CN110110804A - Flight control system method for predicting residual useful life based on CNN and LSTM - Google Patents
Flight control system method for predicting residual useful life based on CNN and LSTM Download PDFInfo
- Publication number
- CN110110804A CN110110804A CN201910403169.4A CN201910403169A CN110110804A CN 110110804 A CN110110804 A CN 110110804A CN 201910403169 A CN201910403169 A CN 201910403169A CN 110110804 A CN110110804 A CN 110110804A
- Authority
- CN
- China
- Prior art keywords
- control system
- flight control
- data set
- data
- mode
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 title claims abstract description 141
- 238000000034 method Methods 0.000 title claims abstract description 64
- 230000003862 health status Effects 0.000 claims abstract description 27
- 230000008859 change Effects 0.000 claims abstract description 14
- 238000013527 convolutional neural network Methods 0.000 claims description 61
- 238000012549 training Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 21
- 230000036541 health Effects 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 13
- 238000012706 support-vector machine Methods 0.000 claims description 11
- 238000012546 transfer Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000010801 machine learning Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 9
- 230000006399 behavior Effects 0.000 description 5
- CIWBSHSKHKDKBQ-JLAZNSOCSA-N Ascorbic acid Chemical compound OC[C@H](O)[C@H]1OC(=O)C(O)=C1O CIWBSHSKHKDKBQ-JLAZNSOCSA-N 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 4
- 239000010410 layer Substances 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 238000004880 explosion Methods 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000002356 single layer Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 206010011906 Death Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 230000007850 degeneration Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention belongs to machine learning, deep learning and aviation fields, and in particular to a kind of flight control system method for predicting residual useful life based on CNN and LSTM, it is intended to solve the problems, such as that flight control system predicting residual useful life network is complicated, accuracy is low.This system method includes obtaining flight control system data set；Work as the health status of forward using the judgement of fault critical point analysis model；The operation mode of flight control system data set is obtained according to change threshold based on operation modal data；If single mode, using flight control system data set as the second data set；Otherwise classify to operation modal data, obtain the statistical result of the frequency of occurrence when forward and each before turn of corresponding mode, mode classification, statistical result are increased in system data, the second data set is obtained；According to the second data set and fault critical point, the remaining life of each flight control system is obtained based on Life Prediction Model.Present invention reduces the complexities of prediction network, improve the accuracy of prediction.
Description
Technical field
The invention belongs to machine learning, deep learning and aviation fields, and in particular to a kind of winged control based on CNN and LSTM
System spare life-span prediction method.
Background technique
Flight control system is the core component of aircraft.Due to its complexity, randomness, safety, reliability and
The considerations of factors such as economy, fast and accurately predicts that the remaining life of flight control system becomes the weight of aviation field research
Point.
Traditional strategy be by regularly preventative maintenance, it is time-consuming, inefficient, cost dearly.With Internet technology
It continues to develop, the information explosion epoch, various data explosion formulas increased, we can be easily from aircraft mounting assembly
Sensor obtains a large amount of vibration data, this makes data-driven method come into being.Data-driven is found according to measured data
Deterioration law can reveal that potential association and the causality between collected sensing data by statistical learning algorithm,
And then it is inferred to corresponding system message.For example, support vector regression, Hidden Markov Model etc. in typical ML algorithm,
Has good estimated performance.
Traditional ML algorithm can be good at the mapping relations between study acquisition data and target output, but usually require
Manual extraction feature, it is extremely complex in large-scale task.Deep learning (DL) is the derivative of neural network, can be passed through
Multilayer, large-scale neural network learn the level characteristics into data automatically, and usual DL handles single data object, and remaining
Task as useful life prediction needs processing sequence information, i.e., previous output is associated with next input, when time sequence
When arranging longer, parameter can be exponentially increased, so that model is more complicated.RNN is the network of special disposal sequence data, but
During all information of input are transmitted along the time filtering is not added, with the growth of time will appear gradient explosion or
The phenomenon that disappearance, i.e., long-term Dependence Problem.And LSTM increases 4 doors in the structure basis of RNN, so that information is being sieved
After choosing processing just flows to next moment, this problem of effective solution.One-dimensional CNN as a kind of deep learning method,
With characteristics such as the shared, convolution operations of weight, the further feature in mass data can be excavated, processing sequence is also applied for
Data.
Therefore, the invention proposes the residues that one kind predicts flight control system in conjunction with the model of one-dimensional CNN and LSTM parallel
Useful life takes relevant counter-measure by constantly monitoring the operating status of flight control system in time, for example repairs, changes
Deng reducing the incidence etc. of accident, restore the normal operation function of flight control system, shorten predicted time, improve the Shandong of prediction
Stick.
Summary of the invention
It is multiple in order to solve aviation flight control system predicting residual useful life network in order to solve the above problem in the prior art
It is pre- to propose a kind of flight control system remaining life based on CNN and LSTM for problem miscellaneous, accuracy is low, first aspect present invention
Survey method, this method comprises:
Step S10, obtain engine when forward and the corresponding flight control system data set of forward, as the first data set；
The flight control system data set includes the serial number, engine revolution, sensing data, operation modal data of flight control system；
Step S20 is based on first data set, and the healthy shape of forward is worked as using the judgement of fault critical point analysis model
State, if health status is that health thens follow the steps S10, if health status is that failure thens follow the steps S30；
Step S30, will currently transfer to as fault critical point, based on when forward and the corresponding operation modal data of forward,
According to the change threshold of operation modal data, the operation mode of first data set is obtained；If the behaviour of first data set
Making mode is single mode, then using the first data set as the second data set；If the operation mode of first data set is multimode
State carries out mode classification to the operation modal data, obtains each flight control system when forward and each before turn of corresponding mode
The statistical result of frequency of occurrence, and mode classification, the statistical result are increased corresponding flight control system and currently turned in system data,
Obtain the second data set；
Step S40 is carried out remaining according to second data set and fault critical point based on Life Prediction Model
Life prediction obtains the remaining life of each flight control system；
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine
Health status；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for by training
The remaining life of corresponding flight control system is obtained according to the corresponding flight control system data set of fault critical point.
In some preferred embodiments, the fault critical point analysis model training process are as follows:
Using total revolution of engine as the life cycle management of flight control system, each flight control system is found out by health and turns to failure
The time point that failure occurs at first in the process, i.e., the corresponding revolution of first " 1 " appearance；Wherein, health status is labeled as
0, failure state is labeled as 1.
In some preferred embodiments, in step S30 " if the operation mode of first data set be it is multi-modal,
Mode classification is carried out to the operation modal data, obtains each flight control system when the appearance of forward and each before turn of corresponding mode
The statistical result of number, and mode classification, the statistical result are increased corresponding flight control system and currently turned in system data, it obtains
Second data set ", method are as follows:
Using K-means method will when the corresponding operation modal data of each flight control system of forward according to default six mode into
Row cluster, the one-hot one-hot coding that the operation modal data of each flight control system corresponds to mode is increased into the second data set；
Based on each flight control system when the one-hot one-hot coding of forward and each before turn of corresponding mode, each is carried out
The statistics of each mode frequency of occurrence of flight control system, is indicated, and the array is increased into the second data set in a manner of array.
In some preferred embodiments, the Life Prediction Model in the training process, by one-dimensional full connection convolution
Neural network CNN and LSTM extract fault signature parallel, and the feature of extraction is spliced；Wherein one-dimensional convolutional neural networks
Time step is in the same size in the length and LSTM of the convolution kernel of CNN.
In some preferred embodiments, the operation modal data includes height above sea level, Mach number and Sea Level Temperature.
The second aspect of the present invention proposes a kind of flight control system predicting residual useful life system based on CNN and LSTM, should
System includes obtaining module, fault flag module, modal data processing module, prediction module；
The acquisition module, be configured to obtain engine when forward and the corresponding flight control system data set of forward, make
For the first data set；The flight control system data set includes the serial number, engine revolution, sensing data, behaviour of flight control system
Make modal data；
The fault flag module is configured to first data set, is sentenced using fault critical point analysis model
The disconnected health status for working as forward executes acquisition module if health status is health, if health status is that failure executes mode
Data processing module；
The modal data processing module is configured to currently to transfer to for fault critical point, based on when forward and before
Turn corresponding operation modal data, according to the change threshold of operation modal data, obtains the operation mode of first data set；
If the operation mode of first data set is single mode, using the first data set as the second data set；If first number
Operation mode according to collection be it is multi-modal, mode classification is carried out to the operation modal data, each flight control system is obtained and works as forward
And the statistical result of the frequency of occurrence of each before turn of corresponding mode, and mode classification, the statistical result are increased and fly control into corresponding
System currently turns in system data, obtains the second data set；
The prediction module is configured to according to second data set and fault critical point, pre- based on the service life
It surveys model and carries out predicting residual useful life, obtain the remaining life of each flight control system；
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine
Health status；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for by training
The remaining life of corresponding flight control system is obtained according to the corresponding flight control system data set of fault critical point.
The third aspect of the present invention proposes a kind of storage device, wherein be stored with a plurality of program, described program apply by
Processor loads and executes the above-mentioned flight control system method for predicting residual useful life based on CNN and LSTM.
The fourth aspect of the present invention proposes a kind of processing setting, including processor, storage device；Processor is suitable for
Execute each program；Storage device is suitable for storing a plurality of program；Described program is suitable for being loaded by processor and being executed above-mentioned
The flight control system method for predicting residual useful life based on CNN and LSTM.
Beneficial effects of the present invention:
The present invention greatly reduces the complexity of prediction network, improves the accuracy of prediction.The present invention is from data-driven
Angle set out, the time point that each flight control system breaks down dynamically is found using support vector machines, according to what is found out
Fault time carries out the label of remaining life, rather than provides unified persistent fault time point for all flight control systems,
Improve the accuracy of label；It is operation data classification using the method for cluster for the operation data of multioperation mode, and tires out
Add, construct new feature, carries out data enhancing, effectively prevent over-fitting；The one-dimensional shallow-layer connected entirely is combined with parallel organization
Convolutional neural networks CNN and single layer LSTM network are predicted after splicing the data characteristics that the two is extracted, quasi- improving prediction
While true rate, network complexity is reduced.
Detailed description of the invention
By reading the detailed description done to non-limiting embodiment done referring to the following drawings, the application other
Feature, objects and advantages will become more apparent upon.
Fig. 1 is the process of the flight control system method for predicting residual useful life based on CNN and LSTM of an embodiment of the present invention
Schematic diagram；
Fig. 2 is the training of the flight control system method for predicting residual useful life based on CNN and LSTM of an embodiment of the present invention
Process schematic；
The frame of the flight control system predicting residual useful life system based on CNN and LSTM of Fig. 3 an embodiment of the present invention shows
It is intended to.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to the embodiment of the present invention
In technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, without
It is whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, shall fall within the protection scope of the present invention.
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is only used for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to just
Part relevant to related invention is illustrated only in description, attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.
Flight control system method for predicting residual useful life based on CNN and LSTM of the invention, as shown in Figure 1, including following step
It is rapid:
Step S10, obtain engine when forward and the corresponding flight control system data set of forward, as the first data set；
The flight control system data set includes the serial number, engine revolution, sensing data, operation modal data of flight control system；
Step S20 is based on first data set, and the healthy shape of forward is worked as using the judgement of fault critical point analysis model
State, if health status is that health thens follow the steps S10, if health status is that failure thens follow the steps S30；
Step S30, will currently transfer to as fault critical point, based on when forward and the corresponding operation modal data of forward,
According to the change threshold of operation modal data, the operation mode of first data set is obtained；If the behaviour of first data set
Making mode is single mode, then using the first data set as the second data set；If the operation mode of first data set is multimode
State carries out mode classification to the operation modal data, obtains each flight control system when forward and each before turn of corresponding mode
The statistical result of frequency of occurrence, and mode classification, the statistical result are increased corresponding flight control system and currently turned in system data,
Obtain the second data set；
Step S40 is carried out remaining according to second data set and fault critical point based on Life Prediction Model
Life prediction obtains the remaining life of each flight control system；
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine
Health status；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for by training
The remaining life of corresponding flight control system is obtained according to the corresponding flight control system data set of fault critical point.
In order to be more clearly illustrated to the flight control system method for predicting residual useful life the present invention is based on CNN and LSTM,
Each step carries out expansion detailed description in a kind of 1 and 2 pair of embodiment of the method for the present invention with reference to the accompanying drawing.
Hereafter in preferred embodiment, first to the failure in the flight control system method for predicting residual useful life based on CNN and LSTM
Critical point analysis model and Life Prediction Model are described in detail, then again to based on fault critical point analysis model and life prediction
The flight control system method for predicting residual useful life based on CNN and LSTM that model obtains the remaining life of each flight control system carries out detailed
It states.
1, the training of fault critical point analysis model
Step A10 obtains flight control system data set, and is standardized.
Using 21 sensors such as the relevant vibration of flight control system, temperature, pressure, multiple flight control systems full longevity is acquired
Order the sensing data in the period, since the operation mode of engine in flight course constantly changes, the mould of synchronization
State data are also used as a part of initial data to be collected.The total revolution of operation of engine is regarded as to week life-cycle of flight control system
Phase, each turn is a time of running, and at an acquisition moment, each acquisition moment is collected simultaneously current time 21 sensings
The data of device and 3 operation modal datas, final collected data are configured to sensing data, operation modal data, start
Machine revolution, flight control system serial number.
Wherein, sensing data includes pitch angle, roll angle, course, pressure altitude, engine speed, engine blower
Inlet pressure, engine blower inlet temperature；Operating modal data includes height above sea level, Mach number and Sea Level Temperature.
In practical applications, a large amount of raw sensor monitoring data, operation data, load, when failure occurs can be provided
Between etc., the dimension of these data is different, and in order to eliminate influence of the dimension to data, is standardized pretreatment just to data
In the processing and comparison of data.Z- is carried out in the life cycle of current flight control system to the data of collected 21 sensors
Score standardization pretreatment, so that a variety of data such as pressure, temperature, angle, speed from different sensors unify dimension,
Convenient for the processing and comparison of data.
Z-score is standardized as shown in formula (1):
Wherein, X is original sensor data, and X ' is the data after standardization, and μ and σ are respectively sensing data
Mean value and variance.
Step A11, with 30 turns before and after life cycle management in flight control system data set of data Training Support Vector Machines,
It is predicted to obtain flight control system using all data of the support vector machines to life cycle management and starts the time broken down and turned
Number carries out remaining useful life label to the flight control system data of life cycle management according to time revolution.
Total revolution of engine is regarded to the life cycle management of flight control system as, it is assumed that each flight control system life cycle management
Preceding 30 switch to health status (labeled as 0), and rear 30 switch to failure state (labeled as 1).By 60 after each flight control system label
Revolution is trained according to support vector machines is sent into, then the data of life cycle management are sent into support vector machines and are predicted, is found out by being good for
Health turns to the time point that failure occurs at first in failure procedure, i.e. revolution corresponding to first " 1 ", using this time point as state
Turning point continues to mark to the remaining useful life of flight control system.
In actual application, when engine just begins to use, can not have to consider decline problem.Only at some
After time point breaks down, the degeneration of performance, end-of-life when reaching entirely ineffective can just occur for engine.Therefore will
Flight control system remaining useful life is modeled as broken line type.It is turning point (inflection point of broken line) by the above-mentioned time revolution found, with
At this point remaining useful life (revolution of total revolution-current point) label before data, this point after remaining service life
Order linear decrease.
2, the training of Life Prediction Model
Step A20 obtains the flight control system training dataset of label according to fault critical point analysis model, by winged
The operation modal data of the data set of control system is judged, the mode of current data set is obtained, and carries out the service life if single mode
Prediction model training；Otherwise operation modal data is clustered using K-means method according to default six mode, is flown each
The operation modal data of control system corresponds to the one-hot one-hot coding of mode, is added in behind raw data set；Each is flown
The statistics of each mode frequency of occurrence of control system, is added in behind raw data set, then carries out Life Prediction Model training.
It is exactly single behaviour if being no more than change threshold in that corresponding three parameters of entire life cycle operation modal data
Make mode, is otherwise exactly multioperation mode.
Data under multioperation mode, since operation mode does not stop to change, having no rule can be sayed, solve multi-modal ask at present
The common method of topic is exactly divided into multiple single modes for multi-modal, and models respectively to each mode.
Therefore operation modal data is clustered using K-means method according to default six mode and uses one-hot
The form of one-hot coding indicates.One-hot coding is also known as an efficient coding, and method is using N bit status register come to N number of
State is encoded, and each state has its independent register-bit, and when any, wherein only one effective.Example
For example mode 1, then it represents that it is 100000, mode 2 is then expressed as 010000, and so on.By the one-hot coding operation after cluster
The a part of mode classification as new data, is added in behind raw data set, to increase the dimension of characteristic.
According to mode, the mode categorical data after cluster is subjected to the cumulative of life cycle management, so that having no rule originally
Governed modal data has fixed growth trend, is conducive to model with this and acquires useful information.Operation after will be cumulative
The a part of modal data as new data, is added in behind raw data set, increases the dimension of characteristic again, carries out
Data enhancing.As shown in table 1, by taking a flight control system as an example, the revolution of engine life cycle management is 10000 turns, right
Should each turn of corresponding mode and mode accumulation result it is as shown in Table.
Table 1
| Mode | Mode is cumulative | |
| 1st turn | 100000 | 1,0,0,0,0,0 |
| 2nd turn | 100000 | 2,0,0,0,0,0 |
| 3rd turn | 010000 | 2,1,0,0,0,0 |
| 4th turn | 010000 | 2,2,0,0,0,0 |
| 5th turn | 100000 | 3,2,0,0,0,0 |
| … | … | … |
| 10000th turn | 100000 | 1200,800,1900,1100,1500,3500 |
It according to the ratio cut partition of 7:3 is training set and test set by new data, the data of training set are sent into one-dimensional to be connected entirely
It connects in convolution CNN and LSTM network and is trained, then the data in test set are sent into network, output is that each flies control system
It unites the remaining useful life of each time of running.
Step A40 is carried out predicting residual useful life based on one-dimensional CNN and LSTM, is obtained based on the data set that step A30 is obtained
Take the remaining life of each flight control system.
In modern industrial equipment operational process, some failure may correspond to several signs, some sign may also
Inevitably there is multimode step response in the process of running in corresponding multiple failures, system, using LSTM deep learning method into
The multi-modal predicting residual useful life research of row has more far-reaching engineering application requirement.
The inspiration of Life Prediction Model (being constructed by one-dimensional full connection convolution CNN and single layer LSTM) is originated from integrated study.Collection
It is to be learnt using multiple weak learners, then by ballot, majority voting determines classification, to reach and learn by force at study
The same effect of device.LSMT and one-dimensional full connection convolutional neural networks are suitable for processing time series data, therefore this model will
Both is combined with structure arranged side by side, respectively with the two network transaction datas, extracts feature, then the feature extracted is carried out
Splicing, the dimension for carrying out third time increase, and data enhancing further avoids the generation of over-fitting.Structure arranged side by side drops significantly
The low complexity of network, simplifies the composition of model, good prediction effect can be obtained using shallow-layer neural network.
Step A41, convolutional neural networks CNN are usually used in handling two dimensional image, and one-dimensional complete connection convolutional neural networks are then
One-dimensional sequence data can be handled.Substantially, one-dimensional convolutional neural networks are equal to processing X-Y scheme in processing sequence data
Picture, only the first of sequence data dimension is the time, and second dimension is feature.Two data of image data are all pictures
Element value.For one-dimensional full connection convolutional neural networks during feature is extracted in sliding, sliding window is equal to time step in LSTM
Concept guarantee dimension matching therefore in order in merging features, the convolution kernel of full connection convolutional neural networks one-dimensional at this time
Length must be consistent with the size of time step in LSTM, be the difference with common convolutional neural networks herein.Simultaneously
It also needs to be filled, to guarantee that the length of feature extraction context data does not change.
The key of step A42, LSTM are exactly cell state, and on the diagram just through operation, cell state is similar to horizontal line
Conveyer belt.It is directly run on entire chain, only some a small amount of linear reciprocals, information is spread above to be remained unchanged and can hold very much
Easily.
The first step in our LSTM is to determine what information we can abandon from cell state, this decision passes through
One is known as forgetting that gate layer is completed.The gate layer can read ht-1And xt, the numerical value between 0 to 1 is exported to each in cell
Number in state.As shown in formula (2):
ft=σ (Wf·[ht-1,xt]+bf) (2)
Wherein, ftFor the value of the forgetting door of t moment, σ is sigmoid function, WfFor the weighted value for forgeing door, xtFor t moment
Input, ht-1For the output at t-1 moment, bfFor the bias for forgeing door.
It is which type of determining new information is stored in cell state in next step.As shown in formula (3) (4):
it=σ (Wi·[ht-1,xt]+bi) (3)
Wherein, itFor the value of t moment input gate, WiFor the weighted value of input gate, biFor the bias of input gate.
Wherein,For the updated value of t moment cell C,For tanh function, WCFor the weighted value of cell C, bCFor cell C's
Bias.
Determine that the information updated is the time of new and old cell state now, as shown in formula (5):
Wherein, CtFor the state of t moment cell C.
It is final to update cell state, it would be desirable to what value of output determined.This output will be based on our cellular
State, but it is also a filtered version.As shown in formula (6) (7):
ot=σ (Wo·[ht-1,xt]+bo) (6)
Wherein, otFor the value of t moment out gate, WoFor the weighted value of out gate, boFor the bias of out gate.
ht=ot*tanh(Ct) (7)
Wherein, htFor the output of t moment, tanh is tanh function.
Life Prediction Model is based on one-dimensional full connection convolutional Neural according to the flight control system fault data marked in training set
Network C NN and LSTM extract fault signature, and carry out merging features, construct the degradation trend of flight control system data.Pass through training
For obtaining the remaining life of corresponding flight control system according to the corresponding flight control system data set of fault critical point.
3, the method for the flight control system predicting residual useful life based on CNN and LSTM
Step S10, obtain engine when forward and the corresponding flight control system data set of forward, as the first data set；
The flight control system data set includes the serial number, engine revolution, sensing data, operation modal data of flight control system.
In the present embodiment, the sensor based on flight control system obtains sensing data, using z-score to sensing data
Unified dimension is carried out, and obtains the modal data and engine revolution of synchronization, according to the serial number of flight control system, is obtained current
The data set of flight control system.
Step S20 is based on first data set, and the healthy shape of forward is worked as using the judgement of fault critical point analysis model
State, if health status is that health thens follow the steps S10, if health status is that failure thens follow the steps S30.
In the present embodiment, the life cycle management that total revolution of engine is regarded as to flight control system utilizes fault critical point
Each flight control system of analysis model is turned to the time point that failure occurs at first in failure procedure by health, judges the current revolution of engine
Health status.Continue to obtain if health, otherwise predicting residual useful life.
Step S30, will currently transfer to as fault critical point, based on when forward and the corresponding operation modal data of forward,
According to the change threshold of operation modal data, the operation mode of first data set is obtained；If the behaviour of first data set
Making mode is single mode, then using the first data set as the second data set；If the operation mode of first data set is multimode
State carries out mode classification to the operation modal data, obtains each flight control system when forward and each before turn of corresponding mode
The statistical result of frequency of occurrence, and mode classification, the statistical result are increased corresponding flight control system and currently turned in system data,
Obtain the second data set.
In the present embodiment, will currently transfer to as fault critical point, based on when forward and the corresponding operation mode of forward
Data obtain the operation mode of first data set, if first data set according to the change threshold of operation modal data
Operation mode be single mode, then using the first data set as the second data set.
If the operation mode of first data set be it is multi-modal, each flight control system of forward will be worked as using K-means method
Corresponding operation modal data is clustered according to default six mode, and the operation modal data of each flight control system is corresponded to mode
One-hot one-hot coding increase into the second data set；Based on each flight control system when forward and each before turn of corresponding mode
One-hot one-hot coding is carried out the statistics of each mode frequency of occurrence of each flight control system, is indicated in a manner of array, and should
Array increases into the second data set.
Step S40 is carried out remaining according to second data set and fault critical point based on Life Prediction Model
Life prediction obtains the remaining life of each flight control system.
In the present embodiment, the beginning future position of each flight control system remaining life is obtained according to fault critical point, is based on institute
The second data set obtained, is input in Life Prediction Model, extracts feature, root based on LSTM network and parallel one-dimensional CNN
According to the corresponding relationship of fault signature and remaining life, the remaining life of each flight control system is obtained.
A kind of flight control system predicting residual useful life system based on CNN and LSTM of second embodiment of the invention, such as Fig. 3 institute
Show, comprising: obtain module 100, fault flag module 200, modal data processing module 300, prediction module 400；
Obtain module 100, be configured to obtain engine when forward and the corresponding flight control system data set of forward, as the
One data set；The flight control system data set includes the serial number, engine revolution, sensing data, operation mould of flight control system
State data；
Fault flag module 200 is configured to first data set, is worked as using the judgement of fault critical point analysis model
The health status of forward executes if health status is health and obtains module 100, if health status is that failure executes mode number
According to processing module 300；
Modal data processing module 300, configuration will be transferred to currently as fault critical point, based on when forward and forward it is corresponding
Operation modal data the operation mode of first data set is obtained according to the change threshold of operation modal data；If described
The operation mode of first data set is single mode, then using the first data set as the second data set；If first data set
Operate mode be it is multi-modal, mode classification is carried out to the operation modal data, obtains each flight control system when forward and before
The statistical result of the frequency of occurrence of each turn of corresponding mode, and mode classification, the statistical result are increased and worked as into corresponding flight control system
In forward system data, the second data set is obtained；
Prediction module 400 is configured to be based on life prediction mould according to second data set and fault critical point
Type carries out predicting residual useful life, obtains the remaining life of each flight control system；
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine
Health status；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for by training
The remaining life of corresponding flight control system is obtained according to the corresponding flight control system data set of fault critical point.
The technical personnel in the technical field can be clearly understood that, for convenience and simplicity of description, foregoing description
The specific course of work of system and related explanation, can be no longer superfluous herein with reference to the corresponding process in signature embodiment of the method
It states.
It should be noted that the flight control system predicting residual useful life system provided by the above embodiment based on CNN and LSTM,
Only the example of the division of the above functional modules, in practical applications, it can according to need and divide above-mentioned function
With being completed by different functional modules, i.e., by the embodiment of the present invention module or step decompose or combine again, for example,
The module of above-described embodiment can be merged into a module, can also be further split into multiple submodule, to complete above retouch
The all or part of function of stating.For module involved in the embodiment of the present invention, the title of step, it is only for distinguish each
A module or step, are not intended as inappropriate limitation of the present invention.
A kind of storage device of third embodiment of the invention, wherein be stored with a plurality of program, described program be suitable for by
Reason device loads and realizes the above-mentioned flight control system method for predicting residual useful life based on CNN and LSTM.
A kind of processing unit of fourth embodiment of the invention, including processor, storage device；Processor is adapted for carrying out each
Program；Storage device is suitable for storing a plurality of program；Described program is suitable for being loaded by processor and being executed to realize above-mentioned base
In the flight control system method for predicting residual useful life of CNN and LSTM.
The technical personnel in the technical field can be clearly understood that is do not described is convenienct and succinct, foregoing description
The specific work process and related explanation of storage device, processing unit can refer to the corresponding process in signature method example,
This is repeated no more.
Those skilled in the art should be able to recognize that, mould described in conjunction with the examples disclosed in the embodiments of the present disclosure
Block, method and step, can be realized with electronic hardware, computer software, or a combination of the two, software module, method and step pair
The program answered can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electric erasable and can compile
Any other form of storage well known in journey ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field is situated between
In matter.In order to clearly demonstrate the interchangeability of electronic hardware and software, in the above description according to function generally
Describe each exemplary composition and step.These functions are executed actually with electronic hardware or software mode, depend on technology
The specific application and design constraint of scheme.Those skilled in the art can carry out using distinct methods each specific application
Realize described function, but such implementation should not be considered as beyond the scope of the present invention.
Term " first ", " second " etc. are to be used to distinguish similar objects, rather than be used to describe or indicate specific suitable
Sequence or precedence.
Term " includes " or any other like term are intended to cover non-exclusive inclusion, so that including a system
Process, method, article or equipment/device of column element not only includes those elements, but also including being not explicitly listed
Other elements, or further include the intrinsic element of these process, method, article or equipment/devices.
So far, it has been combined preferred embodiment shown in the drawings and describes technical solution of the present invention, still, this field
Technical staff is it is easily understood that protection scope of the present invention is expressly not limited to these specific embodiments.Without departing from this
Under the premise of the principle of invention, those skilled in the art can make equivalent change or replacement to the relevant technologies feature, these
Technical solution after change or replacement will fall within the scope of protection of the present invention.
Claims (8)
1. a kind of flight control system method for predicting residual useful life based on CNN and LSTM, which is characterized in that the prediction technique includes:
Step S10, obtain engine when forward and the corresponding flight control system data set of forward, as the first data set；It is described
Flight control system data set include the serial number of flight control system, engine revolution, sensing data, operation modal data；
Step S20 is based on first data set, and the health status of forward is worked as using the judgement of fault critical point analysis model, if
Health status is that health thens follow the steps S10, if health status is that failure thens follow the steps S30；
Step S30, will currently transfer to as fault critical point, based on when forward and the corresponding operation modal data of forward, according to
The change threshold of modal data is operated, the operation mode of first data set is obtained；If the operation mould of first data set
State is single mode, then using the first data set as the second data set；If the operation mode of first data set be it is multi-modal, it is right
The operation modal data carries out mode classification, obtains each flight control system when forward and the occurrence out of each before turn of corresponding mode
Several statistical results, and mode classification, the statistical result are increased corresponding flight control system and currently turned in system data, obtain the
Two data sets；
Step S40 carries out remaining life based on Life Prediction Model according to second data set and fault critical point
Prediction, obtains the remaining life of each flight control system；
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine strong
Health state；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for basis by training
The corresponding flight control system data set of fault critical point obtains the remaining life of corresponding flight control system.
2. the flight control system method for predicting residual useful life according to claim 1 based on CNN and LSTM, which is characterized in that
The fault critical point analysis model training process are as follows:
Using total revolution of engine as the life cycle management of flight control system, each flight control system is found out by health and turns to failure procedure
The time point that middle failure occurs at first, i.e., the corresponding revolution of first " 1 " appearance；Wherein, health status is labeled as 0, loses
Imitating status indication is 1.
3. the flight control system method for predicting residual useful life according to claim 1 based on CNN and LSTM, which is characterized in that
In step S30 " if the operation mode of first data set be it is multi-modal, mode point is carried out to the operation modal data
Class obtains each flight control system when forward and the statistical result of the frequency of occurrence of each before turn of corresponding mode, and the mode is classified,
The statistical result increases corresponding flight control system and currently turns in system data, obtains the second data set ", method are as follows:
It will be gathered when the corresponding operation modal data of each flight control system of forward according to default six mode using K-means method
Class increases the one-hot one-hot coding that the operation modal data of each flight control system corresponds to mode into the second data set；
Based on each flight control system when the one-hot one-hot coding of forward and each before turn of corresponding mode, carries out each and fly control
The statistics of each mode frequency of occurrence of system, is indicated, and the array is increased into the second data set in a manner of array.
4. the flight control system method for predicting residual useful life according to claim 1 based on CNN and LSTM, which is characterized in that
The Life Prediction Model extracts failure spy in the training process, by one-dimensional full connection convolutional neural networks CNN and LSTM parallel
Sign, the feature of extraction is spliced；Wherein time step in the length of the convolution kernel of one-dimensional convolutional neural networks CNN and LSTM
It is in the same size.
5. the flight control system method for predicting residual useful life according to claim 1 based on CNN and LSTM, which is characterized in that
The operation modal data includes height above sea level, Mach number and Sea Level Temperature.
6. a kind of flight control system predicting residual useful life system based on CNN and LSTM, which is characterized in that the system includes obtaining mould
Block, fault flag module, modal data processing module, prediction module；
The acquisition module, be configured to obtain engine when forward and the corresponding flight control system data set of forward, as the
One data set；The flight control system data set includes the serial number, engine revolution, sensing data, operation mould of flight control system
State data；
The fault flag module is configured to first data set, is worked as using the judgement of fault critical point analysis model
The health status of forward executes acquisition module if health status is health, if health status is that failure executes modal data
Processing module；
The modal data processing module, is configured to currently to transfer to as fault critical point, based on when forward and forward pair
The operation modal data answered obtains the operation mode of first data set according to the change threshold of operation modal data；If institute
The operation mode for stating the first data set is single mode, then using the first data set as the second data set；If first data set
Operation mode be it is multi-modal, mode classification is carried out to the operation modal data, obtain each flight control system when forward and it
The statistical result of the frequency of occurrence of first each turn corresponding mode, and mode classification, the statistical result are increased into corresponding flight control system
Currently turn in system data, obtains the second data set；
The prediction module is configured to be based on life prediction mould according to second data set and fault critical point
Type carries out predicting residual useful life, obtains the remaining life of each flight control system
Wherein,
The fault critical point analysis model is constructed based on support vector machines, by training for judging each turn of engine strong
Health state；
The Life Prediction Model is based on one-dimensional full connection convolutional neural networks CNN and LSTM building, is used for basis by training
The corresponding flight control system data set of fault critical point obtains the remaining life of corresponding flight control system.
7. a kind of storage device, wherein being stored with a plurality of program, which is characterized in that described program is applied and loaded and held by processor
Row is to realize the described in any item flight control system method for predicting residual useful life based on CNN and LSTM of claim 1-5.
8. a kind of processing setting, including processor, storage device；Processor is adapted for carrying out each program；Storage device is fitted
For storing a plurality of program；It is characterized in that, described program is suitable for being loaded by processor and being executed to realize claim 1-5
Described in any item flight control system method for predicting residual useful life based on CNN and LSTM.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910403169.4A CN110110804B (en) | 2019-05-15 | 2019-05-15 | Flight control system residual life prediction method based on CNN and LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910403169.4A CN110110804B (en) | 2019-05-15 | 2019-05-15 | Flight control system residual life prediction method based on CNN and LSTM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110110804A true CN110110804A (en) | 2019-08-09 |
| CN110110804B CN110110804B (en) | 2021-06-25 |
Family
ID=67490221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910403169.4A Active CN110110804B (en) | 2019-05-15 | 2019-05-15 | Flight control system residual life prediction method based on CNN and LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110110804B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110609524A (en) * | 2019-08-14 | 2019-12-24 | 华中科技大学 | Industrial equipment residual life prediction model and construction method and application thereof |
| CN110705176A (en) * | 2019-09-02 | 2020-01-17 | 北京市燃气集团有限责任公司 | Method and device for predicting residual life of gas pipeline |
| CN112308306A (en) * | 2020-10-27 | 2021-02-02 | 贵州工程应用技术学院 | Multi-mode input coal and gas outburst risk prediction method |
| CN112488208A (en) * | 2020-12-03 | 2021-03-12 | 上海电力大学 | Method for acquiring remaining life of island pillar insulator |
| CN113052302A (en) * | 2021-04-02 | 2021-06-29 | 重庆大学 | Machine health monitoring method and device based on recurrent neural network and terminal equipment |
| CN113449626A (en) * | 2021-06-23 | 2021-09-28 | 中国科学院上海高等研究院 | Hidden Markov model vibration signal analysis method and device, storage medium and terminal |
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102402727A (en) * | 2011-11-10 | 2012-04-04 | 中联重科股份有限公司 | System and method for predicting residual life of part of engineering machine |
| CN105808957A (en) * | 2016-03-15 | 2016-07-27 | 清华大学 | Method for predicting residual life of aero-engine |
| CN106094570A (en) * | 2016-07-13 | 2016-11-09 | 北京航空航天大学 | A kind of aero-engine complete machine health evaluating method under variable working condition based on operating mode's switch and this distance of paddy |
| CN106155092A (en) * | 2015-04-21 | 2016-11-23 | 高域（北京）智能科技研究院有限公司 | A kind of intelligent multi-control flight capture apparatus and flight control method thereof |
| CN107958325A (en) * | 2017-11-15 | 2018-04-24 | 中国航空工业集团公司西安飞机设计研究所 | Aircraft time related cost evaluation method |
| CN108009674A (en) * | 2017-11-27 | 2018-05-08 | 上海师范大学 | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks |
| CN108038334A (en) * | 2017-12-28 | 2018-05-15 | 西北工业大学 | A kind of airplane cabin door lock System reliability modeling method for considering burst failure threshold and degenerating |
| CN108959778A (en) * | 2018-07-06 | 2018-12-07 | 南京航空航天大学 | A kind of aero-engine method for predicting residual useful life based on degradation modes consistency |
| CN109033499A (en) * | 2018-06-07 | 2018-12-18 | 南京航空航天大学 | A kind of aero-engine method for predicting residual useful life of multistage consistency check |
| CN109272114A (en) * | 2018-08-08 | 2019-01-25 | 西安交通大学 | The deep learning method that mechanized equipment remaining life is quickly and effectively predicted |
| CN109409567A (en) * | 2018-09-17 | 2019-03-01 | 西安交通大学 | Complex device method for predicting residual useful life based on the double-deck shot and long term memory network |
| CN109472110A (en) * | 2018-11-29 | 2019-03-15 | 南京航空航天大学 | A kind of aero-engine remaining life prediction technique based on LSTM network and ARIMA model |
| CN109522600A (en) * | 2018-10-16 | 2019-03-26 | 浙江大学 | Complex equipment remaining life prediction technique based on combined depth neural network |
| CN109685224A (en) * | 2018-12-20 | 2019-04-26 | 威海众成信息科技股份有限公司 | A kind of management method for formulating engine maintenance decision according to lifetime piece |
-
2019
- 2019-05-15 CN CN201910403169.4A patent/CN110110804B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102402727A (en) * | 2011-11-10 | 2012-04-04 | 中联重科股份有限公司 | System and method for predicting residual life of part of engineering machine |
| CN106155092A (en) * | 2015-04-21 | 2016-11-23 | 高域（北京）智能科技研究院有限公司 | A kind of intelligent multi-control flight capture apparatus and flight control method thereof |
| CN105808957A (en) * | 2016-03-15 | 2016-07-27 | 清华大学 | Method for predicting residual life of aero-engine |
| CN106094570A (en) * | 2016-07-13 | 2016-11-09 | 北京航空航天大学 | A kind of aero-engine complete machine health evaluating method under variable working condition based on operating mode's switch and this distance of paddy |
| CN107958325A (en) * | 2017-11-15 | 2018-04-24 | 中国航空工业集团公司西安飞机设计研究所 | Aircraft time related cost evaluation method |
| CN108009674A (en) * | 2017-11-27 | 2018-05-08 | 上海师范大学 | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks |
| CN108038334A (en) * | 2017-12-28 | 2018-05-15 | 西北工业大学 | A kind of airplane cabin door lock System reliability modeling method for considering burst failure threshold and degenerating |
| CN109033499A (en) * | 2018-06-07 | 2018-12-18 | 南京航空航天大学 | A kind of aero-engine method for predicting residual useful life of multistage consistency check |
| CN108959778A (en) * | 2018-07-06 | 2018-12-07 | 南京航空航天大学 | A kind of aero-engine method for predicting residual useful life based on degradation modes consistency |
| CN109272114A (en) * | 2018-08-08 | 2019-01-25 | 西安交通大学 | The deep learning method that mechanized equipment remaining life is quickly and effectively predicted |
| CN109409567A (en) * | 2018-09-17 | 2019-03-01 | 西安交通大学 | Complex device method for predicting residual useful life based on the double-deck shot and long term memory network |
| CN109522600A (en) * | 2018-10-16 | 2019-03-26 | 浙江大学 | Complex equipment remaining life prediction technique based on combined depth neural network |
| CN109472110A (en) * | 2018-11-29 | 2019-03-15 | 南京航空航天大学 | A kind of aero-engine remaining life prediction technique based on LSTM network and ARIMA model |
| CN109685224A (en) * | 2018-12-20 | 2019-04-26 | 威海众成信息科技股份有限公司 | A kind of management method for formulating engine maintenance decision according to lifetime piece |
Non-Patent Citations (4)
| Title |
|---|
| FENG XUE ET AL: "An Instance-Based Method for Remaining Useful Life Estimation for Aircraft Engines", 《TECHNICAL ARTICLE—PEER-REVIEWED》 * |
| 高育林: "基于深度学习的多模态故障诊断及剩余寿命预测", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 黄加阳 等: "民用飞机健康状态评估方法", 《计算机测量与控制》 * |
| 黄强 等: "一种航空发动机传感器故障诊断方法", 《传感技术学报》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110609524A (en) * | 2019-08-14 | 2019-12-24 | 华中科技大学 | Industrial equipment residual life prediction model and construction method and application thereof |
| CN110609524B (en) * | 2019-08-14 | 2020-07-28 | 华中科技大学 | Industrial equipment residual life prediction model and construction method and application thereof |
| CN110705176A (en) * | 2019-09-02 | 2020-01-17 | 北京市燃气集团有限责任公司 | Method and device for predicting residual life of gas pipeline |
| CN112308306A (en) * | 2020-10-27 | 2021-02-02 | 贵州工程应用技术学院 | Multi-mode input coal and gas outburst risk prediction method |
| CN112488208A (en) * | 2020-12-03 | 2021-03-12 | 上海电力大学 | Method for acquiring remaining life of island pillar insulator |
| CN112488208B (en) * | 2020-12-03 | 2023-02-14 | 上海电力大学 | Method for acquiring remaining life of island pillar insulator |
| CN113052302A (en) * | 2021-04-02 | 2021-06-29 | 重庆大学 | Machine health monitoring method and device based on recurrent neural network and terminal equipment |
| CN113449626A (en) * | 2021-06-23 | 2021-09-28 | 中国科学院上海高等研究院 | Hidden Markov model vibration signal analysis method and device, storage medium and terminal |
| CN113449626B (en) * | 2021-06-23 | 2023-11-07 | 中国科学院上海高等研究院 | Method and device for analyzing vibration signal of hidden Markov model, storage medium and terminal |
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117723782B (en) * | 2024-02-07 | 2024-05-03 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110110804B (en) | 2021-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110110804A (en) | Flight control system method for predicting residual useful life based on CNN and LSTM | |
| CN111552609B (en) | Abnormal state detection method, system, storage medium, program and server | |
| de Pater et al. | Alarm-based predictive maintenance scheduling for aircraft engines with imperfect Remaining Useful Life prognostics | |
| CN111274737A (en) | Method and system for predicting remaining service life of mechanical equipment | |
| CN110609524B (en) | Industrial equipment residual life prediction model and construction method and application thereof | |
| Girish et al. | Anomaly detection in cloud environment using artificial intelligence techniques | |
| CN110737732A (en) | electromechanical equipment fault early warning method | |
| Murphree | Machine learning anomaly detection in large systems | |
| CN116577677B (en) | Discharging test system and method for retired power battery | |
| Wu et al. | A framework of dynamic data driven digital twin for complex engineering products: the example of aircraft engine health management | |
| CN109947898B (en) | Equipment fault testing method based on intellectualization | |
| CN111985155A (en) | Circuit health state prediction method and system based on integrated deep neural network | |
| CN113869563A (en) | Method for predicting remaining life of aviation turbofan engine based on fault feature migration | |
| CN103324155A (en) | System monitoring | |
| CN110757510A (en) | Method and system for predicting remaining life of robot | |
| Attoh‐Okine | Grouping Pavement Condition Variables for Performance Modeling Using Self‐Organizing Maps | |
| Shi et al. | Health index synthetization and remaining useful life estimation for turbofan engines based on run-to-failure datasets | |
| CN110580213A (en) | Database anomaly detection method based on cyclic marking time point process | |
| CN112465045A (en) | Supply chain exception event detection method based on twin neural network | |
| CN107977748A (en) | Multivariable distorted time sequence prediction method | |
| Li et al. | Reliability of non-repairable systems with cyclic-mission switching and multimode failure components | |
| Li et al. | An enhanced CNN-LSTM remaining useful life prediction model for aircraft engine with attention mechanism | |
| Gopalakrishnan et al. | IIoT Framework Based ML Model to Improve Automobile Industry Product. | |
| CN112381213A (en) | Industrial equipment residual life prediction method based on bidirectional long-term and short-term memory network | |
| Bond et al. | A hybrid learning approach to prognostics and health management applied to military ground vehicles using time-series and maintenance event data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |