CN116562120A - RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device - Google Patents
RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device Download PDFInfo
- Publication number
- CN116562120A CN116562120A CN202310269210.XA CN202310269210A CN116562120A CN 116562120 A CN116562120 A CN 116562120A CN 202310269210 A CN202310269210 A CN 202310269210A CN 116562120 A CN116562120 A CN 116562120A
- Authority
- CN
- China
- Prior art keywords
- turbine engine
- engine system
- data
- system health
- rve
- 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
- 230000036541 health Effects 0.000 title claims abstract description 96
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 48
- 238000006731 degradation reaction Methods 0.000 claims abstract description 19
- 238000007781 pre-processing Methods 0.000 claims abstract description 14
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 238000012795 verification Methods 0.000 claims abstract description 10
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims abstract description 9
- 238000011156 evaluation Methods 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 31
- 238000013210 evaluation model Methods 0.000 claims description 22
- 230000015654 memory Effects 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 10
- 230000004913 activation Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000007787 long-term memory Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 3
- 238000002759 z-score normalization Methods 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 2
- 230000008685 targeting Effects 0.000 claims 1
- 230000015556 catabolic process Effects 0.000 description 11
- 238000012423 maintenance Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000003449 preventive effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- FESBVLZDDCQLFY-UHFFFAOYSA-N sete Chemical compound [Te]=[Se] FESBVLZDDCQLFY-UHFFFAOYSA-N 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M15/00—Testing of engines
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application relates to the technical field of turbine engine system health evaluation, and discloses a method and a device for evaluating the health condition of a turbine engine system based on RVE, wherein the method comprises the following steps: acquiring various sensor parameter data in the whole life cycle of a turbine engine system, performing basic preprocessing operation on a data set, constructing each batch of data according to a sliding window method, dividing each batch of data into a training set, a verification set and a test set, constructing a model by taking the training set data as input, constructing a loss function aiming at a residual service life prediction task, training a turbine engine system health condition assessment model by adopting an improved loss function and an Adam optimizer, and finally inputting historical data of the turbine engine system to be assessed into the model, and outputting a residual service life (RUL) degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition assessment curve, thereby enabling the assessment of the system health state to be more accurate.
Description
Technical Field
The application relates to the technical field of turbine engine system health assessment, in particular to a method and a device for assessing the health condition of a turbine engine system based on RVE.
Background
Detection of the health of mechanical systems is critical in any field, however traditional strategies, such as regular preventive maintenance or corrective maintenance of faults, are increasingly inadequate to meet the increasing demands of the industry in terms of efficiency and reliability. Therefore, system health indicators such as residual life (RUL) have been established as key elements for maintaining mechanical systems and preventing engineering safety problems.
In the aviation safety background, the problem of overall system faults caused by hardware health condition degradation exists in the turbine engine system. Data is typically collected from various built-in sensors to monitor turbine engine system operating conditions, build system Remaining Useful Life (RUL), which can help engineers formulate preventive maintenance schemes and avoid turbine engines operating under conditions of hardware degradation, effectively increasing turbine engine flight time, and reducing maintenance costs.
However, as the amount of information that can be collected increases and the accuracy requirements for preventive maintenance are continuously improved over the years, the prior art has the problems of low utilization rate of the parameter characteristics of the multi-dimensional sensor, low accuracy of the predicted life index, and the like.
Disclosure of Invention
The utility model aims to overcome the defects of the prior art and provide a method and a device for evaluating the health condition of a turbine engine system based on RVE.
In a first aspect, there is provided a RVE-based turbine engine system health assessment method comprising:
acquiring data X acquired by sensors in the whole life cycle of the turbine engine;
preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set;
constructing an RVE-based turbine engine system health evaluation model;
training a turbine engine system health assessment model by adopting an improved loss function and an Adam optimizer;
the historical data of the turbine engine system to be evaluated is input into a turbine engine system health condition evaluation model which is trained to output the remaining service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve.
Further, the data x= [ X ] 1 ,x 2 ,…,x N ]Wherein, the method comprises the steps of, wherein,n is the batch size, d is the input dimension size, which depends on the number of sensors selected, +.>Representing a real matrix of n rows and d columns.
Further, constructing an RVE-based turbine engine system health assessment model, comprising: the training set is used as input, full life cycle data characteristics of the turbine engine system are mapped to a potential space, residual service lives corresponding to different values of sensor parameter data are learned, a turbine engine system health condition assessment model comprises a variable self-encoder, a potential space and a decoder composed of a regression model, the variable self-encoder comprises a two-way long-short-term memory network layer and two full-connection layers, the potential space comprises heavy parameter operation, and the decoder composed of the regression model comprises a full-connection layer, a tanh activation function and an output layer.
Further, the calculation process of the turbine engine system health evaluation model is as follows:
the variable self-encoder comprises a two-way long-short-period memory network layer and two fully-connected layers with the same structure, wherein the two fully-connected layers with the same structure are respectively used for learning the average value mu= (mu) of training set data 1 ,…,μ l ) Sum variance σ= (σ) 1 ,…,σ l ) L is the size of potential space dimension, and in the calculation process of the two-way long-short-term memory network layer, the dimension of input is (seq_len, batch_size, input_size) to obtain a hidden layer state sequence with the same sequence length as the sensor parameter dataAnd +.>Output +.>In (a) and (b)Performing splicing operation, and respectively inputting two full-connection layers with the same structure to obtain encoder outputs mu and var;
in potential space, the encoder outputs mu and var are changed from non-conductive to conductive by using a heavy parameter skill, firstly, sampling is carried out from Gaussian distribution with a mean value of 0 and a standard deviation of 1, then scaling and translation are carried out to obtain hidden variable Z for forward propagation operation of the second half, and the specific calculation process is as follows:
wherein mu i Is the ith dimension value, sigma of the mean value mu obtained by training set learning i Is to learn the obtained mean sigma through a training setE is the tensor according to normal distribution, z i Is the ith dimension value of the hidden variable Z;
in a decoder composed of a regression model, a fully connected layer, a hyperbolic tangent activation function and a fully connected layer are included as output layers.
Further, before the label trains the turbine engine system health evaluation model, initializing the parameter weight of all network layers to be 0, constructing a loss function and an optimizer aiming at the residual service life prediction task, and obtaining the trained model by taking the minimum loss function as a target.
Further, preprocessing the data X includes: and denoising the data X, removing abnormal values by adopting a Laida criterion, and finally performing Z-Score normalization processing.
In a second aspect, there is provided an RVE-based turbine engine system health assessment apparatus comprising:
the acquisition module is used for acquiring data X acquired by the sensors in the whole life cycle of the turbine engine;
the preprocessing module is used for preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set;
the model building module is used for building an RVE-based turbine engine system health evaluation model;
the model training module is used for training the turbine engine system health condition evaluation model by adopting an improved loss function and an Adam optimizer;
the output module is used for inputting the historical data of the turbine engine system to be evaluated into the turbine engine system health condition evaluation model after training is completed, so as to output the residual service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve.
Further, the turbine engine system health assessment model comprises a variable self-encoder, a potential space and a regression model, wherein the variable self-encoder comprises a two-way long-short-term memory network layer and two fully-connected layers, the potential space comprises heavy parameter operation, and the regression model comprises a fully-connected layer, a tanh activation function and an output layer.
In a third aspect, a computer readable storage medium is provided, the computer readable medium storing program code for execution by a device, the program code comprising steps for performing the method as in any one of the implementations of the first aspect.
In a fourth aspect, there is provided an electronic device comprising a processor, a memory and a program or instruction stored on the memory and executable on the processor, which when executed by the processor implements a method as in any of the implementations of the first aspect.
The application has the following beneficial effects: the method for evaluating the health condition of the turbine engine system by constructing the residual service life degradation curve by using the RVE model is provided creatively, the parameter characteristics of the multidimensional sensor are fully utilized, the health condition evaluation of the turbine engine system is more accurate, and the method can be applied to the health condition management and maintenance of the turbine engine system and has strong practicability.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application, illustrate and explain the application and are not to be construed as limiting the application.
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method of RVE-based turbine engine system health assessment according to an embodiment of the present application;
FIG. 2 is a block diagram of a turbine engine system health assessment model in a RVE-based turbine engine system health assessment method according to an embodiment of the present application;
FIG. 3 is a graph showing a comparison of degradation functions for constructing a health assessment indicator tag for remaining useful life in an RVE-based turbine engine system health assessment method according to an embodiment of the present application;
FIG. 4 is a training iteration schematic of a turbine engine system health assessment model in a RVE-based turbine engine system health assessment method according to an embodiment of the present application;
FIG. 5 is a schematic diagram of the prediction results of a turbine engine system health assessment model in a RVE-based turbine engine system health assessment method according to an embodiment of the present application;
FIG. 6 is a simplified schematic diagram of simulated engine in the IEEE PHM08 turbofan engine degradation simulation dataset in accordance with one embodiment of the present application.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. 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.
Example 1
An RVE-based turbine engine system health assessment method according to an embodiment of the present application includes: acquiring data X acquired by sensors in the whole life cycle of the turbine engine; preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set; constructing an RVE-based turbine engine system health evaluation model; training a turbine engine system health assessment model by adopting an improved loss function and an Adam optimizer; the method evaluates the health condition of the turbine engine system by constructing a residual service life degradation curve by using an RVE model, fully utilizes the parameter characteristics of a multidimensional sensor, greatly improves the accuracy of a predicted life index, can be applied to the health condition management and maintenance of the turbine engine system, and has strong practicability.
Specifically, fig. 1 shows a flowchart of a method for evaluating health of an RVE-based turbine engine system in application embodiment one, including:
s101, acquiring data X acquired by a sensor in the whole life cycle of a turbine engine;
specifically, the data X includes the total HPC outlet temperature (T30), the total LPT outlet temperature (T50), the total HPC outlet pressure (P30), the engine pressure ratio P50/P2 (epr), the static HPC outlet pressure (Ps 30), the ratio of the fuel flow to Ps30 (phi), and the data x= [ X ] 1 ,x 2 ,…,x N ]Wherein, the method comprises the steps of, wherein,n is the batch size, d is the input dimension size, which depends on the number of sensors selected, +.>Representing a real matrix of n rows and d columns.
S102, preprocessing data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set, wherein the method specifically comprises the following steps:
s201, denoising the acquired data, removing outliers by adopting a Laida criterion, and finally performing Z-Score normalization processing, wherein the specific steps are as follows:
s2010, denoising the data set subjected to normalization processing to obtain a data set subjected to data preprocessing.
S2011, carrying out outlier rejection on the acquired data, wherein the Leida criterion is adopted, and only preserving x of the value of each dimension in the range (mu-3 sigma, mu+3 sigma) i Wherein mu j Is the mean value of the jth dimension of X, sigma j Is the standard deviation of the jth dimension of X.
S2012, calculating a normalized data set subjected to Z-Score processing, wherein the normalization process of the data set is as follows:
wherein x is i,j Representing the ith time step x i Values of the j-th feature dimension, mu j Is the mean value of the jth dimension of X, sigma j Is the standard deviation of the jth dimension of X;
s202, constructing a health index of 'residual service life' (RUL) for the preprocessed data set in the step S201 as a data set label, as shown in fig. 3, since the system always tends to deteriorate, a degradation trend is usually required to be presupposed, a target RUL is constructed by using labels based on the assumptions, model training is guided in a supervised manner and prediction accuracy is enhanced, a 'piecewise linear degradation function' is adopted here, the first 125 cycles of full life cycle data of the turbine engine system are taken as maximum RUL labels, the maximum RUL of the system under a single working condition is 130, the maximum RUL of the system under a multiple working conditions is 150, the RUL labels of the single working condition system after the 125 th cycle are adopted, the label range is [130,0], the RUL labels of the multiple working condition system after the 125 th cycle are adopted, and the label range is [150,0], so as to obtain the preprocessed data set with the label of 'residual service life' (RUL);
s203, selecting sensor data [ x ] in each time window with the size of 30 by a sliding window method i ,x i+1 ,…,x i+29 ]Forming a high-dimensional feature vector as an input sample and using x i+30 As a corresponding output of the sample, guide modelingAnd finally, grouping all samples obtained by a sliding window method according to the engine numbers, and dividing the samples according to the ratio of 7:2:1 to obtain a training set, a verification set and a test set.
S103, constructing an RVE-based turbine engine system health evaluation model, referring to FIG. 2, specifically comprising the following steps:
s301, taking training set data divided in the step S102 as input, and learning the corresponding residual service life (RUL) under different values of sensor parameter data by mapping full life cycle data features of a turbine engine system to potential space;
s302, the turbine engine system health evaluation model consists of three parts, namely a variable self-encoder, a potential space and a regression model, wherein the variable self-encoder comprises a two-way long-short-term memory network layer (Bi-LSTM) with a hidden layer size of 300 and two fully-connected layers with an input size of 600 and an output size of 2, the potential space only comprises a heavy parameter operation, the decoder comprises the fully-connected layers with the input size of 2 and the output size of 200, a tanh activation function and the output layer with the input size of 200 and the output size of 1, and the specific calculation process of each part is as follows:
s3021, in the variable self-encoder, a two-way long-short-term memory network layer (Bi-LSTM) is included, and two fully-connected layers with the same structure are respectively used for learning the average value μ= (μ) of the training set data 1 ,…,μ l ) Sum variance σ= (σ) 1 ,…,σ l ) L is the potential space dimension size;
in the calculation of the two-way long and short term memory network layer (Bi-LSTM), the input dimension is (seq_len, batch_size, input_size), we will get hidden layer sequence with the same sequence length as the sensor parameter dataAnd +.>Output +.>Is->And performing splicing operation, and respectively inputting two full-connection layers with the same structure to obtain encoder outputs mu and var.
S3022, in the potential space part, the encoder outputs mu and var of step S3021 are changed from non-conductive to conductive by using the "heavy parameter skill", and the hidden variable Z is obtained by sampling from the gaussian distribution with the mean value of 0 and the standard deviation of 1, scaling and translating the sampled data to be used for the forward propagation operation of the second half, and the specific calculation process is as follows:
wherein mu i Is the ith dimension value, sigma of the mean value mu obtained by training set learning i Is the ith dimension value of the mean sigma obtained through training set learning, epsilon is tensor conforming to normal distribution, z i Is the ith dimension value of the hidden variable Z.
S3023, in the decoder composed of the regression model, a fully connected layer (input dimension is potential space dimension size), hyperbolic tangent activation function, and a fully connected layer (output dimension is 1) are included as output layers.
S104, training a turbine engine system health evaluation model by adopting an improved loss function and an Adam optimizer;
for example, the training set data and the residual life label divided in step S102 are input into a training model in the health condition evaluation model, before the training model, the parameter weight of all network layers is initialized to 0, and a loss function and an optimizer for the residual life prediction task are constructed, and the trained model is obtained with the aim of minimizing the loss function, as shown in fig. 4, specifically the steps are as follows:
s401, initializing the parameter weights of all network layers in the step S302 to 0, and obtaining an initial model before training.
S402, in a loss function aiming at an RUL prediction task, in order to improve model prediction precision, a Root Mean Square Error (RMSE) is added as a part of the loss function, and the loss function is used for learning training set reconstruction by combining with a Kullback-Leibler divergence to realize a better feature extraction effect, wherein the specific formula is as follows:
wherein X is training set data, Z is hidden variable vector, theta and phi respectively represent parameters of an encoder and a decoder,is the predicted Remaining Useful Life (RUL), y i Is a remaining life (RUL) tag;
s403, training the model by using an Optimizer Adam Optimizer to take the minimum value of the loss function of the health condition evaluation model as an optimization target, and obtaining a trained model.
S105, inputting historical data of the turbine engine system to be evaluated into a turbine engine system health condition evaluation model which is trained to output a remaining service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve, as shown in fig. 5.
As shown in FIG. 6, in one particular embodiment, data collection was performed on the IEEE PHM08 turbofan engine degradation simulation data set C-MAPSS in 2008. Firstly, 6 sensor parameters T30, T50, P30, epr, ps30 and phi are selected as input data, denoising, abnormal value removing and normalization operation are completed on the data, a piecewise linear degradation function is adopted to construct a hypothetical health index (RUL) as a data set label, and then the input and the corresponding output contained in each batch are divided according to a sliding window with the size of 30 by a sliding window method, and the simulated data set is divided into a training set and a verification set according to 8:2 because the simulated data set is provided with a turbine engine test set to be evaluated and a label thereof.
Then, building a RVE-based turbine engine system health assessment model, learning sensor parameter characteristics through a variation self-encoder, performing 'heavy parameter' operation in a potential space for forward propagation of the latter half, and finally completing health assessment of the turbine engine system in a decoder formed by a regression model, wherein model prediction effects are optimized by adopting a loss function aiming at RUL prediction tasks, finally, inputting a turbine engine test set to be assessed into the prediction model which completes training, and outputting a residual service life (RUL) degradation process of the turbine engine system until complete failure as a health assessment curve.
Experimental results show that the health condition assessment method provided by the invention has higher reference guiding value, and compared with the label of the test set, the RUL assessment performance index RMSE of the model on the test set is 11.05, and the RUL degradation curve serving as the health condition assessment curve can have higher prediction precision.
Example two
An RVE-based turbine engine system health assessment device according to a second embodiment of the present application includes:
the acquisition module is used for acquiring data X acquired by the sensors in the whole life cycle of the turbine engine;
the preprocessing module is used for preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set;
the model building module is used for building an RVE-based turbine engine system health evaluation model;
the model training module is used for training the turbine engine system health condition evaluation model by adopting an improved loss function and an Adam optimizer;
the output module is used for inputting the historical data of the turbine engine system to be evaluated into the turbine engine system health condition evaluation model after training is completed, so as to output the residual service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve.
In a further embodiment, the turbine engine system health assessment model includes a variational self-encoder including a two-way long and short term memory network layer and two fully connected layers, a latent space including heavy parameter operations, and a regression model including a fully connected layer, a tanh activation function, and an output layer.
Example III
A computer readable storage medium according to a third embodiment of the present application stores program code for execution by a device, the program code including steps for performing the method in any one of the implementations of the first embodiment of the present application;
wherein the computer readable storage medium may be a Read Only Memory (ROM), a static storage device, a dynamic storage device, or a random access memory (random access memory, RAM); the computer readable storage medium may store program code which, when executed by a processor, is adapted to carry out the steps of a method as in any one of the implementations of the first embodiment of the present application.
Example IV
An electronic device according to a fourth embodiment of the present application includes a processor, a memory, and a program or an instruction stored in the memory and executable on the processor, where the program or the instruction implements a method according to any one of the implementations of the first embodiment of the present application when executed by the processor;
the processor may be a general-purpose central processing unit (central processing unit, CPU), microprocessor, application specific integrated circuit (application specific integrated circuit, ASIC), graphics processor (graphics processing unit, GPU) or one or more integrated circuits for executing relevant programs to implement the methods according to any of the implementations of the first embodiment of the present application.
The processor may also be an integrated circuit electronic device with signal processing capabilities. In implementation, each step of the method in any implementation of the first embodiment of the present application may be implemented by an integrated logic circuit of hardware in a processor or an instruction in software form.
The processor may also be a general purpose processor, a digital signal processor, an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (field programmable gatearray, FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The disclosed methods, steps, and logic blocks in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be embodied directly in a hardware decoding processor or in a combination of hardware and software modules in the decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads information in the memory, and in combination with hardware thereof, performs functions required to be performed by units included in the data processing apparatus according to the embodiment of the present application, or performs a method in any implementation manner of the first embodiment of the present application.
The above is only a preferred embodiment of the present application; the scope of protection of the present application is not limited in this respect. Any person skilled in the art, within the technical scope of the present disclosure, shall cover the protection scope of the present application by making equivalent substitutions or alterations to the technical solution and the improved concepts thereof.
Claims (10)
1. A method of RVE-based turbine engine system health assessment, comprising:
acquiring data X acquired by sensors in the whole life cycle of the turbine engine;
preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set;
constructing an RVE-based turbine engine system health evaluation model;
training a turbine engine system health assessment model by adopting an improved loss function and an Adam optimizer;
the historical data of the turbine engine system to be evaluated is input into a turbine engine system health condition evaluation model which is trained to output the remaining service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve.
2. The RVE-based turbine engine system health assessment method of claim 1, wherein the data x= [ X 1 ,x 2 ,…,x N ]Wherein, the method comprises the steps of, wherein,n is the batch size, d is the input dimension size, which depends on the number of sensors selected, +.>Representing a real matrix of n rows and d columns.
3. The RVE-based turbine engine system health assessment method of claim 1, wherein constructing an RVE-based turbine engine system health assessment model comprises: the training set is used as input, full life cycle data characteristics of the turbine engine system are mapped to a potential space, residual service lives corresponding to different values of sensor parameter data are learned, a turbine engine system health condition assessment model comprises a variable self-encoder, a potential space and a decoder composed of a regression model, the variable self-encoder comprises a two-way long-short-term memory network layer and two full-connection layers, the potential space comprises heavy parameter operation, and the decoder composed of the regression model comprises a full-connection layer, a tanh activation function and an output layer.
4. The RVE-based turbine engine system health assessment method according to claim 3, wherein the turbine engine system health assessment model is calculated by:
the variable self-encoder comprises a two-way long-short-period memory network layer and two fully-connected layers with the same structure, wherein the two fully-connected layers with the same structure are respectively used for learning the average value mu= (mu) of training set data 1 ,…,μ l ) Sum variance σ= (σ) 1 ,…,σ l ) L is the size of potential space dimension, and in the calculation process of the two-way long-short-term memory network layer, the dimension of input is (seq_len, batch_size, input_size) to obtain a hidden layer state sequence with the same sequence length as the sensor parameter dataAnd +.>Output +.>In (a) and (b)Performing splicing operation, and respectively inputting two full-connection layers with the same structure to obtain encoder outputs mu and var;
in potential space, the encoder outputs mu and var are changed from non-conductive to conductive by using a heavy parameter skill, firstly, sampling is carried out from Gaussian distribution with a mean value of 0 and a standard deviation of 1, then scaling and translation are carried out to obtain hidden variable Z for forward propagation operation of the second half, and the specific calculation process is as follows:
wherein mu i Is the ith dimension value, sigma of the mean value mu obtained by training set learning i Is the ith dimension value of the mean sigma obtained through training set learning, epsilon is tensor conforming to normal distribution, z i Is the ith dimension value of the hidden variable Z;
in a decoder composed of a regression model, a fully connected layer, a hyperbolic tangent activation function and a fully connected layer are included as output layers.
5. The RVE-based turbine engine system health assessment method of claim 3, wherein initializing all network layer parameter weights to 0 before training the turbine engine system health assessment model by the tag, and constructing a loss function and optimizer for remaining life prediction tasks, the trained model targeting minimizing the loss function.
6. The RVE-based turbine engine system health assessment method of claim 1, wherein preprocessing data X comprises: and denoising the data X, removing abnormal values by adopting a Laida criterion, and finally performing Z-Score normalization processing.
7. An RVE-based turbine engine system health assessment apparatus, comprising:
the acquisition module is used for acquiring data X acquired by the sensors in the whole life cycle of the turbine engine;
the preprocessing module is used for preprocessing the data X, constructing a data set health index as a data tag, dividing the input and the corresponding output contained in each batch by a sliding window method, and dividing the data set into a training set, a verification set and a test set;
the model building module is used for building an RVE-based turbine engine system health evaluation model;
the model training module is used for training the turbine engine system health condition evaluation model by adopting an improved loss function and an Adam optimizer;
the output module is used for inputting the historical data of the turbine engine system to be evaluated into the turbine engine system health condition evaluation model after training is completed, so as to output the residual service life degradation process of the turbine engine system until the turbine engine system is completely failed as a health condition evaluation curve.
8. The RVE-based turbine engine system health assessment device of claim 7, wherein the turbine engine system health assessment model comprises a variational self-encoder comprising a two-way long and short term memory network layer and two fully connected layers, a latent space comprising heavy parameter operations, and a decoder comprising a regression model comprising fully connected layers, a tanh activation function, and an output layer.
9. A computer readable storage medium storing program code for execution by a device, the program code comprising steps for performing the method of any one of claims 1-6.
10. An electronic device comprising a processor, a memory, and a program or instruction stored on the memory and executable on the processor, which when executed by the processor, implements the method of any of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310269210.XA CN116562120A (en) | 2023-03-20 | 2023-03-20 | RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310269210.XA CN116562120A (en) | 2023-03-20 | 2023-03-20 | RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116562120A true CN116562120A (en) | 2023-08-08 |
Family
ID=87485110
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310269210.XA Pending CN116562120A (en) | 2023-03-20 | 2023-03-20 | RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116562120A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117521528A (en) * | 2024-01-03 | 2024-02-06 | 中国核动力研究设计院 | Turbine equipment simulation model evolution method, device, medium and computing equipment |
CN118211190A (en) * | 2024-05-15 | 2024-06-18 | 中国长江电力股份有限公司 | Section data analysis method and system based on working condition grid |
-
2023
- 2023-03-20 CN CN202310269210.XA patent/CN116562120A/en active Pending
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117521528A (en) * | 2024-01-03 | 2024-02-06 | 中国核动力研究设计院 | Turbine equipment simulation model evolution method, device, medium and computing equipment |
CN117521528B (en) * | 2024-01-03 | 2024-03-15 | 中国核动力研究设计院 | Turbine equipment simulation model evolution method, device, medium and computing equipment |
CN118211190A (en) * | 2024-05-15 | 2024-06-18 | 中国长江电力股份有限公司 | Section data analysis method and system based on working condition grid |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Xiang et al. | LSTM networks based on attention ordered neurons for gear remaining life prediction | |
US20220358363A1 (en) | Engine surge fault prediction system and method based on fusion neural network model | |
CN116562120A (en) | RVE-based turbine engine system health condition assessment method and RVE-based turbine engine system health condition assessment device | |
Ayodeji et al. | Causal augmented ConvNet: A temporal memory dilated convolution model for long-sequence time series prediction | |
CN112598248B (en) | Load prediction method, load prediction device, computer equipment and storage medium | |
CN111814956B (en) | Multi-task learning air quality prediction method based on multi-dimensional secondary feature extraction | |
CN110264270B (en) | Behavior prediction method, behavior prediction device, behavior prediction equipment and storage medium | |
CN111784061B (en) | Training method, device and equipment for power grid engineering cost prediction model | |
CN115017826B (en) | Method for predicting residual service life of equipment | |
CN116821646A (en) | Data processing chain construction method, data reduction method, device, equipment and medium | |
CN112801315A (en) | State diagnosis method and device for power secondary equipment and terminal | |
CN114330934A (en) | Model parameter self-adaptive GRU new energy short-term power generation power prediction method | |
CN115935807A (en) | Diagnostic model training method based on graph Markov neural network | |
CN115017819A (en) | Engine remaining service life prediction method and device based on hybrid model | |
Falcon et al. | Neural turing machines for the remaining useful life estimation problem | |
CN112560252A (en) | Prediction method for residual life of aircraft engine | |
CN114911788B (en) | Data interpolation method and device and storage medium | |
CN110991741A (en) | Section constraint probability early warning method and system based on deep learning | |
CN115217152A (en) | Method and device for predicting opening and closing deformation of immersed tunnel pipe joint | |
CN116089822B (en) | Equipment RUL prediction method and system based on space-time attention network | |
CN118137492B (en) | Short-term power load prediction method and system | |
CN117725543B (en) | Multi-element time sequence anomaly prediction method, electronic equipment and storage medium | |
Zhai et al. | Reconstruction of missing points in agricultural machinery trajectory based on bidirectional adjacent information | |
CN115730245A (en) | Fault diagnosis method, device, equipment and medium for oil-immersed power transformer | |
CN116307159A (en) | Load prediction method and device, electronic equipment and storage medium |
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 |