CN109253826B - Heat meter residual life prediction method based on multi-degradation sample data fusion - Google Patents
Heat meter residual life prediction method based on multi-degradation sample data fusion Download PDFInfo
- Publication number
- CN109253826B CN109253826B CN201810863794.2A CN201810863794A CN109253826B CN 109253826 B CN109253826 B CN 109253826B CN 201810863794 A CN201810863794 A CN 201810863794A CN 109253826 B CN109253826 B CN 109253826B
- Authority
- CN
- China
- Prior art keywords
- performance degradation
- rul
- heat meter
- heat
- dci
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K19/00—Testing or calibrating calorimeters
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses a method for predicting the remaining life of a heat meter based on multi-degradation sample data fusion, which comprises the steps of obtaining performance degradation data of each heat meter, establishing a performance degradation index-remaining life sequence set R of N heat meters, using an One-class SVM to perform clustering processing on the performance degradation index-remaining life sequence set R of the N heat meters, using a SVR to establish a regression model between the degradation index and the remaining life, determining the remaining life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining the regression model, and realizing the prediction of the remaining life of the heat meter. The invention has reliable analysis result, can promote the improvement of the reliability of the heat meter, and can provide a reference means for the quality verification of the heat meter in China.
Description
Technical Field
The invention belongs to the technical field of equipment safety, and particularly relates to a method for predicting the residual life of a heat meter based on multi-degradation sample data fusion.
Background
The calorimeter is used as a measuring instrument and is the basis for realizing accurate measurement of a heating system. The performance change of the heat meter not only affects the service life of the heat meter, but also has great influence on a heat metering system in which the heat meter is positioned. The method has the advantages that the residual life prediction precision of the heat meter is improved, the heat meter can be guaranteed to stably run for a long time in the using process, the problems of the heat meter can be further determined, corresponding improvement measures are taken aiming at the problems to improve the reliability of the heat meter, the reliability of the heat meter is promoted to be improved, and reference can be provided for quality verification of the heat meter in China. Therefore, the residual life prediction technology of the heat meter needs to be researched, more and more accurate references are provided for maintenance and repair of the heat meter, stable and reliable operation of the heat meter is guaranteed, and the method has very important practical significance for more accurate residual life prediction of the heat meter.
Currently, few researches are carried out on the service life prediction problem of the heat meter, the service life analysis of the conventional heat meter is mainly determined by a durability test according to the European standard, namely that the 2400h durability test is generally considered to be equivalent to the service life of 5 years under the actual working condition, the 300h additional durability test is equivalent to the service life of 3 years under the working condition, and the heat meter passing the 2400h +300h durability test can at least meet the service life of 8 years under the normal working condition. However, the durability test can only determine whether the life of the heat meter is 8 years, and only rough interval estimation is performed on the remaining life, so that the remaining life cannot be accurately estimated. In addition, although actual case data of the calorimeter performance degradation are currently accumulated, how to establish a performance degradation evaluation benchmark of the calorimeter according to a plurality of degradation data is also a problem which is not solved by the existing method, and the residual life prediction requirement based on the actual case data of the calorimeter performance degradation is difficult to meet.
Disclosure of Invention
The invention aims to solve the technical problem that the method for predicting the remaining service life of the heat meter based on the fusion of multiple degradation sample data is provided aiming at the defects in the prior art, and the problem of low accuracy of predicting the service life of the heat meter is solved on the basis of a large amount of performance degradation actual case data.
The invention adopts the following technical scheme:
a method for predicting the remaining life of a heat meter based on multi-degradation sample data fusion comprises the steps of obtaining performance degradation data of each heat meter, establishing a performance degradation index-remaining life sequence set R of N heat meters, using an One-class SVM to perform clustering processing on the performance degradation index-remaining life sequence set R of the N heat meters, using a SVR to establish a regression model between the degradation index and the remaining life, determining the remaining life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining the regression model, and achieving the remaining life prediction of the heat meter.
Specifically, the method comprises the following steps:
s1, based on the performance degradation monitoring sample test data of N heat meters, giving the performance degradation index value of the heat meterAnd its failure index valueIntercepting the test data of each heat meter to obtain the performance degradation data of each heat meter;
s2, combining the performance degradation data of each heat meter obtained in the step S1 with the residual life time sample, and establishing a performance degradation index-residual life sequence set R of N heat meters;
s3, clustering the performance degradation index-residual life sequence set R of the N heat meters by using One-class SVM, and acquiring the performance degradation reference R of the N heat meters in statistical senseDCI-rul；
S4, performing SVR training on the M groups of sample pairs, and establishing a remaining life prediction model rul ═ f (DCI) of the calorimeter;
s5, determining the remaining life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining with the remaining life prediction model rul ═ f (dci).
Further, in step S1, letA performance degradation index sequence corresponding to the performance degradation data of the kth heat meter in the track library, whereinFor the heat meter at tiPerformance degradation index value corresponding to time, performance degradation index valueAnd a failure index valueOf performance degradation data set S'kThe following were used:
wherein, tDAnd tFRespectively the time when the heat meter starts to degrade and fail,andis the corresponding index value.
Further, in step S2, R is a mapping set of performance degradation indexes and remaining lives obtained from performance degradation data of the N heat meters, and the performance degradation index-remaining life sequence set R of the N heat meters is as follows:
R＝{R1,R2,...,Rk,...,RN}
wherein R iskIs a performance degradation index-residual life sequence of the heat meter k.
Further, the degradation index is DIiThe corresponding remaining life was set to ruliAnd defining the residual life of the heat meter when the heat meter fails as 0, and obtaining a performance degradation index-residual life sequence corresponding to the heat meter k as follows:
wherein i is the time tiIn Performance degraded data set S'kWherein n' is the performance degradation index and the corresponding number of the remaining life in the mapping relation of the performance degradation index and the remaining life corresponding to the heat meter k, (DI)i,ruli) Represents tiPerformance degradation indexes of the heat meter at the moment and corresponding residual service life of the heat meter.
Further, step S3 is specifically: let rulpFor a remaining life point in the set R, the remaining life rul is obtained from the set RpCorresponding set DIE of N performance degradation index valuesp(ii) a Then the residual life rul is obtainedpSet DI of corresponding performance degradation index valuespPerforming clustering analysis on the obtained performance degradation index value and residual service life by using One-class SVM algorithm to obtain a mapping set RDCI-rul，。
Further, the cluster analysis comprises the following specific steps:
DIE set by using One-class SVM algorithmpPerforming cluster analysis, removing abnormal points in the set, and setting the performance degradation index set after screening as follows:
DIEp′＝{DIE1′,DIE2′,...,DIEj′,...,DIEm′}
wherein j is 1,2, and m is set DIEp' the number of elements;
obtaining a set DIEpThe cluster centers of' are as follows:
wherein j is 1,2, and m is set DIEp' the number of elements;
processing the remaining life sequence R of the heat meter to obtain a mapping set R of a performance degradation index value and the remaining lifeDCI-rul，Namely, the performance degradation criteria in the statistical sense of the N heat meters are as follows:
RDCI-rul＝{(DCI1,rul1),...,(DCIq,rulq),...,(DCIn′,ruln′)}
wherein q is 1, 2.
Further, aggregate DIEpThe following were used:
wherein, DIEjRul for remaining lifepThe corresponding j th heat meter performance degradation index value, N is the set DIEpThe number of the performance degradation indexes is n', the performance degradation indexes and the corresponding surplus in the mapping sequence of the heat meter k performance degradation indexes and the surplus service life are nNumber of life points.
Further, in step S4, it is assumed that at time tiThe performance degradation index value of the heat meter is DCIiCorresponding to a remaining life of ruliAccording to DCIiAnd ruliThe mapping relationship between the two training samples is as follows to obtain M groups of training sample pairs of the SVR prediction model:
DCI＝[DCI1 DCI2 ... DCIi ... DCIM]T
rul＝[rul1 rul2 ... ruli ... rulM]T
wherein, M is the number of samples, DCI is input, rul is output, and after SVR training, the remaining life prediction model rul ═ f (DCI) of the calorimeter is obtained.
Compared with the prior art, the invention has at least the following beneficial effects:
the invention discloses a method for predicting the remaining life of a heat meter based on multi-degradation sample data fusion, which is characterized by obtaining performance degradation data of each heat meter, establishing a performance degradation index-remaining life sequence set R of N heat meters, clustering the performance degradation index-remaining life sequence set R of the N heat meters by using One-class SVM, establishing a regression model between the degradation index and the remaining life by using SVR, determining the remaining life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining the regression model, and realizing the remaining life prediction of the heat meter. The method has reliable analysis result, can promote the improvement of the reliability of the heat meter, and can provide a reference means for the quality verification of the heat meter in China.
Further, step S1 may obtain performance degradation data of each heat meter, facilitating subsequent data analysis.
Further, in step S2, a set R of performance degradation indicators — remaining lifetime sequences of the N heat meters can be obtained, which is used as a basis for the next clustering process and facilitates the establishment of a benchmark.
Further, step S3 comprehensively considers individual differences of the heat meters to perform clustering processing on the performance degradation index-remaining life sequence sets R of the N heat meters, so as to obtain a performance degradation reference RDCI-rul in the statistical sense of the N heat meters, and to base subsequent model establishment.
Further, step S4 trains the sample by using SVR to obtain the remaining life prediction model rul ═ f (dci), and the training time can be reduced by combining step S3 on the premise of ensuring the accuracy.
In conclusion, the method can obtain the performance degradation data of each heat meter, is convenient for subsequent data analysis, can obtain the performance degradation indexes-residual life sequence sets R of the N heat meters, is used as a basis for next clustering processing, is convenient for establishing a benchmark, comprehensively considers the individual difference of each heat meter to cluster the performance degradation indexes-residual life sequence sets R of the N heat meters, and can obtain the performance degradation benchmark R in the statistical significance of the N heat metersDCI-rulBased on the establishment of the subsequent model, the SVR is used to train the sample to obtain the remaining life prediction model rul ═ f (dci), and in combination with step S3, the training time can be reduced on the premise of ensuring the accuracy, so that the possibility is provided for realizing real-time remaining life prediction.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a flow chart of the operation of the present invention;
fig. 2 is a diagram of a remaining life prediction result of the heat meter 50216151;
fig. 3 is a diagram showing the remaining life prediction result of the heat meter 50215885.
Detailed Description
The Degradation Index (DI) is a reference which is established for researching the performance Degradation degree of the equipment and quantitatively describes the performance Degradation degree of the equipment, and is widely applied to the field of equipment performance Degradation and residual life prediction.
The Support Vector Machine (SVM) is a powerful tool for realizing the principle of minimizing the structural risk developed under a statistical learning theory system, mainly realizes the principle of minimizing the structural risk by keeping an empirical risk value fixed and minimizing a confidence range, and is suitable for learning small samples. The SVM is mainly applied to the field of pattern recognition (classification), and Support Vector Regression (SVR) is obtained by expanding on the basis of SVM theory, and is widely applied to the field of prediction as a Regression algorithm under the condition of known results.
A class of support vector machines (One-class SVM) is developed on the basis of SVM theory and is a method for single cluster analysis. The minimum support area containing most samples in the sample set can be obtained through an One-class SVM algorithm, and the method is as follows:
1) mapping data of an original space into a high-dimensional feature space by adopting a Gaussian radial basis kernel function, wherein the expression of the Gaussian radial basis kernel function is as follows:
wherein, K (x)i,xj) Is a Gaussian radial basis kernel function, xi,xjFor a set of variables in space, P is the width of the radial basis function.
2) Constructing a hyperplane, separating the sample data from the origin in the high-dimensional feature space as much as possible, namely converting one kind of problem in the original sample space into two kinds of Support Vector Machine (SVM) problems in the high-dimensional feature space, and calling the region containing most target sample points in the sample set at one side of the hyperplane as a normal region, wherein the sample points are called 'positive samples' and marked as '1', and the sample points at the other side of the hyperplane are called 'negative samples' and marked as '1'.
According to the invention, a regression model between the degradation index and the residual life is established by using the SVR, so that the residual life of the heat meter is predicted.
Referring to fig. 1, the method for predicting the remaining life of a heat meter based on multi-degradation sample data fusion according to the present invention includes the following steps:
s1, based on the performance degradation monitoring sample test data of N heat meters, giving the performance degradation index of the heat metersValue ofAnd its failure index valueIntercepting the test data of each heat meter to obtain the performance degradation data of each heat meter;
is provided withA performance degradation index sequence corresponding to the performance degradation data of the kth heat meter in the track library, whereinFor the heat meter at tiThe performance degradation index value corresponding to the time. According to a given performance degradation index valueAnd a failure index valueIntercepting the performance degradation data between the two to obtain:
of formula (II) S'kFor a performance-degrading data set, tDAnd tFThe time when the heat meter starts to degrade and fail, respectivelyAndis the corresponding index value.
S2, on the basis of the step S1, combining the residual life time samples to establish a performance degradation index-residual life sequence set R of the N heat meters, which is specifically as follows:
and (5) converting the formula (8) to obtain the corresponding relation between the residual life of the heat meter and the degradation index value. Index of degradation as DIiThe corresponding remaining life was set to ruliAnd defining the residual life of the heat meter when the heat meter fails as 0, and obtaining a performance degradation index-residual life sequence corresponding to the heat meter k, namely:
in which i is the time tiIn Performance degraded data set S'kWherein n' is the performance degradation index and the corresponding number of the remaining life in the mapping relation of the performance degradation index and the remaining life corresponding to the heat meter k, (DI)i,ruli) Represents tiPerformance degradation index of time calorimeter and its corresponding residual life, RkIs a performance degradation index-residual life sequence of the heat meter k.
Establishing a performance degradation index-residual life sequence set R ═ { R } of N heat meters1,R2,...,Rk,...,RNR is a mapping set of performance degradation indexes and residual lives obtained from performance degradation data of N heat meters, wherein R iskAs shown in equation (9), the mapping sequence of the performance degradation index and the remaining life of the heat meter k is shown.
S3, on the basis of the step S2, clustering the performance degradation indexes-residual life sequence set R of the N heat meters by using the One-class SVM, and acquiring the performance degradation reference R of the N heat meters in the statistical senseDCI-rulThe specific method comprises the following steps:
let rulpFor a remaining life point in the set R, to implement the cluster analysis, it is first necessary to obtain rul the remaining life from the set RpCorresponding set DIE of N performance degradation index valuesp:
In the formula, DIEjRul for remaining lifepThe corresponding j th heat meter performance degradation index value, N is the set DIEpThe number of the performance degradation indexes is n ', and the n' is the performance degradation indexes in the mapping sequence of the heat meter k performance degradation indexes and the residual service life and the number of corresponding residual service life points.
The remaining life rul is obtained according to the formula (10)pSet DI of corresponding performance degradation index valuespAnd performing cluster analysis on the data by using an One-class SVM algorithm, wherein the method comprises the following specific steps:
(a) DIE set by using One-class SVM algorithmpPerforming cluster analysis, removing abnormal points in the set, and setting the performance degradation index set after screening as follows:
DIEp′＝{DIE1′,DIE2′,...,DIEj′,...,DIEm′}
wherein j is 1,2, and m is set DIEp' the number of elements;
(b) obtaining a set DIEp' clustering center:
processing the remaining life sequence R of the heat meter to obtain a mapping set R of a performance degradation index value and the remaining lifeDCI-rul，Namely, the performance degradation benchmark in the statistical sense of the N heat meters:
RDCI-rul＝{(DCI1,rul1),...,(DCIq,rulq),...,(DCIn′,ruln′)}
wherein q is 1, 2.
S4, on the basis of step S3, performing SVR training on M groups of sample pairs to establish a remaining life prediction model rul ═ f (dci) for the calorimeter, specifically as follows:
suppose that at time tiThe performance degradation index value of the heat meter is DCIiCorresponding to a remaining life of ruli. According to DCIiAnd ruliA mapping relation between them, canObtaining M groups of training sample pairs of the SVR prediction model, which is shown as the following formula:
DCI＝[DCI1 DCI2 ... DCIi ... DCIM]T (12)
rul＝[rul1 rul2 ... ruli ... rulM]T (13)
in the formula, M is the number of samples, and the remaining life prediction model rul ═ f (dci) of the calorimeter is obtained by SVR training using formula (12) as an input and formula (13) as an output.
And S5, on the basis of the step S4, determining the residual life rul of the heat meter to be predicted at the current moment by using the performance degradation data of the heat meter to be predicted at the current moment and combining a residual life prediction model rul ═ f (DCI).
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
From 12 months in 2017, 4000 times of cold and heat shock tests of the heat meters are carried out by a cooperative unit measuring institute in Shaanxi province, the heat meters used in each test are from the same manufacturer, and the models of the heat meters are the same. When 4000 times of cold and hot impact tests are carried out, the data acquisition period is set to be 15s, the cycle time of each cold and hot impact is 5min, the test lasts for about 333.3 hours, and about 80000 points can be obtained on each tested table. In 4 tests, 17 heat meters fail, wherein 3 heat meters have water leakage, and the durability indication errors of the other 14 heat meters exceed the standard error (namely the indication errors exceed 2MPE, or the variation of the errors before and after the test exceeds 1.5 MPE).
The heat meters 50216151 and 50215885 are selected as test objects, and each 600 points are taken as a sample, so that 133 samples can be obtained respectively. And (4) carrying out the same processing on the rest 13 heat meters, taking the rest 13 heat meters as historical degradation tracks of the heat meters, intercepting the parts used for residual life prediction in the degradation tracks, and carrying out clustering processing on the parts. And establishing a residual life prediction model of the heat meter by using the SVR on the basis of the clustered heat meter performance degradation indexes and the mapping sequence of the residual life. For the heat meter 50216151, the residual life thereof is predicted from the 64 th sample (starting point of prediction of residual life), and for the heat meter 50215885, the prediction of residual life is started from the 66 th sample, as shown in fig. 2, the result of prediction of residual life of the heat meter 50216151 is shown, and as shown in fig. 3, the result of prediction of residual life of the heat meter 50215885 is shown.
Table 1 shows the corresponding residual life prediction error:
the horizontal axis of each of fig. 2 and 3 is a sample point, the vertical axis is the remaining life of the heat meter, the broken line in the graph is the predicted value of the remaining life of each sample of the heat meter, and the straight line is the actual value of the remaining life of the heat meter. As can be seen from fig. 2 and 3, the predicted value of the remaining life of the heat meter shows a decreasing trend, and the decreasing trend coincides with the actual remaining life thereof, and the error between the predicted value and the actual value of the remaining life of the heat meter is shown in table 1.
In summary, the method for predicting the remaining life of the heat meter based on the fusion of the multiple degradation sample data, which is provided by the invention, can predict the remaining life of the heat meter to a certain extent, reduce training time on the premise of ensuring precision, and provide possibility for realizing real-time remaining life prediction.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.
Claims (4)
1. A method for predicting the residual life of a heat meter based on multi-degradation sample data fusion is characterized by comprising the following steps of obtaining performance degradation data of each heat meter, establishing a performance degradation index-residual life sequence set R of N heat meters, using an One-class SVM to perform clustering processing on the performance degradation index-residual life sequence set R of the N heat meters, using a SVR to establish a regression model between a degradation index and a residual life, determining the residual life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining the regression model, and realizing the residual life prediction of the heat meter, wherein the method comprises the following steps:
s1, based on the performance degradation monitoring sample test data of N heat meters, giving the performance degradation index value of the heat meterAnd its failure index valueIntercepting the test data of each heat meter to obtain the performance degradation data of each heat meter;
s2, combining the performance degradation data of each heat meter obtained in the step S1 with the residual life time sample, and establishing a performance degradation index-residual life sequence set R of the N heat meters, wherein the R is a mapping set of the performance degradation index and the residual life obtained from the performance degradation data of the N heat meters, and the performance degradation index-residual life sequence set R of the N heat meters is as follows:
R＝{R1,R2,...,Rk,...,RN}
wherein R iskIs a performance degradation index-residual life sequence of the heat meter k, and the degradation index is DIiThe corresponding remaining life was set to ruliCalorimeterThe remaining life when the heat meter k fails is defined as 0, and the performance degradation index-remaining life sequence corresponding to the obtained heat meter k is as follows:
wherein i is the time tiIn Performance degraded data set S'kWherein n' is the performance degradation index and the corresponding number of the remaining life in the mapping relation of the performance degradation index and the remaining life corresponding to the heat meter k, (DI)i,ruli) Represents tiPerformance degradation indexes of the heat meter at the moment and corresponding residual life of the heat meter;
s3, clustering the performance degradation index-residual life sequence set R of the N heat meters by using One-class SVM, and acquiring the performance degradation reference R of the N heat meters in statistical senseDCI-rul(ii) a Let rulpFor a remaining life point in the set R, the remaining life rul is obtained from the set RpCorresponding set DIE of N performance degradation index valuesp(ii) a Then the residual life rul is obtainedpSet DI of corresponding performance degradation index valuespPerforming clustering analysis on the obtained performance degradation index value and residual service life by using One-class SVM algorithm to obtain a mapping set RDCI-rul，(ii) a Clustering analysis, comprising the following specific steps:
DIE set by using One-class SVM algorithmpPerforming cluster analysis, removing abnormal points in the set, and setting the performance degradation index set after screening as follows:
DIEp′＝{DIE1′,DIE2′,...,DIEj′,...,DIEm′}
wherein j is 1,2, and m is set DIEp' the number of elements;
obtaining a set DIEpThe cluster centers of' are as follows:
wherein j is 1,2, and m is set DIEp' the number of elements;
processing the remaining life sequence R of the heat meter to obtain a mapping set R of a performance degradation index value and the remaining lifeDCI-rulNamely, the performance degradation criteria in the statistical sense of the N heat meters are as follows:
RDCI-rul＝{(DCI1,rul1),...,(DCIq,rulq),...,(DCIn′,ruln′)}
wherein q ═ 1, 2.,. n'
S4, performing SVR training on the M groups of sample pairs, and establishing a remaining life prediction model rul ═ f (DCI) of the calorimeter;
s5, determining the remaining life rul of the heat meter to be predicted at the current time by using the performance degradation data of the heat meter to be predicted at the current time and combining with the remaining life prediction model rul ═ f (dci).
2. The method for predicting the remaining life of the heat meter based on the fusion of the multiple degradation sample data as claimed in claim 1, wherein in step S1, the method setsA performance degradation index sequence corresponding to the performance degradation data of the kth heat meter in the track library, whereinFor the heat meter at tiPerformance degradation index value corresponding to time, performance degradation index valueAnd a failure index valueOf performance degradation data set S'kThe following were used:
wherein, tDAnd tFRespectively the time when the heat meter starts to degrade and fail,andis the corresponding index value.
3. The method for predicting the remaining life of the heat meter based on the fusion of the multi-degradation sample data as claimed in claim 1, wherein in step S3, a DIE is setpThe following were used:
wherein, DIEjRul for remaining lifepThe corresponding j th heat meter performance degradation index value, N is the set DIEpThe number of the performance degradation indexes is n ', and the n' is the performance degradation indexes in the mapping sequence of the heat meter k performance degradation indexes and the residual service life and the number of corresponding residual service life points.
4. The method for predicting the remaining life of the heat meter based on the fusion of the multiple degradation sample data as claimed in claim 1, wherein in step S4, it is assumed that at time tiThe performance degradation index value of the heat meter is DCIiCorresponding to a remaining life of ruliAccording to DCIiAnd ruliThe mapping relationship between the two training samples is as follows to obtain M groups of training sample pairs of the SVR prediction model:
DCI＝[DCI1 DCI2...DCIi...DCIM]T
rul＝[rul1 rul2...ruli...rulM]T
wherein, M is the number of samples, DCI is input, rul is output, and after SVR training, the remaining life prediction model rul ═ f (DCI) of the calorimeter is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810863794.2A CN109253826B (en) | 2018-08-01 | 2018-08-01 | Heat meter residual life prediction method based on multi-degradation sample data fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810863794.2A CN109253826B (en) | 2018-08-01 | 2018-08-01 | Heat meter residual life prediction method based on multi-degradation sample data fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109253826A CN109253826A (en) | 2019-01-22 |
| CN109253826B true CN109253826B (en) | 2019-12-24 |
Family
ID=65048701
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810863794.2A Active CN109253826B (en) | 2018-08-01 | 2018-08-01 | Heat meter residual life prediction method based on multi-degradation sample data fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109253826B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112084459A (en) * | 2019-06-12 | 2020-12-15 | 电计贸易(上海)有限公司 | Method and device for predicting battery charge-discharge cycle life, electronic terminal and storage medium |
| CN111678622B (en) * | 2020-06-18 | 2021-09-14 | 河南省计量科学研究院 | Calorimeter life evaluation method based on step time durability test |
| CN111678621B (en) * | 2020-06-18 | 2021-09-14 | 河南省计量科学研究院 | Calorimeter life evaluation method based on cold-hot alternating acceleration method |
| CN112197890B (en) * | 2020-09-08 | 2021-07-13 | 西安交通大学 | Calorimeter performance degradation evaluation method, storage medium and equipment |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102789545B (en) * | 2012-07-12 | 2015-08-19 | 哈尔滨工业大学 | Based on the Forecasting Methodology of the turbine engine residual life of degradation model coupling |
| TWI463334B (en) * | 2012-07-20 | 2014-12-01 | Univ Nat Cheng Kung | Baseline predictive maintenance method for target device and computer program product thereof |
| WO2017216647A1 (en) * | 2016-06-17 | 2017-12-21 | Tata Consultancy Services Limited | Health monitoring and prognostics of a system |
| CN107103176B (en) * | 2017-01-17 | 2020-10-20 | 东南大学 | Coal-fired boiler SCR catalyst life evaluation method based on multi-source information fusion technology |
| CN107358347A (en) * | 2017-07-05 | 2017-11-17 | 西安电子科技大学 | Equipment cluster health state evaluation method based on industrial big data |
| CN107357994B (en) * | 2017-07-14 | 2020-04-28 | 哈尔滨工业大学（威海） | Staged mining method for aircraft engine performance decline mode |
-
2018
- 2018-08-01 CN CN201810863794.2A patent/CN109253826B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109253826A (en) | 2019-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109253826B (en) | Heat meter residual life prediction method based on multi-degradation sample data fusion | |
| CN104390657B (en) | A kind of Generator Unit Operating Parameters measurement sensor fault diagnosis method and system | |
| CN109948860A (en) | A kind of mechanical system method for predicting residual useful life and system | |
| KR20180061769A (en) | Semiconductor Manufacturing Yield Prediction System and Method based on Machine Learning | |
| KR101843365B1 (en) | Integrated Diagnostic System and Database based on rules and cases | |
| CN110471395B (en) | Fault detection method, device, equipment and storage medium | |
| CN109409425A (en) | A kind of fault type recognition method based on neighbour's constituent analysis | |
| US6763482B2 (en) | Printer diagnostics method | |
| JP7006282B2 (en) | Equipment abnormality diagnostic equipment | |
| CN114545234A (en) | Method for monitoring battery state based on battery temperature gradient change | |
| CN117115169A (en) | Intelligent recognition method for abnormal deformation of surface of die-casting die of automobile part | |
| CN103092762A (en) | Real-time software defect detection method applicable to rapid software development model | |
| CN111354496B (en) | Nuclear power plant accident online diagnosis and state tracking prediction method | |
| CN108470699B (en) | intelligent control system of semiconductor manufacturing equipment and process | |
| CN110414086A (en) | A kind of combined stress accelerated factor calculation method based on sensitivity | |
| CN102023636B (en) | Accelerated life testing method of machine tool numerical control system | |
| CN115544803B (en) | Transformer residual life prediction method, device, equipment and storage medium | |
| US20170132579A1 (en) | Offshore plant preventive maintenance system and offshore plant preventing maintenance method using the same | |
| CN111443602A (en) | Hidden Markov-based valve health degree estimation and life prediction method | |
| CN115374646A (en) | Method and device for predicting service life of electric meter in transit | |
| CN109556861A (en) | A kind of bearing real-time fault diagnosis system of case-based reasioning | |
| CN107121616B (en) | Method and device for fault positioning of intelligent instrument | |
| CN109389313A (en) | A kind of failure modes diagnostic method based on weighting neighbour's decision | |
| CN110532698B (en) | Industrial equipment vibration characteristic value trend prediction method based on data model | |
| US20220283574A1 (en) | Method for Controlling a Production Process for Producing Components |
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 |