehicles and different fault types into
The processing of row hierarchical clustering obtains data after hierarchical clustering processing；Wherein, after the hierarchical clustering processing data include cluster information and
Center position information；
K-means clustering processing subelement, for carrying out K- using the cluster information and the center position information
Means clustering processing determines the time range of data exception before failure.
Optionally, the physical fault predicting unit further include:
Expert Rules determine subelement, for judging whether repeatedly receive identical warning information within a preset time；
Expert Rules determine through subelement, for determining that it is abnormal that the train exists, and by the train in the presence of abnormal
Judgement result be sent to train driver.
Optionally, the early warning system further include:
Storage unit is recorded, for recording and saving all warning information received, generates warning information day
Will, to carry out subsequent analysis according to the warning information log and to trace.
The method for early warning of a kind of traction converter failure provided herein, from the traction convertor historical data of train
It is extracted respectively with network history data and obtains the commutation inversion module operation data and network operation data of the train, and utilized
The commutation inversion module operation data building EOVW nergy Index obtains the data characteristics and fault message of commutation inversion module；
The data characteristics and the network operation data are associated by the different automobile types of train, license number and time, closed
Data after connection；Data after the association are executed into clustering operation by different vehicles and the fault type, obtain failure
The time range of preceding data exception, and to the abnormal data additional fault early warning label in the time range；It will add
The abnormal data and normal data for stating fault pre-alarming label introduce random forest grader and are iterated training, until all random
The classification accuracy of forest classified device reaches threshold value, random forest grader after being trained；Using random after the training
Forest classified device analyzes train actual operating data, if not passed through by random forest grader after the training
Preset path sends warning information.
Obviously, technical solution provided herein is adopted first against commutation inversion module historical data with network data
The inconsistent situation of collection frequency is handled, and data after processing are associated, and and then press the data after association
Vehicle and fault type information divide and pass sequentially through two kinds of hierarchical clustering algorithms, with obtain abnormal data before failure occurs when
Between range, and be abnormal data additional fault early warning label, normal data and abnormal data are uniformly finally put into fault type
Two points of judgements are carried out in the random forest classification of generation.The method for early warning by using data-driven model pre-warning method, one
Aspect accuracy with higher, on the other hand can be excavated from mass data it is not noticeable to failure and device it is hidden
Shape connection, improves the service life of traction convertor, reduces operation expense, and to stability, the peace for promoting train
Quan Xing, reliability play facilitation.The application additionally provides a kind of early warning system of traction converter failure simultaneously, has
Above-mentioned beneficial effect, details are not described herein.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of application for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow chart of the method for early warning of traction converter failure provided by the embodiment of the present application；
Fig. 2 is the flow chart of the method for early warning of another kind traction converter failure provided by the embodiment of the present application；
Fig. 3 is the flow chart of the method for early warning of another traction converter failure provided by the embodiment of the present application；
Fig. 4 is a kind of structural block diagram of the early warning system of traction converter failure provided by the embodiment of the present application；
Fig. 5 is a kind of practical process signal of the early warning system of traction converter failure provided by the embodiment of the present application
Figure.
Specific embodiment
The core of the application is to provide the method for early warning and system of a kind of traction converter failure, drives by using data
Dynamic model pre-warning method, on the one hand accuracy with higher, on the other hand can excavate from mass data and be not easy to examine
The stealth of the failure and device felt contacts, and improves the service life of traction convertor, reduces operation expense, and is right
Stability, safety, the reliability for promoting train play facilitation.
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
All other embodiment obtained without making creative work, shall fall in the protection scope of this application.
Below in conjunction with Fig. 1, Fig. 1 is a kind of stream of the method for early warning of traction converter failure provided by the embodiment of the present application
Cheng Tu.
Itself specifically includes the following steps:
S101: extracted respectively from the traction convertor historical data and network history data of train obtain train rectification it is inverse
Become module operation data and network operation data, and is obtained using commutation inversion module operation data building EOVW nergy Index whole
Flow the data characteristics and fault message of inverter module；Wherein, fault message includes fault type；
It is whole that this step is intended to get the operation data of most important commutation inversion module and train in traction convertor
The network operation data of vehicle, and commutation inversion module is obtained using the operation data of commutation inversion module building EOVW nergy Index
Data characteristics and according to the collected fault message of the sensor for being mounted on commutation inversion module.
In actual operation, since the frequency of sensor acquisition commutation inversion module related data is higher, after some time it is possible to reach every
Second thousands of times, and possible per second of the frequency acquisition of network operation data is once, if simply by rectification adverser mould
The data of block are all associated with network operation data and carry out subsequent analysis, it is clear that it can expend a large amount of processing capacity, and by
Not all data are all representative in the collected data of sensor and use value, because commutation inversion module exists
Collected data generally have fluctuation under working condition, therefore it is thousands of to go out this by wavelet transformation come induction and conclusion
The data characteristics of a data, and qualitative fault message out is combined to use in subsequent analysis, it can quickly and effectively obtain
Representative characteristic, and it is conveniently associated with network operation data.
Wherein, EOVW (Energy of Variation Wavelet) nergy Index, that is, the wavelet decomposition energy adjusted
Value, juche idea and structure still rely on wavelet analysis, but are no longer simply to calculate when extracting signal characteristic index
Decomposition coefficient quadratic sum obtains energy value (characteristic quantity), but calculates the adjustment coefficient of variation of coefficient of wavelet decomposition as new spy
Sign amount.And the calculating for adjusting the coefficient of variation does not rely on the mean value and standard deviation of overall signal (containing failure) instead of, normally to believe
Number calculate adjustment mean value and standard deviation be standard carry out calculate.
And " small echo " is exactly small waveform in wavelet transformation, it is so-called " small " to refer to that it has Decay Rate, and is referred to as " wave "
Then refer to its fluctuation, the concussion form of amplitude alternate positive and negative.Compared with Fourier transformation, wavelet transformation is that the time is (empty
Between) analysis of the localization of frequency, it gradually carries out multi-scale refinement to signal (function) by flexible shift operations, is finally reached
High frequency treatment time subdivision, frequency subdivision at low frequency, can adapt to the requirement of time frequency signal analysis, automatically so as to focus on signal
Any details solves the difficulty of Fourier transformation encountered here.
Certainly, as this step how from processing obtain the data characteristics of commutation inversion module, can be used to small form
The wavelet transformation of analysis more in place, can choose such as the methods of maximum value, minimum value, mean value, variance, quantile also to mention
EOVW nergy Index is obtained, the specific EOVW nergy Index that distinct methods finally extract slightly has deviation, but still has
It is representative.And analysis time, the computational complexity difference of each analysis method, in a practical situation factory can be used according to each
The performance power of equipment provisioned in family comprehensively considers, and herein and is not specifically limited.
S102: data characteristics and network operation data are associated by the different automobile types of train, license number and time, obtained
Data after to association；
On the basis of S101, this step be intended to will the obtained data characteristics of analysis and the network operation data extracted by
It is associated on the basis of time shaft according to the vehicle of different trains, license number, because the traction installed on the train of different automobile types becomes
The diversified in specifications of stream device fixed identical, other may also be not quite similar to the equipment that traction convertor impacts.It is different simultaneously
Existing otherness causes different when the train of number but same vehicle may be manufactured because of the equipment part of same size
It influences.
The purpose of this step just allows for these difference sexual factors that may be present, will run number from commutation inversion module
It is associated, obtains convenient for subsequent place on a timeline according to the data characteristics and network operation data that are obtained after converting by analysis
Data after the association of reason.
It is of course possible to which existing difference sexual factor not only includes vehicle, license number, there is likely to be it in practical situations
Its influence factor obtains association results to be more accurate to obtain the smallest judgement of error in final two points of judgement and to tie
Fruit can be comprehensively considered and be selected according to the performance of each requirement combination on-board processing equipment using manufacturer, herein not
It is specifically limited.
S103: data after association are executed into clustering operation by different vehicles and fault type, obtain number before failure
According to abnormal time range, and to the abnormal data additional fault early warning label in time range；
On the basis of S102, this step is intended to data after association carrying out cluster point by different vehicle and fault type
Analysis to operate by clustering from searching out the abnormal data for including after association in data, and orients this on a timeline
Time range locating for abnormal data, while being abnormal data additional fault early warning label.
Wherein, clustering refers to that the set by physics or abstract object is grouped into the multiple classes being made of similar object
Analytic process is a kind of feature according to research object (sample or index), the method classified to it, it is intended to reduce research
The number of object.The purpose of clustering is the close things of property to be included into one kind, and the mode relied on is then between each index
The correlativity more or less having.
Wherein, the concept of cluster is involved, cluster is to sort data into different such a processes of class or cluster,
Therefore the object in the same cluster has very big similitude, and the object between different clusters has very big diversity.From statistical sight
Point sees that clustering is to simplify a kind of method of data by data modeling.Traditional Statistical Clustering Analysis analysis method includes system
Clustering procedure, addition method, dynamic state clustering, clustering ordered samples, has overlapping to cluster and fuzzy clustering etc. at decomposition method.It is equal using k-
Value, k-means scheduling algorithm clustering tool interpolated in many famous statistics analysis software packages, such as SPSS, SAS
Deng.
Say that cluster is equivalent to stealth mode from the angle of machine learning.Cluster is to search for the unsupervised learning process of cluster.With point
Class is different, and unsupervised learning does not depend on class predetermined or the training example with class label, needs by cluster learning algorithm certainly
Dynamic to determine label, i.e., cluster is observation type study, rather than the study of example.
That is the clustering analysis that is a kind of exploration, during classification, people need not provide a classification in advance
Standard, clustering can classify automatically from sample data.The difference of clustering institute application method, usually
It can obtain different conclusions.Different researchers carry out clustering for same group of data, and obtained cluster numbers may not be consistent.
From the point of view of practical application, clustering is one of main task of data mining, and clustering being capable of conduct
One independent tool obtains the distribution situation of data, observes the feature of each cluster data, concentrates to the collection cooperation that specifically clusters
It further analyzes, clustering is also used as the pre-treatment step of other algorithms (such as classification and qualitative inductive algorithm).
The application is intended to, from analysis obtains abnormal data in data after association, and pass through number by using clustering
Correlation between searches out from when occurring as soon as and eventually leading to the sign variable of failure, summarizes abnormal data set,
Fault pre-alarming is carried out by the time range of the abnormal data set on a timeline.Clustering method as described above has
Very much, most suitable algorithm can be selected according to the actual situation, because different clustering algorithms may obtain identical data
Different conclusion, it may be necessary to carry out test and search out most suitable clustering algorithm, herein and be not specifically limited, only needing can
The time range of the abnormal data is obtained according to the clustering algorithm.
S104: the abnormal data for being attached with fault pre-alarming label and normal data are introduced into random forest grader and changed
Generation training, until the classification accuracy of all random forest graders reaches threshold value, random forest grader after being trained；
Wherein, random forest grader constructs to obtain according to failure mode and dichotomy；
On the basis of S103, what this step was intended to handle by clustering is attached with fault pre-alarming label
The related normal data of abnormal data set introduces random forest grader together, which is to utilize historical data point
Analyse what obtained all fault type combination dichotomies being likely to occur were established, it is therefore intended that classify using the random forest
Device is successively determined abnormal data and normal data one by one by all fault types, determines that by learning abnormal data
Which kind of fault type is a little data be, study normal data determines those data for normal data, finally makes all random forests point
The classification accuracy of class device reaches threshold value, and it is abnormal data which, which can accurately be determined in train actual operating data,
And provide the specific fault type of the abnormal data.
Random forest is a kind of machine learning model that comparison is new.Classical machine learning model is neural network, nerve
Neural network forecast is accurate, but calculation amount is very big.The algorithm of last century the eighties Breiman et al. invention classification tree, by anti-
Multiple two divided datas are classified or are returned, and calculation amount can be made to substantially reduce.Breiman in 2001 is combined into classification tree random gloomy
Woods is randomized in the use of variable (column) and the use of data (row), generates many classification trees, then pooled classification tree
Result.Random forest improves precision of prediction under the premise of operand does not significantly improve.And random forest is to polynary total
Linear insensitive, it is more steady to missing data and nonequilibrium data to cause, and can predict that up to thousands of are explained well
The effect of variable.
Random forest as its name suggests, is to establish a forest with random manner, has many decision tree groups inside forest
At being not associated between each decision tree of random forest.After obtaining forest, when there is a new input sample
Into when, just allow each decision tree in forest once to be judged respectively, look at which this sample should belong to
Class (for sorting algorithm) then looks at which kind of at most, just predicts that this sample is that is a kind of by selection.
S105: train actual operating data is analyzed using random forest grader after training, if not passing through training
Random forest grader afterwards then sends warning information by preset path.
On the basis of S104, this step is intended to obtain using trained random forest grader in train actual motion
To traction convertor related data and network data carry out classification judgement, if not passed through by the judgement of all classifiers
Preset path sends warning information.Because each fault type of random forest grader all establishes a decision tree, as long as
One decision tree, which does not pass through just to represent certainly, to be existed abnormal, and only all decision trees pass through can just think that equipment is in just
Normal state.
The manifestation mode of the preset path is varied, can be led to by the network that train content is connected to according to default network
Road is transmitted, and also be can use wireless communication module while being sent to the display screen conduct backup of drivers' cab, to prevent which bar line
Road breaks down and does not receive warning message etc., can be comprehensively considered and be selected according to the actual situation, herein not
It is specifically limited.
Further, it removes using random forest grader as training pattern, GBDT also can be used, Xgboost is equivalent
Classifier as fruit, equally, determine the selection of abnormal data time range is that hierarchical clustering is combined with K-means cluster
Method also can be used other clustering algorithms or carry out analysis using related coefficient and achieve the goal, herein and is not specifically limited.
Further, due to sensor short-duration failure that may be present, the data for collecting is caused such as " hair occur
Thorn " " spike " etc. is abnormal, so that relevant Decision tree is alarmed, aiming at the problem that this be likely to occur, if by each alarm signal
A possibility that number being all directly sent to drivers' cab, erroneous judgement may be greatly increased, it can take connect whether in the given time at this time
It is continuous to receive identical and alarm signal for predetermined quantity to determine whether really to occur exception, rather than judge by accident.Certainly, this is predetermined
Time can with sets itself because different faults type there is the probability judged by accident may be different, specific predetermined quantity can also be with
Sets itself herein and is not specifically limited with meeting different particular/special requirements and standard using manufacturer.
Further, can also all warning information received be recorded and is saved, by all relevant informations
Detailed warning information log is recorded and generates, all in order to carry out subsequent analysis and trace to make according to the warning information log
With.
Based on the above-mentioned technical proposal, the method for early warning of a kind of traction converter failure provided by the embodiments of the present application, first
It is handled for the commutation inversion module historical data situation inconsistent with network data acquisition frequency, and after processing
Data are associated, and are and then divided the data after association by vehicle and fault type information and are passed sequentially through two kinds of clusters and calculate
Method to obtain the time range of abnormal data before failure occurs, and is abnormal data additional fault early warning label, finally will be normal
Data and abnormal data are uniformly put into the random forest classification of fault type generation and carry out two points of judgements.The method for early warning passes through
Using the model pre-warning method of data-driven, on the one hand accuracy with higher, on the other hand can dig from mass data
Excavate it is not noticeable to failure and the stealth of device contact, improve the service life of traction convertor, reduce operation dimension
Cost is protected, and facilitation is played to the stability, safety, the reliability that promote train.
It is the method for early warning of another kind traction converter failure provided by the embodiment of the present application below in conjunction with Fig. 2, Fig. 2
Flow chart.
The present embodiment is to obtain commutation inversion module for how constructing EOVW nergy Index in S101 in a upper embodiment
Data characteristics and fault message and S103 how to carry out one that hierarchical clustering operation is made it is specific limit, Qi Tabu
Suddenly it is substantially the same with a upper embodiment, same section can be found in an embodiment relevant portion, and details are not described herein.
Itself specifically includes the following steps:
S201: commutation inversion module operation data is passed through into wavelet transformation in seconds and extracts each sensor information
EOVW nergy Index；
S202: using EOVW nergy Index as the data characteristics of the commutation inversion module of current second, and fault message is obtained；
The S201 and S202 of the present embodiment are intended to using the preferable wavelet transformation of waveform analysis result to minor swing to whole
Stream inverter module operation data is extracted obtains the EOVW nergy Index of each sensor information in seconds, and by the EOVW energy
Data characteristics of the index as the commutation inversion module of current second, and obtain from each sensor comprising qualitative fault type out
Fault message.
S203: data characteristics and network operation data are associated by the different automobile types of train, license number and time, obtained
Data after to association；
This step is identical as S102, and related content may refer to the description content of S102, and details are not described herein.
S204: data after association are subjected to hierarchical clustering processing by different vehicles and different fault types, obtain layer
Data after secondary clustering processing；Wherein, data include cluster information and center position information after hierarchical clustering processing；
S205: K-means clustering processing is carried out using cluster information and center position information, determines data exception before failure
Time range.
The S204 and S205 of the present embodiment, first with hierarchical clustering by data after association by different vehicle and different
Fault type is handled, obtain include cluster information and center position information processing after data, believe followed by cluster
Breath and center position information carry out K-means clustering processing, determine the time range of data exception before failure.
Hierarchy clustering method (Hierarchical Clustering) is exactly by carrying out to data set according to some way
Hierachical decomposition, until meeting certain condition.According to the difference of principle of classification, two methods of cohesion and division can be divided into.
Wherein, the hierarchical clustering of cohesion is a kind of bottom-up strategy, first using each object as a cluster, then
Merging these clusters is increasing cluster, until all objects are all in a cluster or some finish condition is expired
Foot, most hierarchy clustering methods belong to this kind, they are different in the definition of similarity between cluster；Division
On the contrary, using top-down strategy, all objects are placed in the same cluster first by it for hierarchical clustering and the hierarchical clustering of cohesion
In, it is then gradually subdivided into smaller and smaller cluster, until each object self-contained cluster, or has reached some termination condition.
K-means algorithm is hard clustering algorithm, is the representative of the typically objective function clustering method based on prototype, it is
Data point obtains the tune of interative computation using the method that function seeks extreme value to certain objective function of distance as optimization of prototype
Whole rule.For K-means algorithm using Euclidean distance as similarity measure, it is to ask corresponding a certain initial cluster center vector optimal
Classification, so that evaluation index is minimum, the algorithm is using error sum of squares criterion function as clustering criteria function.
K-means algorithm is the evaluation index very typically based on the clustering algorithm of distance, using distance as similitude,
Think that the distance of two objects is closer, similarity is bigger.The algorithm think cluster by forming apart from close object,
Therefore handle obtains compact and independent cluster as final goal.The selection of k initial classes cluster centre point has cluster result
Large effect, because in the algorithm first step being center of the random any k object of selection as initial clustering, initially
Ground represents a cluster.The algorithm concentrates remaining each object to data in each iteration, according to itself and each cluster center
Each object is assigned to nearest cluster by distance again.After having investigated all data objects, an iteration operation is completed, and new is poly-
Class center is computed.If before and after an iteration, value there is no variation, illustrate that algorithm has been restrained.
Preliminary clustering is carried out by hierarchical clustering first to handle, and obtains using convenient for subsequent K-means algorithm
Cluster information and center position information, finally using the K-means algorithm summarize it is intended to abnormal data set.
It is the method for early warning of another traction converter failure provided by the embodiment of the present application below in conjunction with Fig. 3, Fig. 3
Flow chart.
The present embodiment is increased a part of content on the basis of the above embodiments, other steps and a upper embodiment
It is substantially the same, same section can be found in an embodiment relevant portion, and details are not described herein.
Itself specifically includes the following steps:
S301: train actual operating data is analyzed using random forest grader after training, if not passing through training
Random forest grader afterwards then sends warning information by preset path；
S302: whether judgement repeatedly receives identical warning information within a preset time；
S303: it is abnormal to determine that train exists, and train is had into abnormal judgement result and is sent to train driver；
S304: determine that train there is no exception, does not execute any operation；
S305: all warning information received are recorded and saved, and generate warning information log, so as to according to pre-
Alert information log carries out subsequent analysis and traces.
Based on the above-mentioned technical proposal, the method for early warning of a kind of traction converter failure provided by the embodiments of the present application utilizes
Wavelet transformation obtains representative EOVW nergy Index, and to be formed by data characteristics inconsistent to solve information collection frequency
The case where, and the data after association are passed sequentially through into two kinds of clustering algorithms of hierarchical clustering and K-means, to be calculated using two kinds of clusters
The advantages of method, more accurately obtains the time range of abnormal data before failure occurs, and finally normal data and abnormal data are united
One is put into two points of judgements of progress in the random forest classification generated with fault type.The method for early warning is by using data-driven
Model pre-warning method, on the one hand accuracy with higher, on the other hand can be excavated from mass data it is not noticeable to
Failure and the stealth of device contact, improve the service life of traction convertor, reduce operation expense, and to promotion
Stability, safety, the reliability of train play facilitation.
Because situation is complicated, it can not enumerate and be illustrated, those skilled in the art should be able to recognize more the application
The basic skills principle combination actual conditions of offer may exist many examples, in the case where not paying enough creative works,
It should within the scope of protection of this application.
Fig. 4 is referred to below, and Fig. 4 is a kind of early warning system of traction converter failure provided by the embodiment of the present application
Structural block diagram.
The early warning system may include:
Operation data obtains and analytical unit 100, for the traction convertor historical data and network history number from train
The commutation inversion module operation data and network operation data of train are obtained according to extracting respectively, and is run using commutation inversion module
Data building EOVW nergy Index obtains the data characteristics and fault message of commutation inversion module；Wherein, fault message includes event
Hinder type；
Data associating unit 200, for by data characteristics and network operation data by the different automobile types of train, license number and
Time is associated, data after being associated with；
Clustering operation execution unit 300 is executed by different vehicles and fault type for data after being associated with and is gathered
Alanysis operation, obtains the time range of data exception before failure, and to the abnormal data additional fault early warning in time range
Label；
Classifier training unit 400, for will be attached with fault pre-alarming label abnormal data and normal data introduce with
Machine forest classified device is iterated training, until the classification accuracy of all random forest graders reaches threshold value, is instructed
Random forest grader after white silk；Wherein, random forest grader constructs to obtain according to failure mode and dichotomy；
Physical fault predicting unit 500, for using training after random forest grader to train actual operating data into
Row analysis, if not sending warning information by preset path by random forest grader after training.
Wherein, operation data obtains and analytical unit 100 includes:
Wavelet transformation subelement, for extracting commutation inversion module operation data by wavelet transformation in seconds
The EOVW nergy Index of each sensor information；
Subelement is handled, for using EOVW nergy Index as the data characteristics of the commutation inversion module of current second, and
To fault message.
Wherein, clustering operation execution unit 300 includes:
Hierarchical clustering handles subelement, carries out layer by different vehicles and different fault types for data after being associated with
Secondary clustering processing obtains data after hierarchical clustering processing；Wherein, data include cluster information and center point after hierarchical clustering processing
Confidence breath；
K-means clustering processing subelement, for being carried out at K-means cluster using cluster information and center position information
Reason, determines the time range of data exception before failure.
Wherein, physical fault predicting unit 500 can also include:
Expert Rules determine subelement, for judging whether repeatedly receive identical warning information within a preset time；
Expert Rules determine for determining that it is abnormal that train exists, and train is existed to abnormal judgement knot by subelement
Fruit is sent to train driver.
Further, which can also include:
Storage unit is recorded, for recording and saving all warning information received, generates warning information day
Will, to carry out subsequent analysis according to warning information log and to trace.
The above each unit can be applied in one below specific concrete instance, may refer to lower Fig. 5, and Fig. 5 is this
A kind of practical flow diagram of the early warning system of traction converter failure provided by applying:
Extract train fortune from train traction converter module historical data and train network historical data respectively first
Data when row.Since the sample frequency of both sides data is inconsistent, in order to be associated with both sides data, need after data extraction to change
It flows device module data and carries out feature extraction operation, selection is that wavelet transformation extracts EOVW of each sensor signal in one second
Nergy Index.Train network data and converter module characteristic are pressed into vehicle, license number and time after completion feature extraction
It associates.Then by vehicle and fault type information division data and by using hierarchical clustering and K-means clustering algorithm
Cluster obtains the time range of abnormal data before failure twice.It is pre- that failure finally is stamped to corresponding abnormal data by fault type
After alert label, by vehicle by it is all normally merges with abnormal data after be put into random forest grader carry out abnormal data with normally
Two classification of data determine, and are iterated training until accuracy rate reaches requirement by label.
Since train traction current transformer related sensor sample frequency will be much higher than train network data sampling frequency, consider
The time for needing abundance to early warning allows driver to make a decision, and traction convertor data are that unit passes through small echo with " second " by the present invention
It is sent to train network end after the EOVW nergy Index of each sensor signal of transformation extraction, is inputted after being associated with network end data
Trained two classifier of vehicle-mounted random forest carries out data classification judgement.Due to for each failure have one it is random gloomy
Two disaggregated model of woods, therefore each prediction data needs to carry out prediction judgement by two all disaggregated models.If wherein institute
There is classifier result to show normally, then it is assumed that current traction convertor is working properly, conversely, if wherein there are one or more classification
Device prediction is out of order, then needs to further determine that by Expert Rules.Since the data value of input model may be extremely because passing
Sensor problem causes data burr, spike etc. occur, therefore expert generally takes the mode of " same fault is continuously quoted repeatedly "
Carry out abnormal make a definite diagnosis.
It is had the following advantages in the above manner, can gather around:
(1) make drag current transformer module exception judgment threshold more scientific and reasonable.Based on data-driven, by history number
According to analysis modeling is carried out, the judgment rule of failure and unit exception is obtained, the judgement of Heuristics given threshold is leaned in comparison at present
Mode is more scientific and reasonable；
(2) it solves since drag current transformer module data sampling frequency is excessively high, data waveform changes too fast, fault moment
And abnormal time data is difficult to carry out effectively analyzing the problems such as making a definite diagnosis before failure.The present invention is using association drag current transformer module
The method of data and train network data, model can comprehensively consider influence of the both sides data to failure.When converter module data
When off-note is unobvious, model can influence of the Main Analysis consideration lower network data of sample frequency to failure；Conversely, mould
Type then can mainly consider drag current transformer module data characteristics；
(3) in terms of modeling, the Annual distribution of abnormal data before failure is determined using clustering algorithm, and (e.g. failure is former
There are unusual fluctuations in second to several divided datas), determine that abnormal data range is more accurate and reliable by Heuristics compared to traditional.
Each embodiment is described in a progressive manner in specification, the highlights of each of the examples are with other realities
The difference of example is applied, the same or similar parts in each embodiment may refer to each other.For device disclosed in embodiment
Speech, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method part illustration
?.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond scope of the present application.
Specific examples are used herein to illustrate the principle and implementation manner of the present application, and above embodiments are said
It is bright to be merely used to help understand the present processes and its core concept.It should be pointed out that for the ordinary skill of the art
For personnel, under the premise of not departing from the application principle, can also to the application, some improvement and modification can also be carried out, these improvement
It is also fallen into the protection scope of the claim of this application with modification.
It should also be noted that, in the present specification, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also other elements including being not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or equipment for including element.
Claims (10)
1. a kind of method for early warning of traction converter failure characterized by comprising
It is extracted respectively from the traction convertor historical data and network history data of train and obtains the commutation inversion mould of the train
Block operation data and network operation data, and using the commutation inversion module operation data building EOVW nergy Index obtain it is whole
Flow the data characteristics and fault message of inverter module；Wherein, the fault message includes fault type；
The data characteristics and the network operation data are associated by the different automobile types of train, license number and time, obtained
Data after to association；
Data after the association are executed into clustering operation by different vehicles and the fault type, obtain data before failure
Abnormal time range, and to the abnormal data additional fault early warning label in the time range；
The abnormal data for being attached with the fault pre-alarming label and normal data are introduced into random forest grader and are iterated instruction
Practice, until the classification accuracy of all random forest graders reaches threshold value, random forest grader after being trained；Its
In, the random forest grader constructs to obtain according to failure mode and dichotomy；
Train actual operating data is analyzed using random forest grader after the training, if after not by the training
Random forest grader then sends warning information by preset path.
2. method for early warning according to claim 1, which is characterized in that constructed using the commutation inversion module operation data
EOVW nergy Index obtains the data characteristics and fault message of commutation inversion module, comprising:
The commutation inversion module operation data is passed through into the EOVW that wavelet transformation extracts each sensor information in seconds
Nergy Index；
Using the EOVW nergy Index as the data characteristics of the commutation inversion module of current second, and obtain the fault message.
3. method for early warning according to claim 1 or 2, which is characterized in that data after the association are pressed different vehicles
Clustering operation is executed with the fault type, obtains the time range of data exception before failure, comprising:
Data after the association are subjected to hierarchical clustering processing by different vehicles and different fault types, obtain hierarchical clustering
Data after processing；Wherein, data include cluster information and center position information after the hierarchical clustering processing；
K-means clustering processing is carried out using the cluster information and the center position information, determines data exception before failure
Time range.
4. method for early warning according to claim 3, which is characterized in that after sending warning information by preset path,
Further include:
Whether judgement repeatedly receives identical warning information within a preset time；
If so, it is abnormal to determine that the train exists, and be there is into abnormal judgement result in the train and be sent to train driving
Member.
5. method for early warning according to claim 4, which is characterized in that further include；
All warning information received are recorded and saved, warning information log is generated, to be believed according to the early warning
Breath log carries out subsequent analysis and traces.
6. a kind of early warning system of traction converter failure characterized by comprising
Operation data obtains and analytical unit, for mentioning respectively from the traction convertor historical data and network history data of train
The commutation inversion module operation data and network operation data of the train are obtained, and is run using the commutation inversion module
Data building EOVW nergy Index obtains the data characteristics and fault message of commutation inversion module；Wherein, the failure information package
Include fault type；
Data associating unit, for by the data characteristics and the network operation data by the different automobile types of train, license number with
And the time is associated, data after being associated with；
Clustering operation execution unit, it is poly- for executing data after the association by different vehicles and the fault type
Alanysis operation, obtains the time range of data exception before failure, and to the abnormal data additional fault in the time range
Early warning label；
Classifier training unit, abnormal data and normal data for that will be attached with the fault pre-alarming label introduce random gloomy
Woods classifier is iterated training, until the classification accuracy of all random forest graders reaches threshold value, after being trained
Random forest grader；Wherein, the random forest grader constructs to obtain according to failure mode and dichotomy；
Physical fault predicting unit, for being divided using random forest grader after the training train actual operating data
Analysis, if not sending warning information by preset path by random forest grader after the training.
7. early warning system according to claim 6, which is characterized in that the operation data obtains and analytical unit includes:
Wavelet transformation subelement, for extracting the commutation inversion module operation data by wavelet transformation in seconds
The EOVW nergy Index of each sensor information；
Subelement is handled, for using the EOVW nergy Index as the data characteristics of the commutation inversion module of current second, and
To the fault message.
8. early warning system according to claim 6 or 7, which is characterized in that the clustering operation execution unit includes:
Hierarchical clustering handles subelement, for data after the association to be carried out layer by different vehicles and different fault types
Secondary clustering processing obtains data after hierarchical clustering processing；Wherein, data include cluster information and center after the hierarchical clustering processing
Dot position information；
K-means clustering processing subelement, it is poly- for carrying out K-means using the cluster information and the center position information
Class processing, determines the time range of data exception before failure.
9. early warning system according to claim 8, which is characterized in that the physical fault predicting unit further include:
Expert Rules determine subelement, for judging whether repeatedly receive identical warning information within a preset time；
Expert Rules determine to sentence for determining that it is abnormal that the train exists, and by the train in the presence of abnormal by subelement
Determine result and is sent to train driver.
10. early warning system according to claim 9, which is characterized in that further include:
Storage unit is recorded, for all warning information received to be recorded and saved, generation warning information log, with
Just subsequent analysis is carried out according to the warning information log and traced.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711094900.7A CN109754110B (en) | 2017-11-03 | 2017-11-03 | Early warning method and system for traction converter faults |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711094900.7A CN109754110B (en) | 2017-11-03 | 2017-11-03 | Early warning method and system for traction converter faults |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109754110A true CN109754110A (en) | 2019-05-14 |
| CN109754110B CN109754110B (en) | 2023-07-11 |
Family
ID=66402281
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711094900.7A Active CN109754110B (en) | 2017-11-03 | 2017-11-03 | Early warning method and system for traction converter faults |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109754110B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110991472A (en) * | 2019-08-01 | 2020-04-10 | 南京航空航天大学 | Micro fault diagnosis method for high-speed train traction system |
| CN110991240A (en) * | 2019-10-31 | 2020-04-10 | 湖南世优电力科技股份有限公司 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
| CN111060844A (en) * | 2019-12-09 | 2020-04-24 | 南京航空航天大学 | Interturn short-circuit fault diagnosis method and device for high-speed train traction transmission system |
| CN111259947A (en) * | 2020-01-13 | 2020-06-09 | 国网浙江省电力有限公司信息通信分公司 | Power system fault early warning method and system based on multi-mode learning |
| CN111985651A (en) * | 2019-05-22 | 2020-11-24 | 中国移动通信集团福建有限公司 | Operation and maintenance method and device for business system |
| CN112801315A (en) * | 2021-01-28 | 2021-05-14 | 国网河北省电力有限公司电力科学研究院 | State diagnosis method and device for power secondary equipment and terminal |
| CN112819386A (en) * | 2021-03-05 | 2021-05-18 | 中国人民解放军国防科技大学 | Method, system and storage medium for generating time series data with abnormity |
| WO2021097604A1 (en) * | 2019-11-18 | 2021-05-27 | 株洲中车时代电气股份有限公司 | Multi-information fusion-based fault early warning method and device for converter |
| CN113034732A (en) * | 2021-03-29 | 2021-06-25 | 南京格物智能科技有限公司 | Method and device for diagnosing blockage of traction converter filter screen |
| CN113071537A (en) * | 2020-01-06 | 2021-07-06 | 株洲中车时代电气股份有限公司 | Early warning method, device and system for cooling system |
| CN113809786A (en) * | 2020-07-23 | 2021-12-17 | 广东毓秀科技有限公司 | Method for predicting faults of UPS rectification and inversion module through big data |
| CN114492588A (en) * | 2021-12-29 | 2022-05-13 | 中国大唐集团科学技术研究总院有限公司华东电力试验研究院 | Method, system, equipment and storage medium for identifying faults of auxiliary equipment of thermal power plant |
| CN115792476A (en) * | 2023-02-06 | 2023-03-14 | 石家庄科林电气股份有限公司 | Charging pile rectification module abnormity early warning method and device, terminal and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221655A (en) * | 2011-06-16 | 2011-10-19 | 河南省电力公司济源供电公司 | Random-forest-model-based power transformer fault diagnosis method |
| CN103901882A (en) * | 2014-04-15 | 2014-07-02 | 北京交通大学 | Online monitoring fault diagnosis system and method of train power system |
| CN104236933A (en) * | 2013-06-13 | 2014-12-24 | 同济大学 | Hidden fault danger warning method for train hauling system |
| CN106338406A (en) * | 2016-10-19 | 2017-01-18 | 北京交通大学 | On-line monitoring and fault early-warning system and method for traction electric transmission system of train |
| CN106338664A (en) * | 2016-08-11 | 2017-01-18 | 中车株洲电力机车研究所有限公司 | Train current transformer fault diagnosis method and device |
-
2017
- 2017-11-03 CN CN201711094900.7A patent/CN109754110B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221655A (en) * | 2011-06-16 | 2011-10-19 | 河南省电力公司济源供电公司 | Random-forest-model-based power transformer fault diagnosis method |
| CN104236933A (en) * | 2013-06-13 | 2014-12-24 | 同济大学 | Hidden fault danger warning method for train hauling system |
| CN103901882A (en) * | 2014-04-15 | 2014-07-02 | 北京交通大学 | Online monitoring fault diagnosis system and method of train power system |
| CN106338664A (en) * | 2016-08-11 | 2017-01-18 | 中车株洲电力机车研究所有限公司 | Train current transformer fault diagnosis method and device |
| CN106338406A (en) * | 2016-10-19 | 2017-01-18 | 北京交通大学 | On-line monitoring and fault early-warning system and method for traction electric transmission system of train |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111985651A (en) * | 2019-05-22 | 2020-11-24 | 中国移动通信集团福建有限公司 | Operation and maintenance method and device for business system |
| CN110991472A (en) * | 2019-08-01 | 2020-04-10 | 南京航空航天大学 | Micro fault diagnosis method for high-speed train traction system |
| CN110991472B (en) * | 2019-08-01 | 2023-10-03 | 南京航空航天大学 | Method for diagnosing minor faults of high-speed train traction system |
| CN110991240B (en) * | 2019-10-31 | 2023-06-09 | 湖南世优电力科技股份有限公司 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
| CN110991240A (en) * | 2019-10-31 | 2020-04-10 | 湖南世优电力科技股份有限公司 | Single-phase fault identification and classification method for low-voltage neutral point ungrounded system |
| WO2021097604A1 (en) * | 2019-11-18 | 2021-05-27 | 株洲中车时代电气股份有限公司 | Multi-information fusion-based fault early warning method and device for converter |
| CN111060844A (en) * | 2019-12-09 | 2020-04-24 | 南京航空航天大学 | Interturn short-circuit fault diagnosis method and device for high-speed train traction transmission system |
| CN113071537A (en) * | 2020-01-06 | 2021-07-06 | 株洲中车时代电气股份有限公司 | Early warning method, device and system for cooling system |
| CN111259947A (en) * | 2020-01-13 | 2020-06-09 | 国网浙江省电力有限公司信息通信分公司 | Power system fault early warning method and system based on multi-mode learning |
| CN113809786A (en) * | 2020-07-23 | 2021-12-17 | 广东毓秀科技有限公司 | Method for predicting faults of UPS rectification and inversion module through big data |
| CN112801315A (en) * | 2021-01-28 | 2021-05-14 | 国网河北省电力有限公司电力科学研究院 | State diagnosis method and device for power secondary equipment and terminal |
| CN112819386A (en) * | 2021-03-05 | 2021-05-18 | 中国人民解放军国防科技大学 | Method, system and storage medium for generating time series data with abnormity |
| CN113034732A (en) * | 2021-03-29 | 2021-06-25 | 南京格物智能科技有限公司 | Method and device for diagnosing blockage of traction converter filter screen |
| CN113034732B (en) * | 2021-03-29 | 2022-12-27 | 南京格物智能科技有限公司 | Method and device for diagnosing blockage of filter screen of traction converter |
| CN114492588A (en) * | 2021-12-29 | 2022-05-13 | 中国大唐集团科学技术研究总院有限公司华东电力试验研究院 | Method, system, equipment and storage medium for identifying faults of auxiliary equipment of thermal power plant |
| CN115792476A (en) * | 2023-02-06 | 2023-03-14 | 石家庄科林电气股份有限公司 | Charging pile rectification module abnormity early warning method and device, terminal and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109754110B (en) | 2023-07-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109754110A (en) | A kind of method for early warning and system of traction converter failure | |
| CN102765643B (en) | Elevator fault diagnosis and early-warning method based on data drive | |
| CN104091070B (en) | Rail transit fault diagnosis method and system based on time series analysis | |
| CN103699698B (en) | A kind of being based on improves Bayesian rail transit fault identification method and system | |
| CN104933869B (en) | Anti-noise type satellite positioning data-based method for identifying vehicle violation behaviors | |
| CN105303197B (en) | A kind of vehicle follow the bus safety automation appraisal procedure based on machine learning | |
| CN104598813A (en) | Computer intrusion detection method based on integrated study and semi-supervised SVM | |
| CN108248610B (en) | It is a kind of adaptively divert one's attention driving behavior monitoring and tampering devic | |
| CN103760901A (en) | Rail transit fault identification method based on association rule classifier | |
| CN104050361A (en) | Intelligent analysis early warning method for dangerousness tendency of prison persons serving sentences | |
| CN108682149A (en) | The linear causation analysis method in highway accident stain section based on binary logistic regression | |
| CN104951764A (en) | Identification method for behaviors of high-speed vehicle based on secondary spectrum clustering and HMM (Hidden Markov Model)-RF (Random Forest) hybrid model | |
| CN111914687B (en) | Method for actively identifying accidents based on Internet of vehicles | |
| CN109993966A (en) | A kind of method and device of building user portrait | |
| Verwer et al. | Learning driving behavior by timed syntactic pattern recognition | |
| CN110562261A (en) | Method for detecting risk level of driver based on Markov model | |
| CN106384507A (en) | Travel time real-time estimation method based on sparse detector | |
| Paul et al. | Prediction of road accident and severity of Bangladesh applying machine learning techniques | |
| He et al. | An individual driving behavior portrait approach for professional driver of HDVs with naturalistic driving data | |
| CN110596490A (en) | Intelligent detection method for railway turnout fault | |
| Zhang et al. | Car-following behavior model learning using timed automata | |
| Yang et al. | Developing a new real-time traffic safety management framework for urban expressways utilizing reinforcement learning tree | |
| CN109035763A (en) | Expressway traffic accident primary and secondary based on C4.5 is because of analysis and accident pattern judgment method | |
| CN115860582B (en) | Impact risk intelligent early warning method based on self-adaptive lifting algorithm | |
| CN115192026B (en) | Tunnel driving load monitoring method and terminal |
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 |