WO2023165006A1

WO2023165006A1 - Predictive maintenance method and apparatus for industrial equipment based on health status index, and electronic device

Info

Publication number: WO2023165006A1
Application number: PCT/CN2022/089549
Authority: WO
Inventors: 郭晓辉; 牟许东; 王瑞; 刘旭东
Original assignee: 北京航空航天大学杭州创新研究院
Priority date: 2022-03-03
Filing date: 2022-04-27
Publication date: 2023-09-07
Also published as: CN114298443A; CN114298443B

Abstract

A predictive maintenance method and apparatus for industrial equipment based on a health status index, and an electronic device. The method comprises: acquiring operation data at 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 status index according to the reconstruction data and the operation data, and determining a degradation point according to the health status index; and importing a health status index corresponding to a time point after the degradation point into a pre-trained prediction model for fitting, obtaining an extension curve, comparing the predicted health status data at each time point in the extension curve with a preset threshold, and determining a time point when the predicted health status data is consistent with the preset threshold as a failure time point. The pre-trained reconstruction model and prediction model are used, so that the determination of the degradation point and the prediction of the data can be accurately realized by learning the characteristics of the operation data, and the method can be suitable for predictive maintenance based on a small amount of data.

Description

Method, device and electronic equipment for predictive maintenance of industrial equipment based on health status index

Cross References to Related Applications

This application claims the priority of the Chinese patent application with the application number 202210200518.4 and the title "Predictive Maintenance Method, Device and Electronic Equipment for Industrial Equipment Based on Health Status Index" submitted to the State Intellectual Property Office of China on March 3, 2022. The entire contents of which are incorporated by reference in this application.

technical field

The present application relates to the technical field of industrial equipment management, in particular, to a health state index-based predictive maintenance method, device and electronic equipment for industrial equipment.

Background technique

Industrial equipment maintenance is of great significance to the economic efficiency of manufacturing industry. Preventive maintenance, that is, regular maintenance, can fully prevent the economic loss caused by the downtime of the robot during the production process, but it increases the workload of the operation and maintenance personnel and causes a lot of waste of parts, which increases the maintenance cost of the robot. Therefore, the industry is exploring predictive maintenance technology for industrial equipment in order to perform maintenance when the performance of the equipment drops to the minimum, thereby saving maintenance costs.

In related technologies, the predictive maintenance of industrial equipment mainly uses multi-sensors to collect multi-source data, and constructs single or composite health status indicators, so as to predict the failure of industrial equipment. However, the method of setting multiple sensors to collect data for prediction in the related art is mainly applicable to large-scale equipment. When performing predictive maintenance on small equipment such as industrial robots, it becomes unrealistic to carry a large number of sensors due to the limitations of the actual working environment and the cost of the process itself and sensors. Therefore, for small equipment, it is very important to accurately implement predictive maintenance without setting up a large number of sensors to collect multi-source data.

Contents of the invention

The present application provides a health state index-based predictive maintenance method, device and electronic equipment for industrial equipment, which can accurately determine degradation points and predict data.

The embodiment of the application can be realized like this:

Some embodiments of the present application provide a health state index-based predictive maintenance method for industrial equipment, the method may include:

Obtaining operation data at various time points during the operation of the device under test, importing the operation data into a pre-trained reconstruction model, and obtaining reconstruction data corresponding to the operation data;

Obtaining a health status index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health status index, and the degradation point represents a time point when the health status of the device under test begins to degrade;

Import the health status index corresponding to the time point after the degradation point into the prediction model obtained in advance to fit the health status index, and obtain an extension curve based on the fitting curve, and each of the extension curves The predicted health state index at the time point is compared with the preset threshold, and the time point at which the predicted health state index is the same as the preset threshold is determined as the failure time point.

Health status index In an optional implementation manner, the step of importing the operation data into the pre-trained reconstruction model to obtain the reconstruction data corresponding to the operation data may include:

For the obtained operating data of the equipment under test at multiple consecutive time points, the operating data is intercepted according to the preset step size and the preset window length, and the operating data in multiple time windows are obtained;

For the operating data in each time window, scaling the operating data to a preset range;

Extracting time-domain feature vectors, frequency-domain feature vectors and time-frequency-domain feature vectors of the scaled operating data;

Importing the time-domain feature vector, frequency-domain feature vector, and time-frequency-domain feature vector into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operating data.

In an optional implementation manner, the preset window length is greater than the preset step size.

In an optional embodiment, the step of obtaining the health status index according to the reconstruction data and the running data, and determining the degradation point in time according to the health status index may include:

Obtaining difference data between the reconstruction data and the running data, and using the difference data as a health status index;

The health status index is compared with the health status threshold, and the time point at which the health status index starts to deviate from the health status threshold is determined as a degradation point.

In an optional implementation manner, the step of determining the time point when the health status index starts to deviate from the health status threshold as the degradation point may include:

Obtain the time point when the health status index starts to deviate from the health status threshold;

Detecting whether the health state indices corresponding to the set number of time points after the time point deviate from the health state threshold, and if they deviate, determine that the time point is a degradation point.

In an optional embodiment, the method also includes the step of obtaining the reconstructed model based on pre-built neural network model training, the neural network model includes an encoder and a decoder, and this step may include:

collecting sample data, where the sample data includes data corresponding to multiple consecutive time points;

Importing the sample data into the encoder for encoding processing to obtain feature data;

Importing the feature data and sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;

After adjusting the model parameters of the encoder and decoder based on the loss function constructed according to the sample data and the sample reconstruction data, the training is continued until the preset requirements are met, and the reconstruction model is obtained.

In an optional implementation manner, the health status threshold is obtained in the following manner:

calculating the difference between the sample data and the sample reconstructed data;

Calculate the difference mean value and difference standard deviation based on the difference value;

The health status threshold is obtained according to the difference mean value and the difference standard deviation.

In an optional embodiment, the step of importing the health status index corresponding to the time point after the degradation point into the pre-trained prediction model to obtain the prediction data in the next prediction period may include:

Obtaining the health status index corresponding to the time point after the degradation point of the device under test;

Dividing the health status index into time windows according to time series to obtain health status indices in multiple time windows;

Normalize the health status index in each time window;

The data features of the health status index after normalization are extracted, and the data features are imported into the pre-trained prediction model to fit the health status index.

Other embodiments of the present application provide a device for predictive maintenance of industrial equipment based on a health state index, the device comprising:

An acquisition module, configured to acquire operating data at various time points during the operation of the device to be tested, import the operating data into a pre-trained reconstruction model, and obtain reconstruction data corresponding to the operating data;

A determining module, configured to obtain a health status index according to the reconstruction data and the operating data, and determine a degradation point in a time point according to the health status index, and the degradation point indicates that the health status of the device under test begins to degrade point in time;

The prediction module is used to import the health status index corresponding to the time point after the degradation point into the prediction model obtained by pre-training to fit the health status index, and obtain an extension curve based on the fitting curve, and the The predicted health status index at each time point in the extension curve is compared with a preset threshold, and the time point at which the predicted health status index is the same as the preset threshold is determined as the failure time point.

In an optional implementation manner, the acquisition module may also be configured to:

In an optional implementation manner, the determining module may also be configured to:

In an optional embodiment, the device for predictive maintenance of industrial equipment based on the health state index further includes a building block for obtaining the reconstruction model based on the training of the constructed neural network model in advance, and the neural network model includes an encoder and decoders, the building blocks can also be configured to:

In an optional embodiment, the device for predictive maintenance of industrial equipment based on the health state index further includes an obtaining module for obtaining the health state threshold, and the obtaining module may be configured to:

In an optional implementation manner, the prediction module may be configured to:

Normalize the health status index in each time window;

Still other embodiments of the present application provide an electronic device, which may include one or more storage media and one or more processors communicating with the storage medium, and one or more storage media store a machine-executable program executable by the processor. Instructions, when the electronic device is running, the processor executes the machine-executable instructions, so as to execute the method steps described in any one of the foregoing embodiments.

The beneficial effects of the embodiments of the present application at least include, for example:

This application provides a method, device, and electronic equipment for predictive maintenance of industrial equipment based on the health state index. By obtaining the operating data at various time points during the operating process of the equipment to be tested, the operating data is imported into the pre-trained reconstruction model to obtain The reconstructed data corresponding to the operating data, the health status index is obtained according to the reconstructed data and the operating data, and the degradation point is determined according to the health status index. Then import the health status index corresponding to the time point after the degradation point into the pre-trained prediction model to fit and obtain the extension curve, compare the predicted health status index at each time point on the extension curve with the preset threshold, and compare the two The consistent time point is determined as the failure time point. In this solution, the pre-trained reconstruction model and prediction model can be used to accurately determine the degradation point and predict the data by learning the characteristics of the operating data, which is suitable for predictive maintenance based on a small amount of data.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the accompanying drawings that are required in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and thus It should be regarded as a limitation on the scope, and those skilled in the art can also obtain other related drawings based on these drawings without creative work.

Fig. 1 is a flow chart of the predictive maintenance method provided by the embodiment of the present application;

Fig. 2 is the schematic diagram of fitting curve and extension curve provided by the embodiment of the present application;

Fig. 3 is the flowchart of the substep that step S101 comprises in Fig. 1;

FIG. 4 is a schematic diagram of time window data interception in the embodiment of the present application;

FIG. 5 is a schematic diagram of the processing of the reconstruction model provided in the embodiment of the present application;

FIG. 6 is a flow chart of the reconstruction model training method provided by the embodiment of the present application;

Fig. 7 is the flowchart of the substep that step S102 comprises in Fig. 1;

FIG. 8 is a flow chart of the substeps included in step S1022 in FIG. 7;

Fig. 9 is the flowchart of the substep that step S103 comprises in Fig. 1;

FIG. 10 is a structural block diagram of an electronic device provided by an embodiment of the present application;

FIG. 11 is a block diagram of functional modules of a device for predictive maintenance of industrial equipment based on a health state index provided by an embodiment of the present application.

Icons: 110-storage medium; 120-processor; 130-predictive maintenance device for industrial equipment based on health status index; 131-acquisition module; 132-determination module; 133-prediction module; 140-communication interface.

Detailed ways

In order to make the purposes, 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 in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments It is a part of the embodiments of this application, not all of them. The components of the embodiments of the application generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations.

Accordingly, the following detailed description of the embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the application. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of this application.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

In the description of the present application, it should be noted that the features in the embodiments of the present application can be combined without conflict.

Please refer to Fig. 1, which is a flow chart of the method for predictive maintenance of industrial equipment based on the health status index provided by the embodiment of the present application. implemented by the device. The electronic device may be a computer device, or a server on which a platform for maintaining relevant functions of industrial equipment is located. The specific process shown in FIG. 1 will be described in detail below.

S101. Obtain operating data at various time points during the operating process of the device under test, import the operating data into a pre-trained reconstruction model, and obtain reconstruction data corresponding to the operating data.

S102. Obtain a health status index according to the reconstruction data and the running data, and determine a degradation point in a time point according to the health status index.

S103. Import the health status index corresponding to the time point after the degradation point into the prediction model obtained in advance to fit the health status index, and obtain an extension curve based on the fitting curve, and insert the extension curve into The predicted health status index at each time point is compared with the preset threshold, and the time point at which the predicted health status index is the same as the preset threshold is determined as the failure time point.

In this embodiment, the device under test may be an industrial device, such as an industrial robot. After the device under test is put into use, generally, the performance status of the device under test is normal within a period of time. However, with the increase of the time in use, the performance status of the equipment under test may begin to degrade, and eventually fail.

In this embodiment, for the equipment under test put into use, the current time point of predictive maintenance can be used as a node, and the operation data of each time point in the operation process of the equipment under test before the node can be obtained. The time point may be a set sampling point, for example, intervals of 1 minute, 1 hour, etc. are not limited as a sampling point.

The obtained operation data may include dynamic operation data and static operation data, wherein the dynamic operation data may include real-time current, torque, shaft angular position and other data during operation of the device under test. The static operating data may include the body parameters of the device under test, such as the number of axes, degrees of freedom, and so on.

In this embodiment, the predictive maintenance is realized by using the electromechanical control data inside the industrial equipment, in which the current data and the like play an important role in the control.

In this embodiment, the reconstructed model may also be pre-trained, and the reconstructed model is obtained by pre-training based on sample data. The reconstruction model can reflect the health state of the device in the form of data change characteristics by learning the characteristics of the change in the health state of the device reflected in the sample data. Therefore, in this embodiment, for the device under test, the operation data of the device under test can be imported into the pre-trained reconstruction model, so as to output the reconstruction data corresponding to the operation data.

From the above, it can be seen that the reconstructed data can reflect the health status of the device under test. The situation related to the health status can be reflected as the difference between the reconstructed data and the running data. That is to say, the reconstructed data can be understood as the characteristic data conforming to the normal health status, therefore, the difference between the running data and the reconstructed data can reflect the health status of the device under test.

In this embodiment, the health status index is obtained according to the reconstructed data and the running data. The health status index is a series of data in time series, that is, includes the health status index corresponding to each time point. Based on the analysis of the health status index corresponding to each time point, the degradation point in the time point can be determined.

In this embodiment, the degradation point represents the time point when the health state of the device under test begins to degrade. That is to say, it can be understood that at each time point before the degradation point, the operating data of the device under test is data belonging to a normal health state, and at each time point after the degradation point, the operation of the device under test begins to decline The phenomenon. However, decline does not mean failure. It will generally take a period of time from the operation of the equipment under test to decline to failure. The operating data after the time point of decline can provide an effective data prediction basis for the prediction of the failure point.

In this embodiment, a prediction model may be pre-trained, and the prediction model may be pre-trained based on sample data. The sample data may be related data after the degradation point of the device as a sample. Therefore, the prediction model can learn the relevant features of the data after the degradation point, so as to accurately predict the failure point based on the learned relevant features.

Therefore, in this embodiment, for the device under test, after determining the degradation point during operation of the device under test, the health status index corresponding to the time point after the degradation point can be imported into the prediction model for prediction. The prediction model can fit the health status data. On the basis of the fitted curve, it can be further extended based on the fitted curve to obtain the extended curve. And the extension curve also includes the predicted health status index at multiple time points.

A preset threshold may be set, and based on the preset threshold, it may be judged whether the predicted health status index indicates that the device under test is in a failure state. For example, on the prediction curve, if the predicted health status index reaches the same as the preset threshold, the corresponding time point can be determined as the failure time point.

For example, please refer to Figure 2. For example, the time point of 20-06-06 is the determined degradation point, and the data collected from the degradation point during the period from 20-06-06 to 21-11-28 is after the degradation point Health status index at each time point. The prediction model can fit the health status index in this time period, and then obtain the fitting curve in the time period from 20-06-06 to 21-11-28 as shown in the figure.

Based on the curve trend of the fitted curve, an extension can be performed to obtain an extended curve, for example, the extended curve in the time period from 21-11-28 to 23-05-22 in the figure. Wherein, the constructed extension curve may also be a plurality of extension curves within a certain error range.

The value indicated by the horizontal dotted line in the figure can be the preset threshold value, and at the time point when the extension curve intersects the preset threshold value, that is, the time point when the predicted health status index is consistent with the preset threshold value, it can be determined as failure point in time. That is, the device under test is predicted to fail at that point in time.

The predictive maintenance method provided in this embodiment, using the pre-trained reconstruction model and prediction model, can accurately realize the determination of the degradation point and the prediction of the data by learning the characteristics of the operating data, and can be applied to the situation based on a small amount of data predictive maintenance.

In this embodiment, the obtained operating data is pure data such as current and torque of the device under test, and it is difficult to reflect the characteristics of the data in time series. In order to facilitate model learning or obtain the characteristics of time series data, before importing the operating data into the reconstruction model, some processing can be performed on the operating data. Please refer to Figure 3. In this embodiment, when processing the running data based on the reconstruction model, it can be implemented in the following ways:

S1011, for the obtained operation data of the equipment under test at multiple consecutive time points, intercept the operation data according to the preset step size and the preset window length, and obtain the operation data in multiple time windows.

S1012. For the operating data in each time window, scale the operating data to a preset range.

S1013, extracting time-domain feature vectors, frequency-domain feature vectors, and time-frequency-domain feature vectors of the scaled running data.

S1014. Import the time-domain feature vector, frequency-domain feature vector, and time-frequency-domain feature vector into a pre-trained reconstruction model to obtain reconstruction data corresponding to the running data.

In this embodiment, formal definition processing can be performed on the obtained operating data first, and for each time point t, the data of various types of operating data at the time point t can be expressed as the following form:

Among them, m represents the number of components to be tested in the device under test, and the health status at this time point t can be represented by h _k , _which can indicate that the health status of the device under test at this time point is normal or faulty. For example, when h _k is 1, it means that the device under test is in a normal state, and when h _k is 0, it means that the device under test is in a fault state. This is only for illustration, and this embodiment is not limited thereto.

In this way, the operation data of the device under test at multiple time points continuous in time series can be obtained. On this basis, in order to facilitate the processing of data by the model, the operating data can be divided into data segments within multiple time windows for input into the model. In this embodiment, the operation data can be intercepted according to the preset step size and the preset window length, so as to obtain the operation data in multiple time windows.

In order to ensure that the intercepted operating data is generally continuous, in this embodiment, the preset window length may be greater than the preset step size. In this way, in two adjacent time windows, the data in the last part of the previous time window overlaps with the data in the front part of the next time window, which can effectively ensure that the running data in different time windows is continuously saved. 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 various operating data, the operating data can also be standardized, and the operating data in each time window can be scaled to a preset range.

In a possible implementation manner, the running data may be standardized 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 running data falls within the interval with a mean of 0 and a standard deviation of 1.

In this embodiment, the running data can be scaled according to the following scaling formula.

Among them, x represents the running data before scaling, z represents the running data after scaling, N represents the total number of running data, μ represents the average value of the running data before scaling, and σ represents the standard deviation of the running data before scaling.

In this embodiment, z-score normalization may be performed on various types of operating data in the above manner. The distribution of the scaled operating data itself has not changed, but the data distribution interval can remain basically the same after scaling, and the data can be mainly distributed in the [-2,2] interval. By scaling the running data, subsequent models can focus more on analyzing the distribution of the data itself.

In order to further enable the model to analyze and obtain the distribution of operating data from multiple dimensions, in this embodiment, features on multiple dimensions of the scaled operating data can be extracted, including time-domain feature vectors, frequency-domain feature vectors, and time-domain feature vectors. Frequency-domain feature vectors.

In this embodiment, for time-domain feature extraction, feature analysis can be directly performed on the operating data of each time window. Mainly from the perspective of time domain, multiple time domain features can be extracted, including effective value, square root amplitude, peak-to-peak value, crest factor, margin index, skewness index, kurtosis index, form factor, pulse factor, information entropy, correlation sex coefficient. The multiple time-domain features are concatenated into a vector, and the following time-domain feature vector is obtained:

In addition, frequency domain features can also be extracted. It can be seen from Basselval's theorem that whether it is a real signal or a complex signal, the integral of the square of the signal amplitude is equal to the energy of the signal, and equal to the square of the modulus of the signal spectral density. It can be expressed in formula as follows:

Among them, E represents the signal energy, x(t) represents the time domain value of the signal, and X(f) represents the frequency domain value of the signal.

Therefore, the energy of the high-frequency discrete signal can be obtained by accumulating the squares of each value in the running data, as follows:

Wherein, xf _e represents a characteristic component as a whole, and e represents an energy signal, which corresponds to rms and sra in x _rms and x _sra mentioned above. f(i) represents the i-th value of the time-domain signal such as x _rms . The above formula can be understood as the frequency-domain feature component on the left side of the equal sign is equal to the sum of the modular squares of the corresponding time-domain feature component.

In this way, the frequency domain feature vector can be obtained as follows:

F _f ＝{xf _rms ,xf _sra ,xf _ppv ,xf _cf ,xf _mf ,

xf _skf ,xf _kf ,xf _shf ,xf _if ,xf _e }

On the basis of the above, time-frequency domain feature analysis can be carried out. In this embodiment, the time-frequency analysis can be performed by using the EDM method and the short-time Fourier transform method. Firstly, n intrinsic mode functions (IMFs) related to the faults of the equipment under test in each time window are obtained by EDM. Then through EDM, four types of eigenvalues are taken from the n IMFs that are screened: energy xtf _e , variance xtf _sd , skewness index xtf _sf and kurtosis index xtf _kf , and STFT is used to obtain the standard deviation of instantaneous frequency σ _std and the value of instantaneous frequency The signal-to-noise ratio SNR2 eigenvalues are concatenated to obtain the 4n+2-dimensional time-frequency domain characteristics of the signal:

Wherein, the above-mentioned xtf as a whole represents a component.

Import the time-domain feature vector, frequency-domain feature vector, and time-frequency-domain feature vector obtained above into the reconstruction model to obtain the reconstruction data corresponding to the running data.

In this embodiment, the reconstruction model is obtained by pre-training based on sample data. The predictive maintenance method provided in this embodiment also includes the step of obtaining the reconstruction model based on the training of the constructed neural network model in advance, wherein the neural network model Can be an LSTM model. This neural network model includes an encoder and a decoder, as shown in Figure 5. Please refer to Figure 6, the step of pre-training to obtain the reconstructed model can be implemented in the following ways:

S201. Collect sample data, where the sample data includes data corresponding to multiple consecutive time points.

S202. Import the sample data into the encoder for encoding processing to obtain feature data.

S203. Import the feature data and sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data.

S204. Based on the loss function constructed according to the sample data and the sample reconstruction data, the model parameters of the encoder and the decoder are adjusted, and then the training is continued until the preset requirements are met, and the reconstruction model is obtained.

In this embodiment, the sample data may be operation data corresponding to consecutive time points during the operation of the industrial equipment. Similarly, the sample data can be processed according to the above-mentioned preprocessing, scaling processing, time-domain feature extraction, frequency-domain feature extraction, and time-frequency domain feature extraction.

Import the above-mentioned processed sample data into the encoder of the neural network model for encoding processing to obtain feature data. In this embodiment, the structures of the encoder and the decoder are respectively an LSTM unit. LSTM can take a period of time sequence data as input, and then update its hidden state until the last step of the time series, denoted as t2, the cell state generated by LSTM contains all the information of the previous sequence, that is

The cell state can also be called context vectors (Context Vectors), and the decoder reconstructs the input of the encoder through 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 hidden state of the decoder can be described as

The feature data obtained by the encoder is combined with the sample data, including the time domain feature, frequency domain feature and time-frequency domain feature of the sample data, and is fused and decoded in the decoder to obtain the sample reconstruction data.

The loss function can be constructed based on sample data and sample reconstruction data, and the constructed loss function can be as follows:

Among them, _xi represents the sample data,

Represents the sample reconstruction data, t ₁ and t ₂ represent the start time point and end time point of the time series data, respectively.

The training of the encoder and the decoder can have multiple iterations, and the function value of the above loss function can be calculated after each iteration, and the model parameters of the encoder and the decoder can be adjusted to continue the training. When the number of iterations reaches the set maximum number, or the loss function reaches convergence and no longer decreases, or the iteration time reaches the set maximum time, it can be judged to meet the preset requirements, so that the neural network model obtained at this time can be obtained reconstruction model.

The above is the process of obtaining the reconstruction model through pre-training. When using the reconstruction model to reconstruct the operating data of the equipment to be tested and determine the degradation point, first use the reconstruction model to obtain the reconstruction data corresponding to the operating data, and then based on the reconstruction The data is obtained with a state of health index, and based on the state of health index, the degradation point in the time point is determined. Please refer to Figure 7, in this embodiment, the step of determining the degradation point can be implemented in the following ways:

S1021. Obtain difference data between the reconstruction data and the running data, and use the difference data as a health status index.

S1022. Compare the health status index with a health status threshold, and determine the time point corresponding to which the health status index starts to deviate from the health status threshold as a degradation point.

In this embodiment, the health status index may be the difference between the actual operating data and the reconstructed (regarded as normal) data by the reconstruction model. Therefore, an increase in the reconstruction error means that the operating state deviates from the normal state.

In this embodiment, a health state threshold may be preset as a criterion for judging whether the health state of the device under test is abnormal. The health status threshold may be set in advance based on relevant data in the process of constructing the reconstruction model based on sample data. In this embodiment, the health status threshold can be constructed in the following ways:

Calculate the difference between the sample data and the sample reconstruction data, calculate the difference average and the difference standard deviation based on the difference, and obtain the health status threshold according to the difference difference and the difference standard deviation.

In this embodiment, the specific calculation formula of the health status threshold may be as follows:

Among them, mean means to take the average value, std means to take the standard deviation, and ‖‖ ₂ means to calculate the L2 norm.

When the health status index deviates from the health status threshold, that is, when the difference between the running data and the reconstructed data exceeds the health status threshold, the corresponding time point can be determined as a degradation point.

Considering that in the actual processing process, there may be some mutation points in the running data, resulting in some mutation points in the obtained health status index. If the health status index corresponding to the mutation point may deviate from the health status threshold at the corresponding time point due to a sudden change in data characteristics, it is misjudged as a degradation point. Therefore, referring to FIG. 8, in this embodiment, in the above step of determining the degradation point, it can be implemented in the following manner:

S10221. Obtain the time point when the health status index starts to deviate from the health status threshold.

S10222. Detect whether the health status indices corresponding to the set number of time points after the time point deviate from the health status threshold. If they all deviate, execute the following step S10223. If not, execute 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 corresponding health status index starts to deviate from the health status threshold from a certain time point, it can be re-determined that there is no limit to 5 time points, 10 time points, etc. after the time point. The health status index corresponding to each subsequent time point can be obtained, and then detect whether each health status index deviates from the health status threshold. If each health status index deviates from the health status threshold, it indicates that there is a long period of time. The data of the continuous deviation from the health status threshold is not due to the accidental deviation caused by the mutation of the data. Therefore, in this situation, it can be determined that the above-mentioned time point when it starts to deviate from the health state threshold is the degradation point.

However, if the health status index at the set number of time points after the departure from the health status threshold does not all deviate from the health status threshold, it indicates that the health status index at the above time point may only be due to data mutations. resulting deviation. Therefore, in this case, it can be determined that the above time point is not a degradation point.

In a possible implementation manner, Raida's rule can be used to eliminate abnormal points in a series of health status indices in time series, that is, to eliminate health status indices with data mutations. In this way, the real degradation point where the equipment under test begins to degrade can be found, so that the subsequent failure point can be predicted based on the health status index after the real degradation point.

Please refer to Figure 9. In this embodiment, when performing data fitting prediction based on the health status index corresponding to the time point after the degradation point, it can be implemented in the following manner:

S1031. Obtain a health status index corresponding to a time point after the degradation point of the device under test.

S1032. Divide the health status index into time windows according to time series to obtain health status indices in multiple time windows.

S1033. Normalize the health status index in each time window.

S1034. Extract data features of the health status index after normalization processing, and import the data features into a pre-trained prediction model to fit the health status index.

In this embodiment, for the health status index at each time point after the degradation point of the device under test, the health status index may be intercepted according to a certain window length and a certain interception step. Wherein, similarly, in order to ensure the consistency of the intercepted health status index of each time window, the window length may be greater than the interception step size.

For the intercepted health status index in each time window, the health status index can be normalized to a certain unified numerical range according to the above-mentioned scaling processing method for the operation data. So that the prediction model can focus on the distribution characteristics of the data itself.

The data features of the normalized health status index can be extracted, and the data features can include, for example, time domain features, frequency domain features, and time-frequency domain features. Import the data characteristics of the health status index into the pre-trained prediction model, and the prediction model can fit the health status index to obtain a fitting curve. Then, based on the extension curve of the fitting curve, the failure time point is determined.

In this embodiment, the predictive model may be obtained by training a constructed neural network model based on sample data in advance. For example, the health status index corresponding to the time point after the degradation point of industrial equipment is used as sample data, and the neural network model can be a GRU (Gate Recurrent Unit) network model.

On this basis, the predicted health status index obtained each time is compared with a preset threshold, which can be set based on known operating conditions of the same type of industrial equipment running the same process as the equipment under test. When the predicted health status index is consistent with the preset threshold, the corresponding time point can be considered as the failure time point.

Based on the failure time point, the remaining service life of the equipment under test can be determined. For example, the time point when the health state index is fitted by the prediction model is used as the node, and the time period from this node to the predicted failure time point , which is the remaining service life of the device under test.

The overall flow of the predictive maintenance method provided by this embodiment is introduced below.

In this embodiment, sample data may be collected in advance, and the sample data may be data corresponding to multiple consecutive time points. The sample data can be the real-time current, torque, and shaft angular position values during the operation of the industrial equipment, as well as the body parameters of the industrial equipment, etc.

Data preprocessing can be performed on the sample data, such as using a certain step size and intercepting the sample data according to a certain window size to obtain sample data in multiple windows. In addition, data standardization can be performed, for example, the sample data in each window is scaled to a preset range, such as a certain mean value and a certain standard deviation range.

Then perform feature extraction on the scaled data, including time-domain feature extraction, frequency-domain feature extraction and time-frequency domain feature extraction.

Import the time-domain features, frequency-domain features, and time-frequency-domain features obtained based on the sample data into the constructed neural network model to train the neural network model to obtain a reconstructed model. During the training process, the loss function constructed based on the difference information between the input and output can be used as a training guide, and the training is stopped when the iteration meets certain requirements.

During the process of training the reconstruction model, the health status threshold can also be constructed based on the difference between the sample data input into the reconstruction model and the sample reconstruction data output by the reconstruction model. This state of health threshold can then be used to determine the degradation point of the industrial equipment.

On this basis, based on the sample reconstruction data and sample data obtained by the reconstruction model, the health status index can be obtained, and then the time point when the health status index first begins to degenerate is found as the degradation point.

Based on the health status index after the degradation point, the deep model of the GRU network can be trained to obtain a prediction model.

On this basis, in the actual application stage, the operating data of the device under test can be obtained for the device under test. Perform the above-mentioned data preprocessing, time window extraction processing, scaling processing, and time domain feature processing, frequency domain feature processing, and time-frequency domain feature processing on the running data.

Furthermore, the above-mentioned time-domain features, frequency-domain features, and time-frequency-domain features are imported into the reconstruction model to obtain corresponding reconstruction data. The health status index of the device under test is obtained by combining the reconstructed data and the sample data of the device under test.

When performing predictive maintenance on the equipment under test, the operating data of the equipment under test can be imported into the reconstruction model to obtain the corresponding reconstruction data. The health status index can be obtained according to the reconstructed data and the running data. By analyzing and processing the health state index, the degradation point that can represent the health state of the equipment under test begins to degenerate is obtained.

Import the health status index after the degradation point into the prediction model to fit the health status index to obtain a fitting curve. The extension curve is obtained based on the fitting curve, and each time point on the extension curve has a corresponding predicted health status index.

Each predicted health status index is compared with a preset threshold, and when the predicted health status index is the same as the preset threshold, the corresponding time point is determined as the predicted failure time point. The connection point between the fitting curve and the extension curve is obtained, that is, the difference between the time point predicted by the prediction model and the predicted failure time point is the predicted remaining service life of the equipment under test.

The predictive maintenance method provided in this embodiment adopts the health status index as the indicator for the predictive maintenance of industrial equipment based on the health status index, which reduces the dependence of predictive maintenance on various sensors and reduces the practicality of predictive maintenance technology. application cost.

In addition, the reconstruction model including the encoder and the decoder is used to output the reconstructed data and then construct the health status index, the value of the health status index can be extracted from the time series signal, which reduces the cost of building the model, and then on the basis of accurately determining the degradation point It can provide a data basis for the accurate prediction of subsequent failure points.

When predicting the failure point, GRU deep learning is used to extract the features of the time series basis, to predict the health status index and calculate the remaining service life, so as to improve the accuracy of health status monitoring.

Please refer to FIG. 10 , which is a schematic diagram of exemplary components of an electronic device provided by an embodiment of the present application. The electronic device may include a storage medium 110 , a processor 120 , a device for predictive maintenance of industrial equipment based on health status index 130 and a communication interface 140 . In this embodiment, both the storage medium 110 and the processor 120 are located in the electronic device and are set separately. However, it should be understood that the storage medium 110 may also be independent from the electronic device, and may be accessed by the processor 120 through the bus interface. Alternatively, the storage medium 110 may also be integrated into the processor 120, for example, may be a cache and/or a general-purpose register.

The industrial equipment predictive maintenance device 130 based on the health status index can be understood as the above-mentioned electronic equipment, or the processor 120 of the electronic equipment, and can also be understood as realizing the above-mentioned electronic equipment under the control of the electronic equipment independently of the above-mentioned electronic equipment or the processor 120. Software function modules for predictive maintenance methods.

As shown in FIG. 11 , the aforementioned health state index-based predictive maintenance device 130 for industrial equipment may include an acquisition module 131 , a determination module 132 and a prediction module 133 . The functions of each functional module of the health state index-based predictive maintenance device 130 for industrial equipment will be described in detail below.

The obtaining module 131 may be configured to obtain operating data at various time points during the operation of the device under test, import the operating data into a pre-trained reconstruction model, and obtain reconstruction data corresponding to the operating data.

It can be understood that the acquisition module 131 can be used to execute the above step S101, and for the detailed implementation of the acquisition module 131, please refer to the content related to the above step S101.

The determining module 132 may be configured to obtain a health status index according to the reconstruction data and the running data, and determine a degradation point in a time point according to the health status index, and the degradation point represents the health of the device under test The point at which state degradation begins.

It can be understood that the determination module 132 can be used to execute the above step S102, and for the detailed implementation manner of the determination module 132, reference can be made to the content related to the above step S102.

The prediction module 133 can be configured to import the health status index corresponding to the time point after the degradation point into the prediction model obtained by pre-training to fit the health status index, and obtain an extension curve based on the fitting curve Comparing the predicted health status index at each time point in the extension curve with a preset threshold, and determining the time point at which the predicted health status index is the same as the preset threshold as the failure time point.

It can be understood that the prediction module 133 can be used to execute the above step S103, and for the detailed implementation of the prediction module 133, please refer to the content related to the above step S103.

In a possible implementation, the acquisition module 131 may be configured to:

In a possible implementation manner, the preset window length is greater than the preset step size.

In a possible implementation manner, the determination module 132 may be configured to:

In a possible implementation manner, the device 130 for predictive maintenance of industrial equipment based on the health state index further includes a building block for obtaining the reconstruction model based on the training of the neural network model constructed in advance, and the neural network model includes encoder and decoder, this building block can be configured for:

In a possible implementation manner, the health state index-based industrial equipment predictive maintenance device 130 also includes an obtaining module for obtaining the health state threshold, and the obtaining module can be configured to:

In a possible implementation manner, the prediction module 133 may be configured to:

Normalize the health status index in each time window;

For the description of the processing flow of each module in the device and the interaction flow between the modules, reference may be made to the relevant description in the above method embodiment, and details will not be described here.

Further, the embodiment of the present application also provides a computer-readable storage medium, which stores machine-executable instructions, and implements the predictive maintenance method provided by the above-mentioned embodiments when the machine-executable instructions are executed.

Specifically, the computer-readable storage medium can be a general-purpose storage medium, such as a removable disk, a hard disk, etc. When the computer program on the computer-readable storage medium is run, the above predictive maintenance method can be executed. As for the processes involved when the executable instructions in the computer-readable storage medium are executed, reference may be made to relevant descriptions in the foregoing method embodiments, and no further details are given here.

To sum up, the health state index-based predictive maintenance method, device and electronic equipment for industrial equipment provided by the embodiment of the present application acquire the operating data at various time points during the operation of the equipment under test, and import the operating data into pre-training to obtain According to the reconstruction model, the reconstruction data corresponding to the operation data is obtained, the health status index is obtained according to the reconstruction data and the operation data, and the degradation point is determined according to the health status index. Then import the health status index corresponding to the time point after the degradation point into the pre-trained prediction model to fit and obtain the extension curve, compare the predicted health status index at each time point on the extension curve with the preset threshold, and compare the two The consistent time point is determined as the failure time point. In this solution, the pre-trained reconstruction model and prediction model can be used to accurately determine the degradation point and predict the data by learning the characteristics of the operating data, which is suitable for predictive maintenance based on a small amount of data.

The above is only a specific embodiment of the application, but the scope of protection of the application is not limited thereto. Any person familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the application. All should be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be determined by the protection scope of the claims.

Industrial Applicability

This application provides a method, device and electronic equipment for predictive maintenance of industrial equipment based on the health state index. By obtaining the operation data at various time points during the operation of the equipment to be tested, the operation data is imported into the pre-trained reconstruction model, The reconstruction data corresponding to the operation data is obtained, the health status index is obtained according to the reconstruction data and the operation data, and the degradation point is determined according to the health status index. Then import the health status index corresponding to the time point after the degradation point into the pre-trained prediction model to fit and obtain the extension curve, compare the predicted health status data at each time point on the extension curve with the preset threshold, and compare the two The consistent time point is determined as the failure time point. In this solution, the pre-trained reconstruction model and prediction model can be used to accurately determine the degradation point and predict the data by learning the characteristics of the operating data, which is suitable for predictive maintenance based on a small amount of data.

In addition, it can be understood that the method, device and electronic equipment for predictive maintenance of industrial equipment based on the health status index of the present application are reproducible and can be used in various industrial applications. For example, the method, device and electronic equipment for predictive maintenance of industrial equipment based on the health status index of the present application can be used in the technical field of industrial equipment management.

Claims

A method for predictive maintenance of industrial equipment based on health status index, characterized in that the method comprises:

Obtaining operation data at various time points during the operation of the device under test, importing the operation data into a pre-trained reconstruction model, and obtaining reconstruction data corresponding to the operation data;

Obtaining a health status index according to the reconstruction data and the operation data, and determining a degradation point in a time point according to the health status index, and the degradation point represents a time point when the health status of the device under test begins to degrade;

Import the health status index corresponding to the time point after the degradation point into the prediction model obtained in advance to fit the health status index, and obtain an extension curve based on the fitting curve, and each of the extension curves The predicted health state index at the time point is compared with the preset threshold, and the time point at which the predicted health state index is the same as the preset threshold is determined as the failure time point.
The method for predictive maintenance of industrial equipment based on the health state index according to claim 1, wherein the operation data is imported into a pre-trained reconstruction model to obtain the reconstruction data corresponding to the operation data steps, including:

For the obtained operating data of the equipment under test at multiple consecutive time points, the operating data is intercepted according to the preset step size and the preset window length, and the operating data in multiple time windows are obtained;

For the operating data in each time window, scaling the operating data to a preset range;

Extracting time-domain feature vectors, frequency-domain feature vectors and time-frequency-domain feature vectors of the scaled operating data;

Importing the time-domain feature vector, frequency-domain feature vector, and time-frequency-domain feature vector into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operating data.
The health state index-based predictive maintenance method for industrial equipment according to claim 2, wherein the preset window length is greater than the preset step size.
The method for predictive maintenance of industrial equipment based on the health status index according to any one of claims 1 to 3, wherein the health status index is obtained according to the reconstruction data and the operation data, and the health status index is obtained according to the health status index. The state index determines the steps of the degradation point in the time point, including:

Obtaining difference data between the reconstruction data and the running data, and using the difference data as a health status index;

The health status index is compared with the health status threshold, and the time point at which the health status index starts to deviate from the health status threshold is determined as a degradation point.
The method for predictive maintenance of industrial equipment based on the health status index according to claim 4, wherein the step of determining the time point when the health status index starts to deviate from the health status threshold as a degradation point includes: :

Obtain the time point when the health status index starts to deviate from the health status threshold;

Detecting whether the health state indices corresponding to the set number of time points after the time point deviate from the health state threshold, and if they deviate, determine that the time point is a degradation point.
The method for predictive maintenance of industrial equipment based on health status index according to claim 4 or 5, characterized in that, the method also includes the step of obtaining the reconstruction model based on the neural network model training in advance, the neural network The network model includes an encoder and a decoder, and this step includes:

collecting sample data, where the sample data includes data corresponding to multiple consecutive time points;

Importing the sample data into the encoder for encoding processing to obtain feature data;

Importing the feature data and sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;

After adjusting the model parameters of the encoder and decoder based on the loss function constructed according to the sample data and the sample reconstruction data, the training is continued until the preset requirements are met, and the reconstruction model is obtained.
The method for predictive maintenance of industrial equipment based on the state of health index according to claim 6, wherein the state of health threshold is obtained in the following manner:

calculating the difference between the sample data and the sample reconstructed data;

Calculate the difference mean value and difference standard deviation based on the difference value;

The health status threshold is obtained according to the difference mean value and the difference standard deviation.
The health state index-based predictive maintenance method for industrial equipment according to any one of claims 1 to 7, wherein the health state index corresponding to the time point after the degradation point is introduced into pre-training to obtain The predictive model is used to carry out the step of fitting health status index to described health status index, comprising:

Obtaining the health status index corresponding to the time point after the degradation point of the device under test;

Dividing the health status index into time windows according to time series to obtain health status indices in multiple time windows;

Normalize the health status index in each time window;

The data features of the health status index after normalization are extracted, and the data features are imported into the pre-trained prediction model to fit the health status index.
A device for predictive maintenance of industrial equipment based on health status index, characterized in that the device includes:

An acquisition module, configured to acquire operating data at various time points during the operation of the device to be tested, import the operating data into a pre-trained reconstruction model, and obtain reconstruction data corresponding to the operating data;

A determining module, configured to obtain a health status index according to the reconstruction data and the operating data, and determine a degradation point in a time point according to the health status index, and the degradation point indicates that the health status of the device under test begins to degrade point in time;

The prediction module is used to import the health status index corresponding to the time point after the degradation point into the prediction model obtained by pre-training to fit the health status index, and obtain an extension curve based on the fitting curve, and the The predicted health status index at each time point in the extension curve is compared with a preset threshold, and the time point at which the predicted health status index is the same as the preset threshold is determined as the failure time point.
The device for predictive maintenance of industrial equipment based on the health state index according to claim 9, wherein the acquisition module is further configured to:

For the obtained operating data of the equipment under test at multiple consecutive time points, the operating data is intercepted according to the preset step size and the preset window length, and the operating data in multiple time windows are obtained;

For the operating data in each time window, scaling the operating data to a preset range;

Extracting time-domain feature vectors, frequency-domain feature vectors and time-frequency-domain feature vectors of the scaled operating data;

Importing the time-domain feature vector, frequency-domain feature vector, and time-frequency-domain feature vector into a pre-trained reconstruction model to obtain reconstruction data corresponding to the operating data.
The health state index-based predictive maintenance device for industrial equipment according to claim 10, wherein the preset window length is greater than the preset step size.
The health state index-based predictive maintenance device for industrial equipment according to any one of claims 9 to 11, wherein the determination module is further configured to:

Obtaining difference data between the reconstruction data and the running data, and using the difference data as a health status index;

The health status index is compared with the health status threshold, and the time point at which the health status index starts to deviate from the health status threshold is determined as a degradation point.
The device for predictive maintenance of industrial equipment based on the health status index according to claim 12, wherein the determination module is further configured to:

Obtain the time point when the health status index starts to deviate from the health status threshold;

Detecting whether the health state indices corresponding to the set number of time points after the time point deviate from the health state threshold, and if they deviate, determine that the time point is a degradation point.
The device for predictive maintenance of industrial equipment based on the health state index according to claim 12 or 13, characterized in that, the device for predictive maintenance of industrial equipment based on the state of health index also includes a neural network model for pre-built training The building blocks of the reconstruction model are obtained, the neural network model includes an encoder and a decoder, and the building blocks are also configured to:

collecting sample data, where the sample data includes data corresponding to multiple consecutive time points;

Importing the sample data into the encoder for encoding processing to obtain feature data;

Importing the feature data and sample data into the decoder for fusion and decoding processing to obtain sample reconstruction data;

After adjusting the model parameters of the encoder and decoder based on the loss function constructed according to the sample data and the sample reconstruction data, the training is continued until the preset requirements are met, and the reconstruction model is obtained.
The device for predictive maintenance of industrial equipment based on health status index according to claim 14, characterized in that, the device for predictive maintenance of industrial equipment based on health status index further comprises an obtaining module for obtaining the health status threshold, The obtaining module is configured to:

calculating the difference between the sample data and the sample reconstructed data;

Calculate the difference mean value and difference standard deviation based on the difference value;

The health status threshold is obtained according to the difference mean value and the difference standard deviation.
The health state index-based predictive maintenance device for industrial equipment according to any one of claims 9 to 15, wherein the prediction module is configured to:

Obtaining the health status index corresponding to the time point after the degradation point of the device under test;

Dividing the health status index into time windows according to time series to obtain health status indices in multiple time windows;

Normalize the health status index in each time window;

The data features of the health status index after normalization are extracted, and the data features are imported into the pre-trained prediction model to fit the health status index.
An electronic device, characterized in that it includes one or more storage media and one or more processors communicating with the storage media, one or more storage media stores machine-executable instructions executable by the processor, when the electronic device In operation, the processor executes the machine-executable instructions to perform the method steps of any one of claims 1-8.