CN109325310A - A kind of bullet train intermittent fault detection method based on the multiple side's T control figure - Google Patents
A kind of bullet train intermittent fault detection method based on the multiple side's T control figure Download PDFInfo
- Publication number
- CN109325310A CN109325310A CN201811250111.2A CN201811250111A CN109325310A CN 109325310 A CN109325310 A CN 109325310A CN 201811250111 A CN201811250111 A CN 201811250111A CN 109325310 A CN109325310 A CN 109325310A
- Authority
- CN
- China
- Prior art keywords
- intermittent fault
- time
- steps
- fault
- failure
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/08—Railway vehicles
Abstract
The invention discloses a kind of bullet train intermittent fault detection methods based on the multiple side's T control figure, belong to fault diagnosis field；This method can detecte out that amplitude is smaller and the intermittent fault of limited duration, and infer the time of origin and extinction time of intermittent fault, significantly improve data-driven method to the detection effect of intermittent fault；This method comprises: training data is collected and pre-processed, bullet train critical system intermittent fault parameter is analyzed, online fault detection and wrong report exclude, and time window selection and intermittent fault generation, extinction time are inferred in real time.The practical application request of effective guarantee of the present invention bullet train critical system intermittent fault detection.
Description
Technical field
The invention belongs to fault diagnosis fields, and in particular to a kind of bullet train interval event based on the multiple side's T control figure
Hinder detection method.
Background technique
In recent years, hot research problem is had become to the fault detection of bullet train critical system.However for many years, people
It is concerned only with the test problems of sustained fault mostly, and the test problems of intermittent fault are studied less.On the other hand, with electricity
The high speed development of the technologies such as son, information, a kind of failure and special type different from conventional persistence failure mode, i.e. intermittent fault by
Gradually cause the attention of people.High-speed train braking control system is run under complex environment, easily generation intermittent fault.One side
Face, electronic brake control unit are made of complex electronic circuit, and rosin joint aging etc. can all cause control circuit to loosen and then cause
Controller intermittent fault.In addition, bullet train running environment is complicated, onboard sensor is influenced vulnerable to vibration, electromagnetic interference etc., is led
Cause sensor intermittent failure.
Intermittent fault refers to a kind of limited duration, and no external compensation measure still can die away and make system weight
The new failure for restoring acceptable performance.Compared with sustained fault, the generation of intermittent fault has randomness and repeatability, and has
Apparent cumulative effect, i.e., as time goes by, fault occurrence frequency and duration gradually increase, and occur finally to drill by interval
Become sustained fault.Since the generation and disappearance of intermittent fault have randomness, diagnostic criteria requires while detecting event
The generation of barrier and disappearance moment, diagnosis performance meets determines that moment occurs in its before this failure vanishes, sends out in failure next time
Determine that it disappears the moment before death.Such testing result is more advantageous to the operating status of our analysis systems, formulates reasonable dimension
Shield and maintenance policy and the design that faults-tolerant control rule is carried out for intermittent fault.
Intermittent fault amplitude is smaller and the characteristic of limited duration, so that traditional data-driven fault detection method is very
Difficulty directly applies to the detection of intermittent fault
Summary of the invention
For the above-mentioned technical problems in the prior art, the invention proposes a kind of based on the multiple side's T control figure
Bullet train intermittent fault detection method, design rationally, overcome the deficiencies in the prior art, have good effect.
To achieve the goals above, the present invention adopts the following technical scheme:
Bullet train intermittent fault detection method based on the multiple side's T control figure, includes the following steps:
Step 1: off-line training；Specifically comprise the following steps:
Step 1.1: construction nominal situation calculation matrix:
[X1,X2,…,Xk,…,XN]∈Rm×N；
Wherein, Xk∈Rm×1For column vector；M is the number of probes that system detected includes；N is each sensor packet
The independent sample number included；
Step 1.2: calculating the sample average of training dataWith covariance matrix S, i.e.,
Step 1.3: rule of thumb or by analysis of history fault data, providing bullet train critical system intermittent fault
Direction ξq, failure amplitude lower boundTrouble duration lower boundWith failure vanishes time lower boundAnd calculate following formula:
Wherein, W*And W#For two time windows；W=1,2 ..., W#；Theoretically to be detected to failure
The upper bound of delay；For theoretically to the upper bound of failure vanishes detection delay；
Step 1.4: according to given confidence level α, to time window W=1,2 ..., W#, calculate separatelyI.e.
Wherein, FaIt is p that (p, N-p), which is freedom degree, the upper quantile when F distribution confidence level of N-p is α, note
Step 2: online fault detection；Specifically comprise the following steps:
Step 2.1: online acquisition newly measures sampleAnd to time window W=1,2 ..., W#, each side T is calculated in real time
The statistic of control figureI.e.
Step 2.2: to time window W=1,2 ..., W#, the alarm moment of each side's the T control figure of real-time update: failure hair
Raw alarmIt alarms with failure vanishesI.e.
Step 2.3: it is online to exclude wrong report, specifically comprise the following steps:
Step 2.3.1: if to W=W#, following formula is invalid
Then restart step 2；It is no to then follow the steps 2.4；
Step 2.3.2: W ∈ [W if it exists*,W#] make (1) formula invalid, then restart step 2；It is no to then follow the steps
2.4；
Step 2.3.3: W if it exists, W ' ∈ [W*,W#] following formula is set up
WhereinFor empty set, then restart step 2, it is no to then follow the steps 2.4；
Step 2.4: selection time window specifically comprises the following steps:
Step 2.4.1: W={ W is successively examined*..., 1 }, if (1) formula is invalid, set Wo=W+1；Otherwise Wo=1；
Step 2.4.2: W={ W is successively examined*..., Wo }, W ' ∈ (W, W if it exists#], so that (2) formula is set up, then reset
Wo=W+1；
Step 2.4.3: to W={ W*,…,Wo, it calculates
Step 2.4.4: W={ W is successively examined*,…,Wo, W ' ∈ (W, W if it exists#] following formula is set up, then reset Wo
=W+1；
Step 2.5: intermittent fault occurs and extinction time is inferred, specifically comprises the following steps:
Step 2.5.1: intermittent fault generation and the final deduction of extinction time are calculatedWithI.e.
Step 2.5.2: infer that this intermittent fault existsOccur in moment,It disappears in moment.
Advantageous effects brought by the present invention:
The present invention provides a kind of new method of intermittent fault detection, and this method give the selection of multiple temporal window standards
Then, it can detecte out that amplitude is smaller and the intermittent fault of limited duration, and infer time of origin and the disappearance of intermittent fault
Time significantly improves data-driven method to the detection effect of intermittent fault.
Detailed description of the invention
Fig. 1 is the simulation result schematic diagram that intermittent fault repeated and disappear.
Fig. 2 is the original side the T control testing result schematic diagram not carried out when wrong report is eliminated.
Fig. 3 is the intermittent fault testing result schematic diagram after wrong report is eliminated and time window selects.
Fig. 4 is the flow chart of the method for the present invention.
Specific embodiment
With reference to the accompanying drawing and specific embodiment invention is further described in detail:
1, a kind of bullet train intermittent fault detection method based on the multiple side's T control figure, process is as shown in figure 4, packet
Include following steps:
Step 1: off-line training；Specifically comprise the following steps:
Step 1.1: construction nominal situation calculation matrix:
[X1,X2,…,Xk,…,XN]∈Rm×N；
Wherein, Xk∈Rm×1For column vector；M is the number of probes that system detected includes；N is each sensor packet
The independent sample number included；
Step 1.2: calculating the sample average of training dataWith covariance matrix S, i.e.,
Step 1.3: rule of thumb or by analysis of history fault data, providing bullet train critical system intermittent fault
Direction ξq, failure amplitude lower boundTrouble duration lower boundWith failure vanishes time lower boundAnd calculate following formula:
Wherein, W*And W#For two time windows；W=1,2 ..., W#；Theoretically to be detected to failure
The upper bound of delay；For theoretically to the upper bound of failure vanishes detection delay；
Step 1.4: according to given confidence level α, to time window W=1,2 ..., W#, calculate separatelyI.e.
Wherein, FaIt is p that (p, N-p), which is freedom degree, the upper quantile when F distribution confidence level of N-p is α, note
Step 2: online fault detection；Specifically comprise the following steps:
Step 2.1: online acquisition newly measures sampleAnd to time window W=1,2 ..., W#, each side T is calculated in real time
The statistic of control figureI.e.
Step 2.2: to time window W=1,2 ..., W#, the alarm moment of each side's the T control figure of real-time update: failure hair
Raw alarmIt alarms with failure vanishesI.e.
Step 2.3: it is online to exclude wrong report, specifically comprise the following steps:
Step 2.3.1: if to W=W#, following formula is invalid
Then restart step 2；It is no to then follow the steps 2.4；
Step 2.3.2: W ∈ [W if it exists*,W#] make (1) formula invalid, then restart step 2；It is no to then follow the steps
2.4；
Step 2.3.3: W if it exists, W ' ∈ [W*,W#] following formula is set up
WhereinFor empty set, then restart step 2, it is no to then follow the steps 2.4；
Step 2.4: selection time window specifically comprises the following steps:
Step 2.4.1: W={ W is successively examined*..., 1 }, if (1) formula is invalid, set Wo=W+1；Otherwise Wo=1；
Step 2.4.2: W={ W is successively examined*,…,Wo, W ' ∈ (W, W if it exists#], so that (2) formula is set up, then reset
Wo=W+1；
Step 2.4.3: to W={ W*,…,Wo, it calculates
Step 2.4.4: W={ W is successively examined*..., Wo }, W ' ∈ (W, W if it exists#] following formula is set up, then reset Wo
=W+1；
Step 2.5: intermittent fault occurs and extinction time is inferred, specifically comprises the following steps:
Step 2.5.1: intermittent fault generation and the final deduction of extinction time are calculatedWithI.e.
Step 2.5.2: infer that this intermittent fault existsOccur in moment,It disappears in moment.
2, simulation study
Simulation model is chosen as follows:
5000 training datas are generated according to above-mentioned model first, represent the observation data under nominal situation；Then, then root
500 test datas are generated according to above-mentioned model, and it is ξ that direction, which is added, from the 201st dataq=[0.2425,0.9701]T's
Intermittent fault；Intermittent fault amplitude lower bound isIntermittent fault continues and extinction time lower bound isIt is imitative
True result is as shown in Figure 1, 2, 3.
Left side ordinate illustrates the true value of 1/4 failure amplitude in Fig. 1, and solid line represents the hair repeatedly of intermittent fault in figure
Raw and disappearance.Ordinate represents the testing result of 1/10 length of window on the right side of Fig. 2, and dotted line expression does not carry out wrong report elimination in figure
When the original side T control testing result.Fig. 3 dotted line indicates the intermittent fault detection after wrong report is eliminated and time window selects
As a result, representing us if corresponding time window takes place without dotted line in failure and speculating failure generation in real time and disappearing
Lose non-selected time window when the time.
Table 1 gives our inferred results to each secondary intermittent fault generation and extinction time, wherein μqRepresent interval event
The true generation moment of barrier two is classified as our deductions of the method to the generation moment thereafter, i.e., in above-mentioned algorithm stepsLikewise, νqThe true disappearance moment of intermittent fault is represented, two our methods is classified as thereafter the disappearance moment is pushed away
It is disconnected, i.e., in above-mentioned algorithm steps
Table 1
It can easily be seen that generation and the extinction time for reporting and deducing intermittent fault by mistake can be effectively reduced in our method,
And this method can detecte out that amplitude is smaller and the intermittent fault of limited duration, significantly improve between data-driven method pair
The detection effect for failure of having a rest.
Certainly, the above description is not a limitation of the present invention, and the present invention is also not limited to the example above, this technology neck
The variations, modifications, additions or substitutions that the technical staff in domain is made within the essential scope of the present invention also should belong to of the invention
Protection scope.
Claims (1)
1. a kind of bullet train intermittent fault detection method based on the multiple side's T control figure, it is characterised in that: including walking as follows
It is rapid:
Step 1: off-line training；Specifically comprise the following steps:
Step 1.1: construction nominal situation calculation matrix:
[X1,X2,…,Xk,…,XN]∈Rm×N；
Wherein, Xk∈Rm×1For column vector；M is the number of probes that system detected includes；The each sensor of N includes
Independent sample number；
Step 1.2: calculating the sample average of training dataWith covariance matrix S, i.e.,
Step 1.3: rule of thumb or by analysis of history fault data, providing the direction of bullet train critical system intermittent fault
ξq, failure amplitude lower boundTrouble duration lower boundWith failure vanishes time lower boundAnd calculate following formula:
Wherein, W*And W#For two time windows；W=1,2 ..., W#；For detection delay theoretically occurs to failure
The upper bound；For theoretically to the upper bound of failure vanishes detection delay；
Step 1.4: according to given confidence level α, to time window W=1,2 ..., W#, calculate separatelyI.e.
Wherein, FaIt is p that (p, N-p), which is freedom degree, and the upper quantile when F distribution confidence level of N-p is α remembers δ2=δ1 2；
Step 2: online fault detection；Specifically comprise the following steps:
Step 2.1: online acquisition newly measures sampleAnd to time window W=1,2 ..., W#, each side's T control is calculated in real time
The statistic of figureI.e.
Step 2.2: to time window W=1,2 ..., W#, the alarm moment of each side's the T control figure of real-time update: failure is reported
It is alertIt alarms with failure vanishesI.e.
Step 2.3: it is online to exclude wrong report, specifically comprise the following steps:
Step 2.3.1: if to W=W#, following formula is invalid
Then restart step 2；It is no to then follow the steps 2.4；
Step 2.3.2: W ∈ [W if it exists*,W#] make (1) formula invalid, then restart step 2；It is no to then follow the steps 2.4；
Step 2.3.3: W if it exists, W ' ∈ [W*,W#] following formula is set up
WhereinFor empty set, then restart step 2, it is no to then follow the steps 2.4；
Step 2.4: selection time window specifically comprises the following steps:
Step 2.4.1: W={ W is successively examined*..., 1 }, if (1) formula is invalid, set Wo=W+1；Otherwise Wo=1；
Step 2.4.2: W={ W is successively examined*,…,Wo, W ' ∈ (W, W if it exists#], so that (2) formula is set up, then reset Wo=W+
1；Step 2.4.3: to W={ W*,…,Wo, it calculates
Step 2.4.4: W={ W is successively examined*,…,Wo, W ' ∈ (W, W if it exists#] following formula is set up, then reset Wo=W+1；
Step 2.5: intermittent fault occurs and extinction time is inferred, specifically comprises the following steps:
Step 2.5.1: the final deduction μ of intermittent fault generation and extinction time is calculatedq,And νq,I.e.
Step 2.5.2: infer that this intermittent fault existsOccur in moment,It disappears in moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811250111.2A CN109325310B (en) | 2018-10-25 | 2018-10-25 | High-speed train intermittent fault detection method based on multiple T-square control diagram |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811250111.2A CN109325310B (en) | 2018-10-25 | 2018-10-25 | High-speed train intermittent fault detection method based on multiple T-square control diagram |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109325310A true CN109325310A (en) | 2019-02-12 |
| CN109325310B CN109325310B (en) | 2023-05-05 |
Family
ID=65261847
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811250111.2A Active CN109325310B (en) | 2018-10-25 | 2018-10-25 | High-speed train intermittent fault detection method based on multiple T-square control diagram |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109325310B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109855855A (en) * | 2019-03-13 | 2019-06-07 | 山东科技大学 | Bullet train closed loop brake system intermittent fault detection method |
| CN114664058A (en) * | 2022-01-29 | 2022-06-24 | 上海至冕伟业科技有限公司 | Integral fault early warning system and method for fire water system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0479718A (en) * | 1990-07-20 | 1992-03-13 | Kansai Electric Power Co Inc:The | Detecting method of intermittent ground fault |
| WO2010115474A1 (en) * | 2009-04-10 | 2010-10-14 | Areva T&D Uk Ltd | Method and system for transient and intermittent earth fault detection and direction determination in a three-phase median voltage electric power distribution system |
| CN104697804A (en) * | 2015-03-24 | 2015-06-10 | 清华大学 | Method and system for detecting and separating intermittent faults of train active suspension system |
| CN107703740A (en) * | 2017-07-10 | 2018-02-16 | 山东科技大学 | A kind of robust interval sensor fault diagnosis method of bullet train critical system |
-
2018
- 2018-10-25 CN CN201811250111.2A patent/CN109325310B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0479718A (en) * | 1990-07-20 | 1992-03-13 | Kansai Electric Power Co Inc:The | Detecting method of intermittent ground fault |
| WO2010115474A1 (en) * | 2009-04-10 | 2010-10-14 | Areva T&D Uk Ltd | Method and system for transient and intermittent earth fault detection and direction determination in a three-phase median voltage electric power distribution system |
| CN104697804A (en) * | 2015-03-24 | 2015-06-10 | 清华大学 | Method and system for detecting and separating intermittent faults of train active suspension system |
| CN107703740A (en) * | 2017-07-10 | 2018-02-16 | 山东科技大学 | A kind of robust interval sensor fault diagnosis method of bullet train critical system |
Non-Patent Citations (1)
| Title |
|---|
| 鄢镕易;何潇;周东华: "一类存在参数摄动的线性随机系统的鲁棒间歇故障诊断方法" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109855855A (en) * | 2019-03-13 | 2019-06-07 | 山东科技大学 | Bullet train closed loop brake system intermittent fault detection method |
| CN114664058A (en) * | 2022-01-29 | 2022-06-24 | 上海至冕伟业科技有限公司 | Integral fault early warning system and method for fire water system |
| CN114664058B (en) * | 2022-01-29 | 2023-08-18 | 上海至冕伟业科技有限公司 | Overall fault early warning system and method for fire fighting water system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109325310B (en) | 2023-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106006344B (en) | Staircase On-line Fault early warning system and method for diagnosing faults | |
| CN103853152B (en) | A kind of batch process fault monitoring method based on AR-PCA | |
| CN103178615B (en) | Electrical equipment fault monitoring method and system thereof | |
| CN105259895B (en) | A kind of detection of industrial process small fault and separation method and its monitoring system | |
| CN109886475A (en) | The information security Situation Awareness System of metering automation system based on AI | |
| CN103713628B (en) | Fault diagnosis method based on signed directed graph and data constitution | |
| CN106338406A (en) | On-line monitoring and fault early-warning system and method for traction electric transmission system of train | |
| CN109858104A (en) | A kind of rolling bearing health evaluating and method for diagnosing faults and monitoring system | |
| CN105868629B (en) | Security threat situation assessment method suitable for electric power information physical system | |
| CN109213121A (en) | A kind of fan braking system card cylinder method for diagnosing faults | |
| CN110163075A (en) | A kind of multi-information fusion method for diagnosing faults based on Weight Training | |
| CN107272651B (en) | A kind of Robust Detection Method of Braking System for Multiple Units interval multiplying property sensor fault | |
| CN111474919B (en) | Aeroengine control system sensor fault diagnosis method based on AANN network group | |
| CN109325310A (en) | A kind of bullet train intermittent fault detection method based on the multiple side's T control figure | |
| CN106446339B (en) | Rotation Mechanical Running Condition method for detecting abnormality based on Di Li Cray mixed model | |
| CN106679847A (en) | Electric power equipment fault diagnosing method and apparatus | |
| CN109489931A (en) | A kind of abnormal impact real-time detection method | |
| CN112632845B (en) | Data-based mini-reactor online fault diagnosis method, medium and equipment | |
| CN108491762A (en) | A kind of detection method and device that human body is fallen | |
| CN103400040A (en) | Fault diagnosis and prediction method utilizing multistep time domain difference value learning | |
| CN111776006B (en) | Pre-alarming method and device for train axle temperature | |
| CN106647650A (en) | Distributed industrial process monitoring method based variable weighting PCA (Principal Component Analysis) model | |
| CN114519923A (en) | Intelligent diagnosis and early warning method and system for power plant | |
| CN108345711A (en) | Based on event driven EMU robust sensor intermittent fault detection method | |
| CN115600067A (en) | Building structure deformation monitoring system and method based on Internet of things |
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 |