CN116561927A - Digital twin-driven small sample rotary machine residual life prediction method and system - Google Patents
Digital twin-driven small sample rotary machine residual life prediction method and system Download PDFInfo
- Publication number
- CN116561927A CN116561927A CN202310619377.4A CN202310619377A CN116561927A CN 116561927 A CN116561927 A CN 116561927A CN 202310619377 A CN202310619377 A CN 202310619377A CN 116561927 A CN116561927 A CN 116561927A
- Authority
- CN
- China
- Prior art keywords
- rotary machine
- signal
- small sample
- predicting
- encoder
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000006731 degradation reaction Methods 0.000 claims abstract description 67
- 230000015556 catabolic process Effects 0.000 claims abstract description 62
- 230000036541 health Effects 0.000 claims abstract description 60
- 238000013507 mapping Methods 0.000 claims abstract description 53
- 238000012360 testing method Methods 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 230000008569 process Effects 0.000 claims abstract description 16
- 230000006870 function Effects 0.000 claims description 87
- 238000001914 filtration Methods 0.000 claims description 17
- 238000007781 pre-processing Methods 0.000 claims description 16
- 238000011478 gradient descent method Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000005259 measurement Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000011156 evaluation Methods 0.000 claims description 8
- 230000009467 reduction Effects 0.000 claims description 7
- 230000007704 transition Effects 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 4
- 239000012634 fragment Substances 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 description 15
- 238000003860 storage Methods 0.000 description 13
- 238000012545 processing Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 238000009826 distribution Methods 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000000670 limiting effect Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002542 deteriorative effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Optimization (AREA)
- Evolutionary Computation (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method and a system for predicting the residual life of a small sample rotary machine driven by digital twinning, which use a rotary machine early degradation signal training convolution self-encoder to learn an early degradation signal mode; then reconstructing the test signal by using a convolution self-encoder, calculating a reconstruction error, mapping the reconstruction error to a [0,1] interval as a health factor, fitting a Weibull reliability function according to the health factor, and predicting the residual service life of the rotary machine; and repeating the steps based on the real-time data in the continuous running process of the rotary machine to realize the real-time updating of the residual life of the rotary machine. The prediction algorithm for the residual life of the rotary machine constructed by the invention can predict the residual life of the bearing on the premise of not needing an end degradation signal, can adaptively update a model along with the increase of data quantity so as to improve the prediction precision in real time, and has the feasibility of predicting the life of the rotary machine with a small sample in an actual industrial scene.
Description
Technical Field
The invention belongs to the technical field of life prediction of rotary machinery, and particularly relates to a method and a system for predicting residual life of a digital twin-driven small-sample rotary machinery.
Background
The method for predicting the residual service life of the rotary machine mainly comprises two steps: a prediction method based on a mechanism model and a data-driven prediction method. The former analyzes the degradation process of the rotating machinery by using failure mechanism, probability statistics and other methods, builds a degradation mechanism model by using theoretical analysis, experimental verification and other methods, and predicts the residual life. The data-driven residual life prediction is to construct an end-to-end model between the sensor signal and the residual service life of the rotary machine based on various prediction algorithms, and the residual life of the rotary machine can be predicted without expert experience and domain knowledge under the support of big data.
The method only can predict the static average life of the rotary machine at a specific moment in the design stage or the use process, the dynamic degradation process of the rotary machine can not be described, and the real-time online prediction of the residual life is difficult to realize. In addition, when a traditional mechanism model or a data driving model is used, a large amount of degradation data is needed for model construction, when the data for model construction is insufficient, the problem of model prediction accuracy is solved, however, due to the limitation of data acquisition capacity and the requirement of safe production, a large amount of complete degradation data is often difficult to collect in the actual process for constructing a prediction model.
Disclosure of Invention
The invention aims to solve the technical problems in the prior art, and provides a method and a system for predicting the residual life of a digital twin-driven small-sample rotary machine, which are used for solving the technical problem that the residual life of the rotary machine is difficult to accurately predict under small-sample data by constructing a data-driven rotary machine degradation behavior model through fusion convolution self-encoder and Weibull distribution, and realizing the accurate real-time prediction of the residual life of the rotary machine under the condition of small samples.
The invention adopts the following technical scheme:
a method for predicting the residual life of a digital twin-driven small sample rotary machine comprises the following steps:
s1, preprocessing signals of rotary mechanical equipment;
s2, building a training convolution self-encoder, and evaluating the unhealthy degree of the signals obtained in the step S1;
s3, establishing an index mapping function, and mapping the unhealthy degree obtained in the step S2 into a health factor capable of directly reflecting the health state of the equipment;
s4, carrying out real-time iterative prediction on the residual service life of the rotary machine based on the Weibull reliability function and the health factor obtained in the step S3 by combining a gradient descent method.
Specifically, step S1 specifically includes:
S101, for a section of measured equipment signal z (t), a one-dimensional Kalman filtering algorithm is used for establishing a state equation of the equipment signal, and an optimal estimated value is obtained by superposing a measured value of an estimated signal with error and an observed value with error;
s102, dividing the filtered signal x (t) obtained in the step S101 into a plurality of signal segments with the length of N, recognizing the signal within the first 30% of time as an early degradation signal of the rotary machine, and using the early degradation signal for training a subsequent algorithm model, wherein the rest is test data.
Further, in step S101, z (t) is set as the observed value z k And (3) with the initial estimated value at the moment k=0, giving an initial value of a state transition matrix A, a process excitation noise covariance Q, a measurement noise covariance R and a conversion matrix H, and carrying out iterative calculation to obtain a signal x (t) of z (t) after noise reduction through Kalman filtering.
Further, in step S102, the division length N of the device signal is:
where k is a given scaling factor, f rotate For the rotation frequency of the rotary machine, f collect Is the acquisition frequency of the signal.
Specifically, step S2 specifically includes:
s201, constructing a convolutional self-encoder neural network model;
s202, the early degradation signal obtained after the pretreatment in the step S1 is x early (t) obtaining a reconstructed signal after encoding and decoding by the input convolution self-encoderCalculating a reconstruction error using a squared difference loss function;
s203, inputting the test signal into a trained convolution self-encoder for signal reconstruction, calculating a reconstruction error of the test signal and the reconstruction signal, and taking the reconstruction error as the unhealthy degree of the equipment corresponding to the test signal.
Specifically, the step S3 specifically includes:
s301, constructing an index mapping function based on the definition of unhealthy degree and health factor HI and determining an index mapping function super-parameter;
s302, obtaining M test signal fragments x= { x after pretreatment 1N ,x 2N ,...,x MN Sequentially inputting the convolutional self-encoder and the exponential mapping function after the sequence of time to obtain a health factor HI set HI= { HI of the rotating machine 1N ,HI 2N ,...,HI MN }。
Further, in step S301, the index mapping function data specifically includes:
wherein k and b are the shape factor and bias constant of the exponential mapping function respectively,and reconstructing errors.
Specifically, step S4 specifically includes:
s401, initializing a Weibull reliability function;
s402, calculating the residual service life of the rotary machine based on a Weibull reliability function;
and S403, updating parameters of the Weibull reliability function by using a gradient descent method according to the HI set obtained from the real-time measurement data in the step S3, and repeating the step S402 and the step S403 to realize real-time iterative prediction of the residual service life of the rotary machine under the condition of a small sample.
Specifically, in step S402, the remaining service life RUL of the rotating machine at time t t The method comprises the following steps:
RUL t ＝T f -T t ＝η[(-ln R f ) 1/β -(-ln R t ) 1/β ]
wherein T is f For the theoretical failure working time of the rotary machine, T t For the theoretical working time of the rotary machine, eta is a scale parameter, R f For failure threshold reliability, R t For reliability, β is a shape parameter.
In a second aspect, embodiments of the present invention provide a digital twinned small sample rotary machine residual life prediction system, comprising:
the preprocessing module is used for preprocessing signals of the rotating mechanical equipment;
the evaluation module is used for establishing a training convolution self-encoder and evaluating the unhealthy degree of the signals obtained by the preprocessing module;
the mapping module is used for establishing an index mapping function and mapping the unhealthy degree obtained by the evaluation module into a health factor capable of directly reflecting the health state of the equipment;
and the prediction module is used for carrying out real-time iterative prediction on the residual service life of the rotary machine by combining a gradient descent method based on the Weibull reliability function and the health factor obtained by the mapping module.
Compared with the prior art, the invention has at least the following beneficial effects:
a digital twin-driven small sample rotary machine residual life prediction method is provided, and the problems that a rotary machine is long in degradation period, high in data density and high in acquisition cost, a large number of full life cycle degradation signals are difficult to acquire in practice, a mechanism prediction model or a data driving prediction model is trained to accurately predict the residual service life of the rotary machine in a small sample scene are solved, so that a Weibull reliability function and a convolution self-encoder combined rotary machine residual life prediction method is established. The convolutional self-encoder (Convolutional Autoencoder, CAE) need only be trained during the training phase using the rotating machinery early degradation signal, evaluating the degree of real-time degradation of the device and constructing Health Index (HI) by calculating the reconstruction error of the CAE for the rotating machinery real-time signal. And introducing a Weibull reliability function as a degradation model to describe the degradation process of the rotary machine, updating the Weibull reliability function parameters in real time according to the health factors, and predicting the residual service life based on the Weibull reliability function. The method for predicting the residual life of the rotary machine, which is constructed by the invention, can well solve the dynamic description of the degradation process of the rotary machine and the real-time prediction of the residual life of the rotary machine under the condition of a small sample, and provides a new thought for the application of digital twin technology in the field of monitoring operation and maintenance of the rotary machine equipment.
Furthermore, the original measurement signal of the rotary machine is preprocessed based on a one-dimensional Kalman filtering algorithm and signal segment interception, so that noise and abnormal signals in the original measurement signal can be removed, the original signal is converted into the signal segment with the same data format, and unified calculation processing of subsequent steps of the algorithm is facilitated.
Further, the one-dimensional Kalman filtering algorithm utilizes the estimated value of the rotating mechanical signal at the previous moment and the signal measured value at the current moment to acquire the optimal estimation of the rotating mechanical signal at the current moment, so that a large amount of noise in the rotating mechanical measured signal is removed, and errors caused by the noise in the subsequent reconstruction errors and the health factor construction step are avoided. Meanwhile, the one-dimensional Kalman filtering algorithm has the characteristic of small calculation amount, can rapidly filter and reduce noise of original equipment signals, and improves the instantaneity of the prediction of the residual life of the rotary machine.
Further, the conventional signal segment intercepting method is to manually set a length, and a signal segment with an excessively small length cannot reflect the process of rotating a rotary machine for one turn; when the length is too large, the rotating machinery rotates for a plurality of circles within the corresponding time of one signal segment, so that signal redundancy is caused, and the complexity and the calculated amount of a prediction algorithm are increased. In the patent, the rotating frequency of the rotating machine and the signal measuring frequency of the sensor are considered to calculate the signal length corresponding to the rotating circle of the rotating machine, the rationality of the signal segment length setting is guaranteed by combining the scaling coefficient, and the signal segment not only contains data of at least one complete rotating period, but also does not contain redundant data.
Furthermore, the convolutional neural network can effectively extract the signal characteristics of the degradation signals of the rotary machine, learn the data modes of the degradation signals, and quantitatively estimate the unhealthy degree of equipment corresponding to the degradation signals of each period of the rotary machine under the condition that only the degradation signals of the early stage are available by utilizing the characteristic that the reconstruction errors of the signals of different degradation periods of the rotary machine are different by using a convolutional self-encoder trained by using the degradation signals of the early stage.
Furthermore, due to the difference between the working condition of the rotary machine and the type of equipment, the unhealthy degree distribution ranges of different test signals obtained by the evaluation of the convolution self-encoder are different, and the unhealthy degree of the test signals is mapped into the healthy factors on the [0,1] interval by using the exponential mapping function, so that the subsequent parameter updating and service life prediction for the Weibull reliability function are facilitated.
Furthermore, the index mapping function has the characteristics of continuous conduction, monotonic decrease, saturation and the like, and can flexibly set corresponding shape coefficients and bias constants according to the working condition and the equipment type of the rotary machine, so that the interpretability and the calculation efficiency of the health factor mapping are improved.
Furthermore, the Weibull reliability function is used as a degradation behavior model of the rotating machine, so that on one hand, the degradation trend of the rotating machine can be intuitively reflected, and the interpretation of the residual life prediction of the rotating machine is improved; and on the other hand, the residual life is predicted after the Weibull reliability function parameters are updated according to the real-time measurement signals, so that the accuracy and the instantaneity of the residual life prediction process of the rotary machine are ensured.
Furthermore, the reliability of the failure threshold can be flexibly set according to the requirements of the life prediction of the rotary machine and the actual application scene of the rotary machine, so that the residual service life of the rotary machine can be predicted more accurately and efficiently.
It will be appreciated that the advantages of the second aspect may be found in the relevant description of the first aspect, and will not be described in detail herein.
In summary, the present invention trains the convolutional self-encoder using the early degradation signal of the rotary machine, evaluates the corresponding unhealthy degree of the degradation signal of the rotary machine, and maps it to the health factor using an exponential mapping function. By introducing the Weibull reliability function to describe the degradation trend of the rotary machine, the parameter updating and the residual life prediction of the Weibull reliability function are carried out based on real-time data, so that the residual life prediction of the rotary machine is carried out more accurately and in real time.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
FIG. 1 is a unitary frame diagram of the present invention;
FIG. 2 is an overall flow of residual life prediction according to the present invention;
FIG. 3 is a graph comparing the bearing vibration raw signal with the Kalman filtered vibration signal;
FIG. 4 is a graph comparing bearing vibration signals with reconstructed signals from convolutional encoders;
FIG. 5 is a graph of error in reconstructing vibration signals of a training set bearing sample;
FIG. 6 is a graph of training set bearing sample health factor HI and Weibull reliability function;
FIG. 7 is a graph of real-time iterative prediction results of residual life of test set bearing samples 1_3;
fig. 8 is a graph of the residual life prediction results of test set bearing samples 1_3.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it will be understood that the terms "comprises" and "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In the present invention, the character "/" generally indicates that the front and rear related objects are an or relationship.
It should be understood that although the terms first, second, third, etc. may be used to describe the preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish one preset range from another. For example, a first preset range may also be referred to as a second preset range, and similarly, a second preset range may also be referred to as a first preset range without departing from the scope of embodiments of the present invention.
Depending on the context, the word "if" as used herein may be interpreted as "at … …" or "at … …" or "in response to a determination" or "in response to detection". Similarly, the phrase "if determined" or "if detected (stated condition or event)" may be interpreted as "when determined" or "in response to determination" or "when detected (stated condition or event)" or "in response to detection (stated condition or event), depending on the context.
Various structural schematic diagrams according to the disclosed embodiments of the present invention are shown in the accompanying drawings. The figures are not drawn to scale, wherein certain details are exaggerated for clarity of presentation and may have been omitted. The shapes of the various regions, layers and their relative sizes, positional relationships shown in the drawings are merely exemplary, may in practice deviate due to manufacturing tolerances or technical limitations, and one skilled in the art may additionally design regions/layers having different shapes, sizes, relative positions as actually required.
The invention provides a method for predicting the residual service life of a digital twin-driven small sample rotary machine, which aims at solving the problem that the residual service life of the rotary machine is difficult to accurately predict under small sample data and constructs a data-driven rotary machine degradation behavior model through fusion convolution self-encoder and Weibull distribution. Firstly, training a convolution self-encoder by using a rotating machinery early degradation signal to learn an early degradation signal mode; then reconstructing the test signal by using a convolution self-encoder, calculating a reconstruction error, mapping the reconstruction error to a [0,1] interval as a health factor, fitting a Weibull reliability function according to the health factor, and predicting the residual service life of the rotary machine; and repeating the steps based on the real-time data in the continuous running process of the rotary machine to realize the real-time updating of the residual life of the rotary machine. The prediction algorithm for the residual life of the rotary machine constructed by the invention can predict the residual life of the bearing on the premise of not needing an end degradation signal, can adaptively update a model along with the increase of data quantity so as to improve the prediction precision in real time, and has the feasibility of predicting the life of the rotary machine with a small sample in an actual industrial scene.
Referring to fig. 2, the method for predicting the residual life of a digital twin-driven small sample rotary machine according to the present invention includes the following steps:
s1, preprocessing signals of rotary mechanical equipment
The original equipment signal of the rotating machine is usually a continuous time domain signal with high frequency and low density, and the noise reduction filtering treatment is needed to be used for subsequent model training and prediction. Firstly, filtering and denoising an original signal by using a one-dimensional Kalman filtering algorithm; and then determining the length of a time window by comparing the rotating speed of the rotating machine with the signal acquisition frequency, and dividing the equipment signal into equal-length signal fragments by using the fixed-length time window.
S101, for a section of measured equipment signal z (t), a state equation of the equipment signal is established by using a one-dimensional Kalman filtering algorithm, and an optimal estimated value obtained by superposing a measured value of a signal band error and an observed value of the band error is estimated, so that filtering and noise reduction of the equipment signal are realized.
The one-dimensional Kalman filter prediction formula used is as follows:
wherein,,the equation is the posterior state estimated value of the k-1 moment signal, is the filtering result of the Kalman filtering algorithm and is also called the optimal estimated value; a is a state transition matrix; />Is a priori estimated state value at time k; / >Is the a priori estimated covariance at time k; p (P) k-1 Is the posterior estimated covariance at time k-1; q is the process excitation noise covariance used to characterize the error between the state transition matrix and the actual process.
The one-dimensional kalman filter update used is known as follows:
wherein K is k The formula is a filter gain matrix at k time; h is a state variable to observation transition matrix; r is the measurement noise covariance; z k Is the signal measurement at time k.
Taking z (t) as the observed value z k And (3) with the initial estimated value at the moment of k=0, the signal x (t) of z (t) after noise reduction through Kalman filtering can be obtained by using the five formulas to carry out iterative calculation by giving initial values of the state transition matrix A, the process excitation noise covariance Q, the measurement noise covariance R and the conversion matrix H.
S102, intercepting and dividing signals of the rotary mechanical equipment;
if the rotation frequency of the rotary machine is f rotate The acquisition frequency of the signal is f collect The division length N of the device signal is:
where k is a scaling factor given for adjusting the division length N in accordance with the actual characteristics of the device signal.
After the division length N is obtained, the filtered signal x (t) obtained in the step S101 is divided into a plurality of signal segments with the length N, and the signal within the first 30% of time is considered as an early degradation signal of the rotary machine, and is used for training a subsequent algorithm model, and the rest is test data.
S2, realizing unhealthy degree assessment of input signals by building training convolution self-encoder
The health state of the rotating machine shows a tendency of gradually deteriorating along with the aggravation of the degradation process, and the device signals reflected on the device signals, namely, the device signals in different degradation stages, have different data modes, so that the unhealthy degree of the device corresponding to the test signal can be reflected by measuring the difference degree between the test signal and the early degradation signal. To achieve this, a convolutional self-encoder neural network model is first constructed; then training a convolution self-encoder by using the early degradation signal to learn an early degradation signal mode, so that high-precision reconstruction of the early degradation signal can be realized; and finally, inputting the test signal into a trained convolution self-encoder to reconstruct the signal, and calculating the reconstruction error of the test signal and the reconstruction signal, wherein the convolution self-encoder only learns the data mode of the early degradation signal, the test signal in other degradation stages cannot be accurately restored, and the higher the degradation degree is, the larger the difference degree between the test signal and the early degradation signal is, namely, the larger the reconstruction error is, so that the reconstruction error can be directly used as the unhealthy degree of equipment corresponding to the test signal.
S201, constructing a convolution self-encoder model;
the self-encoder comprises an encoder and a decoder, wherein the encoder reduces the dimension of an input signal to a feature space, and the decoder encodes and reconstructs and outputs the signal in the feature space. Convolutional self-encoder, i.e. a self-encoder using a convolutional neural network as encoder and decoder, for a convolutional self-encoder with k convolutional kernels, the formula for the encoding and decoding stages is as follows:
h k ＝σ(x*W k +b k ) (7)
wherein h is k Is the feature code corresponding to the kth convolution kernel; w (W) k Is the kth convolution kernel; b k Is the offset constant corresponding to the kth convolution kernel; σ () is an activation function; u (U) k Is the kth deconvolution core; c k Is the bias constant corresponding to the kth deconvolution core.
S202, training convolution self-encoder based on early degraded signal
The early degradation signal obtained after the pretreatment in the step S1 is x early (t) obtaining a reconstructed signal after encoding and decoding by the input convolution self-encoderCalculating a reconstruction error using a squared difference loss function:
the reconstruction error of the early degraded signal is counter-propagated in the convolutional self-encoder, and the parameters in the self-encoder are updated to reduce the reconstruction error of the early degraded signal. When the reconstruction error is smaller after multiple parameter updating, the convolution self-encoder can be considered to learn the mapping relation between the input signal and the characteristic signal, and the data mode characteristic extraction of the input signal can be realized on the premise of no need of additional information.
S203, calculating the reconstruction error of the test signal.
The test signal obtained after the pretreatment in the step S1 is x (t), and the reconstruction test signal is obtained after the input convolution self-encoder codes and decodesCalculating a reconstruction error using a squared difference loss function:
the unhealthy degree of the equipment corresponding to the test signal is
S3, establishing an index mapping function, and mapping the unhealthy degree into a health factor capable of directly reflecting the health state of the equipment
According to the definition and calculation process of the unhealthy degree of the device, the value interval of the unhealthy degree is [0, + -infinity), and the unhealthy degree approaches to 0 when the degradation degree of the device is lower. The rotary machine has multiple degradation failure modes, and CAE unhealthy degree ranges corresponding to different failure conditions are different. To unify the different failure cases, using the mapping function will be 0, mapping of unhealthy degree over a+ -infinity) interval is the health factor HI over the [0,1] interval. HI may quantitatively describe the state of health of the rotating machine, hi=1 indicating that the device has not yet begun to degrade, hi=0 indicating that the device has completely degraded to failure.
S301, constructing an index mapping function based on the definition of unhealthy degree and health factor HI and determining an index mapping function super-parameter;
According to the degree of unhealthy, the definition of the health factor and the degradation law of the rotating machine, the mapping function should meet the following requirements:
(1) The definition field is a field of [0 ], ++ infinity a) of the above-mentioned components, the value range is [0,1].
(2) Monotonically decreasing and smooth and conductive in the definition domain, and is convenient for calculation.
(3) The saturation is achieved, that is, the function value is approximately 1 when the independent variable is approximately 0, and the function value is approximately 0 when the independent variable is approximately +.
Improving the sigmoid function results in an exponential mapping function that meets the above requirements:
where k and b are the shape factor and bias constant of the exponential mapping function, respectively.
The bias constant b is used to adjust the health factor HI of the device at the initial time and the form factor k is used to control the rate of decrease of the health factor HI as the degree of unhealthy increases. The two parameters need to be artificially given according to the change rule of the initial health state and unhealthy degree of the rotary machine.
S302, rotating machinery health factor calculation
The M test signal fragments x= { x obtained after pretreatment 1N ,x 2N ,...,x MN Sequentially inputting the convolutional self-encoder and the exponential mapping function after the sequence of time to obtain a health factor HI set HI= { HI of the rotating machine 1N ,HI 2N ,...,HI MN }。
S4, carrying out real-time iterative prediction on the residual service life of the rotary machine based on the Weibull reliability function and by combining a gradient descent method.
The method for carrying out real-time iterative prediction on the residual life of the rotary machine by using the Weibull reliability function mainly comprises the following steps: initializing a Weibull reliability function; calculating the residual service life of the rotary machine based on the Weibull reliability function; according to the HI set obtained from the real-time measurement data in the step S3, carrying out parameter updating on the Weibull reliability function by using a gradient descent method; the above two steps are repeated. The specific process is as follows:
s401, initializing Weibull reliability function
The Weibull distribution is a common distribution in reliability theory, and the expression of the two-parameter Weibull reliability function is:
wherein R (t) is Weibull reliability, the reliability of the mechanical equipment is represented, and beta and eta are Weibull reliability function shape parameters and scale parameters respectively.
The initial stage can artificially give initial values of shape parameters and scale parameters according to specific working conditions and historical degradation rules of the rotary machine, and can randomly initialize the shape parameters and the scale parameters.
S402, giving a failure threshold reliability R f Substituting into S401 in the formula (12) to obtain R f Corresponding theoretical failure working time length T of rotary machine f ；
T f ＝η(-ln R f ) 1/β (13)
According to step S3, obtaining health factor HI corresponding to rotating machinery equipment at t moment t . Weibull reliability and health factor distribution interval are 0,1]And both the health factor HI and reliability can be used to characterize the health of the rotating machine, thus HI can be used t As reliability R t HI is taken t Substituting into S401 Chinese formula (12) to obtain HI t Corresponding theoretical working time length T of rotary machine t The method comprises the following steps:
T t ＝η(-ln R t ) 1/β (14)
combining equation (13) and equation (14) to obtain the remaining service life RUL of the rotating machine at time t t The method comprises the following steps:
RUL t ＝T f -T t ＝η[(-ln R f ) 1/β -(-ln R t ) 1/β ] (15)
s403, sequentially solving a health factor set HI= { HI from the beginning of degradation of the rotary machine to the moment t according to S3 1N ,HI 2N ,...,HI MN Substituting the health factor and the corresponding time into formula (12), and selecting a square difference loss function to calculate the fitting error of the Weibull reliability function at time t as follows:
the partial derivatives of the fitting errors on the shape parameters and the scale parameters are respectively calculated as follows:
given a learning rate a, the shape parameter and the scale parameter can be updated according to the learning rate and the partial derivative:
every time a HI is newly calculated according to S3 t S402 and S403 are repeated to implement real-time iterative prediction of the remaining service life of the rotary machine in the case of small samples.
In still another embodiment of the present invention, a digital twin-driven small sample rotary machine residual life prediction system is provided, which can be used to implement the above-mentioned digital twin-driven small sample rotary machine residual life prediction method, and in particular, the digital twin-driven small sample rotary machine residual life prediction system includes a preprocessing module, an evaluation module, a mapping module, and a prediction module.
The preprocessing module is used for preprocessing signals of the rotating mechanical equipment;
the evaluation module is used for establishing a training convolution self-encoder and evaluating the unhealthy degree of the signals obtained by the preprocessing module;
the mapping module is used for establishing an index mapping function and mapping the unhealthy degree obtained by the evaluation module into a health factor capable of directly reflecting the health state of the equipment;
and the prediction module is used for carrying out real-time iterative prediction on the residual service life of the rotary machine by combining a gradient descent method based on the Weibull reliability function and the health factor obtained by the mapping module.
In yet another embodiment of the present invention, a terminal device is provided, the terminal device including a processor and a memory, the memory for storing a computer program, the computer program including program instructions, the processor for executing the program instructions stored by the computer storage medium. The processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processor, digital signal processor (Digital Signal Processor, DSP), application specific integrated circuit (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc., which are the computational core and control core of the terminal adapted to implement one or more instructions, in particular to load and execute one or more instructions to implement the corresponding method flow or corresponding functions; the processor according to the embodiment of the invention can be used for the operation of a residual life prediction method of a digital twin-driven small sample rotary machine, and comprises the following steps:
Preprocessing signals of rotating mechanical equipment; building a training convolution self-encoder to evaluate the unhealthy degree of the signals; establishing an index mapping function, and mapping unhealthy degree into health factors capable of directly reflecting the health state of equipment; based on Weibull reliability function and health factor, the gradient descent method is combined to conduct real-time iterative prediction of the residual service life of the rotary machine.
In a further embodiment of the present invention, the present invention also provides a storage medium, in particular, a computer readable storage medium (Memory), which is a Memory device in a terminal device, for storing programs and data. It will be appreciated that the computer readable storage medium herein may include both a built-in storage medium in the terminal device and an extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space storing an operating system of the terminal. Also stored in the memory space are one or more instructions, which may be one or more computer programs (including program code), adapted to be loaded and executed by the processor. The computer readable storage medium may be a high-speed RAM Memory or a Non-Volatile Memory (Non-Volatile Memory), such as at least one magnetic disk Memory.
One or more instructions stored in a computer-readable storage medium may be loaded and executed by a processor to implement the respective steps of the method for predicting remaining life of a small sample rotary machine in connection with digital twinning driving in the above-described embodiments; one or more instructions in a computer-readable storage medium are loaded by a processor and perform the steps of:
preprocessing signals of rotating mechanical equipment; building a training convolution self-encoder to evaluate the unhealthy degree of the signals; establishing an index mapping function, and mapping unhealthy degree into health factors capable of directly reflecting the health state of equipment; based on Weibull reliability function and health factor, the gradient descent method is combined to conduct real-time iterative prediction of the residual service life of the rotary machine.
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
The method for predicting the residual life of the digital twin-driven small sample rotary machine disclosed by the invention is further described below by substituting a specific bearing vibration signal.
The invention discloses a method for predicting the residual life of a digital twin-driven small sample rotary machine, which comprises the following steps:
s1, preprocessing signals of rotating mechanical equipment:
and collecting the full life cycle vibration signals of the bearings, and filtering the vibration signals by using a Kalman filtering algorithm to obtain noise-reduced signals, wherein the signal of one bearing is shown in figure 3. The bearing rotating speed is 1800r/min, the sampling frequency of the vibration sensor is 25600Hz, the scaling factor is 1.5, and the segment length of the vibration signal of the bearing is 1280. After the noise reduction in fig. 1, the amplitude of the bearing vibration signal is smaller and has no obvious change, the amplitude of the bearing vibration signal slowly increases after 200 seconds and rapidly increases after 280 seconds, which indicates that the bearing is completely failed, and based on the amplitude, the first 40% of time, that is, the first 100 seconds, of the bearing is in an early degradation stage, and the bearing signal segment in the period is selected as a training data set. And processing the collected full life cycle degradation signals of the rest bearings by using the same method and parameters to obtain a training data set consisting of early degradation signals.
S2, realizing unhealthy degree assessment of the input signal by establishing a training convolution self-encoder:
the convolution kernel sizes of a convolution layer and a transposition convolution layer in a convolution self-encoder model are set to be 5, and the step length is set to be 2; the pooling size and the step length of the pooling layer and the reverse pooling layer are both set to be 4, the batch size is 64 when the network is trained, the training round number is 50, and the learning rate is set to be 0.001. The early degradation data training data set in S1 is used to train the convolutional self-encoder, and the early degradation signal and the late degradation signal are input to the convolutional self-encoder for reconstruction, and the pair of the original signal and the reconstructed signal is as shown in fig. 4. As can be seen from the comparison, for the early degraded signal, the convolutional self-encoder well learns the data pattern of the early degraded data, and the original signal can be reconstructed with higher accuracy, but the end-degraded signal cannot be reconstructed by reduction.
The trained convolution self-encoder is used for calculating the reconstruction error of the bearing full life cycle vibration signal, and the calculation result of the reconstruction error is shown in fig. 5. As can be seen from the graph, the reconstruction errors of the bearing signals are close to 0 in early degradation, are stable, and the reconstruction errors are increased sharply in near failure, for example, the reconstruction errors of the three bearing signals in fig. 5 are all greater than 1 in failure, and the reconstruction errors of the bearing samples 1_1 exceed 5.
S3, establishing an index mapping function, and mapping the unhealthy degree into a health factor capable of directly reflecting the health state of the equipment:
as can be seen from the reconstruction errors corresponding to the full life cycle vibration signals of the bearing shown in fig. 3, the reconstruction errors at the initial time of the bearing are all close to 0, i.e., no degradation occurs, and the health factor at this time is considered to be 0. When the bearing is in the rapid degradation or failure phase, the reconstruction error is close to or far greater than 1, and the HI of the bearing is considered to be less than 0.8 at the moment, namely HI (1) <0.8. From the above observation, when b= -4, -5, -6, HI (0) = 0.982, 0.993,0.997, the bias constant is set to-5 considering the effect of the bias constant; from HI (1) <0.8, k >3.61 is solved, and thus the shape parameter is set to 4. Substituting k=4 and b= -6 as the index mapping function parameters into S2 can obtain health factor curves corresponding to three bearing sample vibration signals according to the reconstruction error calculated by the convolution self-encoder as shown in fig. 6.
S4, carrying out real-time iterative prediction on the residual service life of the rotary machine based on Weibull reliability function and by combining a gradient descent method:
the Weibull fail reliability threshold was set to 0.05, and bearing health factor HI was considered to be less than 0.05 for bearing failure, with the shape parameter beta and the size parameter eta of the Weibull fail reliability function initialized to 2 and 15000, respectively. Taking test bearing sample 1_3 as an example, the test data comprises a signal of 18010 seconds before the bearing is in the whole life cycle, the signal of 2700 seconds before the test data is selected as CAE training data, the Weibull reliability function is updated every 1000 seconds after that, the current bearing residual life is predicted, and the iterative prediction result is shown in fig. 7. Fig. 7 shows that as the working time increases and the parameters are updated iteratively, the error between the predicted result of the residual life of the bearing and the actual residual life of the Weibull reliability function gradually decreases, which indicates that the Weibull reliability function learns the mapping relationship between the vibration signal of the bearing and the residual life.
Fig. 8 is a graph of the result of predicting the remaining life of the Weibull reliability function for test set bearing sample 1_3 at 18000 seconds, and fig. 8 shows that the result of predicting the remaining life at this time is 6876 seconds, the actual remaining life is 5730 seconds, and the prediction percentage error is 20%. The remaining life of the remaining test bearing samples was predicted using the same method and procedure, resulting in the predicted results of the method and the results versus other commonly used conventional algorithms shown in Table 1.
TABLE 1
By comparison, the method can realize high-precision prediction of the residual service life of the rotary machine under the condition that only early degradation data are used.
In summary, according to the method and system for predicting the residual life of the digital twin-driven small sample rotary machine, only the early degradation signal is used for training the convolution self-encoder, and the health factor is constructed by using the convolution self-encoder and the exponential mapping function. The Weibull reliability function is introduced as a degradation behavior model to describe the degradation trend of the rotary machine, and the parameter updating of the Weibull reliability function and the calculation of the residual life of the rotary machine are carried out based on the health factor, so that the accurate and real-time prediction of the residual life of the rotary machine under the condition of a small sample is realized.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal and method may be implemented in other manners. For example, the apparatus/terminal embodiments described above are merely illustrative, e.g., the division of the modules or units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated modules/units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RandomAccess Memory, RAM), an electrical carrier wave signal, a telecommunications signal, a software distribution medium, etc., it should be noted that the computer readable medium may contain content that is appropriately increased or decreased according to the requirements of jurisdictions and patent practices, such as in certain jurisdictions, according to the jurisdictions and patent practices, the computer readable medium does not contain electrical carrier wave signals and telecommunications signals.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above is only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited by this, and any modification made on the basis of the technical scheme according to the technical idea of the present invention falls within the protection scope of the claims of the present invention.
Claims (10)
1. The method for predicting the residual life of the digital twin-driven small sample rotary machine is characterized by comprising the following steps of:
s1, preprocessing signals of rotary mechanical equipment;
s2, building a training convolution self-encoder, and evaluating the unhealthy degree of the signals obtained in the step S1;
s3, establishing an index mapping function, and mapping the unhealthy degree obtained in the step S2 into a health factor capable of directly reflecting the health state of the equipment;
s4, carrying out real-time iterative prediction on the residual service life of the rotary machine based on the Weibull reliability function and the health factor obtained in the step S3 by combining a gradient descent method.
2. The method for predicting the remaining life of a digital twin-driven small sample rotary machine according to claim 1, wherein step S1 is specifically:
s101, for a section of measured equipment signal z (t), a one-dimensional Kalman filtering algorithm is used for establishing a state equation of the equipment signal, and an optimal estimated value is obtained by superposing a measured value of an estimated signal with error and an observed value with error;
s102, dividing the filtered signal x (t) obtained in the step S101 into a plurality of signal segments with the length of N, recognizing the signal within the first 30% of time as an early degradation signal of the rotary machine, and using the early degradation signal for training a subsequent algorithm model, wherein the rest is test data.
3. The method for predicting remaining life of a digital twin-driven small sample rotary machine according to claim 2, wherein in step S101, z (t) is taken as an observed value z k And (3) with the initial estimated value at the moment k=0, giving an initial value of a state transition matrix A, a process excitation noise covariance Q, a measurement noise covariance R and a conversion matrix H, and carrying out iterative calculation to obtain a signal x (t) of z (t) after noise reduction through Kalman filtering.
4. The method for predicting the remaining life of a digital twin-driven small sample rotary machine according to claim 2, wherein in step S102, the dividing length N of the device signal is:
Where k is a given scaling factor, f rotate For the rotation frequency of the rotary machine, f collect Is the acquisition frequency of the signal.
5. The method for predicting the remaining life of a digital twin-driven small sample rotary machine according to claim 1, wherein step S2 is specifically:
s201, constructing a convolutional self-encoder neural network model;
s202, the early degradation signal obtained after the pretreatment in the step S1 is x early (t) obtaining a reconstructed signal after encoding and decoding by the input convolution self-encoderCalculating a reconstruction error using a squared difference loss function;
s203, inputting the test signal into a trained convolution self-encoder for signal reconstruction, calculating a reconstruction error of the test signal and the reconstruction signal, and taking the reconstruction error as the unhealthy degree of the equipment corresponding to the test signal.
6. The method for predicting the remaining life of a digital twin-driven small sample rotary machine according to claim 1, wherein step S3 is specifically:
s301, constructing an index mapping function based on the definition of unhealthy degree and health factor HI and determining an index mapping function super-parameter;
s302, obtaining M test signal fragments x= { x after pretreatment 1N ,x 2N ,...,x MN Sequentially inputting the convolutional self-encoder and the exponential mapping function after the sequence of time to obtain a health factor HI set HI= { HI of the rotating machine 1N ,HI 2N ,...,HI MN }。
7. The method for predicting remaining life of a digital twin-driven small sample rotary machine according to claim 6, wherein in step S301, the exponential mapping function data is specifically:
wherein k and b are respectively the shapes of the exponential mapping functionsThe shape factor and the bias constant are used,and reconstructing errors.
8. The method for predicting the remaining life of a digital twin-driven small sample rotary machine according to claim 1, wherein step S4 is specifically:
s401, initializing a Weibull reliability function;
s402, calculating the residual service life of the rotary machine based on a Weibull reliability function;
and S403, updating parameters of the Weibull reliability function by using a gradient descent method according to the HI set obtained from the real-time measurement data in the step S3, and repeating the step S402 and the step S403 to realize real-time iterative prediction of the residual service life of the rotary machine under the condition of a small sample.
9. The method for predicting remaining life of digital twin-driven small sample rotary machine as claimed in claim 1, wherein in step S402, the remaining life RUL of the rotary machine is set at time t t The method comprises the following steps:
RUL t ＝T f -T t ＝η[(-lnR f ) 1/β -(-lnR t ) 1/β ]
wherein T is f For the theoretical failure working time of the rotary machine, T t For the theoretical working time of the rotary machine, eta is a scale parameter, R f For failure threshold reliability, R t For reliability, β is a shape parameter.
10. A digital twinned small sample rotary machine residual life prediction system, comprising:
the preprocessing module is used for preprocessing signals of the rotating mechanical equipment;
the evaluation module is used for establishing a training convolution self-encoder and evaluating the unhealthy degree of the signals obtained by the preprocessing module;
the mapping module is used for establishing an index mapping function and mapping the unhealthy degree obtained by the evaluation module into a health factor capable of directly reflecting the health state of the equipment;
and the prediction module is used for carrying out real-time iterative prediction on the residual service life of the rotary machine by combining a gradient descent method based on the Weibull reliability function and the health factor obtained by the mapping module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310619377.4A CN116561927A (en) | 2023-05-29 | 2023-05-29 | Digital twin-driven small sample rotary machine residual life prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310619377.4A CN116561927A (en) | 2023-05-29 | 2023-05-29 | Digital twin-driven small sample rotary machine residual life prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116561927A true CN116561927A (en) | 2023-08-08 |
Family
ID=87500010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310619377.4A Pending CN116561927A (en) | 2023-05-29 | 2023-05-29 | Digital twin-driven small sample rotary machine residual life prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116561927A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117077327A (en) * | 2023-10-18 | 2023-11-17 | 国网山东省电力公司鱼台县供电公司 | Bearing life prediction method and system based on digital twin |
| CN118052255A (en) * | 2024-03-19 | 2024-05-17 | 广东石油化工学院 | First prediction time determining method, device, equipment and storage medium |
-
2023
- 2023-05-29 CN CN202310619377.4A patent/CN116561927A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117077327A (en) * | 2023-10-18 | 2023-11-17 | 国网山东省电力公司鱼台县供电公司 | Bearing life prediction method and system based on digital twin |
| CN118052255A (en) * | 2024-03-19 | 2024-05-17 | 广东石油化工学院 | First prediction time determining method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116561927A (en) | Digital twin-driven small sample rotary machine residual life prediction method and system | |
| CN110705181B (en) | Rolling bearing residual life prediction method based on convolution length-time memory cyclic neural network | |
| CN113344295B (en) | Method, system and medium for predicting residual life of equipment based on industrial big data | |
| CN105593864B (en) | Analytical device degradation for maintenance device | |
| Chandak et al. | LFZip: Lossy compression of multivariate floating-point time series data via improved prediction | |
| CN113537160B (en) | Ball mill load measuring method, ball mill load measuring device, electronic equipment and medium | |
| CN113288131B (en) | Non-invasive blood glucose detection method, processor and device based on graph convolution network | |
| CN116595465A (en) | High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement | |
| CN114648060A (en) | Fault signal standardization processing and classification method based on machine learning | |
| CN110543869A (en) | Ball screw service life prediction method and device, computer equipment and storage medium | |
| CN117272055B (en) | Electric energy meter abnormality detection method and device based on filtering enhancement self-encoder | |
| CN117951455B (en) | On-line monitoring method for operation faults of scraper conveyor | |
| CN118506525A (en) | Information fusion early warning method and system based on cloud computing | |
| CN117312255B (en) | Electronic document splitting optimization management method and system | |
| CN116504314B (en) | Gene regulation network construction method based on cell dynamic differentiation | |
| CN117951456A (en) | Equipment state intelligent monitoring system based on digital twinning | |
| CN114942066B (en) | Electric spindle error indirect prediction method and device | |
| CN117591860A (en) | Data anomaly detection method and device | |
| CN112152633B (en) | Improved convolution self-encoder method and device for vibration signal compression | |
| CN114993671A (en) | Vibration fault diagnosis method and system based on Q factor wavelet transform | |
| CN114595525A (en) | Rolling bearing degradation trend prediction method based on linear regression and TCN | |
| CN118193955B (en) | Method, device, medium and product for acquiring pneumatic noise of air compressor | |
| CN118467935B (en) | Motor data denoising method and system | |
| CN113963085B (en) | State characterization method and device of industrial system and electronic equipment | |
| CN115859201B (en) | Chemical process fault diagnosis method and system |
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 |