CN108860211A - A kind of wrong report recognition methods and device based on shaft temperature sensor - Google Patents
A kind of wrong report recognition methods and device based on shaft temperature sensor Download PDFInfo
- Publication number
- CN108860211A CN108860211A CN201810515666.9A CN201810515666A CN108860211A CN 108860211 A CN108860211 A CN 108860211A CN 201810515666 A CN201810515666 A CN 201810515666A CN 108860211 A CN108860211 A CN 108860211A
- Authority
- CN
- China
- Prior art keywords
- temperature
- axis
- shaft
- shaft end
- data
- 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
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
- B61K9/04—Detectors for indicating the overheating of axle bearings and the like, e.g. associated with the brake system for applying the brakes in case of a fault
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L15/00—Indicators provided on the vehicle or vehicle train for signalling purposes ; On-board control or communication systems
- B61L15/0072—On-board train data handling
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L15/00—Indicators provided on the vehicle or vehicle train for signalling purposes ; On-board control or communication systems
- B61L15/0081—On-board diagnosis or maintenance
Abstract
The embodiment of the present invention provides a kind of wrong report recognition methods based on shaft temperature sensor and device.The method includes：The current shaft end temperature and axis temperature related data that the shaft temperature sensor of target axle position reports when obtaining train operation；The axis temperature related data is input to the corresponding axis temperature prediction model of the shaft temperature sensor, obtains the corresponding shaft end temperature prediction value of the axis temperature related data, the axis temperature prediction model is obtained by neural metwork training；If judgement knows that the difference of the current shaft end temperature and the shaft end temperature prediction value meets default wrong report recognition rule, the current shaft end temperature is identified as wrong report data.The method of the embodiment of the present invention can effectively identify that shaft temperature sensor reports situation by mistake, provide safeguard for PHM system to train operation state real time monitoring.
Description
Technical field
The present embodiments relate to train fault detection field more particularly to a kind of wrong report identifications based on shaft temperature sensor
Method and device.
Background technique
During high-speed cruising, the temperature of each bearing in travelled by vehicle part can be increased constantly railroad train, when axis temperature mistake
When high, it will cause hot axis, cut phenomena such as axis, seriously affect safety of railway traffic.Currently available technology is using resistance
Formula temperature sensor is welded on the acquisition of the enterprising trip temperature data of each bearing, confirms axis by shaft temperature sensor alarm system
Whether temperature is higher than hot axis threshold value, that is, confirms hot axis accident, to trigger alarm, prompts train operator's axis temperature exceeded, with season column
Vehicle limiting operation.
But shaft temperature sensor and welding position is caused to loosen when shaft temperature sensor itself damage or because train shakes
When, the temperature data that will cause shaft temperature sensor is abnormal, this may trigger hot axis alarm system, so as to cause unnecessary
Train speed limit operation.
In order to accurately grasp the actual temperature of high-speed rail, EMU operational process middle (center) bearing, avoid sensing due to axis temperature
Device failure or exception lead to the unnecessary reduction of speed of train, influence the high efficiency of transport, and there is an urgent need to one kind to be based on shaft temperature sensor
Wrong report recognition methods, exclude the improper axis temperature data (i.e. wrong report data) of the shaft temperature sensor of non-faulting state, guarantee
Train safety is run at high speed, is arrived punctually at the destination.
Summary of the invention
In view of the problems of the existing technology, the embodiment of the present invention provides a kind of wrong report identification side based on shaft temperature sensor
Method and device carry out big data mining analysis based on vehicle-mounted WTD equipment, can effectively identify that shaft temperature sensor reports situation by mistake, be
PHM system provides safeguard train operation state real time monitoring.
In a first aspect, the embodiment of the present invention provides a kind of wrong report recognition methods based on shaft temperature sensor, including：
The current shaft end temperature and axis temperature related data that the shaft temperature sensor of target axle position reports when obtaining train operation；
The axis temperature related data is input to the corresponding axis temperature prediction model of the shaft temperature sensor, obtains the axis temperature
The corresponding shaft end temperature prediction value of related data, the axis temperature prediction model are obtained by neural metwork training；
If judgement knows that the difference of the current shaft end temperature and the shaft end temperature prediction value meets default wrong report identification
The current shaft end temperature is then identified as wrong report data by rule.
Second aspect, the embodiment of the present invention provide a kind of wrong report identification device based on shaft temperature sensor, including：
Data acquisition module, the current shaft end temperature that the shaft temperature sensor of target axle position reports when for obtaining train operation
With axis temperature related data；
Model prediction module is predicted for the axis temperature related data to be input to the corresponding axis temperature of the shaft temperature sensor
Model, obtains the corresponding shaft end temperature prediction value of the axis temperature related data, and the axis temperature prediction model is instructed by neural network
It gets；
Identification module is reported by mistake, if for judging to know the difference of the current shaft end temperature and the shaft end temperature prediction value
Meet default wrong report recognition rule, then the current shaft end temperature is identified as wrong report data.
The third aspect, the embodiment of the invention provides a kind of electronic equipment, including：
At least one processor；And
At least one processor being connect with the processor communication, wherein：
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to refer to
Enable the wrong report recognition methods and its any optional reality being able to carry out described in first aspect of the embodiment of the present invention based on shaft temperature sensor
Apply method described in example.
Fourth aspect provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage medium
Matter stores computer instruction, and the computer instruction executes the mistake based on shaft temperature sensor described in first aspect of the embodiment of the present invention
The method for reporting recognition methods and its any alternative embodiment.
A kind of wrong report recognition methods based on shaft temperature sensor provided in an embodiment of the present invention, according to target when train operation
The current shaft end temperature and axis temperature related data that the shaft temperature sensor of axle position reports obtain axis temperature phase using axis temperature prediction model
The corresponding shaft end temperature prediction value of data is closed, then current shaft end temperature is compared with shaft end temperature prediction value, if the two
Difference meets default wrong report recognition rule, and the current shaft end temperature that can report shaft temperature sensor is identified as wrong report data.This
Inventive embodiments the method can effectively identify that shaft temperature sensor reports situation by mistake, be that PHM system is real-time to train operation state
Monitoring provides safeguard, and provides safeguard for train operating safety.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, 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 this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of wrong report recognition methods flow chart based on shaft temperature sensor of the embodiment of the present invention；
Fig. 2 is 30 seconds shaft end temperature change histogram schematic diagrames of the embodiment of the present invention；
Fig. 3 is the shaft end temperature of the shaft end temperature that shaft temperature sensor of the embodiment of the present invention reports and the output of axis temperature prediction model
The fitting effect schematic diagram of predicted value；
Fig. 4 is axis of embodiment of the present invention temperature error prediction model distribution histogram schematic diagram；
Fig. 5 is a kind of wrong report identification device schematic diagram based on shaft temperature sensor of the embodiment of the present invention；
Fig. 6 is the block schematic illustration of a kind of electronic equipment of the embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical solution in the embodiment of the present invention is explicitly described, it is clear that described embodiment is the present invention
A part of the embodiment, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not having
Every other embodiment obtained under the premise of creative work is made, shall fall within the protection scope of the present invention.
Fig. 1 is a kind of wrong report recognition methods flow chart based on shaft temperature sensor of the embodiment of the present invention, base as shown in Figure 1
In the wrong report recognition methods of shaft temperature sensor, including：
100, the current shaft end temperature and axis temperature dependency number that the shaft temperature sensor of target axle position reports when obtaining train operation
According to；
Specifically, the target axle position is that shaft end, pinion gear motor side, pinion gear axle side, motor drive end, motor are non-
Any sort in driving end, motor stator, gear wheel motor side and gear wheel axle side, wherein the axis temperature of every one kind axle position passes
Sensor corresponds to an axis temperature prediction model, such as shaft temperature sensor of above-mentioned eight classes axle position, respectively corresponds eight axis temperature prediction models.
The axis temperature related data includes master data, derivative variable data and continuous variable data, wherein described basic
Data include：Vehicle, row number, train number, railway carriage number, axle position id, time, car types, brake range, acceleration-deceleration, empty spring pressure
Power, sensor states, sensor temperature and speed.
The derivative variable data includes：Speed mean value, speed maximum value, speed minimum value, speed median, axis Wen Jun
Value, axis temperature maximum value, axis temperature minimum value, axis middle benefit gas digit and outer temperature mean value；
The continuous variable data include：Preceding 1 speed packet, preceding 2 speed packet, preceding 3 speed packet, preceding 4 speed packet, preceding 5 packet speed
The mild preceding 6 packet axis middle benefit gas of degree, preceding 6 speed packet, preceding 1 packet axis temperature, preceding 2 packet axis temperature, preceding 3 packet axis temperature, preceding 4 packet axis temperature, preceding 5 packet axis
It is a variety of.
It should be noted that the shaft temperature sensor of target axle position reports in the present embodiment current shaft end temperature and axis temperature phase
Data are closed, are the real time datas of the train operation state of train-installed WTD equipment detection, vehicle-mounted WTD equipment passes real time data
Ground is gone back to, ground is based on train health big data platform PHM system and is received, parsed and stored to the real time data.
It should be noted that real time data is generated data packet and is carried out when vehicle-mounted WTD equipment passes real time data back ground
Transmission.For continuous variable data, the speed of available preceding 7 data packets and the axis temperature of preceding 7 data packets.Certainly originally
Embodiment, which is not intended to limit, will be less than the continuous data of 7 data packets or the continuous data of extra 7 data packets as continuous data
Variable, for example, choosing the speed of preceding 5 data packets and the axis temperature of preceding 5 data packets；Alternatively, the speed of preceding 10 data packets
The axis temperature of degree and preceding 10 data packets can specifically determine that the present embodiment is not especially limited this as needed.
101, the axis temperature related data is input to the corresponding axis temperature prediction model of the shaft temperature sensor, described in acquisition
The corresponding shaft end temperature prediction value of axis temperature related data, the axis temperature prediction model are obtained by neural metwork training；
The real time data and then train health big data platform PHM system sent the present invention is based on vehicle-mounted WTD equipment, utilizes
Data mining to train operation state data carry out analysis and modeling, the train operation state data that PHM system is stored as
Training data is trained using neural network, obtains shaft end temperature prediction model.
As previously mentioned, the corresponding shaft temperature sensor of every one kind axle position respectively corresponds an axis temperature prediction model, but this implementation
Example does not limit the corresponding shaft temperature sensor of every a kind of axle position, and to respectively correspond an axis temperature prediction model identical or not identical, because
Axis temperature prediction model after being modeled by real-time data analysis may be identical, it is also possible to be different.But the present embodiment exists
It when modeling, is modeled respectively according to the corresponding shaft temperature sensor of every one kind axle position；The result of modeling, that is, axis temperature prediction model
It may be identical, it is also possible to be different.
Further, by the axis temperature related data be input to the corresponding axis temperature prediction model of the shaft temperature sensor it
Before, it further include the data in data filtering, Feature Engineering processing and labeling processing etc., with axis temperature prediction model training embodiment
Filtering, Feature Engineering processing and labeling processing etc. are identical, please refer to and are hereinafter described.
102, if judgement knows that the difference of the current shaft end temperature and the shaft end temperature prediction value meets default wrong report
The current shaft end temperature is then identified as wrong report data by recognition rule.
Specifically, the wrong report recognition methods described in the present embodiment based on shaft temperature sensor includes model prediction and wrong report identification
Two parts, model prediction mainly pass through step 101 and realize, i.e., are passed using the axis temperature that axis temperature prediction model obtains current goal axle position
The corresponding shaft end temperature prediction value of axis temperature related data that sensor reports；Wrong report identification is mainly realized by step 102, that is, is compared
The difference of current shaft end temperature and shaft end temperature prediction value that the shaft temperature sensor of target axle position reports, presets if the difference meets
It reports recognition rule by mistake, then determines that current shaft end temperature is abnormal data, it should be caused by shaft temperature sensor wrong report, then it will be described
Current shaft end temperature is identified as wrong report data.
A kind of wrong report recognition methods based on shaft temperature sensor provided in an embodiment of the present invention, according to target when train operation
The current shaft end temperature and axis temperature related data that the shaft temperature sensor of axle position reports obtain axis temperature phase using axis temperature prediction model
The corresponding shaft end temperature prediction value of data is closed, then current shaft end temperature is compared with shaft end temperature prediction value, if the two
Difference meets default wrong report recognition rule, and the current shaft end temperature that can report shaft temperature sensor is identified as wrong report data.This
Inventive embodiments the method can effectively identify that shaft temperature sensor reports situation by mistake, be that PHM system is real-time to train operation state
Monitoring provides safeguard, and provides safeguard for train operating safety.
Specifically, the default wrong report recognition rule includes：
The difference of the current shaft end temperature and the shaft end temperature prediction value is more than the error of the axis temperature prediction model
Threshold value, and the difference and the current shaft end temperature and latter packet temperature of the current shaft end temperature and previous packet temperature value
The difference of value is more than variation range of the target axle position in preset duration inner shaft temperature.
It is understood that including several critical datas in the default wrong report recognition rule, first critical data is
The error threshold of axis temperature prediction model, the error threshold can be obtained by model training and assessment；Second critical data
It is variation range of the target axle position in preset duration inner shaft temperature, the continuous change that can be detected by the shaft temperature sensor of target axle position
Amount data obtain.
In the embodiment of the present invention, if the difference of current shaft end temperature and the shaft end temperature prediction value is more than that the axis temperature is pre-
Survey the error threshold of model, and the difference and the current shaft end temperature of the current shaft end temperature and previous packet temperature value
Difference with latter packet temperature value is more than the target axle position in the variation range of preset duration inner shaft temperature, then determines current axis
End temperature is abnormal data, it should be that the current shaft end temperature is then identified as wrong report number caused by shaft temperature sensor wrong report
According to.
Based on the above embodiment, the axis temperature prediction model is obtained by neural metwork training, is specifically included：
Step 1, for every a kind of axle position, the shaft end temperature that shaft temperature sensor when train operation is reported and axis temperature dependency number
According to as training data, the corresponding training data of shaft temperature sensor of every a kind of axle position is obtained；
Step 2, the corresponding training data of shaft temperature sensor based on every a kind of axle position is obtained using correlation analysis algorithm
Take the variation model of shaft end temperature and speed, brake range, the relative coefficient between acceleration-deceleration and preset duration inner shaft temperature
It encloses；
Step 3, by the corresponding training data of shaft temperature sensor, relative coefficient and preset duration of every a kind of axle position
The variation range of axis temperature, is trained neural network, obtains the corresponding axis temperature prediction model of every a kind of axle position；
Wherein, the neural network is feedforward neural network, and learning algorithm is BP back-propagation algorithm, has input
Three layers of full structure in succession of layer, hidden layer and output layer.
Specifically, the training data in step 1 can be the received axis temperature sensing of train health big data platform PHM system
The real time data that device reports carries out mining analysis and obtains.
Specifically, using correlation analysis algorithm, obtaining the shaft end temperature and speed, braking in preset duration in step 2
Relative coefficient between gear, acceleration-deceleration etc., it is for statistical analysis by the situation of change to preset duration inner shaft temperature, it obtains
The variation range of preset duration inner shaft temperature is obtained, which is the normal variation range of axis temperature, and every one kind axle position is one corresponding
The variation range of preset duration inner shaft temperature.Wherein, preset duration can be selected according to actual needs, and the present embodiment does not limit specifically
It is fixed.
In an alternative embodiment, it is obtained in the present embodiment by analyzing 30 seconds axis temperature delta datas
Relative coefficient such as following table：
| Correlation | Brake range | Acceleration-deceleration | Speed | Outer temperature mean value |
| Shaft end temperature | 0.039249624 | -0.021929341 | 0.283792992 | 0.680123837 |
Fig. 2 is 30 seconds shaft end temperature change histogram schematic diagrames of the embodiment of the present invention, and shaft end temperature variations are asked within 30 seconds
With reference to Fig. 2, wherein horizontal axis is temperature, and the longitudinal axis is the corresponding temperature change probability of each temperature range.
It should be noted that establishing axis temperature prediction model is a regression problem in fact, and should be according to shaft end, small tooth
Turbin generator side, pinion gear wheel side, motor drive end, motor non-transmision end, motor stator, gear wheel motor side, gear wheel vehicle
Axis temperature prediction model is established in wheel side respectively.In order to guarantee the accuracy of prediction model, there is no using general for the embodiment of the present invention
Regression algorithm, but use more special neural network algorithm and establish axis temperature prediction model.Specifically, the present invention is real
The type for applying the neural network of example is feedforward neural network, and learning algorithm is BP back-propagation algorithm；Network structure is three layers
Full structure, including input layer, hidden layer and output layer in succession, wherein input layer includes 128 neurons, and hidden layer includes 64
Neuron, output layer include 1 neuron；Trained study, learning rate 0.0013；Loss function is mean square error, excellent
Change device is Adam optimizer, and anti-over-fitting technology is Dropout.
The axis temperature of brake range, acceleration-deceleration, speed, outer temperature when the embodiment of the present invention is according to train operation and each axle position
Data and shaft temperature sensor status data etc. carry out comprehensive analysis to the axis temperature related data of train, establish the axis temperature of each axle position
The axis temperature prediction model of sensor, then the shaft end temperature value transmitted according to shaft temperature sensor judge whether the shaft temperature sensor is deposited
It is reporting by mistake.
Based on the above embodiment, step 1, described for every a kind of axle position, shaft temperature sensor when train operation is reported
Shaft end temperature and axis temperature related data obtain the corresponding training data of shaft temperature sensor of every a kind of axle position as training data,
It specifically includes：
Step 1.1, the shaft end temperature that the shaft temperature sensor of every a kind of axle position reports when obtaining train operation is related to axis temperature
Data；
Step 1.2, data filtering, Feature Engineering are carried out to the corresponding shaft end temperature of every one kind axle position and axis temperature related data
After processing and labeling processing, prediction label is obtained；
Step 1.3, it according to the corresponding shaft end temperature of every one kind axle position, axis temperature related data and the prediction label, obtains
The corresponding training data of shaft temperature sensor of every one kind axle position.
The real time data that the received shaft temperature sensor of train health big data platform PHM system of the embodiment of the present invention reports into
Row mining analysis, specifically, the step 1.1 shaft end temperature that the shaft temperature sensor of every one kind axle position reports when obtaining train operation
When with axis temperature related data, since initial data is stored in the HBase data warehouse of PHM system, in order to facilitate the pumping of data
It takes and guarantees that data are standardized data structures, it is possible to first establish hive table, the data for needing to use according to each
Database table is established, data see that table includes field, annotation, data type and illustrates；The data for needing to use are as previously mentioned, packet
Master data, derivative variable data and continuous variable data are included, are not being repeated herein；It is again very huge due to being related to data volume,
So also to carry out partition management according to month and date.It is then possible to which the hql sentence of data from the sample survey is write into shell script
In, according to the data in 3 hours before the decimation in time of execution, because having 8 major class axle positions (shaft end, pinion gear motor side, small tooth
Take turns axle side, motor drive end, motor non-transmision end, motor stator, gear wheel motor side and gear wheel axle side), so one
8 data from the sample survey scripts are shared, distinguish different shaft temperature sensors pair according to the filter condition of axle position id in data from the sample survey script
The training data for the axis temperature prediction model answered.
Based on the above embodiment, data filtering described in step 1.2 includes：Null value filtering, forbidden character filtering, empty spring pressure
The filtering of power exceptional value, axis temperature data -50 spend one of filter or a variety of；
The Feature Engineering is handled：Car types are mapped to numerical value；First kind brake range is mapped to first
Second class brake range is mapped to the value within the scope of second value by the value in numberical range, and third class brake range is mapped
For third value.
Specifically, including T and M by car types, T being mapped to 0, M is mapped to for existing type of train
1；First kind brake range can be B1~B7, B1~B7 can be mapped to -1~-7；Second class brake range can be P1
P1~P10 can be mapped to 1~10 by~P10, and third class brake range is other gears, other gears can be mapped as
0。
The labeling is handled：Using vehicle, row number, license number and axle position id as keyword, the determining and keyword
Unique corresponding axle box bearing sensor, using the shaft end temperature of the shaft temperature sensor subsequent time of every a kind of axle position as it is current when
The prediction label at quarter.
Finally, by the corresponding shaft end temperature of every one kind axle position, axis temperature related data and the prediction label, as every one kind
The corresponding training data of the shaft temperature sensor of axle position.
Based on the above embodiment, the embodiment of the present invention by the corresponding training data of shaft temperature sensor of every a kind of axle position,
The variation range of relative coefficient and preset duration inner shaft temperature, is trained neural network, and it is corresponding to obtain every a kind of axle position
Axis temperature prediction model.
In general, model quality mainly sees two aspect：One be model accuracy, the other is the stabilization of model
Property；Accuracy refers to model in the output on sample and the error between true value, and stability refers to that model is defeated each time
The fluctuation size of error between result and true value out.
The embodiment of the present invention selects mean absolute error (MAE) and mean square error (MSE) to refer to as main model evaluation
Mark assesses trained axis temperature prediction model.Mean absolute error can embody the accuracy of model, mean square error
It can show the stability of model.When mean absolute error and mean square error are all in small value, the essence of model has been reacted
Exactness is high, and stability is high, and whole Error (error) is small, shows that the estimated performance of neural network model is good.In addition it also chooses
R2 model evaluating index, it can reflect the correlation degree between the predicted value of model and the true value of actual conditions.
Fig. 3 is the shaft end temperature of the shaft end temperature that shaft temperature sensor of the embodiment of the present invention reports and the output of axis temperature prediction model
The fitting effect schematic diagram of predicted value, the fitting effect of Visualization Model, predicted value and the reality for drawing axle box axis temperature model are right
The curve graph for the axis temperature value answered, referring to FIG. 3, wherein horizontal axis is packet number, the longitudinal axis is temperature.From figure 3, it can be seen that axis temperature is pre-
The fitting effect for surveying the shaft end temperature prediction value that model exports and the shaft end temperature that shaft temperature sensor reports is good, can be with table
In achievement data mutually confirm.It is possible thereby to confirm that the precision of prediction of axis temperature prediction model of the embodiment of the present invention is very high.
Based on the above embodiment, the range of the error threshold of axis of embodiment of the present invention temperature prediction model is 1-2 degree.
The variation range of the preset duration inner shaft temperature includes：30 seconds inner shaft temperature variation ranges are 1-2 degree；
Correspondingly, the default wrong report recognition rule includes：
The absolute value of the difference of the current shaft end temperature and the shaft end temperature prediction value is greater than 2 degree, and described current
The difference of the absolute value and the current shaft end temperature and latter packet temperature value of shaft end temperature and the difference of previous packet temperature value
Absolute value more than 2 degree.
Fig. 4 is axis of embodiment of the present invention temperature error prediction model distribution histogram schematic diagram, referring to FIG. 4, according to one week
The test data of left and right, between 1~2 degree, between 2~3 degree, between 3~4 degree, is situated between as a result, according to absolute value error less than 1 degree
It in 4~5 degree, between 5~6 degree, is divided greater than 6 degree, draws ratio shared by data in each error range, as a result as schemed
Shown in 4, wherein horizontal axis is temperature, and the longitudinal axis is error rate.As can be seen that missing from axis temperature error prediction model distribution histogram
Accounting distribution of the difference greater than 2 degree is relatively average and belongs to the same magnitude, and both less than a ten thousandth, so axis temperature prediction model
Reasonable error threshold value is 2 degree.
Specific default wrong report recognition rule can be obtained as a result, is, when the prediction of shaft end temperature value and axis temperature prediction model
When the Error Absolute Value of value is greater than 2 degree, and the absolute difference of shaft end temperature value and previous packet temperature value is greater than 2 degree, and rear
When the absolute difference of one packet temperature value is also greater than 2 degree, then assert that shaft end temperature value is caused by temperature sensor wrong report.
The embodiment of the present invention passes the running state data of PHM system back based on vehicle-mounted WTD equipment, to the shaft end temperature of train
It is analyzed with shaft temperature sensor state, studied and axle box bearing temperature is predicted using neural network algorithm, according to pre-
Measured value wrong report recognition rule corresponding with the generation of actual value difference is finally developed in big data platform and applies this rule.This hair
Wrong report recognition methods described in bright embodiment based on shaft temperature sensor, which can report train axle box bearing temperature sensor by mistake, to be carried out
Effectively identification repairs for service personnel and provides guidance, carries out data analysis for business personnel and excavation provides safeguard；?
Under some axle box bearing temperature conditions, reached using a packet axis temperature temperature value, predictablity rate under Neural Network Prediction
99%, train operation state real time monitoring is provided safeguard for PHM system, is provided safeguard for train operating safety.
Fig. 5 is a kind of wrong report identification device schematic diagram based on shaft temperature sensor of the embodiment of the present invention, as shown in figure 5, this
Inventive embodiments also provide a kind of wrong report identification device based on shaft temperature sensor, including：
Data acquisition module 500, the current shaft end that the shaft temperature sensor of target axle position reports when for obtaining train operation
Temperature and axis temperature related data；
Model prediction module 501, for the axis temperature related data to be input to the corresponding axis temperature of the shaft temperature sensor
Prediction model, obtains the corresponding shaft end temperature prediction value of the axis temperature related data, and the axis temperature prediction model passes through nerve net
Network training obtains；
Identification module 502 is reported by mistake, if for judging to know the current shaft end temperature and the shaft end temperature prediction value
Difference meets default wrong report recognition rule, then the current shaft end temperature is identified as wrong report data.
It is real to can be used for executing the wrong report recognition methods shown in FIG. 1 based on shaft temperature sensor for the device of the embodiment of the present invention
The technical solution of example is applied, it is similar that the realization principle and technical effect are similar, and details are not described herein again.
Fig. 6 is the block schematic illustration of a kind of electronic equipment of the embodiment of the present invention.Referring to FIG. 6, the embodiment of the present invention provides
A kind of electronic equipment, including：Processor (processor) 610, communication interface (Communications Interface) 620,
Memory (memory) 630 and bus 640, wherein processor 610, communication interface 620, memory 630 are complete by bus 640
At mutual communication.Processor 610 can call the logical order in memory 630, to execute following method, including：It obtains
The current shaft end temperature and axis temperature related data that the shaft temperature sensor of target axle position reports when taking train operation；By the axis temperature phase
It closes data and is input to the corresponding axis temperature prediction model of the shaft temperature sensor, obtain the corresponding shaft end temperature of the axis temperature related data
Predicted value is spent, the axis temperature prediction model is obtained by neural metwork training；If the current shaft end temperature and institute are known in judgement
The difference for stating shaft end temperature prediction value meets default wrong report recognition rule, then the current shaft end temperature is identified as wrong report number
According to.
The embodiment of the present invention discloses a kind of computer program product, and the computer program product is non-transient including being stored in
Computer program on computer readable storage medium, the computer program include program instruction, when described program instructs quilt
When computer executes, computer is able to carry out method provided by above-mentioned each method embodiment, for example including：Obtain train operation
When target axle position the shaft temperature sensor current shaft end temperature and axis temperature related data that report；The axis temperature related data is inputted
To the corresponding axis temperature prediction model of the shaft temperature sensor, the corresponding shaft end temperature prediction value of the axis temperature related data is obtained,
The axis temperature prediction model is obtained by neural metwork training；If the current shaft end temperature and the shaft end temperature are known in judgement
The difference of predicted value meets default wrong report recognition rule, then the current shaft end temperature is identified as wrong report data.
The embodiment of the present invention provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage
Medium storing computer instruction, the computer instruction make the computer execute side provided by above-mentioned each method embodiment
Method, for example including：The current shaft end temperature and axis temperature dependency number that the shaft temperature sensor of target axle position reports when obtaining train operation
According to；The axis temperature related data is input to the corresponding axis temperature prediction model of the shaft temperature sensor, it is related to obtain the axis temperature
The corresponding shaft end temperature prediction value of data, the axis temperature prediction model are obtained by neural metwork training；If judgement is known described
The difference of current shaft end temperature and the shaft end temperature prediction value meets default wrong report recognition rule, then by the current shaft end temperature
Degree is identified as wrong report data.
Those of ordinary skill in the art will appreciate that：Realize that above equipment embodiment or embodiment of the method are only schematic
, wherein the processor and the memory can be physically separate component may not be it is physically separated, i.e.,
It can be located in one place, or may be distributed over multiple network units.It can select according to the actual needs therein
Some or all of the modules achieves the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying creative labor
In the case where dynamic, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as USB flash disk, mobile hard disk, ROM/RAM, magnetic disk, CD
Deng, including some instructions use is so that a computer equipment (can be personal computer, server or the network equipment etc.)
Execute method described in certain parts of each embodiment or embodiment.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that：It still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of wrong report recognition methods based on shaft temperature sensor, which is characterized in that including：
The current shaft end temperature and axis temperature related data that the shaft temperature sensor of target axle position reports when obtaining train operation；
The axis temperature related data is input to the corresponding axis temperature prediction model of the shaft temperature sensor, it is related to obtain the axis temperature
The corresponding shaft end temperature prediction value of data, the axis temperature prediction model are obtained by neural metwork training；
If judgement knows that the difference of the current shaft end temperature and the shaft end temperature prediction value meets default wrong report recognition rule,
The current shaft end temperature is then identified as wrong report data.
2. the method according to claim 1, wherein the target axle position is shaft end, pinion gear motor side, small tooth
Take turns any in axle side, motor drive end, motor non-transmision end, motor stator, gear wheel motor side and gear wheel axle side
Class, wherein the corresponding axis temperature prediction model of shaft temperature sensor of every one kind axle position；
The axis temperature related data includes master data, derivative variable data and continuous variable data, wherein the master data
Including：Vehicle, train number, railway carriage number, axle position id, the time, car types, brake range, acceleration-deceleration, empty spring pressure, passes row number
Sensor state, sensor temperature and speed；
The derivative variable data includes：Speed mean value, speed maximum value, speed minimum value, speed median, axis temperature mean value,
Axis temperature maximum value, axis temperature minimum value, axis middle benefit gas digit and outer temperature mean value；
The continuous variable data include：It is preceding 1 speed packet, preceding 2 speed packet, preceding 3 speed packet, preceding 4 speed packet, preceding 5 speed packet, preceding
The mild preceding 6 packet axis middle benefit gas of 6 speed packets, preceding 1 packet axis temperature, preceding 2 packet axis temperature, preceding 3 packet axis temperature, preceding 4 packet axis temperature, preceding 5 packet axis it is a variety of.
3. method according to claim 1 or 2, which is characterized in that the axis temperature prediction model passes through neural metwork training
It obtains, specifically includes：
For every a kind of axle position, the shaft end temperature and axis temperature related data that shaft temperature sensor when train operation is reported are as training
Data obtain the corresponding training data of shaft temperature sensor of every a kind of axle position；
The corresponding training data of shaft temperature sensor based on every a kind of axle position obtains shaft end temperature using correlation analysis algorithm
The variation range of relative coefficient and preset duration inner shaft temperature between speed, brake range, acceleration-deceleration；
Pass through the variation of the corresponding training data of shaft temperature sensor of every a kind of axle position, relative coefficient and preset duration inner shaft temperature
Range is trained neural network, obtains the corresponding axis temperature prediction model of every a kind of axle position；
Wherein, the neural network is feedforward neural network, and learning algorithm is BP back-propagation algorithm, has input layer, hidden
Hide three layers of full structure in succession of layer and output layer.
4. according to the method described in claim 3, it is characterized in that, described for every a kind of axle position, by axis temperature when train operation
For the shaft end temperature and axis temperature related data that sensor reports as training data, the shaft temperature sensor for obtaining every a kind of axle position is corresponding
Training data, specifically include：
The shaft end temperature and axis temperature related data that the shaft temperature sensor of every a kind of axle position reports when obtaining train operation；
Data filtering, Feature Engineering processing and labeling are carried out to the corresponding shaft end temperature of every one kind axle position and axis temperature related data
After processing, prediction label is obtained；
According to the corresponding shaft end temperature of every one kind axle position, axis temperature related data and the prediction label, every a kind of axle position is obtained
The corresponding training data of shaft temperature sensor.
5. according to the method described in claim 4, it is characterized in that, the data filtering includes：Null value filtering, forbidden character mistake
Filter, empty spring pressure anomaly value filtering, axis temperature data -50 spend one of filter or a variety of；
The Feature Engineering is handled：Car types are mapped to numerical value；First kind brake range is mapped to the first numerical value
Second class brake range is mapped to the value within the scope of second value by the value in range, and third class brake range is mapped as
Three numerical value；
The labeling is handled：Using vehicle, row number, license number and axle position id as keyword, determination is unique with the keyword
Corresponding axle box bearing sensor, using the shaft end temperature of the shaft temperature sensor subsequent time of every a kind of axle position as current time
Prediction label.
6. the method according to claim 1, wherein the default wrong report recognition rule includes：
The difference of the current shaft end temperature and the shaft end temperature prediction value is more than the error threshold of the axis temperature prediction model,
And the current shaft end temperature and the difference of previous packet temperature value and the difference of the current shaft end temperature and latter packet temperature value
Value is more than variation range of the target axle position in preset duration inner shaft temperature.
7. method according to claim 1 or 6, which is characterized in that the range of the error threshold of the axis temperature prediction model
For 1-2 degree；The variation range of the preset duration inner shaft temperature includes：30 seconds inner shaft temperature variation ranges are 1-2 degree；
Correspondingly, the default wrong report recognition rule includes：
The absolute value of the difference of the current shaft end temperature and the shaft end temperature prediction value is greater than 2 degree, and the current shaft end
The absolute value of temperature and the difference of previous packet temperature value and the current shaft end temperature are exhausted with the difference of latter packet temperature value
To value more than 2 degree.
8. a kind of wrong report identification device based on shaft temperature sensor, which is characterized in that including：
Data acquisition module, the current shaft end temperature and axis that the shaft temperature sensor of target axle position reports when for obtaining train operation
Warm related data；
Model prediction module predicts mould for the axis temperature related data to be input to the corresponding axis temperature of the shaft temperature sensor
Type, obtains the corresponding shaft end temperature prediction value of the axis temperature related data, and the axis temperature prediction model passes through neural metwork training
It obtains；
Identification module is reported by mistake, if for judging to know that the difference of the current shaft end temperature and the shaft end temperature prediction value meets
Default wrong report recognition rule, then be identified as wrong report data for the current shaft end temperature.
9. a kind of electronic equipment, which is characterized in that including：
At least one processor；And
At least one processor being connect with the processor communication, wherein：
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to instruct energy
Enough methods executed as described in claim 1 to 7 is any.
10. a kind of non-transient computer readable storage medium, which is characterized in that the non-transient computer readable storage medium is deposited
Computer instruction is stored up, the computer instruction makes the computer execute the method as described in claim 1 to 7 is any.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810515666.9A CN108860211A (en) | 2018-05-25 | 2018-05-25 | A kind of wrong report recognition methods and device based on shaft temperature sensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810515666.9A CN108860211A (en) | 2018-05-25 | 2018-05-25 | A kind of wrong report recognition methods and device based on shaft temperature sensor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108860211A true CN108860211A (en) | 2018-11-23 |
Family
ID=64333800
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810515666.9A Pending CN108860211A (en) | 2018-05-25 | 2018-05-25 | A kind of wrong report recognition methods and device based on shaft temperature sensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108860211A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978228A (en) * | 2019-01-31 | 2019-07-05 | 中南大学 | A kind of PM2.5 concentration prediction method, apparatus and medium |
| CN111351664A (en) * | 2020-02-13 | 2020-06-30 | 成都运达科技股份有限公司 | Bearing temperature prediction and alarm diagnosis method based on LSTM model |
| CN111829693A (en) * | 2020-07-30 | 2020-10-27 | 成都运达科技股份有限公司 | Bearing temperature detection method and system based on LSTM model of time and space dimensions |
| CN112183830A (en) * | 2020-09-16 | 2021-01-05 | 新奥数能科技有限公司 | Method and device for predicting temperature of chilled water |
| CN112446411A (en) * | 2019-08-12 | 2021-03-05 | 美光科技公司 | Storage and access of neural network inputs in automotive predictive maintenance |
| CN112576454A (en) * | 2020-12-11 | 2021-03-30 | 龙源（北京）风电工程技术有限公司 | Wind turbine generator main shaft temperature early warning method and device based on multi-dimensional early warning strategy |
| CN112572522A (en) * | 2020-11-10 | 2021-03-30 | 中车青岛四方机车车辆股份有限公司 | Early warning method and device for axle temperature fault of vehicle bearing |
| CN114577364A (en) * | 2020-12-01 | 2022-06-03 | 株洲中车时代电气股份有限公司 | Train axle temperature sensor fault diagnosis method, system and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1688874A (en) * | 2002-08-30 | 2005-10-26 | 日本精工株式会社 | Method and device for monitoring status of mechanical equipment and abnormality diagnosing device |
| CN106061821A (en) * | 2013-12-24 | 2016-10-26 | 阿母斯替德铁路公司 | System and method for detecting operational anomalies in train consists and railcars |
-
2018
- 2018-05-25 CN CN201810515666.9A patent/CN108860211A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1688874A (en) * | 2002-08-30 | 2005-10-26 | 日本精工株式会社 | Method and device for monitoring status of mechanical equipment and abnormality diagnosing device |
| CN106061821A (en) * | 2013-12-24 | 2016-10-26 | 阿母斯替德铁路公司 | System and method for detecting operational anomalies in train consists and railcars |
Non-Patent Citations (2)
| Title |
|---|
| 张士存: "基于动车组车载WTD设备的数据实时接收和解析技术机制的研究与应用", 《电子世界》 * |
| 顾佳: "基于神经网络的动车组轴端温度传感器误报识别", 《数字通信世界》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978228A (en) * | 2019-01-31 | 2019-07-05 | 中南大学 | A kind of PM2.5 concentration prediction method, apparatus and medium |
| CN109978228B (en) * | 2019-01-31 | 2023-12-12 | 中南大学 | PM2.5 concentration prediction method, device and medium |
| CN112446411A (en) * | 2019-08-12 | 2021-03-05 | 美光科技公司 | Storage and access of neural network inputs in automotive predictive maintenance |
| CN111351664A (en) * | 2020-02-13 | 2020-06-30 | 成都运达科技股份有限公司 | Bearing temperature prediction and alarm diagnosis method based on LSTM model |
| CN111829693A (en) * | 2020-07-30 | 2020-10-27 | 成都运达科技股份有限公司 | Bearing temperature detection method and system based on LSTM model of time and space dimensions |
| CN112183830A (en) * | 2020-09-16 | 2021-01-05 | 新奥数能科技有限公司 | Method and device for predicting temperature of chilled water |
| CN112572522A (en) * | 2020-11-10 | 2021-03-30 | 中车青岛四方机车车辆股份有限公司 | Early warning method and device for axle temperature fault of vehicle bearing |
| CN114577364A (en) * | 2020-12-01 | 2022-06-03 | 株洲中车时代电气股份有限公司 | Train axle temperature sensor fault diagnosis method, system and device |
| CN114577364B (en) * | 2020-12-01 | 2022-11-08 | 株洲中车时代电气股份有限公司 | Train axle temperature sensor fault diagnosis method, system and device |
| CN112576454A (en) * | 2020-12-11 | 2021-03-30 | 龙源（北京）风电工程技术有限公司 | Wind turbine generator main shaft temperature early warning method and device based on multi-dimensional early warning strategy |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108860211A (en) | A kind of wrong report recognition methods and device based on shaft temperature sensor | |
| CN106023586B (en) | For detecting the abnormal method and apparatus in vehicular traffic mode | |
| CN109800127A (en) | A kind of system fault diagnosis intelligence O&M method and system based on machine learning | |
| CN109581871B (en) | Industrial control system intrusion detection method of immune countermeasure sample | |
| Zorin | Assessment of products risks of mechanical engineering by results of diagnosing | |
| JP2020507748A (en) | Cloud-based vehicle failure diagnosis method, apparatus and system | |
| CN103544389A (en) | Fault tree and fuzzy neural network based automobile crane fault diagnosis method | |
| KR20180108446A (en) | System and method for management of ict infra | |
| CN107437147A (en) | Reduce the vehicle travel risk dynamic assessment method and its system of freight logistics scene | |
| CN107776606A (en) | A kind of shaft temperature monitoring system fault detection method | |
| CN113177650A (en) | Predictive maintenance method and device for wagon compartment | |
| Biroon et al. | Real-time false data injection attack detection in connected vehicle systems with pde modeling | |
| CN116893643A (en) | Intelligent robot driving track safety control system based on data analysis | |
| US20200156680A1 (en) | Railway vehicle major component and system diagnosis apparatus | |
| CN110673588A (en) | Wireless overtime degradation fault diagnosis method for CTCS-3 train control system | |
| CN111078503B (en) | Abnormality monitoring method and system | |
| CN108248641A (en) | A kind of urban track traffic data processing method and device | |
| CN110119759B (en) | Gain tensor-based infrared image monitoring method for high-temperature pressure-bearing element | |
| CN112572522A (en) | Early warning method and device for axle temperature fault of vehicle bearing | |
| Stuchlý et al. | Reliability evaluation as a means of increasing the efficiency of equipment maintenance | |
| CN110097219A (en) | A kind of electric vehicle O&M optimization method based on security tree model | |
| Susarev et al. | Use of previous conditions matrixes for the vehicle on the basis of operational information and dynamic models of systems, Nodes and Units | |
| CN105910804A (en) | Method and device for monitoring performance trends of parts and accessories of airplane | |
| Li et al. | A hazard causation analysis method for battery electric vehicles based on STPA and complex network | |
| Bleisinger et al. | Machine Learning Based Simulation for Wear Estimation in Commercial Vehicle Applications |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181123 |
|
| RJ01 | Rejection of invention patent application after publication |