CN114239752B - Method, device, equipment and medium for constructing residual life prediction model of relay - Google Patents
Method, device, equipment and medium for constructing residual life prediction model of relay Download PDFInfo
- Publication number
- CN114239752B CN114239752B CN202210079835.5A CN202210079835A CN114239752B CN 114239752 B CN114239752 B CN 114239752B CN 202210079835 A CN202210079835 A CN 202210079835A CN 114239752 B CN114239752 B CN 114239752B
- Authority
- CN
- China
- Prior art keywords
- data
- relay
- time window
- time
- residual life
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The embodiment of the invention discloses a method, a device, equipment and a medium for constructing a residual life prediction model of a relay, wherein the method comprises the following steps: obtaining the operating parameters of the relay in a period of time, namely historical operating data; selecting data with different time window coefficients in corresponding proportions from historical operating data to serve as time window data corresponding to different time window coefficients; calculating the data quality by utilizing a preset quality evaluation rule aiming at each time window data; and taking the time window data with the best data quality as an optimal data set, and constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay. Therefore, the embodiment of the invention avoids the situation that all historical operating data are needed in the process of constructing the residual life prediction model of the relay based on the extraction of the time window data with the best quality, improves the construction speed of the residual life prediction model of the relay, and ensures the accuracy of the residual life prediction.
Description
Technical Field
The invention relates to the field of life estimation, in particular to a method, a device, equipment and a medium for constructing a residual life prediction model of a relay.
Background
As an indispensable component in an automatic subway control system, the reliability of a relay directly influences the safety of subway operation, so that the reliability research of the relay gradually becomes the focus of academic attention, and one of important research contents of the reliability is the service life prediction of the relay.
In the existing research on residual life prediction of relays, most students use all historical data of relays to construct a life prediction model of the relays. And under the condition that the data quality of the historical data of the relay is poor, the problems of low prediction precision and low calculation speed of a service life prediction model of the relay are caused.
Disclosure of Invention
In view of the above, the present invention provides a method, an apparatus, a device, and a medium for constructing a residual life prediction model of a relay, so as to improve the current situations of low prediction accuracy and low computation speed of the residual life prediction model of the relay.
In a first aspect, an embodiment of the present invention provides a method for constructing a residual life prediction model of a relay, including:
acquiring historical operating data of the relay;
obtaining time window data corresponding to each time window coefficient according to the historical operating data based on a preset number of time window coefficients, wherein the time window coefficients represent the proportion of the corresponding time window data in the historical operating data;
obtaining the data quality of each time window data by using a preset quality evaluation rule;
taking the time window data with the best data quality as an optimal data set;
and constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay.
Optionally, in an implementation manner provided by the embodiment of the present invention, the historical operating data includes a preset number of operating parameters of the relay in each action;
the obtaining of the data quality of each time window data by using a preset quality evaluation rule includes:
aiming at the operation parameters in each time window data, calculating a time sequence correlation index of each operation parameter and the action times based on a preset first preset formula, and calculating a monotonicity index of each operation parameter under the action times based on a second preset formula;
calculating a data quality value corresponding to each operating parameter according to the time sequence correlation index and the monotonicity index of each operating parameter;
the taking the time window data with the best data quality as an optimal data set comprises the following steps:
and taking the time window data of the operating parameters with the maximum data quality values as the optimal data set.
Further, in an implementation manner provided by the embodiment of the present invention, the first preset formula includes:
wherein Corr () represents a time-series correlation index, X j A parameter value representing the jth operating parameter, k representing the number of actuations of the relay, and T representing the number of actuations represented by T k The time matrix of composition, t k Indicating the time of the kth action of the relay，X jT (t k ) Indicating the j-th operating parameter at time t k A trend term of;
the second preset formula includes:
in the formula, mon () represents a monotonicity index, and δ represents a unit step function.
Optionally, in an implementation manner provided by the embodiment of the present invention, the obtaining historical operation data of the relay includes:
acquiring original historical operating data of the relay;
and carrying out preset wavelet denoising processing on the original historical operating data to obtain historical operating data.
Optionally, in an implementation manner provided by the embodiment of the present invention, data in the historical operating data is sorted according to a time sequence;
the obtaining of the time window data corresponding to each time window coefficient according to the historical operating data based on the preset number of time window coefficients includes:
extracting data corresponding to each time window coefficient from historical operating data in a reverse-order extraction mode based on a preset number of time window coefficients;
and sequencing the data corresponding to each time window coefficient according to the time sequence to obtain the time window data corresponding to each time window coefficient.
Optionally, in an implementation manner provided by the embodiment of the present invention, the historical operating data includes a preset number of operating parameters corresponding to each contact unit;
the step of constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay comprises the following steps:
and according to the preset number of operating parameters corresponding to each contact unit of the optimal data set, constructing a wiener degradation model corresponding to the relay to obtain a residual life prediction model of the relay.
Further, in an implementation manner provided by the embodiment of the present invention, the contact unit includes at least one normally closed contact and at least one normally open contact, and the preset number of operation parameters include a normally open contact resistance, a normally closed contact resistance, pull-in time, and release time.
In a second aspect, an embodiment of the present invention provides a device for constructing a residual life prediction model of a relay, including:
the acquisition module is used for acquiring historical operating data of the relay;
the window data acquisition module is used for acquiring time window data corresponding to each time window coefficient according to the historical operating data based on a preset number of time window coefficients, wherein the time window coefficients represent the proportion of the corresponding time window data in the historical operating data;
the evaluation module is used for obtaining the data quality of each time window data by using a preset quality evaluation rule;
the selection module is used for taking the time window data with the best data quality as an optimal data set;
and the modeling module is used for constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay.
In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and the computer program, when running on the processor, executes the method for constructing the relay residual life prediction model disclosed in any one of the first aspects.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when running on a processor, executes the method for constructing the relay residual life prediction model disclosed in any one of the first aspects.
In the method for constructing the residual life prediction model of the relay, provided by the embodiment of the invention, firstly, the operation parameters of the relay in a period of time, namely historical operation data, are obtained; selecting data with different proportions from historical operating data to serve as time window data corresponding to different time window coefficients; then, aiming at each time window data, calculating the data quality by using a preset quality evaluation rule; and then, the time window data with the best data quality is used as an optimal data set, and a performance degradation model corresponding to the relay is constructed according to the optimal data set to obtain a residual life prediction model of the relay, so that the condition that all historical operating data are adopted in the construction process of the residual life prediction model of the relay is avoided.
Therefore, the embodiment of the invention avoids the condition that all historical data are needed to be used in the process of constructing the residual life prediction model of the relay based on the extraction of the time window data with the best quality, and improves the construction speed of the residual life prediction model of the relay. Moreover, because the data for constructing the residual life prediction model of the relay is the time window data with the best data quality in all the time window data, the residual life prediction model of the relay can be effectively supported by the data, and the residual life prediction accuracy of the relay is further ensured.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings required in the embodiments will be briefly described below, and it should be understood that the following drawings only illustrate some embodiments of the present invention, and therefore should not be considered as limiting the scope of the present invention. Like components are numbered similarly in the various figures.
Fig. 1 is a schematic flow chart illustrating a method for constructing a residual life prediction model of a relay according to a first embodiment of the present invention;
fig. 2 is a schematic flow chart illustrating a method for constructing a residual life prediction model of a second relay according to an embodiment of the present invention;
fig. 3 is a schematic flowchart illustrating a method for constructing a residual life prediction model of a third relay according to an embodiment of the present invention;
fig. 4 shows a schematic structural diagram of a device for constructing a residual life prediction model of a relay according to an embodiment of the present invention.
Detailed Description
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 only a part of the embodiments of the present invention, and not all of the embodiments.
The components of 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 of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
Hereinafter, the terms "including", "having", and their derivatives, which may be used in various embodiments of the present invention, are only intended to indicate specific features, numbers, steps, operations, elements, components, or combinations of the foregoing, and should not be construed as first excluding the existence of, or adding to, one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing.
Furthermore, the terms "first," "second," "third," and the like are used solely to distinguish one from another and are not to be construed as indicating or implying relative importance.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of the present invention belong. The terms (such as those defined in commonly used dictionaries) should be interpreted as having a meaning that is consistent with their contextual meaning in the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein in various embodiments of the present invention.
Referring to fig. 1, fig. 1 shows a schematic flow chart of constructing a first relay remaining life prediction model provided in an embodiment of the present invention, that is, the constructing of the relay remaining life prediction model provided in an embodiment of the present invention includes:
and S110, acquiring historical operation data of the relay.
It will be appreciated that the historical operating data represents the operating condition of the relay over a period of time, and may indicate the amount of change in the state of the relay, so that a model for predicting the remaining life of the relay may be constructed based on the historical operating data.
It will also be appreciated that the data composition of the historical operating data may be set as a function of the actual circumstances, such as in one possible approach, the historical operating data includes coil voltage and coil resistance values corresponding to the contacts.
In yet another possible approach, the historical operating data includes the normally open contact resistance, the normally closed contact resistance, the pull-in time, and the release time of the relay at each actuation.
It should be noted that, in the embodiment of the present invention, a specific obtaining manner of the historical operation data of the relay is not limited, and the specific obtaining manner of the relay may be selected according to an actual situation. As one possibility, historical operating data may be obtained by installing test equipment in the relay to monitor and collect data from the relay in real time.
Further, it is understood that noise may be included in the acquired historical operating data. Therefore, to improve the quality of the historical operating data, in an implementation manner provided by the embodiment of the present invention, specifically referring to fig. 2, fig. 2 shows a flowchart of a method for constructing a second relay remaining life prediction model provided by the embodiment of the present invention, that is, S110 in this implementation manner includes:
s111, acquiring original historical operation data of the relay;
and S112, performing preset wavelet denoising processing on the original historical operating data to obtain historical operating data.
That is, after the original historical operating data is acquired, the historical operating data is subjected to noise reduction processing to remove noise data contained in the data, and then a data set with a relatively obvious quantity change trend is obtained.
The wavelet denoising process provided by the embodiment of the invention can comprise the following steps: according to the characteristics that the intensity distribution conditions corresponding to the wavelet decomposition coefficients of the noise and the signal on different frequency bands are different, the wavelet decomposition coefficients corresponding to the noise on each frequency band are removed so as to reserve the wavelet decomposition coefficients of the original signal, and then wavelet reconstruction operation is carried out on the reserved wavelet decomposition coefficients of the original signal so as to obtain a pure signal, thereby obtaining historical operating data.
And S120, based on a preset number of time window coefficients, obtaining time window data corresponding to each time window coefficient according to the historical operating data, wherein the time window coefficients represent the proportion of the corresponding time window data in the historical operating data.
That is, in the embodiment of the present invention, data of a corresponding proportion is extracted from historical operating data according to the magnitude of each time window coefficient.
It can be understood that the time window coefficients may be set according to actual situations, for example, in a feasible manner, the preset number of time window coefficients include a first time window coefficient with a size of 0.1, a second time window coefficient with a size of 0.5, and a third time window coefficient with a size of 1, and the process of respectively obtaining corresponding time window data according to the first time window coefficient, the second time window coefficient, and the third time window coefficient is as follows: taking 0.1 percent of data in the historical operating data, namely 10 percent of data, as time window data corresponding to a first time window coefficient, taking 50 percent of data in the historical operating data as time window data corresponding to a second time window coefficient, and taking 100 percent of data in the historical operating data, namely all data, as time window data corresponding to a third time window coefficient.
In yet another possible way, the predetermined number of time window coefficients is a set with set elements [0.1,0.2,0.3, \ 8230; \ 8230;, 1], and the difference between the magnitudes of two adjacent time window coefficients is 0.1.
Optionally, in order to better reflect a data change trend in time window data and improve the prediction accuracy of the residual life prediction model of the relay, in a feasible implementation manner, the data in the historical operating data are sorted according to a time sequence;
the S120 includes:
extracting data corresponding to each time window coefficient from historical operating data in a reverse extraction mode based on a preset number of time window coefficients;
and sequencing the data corresponding to each time window coefficient according to the time sequence to obtain the time window data corresponding to each time window coefficient.
That is, in this embodiment, the data corresponding to each time window coefficient is extracted from the historical operating data in a reverse order extraction manner, and the data corresponding to each time window coefficient is sorted in a forward order to obtain the time window data.
Exemplarily, when the size of one time window coefficient is 0.5 and the data volume of the historical operating data is 1000, when the time window data is extracted, 500 data are extracted from the last data of the historical operating data, namely the 1000 th data, from the back to the front, and then the extracted 500 data are sequenced according to the time sequence, so that the time window data corresponding to the time window coefficient with the size of 0.5 is obtained, namely the 501 th to 1000 th data in the historical operating data are used as the time window data.
It is to be understood that the time window data obtained by the decimation provided in this embodiment can relatively effectively reflect future changes in the data.
For example, after the 1 st to 500 th data in the historical operating data are used as the time window data and the relay residual life prediction model is obtained based on the time window data, even if the relay residual life prediction model more accurately predicts the 501 th to 1000 th subsequent data, the relay residual life prediction model cannot be guaranteed to accurately predict the future change condition of the data. And if the predicted 501 th to 1000 th subsequent data and the actual 501 th to 1000 th subsequent data have errors, the error of the future change situation of the data predicted by the residual life prediction model of the relay is larger.
And if the residual life prediction model of the relay cannot accurately predict the 501 th to 1000 th subsequent data, the change trends of the former 500 data and the latter 500 data of the relay are inconsistent, and further the future change trend of the data cannot be predicted.
Therefore, the embodiment of the invention completes the extraction of the time window data based on the implementation mode, so that the future change condition of the data can be predicted by the relay residual life prediction model, the data quantity required for constructing the relay residual life prediction model is reduced, and the model training efficiency is improved.
And S130, obtaining the data quality of the data of each time window by using a preset quality evaluation rule.
That is, the embodiment of the present invention calculates the data quality of each time window data to determine the quality of the historical operating data under different time window coefficients.
It is understood that the specific calculation process of the data quality of each time window data can be set according to practical situations, such as in a feasible manner, the calculation process of the data quality of each time window data comprises: and carrying out preset data cleaning operation on each time window data, and calculating a data retention rate according to the data volume of the cleaned time window data and the data volume of the time window data before cleaning, so as to obtain the data quality of each time window data.
And S140, taking the time window data with the best data quality as an optimal data set.
That is, after the quality of each time window data is obtained, the time window data with the best quality, that is, the optimal data set, is used to obtain a residual life prediction model of the relay.
If the time window coefficient corresponding to the time window data is not larger than 1, the data amount required for obtaining the residual life prediction model of the relay is reduced, and the generation efficiency of the model is improved.
In addition, due to the fact that the data quality in the optimal data set is the best, effective data support can be provided for the construction process of the residual life prediction model of the relay, and effective training of the residual life prediction model of the relay is guaranteed.
And S150, constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay.
Namely, a performance degradation model of the relay is constructed according to the data in the optimal data set, namely, a model expression that the performance of the relay is gradually degraded along with time or action times is realized, so that the residual life of the relay is predicted, and a residual life prediction model of the relay is obtained.
It can be understood that the process of the method for constructing the performance degradation model corresponding to the relay is more complete, that is, the process for constructing the performance degradation model corresponding to the relay can be set as required. As one possible approach, the process of constructing the performance degradation model includes: and inputting the optimal data set into a preset neural network model, and training the neural network model to obtain a trained performance degradation model, namely a residual life prediction model of the relay.
In a possible embodiment, the historical operating data includes a preset number of operating parameters corresponding to each contact unit;
further, the S150 includes:
and according to the preset number of operating parameters corresponding to each contact unit of the optimal data set, constructing a wiener degradation model corresponding to the relay to obtain a residual life prediction model of the relay.
That is, the historical operating state of the relay in this embodiment is described by a preset number of operating parameters corresponding to each contact unit. Furthermore, according to the embodiment of the invention, a corresponding wiener degradation model is constructed based on the preset number of operation parameters corresponding to each contact unit, so that a residual life prediction model of the relay is obtained.
It is understood that wiener degradation models are often used in residual life prediction research to effectively characterize changes in the state of a device or component.
It should be further understood that the working condition of the relay is actually related to the working condition of any one contact, and the failure of any one contact will cause the relay to work abnormally, so that the remaining life of the relay can be understood as the minimum life of the contact units corresponding to all the contact units of the relay.
It will be appreciated that the data composition of the historical operating data may be set according to the actual situation, as in the foregoing possible manner, the historical operating data includes the coil voltage and the coil resistance corresponding to the contact.
In another possible mode, the contact unit includes at least one normally closed contact and at least one normally open contact, and the preset number of operating parameters includes a normally open contact resistance, a normally closed contact resistance, pull-in time, and release time.
The pull-in time of the relay is the time required from the energization of a coil of the relay to the working state of all contacts of the relay. The release time of the relay refers to the time required from when the coil is de-energized to when all the contacts transition to the released state. The pull-in time and the release time of the relay can effectively express the switching speed of the working state of the relay, and further the aging condition of the relay can be effectively represented.
Normally open contact resistance and normally closed contact resistance all change because of the oxidation degree and the roughness on contact surface, and oxidation degree and roughness all are relevant with the live time and the action number of times of relay, and then normally open contact resistance and normally closed contact resistance all can express the operating condition of relay to can calculate the remaining life of relay according to normally open contact resistance and normally closed contact resistance.
Therefore, the method and the device complete the construction of the residual life prediction model of the relay based on the pull-in time, the release time, the normally open contact resistance and the normally closed contact resistance, so that the residual life prediction model of the relay can accurately predict the residual life of the relay.
In the method for constructing the residual life prediction model of the relay, provided by the embodiment of the invention, firstly, the operation parameters of the relay in a period of time, namely historical operation data, are obtained; selecting data with different proportions from historical operating data to serve as time window data corresponding to different time window coefficients; then, aiming at each time window data, calculating the data quality by using a preset quality evaluation rule; and then, the time window data with the best data quality is used as an optimal data set, and a performance degradation model corresponding to the relay is constructed according to the optimal data set to obtain a residual life prediction model of the relay, so that the condition that all historical operating data are adopted in the construction process of the residual life prediction model of the relay is avoided.
Therefore, the embodiment of the invention avoids the condition that all historical data are needed to be used in the process of constructing the residual life prediction model of the relay based on the extraction of the time window data with the best quality, and improves the construction speed of the residual life prediction model of the relay. Moreover, because the data for constructing the residual life prediction model of the relay is the time window data with the best data quality in all the time window data, the residual life prediction model of the relay can be effectively supported by the data, and the accuracy of the residual life prediction of the relay is further ensured.
Optionally, in an implementation manner provided by the embodiment of the present invention, specifically referring to fig. 3, fig. 3 is a schematic flowchart illustrating a method for constructing a residual life prediction model of a third relay provided by the embodiment of the present invention, that is, the historical operating data in this implementation manner includes a preset number of operating parameters of the relay during each action;
further, the S130 includes:
s131, aiming at the operation parameters in each time window data, calculating a time sequence correlation index of each operation parameter and the action times based on a preset first preset formula, and calculating a monotonicity index of each operation parameter under the action times based on a second preset formula;
s132, calculating a data quality value corresponding to each operation parameter according to the time sequence correlation index and the monotonicity index of each operation parameter;
further, the S140 includes:
and S141, taking the time window data containing the operation parameters with the maximum data quality values as an optimal data set.
It is understood that the historical operating data and the time window data obtained from the historical operating data in this embodiment are both comprised of a predetermined number of operating parameters.
It will also be appreciated that a predetermined number of operating parameters may be set as a function of circumstances, such as in one possible approach, the predetermined number of operating parameters including the coil current for each contact. In another possible manner, the preset number of operating parameters includes a normally open contact resistance, a normally closed contact resistance, a pull-in time, and a release time.
It will be appreciated that the evaluation of data quality includes evaluation of the relevance of individual data in addition to the integrity, normalization, and consistency of the data. Therefore, the embodiment of the invention evaluates the relevance of the data based on the time sequence correlation index of each operation parameter and the action times and the monotonicity index of each operation parameter under the action times, and further evaluates the quality of the data.
Optionally, the calculation manners of the timing correlation index and the monotonicity index, that is, the first preset formula and the second preset formula, may be set/selected according to actual situations, for example, in an implementation manner provided in the embodiment of the present invention, the first preset formula includes:
wherein Corr () represents a time-series correlation index, X j A parameter value representing the jth operating parameter, k representing the number of actuations of the relay, and T representing the number of actuations represented by T k The time matrix of composition, t k Indicating the time of the kth action of the relay, X jT (t k ) Indicating the j-th operating parameter at time t k A trend term of;
the second preset formula includes:
in the formula, mon () represents a monotonicity index, and δ represents a unit step function.
Optionally, the process of calculating the data quality according to the timing correlation index and the monotonicity index may be set according to actual needs, for example, in a feasible manner, after the timing correlation index and the monotonicity index of each operating parameter are obtained, the data quality of each operating parameter is the sum of the corresponding timing correlation index and the corresponding monotonicity index.
In yet another possible approach, the process of calculating the data quality according to the time-series correlation index and the monotonicity index includes: multiplying the time sequence correlation index of each operation parameter by a preset first weight to obtain first data; multiplying the monotonicity index of each operation parameter by a preset second weight to obtain second data; and adding the first data and the second data to obtain the data quality of the operating parameter.
Therefore, the embodiment of the invention determines the correlation between each operation parameter and the action times in each time window data based on the time sequence correlation index and the monotonicity index of each operation parameter. It should be understood that the remaining life of the relay may be understood as the minimum number of times that the relay can normally operate under the preset condition, and thus, if the data effectively represents the correlation between the relay and the number of times of operation, the prediction accuracy of the model for predicting the remaining life of the relay is higher.
Therefore, the embodiment of the invention selects the time window data with the strongest relevance with the action times of the relay, namely the best data quality to construct the residual life prediction model of the relay based on the time sequence relevance index and the monotonicity index of each operation parameter, thereby ensuring the prediction accuracy of the residual life prediction model of the relay.
In addition, to better explain the obtaining process of the residual life prediction model of the relay, the embodiment of the invention also provides a derivation process of the residual life prediction model of the relay, which comprises the following steps:
setting historical operating data to comprise a preset number of operating parameters corresponding to each contact unit, wherein the preset number of operating parameters comprise normally open contact resistance, normally closed contact resistance, pull-in time and release time, and assuming Z j (t) is the amount of degradation of the relay at time t, the degradation process of the relay can be expressed as:
Z j (t)＝Z j (0)+μ j t+σ j B j (t) (1)
wherein Z is j (0) Representing the initial amount of degradation of the jth operating parameter; mu.s j Represents the diffusivity, i.e., the drift parameter, of the j-th operating parameter; sigma j A diffusion parameter representing a jth operating parameter and being constant; b j (t) denotes that the j-th operating parameter follows standard Brownian motion, characterizes the dynamics of the degradation process of the j-th operating parameter, and B j (t) obeys a normal distribution with a mean of 0 and a variance of t, i.e. B j (t)～N(0,t)。
And for the degradation process described by the formula (1), setting the preset failure threshold corresponding to the jth operation parameter as the failure threshold of the relay as D j The service life value corresponding to the jth operation parameter is T j I.e. amount of degeneration Z j (t) first reaching a failure threshold D j Time of (d), and thus the lifetime T corresponding to the jth operating parameter j The definition formula of (1) is as follows:
T j ＝inf{t|Z j (t)＝D j ,t≥0} (2)
according to the formula (2), the service life of the relay follows inverse Gaussian distribution, and then the cumulative failure probability function F corresponding to the jth operation parameter j (t) is:
thus, the reliability R corresponding to the jth operating parameter j The formula of (t) is:
R j (t)＝1-F j (t) (4)
understandably, the relay failure is competitive failure, namely, the failure of any one contact unit in the relay is equivalent to the failure of the relay, and the reliability R 'of the ith unit of the relay' i (t) is the product of the reliabilities corresponding to all the operating parameters, namely:
in the formula, R ij (t) represents the reliability of the jth operation parameter corresponding to the ith unit at the time t.
That is, the reliability R corresponding to the 4 operation parameters, i.e. the normally open contact resistance, the normally closed contact resistance, the pull-in time and the release time respectively j (t) are multiplied to obtain the reliability R 'of the ith unit' i (t)。
Further, the relay is composed of a plurality of units, and the units are mostly in series relation, so the overall reliability R (t) of the relay is the product of the reliabilities of all the units, and the expression is:
wherein n represents the number of contact units comprised by the relay.
Therefore, the expression of the cumulative failure probability function for a relay is:
F(t)＝1-R(t) (7)
let t k The current time, i.e. the time when the number of times the relay performs an action has accumulated to k, t j The service life value corresponding to the jth operation parameter is the jth operation parameter at the current time t k Residual life value l jk The calculation formula of (2) is as follows:
l jk ＝t j -t jk (8)
wherein, t jk Represents that the j-th operation parameter is matched when the action frequency of the relay reaches kThe time of response;
from the nature of the wiener degradation process, and in conjunction with equation (1):
Y j (l k )＝Y j (0)+μ j l jk +σ j B j (l jk ) (9)
wherein, Y j (l k )＝Z j (l k +t k )-Z j (t k )；T j (l k ) Indicates that the remaining life value l is reached when the number of operations reaches k k The corresponding amount of degradation; y is j (0) Indicates the residual life value l of the jth unit k A corresponding degradation increment of 0, and Y j (0)＝0。
Thus, the jth operating parameter is at the current time t k Probability density function f (l) of corresponding remaining life jk ) Comprises the following steps:
wherein z is jk Indicating that the jth operating parameter is at the current time t k The corresponding amount of degradation.
Furthermore, it will be appreciated that to accurately estimate the relevant unknown parameters in the degradation model and the model parameters in the wiener process, an estimate is made for each of the drift parameters and the diffusion parameters. I.e. for the diffusion parameter σ in equation (2) j And a drift parameter mu j And (6) estimating.
It will be appreciated that for any given time t +. DELTA.t > t >0, the increment of degradation between t +. DELTA.t and t follows a normal distribution, and in turn:
wherein, delta Z j (t) represents the corresponding degradation increment of the jth operating parameter at time t.
As can be seen from the nature of the wiener process, each operating parameter has an independent amount of degradation, Δ Z j Probability density function f (Δ Z) of (t) j ) Can be expressed as:
while in the wiener process, the drift parameter mu j And a diffusion parameter σ j Likelihood function L (μ) j ，σ j ) Comprises the following steps:
L(μ j ，σ j )＝f(△Z 1j ，△Z 2j ，…△Z nj )＝f(△Z 1j )f(△Z 2j )…f(△Z nj ) (13)
logarithm is taken to the likelihood function, and derivation is carried out to the likelihood function after logarithm taking to obtain a drift parameter mu j Corresponding maximum likelihood estimator
And a diffusion parameter σ j Corresponding maximum likelihood estimator
Respectively as follows:
wherein r represents the data volume of a data set for constructing a residual life prediction model of the relay, namely the data volume of an optimal data set; delta Z pj A degradation delta representing a jth operating parameter; delta t pj Representing the amount of time variation of the jth operating parameter.
To sum up, the jth operation parameter is processed at the current time t k Probability density function f (l) of corresponding remaining life jk ) Obtaining the expectation to obtain the jth operation parameter at the current time t k Residual life value of E (l) jk ) Comprises the following steps:
where E (-) represents the expected value of the parameter.
It will be appreciated that the relay includes a plurality of contact elements, and that each contact element corresponds to 4 operating parameters, namely, a normally open contact resistance, a normally closed contact resistance, a pull-in time, and a release time. Thus, one relay corresponds to the remaining life value of 4 operating parameters. And then the relay is considered to be failed due to the fact that the relay has the situation of competitive failure, namely when any parameter of any contact unit of the relay exceeds a threshold value. Therefore, the minimum residual life value corresponding to each operation parameter is taken as the residual life value T of the relay 1 The calculation formula is as follows:
T 1 ＝min{E 1 ，E 2 ，E 3 ，…，E M } (17)
wherein min {. Cndot } represents a minimum function, i.e., taking the minimum of the remaining life values of all operating parameters, E M And the residual life value of the Mth operation parameter is represented, and M represents the parameter number of the operation parameter corresponding to the relay.
The relay is composed of a plurality of contact units, each unit corresponds to 4 operating parameters, and the number of all the operating parameters corresponding to one relay is as follows:
M＝4×m (18)
wherein m represents the number of contact units included in the relay.
Corresponding to the method for constructing the relay residual life prediction model provided by the embodiment of the present invention, the embodiment of the present invention further provides a device for constructing the relay residual life prediction model, referring to fig. 4, fig. 4 shows a schematic structural diagram of the device for constructing the relay residual life prediction model provided by the embodiment of the present invention, and the device 200 for constructing the relay residual life prediction model provided by the embodiment of the present invention includes:
an obtaining module 210, configured to obtain historical operating data of the relay;
a window data obtaining module 220, configured to obtain, based on a preset number of time window coefficients, time window data corresponding to each time window coefficient according to the historical operating data, where the time window coefficient indicates a ratio of the corresponding time window data to the historical operating data;
the evaluation module 230 is configured to obtain data quality of each time window data according to a preset quality evaluation rule;
a selecting module 240, configured to use the time window data with the best data quality as an optimal data set;
and the modeling module 250 is used for constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay.
Optionally, in an implementation manner provided by the embodiment of the present invention, the historical operating data includes a preset number of operating parameters of the relay in each action;
further, the evaluation module includes:
the index calculation submodule is used for calculating a time sequence correlation index of each operation parameter and action times based on a preset first preset formula and calculating a monotonicity index of each operation parameter under the action times based on a second preset formula aiming at the operation parameter in each time window data;
the quality calculation submodule is used for calculating a data quality value corresponding to each operating parameter according to the time sequence correlation index and the monotonicity index of each operating parameter;
and the selection module is further used for taking the time window data containing the operation parameters with the maximum data quality values as the optimal data set.
Further, in an implementation manner provided by the embodiment of the present invention, the first preset formula includes:
wherein Corr () represents a time-series correlation index, X j A parameter value representing the jth operating parameter, k representing the number of actuations of the relay, and T representing the number of actuations represented by T k The time matrix of composition, t k Indicates the time of the kth action of the relay, X jT (t k ) Indicating the j-th operating parameter at time t k A trend term of;
the second preset formula includes:
in the formula, mon () represents a monotonicity index, and δ represents a unit step function.
Optionally, in an implementation manner provided by the embodiment of the present invention, the obtaining module includes:
the original data acquisition submodule is used for acquiring original historical operating data of the relay;
and the denoising submodule is used for performing preset wavelet denoising processing on the original historical operating data to obtain historical operating data.
Optionally, in an implementation manner provided by the embodiment of the present invention, data in the historical operating data is sorted according to a time sequence;
furthermore, the window data acquisition module includes:
the extraction submodule is used for extracting data corresponding to each time window coefficient from historical operating data in a reverse extraction mode based on a preset number of time window coefficients;
and the sequencing submodule is used for sequencing the data corresponding to each time window coefficient according to the time sequence to obtain the time window data corresponding to each time window coefficient.
Optionally, in an implementation manner provided by the embodiment of the present invention, the historical operating data includes a preset number of operating parameters corresponding to each contact unit;
furthermore, the modeling module is further configured to construct a wiener degradation model corresponding to the relay according to a preset number of operating parameters corresponding to each contact unit of the optimal data set, so as to obtain a residual life prediction model of the relay.
Further, in an implementation manner provided by the embodiment of the present invention, the contact unit includes at least one normally closed contact and at least one normally open contact, and the preset number of operation parameters includes a normally open contact resistance, a normally closed contact resistance, pull-in time, and release time.
The device for constructing the residual life prediction model of the relay, provided by the embodiment of the application, can realize each process of the method for constructing the residual life prediction model of the relay in the method embodiment disclosed in fig. 1, can achieve the same technical effect, and is not repeated here in order to avoid repetition.
The embodiment of the present invention further provides a computer device, which includes a memory and a processor, where the memory stores a computer program, and when the computer program runs on the processor, the method for constructing the residual life prediction model of the relay disclosed in the method embodiment corresponding to fig. 1 is executed.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program runs on a processor, the method for constructing a residual life prediction model of a relay disclosed in the method embodiment corresponding to fig. 1 is executed.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus embodiments described above are merely illustrative and, for example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, each functional module or unit in each embodiment of the present invention may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention or a part of the technical solution that contributes to the prior art in essence can be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention.
Claims (8)
1. A method for constructing a residual life prediction model of a relay is characterized by comprising the following steps:
obtaining historical operation data of a relay, wherein the historical operation data comprises a preset number of operation parameters of the relay during each action;
obtaining time window data corresponding to each time window coefficient according to the historical operating data based on a preset number of time window coefficients, wherein the time window coefficients represent the proportion of the corresponding time window data in the historical operating data;
obtaining the data quality of each time window data by using a preset quality evaluation rule;
taking the time window data with the best data quality as an optimal data set;
constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay;
the obtaining the data quality of each time window data by using a preset quality evaluation rule includes:
for the operation parameters in each time window data, calculating a time sequence correlation index of each operation parameter and the action times based on a preset first preset formula, and calculating a monotonicity index of each operation parameter under the action times based on a second preset formula, wherein the first preset formula comprises the following steps:
wherein Corr () represents a time-series correlation index, X j A parameter value representing the jth operating parameter, k representing the number of actuations of the relay, and T representing the number of actuations represented by T k The time matrix of composition, t k Indicating the time of the kth action of the relay, X jT (t k ) Indicating the jth operating parameter at time t k A trend term of;
the second preset formula includes:
in the formula, mon () represents a monotonicity index, and δ represents a unit step function;
calculating a data quality value corresponding to each operating parameter according to the time sequence correlation index and the monotonicity index of each operating parameter;
the taking the time window data with the best data quality as the optimal data set comprises the following steps:
and taking the time window data of the operating parameters with the maximum data quality values as an optimal data set.
2. The method for constructing the residual life prediction model of the relay according to claim 1, wherein the step of obtaining historical operation data of the relay comprises the following steps:
acquiring original historical operating data of the relay;
and carrying out preset wavelet denoising processing on the original historical operating data to obtain historical operating data.
3. The construction method of the relay residual life prediction model according to claim 1, characterized in that data in the historical operating data are sorted according to time sequence;
the obtaining of the time window data corresponding to each time window coefficient according to the historical operating data based on the preset number of time window coefficients includes:
extracting data corresponding to each time window coefficient from historical operating data in a reverse extraction mode based on a preset number of time window coefficients;
and sequencing the data corresponding to each time window coefficient according to the time sequence to obtain the time window data corresponding to each time window coefficient.
4. The method for constructing the residual life prediction model of the relay according to claim 1, wherein the historical operating data comprises a preset number of operating parameters corresponding to each contact unit;
the method for constructing the performance degradation model corresponding to the relay according to the optimal data set to obtain the residual life prediction model of the relay comprises the following steps:
and according to the preset number of operating parameters corresponding to each contact unit of the optimal data set, constructing a wiener degradation model corresponding to the relay to obtain a residual life prediction model of the relay.
5. The method for constructing a model for predicting the residual life of a relay according to claim 4, wherein the contact unit comprises at least one normally closed contact and at least one normally open contact, and the preset number of operating parameters comprise normally open contact resistance, normally closed contact resistance, pull-in time and release time.
6. A device for constructing a residual life prediction model of a relay is characterized by comprising the following components:
the relay control system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring historical operation data of a relay, and the historical operation data comprises a preset number of operation parameters of the relay during each action;
the window data acquisition module is used for acquiring time window data corresponding to each time window coefficient according to the historical operating data based on a preset number of time window coefficients, wherein the time window coefficients represent the proportion of the corresponding time window data in the historical operating data;
the evaluation module is used for obtaining the data quality of each time window data by using a preset quality evaluation rule;
the selection module is used for taking the time window data with the best data quality as an optimal data set;
the modeling module is used for constructing a performance degradation model corresponding to the relay according to the optimal data set to obtain a residual life prediction model of the relay;
the evaluation module comprises:
the index calculation submodule is used for calculating a time sequence correlation index of each operation parameter and action times based on a preset first preset formula and calculating a monotonicity index of each operation parameter under the action times based on a second preset formula aiming at the operation parameters in each time window data, wherein the first preset formula comprises the following steps:
wherein Corr () represents a time-series correlation index, X j A parameter value representing the jth operating parameter, k representing the number of actuations of the relay, and T representing the number of actuations represented by T k The time matrix of composition, t k Indicates the time of the kth action of the relay, X jT (t k ) Indicating the j-th operating parameter at time t k A trend term of;
the second preset formula includes:
where Mon () represents a monotonicity index, and δ represents a unit step function;
the quality calculation submodule is used for calculating a data quality value corresponding to each operating parameter according to the time sequence correlation index and the monotonicity index of each operating parameter;
the selection module is further used for taking the time window data containing the operation parameters with the maximum data quality values as the optimal data set.
7. A computer device, characterized by comprising a memory and a processor, the memory storing a computer program which, when run on the processor, performs the method of constructing a relay residual life prediction model according to any one of claims 1-5.
8. A computer-readable storage medium, having stored thereon a computer program which, when run on a processor, performs the method of constructing a relay residual life prediction model according to any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079835.5A CN114239752B (en) | 2022-01-24 | 2022-01-24 | Method, device, equipment and medium for constructing residual life prediction model of relay |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079835.5A CN114239752B (en) | 2022-01-24 | 2022-01-24 | Method, device, equipment and medium for constructing residual life prediction model of relay |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114239752A CN114239752A (en) | 2022-03-25 |
| CN114239752B true CN114239752B (en) | 2023-01-31 |
Family
ID=80746992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210079835.5A Active CN114239752B (en) | 2022-01-24 | 2022-01-24 | Method, device, equipment and medium for constructing residual life prediction model of relay |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114239752B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106503794A (en) * | 2016-11-08 | 2017-03-15 | 上海电机学院 | A kind of gear case of blower method for predicting residual useful life |
| WO2020137260A1 (en) * | 2018-12-28 | 2020-07-02 | オムロン株式会社 | Relay state prediction device, relay state prediction system, relay state prediction method, and program |
| CN111507046A (en) * | 2020-04-16 | 2020-08-07 | 哈尔滨工程大学 | Method and system for predicting remaining service life of electric gate valve |
| CN112734002A (en) * | 2020-12-28 | 2021-04-30 | 北京航空航天大学 | Service life prediction method based on data layer and model layer joint transfer learning |
| CN113297787A (en) * | 2021-05-12 | 2021-08-24 | 北京航空航天大学 | Method for predicting remaining life of aircraft engine based on transfer learning |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10699040B2 (en) * | 2017-08-07 | 2020-06-30 | The Boeing Company | System and method for remaining useful life determination |
| EP4029046A1 (en) * | 2019-09-11 | 2022-07-20 | Arc Suppression Technologies | Power contact end-of-life predictor apparatus |
| CN112629841A (en) * | 2020-11-23 | 2021-04-09 | 江苏科技大学 | Method for predicting service life of relay |
-
2022
- 2022-01-24 CN CN202210079835.5A patent/CN114239752B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106503794A (en) * | 2016-11-08 | 2017-03-15 | 上海电机学院 | A kind of gear case of blower method for predicting residual useful life |
| WO2020137260A1 (en) * | 2018-12-28 | 2020-07-02 | オムロン株式会社 | Relay state prediction device, relay state prediction system, relay state prediction method, and program |
| CN111507046A (en) * | 2020-04-16 | 2020-08-07 | 哈尔滨工程大学 | Method and system for predicting remaining service life of electric gate valve |
| CN112734002A (en) * | 2020-12-28 | 2021-04-30 | 北京航空航天大学 | Service life prediction method based on data layer and model layer joint transfer learning |
| CN113297787A (en) * | 2021-05-12 | 2021-08-24 | 北京航空航天大学 | Method for predicting remaining life of aircraft engine based on transfer learning |
Non-Patent Citations (3)
| Title |
|---|
| Life Prediction of Electromagnetic Relay Based on Bayesian Method;Jiang Wenman;《2018 IEEE International Conference of Safety Produce Information(IICSPI)》;20190414;340-345 * |
| Life Prediction of Residual Current Device Based on Wiener Process with GM(1,1) Model;Guojin Liu;《IEEJ Transactions on Electrical and Electronic Enginnering》;20210930;1165-1173 * |
| 基于Wiener过程的步退加速退化试验方法研究;李文华;《铁道科学与工程学报》;20180731;第15卷(第7期);1886-1892 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114239752A (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A similarity-based prognostics approach for remaining useful life estimation of engineered systems | |
| EP2098878B1 (en) | Battery service life estimation using a regression model and state estimation | |
| EP1416348B1 (en) | Methodology for temporal fault event isolation and identification | |
| Cartella et al. | Hidden Semi‐Markov Models for Predictive Maintenance | |
| Carr et al. | Modeling failure modes for residual life prediction using stochastic filtering theory | |
| Louen et al. | A new framework for remaining useful life estimation using support vector machine classifier | |
| CN101739819A (en) | Method and device for predicting traffic flow | |
| US12001180B2 (en) | Condition-based method for malfunction prediction | |
| Garthwaite et al. | Generating Monte Carlo confidence intervals by the Robbins–Monro process | |
| CN112364012B (en) | Data feature determination method and device and electronic equipment | |
| CN115470936B (en) | NWDAF-based machine learning model updating method and device | |
| CN112214934B (en) | Suspension system life prediction method based on multiple sensors and related device | |
| CN111783242B (en) | RVM-KF-based rolling bearing residual life prediction method and device | |
| CN111460692A (en) | Equipment residual life prediction method and system considering degradation rate mutual influence | |
| CN114924203A (en) | Battery SOH prediction analysis method and electric automobile | |
| US20160314405A1 (en) | Method of maintenance of equipment | |
| CN110851333B (en) | Root partition monitoring method and device and monitoring server | |
| CN111523727A (en) | Method for predicting remaining life of battery by considering recovery effect based on uncertain process | |
| CN115660484A (en) | Train health state evaluation method, device, system, electronic equipment and medium | |
| CN114239752B (en) | Method, device, equipment and medium for constructing residual life prediction model of relay | |
| CN116011214A (en) | Complex equipment residual life prediction method based on sparse variable decibel leaf | |
| CN113962020B (en) | Vehicle bearing residual life prediction method based on wiener process with measurement error | |
| Jouin et al. | Combined predictions for prognostics and predictive control of transportation PEMFC | |
| Weems et al. | Envelope peaks over threshold (EPOT) application and verification | |
| Mohril et al. | Residual life prediction in the presence of human error using machine learning |
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 |