CN105651323A - Active prediction method for soft fault of sensor based on GM (1, 1) model - Google Patents
Active prediction method for soft fault of sensor based on GM (1, 1) model Download PDFInfo
- Publication number
- CN105651323A CN105651323A CN201410717102.5A CN201410717102A CN105651323A CN 105651323 A CN105651323 A CN 105651323A CN 201410717102 A CN201410717102 A CN 201410717102A CN 105651323 A CN105651323 A CN 105651323A
- Authority
- CN
- China
- Prior art keywords
- grey
- sensor
- sensing data
- automobile sensor
- automobile
- 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
Landscapes
- Combined Controls Of Internal Combustion Engines (AREA)
Abstract
The invention relates to an active prediction method for soft fault of a sensor based on a GM (1, 1) model, and belongs to the technical field of sensor soft fault detection. An active prediction manner is used to predict the precision failure trend in advance, unknown faults in an engine and an automobile control system caused by precision failure of the sensor are avoided, precautions are taken against calamity, the utilization rate of fuel oil and the working efficiency of other control systems are improved, the characteristic that shot-term accurate prediction can be achieved via a small amount of data of a gray system is used, the operation stability of an automobile system is ensured, and dynamic response is rapid and accurate.
Description
Technical field
The present invention relates to a kind of soft fault active predicting method of sensor based on GM (1,1) model, belong to sensor soft-defect detection field.
Background technology
Automobile sensor promotes that automobile is superior, one of gordian technique of electronization and intelligence automation development, general automobile is equipped with tens or nearly hundred sensors at present, and senior luxury car approximately uses two or three hundred, can they normally run is the key that can automobile normally travel, so Transducer fault detection and diagnosis are just seemed very important. sensor fault is mainly divided into hard fault and soft fault, hard fault randomness is stronger, also more difficult prediction, and the soft fault mainly aging precise decreasing etc. caused of sensor, cause signal distortion. with regard to the oxygen sensor that automobile engine control system is relevant, intake air temperature sensor, throttle opening sensor, bent axle position and tachogenerator, exhaust gas temperature sensor, detonation sensor, vehicle speed sensor, coolant water temperature sensor etc. is example, these sensors all play most important effect in fuel charge control and ignition control etc., any link loses precision, the decline of enngine control system control accuracy is caused in capital, cause automobile before not arriving the turnaround, fuel oil consumption improves, wattful power declines, even cannot start time serious. through retrieval and investigation, what the current method for automobile sensor soft defect detection mostly adopted is passive detection, namely fault just can carry out detecting or regularly arriving specialized maintenance Spot detection according to maintenance instruction after occurring.
In sum, also there is following problem in Hyundai Motor field Transducer fault detection:
Take the strategy of regularly fixed point maintenance, carry out the investigation of the soft fault of sensor, cause the seriously delayed of reply catastrophic failure response, there is potential safety hazard, refer more particularly to enngine control system, indirectly cause the unnecessary consumption of fuel oil and the decline of wattful power.
Summary of the invention
The technical problem to be solved in the present invention is: the some shortcomings existed for the automobile sensor soft defect detection existed in prior art, especially the enngine control system deficiency existed in sensor soft defect detection, propose the soft fault active predicting method of sensor of a kind of GM based on gray system theory (1,1) model.
The technical solution adopted for the present invention to solve the technical problems is:
Based on the sensor soft fault active predicting side of GM (1,1) model, it is characterised in that:
Step is as follows:
1.1 determine the sensing data of cycle continuous sampling automobile sensor, whereinFor the unit periodic sampling moment;
1.2 for the sensing data sequence of automobile sensorCarry out the grey one-accumulate formation sequence of the sensing data sequence of a grey Accumulating generation acquisition automobile sensor
Wherein,For the total number of unit periodic sampling, and;
1.3 for the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor
Carry out the average generation of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor, obtain the average generation sequence of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor
Wherein, the general expression formula of average generation is;
1.4 set up automobile sensor soft fault grey GM (1,1) model based on the sensing data of automobile sensor, and concrete expression formula is:
;
Wherein ash actuating quantityConcrete expression formula as follows:
,
;
1.5 obtain the initial active predicting grey derivative value of automobile sensor sensing data next cycle according to soft fault grey GM (1,1) model of the automobile sensor in step 1.4, and once obtained the sensing data actual prediction value of automobile sensor against Accumulating generation by grey, concrete expression formula is:
;
1.6 predicting the outcome according to step 1.5, compare actual to itself and next cycle initial sampled value, if it exceeds standard setting error, then can judge that automobile sensor exists precision failure tends, should overhaul in time.
Compared with prior art, the present invention is based on the soft fault active predicting method of sensor of GM (1,1) model, and the useful effect having is:
The present invention based on GM (1,1) what the soft fault active predicting method of the sensor of model was implemented is active predicting, look-ahead precision failure tends, avoid the unknown engine because causing after sensor accuracy inefficacy and automotive control system internal fault, accomplish to prevent trouble before it happens, improve the working efficiency of fuel utilization ratio and other control module, and the feature of short-term accurately predicting just can be carried out in conjunction with a small amount of data of gray system, ensure that the stability of automobile system cloud gray model, dynamic response rapidity and accuracy.
Accompanying drawing explanation
Fig. 1 is the FB(flow block) of the present invention based on the soft fault active predicting method of sensor of GM (1,1) model.
Fig. 2 is the design sketch of automobile sensor fault active predicting of the present invention.
Embodiment
Below in conjunction with Fig. 1-2, the soft fault active predicting method of the sensor based on GM (1,1) model of the present invention is described in further detail.
Being illustrated in figure 1 FB(flow block) of the present invention, concrete steps are as follows:
Step 1: the sensing data determining cycle continuous sampling automobile sensor, whereinFor the unit periodic sampling moment, determining the cycle is 200 ~ 500ms.
Step 2: for the sensing data sequence of automobile sensorCarry out the grey one-accumulate formation sequence of the sensing data sequence of a grey Accumulating generation acquisition automobile sensor:
;
Wherein,For the total number of unit periodic sampling, and, the concrete computing expression formula that grey one-accumulate generates is:
;
Step 3: the grey one-accumulate formation sequence for the sensing data sequence of automobile sensor:
;
Carry out the average generation of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor, obtain the average generation sequence of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor:
;
Wherein, the general expression formula of average generation is��
Step 4: setting up automobile sensor soft fault grey GM (1,1) model based on the sensing data of automobile sensor, concrete expression formula is:
Wherein ash actuating quantityConcrete expression formula as follows
,
;
Step 5: obtain the initial active predicting grey derivative value of automobile sensor sensing data next cycle according to soft fault grey GM (1,1) model of the automobile sensor in step 4, and once obtained the sensing data actual prediction value of automobile sensor against Accumulating generation by grey, concrete expression formula is:
��
Step 6: according to predicting the outcome in step 5, compares actual to itself and next cycle initial sampled value, if it exceeds standard setting error, then can judge that automobile sensor exists precision failure tends, should overhaul in time.
Embodiment:
Embodiment of the present invention Object Selection be that Nanfeng Electromechanical Equipment Manufacturing Co., Ltd., Luoyang manufactures and produces supporting petrol engine test-bed, this equipment can pass through current vortex dynamometer machine, the various driving cycle of accurate simulation.
Below in conjunction with the temperature of cooling water sensor that the present invention is directed to embodiment object engine, the present invention is elaborated based on the soft fault active predicting method seat of automobile sensor of gray system theory, it should be noted that the accuracy in order to embodiment, adopt two temperature of cooling water sensor, one of them can pass through potentiometer analog distortion phenomenon, another is as reference value, and the maximum induction error of setting sensor is �� 3%:
The first step, unit period continuous sampling temperature of cooling water data amount check is 5, and the temperature of cooling water sensing data of acquisition is as shown in table 1:
2nd step, the temperature of cooling water sensing data for table 1 carries out grey one-accumulate generation, and according to the formula that one-accumulate generatesObtain the grey one-accumulate formation sequence of temperature of cooling water sensing data, as shown in table 2:
3rd step, according to formula, the data in associative list 2, calculate the average generation sequence of the grey one-accumulate formation sequence of temperature of cooling water sensing data:
As shown in table 3:
4th step, sets up grey GM (1,1) model of the sensing data of temperature of cooling water, and concrete expression formula is:
According to formula
,
;
The grey actuating quantity of Confirming model is respectively, carry it into and model obtain subsequent time coolant water temperature grey Derivative prediction value, it is 99.1682%. by the precision of forecasting model of residual test
Step 5: according to formulaThe subsequent time actual water temperature obtaining temperature of cooling water is��
Step 6: the initial actual coolant water temperature sampled value (potentiometer simulated sensors) of next cycle is 82, itself and predictorAbsolute difference exceed the permission work error of default �� 3%, and the sampled value combining actual temperature of cooling water sensor (without potentiometer simulated sensors) is the fact of 91, it can be seen that the present invention well predicts this phenomenon of failure tends.
It is illustrated in figure 2 the automobile sensor fault active predicting design sketch of the present invention based on the soft fault active predicting method of automobile sensor of gray system theory, in Fig. 2, data are the temperature of cooling water sensor actual effect test in another sampling period, as can be seen from Figure 2 the prediction effect of the present invention is ideal, sampling distortion trend is obviously there is in data sampling point when arriving the 5th point, the present invention well predicts this kind of trend, from 6th o'clock to the sampling of the 10th point, more confirm to predict the trend lost efficacy, accomplish that active detecting suppresses the generation of failure tends in advance, the predicated error of the present invention is within 0.75%.
The above is only the better embodiment of the present invention, is not the restriction that the present invention does other form, and any those skilled in the art may utilize the technology contents of above-mentioned announcement to be changed or be modified as the equivalent embodiment of equivalent variations. In every case being do not depart from technical solution of the present invention content, any simple modification, equivalent variations and the remodeling above embodiment done according to the technical spirit of the present invention, still belongs to the protection domain of technical solution of the present invention.
Claims (1)
1. based on the soft fault active predicting method of sensor of GM (1,1) model, it is characterised in that:
Step is as follows:
1.1 determine the sensing data of cycle continuous sampling automobile sensor, whereinFor the unit periodic sampling moment;
1.2 for the sensing data sequence of automobile sensorCarry out the grey one-accumulate formation sequence of the sensing data sequence of a grey Accumulating generation acquisition automobile sensor
Wherein,For the total number of unit periodic sampling, and;
1.3 for the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor
Carry out the average generation of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor, obtain the average generation sequence of the grey one-accumulate formation sequence of the sensing data sequence of automobile sensor
Wherein, the general expression formula of average generation is;
1.4 set up automobile sensor soft fault grey GM (1,1) model based on the sensing data of automobile sensor, and concrete expression formula is:
Wherein ash actuating quantityConcrete expression formula as follows
,
1.5 obtain the initial active predicting grey derivative value of automobile sensor sensing data next cycle according to soft fault grey GM (1,1) model of the automobile sensor in step 1.4, and once obtained the sensing data actual prediction value of automobile sensor against Accumulating generation by grey, concrete expression formula is:
;
1.6 predicting the outcome according to step 1.5, compare actual to itself and next cycle initial sampled value, if it exceeds standard setting error, then can judge that automobile sensor exists precision failure tends, should overhaul in time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410717102.5A CN105651323A (en) | 2014-12-02 | 2014-12-02 | Active prediction method for soft fault of sensor based on GM (1, 1) model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410717102.5A CN105651323A (en) | 2014-12-02 | 2014-12-02 | Active prediction method for soft fault of sensor based on GM (1, 1) model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105651323A true CN105651323A (en) | 2016-06-08 |
Family
ID=56480585
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410717102.5A Pending CN105651323A (en) | 2014-12-02 | 2014-12-02 | Active prediction method for soft fault of sensor based on GM (1, 1) model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105651323A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256684A (en) * | 2018-01-16 | 2018-07-06 | 安徽理工大学 | A kind of Seepage Prediction method based on chemicla plant |
| CN112013878A (en) * | 2020-09-04 | 2020-12-01 | 南京理工大学 | Star sensor on-line calibration, correction and error reporting method based on gray model |
-
2014
- 2014-12-02 CN CN201410717102.5A patent/CN105651323A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256684A (en) * | 2018-01-16 | 2018-07-06 | 安徽理工大学 | A kind of Seepage Prediction method based on chemicla plant |
| CN112013878A (en) * | 2020-09-04 | 2020-12-01 | 南京理工大学 | Star sensor on-line calibration, correction and error reporting method based on gray model |
| CN112013878B (en) * | 2020-09-04 | 2022-06-24 | 南京理工大学 | Star sensor on-line calibration, correction and error reporting method based on gray model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103776654B (en) | The method for diagnosing faults of multi-sensor information fusion | |
| CN106202635B (en) | A kind of bullet train dynamic shaft temperature prediction technique based on multivariate regression models | |
| CN105966382B (en) | The height above sea level evaluation method and device of electric car and its local environment | |
| US10296674B2 (en) | Method for a vehicle | |
| JP2014511970A5 (en) | ||
| WO2011020063A3 (en) | Apparatus, system, and method for adaptive engine system control with integrated global position sensing | |
| US10317312B2 (en) | Method for reducing vibrations in a test bed | |
| CN107002578A (en) | Method and apparatus for determining boost torque | |
| CN106647288A (en) | Method for estimating indicating torque of engine based on nonsingular terminal sliding mode observer | |
| Wang et al. | Real-time misfire detection via sliding mode observer | |
| CN113176739A (en) | Vehicle control device, vehicle control method, and non-transitory computer readable medium storing vehicle control program | |
| CN105240123B (en) | The engine RPM monitoring method and ERM controller of use priority mode | |
| CN105651323A (en) | Active prediction method for soft fault of sensor based on GM (1, 1) model | |
| CN102109363A (en) | Method for actively predicting soft fault of sensor for automobile based on grey system prediction theory | |
| CN108414230B (en) | A kind of internal combustion engine preignition detecting method | |
| KR102185572B1 (en) | Method and device for performing a reference measurement on a sensor of an internal combustion engine | |
| WO2009060155A3 (en) | Method and device for diagnosing the state of a moving-vane control system for the turbine of a turbocharger | |
| Malikopoulos et al. | Optimal engine calibration for individual driving styles | |
| CN112096532B (en) | Method and system for limiting fuel injection quantity | |
| CN110529278B (en) | Engine misfire detection method, device and system | |
| Wan et al. | Engine modelling architecture study for hybrid electric vehicle diagnosis application | |
| US11339703B2 (en) | Engine emission prediction system | |
| Chung et al. | Real-time empirical NO x model based on In-cylinder pressure measurements for light-duty diesel engines | |
| KR102172012B1 (en) | Railway vehicle safety component diagnosis apparatus | |
| CN201429503Y (en) | Fault diagnostic instrument for main bearings of locomotive diesel engines |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160608 |
|
| WD01 | Invention patent application deemed withdrawn after publication |