CN114298443A - Industrial equipment predictive maintenance method and device based on health state index and electronic equipment - Google Patents
Industrial equipment predictive maintenance method and device based on health state index and electronic equipment Download PDFInfo
- Publication number
- CN114298443A CN114298443A CN202210200518.4A CN202210200518A CN114298443A CN 114298443 A CN114298443 A CN 114298443A CN 202210200518 A CN202210200518 A CN 202210200518A CN 114298443 A CN114298443 A CN 114298443A
- Authority
- CN
- China
- Prior art keywords
- data
- health state
- index
- time
- health
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000036541 health Effects 0.000 title claims abstract description 187
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000012423 maintenance Methods 0.000 title claims abstract description 56
- 230000015556 catabolic process Effects 0.000 claims abstract description 79
- 238000006731 degradation reaction Methods 0.000 claims abstract description 79
- 238000012549 training Methods 0.000 claims abstract description 44
- 230000003862 health status Effects 0.000 claims description 53
- 239000013598 vector Substances 0.000 claims description 36
- 238000012545 processing Methods 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 20
- 238000003860 storage Methods 0.000 claims description 16
- 238000003062 neural network model Methods 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 description 16
- 230000006870 function Effects 0.000 description 14
- 238000000605 extraction Methods 0.000 description 11
- 238000009826 distribution Methods 0.000 description 6
- 239000000126 substance Substances 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000035772 mutation Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008676 import Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- 238000002759 z-score normalization Methods 0.000 description 1
Images
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Quality & Reliability (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a health state index-based industrial equipment predictive maintenance method, a health state index-based industrial equipment predictive maintenance device and electronic equipment. And then, importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training for fitting to obtain an extension curve, comparing the predicted health state data of each time point on the extension curve with a preset threshold value, and determining the time point with the same data as the failure time point. In the scheme, by using a pre-trained reconstruction model and a pre-trained prediction model, the determination of the degradation point and the prediction of the data can be accurately realized by learning the characteristics of the operating data, and the method can be suitable for predictive maintenance based on a small amount of data.
Description
Technical Field
The application relates to the technical field of industrial equipment management, in particular to a health state index-based industrial equipment predictive maintenance method and device and electronic equipment.
Background
The maintenance of industrial equipment is of great significance to the economic benefits of the manufacturing industry. The preventive maintenance, namely the method for regular maintenance, can fully prevent the economic loss caused by the downtime of the robot in the production process, but increases the workload of operation and maintenance personnel, causes the waste of a large amount of parts and improves the maintenance cost of the robot. The industry is therefore exploring predictive maintenance techniques for industrial equipment in an attempt to save maintenance costs by performing maintenance when equipment performance is minimized.
In the prior art, predictive maintenance of industrial equipment mainly adopts multi-sensor acquisition of multi-source data and constructs a single or composite health state index, so that failure prediction of the industrial equipment is performed. However, in the prior art, the method for setting a plurality of sensors to collect data for prediction is mainly suitable for large-scale equipment. In predictive maintenance for small devices such as industrial robots, the portability of a large number of sensors becomes impractical due to practical operating environment limitations as well as the process itself and cost limitations of the sensors being performed. Therefore, for small-sized equipment, how to accurately realize predictive maintenance is very important under the condition that a large number of sensors cannot be arranged to collect multi-source data.
Disclosure of Invention
Objects of the present application include, for example, providing a health status index-based industrial equipment predictive maintenance method, apparatus, and electronic device that enables accurate determination of degradation points and prediction of data.
The embodiment of the application can be realized as follows:
in a first aspect, the present application provides a method for predictive maintenance of industrial equipment based on a health status index, the method comprising:
acquiring operation data of each time point in the operation process of equipment to be tested, and importing the operation data into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operation data;
obtaining a health state index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health state index, wherein the degradation point represents the time point when the health state of the equipment to be tested begins to degrade;
and importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training to fit the health state index, obtaining an extension curve based on the fit curve, comparing the predicted health state index on each time point in the extension curve with a preset threshold value, and determining the time point with the same predicted health state index and the preset threshold value as a failure time point.
In an optional embodiment, the step of importing the operating data into a pre-trained reconstructed model to obtain reconstructed data corresponding to the operating data includes:
intercepting the operation data of the equipment to be tested at a plurality of continuous time points according to a preset step length and a preset window length to obtain the operation data in a plurality of time windows;
for the operation data in each time window, scaling the operation data to be in a preset range;
extracting a time domain characteristic vector, a frequency domain characteristic vector and a time-frequency domain characteristic vector of the scaled operation data;
and importing the time domain feature vector, the frequency domain feature vector and the time-frequency domain feature vector into a reconstruction model obtained by pre-training to obtain reconstruction data corresponding to the operating data.
In an optional embodiment, the preset window length is greater than the preset step size.
In an optional embodiment, the step of obtaining a health status index from the reconstruction data and the operating data, and determining a degradation point in a time point according to the health status index includes:
obtaining difference data between the reconstruction data and the operation data, and taking the difference data as a health state index;
and comparing the health state index with a health state threshold value, and determining a time point when the health state index begins to deviate from the health state threshold value as a degradation point.
In an alternative embodiment, the step of determining a point in time at which the health status index starts to deviate from the health status threshold as a degradation point includes:
acquiring a time point in the health status index at which the health status index begins to deviate from the health status threshold;
and detecting whether the health state indexes respectively corresponding to the set number of time points after the time point all deviate from the health state threshold value, and if so, determining that the time point is a degradation point.
In an optional embodiment, the method further includes a step of obtaining the reconstructed model based on a training of a constructed neural network model in advance, where the neural network model includes an encoder and a decoder, and the step includes:
collecting sample data, wherein the sample data comprises data corresponding to a plurality of continuous time points;
importing the sample data into the encoder to perform encoding processing to obtain characteristic data;
importing the characteristic data and the sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;
and continuing training after adjusting the model parameters of the encoder and the decoder based on the loss function constructed according to the sample data and the sample reconstruction data until the preset requirements are met, and obtaining the reconstruction model.
In an alternative embodiment, the state of health threshold is obtained by:
calculating a difference between the sample data and sample reconstruction data;
calculating to obtain a difference average value and a difference standard deviation based on the difference value;
and obtaining the health state threshold value according to the difference average value and the difference standard deviation.
In an optional embodiment, the step of importing a pre-trained prediction model into the health state index corresponding to the time point after the degradation point to obtain prediction data in a next prediction period includes:
acquiring a health state index corresponding to a time point after a degradation point of the equipment to be tested;
dividing the health state indexes into time windows according to a time sequence to obtain the health state indexes in a plurality of time windows;
normalizing the health state index in each time window;
and extracting the data characteristics of the health state index after normalization processing, and introducing the data characteristics into a prediction model obtained by pre-training so as to fit the health state index.
In a second aspect, the present application provides a health status index-based predictive maintenance apparatus for industrial equipment, the apparatus comprising:
the acquisition module is used for acquiring operation data of each time point in the operation of the equipment to be tested, and importing the operation data into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operation data;
the determining module is used for obtaining a health state index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health state index, wherein the degradation point represents the time point when the health state of the equipment to be tested begins to degrade;
and the prediction module is used for importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training so as to fit the health state index, obtaining an extension curve based on the fit curve, comparing the predicted health state index on each time point in the extension curve with a preset threshold value, and determining the time point with the same predicted health state index and the preset threshold value as a failure time point.
In a third aspect, the present application provides an electronic device comprising one or more storage media and one or more processors in communication with the storage media, the one or more storage media storing processor-executable machine-executable instructions that, when executed by the electronic device, are executed by the processors to perform the method steps of any one of the preceding embodiments.
The beneficial effects of the embodiment of the application include, for example:
the application provides a health state index-based industrial equipment predictive maintenance method, a health state index-based industrial equipment predictive maintenance device and electronic equipment. And then, importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training for fitting to obtain an extension curve, comparing the predicted health state index of each time point on the extension curve with a preset threshold value, and determining the time point with the same index as the failure time point. In the scheme, by using a pre-trained reconstruction model and a pre-trained prediction model, the determination of the degradation point and the prediction of the data can be accurately realized by learning the characteristics of the operating data, and the method can be suitable for predictive maintenance based on a small amount of data.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
FIG. 1 is a flow chart of a predictive maintenance method provided by an embodiment of the present application;
FIG. 2 is a schematic diagram of a fitted curve and an extended curve provided by an embodiment of the present application;
FIG. 3 is a flowchart of sub-steps included in step S101 of FIG. 1;
FIG. 4 is a schematic diagram of time window data interception performed in the embodiment of the present application;
FIG. 5 is a schematic illustration of a process for reconstructing a model provided in an embodiment of the present application;
FIG. 6 is a flowchart of a reconstruction model training method according to an embodiment of the present disclosure;
FIG. 7 is a flowchart of sub-steps included in step S102 of FIG. 1;
fig. 8 is a flowchart of sub-steps included in step S1022 in fig. 7;
FIG. 9 is a flowchart of sub-steps involved in step S103 of FIG. 1;
fig. 10 is a block diagram of an electronic device according to an embodiment of the present application;
fig. 11 is a functional block diagram of an industrial equipment predictive maintenance device based on a health status index according to an embodiment of the present application.
Icon: 110-a storage medium; 120-a processor; 130-a state of health index based industrial equipment predictive maintenance device; 131-an acquisition module; 132-a determination module; 133-a prediction module; 140-communication interface.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. 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 application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present application, it should be noted that the features in the embodiments of the present application may be combined with each other without conflict.
Referring to fig. 1, a flowchart of a method for predictive maintenance of industrial equipment based on a health status index according to an embodiment of the present application is shown, where method steps defined by a flow related to the predictive maintenance method can be implemented by an electronic device having data analysis and processing functions. The electronic device may be a computer device or a server on which a platform for maintaining functions related to the industrial device is located. The specific process shown in FIG. 1 will be described in detail below.
S101, obtaining operation data of each time point in the operation process of the equipment to be tested, importing the operation data into a reconstruction model obtained through pre-training, and obtaining reconstruction data corresponding to the operation data.
S102, obtaining a health state index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health state index.
S103, importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training to fit the health state index, obtaining an extension curve based on the fit curve, comparing the predicted health state index on each time point in the extension curve with a preset threshold value, and determining the time point with the same predicted health state index and the preset threshold value as a failure time point.
In this embodiment, the device under test may be an industrial device, such as an industrial robot. From the time the device under test is put into use, generally, the performance state of the device under test is normal for a period of time from the start. However, as the time spent in service increases, the performance state of the device under test may begin to degrade and eventually fail.
In this embodiment, for the device to be tested that is put into use, the current time point of predictive maintenance may be used as a node, and the operation data of each time point in the operation process of the device to be tested before the node is obtained. The time point may be a set sampling point, and for example, the interval is 1 minute, 1 hour, and the like, but is not limited to be one sampling point.
The obtained operation data may include dynamic operation data and static operation data, wherein the dynamic operation data may include data such as real-time current, torque, shaft angle position, etc. during the operation of the device under test. The static operation data may include parameters of the device under test, such as the number of axes, degrees of freedom, and the like.
In this embodiment, the electromechanical control data inside the industrial equipment is used to implement predictive maintenance, and the current data and the like in the predictive maintenance have relatively important roles in control.
In this embodiment, the reconstruction model obtained by training in advance may be obtained by training based on sample data in advance. The reconstruction model can reflect the health state of the equipment in a data change characteristic mode by learning the characteristic of the health state change of the equipment reflected by the sample data. Therefore, in this embodiment, for the device to be tested, the operation data of the device to be tested may be imported into the pre-trained reconstruction model, so as to output the reconstruction data corresponding to the operation data.
Therefore, the reconstructed data can reflect the relevant condition of the health state of the equipment to be tested in the operation process. And the health-related condition may be embodied as a difference between the reconstructed data and the operational data. That is, the reconstructed data may be understood as feature data that fits a normal health state, and thus, a difference between the operating data and the reconstructed data may reflect a health state of the device under test.
In this embodiment, the health status index is obtained from the reconstruction data and the operating data. The health status index is a series of data in time sequence, that is, the health status index includes the health status index corresponding to each time point. Based on an analysis of the health status index corresponding to each time point, a degradation point in the time points may be determined.
In this embodiment, the degradation point represents a time point at which the health state of the device under test starts to degrade. That is, it can be understood that the operation data of the device under test is data belonging to a normal healthy state at each time point before the degradation point, and the operation of the device under test starts to deteriorate at each time point after the degradation point. However, degradation does not mean failure, and the degradation to a failure state occurs from the beginning of the operation of the device under test, and generally takes a while. And the operation data after the time point of starting to decline can provide effective data prediction basis for the prediction of the failure point.
In this embodiment, the prediction model may be obtained by training in advance, and the prediction model may be obtained by training in advance based on sample data. The sample data may be data relevant after the device's point of degradation as a sample. Therefore, the prediction model can learn the relevant features of the data after the degradation point, thereby accurately predicting the failure point based on the learned relevant features.
Therefore, in this embodiment, for the device to be tested, after the degradation point of the device to be tested in operation is determined, the health state index corresponding to the time point after the degradation point may be imported into the prediction model for prediction. The predictive model may fit the health status data and, upon obtaining the fitted curve, may further extend based on the fitted curve to obtain an extended curve. The extended curve also includes predicted health indices at a plurality of time points.
Whether the predicted health state index represents the failure state of the equipment to be tested can be judged based on a preset threshold value through setting the preset threshold value. If the predicted health status index is the same as the preset threshold value on the prediction curve, the corresponding time point may be determined as the failure time point.
For example, referring to FIG. 2, for example, 20-06-06 time points are identified as degradation points, and the data collected over the time period from the degradation point 20-06-06 to 21-11-28 is the health status index at each time point after the degradation point. The predictive model may be fitted to the health status index over the time period to obtain a fitted curve as shown in the figure for the time period 20-06-06 to 21-11-28.
Based on the curve trend of the fitted curve, an extension curve may be developed, for example, over the time period 21-11-28 to 23-05-22 in the figure. The constructed extension curve may be a plurality of extension curves within a certain error range.
The value represented by the horizontal dotted line in the figure may be the preset threshold, and the time point at which the extended curve intersects with the preset threshold, that is, the time point at which the predicted health state index is consistent with the preset threshold, may be determined as the failure time point. That is, the device under test is predicted to fail at this point in time.
The predictive maintenance method provided by the embodiment utilizes the reconstruction model and the prediction model obtained by pre-training, can accurately realize the determination of the degradation point and the prediction of data by learning the characteristics of the operating data, and can be suitable for predictive maintenance based on a small amount of data.
In this embodiment, the acquired operation data is the simple data of the device to be tested, such as current, torque, and the like, and it is difficult to embody the characteristics of the data in time sequence. In order to facilitate model learning or obtain the characteristics of the time series data, the operation data may be processed to some extent before being imported into the reconstructed model. Referring to fig. 3, in this embodiment, when processing the operation data based on the reconstruction model, the following method may be implemented:
s1011, intercepting the acquired running data of the equipment to be tested at a plurality of continuous time points according to a preset step length and a preset window length to obtain the running data in a plurality of time windows.
And S1012, zooming the operation data in each time window to a preset range.
And S1013, extracting the time domain feature vector, the frequency domain feature vector and the time-frequency domain feature vector of the scaled operating data.
And S1014, importing the time domain feature vector, the frequency domain feature vector and the time-frequency domain feature vector into a reconstruction model obtained by pre-training to obtain reconstruction data corresponding to the operating data.
In this embodiment, the obtained operation data may be formally defined and processed first, and each time point is targetedtVarious types of operational data may be stored at that point in timetThe data of (a) are expressed in the following form:
wherein the content of the first and second substances,mindicating the number of parts to be tested in the device under test, the point in timetCan be used in the health stateh k To indicate that the user is not in a normal position,h k the health state of the device under test at the time point can be characterized as a normal state or a fault state. For example,h k when the value is 1, the device to be tested is represented to be in a normal state,h k and when the value is 0, the device to be tested is represented as a fault state. The present invention is described herein by way of example only, and the present invention is not limited thereto.
Therefore, the operation data of the equipment to be tested at a plurality of time points which are continuous in time sequence can be obtained. On the basis, in order to facilitate the model to process the data, the operation data can be divided into data segments in a multi-segment time window to be input into the model. In this embodiment, the operation data may be intercepted according to a preset step length and a preset window length, so as to obtain the operation data in a plurality of time windows.
In order to ensure that the intercepted operation data is generally continuous, in this embodiment, the length of the preset window may be greater than the preset step length. Therefore, in two adjacent time windows, the data of the last part of the previous time window is overlapped with the data of the front part of the next time window, and the continuous section of the operation data in different time windows can be effectively guaranteed. As shown in fig. 4, for example, the preset step size may be 2 and the preset window length may be 7.
In this embodiment, in order to better analyze the nature of the distribution of each type of operation data, the operation data may be normalized, and the scaling value of the operation data in each time window may be within a preset range.
In one possible implementation, the operational data may be normalized using a z-score method (zero-mean normalization). The preset range may be an interval range with a mean value of 0 and a standard deviation of 1. That is, the scaled operation data is made to fall within the interval of 0 as the mean and 1 as the standard deviation.
In this embodiment, the scaling processing may be performed on the operation data according to the following scaling formula.
Wherein the content of the first and second substances,xrepresenting the operational data before scaling and,zrepresenting the scaled operational data as a function of time,Nindicates the total number of the operation data,
represents the average of the running data before scaling,
representing the standard deviation of the run data before scaling.
In this embodiment, z-score normalization may be performed on each type of operational data in the manner described above. The distribution of the data of the scaled running data is not changed, but the data distribution interval can be kept basically consistent after the scaling, and the data can be mainly distributed in the range of [ -2, 2 ]. By carrying out scaling processing on the operation data, the subsequent model can be more focused on analyzing the distribution condition of the data.
In order to further enable the model to obtain the distribution of the operating data from multiple dimension analysis, in this embodiment, features of the scaled operating data in multiple dimensions, including, for example, a time-domain feature vector, a frequency-domain feature vector, and a time-frequency-domain feature vector, may be extracted.
In this embodiment, for time domain feature extraction, feature analysis may be directly performed on the operating data of each time window. A plurality of time domain characteristics including effective values, square root amplitudes, peak-to-peak values, crest factors, margin indexes, skewness indexes, kurtosis indexes, form factors, pulse factors, information entropies and correlation coefficients can be extracted from the time domain. The time domain features are spliced into a vector to obtain a time domain feature vector as shown in the following:
in addition, frequency domain feature extraction can also be performed. From the buckivall theorem, the integral of the square of the signal amplitude, whether it is a real or complex signal, is equal to the energy of the signal, equal to the square of the mode of the signal spectral density. The formulation can be as follows:
wherein the content of the first and second substances,Ewhich is indicative of the energy of the signal,x(t) A time-domain value of the signal is represented,X(f) Representing the signal frequency-domain value.
Therefore, the energy of the high frequency discrete signal is obtained by adding the squares of each value in the operation data as follows:
wherein the content of the first and second substances,xf e the whole is characterized by a characteristic component,erepresenting an energy signal, and a method as described abovex rms 、x sra In (1)rms、sraAnd (7) corresponding.f(i) Is shown asx rms Value of the isochronous domain signaliAnd (4) respectively. The above formula can be understood as the frequency domain feature division to the left of the equal signThe quantity, equal to the sum of the modulo squares of the corresponding time-domain feature components.
Thus, a frequency domain feature vector can be obtained as follows:
on the basis of the above, time-frequency domain feature analysis can be performed. In this embodiment, the EDM method, the short-time fourier transform method, and the like may be used to perform time-frequency analysis. First obtaining the correlation with the fault of the device under test in each time window by EDMnThe eigenmode function (IMF). Respectively taking energy from the screened n IMFs by EDMxtf e Variance, variancextf sd Deviation indexxtf sf Index of degree of summitxtf kf 4 kinds of characteristic values are totally used, and the standard deviation of the instantaneous frequency is obtained by using STFT
Signal to noise ratio of instantaneous frequencySNRAnd 2 characteristic values are spliced to obtain the 4n + 2-dimensional time-frequency domain characteristic of the signal:
wherein, the abovextfThe whole represents one component.
And importing the obtained time domain feature vector, frequency domain feature vector and time-frequency domain feature vector into a reconstruction model to obtain reconstruction data corresponding to the operating data.
In this embodiment, the reconstructed model is obtained by training based on sample data in advance, and the predictive maintenance method provided in this embodiment further includes a step of obtaining the reconstructed model based on training of the constructed neural network model in advance, where the neural network model may be an LSTM model. The neural network model includes an encoder and a decoder, as shown in fig. 5. Referring to fig. 6, the step of obtaining the reconstructed model by pre-training may be implemented as follows:
s201, collecting sample data, wherein the sample data comprises data corresponding to a plurality of continuous time points.
And S202, importing the sample data into the encoder for encoding to obtain characteristic data.
And S203, importing the characteristic data and the sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data.
And S204, continuing training after adjusting the model parameters of the encoder and the decoder based on the loss function constructed according to the sample data and the sample reconstruction data until the reconstruction model is obtained when the preset requirements are met.
In this embodiment, the sample data may be operation data corresponding to a continuous time point in an operation process of the industrial device. Similarly, the sample data may be processed according to the above preprocessing, scaling processing, and multiple items of processing such as time domain feature extraction, frequency domain feature extraction, and time-frequency domain feature extraction.
And importing the processed sample data into an encoder of the neural network model for encoding to obtain characteristic data. In this embodiment, the encoder and decoder are each configured as an LSTM unit. LSTM can take time series data as input and then update its hidden state until the last step in the time series, denoted t2, the LSTM generated cell state containing all the information of the previous sequence, i.e. the LSTM generated cell state
. This cell state may also be referred to as Context Vectors (Context Vectors), and the decoder reconstructs the encoder input from the Context Vectors. The decoder, like the encoder, is also an LSTM unit. The input of each step in the decoder is the prediction of the previous step or the label of the previous step, and the updating of the hidden state of the decoder can be described as
。
And combining the characteristic data obtained by the encoder with sample data, including time domain characteristics, frequency domain characteristics and time-frequency domain characteristics of the sample data, in a decoder, and performing fusion and decoding processing to obtain sample reconstruction data.
And constructing a loss function based on the sample data and the sample reconstruction data, wherein the constructed loss function can be as follows:
wherein the content of the first and second substances,
the sample data is represented by a sample data,
representing the reconstructed data of the sample(s),t 1andt 2respectively representing a start time point and an end time point of the time series data.
The training of the encoder and decoder may have a number of iterations, after each iteration the function value of the loss function may be calculated, and the training may continue after the model parameters of the encoder and decoder are adjusted. When the iteration times reach the set maximum times, or the loss function reaches the state that the convergence is not reduced, or the iteration time reaches the set maximum length, the preset requirement can be judged to be met, and therefore the reconstruction model obtained by the neural network model at the moment is obtained.
The above process is a process of obtaining a reconstruction model through pre-training, when the operation data of the equipment to be tested is reconstructed by using the reconstruction model and the degradation point is determined, firstly, the reconstruction model is used for obtaining the reconstruction data corresponding to the operation data, then, the health state index is obtained based on the reconstruction data, and the degradation point in the time point is determined according to the health state index. Referring to fig. 7, in the present embodiment, the step of determining the degradation point may be implemented as follows:
and S1021, obtaining difference data between the reconstruction data and the operation data, and taking the difference data as a health state index.
And S1022, comparing the health state index with a health state threshold, and determining a time point when the health state index begins to deviate from the time point corresponding to the health state threshold as a degradation point.
In this embodiment, the health index may be the difference between actual operating data and data reconstructed (considered normal) by the reconstructed model. Therefore, the reconstruction error increases, which means that the operating state deviates from the normal state.
In this embodiment, a health state threshold may be preset as a criterion for determining whether the health state of the device under test is abnormal. The health state threshold may be set by related data in a process of constructing a reconstruction model based on sample data in advance. In this embodiment, the health state threshold may be constructed in the following manner:
and calculating a difference value between the sample data and the sample reconstruction data, calculating to obtain a difference average value and a difference standard deviation based on the difference value, and obtaining the health state threshold according to the difference average value and the difference standard deviation.
In this embodiment, a specific calculation formula of the health state threshold may be as follows:
wherein the content of the first and second substances,meanit is indicated that the average value is taken,stdrepresenting the standard deviation, |2Representing the L2 norm calculation.
When the health status index deviates from the health status threshold, i.e. the difference between the operational data and the reconstructed data exceeds the health status threshold, the corresponding point in time may be determined to be a degradation point.
Considering that some abrupt points may exist in the operation data in the actual processing process, some abrupt points also exist in the obtained health status index. If the health status index corresponding to the mutation point deviates from the health status threshold value due to the mutation in the data characteristics, the health status index at the corresponding time point may be erroneously determined as the degradation point. Therefore, referring to fig. 8, in the present embodiment, in the step of determining the degradation point, the following may be implemented:
s10221, a time point in the health status index at which the deviation from the health status threshold starts is acquired.
S10222, detecting whether the health status indexes corresponding to the set number of time points after the time point are all deviated from the health status threshold, if yes, executing the following step S10223, and if not, executing the following step S10224.
Step S10223, determining the time point as a degradation point.
Step S10224, determining that the time point is not a degradation point.
In this embodiment, if the health status index corresponding to a certain time point deviates from the health status threshold, it may be determined that there are no limitations to 5 time points, 10 time points, etc. after the certain time point. The health state indexes corresponding to the later time points can be obtained, whether the health state indexes deviate from the health state threshold value or not is detected, and if the health state indexes deviate from the health state threshold value, it is indicated that data continuously deviate from the health state threshold value within a longer time period, and the deviation is not accidental deviation caused by data mutation. Therefore, in this case, the point in time at which the deviation from the health state threshold value starts as the degradation point can be determined.
If the health status indexes at a set number of time points after the time point at which the deviation from the health status threshold value starts are not all deviated from the health status threshold value, it indicates that the health status indexes at the time point may be only the deviation caused by the abrupt change of the data. Therefore, it can be determined in this case that the above-described time point is not the degradation point.
In a possible implementation manner, the method may adopt the law of raydeta to eliminate abnormal points in a series of health status indexes in a time sequence, that is, to eliminate health status indexes with data mutation. Therefore, a degradation point where degradation of the device to be tested begins to occur can be found, and subsequent failure points are predicted based on the health state index after the degradation point.
Referring to fig. 9, in the present embodiment, when performing data fitting prediction based on the health state index corresponding to the time point after the degradation point, the following method may be implemented:
and S1031, acquiring the health state index corresponding to the time point after the degradation point of the equipment to be tested.
S1032, time window division is carried out on the health state indexes according to time sequence, and the health state indexes in a plurality of time windows are obtained.
S1033, normalize the health status index in each time window.
S1034, extracting the data characteristics of the health state index after normalization processing, and importing the data characteristics into a prediction model obtained by pre-training so as to fit the health state index.
In this embodiment, for the health state index of each time point after the degradation point of the device to be measured, the health state index may be intercepted according to a certain window length and a certain interception step length. Wherein, similarly, in order to guarantee the consistency of the health status indexes of the intercepted time windows, the window length may be larger than the interception step length.
For the health state indexes in each intercepted time window, the health state indexes can be normalized to a certain unified numerical range according to the scaling processing mode of the operation data. So that the predictive model can focus on the distribution characteristics of the data itself.
Data features of the normalized health status index may be extracted, and the data features may include, for example, time domain features, frequency domain features, time-frequency domain features, and the like. And importing the data characteristics of the health state index into a prediction model obtained by pre-training, wherein the prediction model can be used for fitting the health state index to obtain a fitting curve. And determining the failure time point based on the extension curve of the fitting curve.
In this embodiment, the prediction model may be obtained by training the constructed neural network model based on sample data in advance. For example, the health state index corresponding to the time point of the industrial equipment after the degradation point is used as sample data, and the neural network model may be a gru (gate recovery unit) network model.
On the basis, the predicted health state index obtained each time is compared with a preset threshold value, wherein the preset threshold value can be set based on the known operation condition of the industrial equipment of the same type which operates the same process as the equipment to be tested. When the predicted health state index is consistent with the preset threshold, the corresponding time point can be considered as a failure time point.
The remaining service life of the device to be tested may be determined based on the failure time point, for example, a time period from the node to the predicted failure time point by using a time point when the health state index fitting is performed by using the prediction model as the node is the remaining service life of the device to be tested.
The overall flow of the predictive maintenance method provided in this embodiment is described below.
In this embodiment, sample data may be collected in advance, and the sample data may be data corresponding to a plurality of continuous time points. The sample data can be real-time current, torque, values of shaft angle position, body parameters of the industrial equipment and the like in the operation process of the industrial equipment.
The data preprocessing may be performed on the sample data, for example, the sample data may be intercepted by a certain step length and a certain window size, so as to obtain the sample data in multiple windows. Moreover, data normalization may be performed, for example, scaling the sample data in each window to a preset range, such as a range of a certain mean value and a certain standard deviation.
And then, carrying out feature extraction on the scaled data, wherein the feature extraction comprises time domain feature extraction, frequency domain feature extraction and time-frequency domain feature extraction.
And importing the time domain characteristics, the frequency domain characteristics and the time-frequency domain characteristics obtained based on the sample data into the constructed neural network model to train the neural network model, so as to obtain a reconstructed model. In the training process, a loss function constructed based on difference information between input and output can be used as a training guide, and the training is stopped under the condition that iteration meets certain requirements.
In the process of training the reconstruction model, the health state threshold value can be further constructed and obtained based on the difference between the sample data input into the reconstruction model and the sample reconstruction data output by the reconstruction model. The state of health threshold can subsequently be used to determine a degradation point of the industrial device.
On the basis, the health state index can be obtained based on sample reconstruction data and sample data obtained by the reconstruction model, and then the time point of first beginning degradation in the health state index is found and used as a degradation point.
The depth model of the GRU network can be trained based on the health state index after the degradation point to obtain a prediction model.
On the basis, in the actual application stage, the operation data of the equipment to be tested can be obtained for the equipment to be tested. The above-mentioned data preprocessing, time window extraction processing, scaling processing, time domain feature processing, frequency domain feature processing, time-frequency domain feature processing, and the like are performed on the operation data.
And further importing the time domain characteristics, the frequency domain characteristics and the time-frequency domain characteristics into a reconstruction model to obtain corresponding reconstruction data. And obtaining the health state index of the equipment to be tested by combining the reconstruction data and the sample data of the equipment to be tested.
When the equipment to be tested is subjected to predictive maintenance, the operation data of the equipment to be tested can be imported into the reconstruction model to obtain corresponding reconstruction data. A health status index may be derived from the reconstructed data and the operational data. And analyzing and processing the health state index to obtain a degradation point which can represent that the health state of the equipment to be tested begins to degrade.
And importing the health state index after the degradation point into a prediction model, and fitting the health state index to obtain a fitting curve. And extending based on the fitted curve to obtain an extended curve, wherein each time point on the extended curve has a corresponding predicted health state index.
And comparing each predicted health state index with a preset threshold value, and determining the corresponding time point as a predicted failure time point when the predicted health state index is the same as the preset threshold value. And obtaining a connection point of the fitting curve and the extension curve, namely a difference value between a time point predicted by using the prediction model and a predicted failure time point, namely the predicted residual service life of the equipment to be tested.
The predictive maintenance method provided by the embodiment adopts the health state index as the index of the predictive maintenance of the industrial equipment based on the health state index, reduces the dependence of the predictive maintenance on various sensors, and reduces the actual application cost of the predictive maintenance technology.
In addition, the reconstruction model comprising the encoder and the decoder is adopted to output reconstruction data so as to construct the health state index, the numerical value of the health state index can be extracted from the time sequence signal, the cost for constructing the model is reduced, and further, on the basis of accurately determining the degradation point, a data basis can be provided for the accurate prediction of the subsequent failure point.
When the failure point is predicted, the characteristics of the time sequence basis are extracted in a GRU deep learning mode, the health state index is predicted, the residual service life is calculated, and the health state monitoring precision is improved.
Referring to fig. 10, a schematic diagram of exemplary components of an electronic device according to an embodiment of the present disclosure is provided, where the electronic device may include a storage medium 110, a processor 120, a health index-based industrial device predictive maintenance apparatus 130, and a communication interface 140. In this embodiment, the storage medium 110 and the processor 120 are both located in the electronic device and are separately disposed. However, it should be understood that the storage medium 110 may be separate from the electronic device and may be accessed by the processor 120 through a bus interface. Alternatively, the storage medium 110 may be integrated into the processor 120, for example, may be a cache and/or general purpose registers.
The health status index-based industrial equipment predictive maintenance device 130 may be understood as the electronic equipment, or the processor 120 of the electronic equipment, or may be understood as a software functional module that is independent of the electronic equipment or the processor 120 and implements the predictive maintenance method under the control of the electronic equipment.
As shown in fig. 11, the aforementioned state of health index-based industrial equipment predictive maintenance device 130 may include an obtaining module 131, a determining module 132, and a predicting module 133. The functions of the functional modules of the health index-based industrial equipment predictive maintenance device 130 are described in detail below.
The obtaining module 131 is configured to obtain operation data of each time point in the operation of the device to be tested, and import the operation data into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operation data.
It is understood that the obtaining module 131 can be used to execute the step S101, and for the detailed implementation of the obtaining module 131, reference can be made to the contents related to the step S101.
A determining module 132, configured to obtain a health state index according to the reconstruction data and the operation data, and determine a degradation point in a time point according to the health state index, where the degradation point represents a time point at which the health state of the device to be tested starts to degrade.
It is understood that the determining module 132 can be used to execute the step S102, and the detailed implementation of the determining module 132 can refer to the content related to the step S102.
The prediction module 133 is configured to import the health state index corresponding to the time point after the degradation point into a prediction model obtained through pre-training to fit the health state index, obtain an extension curve based on the fit curve, compare the predicted health state index at each time point in the extension curve with a preset threshold, and determine a time point at which the predicted health state index is the same as the preset threshold as a failure time point.
It is understood that the prediction module 133 can be used to perform the step S103, and for the detailed implementation of the prediction module 133, reference can be made to the above-mentioned content related to the step S103.
In a possible implementation, the obtaining module 131 may be configured to:
intercepting the operation data of the equipment to be tested at a plurality of continuous time points according to a preset step length and a preset window length to obtain the operation data in a plurality of time windows;
for the operation data in each time window, scaling the operation data to be in a preset range;
extracting a time domain characteristic vector, a frequency domain characteristic vector and a time-frequency domain characteristic vector of the scaled operation data;
and importing the time domain feature vector, the frequency domain feature vector and the time-frequency domain feature vector into a reconstruction model obtained by pre-training to obtain reconstruction data corresponding to the operating data.
In a possible embodiment, the preset window length is greater than the preset step length.
In one possible implementation, the determining module 132 may be configured to:
obtaining difference data between the reconstruction data and the operation data, and taking the difference data as a health state index;
and comparing the health state index with a health state threshold value, and determining a time point when the health state index begins to deviate from the health state threshold value as a degradation point.
In one possible implementation, the determining module 132 may be configured to:
acquiring a time point in the health status index at which the health status index begins to deviate from the health status threshold;
and detecting whether the health state indexes respectively corresponding to the set number of time points after the time point all deviate from the health state threshold value, and if so, determining that the time point is a degradation point.
In a possible implementation, the health index-based industrial equipment predictive maintenance apparatus 130 further includes a building module for obtaining the reconstructed model based on a pre-training of a built neural network model, where the neural network model includes an encoder and a decoder, and the building module is configured to:
collecting sample data, wherein the sample data comprises data corresponding to a plurality of continuous time points;
importing the sample data into the encoder to perform encoding processing to obtain characteristic data;
importing the characteristic data and the sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;
and continuing training after adjusting the model parameters of the encoder and the decoder based on the loss function constructed according to the sample data and the sample reconstruction data until the preset requirements are met, and obtaining the reconstruction model.
In one possible embodiment, the health index-based industrial equipment predictive maintenance device 130 further comprises an obtaining module for obtaining the health threshold, and the obtaining module can be used for:
calculating a difference between the sample data and sample reconstruction data;
calculating to obtain a difference average value and a difference standard deviation based on the difference value;
and obtaining the health state threshold value according to the difference average value and the difference standard deviation.
In a possible implementation, the prediction module 133 may be configured to:
acquiring a health state index corresponding to a time point after a degradation point of the equipment to be tested;
dividing the health state indexes into time windows according to a time sequence to obtain the health state indexes in a plurality of time windows;
normalizing the health state index in each time window;
and extracting the data characteristics of the health state index after normalization processing, and introducing the data characteristics into a prediction model obtained by pre-training so as to fit the health state index.
The description of the processing flow of each module in the device and the interaction flow between the modules may refer to the related description in the above method embodiments, and will not be described in detail here.
Further, embodiments of the present application also provide a computer-readable storage medium, in which machine-executable instructions are stored, and when executed, the machine-executable instructions implement the predictive maintenance method provided by the foregoing embodiments.
In particular, the computer readable storage medium can be a general storage medium, such as a removable disk, a hard disk, etc., and the computer program on the computer readable storage medium can be executed to perform the predictive maintenance method. With regard to the processes involved when the executable instructions in the computer-readable storage medium are executed, reference may be made to the related descriptions in the above method embodiments, which are not described in detail herein.
In summary, according to the method and the device for predictive maintenance of industrial equipment based on health state indexes and the electronic equipment provided by the embodiment of the application, the operation data at each time point in the operation process of the equipment to be tested is acquired, the operation data is imported into the reconstruction model obtained through pre-training, the reconstruction data corresponding to the operation data is obtained, the health state indexes are obtained according to the reconstruction data and the operation data, and the degradation points are determined according to the health state indexes. And then, importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training for fitting to obtain an extension curve, comparing the predicted health state index of each time point on the extension curve with a preset threshold value, and determining the time point with the same index as the failure time point. In the scheme, by using a pre-trained reconstruction model and a pre-trained prediction model, the determination of the degradation point and the prediction of the data can be accurately realized by learning the characteristics of the operating data, and the method can be suitable for predictive maintenance based on a small amount of data.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method for predictive maintenance of industrial equipment based on a state of health index, the method comprising:
acquiring operation data of each time point in the operation process of equipment to be tested, and importing the operation data into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operation data;
obtaining a health state index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health state index, wherein the degradation point represents the time point when the health state of the equipment to be tested begins to degrade;
and importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training to fit the health state index, obtaining an extension curve based on the fit curve, comparing the predicted health state index on each time point in the extension curve with a preset threshold value, and determining the time point with the same predicted health state index and the preset threshold value as a failure time point.
2. The method for predictive maintenance of industrial equipment based on health status index as claimed in claim 1, wherein the step of importing the operation data into a pre-trained reconstructed model to obtain reconstructed data corresponding to the operation data comprises:
intercepting the operation data of the equipment to be tested at a plurality of continuous time points according to a preset step length and a preset window length to obtain the operation data in a plurality of time windows;
for the operation data in each time window, scaling the operation data to be in a preset range;
extracting a time domain characteristic vector, a frequency domain characteristic vector and a time-frequency domain characteristic vector of the scaled operation data;
and importing the time domain feature vector, the frequency domain feature vector and the time-frequency domain feature vector into a reconstruction model obtained by pre-training to obtain reconstruction data corresponding to the operating data.
3. The predictive health-state-index-based maintenance method for industrial equipment of claim 2, wherein the preset window length is greater than the preset step size.
4. The method of claim 1, wherein the step of deriving the health index from the reconstructed data and the operational data and determining the degradation point in time from the health index comprises:
obtaining difference data between the reconstruction data and the operation data, and taking the difference data as a health state index;
and comparing the health state index with a health state threshold value, and determining a time point when the health state index begins to deviate from the health state threshold value as a degradation point.
5. The method of claim 4, wherein the step of determining a point in time at which the health index begins to deviate from the health threshold as a degradation point comprises:
acquiring a time point in the health status index at which the health status index begins to deviate from the health status threshold;
and detecting whether the health state indexes respectively corresponding to the set number of time points after the time point all deviate from the health state threshold value, and if so, determining that the time point is a degradation point.
6. The method for predictive health-state-index-based maintenance of industrial equipment, according to claim 4, further comprising the step of obtaining said reconstructed model based on a pre-constructed neural network model training, said neural network model comprising an encoder and a decoder, said step comprising:
collecting sample data, wherein the sample data comprises data corresponding to a plurality of continuous time points;
importing the sample data into the encoder to perform encoding processing to obtain characteristic data;
importing the characteristic data and the sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;
and continuing training after adjusting the model parameters of the encoder and the decoder based on the loss function constructed according to the sample data and the sample reconstruction data until the preset requirements are met, and obtaining the reconstruction model.
7. The method of claim 6, wherein the health status threshold is obtained by:
calculating a difference between the sample data and sample reconstruction data;
calculating to obtain a difference average value and a difference standard deviation based on the difference value;
and obtaining the health state threshold value according to the difference average value and the difference standard deviation.
8. The method of claim 1, wherein the step of importing the health index corresponding to the time point after the degradation point into a pre-trained predictive model to fit the health index comprises:
acquiring a health state index corresponding to a time point after a degradation point of the equipment to be tested;
dividing the health state indexes into time windows according to a time sequence to obtain the health state indexes in a plurality of time windows;
normalizing the health state index in each time window;
and extracting the data characteristics of the health state index after normalization processing, and introducing the data characteristics into a prediction model obtained by pre-training so as to fit the health state index.
9. An apparatus for predictive maintenance of industrial equipment based on a state of health index, the apparatus comprising:
the acquisition module is used for acquiring operation data of each time point in the operation of the equipment to be tested, and importing the operation data into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operation data;
the determining module is used for obtaining a health state index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health state index, wherein the degradation point represents the time point when the health state of the equipment to be tested begins to degrade;
and the prediction module is used for importing the health state index corresponding to the time point after the degradation point into a prediction model obtained by pre-training so as to fit the health state index, obtaining an extension curve based on the fit curve, comparing the predicted health state index on each time point in the extension curve with a preset threshold value, and determining the time point with the same predicted health state index and the preset threshold value as a failure time point.
10. An electronic device comprising one or more storage media and one or more processors in communication with the storage media, the one or more storage media storing processor-executable machine-executable instructions that, when executed by the electronic device, are executed by the processors to perform the method steps of any of claims 1-8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210200518.4A CN114298443B (en) | 2022-03-03 | 2022-03-03 | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment |
| PCT/CN2022/089549 WO2023165006A1 (en) | 2022-03-03 | 2022-04-27 | Predictive maintenance method and apparatus for industrial equipment based on health status index, and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210200518.4A CN114298443B (en) | 2022-03-03 | 2022-03-03 | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114298443A true CN114298443A (en) | 2022-04-08 |
| CN114298443B CN114298443B (en) | 2022-06-14 |
Family
ID=80978489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210200518.4A Active CN114298443B (en) | 2022-03-03 | 2022-03-03 | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN114298443B (en) |
| WO (1) | WO2023165006A1 (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114841021A (en) * | 2022-07-04 | 2022-08-02 | 北京航空航天大学杭州创新研究院 | Method and device for correcting digital twin model, electronic device and storage medium |
| CN115307896A (en) * | 2022-09-20 | 2022-11-08 | 重庆忽米网络科技有限公司 | Equipment health state detection method based on machine learning |
| CN115563095A (en) * | 2022-10-18 | 2023-01-03 | 上海宇佑船舶科技有限公司 | Data reconstruction method and system based on time sequence |
| CN116401137A (en) * | 2023-06-06 | 2023-07-07 | 中诚华隆计算机技术有限公司 | Core particle health state prediction method and device, electronic equipment and storage medium |
| WO2023165006A1 (en) * | 2022-03-03 | 2023-09-07 | 北京航空航天大学杭州创新研究院 | Predictive maintenance method and apparatus for industrial equipment based on health status index, and electronic device |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117056709A (en) * | 2023-10-11 | 2023-11-14 | 腾讯科技（深圳）有限公司 | Training method and device of time sequence prediction model, storage medium and electronic equipment |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018076475A1 (en) * | 2016-10-26 | 2018-05-03 | 广东产品质量监督检验研究院 | Photovoltaic assembly accelerated degradation model established based on deep approach of learning, and method for predicting photovoltaic assembly lifetime |
| CN109948860A (en) * | 2019-03-26 | 2019-06-28 | 哈工大机器人（合肥）国际创新研究院 | A kind of mechanical system method for predicting residual useful life and system |
| CN111222290A (en) * | 2020-01-13 | 2020-06-02 | 浙江工业大学 | Large-scale equipment residual service life prediction method based on multi-parameter feature fusion |
| CN111984513A (en) * | 2020-08-25 | 2020-11-24 | 浙江天垂科技有限公司 | Predictive maintenance method, device, equipment and storage medium |
| CN112418277A (en) * | 2020-11-03 | 2021-02-26 | 西安电子科技大学 | Method, system, medium, and apparatus for predicting remaining life of rotating machine component |
| CN113240099A (en) * | 2021-07-09 | 2021-08-10 | 北京博华信智科技股份有限公司 | LSTM-based rotating machine health state prediction method and device |
| US20210256400A1 (en) * | 2020-02-19 | 2021-08-19 | Vyber Power Systems, Inc. | Power device with self-health status prediction function and self-health status prediction method thereof and cloud server suitable for a plurality of power devices |
| CN113434970A (en) * | 2021-06-01 | 2021-09-24 | 北京交通大学 | Health index curve extraction and service life prediction method for mechanical equipment |
| CN113469300A (en) * | 2021-09-06 | 2021-10-01 | 北京航空航天大学杭州创新研究院 | Equipment state detection method and related device |
| CN113536681A (en) * | 2021-07-21 | 2021-10-22 | 北京航空航天大学 | Electric steering engine health assessment method based on time sequence extrapolation prediction |
| CN113742178A (en) * | 2021-09-18 | 2021-12-03 | 北京航空航天大学 | Network node health state monitoring method based on LSTM |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104166787B (en) * | 2014-07-17 | 2017-06-13 | 南京航空航天大学 | A kind of aero-engine method for predicting residual useful life based on multistage information fusion |
| US20220004182A1 (en) * | 2020-07-02 | 2022-01-06 | Nec Laboratories America, Inc. | Approach to determining a remaining useful life of a system |
| CN114298443B (en) * | 2022-03-03 | 2022-06-14 | 北京航空航天大学杭州创新研究院 | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment |
-
2022
- 2022-03-03 CN CN202210200518.4A patent/CN114298443B/en active Active
- 2022-04-27 WO PCT/CN2022/089549 patent/WO2023165006A1/en unknown
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018076475A1 (en) * | 2016-10-26 | 2018-05-03 | 广东产品质量监督检验研究院 | Photovoltaic assembly accelerated degradation model established based on deep approach of learning, and method for predicting photovoltaic assembly lifetime |
| CN109948860A (en) * | 2019-03-26 | 2019-06-28 | 哈工大机器人（合肥）国际创新研究院 | A kind of mechanical system method for predicting residual useful life and system |
| CN111222290A (en) * | 2020-01-13 | 2020-06-02 | 浙江工业大学 | Large-scale equipment residual service life prediction method based on multi-parameter feature fusion |
| US20210256400A1 (en) * | 2020-02-19 | 2021-08-19 | Vyber Power Systems, Inc. | Power device with self-health status prediction function and self-health status prediction method thereof and cloud server suitable for a plurality of power devices |
| CN111984513A (en) * | 2020-08-25 | 2020-11-24 | 浙江天垂科技有限公司 | Predictive maintenance method, device, equipment and storage medium |
| CN112418277A (en) * | 2020-11-03 | 2021-02-26 | 西安电子科技大学 | Method, system, medium, and apparatus for predicting remaining life of rotating machine component |
| CN113434970A (en) * | 2021-06-01 | 2021-09-24 | 北京交通大学 | Health index curve extraction and service life prediction method for mechanical equipment |
| CN113240099A (en) * | 2021-07-09 | 2021-08-10 | 北京博华信智科技股份有限公司 | LSTM-based rotating machine health state prediction method and device |
| CN113536681A (en) * | 2021-07-21 | 2021-10-22 | 北京航空航天大学 | Electric steering engine health assessment method based on time sequence extrapolation prediction |
| CN113469300A (en) * | 2021-09-06 | 2021-10-01 | 北京航空航天大学杭州创新研究院 | Equipment state detection method and related device |
| CN113742178A (en) * | 2021-09-18 | 2021-12-03 | 北京航空航天大学 | Network node health state monitoring method based on LSTM |
Non-Patent Citations (3)
| Title |
|---|
| JIANXIN SHI 等: ""A Cloud-edge Collaborative Architecture for Data-driven Health Condition Monitoring of Machines"", 《2021 IEEE INTERNATIONAL CONFERENCE ON JOINT CLOUD COMPUTING》 * |
| RUI WANG 等: ""Cloud-Edge Collaborative Industrial Robotic Intelligent Service Platform"", 《2020 IEEE INTERNATIONAL CONFERENCE ON JOINT CLOUD COMPUTING》 * |
| 韩林洁 等: ""基于一维卷积神经网络的轴承剩余寿命预测"", 《制造业自动化》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023165006A1 (en) * | 2022-03-03 | 2023-09-07 | 北京航空航天大学杭州创新研究院 | Predictive maintenance method and apparatus for industrial equipment based on health status index, and electronic device |
| CN114841021A (en) * | 2022-07-04 | 2022-08-02 | 北京航空航天大学杭州创新研究院 | Method and device for correcting digital twin model, electronic device and storage medium |
| CN115307896A (en) * | 2022-09-20 | 2022-11-08 | 重庆忽米网络科技有限公司 | Equipment health state detection method based on machine learning |
| CN115563095A (en) * | 2022-10-18 | 2023-01-03 | 上海宇佑船舶科技有限公司 | Data reconstruction method and system based on time sequence |
| CN115563095B (en) * | 2022-10-18 | 2023-11-24 | 上海宇佑船舶科技有限公司 | Data reconstruction method and system based on time sequence |
| CN116401137A (en) * | 2023-06-06 | 2023-07-07 | 中诚华隆计算机技术有限公司 | Core particle health state prediction method and device, electronic equipment and storage medium |
| CN116401137B (en) * | 2023-06-06 | 2023-09-26 | 中诚华隆计算机技术有限公司 | Core particle health state prediction method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023165006A1 (en) | 2023-09-07 |
| CN114298443B (en) | 2022-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114298443B (en) | Industrial equipment predictive maintenance method and device based on health state index and electronic equipment | |
| CN113868006B (en) | Time sequence detection method and device, electronic equipment and computer storage medium | |
| CN112101554B (en) | Abnormality detection method and apparatus, device, and computer-readable storage medium | |
| CN109697207A (en) | The abnormality monitoring method and system of time series data | |
| Ping et al. | Degradation feature extraction using multi-source monitoring data via logarithmic normal distribution based variational auto-encoder | |
| CN113671917B (en) | Detection method, system and equipment for abnormal state of multi-modal industrial process | |
| Yu | A nonlinear probabilistic method and contribution analysis for machine condition monitoring | |
| CN107516279B (en) | Automatic early warning method for network public sentiment | |
| WO2014058575A2 (en) | Modeling data generating process | |
| Kaji et al. | Constructing a reliable health indicator for bearings using convolutional autoencoder and continuous wavelet transform | |
| CN114169604A (en) | Performance index abnormality detection method, abnormality detection device, electronic apparatus, and storage medium | |
| Qiao et al. | Fault diagnosis method of rolling bearings based on VMD and MDSVM | |
| WO2023196129A1 (en) | Anomaly detection using multiple detection models | |
| CN114298299A (en) | Model training method, device, equipment and storage medium based on course learning | |
| CN113902260A (en) | Information prediction method, information prediction device, electronic equipment and medium | |
| CN116647405B (en) | Multipath QUIC abnormal flow detection method, system and equipment | |
| Lin | An integrated procedure for bayesian reliability inference using MCMC | |
| Xiao et al. | A new fault feature extraction method of rolling bearings based on the improved self-selection ICEEMDAN-permutation entropy | |
| CN115510901A (en) | Fault identification method and device for carrier roller of belt conveyor | |
| Sun et al. | Leakage aperture identification of natural gas pipeline based on compressed acquisition and DSAE | |
| Lu et al. | Weak monotonicity with trend analysis for unsupervised feature evaluation | |
| Yang et al. | Design and application of deep belief network based on stochastic adaptive particle swarm optimization | |
| CN112270574A (en) | Method, device, equipment and medium for analyzing abnormal motion in motion execution process | |
| Ding et al. | Prediction of remaining useful life of rolling bearing based on fractal dimension and convolutional neural network | |
| CN115222164B (en) | Water pump fault prediction method and system based on empirical coupling function |
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 |