CN114528942A - Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery - Google Patents
Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery Download PDFInfo
- Publication number
- CN114528942A CN114528942A CN202210166694.0A CN202210166694A CN114528942A CN 114528942 A CN114528942 A CN 114528942A CN 202210166694 A CN202210166694 A CN 202210166694A CN 114528942 A CN114528942 A CN 114528942A
- Authority
- CN
- China
- Prior art keywords
- sample library
- data
- target
- monitoring
- library
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Abstract
The invention relates to the technical field of fault diagnosis, in particular to a construction method of a data sample library of an engineering machine, a fault prediction method and the engineering machine, wherein the construction method comprises the steps of obtaining an initial sample library corresponding to a target engineering machine, wherein the initial sample library comprises a normal equipment library and a fault library, and each original data in the initial sample library has corresponding time sequence characteristics; acquiring monitoring data of target engineering machinery; dividing the monitoring data based on a preset time unit to obtain a monitoring time sequence; performing feature extraction on each monitoring time sequence to obtain monitoring features corresponding to the monitoring data; and determining an initial sample library to which the monitoring data belongs based on the similarity of the monitoring characteristics and the time sequence characteristics of each original data in each initial sample library so as to update the initial sample library to determine a target sample library. The method expands an initial sample library based on real data by an inference method, and solves the problem of insufficient fault samples.
Description
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a construction method of a data sample library of engineering machinery, a fault prediction method and the engineering machinery.
Background
In the field of anomaly detection, methods based on various machine learning models are an important technical route, and the methods need to learn fault rules from a large amount of historical data and continuously carry out algorithm iteration and parameter correction so as to construct an accurate fault diagnosis model. The high-quality sample library is the basis of all models, especially for supervised machine learning methods, different types of sample libraries are often required to be constructed, and the sample capacity and quality of the sample libraries can be important factors influencing the final algorithm implementation effect.
The anomaly detection of the engineering mechanical equipment is taken as a sub-field, the objective limitation of the anomaly detection enables a universal sample library construction technology to often fail to meet the requirements of a machine learning algorithm, and the specific problems are as follows:
1) the running time of the engineering mechanical equipment in the fault state usually only occupies a small part of the full life cycle of the engineering mechanical equipment, and the fault sample size is smaller than the healthy sample size from the aspect of appearance, so that the problem of sample imbalance is caused, and the training of an intelligent algorithm is not facilitated;
2) when a general sample library is constructed, no matter manual labeling or semi-automatic label generation is carried out, a set of clear executable methods is usually provided for various samples, the quantity of engineering machinery equipment is large, the distribution is wide, the working hours are long, monitoring of all equipment and collection of fault samples are not practical, screening work for different running states can be carried out on limited equipment only in limited time, and the quantity of available fault samples is further limited.
Due to the limited number of available samples, common sample equalization methods such as down-sampling directly limit the sample capacity to a lower level, which is not favorable for subsequent advanced algorithm development based on a sample library, such as fault prediction and fault identification.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method for constructing a data sample library of a construction machine, a method for predicting a fault, and a construction machine, so as to solve the problem that the number of samples of the construction machine is limited.
According to a first aspect, an embodiment of the present invention provides a method for building a data sample library of a construction machine, including:
acquiring an initial sample library corresponding to target engineering machinery, wherein the initial sample library comprises a normal equipment library and a fault library, and each piece of original data in the initial sample library has a corresponding time sequence characteristic;
acquiring monitoring data of the target engineering machinery;
dividing the monitoring data based on a preset time unit to obtain a monitoring time sequence;
performing feature extraction on each monitoring time sequence to obtain monitoring features corresponding to the monitoring data;
and determining an initial sample library to which the monitoring data belong based on the similarity of the monitoring features and the time sequence features of the original data in the initial sample libraries so as to update the initial sample library to determine a target sample library.
The construction method of the data sample base of the engineering machinery provided by the embodiment of the invention is characterized in that an initial sample base is constructed on the basis of a small amount of original data of a target engineering machinery, the original data is extracted and processed by adopting time sequence characteristics, and then the initial sample base is expanded on the basis of real data by an inference method, so that the problem of insufficient fault samples caused by difficulty in tracking the engineering machinery is solved.
With reference to the first aspect, in a first implementation manner of the first aspect, the determining an initial sample library to which the monitoring data belongs based on a similarity between the monitoring feature and a time-series feature of each piece of raw data in each initial sample library to update the initial sample library to determine a target sample library includes:
calculating the similarity of the monitoring characteristics and the time sequence characteristics of each original data in each initial sample library;
determining the similarity of the monitoring features and each initial sample library based on the mean value of all the similarities in each initial sample library;
determining an initial sample library corresponding to the maximum similarity among the similarities of the initial sample libraries as an initial sample library to which the monitoring data belongs;
and adding the monitoring data and the monitoring characteristics thereof into the initial sample library to update the initial sample library to determine a target sample library.
According to the construction method of the data sample base of the engineering machinery, the initial sample bases are provided with the plurality of corresponding original data and have at least one time sequence characteristic, on the basis, the similarity between the monitoring characteristic and each initial sample base can be determined by calculating the mean value of all similarities in the same initial sample base, the initial sample base to which the monitoring characteristic belongs is further determined, the initial sample base to which the monitoring characteristic belongs is expanded, and the accuracy of the determined initial sample base to which the monitoring characteristic belongs can be guaranteed.
With reference to the first implementation manner of the first aspect, in a second implementation manner of the first aspect, the adding the monitoring data and the monitoring characteristics thereof to the initial sample library to update the initial sample library to determine a target sample library further includes:
and performing down-sampling on the updated initial sample library to obtain each target sample library, wherein the number of samples in each target sample library is the same.
According to the construction method of the data sample library of the engineering machinery, provided by the embodiment of the invention, because the original data in each initial sample library possibly has an unbalanced problem, the obtained samples in each target sample library are ensured to be the same in number by performing down-sampling on the original data.
With reference to the first aspect, in a third implementation manner of the first aspect, the obtaining an initial sample library corresponding to a target engineering machine includes:
acquiring original data of the target engineering machinery to determine the normal equipment library and the fault library, wherein the original data comprises normal operation data and fault operation data with fault type labels;
dividing each original data in the normal equipment library and the fault library respectively based on a preset time unit to obtain an original time sequence;
extracting the characteristics of the original time sequence, and determining the time sequence characteristics corresponding to each original data;
determining the initial sample library based on the raw data and its corresponding timing characteristics.
According to the construction method of the data sample base of the engineering machinery, provided by the embodiment of the invention, each database is constructed by utilizing the original data with the labels, the original time sequence is obtained by time division, the characteristic extraction is carried out to obtain the time sequence characteristic corresponding to each original data, and the characteristic extraction is carried out on each original time sequence through a small amount of original data, so that the abundant time sequence characteristic can be obtained.
With reference to the third implementation manner of the first aspect, in the fourth implementation manner of the first aspect, the performing feature extraction on the original time series to determine a time series feature corresponding to each original data includes:
and performing at least one type of feature extraction on each original time sequence to obtain time sequence features corresponding to each original data.
The construction method of the data sample library of the engineering machinery provided by the embodiment of the invention extracts at least one type of feature for each original time sequence, obtains abundant time sequence features, and ensures the reliability of the subsequent initial sample library matching.
According to a second aspect, an embodiment of the present invention further provides a failure prediction method, including:
acquiring real-time operation data of the target engineering machinery;
inputting the real-time operation data into a target fault prediction model, and determining a target fault type, wherein the target fault prediction model is obtained by training a target sample library constructed based on the construction method of the data sample library of the engineering machinery in the first aspect of the invention or any embodiment of the first aspect.
According to the fault prediction method provided by the embodiment of the invention, the target fault prediction model is obtained by training based on the obtained target sample library, and due to the richness of the target sample library, the prediction accuracy of the trained target fault prediction model is higher, so that the reliability of the determined target fault type can be ensured.
With reference to the second aspect, in a first embodiment of the second aspect, the training process of the target fault prediction model includes:
obtaining the target sample library;
inputting each data in the target sample library into an initial fault prediction model to obtain a predicted fault type;
and performing loss calculation based on the predicted fault type and the fault type corresponding to each data, and updating parameters of the initial fault prediction model to determine the target fault prediction model.
According to a third aspect, an embodiment of the present invention provides an electronic device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the method for constructing a data sample library of a construction machine according to the first aspect or any embodiment of the first aspect, or execute the method for predicting a fault according to the second aspect or any embodiment of the second aspect.
According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the method for constructing a data sample library of a construction machine according to the first aspect or any one of the embodiments of the first aspect, or execute the method for predicting a fault according to the second aspect or any one of the embodiments of the second aspect.
According to a fifth aspect, an embodiment of the present invention further provides a construction machine, including:
a body;
the electronic device according to a third aspect of the present invention is provided in the body, and is configured to predict a failure type of the construction machine.
It should be noted that, for the corresponding beneficial effects of the electronic device, the computer-readable storage medium, and the engineering machine described in the embodiments of the present invention, please refer to the description of the corresponding beneficial effects of the construction method or the fault prediction method of the data sample library of the engineering machine, which is not described herein again.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method of building a data sample library for a work machine according to an embodiment of the present disclosure;
FIG. 2 is a flow chart of a method of building a data sample library for a work machine according to an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of raw data in an initial sample library according to an embodiment of the invention;
FIG. 4 is a flow diagram of a fault prediction method according to an embodiment of the invention;
fig. 5 is a block diagram of a construction apparatus of a data sample library for a construction machine according to an embodiment of the present invention;
fig. 6 is a block diagram of a configuration of a failure prediction apparatus according to an embodiment of the present invention;
fig. 7 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present invention.
Detailed Description
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. 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.
According to the construction method of the data sample of the engineering machine, provided by the embodiment of the invention, the initial sample library is obtained through priori knowledge, and then the unmarked real data of the target engineering machine is inferred to determine the initial sample library corresponding to the real data. The method adopted by the inference is a characteristic similarity method, the acquired unmarked real data is divided into time periods, corresponding characteristics are extracted, similarity calculation is carried out on the extracted characteristics and the corresponding characteristics in each initial sample library, so that the initial sample library corresponding to the unmarked real data is calculated, sample data in each initial sample library is enriched, the sample data is real operation data from the target engineering machinery, and the reliability is high.
The embodiment of the invention also provides a fault prediction method, which is used for obtaining a target fault prediction model by training a target sample library constructed based on the construction method of the data sample library of the engineering machinery. After the real-time operation data of the target engineering machinery is acquired, the fault type of the target engineering machinery can be predicted by using the target fault prediction model.
Furthermore, after the fault type is predicted, the real-time operation data can be added into a corresponding target sample library to update the target sample library so as to further update the parameters of the target fault prediction model, optimize the parameters of the target fault prediction model and improve the accuracy of the fault prediction model.
As a specific application scenario of this embodiment, the target fault prediction model is deployed in the electronic device and used for predicting the fault type of the real-time operation data. The electronic equipment can be arranged in the server, real-time operation data of the target engineering machinery is uploaded to the server in the operation process of the target engineering machinery, the server predicts the fault type of the target engineering machinery, and feeds the prediction result back to the target engineering machinery, so that the target engineering machinery can make corresponding treatment measures, or an operator of the target engineering machinery is reminded. It should be noted that, the server may predict the failure type of one or more target work machines at the same time, and is not limited herein.
As another specific application scenario of this embodiment, a target failure prediction model is deployed in an electronic device, and the electronic device is disposed in a target engineering machine. The target engineering machinery transmits real-time operation data thereof to the electronic equipment in the operation process, the electronic equipment predicts the fault type of the target engineering machinery and feeds the prediction result back to the target engineering machinery so that the target engineering machinery can make corresponding treatment measures or remind an operator of the target engineering machinery.
Based on the above, the embodiment of the invention also provides the engineering machinery, which comprises a body and the electronic equipment arranged in the body. The working machine may be an excavator, a forklift, etc., and the specific type thereof is not limited in any way. The specific structural type of the electronic device will be described in detail below.
It should be noted that, although a logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in an order different from the order shown.
In this embodiment, a method for constructing a data sample library of a construction machine is provided, and may be used in the above-mentioned electronic device, such as a terminal, a server, and the like, fig. 1 is a flowchart of a method for constructing a data sample library of a construction machine according to an embodiment of the present invention, and as shown in fig. 1, the flowchart includes the following steps:
and S11, acquiring an initial sample library corresponding to the target engineering machine.
The initial sample library comprises a normal equipment library and a fault library, and each original data in the initial sample library has corresponding time sequence characteristics.
The normal equipment library includes a plurality of raw data, which are normal operation data of the target construction machine. The failure library in the initial sample library includes at least one type, and the specific type is set according to actual needs, and may be 3 types, 4 types, and so on, and the number of the specific type is not limited at all. For example, the initial sample library includes a normal device library and 3 types of failure libraries, which are a failure library of failure type a, a failure library of failure type B, and a failure library of failure type C, respectively.
Each initial sample library comprises a plurality of original data, and each original data has a corresponding time sequence characteristic. Wherein, for the original data 1, there is a time-series characteristic a 1. Specifically, the time segment division is performed on the original data 1, and the time segment division is performed on the original data 1 to obtain a time sequence. And respectively extracting the features of each time sequence to obtain M features. Thus, for the original data 1, a time series feature matrix of N × M can be obtained.
The time period corresponding to the time sequence feature and the number of the features are specifically set according to actual requirements, and are not limited herein.
Details about this step will be described later.
And S12, acquiring the monitoring data of the target engineering machine.
The monitoring data of the target engineering machinery is data acquired by corresponding data acquisition equipment in the operation process of the target engineering machinery. For example, N sensors are provided in the target construction machine, and each time monitoring data with a time length of T is acquired, N monitoring data with a time length of T can be obtained.
The monitoring data may be index data (including but not limited to vibration, sound, temperature, pressure, etc.) and control signals of the device, which is not limited herein.
And S13, dividing the monitoring data based on the preset time unit to obtain a monitoring time sequence.
As described above, the monitoring data obtained by the electronic device is data for a period of time. The electronic equipment divides the monitoring data based on a preset time unit to obtain a monitoring time sequence. The monitoring time sequence comprises a plurality of monitoring subdata.
For example, the duration of the monitoring data is 1 hour, and the preset time unit is 20 minutes. Therefore, the electronic device can obtain 3 segments of monitoring subdata by dividing the monitoring data, thereby forming a monitoring time sequence.
The specific length of the preset time unit can be set according to actual requirements, and is not limited herein.
And S14, performing feature extraction on each monitoring time sequence to obtain monitoring features corresponding to the monitoring data.
After the electronic equipment obtains the monitoring time sequence, feature extraction is carried out on each monitoring subdata in the monitoring time sequence to obtain monitoring features corresponding to the monitoring data. For example, the duration of the monitoring data 1 is 1 hour, 3 segments of monitoring subdata are obtained by dividing, and the feature a, the feature B and the feature C are extracted for each piece of monitoring subdata. Therefore, 3 eigenvectors are obtained for each monitor subdata, and accordingly, a 3 × 3 monitor eigen matrix is obtained for the monitor data.
And S15, determining an initial sample library to which the monitoring data belong based on the similarity of the monitoring characteristics and the time sequence characteristics of the original data in the initial sample libraries, so as to update the initial sample library to determine a target sample library.
It should be noted that the vector dimension of the time-sequence feature of each original data in the initial sample library is the same as the vector dimension of the monitoring feature. Specifically, a plurality of raw data, and accordingly, a plurality of timing characteristics, are included in each initial sample library. The electronic equipment sequentially calculates the similarity between the monitoring characteristics and each time sequence characteristic in the same initial sample library, and can calculate the mean value of all the similarities as the similarity between the monitoring characteristics and the initial sample library; the weighted average of each similarity may also be calculated as the similarity of the monitored feature to the initial sample library.
After the similarity between the monitoring characteristics and each initial sample library is obtained through calculation, the electronic equipment determines the initial sample library with the highest similarity as the initial sample library to which the monitoring data belongs, and then adds the monitoring data into the initial sample library so as to update the initial sample library to determine a target sample library.
Details about this step will be described later.
According to the construction method of the data sample base of the engineering machinery, the initial sample base is constructed on the basis of a small amount of original data of the target engineering machinery, the original data are extracted and processed through time sequence characteristics, the initial sample base is expanded on the basis of real data through an inference method, and the problem that a fault sample is insufficient due to the fact that the engineering machinery is difficult to track is solved.
In this embodiment, a method for constructing a data sample library of a construction machine is provided, which may be used in the above-mentioned electronic device, such as a terminal, a server, and the like, fig. 2 is a flowchart of a method for constructing a data sample library of a construction machine according to an embodiment of the present invention, and as shown in fig. 2, the flowchart includes the following steps:
and S21, acquiring an initial sample library corresponding to the target engineering machine.
The initial sample library comprises a normal equipment library and a fault library, and each original data in the initial sample library has corresponding time sequence characteristics.
Specifically, S21 includes:
s211, acquiring original data of the target engineering machinery to determine a normal equipment library and a fault library.
The original data comprises normal operation data and fault operation data with fault type labels.
For the target engineering machinery, offline data in the historical operation process of all equipment is used as original data X, wherein the original data X comprises monitoring data of n different sensing equipment and control signals (X) of a control systemaXbXc… …). Selecting a batch of equipment, and operating the equipment by professional personnelMonitoring the working state of the equipment in the time period T to form a record, wherein the record content mainly comprises the normal operation time period T of the equipment0If the equipment fails, recording corresponding failure types (marked as A type, B type and C type … …) and failure time periods (T type)ATBTC) Combining the monitoring records of the selected batches of equipment and the original data to form a normal equipment library M0And various fault libraries MAMBMC。
Normal equipment library M0And various fault libraries MAMBMCIs denoted as M, which includes a normal device library M0Class a fault library MAClass b fault library MBClass c fault library MCAnd each knowledge base comprises the working time period of each device and corresponding monitoring data. As shown in fig. 3, an example presentation is constructed for cases in the knowledge base: in this case, the machine sensor monitoring data with the equipment number 0000001 is shown as three curves, and the maintenance record shows that the equipment is working normally in the time period T _0, and the equipment has a type a fault in the time period T _ a, so the time period T _0 in this case is classified into the normal equipment library in the knowledge base, and the time period T _ a is classified into the type a fault library in the knowledge base.
S212, dividing each original data in the normal equipment library and the fault library respectively based on a preset time unit to obtain an original time sequence.
Since the original data X cannot be directly used to distinguish samples, the original data X is divided into an original time series of minimum units by taking an appropriate preset time unit t.
And S213, extracting the characteristics of the original time sequence and determining the time sequence characteristics corresponding to each original data.
The electronic device performs appropriate feature extraction on these original time series, and the features used in this example are as follows:
the specific selection of the features is based on the final fault distinguishing capability, the specific form is not limited, and in other implementation cases, different tables for data are usedThe current form can be properly adjusted, and different feature extraction modes are adopted.
In some optional implementations of this embodiment, the S213 may include: and performing at least one type of feature extraction on each original time sequence to obtain time sequence features corresponding to each original data.
For example, if a target work machine K has three sensors a, b, and c and has an operating time of 2000 hours, three time series of the lengths Xa, Xb, and Xc are recorded as 2000 hours. When in processing, the long time sequence is firstly segmented by 50 hours (the minimum unit time sequence), and subsequent feature extraction and fault judgment are carried out by taking the segmented long time sequence as a unit. The time sequence characteristics are broadly expressed, three types are exemplified by formula forms, and finally, each 50-hour sequence corresponds to n characteristic values, for example, the ith subsequence characteristic of the device K is an n-dimensional vector [ r [ r ] ](1),(2),(3)……，r(n)。
At least one type of characteristics is extracted for each original time sequence, rich time sequence characteristics are obtained, and reliability of subsequent initial sample library matching is guaranteed.
S214, determining an initial sample library based on the original data and the corresponding time sequence characteristics.
After the feature extraction scheme is determined and feature extraction is performed, corresponding time sequence features exist for each original data in the knowledge base. That is, the raw data and the corresponding time series characteristics form a corresponding initial sample library.
And S22, acquiring the monitoring data of the target engineering machine.
Please refer to S12 in fig. 1, which is not described herein again.
And S23, dividing the monitoring data based on the preset time unit to obtain a monitoring time sequence.
For example, for any target engineering machine without monitoring record, the working hours are divided according to the window T to obtain a plurality of time periods TK1,TK2,TK3……
Please refer to S13 in fig. 1 for further details, which are not repeated herein.
And S24, performing feature extraction on each monitoring time sequence to obtain monitoring features corresponding to the monitoring data.
As in the above example, the monitoring characteristics corresponding to the respective time periods are: rK1RK2RK3……
Please refer to S14 in fig. 1, which is not described herein again.
And S25, determining the initial sample library to which the monitoring data belong based on the similarity of the monitoring characteristics and the time sequence characteristics of the original data in the initial sample libraries, and updating the initial sample library to determine a target sample library.
Specifically, S25 includes:
and S251, calculating the similarity of the monitoring characteristics and the time sequence characteristics of each original data in each initial sample library.
For the calculation of the similarity, a covariance mode, cosine similarity, euclidean distance, manhattan distance, pearson correlation coefficient, and the like may be adopted, and the specific mode adopted herein is not limited at all and may be set according to actual requirements.
For example, the similarity between the monitoring data Ki and the original data Mj in the knowledge base is defined
Where S is a covariance matrix.
And S252, determining the similarity of the monitoring features and each initial sample library based on the mean value of all the similarities in each initial sample library.
Wherein, any monitoring data K of the target engineering machinery KiThe similarity with the class a fault knowledge base can be recorded as
Similarity of other category sample libraries is similar.
And S253, determining the initial sample library corresponding to the maximum similarity in the similarities of the initial sample libraries as the initial sample library to which the monitoring data belongs.
And the electronic equipment takes the initial sample library with the highest similarity as a label corresponding to the monitoring data.
And S254, adding the monitoring data and the monitoring characteristics thereof into the initial sample library to update the initial sample library to determine a target sample library.
Correspondingly, adding the monitoring data and the detection characteristics into the initial sample library, and updating the initial sample library to obtain a sample set G'0,′a,′b… … is added. Wherein, G'0Is an updated normal device library, G'aIs an updated a-type fault library of G'bIs an updated b-type fault library.
In some optional implementations of this embodiment, the S254 may further include: and performing down-sampling on the updated initial sample library to obtain each target sample library. Wherein the number of samples in each target sample library is the same.
Specifically, because the running time of the target engineering machine in a normal state is much higher than the time of failure in the original data of the target engineering machine, various sample sets are processed by adopting a proper down-sampling method, so that the number of samples in each set is kept consistent, and a final sample-balanced sample library G is obtained0,Ga,Gb……
Because the original data in each initial sample library may have an imbalance problem, the obtained samples in each target sample library are guaranteed to be the same in number by down-sampling the original data.
In the construction method of the data sample base of the engineering machine, each database is constructed by using the original data with the tags, the original time sequence is obtained by time division, the time sequence features corresponding to each original data are obtained by feature extraction, and the time sequence features can be obtained by performing feature extraction on each original time sequence through a small amount of original data. Because each initial sample library corresponds to a plurality of original data and has at least one time sequence characteristic, on the basis, the similarity between the monitoring characteristic and each initial sample library can be determined by calculating the mean value of all similarities in the same initial sample library, and then the initial sample library to which the monitoring characteristic belongs is determined, so that the initial sample library to which the monitoring characteristic belongs is expanded, and the accuracy of the determined initial sample library to which the monitoring characteristic belongs can be ensured.
In this embodiment, a failure prediction method is provided, which may be used in the above-mentioned electronic device, such as a terminal, a server, and the like, fig. 4 is a flowchart of the failure prediction method according to the embodiment of the present invention, as shown in fig. 4, where the flowchart includes the following steps:
and S31, acquiring real-time operation data of the target engineering machine.
The real-time operation data of the target engineering machinery is acquired through data acquisition equipment arranged in the target engineering machinery. The real-time operation data includes, but is not limited to, index data such as real-time vibration and temperature, and may further include control signals.
And S32, inputting the real-time operation data into the target fault prediction model, and determining the type of the target fault.
The target fault prediction model is obtained by training a target sample library constructed based on the construction method of the data sample library of the engineering machinery.
The target fault prediction model is obtained based on the training of the target sample library, and for the construction method of the target sample library, please refer to the construction method of the data sample library of the engineering machine described above, which is not described herein again. And the electronic equipment predicts the target fault type of the real-time operation data by using the target fault prediction model, wherein the target fault type comprises normal operation, a type a fault type, a type b fault type and the like.
According to the fault prediction method provided by the embodiment, the target fault prediction model is obtained by training based on the obtained target sample library, and due to the richness of the target sample library, the prediction accuracy of the trained target fault prediction model is higher, so that the reliability of the determined target fault type can be ensured.
As an optional implementation manner of this embodiment, the training process of the target fault prediction model includes:
(1) and acquiring a target sample library.
(2) And inputting each data in the target sample library into the initial fault prediction model to obtain the predicted fault type.
(3) And performing loss calculation based on the predicted fault type and the fault type corresponding to each data, and updating parameters of the initial fault prediction model to determine a target fault prediction model.
Each original data in the target sample library has a corresponding fault type, and the fault types are corresponding labels. And inputting each time sequence characteristic in the original data into an initial fault prediction model to obtain a predicted fault type. And determining a target fault prediction model by performing loss calculation on the predicted fault type and the fault type corresponding to each data and updating parameters of the initial fault prediction model based on the loss calculation result.
According to the construction method of the data sample library of the engineering machinery, provided by the embodiment of the invention, the fault judgment of the engineering machinery depends on professional personnel, and the fault judgment and classification of a certain amount of equipment are completed, so that the foundation of the whole work is provided. In order to ensure that fault data can reflect the real condition of equipment and ensure that a sample library reaches a certain scale, the expansion of each knowledge base needs to be extracted from the real operation data of the engineering machinery. Various sensors are arranged on the target engineering machinery, real-time index data (including but not limited to vibration, sound, temperature, pressure and the like) and control signals of the target engineering machinery are important basis for identifying faults, and effective processing of the time sequence data is a key for ensuring effective implementation of a sample expansion method.
The embodiment also provides a device for constructing a data sample library of the engineering machinery and a fault prediction device, and the device is used for implementing the embodiment and the preferred embodiment, and the description of the device is omitted. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
The present embodiment provides a construction apparatus for a data sample library of a construction machine, as shown in fig. 5, including:
the first obtaining module 41 is configured to obtain an initial sample library corresponding to the target engineering machine, where the initial sample library includes a normal device library and a fault library, and each piece of original data in the initial sample library has a corresponding time sequence characteristic;
a second obtaining module 42, configured to obtain monitoring data of the target engineering machine;
a dividing module 43, configured to divide the monitoring data based on a preset time unit to obtain a monitoring time sequence;
a feature extraction module 44, configured to perform feature extraction on each monitoring time sequence to obtain a monitoring feature corresponding to the monitoring data;
and a determining module 45, configured to determine, based on similarity between the monitoring feature and a time-series feature of each piece of original data in each of the initial sample libraries, an initial sample library to which the monitoring data belongs, so as to update the initial sample library to determine a target sample library.
The present embodiment further provides a failure prediction apparatus, as shown in fig. 6, including:
a third obtaining module 51, configured to obtain real-time operation data of the target engineering machine;
the prediction module 52 is configured to input the real-time operation data into a target fault prediction model, and determine a target fault type, where the target fault prediction model is obtained by training a target sample library constructed based on the construction method of the data sample library of the construction machine in the first aspect of the present invention or any embodiment of the first aspect.
The device for constructing a data sample library of a construction machine and the failure prediction device in this embodiment are presented in the form of functional units, where the units refer to ASIC circuits, processors and memories executing one or more software or fixed programs, and/or other devices capable of providing the above functions.
Further functional descriptions of the modules are the same as those of the corresponding embodiments, and are not repeated herein.
An embodiment of the present invention further provides an electronic device, which includes the above-described construction apparatus for a data sample library of a construction machine shown in fig. 5, or the failure prediction apparatus shown in fig. 6.
Referring to fig. 7, fig. 7 is a schematic structural diagram of an electronic device according to an alternative embodiment of the present invention, and as shown in fig. 7, the electronic device may include: at least one processor 601, such as a CPU (Central Processing Unit), at least one communication interface 603, memory 604, and at least one communication bus 602. Wherein a communication bus 602 is used to enable the connection communication between these components. The communication interface 603 may include a Display (Display) and a Keyboard (Keyboard), and the optional communication interface 603 may also include a standard wired interface and a standard wireless interface. The Memory 604 may be a Random Access Memory (RAM) or a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The memory 604 may optionally be at least one storage device located remotely from the processor 601. Wherein the processor 601 may be in connection with the apparatus described in fig. 5 or fig. 6, the memory 604 stores an application program, and the processor 601 calls the program code stored in the memory 604 for performing any of the above-mentioned method steps.
The communication bus 602 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus 602 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in FIG. 7, but this is not intended to represent only one bus or type of bus.
The memory 604 may include a volatile memory (RAM), such as a random-access memory (RAM); the memory may also include a non-volatile memory (english: non-volatile memory), such as a flash memory (english: flash memory), a hard disk (english: hard disk drive, abbreviated: HDD) or a solid-state drive (english: SSD); the memory 604 may also comprise a combination of the above types of memory.
The processor 601 may be a Central Processing Unit (CPU), a Network Processor (NP), or a combination of a CPU and an NP.
The processor 601 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a Programmable Logic Device (PLD), or a combination thereof. The PLD may be a Complex Programmable Logic Device (CPLD), a field-programmable gate array (FPGA), a General Array Logic (GAL), or any combination thereof.
Optionally, the memory 604 is also used for storing program instructions. The processor 601 may call program instructions to implement a method of constructing a data sample library of a work machine, or a method of predicting a failure, as shown in any of the embodiments of the present application.
The embodiment of the invention also provides a non-transitory computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the construction method or the fault prediction method of the data sample library of the engineering machinery in any method embodiment. The storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a Flash Memory (Flash Memory), a Hard Disk (Hard Disk Drive, abbreviated as HDD), a Solid State Drive (SSD), or the like; the storage medium may also comprise a combination of memories of the kind described above.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Claims (10)
1. A method for constructing a data sample library of a construction machine is characterized by comprising the following steps:
acquiring an initial sample library corresponding to target engineering machinery, wherein the initial sample library comprises a normal equipment library and a fault library, and each original data in the initial sample library has corresponding time sequence characteristics;
acquiring monitoring data of the target engineering machinery;
dividing the monitoring data based on a preset time unit to obtain a monitoring time sequence;
performing feature extraction on each monitoring time sequence to obtain monitoring features corresponding to the monitoring data;
and determining an initial sample library to which the monitoring data belong based on the similarity of the monitoring features and the time sequence features of the original data in the initial sample libraries so as to update the initial sample library to determine a target sample library.
2. The method of claim 1, wherein the determining the initial sample library to which the monitoring data belongs based on the similarity of the monitoring features and the time-series features of the raw data in the initial sample libraries to update the initial sample library to determine a target sample library comprises:
calculating the similarity of the monitoring characteristics and the time sequence characteristics of each original data in each initial sample library;
determining the similarity of the monitoring features and each initial sample library based on the mean value of all the similarities in each initial sample library;
determining an initial sample library corresponding to the maximum similarity among the similarities of the initial sample libraries as an initial sample library to which the monitoring data belongs;
and adding the monitoring data and the monitoring characteristics thereof into the initial sample library to update the initial sample library to determine a target sample library.
3. The method of claim 2, wherein the adding the monitoring data and the monitoring characteristics thereof to the initial sample library to update the initial sample library to determine a target sample library, further comprises:
and performing down-sampling on the updated initial sample library to obtain each target sample library, wherein the number of samples in each target sample library is the same.
4. The method of claim 1, wherein the obtaining an initial sample library corresponding to the target work machine comprises:
acquiring original data of the target engineering machinery to determine the normal equipment library and the fault library, wherein the original data comprises normal operation data and fault operation data with fault type labels;
dividing each original data in the normal equipment library and the fault library respectively based on a preset time unit to obtain an original time sequence;
extracting the characteristics of the original time sequence, and determining the time sequence characteristics corresponding to each original data;
determining the initial sample library based on the raw data and its corresponding timing characteristics.
5. The method according to claim 4, wherein the extracting the features of the original time series and determining the time series feature corresponding to each original data comprises:
and performing at least one type of feature extraction on each original time sequence to obtain time sequence features corresponding to each original data.
6. A method of fault prediction, comprising:
acquiring real-time operation data of the target engineering machinery;
inputting the real-time operation data into a target fault prediction model, and determining a target fault type, wherein the target fault prediction model is obtained by training a target sample library constructed based on the construction method of the data sample library of the engineering machinery in any one of claims 1 to 5.
7. The fault prediction method of claim 6, wherein the training process of the target fault prediction model comprises:
obtaining the target sample library;
inputting each data in the target sample library into an initial fault prediction model to obtain a predicted fault type;
and performing loss calculation based on the predicted fault type and the fault type corresponding to each data, and updating parameters of the initial fault prediction model to determine the target fault prediction model.
8. An electronic device, comprising:
a memory and a processor, wherein the memory and the processor are communicatively connected with each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the method for constructing a data sample base of a construction machine according to any one of claims 1 to 5, or execute the method for predicting a failure according to any one of claims 6 to 7.
9. A computer-readable storage medium storing computer instructions for causing a computer to execute the method for constructing a data sample library for a construction machine according to any one of claims 1 to 5 or the method for predicting a failure according to any one of claims 6 to 7.
10. A work machine, comprising:
a body;
the electronic device of claim 8, disposed within the body for predicting a type of failure of the work machine.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210166694.0A CN114528942A (en) | 2022-02-23 | 2022-02-23 | Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210166694.0A CN114528942A (en) | 2022-02-23 | 2022-02-23 | Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114528942A true CN114528942A (en) | 2022-05-24 |
Family
ID=81624943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210166694.0A Pending CN114528942A (en) | 2022-02-23 | 2022-02-23 | Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114528942A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116841650A (en) * | 2023-08-31 | 2023-10-03 | 腾讯科技（深圳）有限公司 | Sample construction method, device, equipment and storage medium |
-
2022
- 2022-02-23 CN CN202210166694.0A patent/CN114528942A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116841650A (en) * | 2023-08-31 | 2023-10-03 | 腾讯科技（深圳）有限公司 | Sample construction method, device, equipment and storage medium |
| CN116841650B (en) * | 2023-08-31 | 2023-11-21 | 腾讯科技（深圳）有限公司 | Sample construction method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN111210024B (en) | Model training method, device, computer equipment and storage medium | |
| CN108647136B (en) | Hard disk damage prediction method and device based on SMART information and deep learning | |
| JP2019016209A (en) | Diagnosis device, diagnosis method, and computer program | |
| WO2017076154A1 (en) | Method and apparatus for predicting network event and establishing network event prediction model | |
| US11657121B2 (en) | Abnormality detection device, abnormality detection method and computer readable medium | |
| JP2019185422A (en) | Failure prediction method, failure prediction device, and failure prediction program | |
| CN108415810B (en) | Hard disk state monitoring method and device | |
| CN102257448B (en) | Method and device for filtering signal using switching models | |
| CN113010389A (en) | Training method, fault prediction method, related device and equipment | |
| JPWO2020021587A1 (en) | Time series data diagnostic device, additional learning method and program | |
| CN109739904A (en) | A kind of labeling method of time series, device, equipment and storage medium | |
| CN113219341B (en) | Model generation and battery degradation estimation device, method, medium, and apparatus | |
| CN115392782A (en) | Method and system for monitoring and diagnosing health state of process system of nuclear power plant | |
| CN114528942A (en) | Construction method of data sample library of engineering machinery, failure prediction method and engineering machinery | |
| CN109698026B (en) | Component identification in fault handling of medical devices | |
| CN115617604A (en) | Disk failure prediction method and system based on image pattern matching | |
| JP7458268B2 (en) | Information processing device, information processing method, computer program and information processing system | |
| KR20230075150A (en) | Method for managing system health | |
| US20210372647A1 (en) | Method and system for automatic detection of malfunctions/inefficient household electronic heating device | |
| CN112395167A (en) | Operation fault prediction method and device and electronic equipment | |
| CN112737834A (en) | Cloud hard disk fault prediction method, device, equipment and storage medium | |
| JP2021192155A (en) | Program, method and system for supporting abnormality detection | |
| TWI684882B (en) | Automatic optimization method for generating fault characteristics | |
| CN116383016B (en) | Method, device and equipment for monitoring state of magnetic disk and predicting fault |
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 |