CN114741948A - Aero-engine degradation trend prediction method based on residual stacked convolution network of sequence reconstruction - Google Patents
Aero-engine degradation trend prediction method based on residual stacked convolution network of sequence reconstruction Download PDFInfo
- Publication number
- CN114741948A CN114741948A CN202210223893.0A CN202210223893A CN114741948A CN 114741948 A CN114741948 A CN 114741948A CN 202210223893 A CN202210223893 A CN 202210223893A CN 114741948 A CN114741948 A CN 114741948A
- Authority
- CN
- China
- Prior art keywords
- output
- sequence
- model
- prediction
- representing
- 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
- 238000006731 degradation reaction Methods 0.000 title claims abstract description 63
- 230000015556 catabolic process Effects 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 230000004913 activation Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000000593 degrading effect Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for predicting degradation trend of an aeroengine based on a sequence reconstruction residual stacked convolution network, which comprises the following steps: step one, preprocessing data; step two, constructing a basic unit block; step three, constructing and training an integral model; fourthly, predicting the degradation trend of the aero-engine by using the model; the method proposes a basic unit block for sequence reconstruction, the basic unit block output comprising a re-prediction value for an input sequence and a prediction value for a future sequence. Meanwhile, based on the basic unit block, the method for predicting the degradation trend of the aero-engine based on the residual stacked convolutional network is provided, the provided model fully excavates and utilizes historical sequence information, the problem that the gradient of the network model disappears is effectively avoided, and the accuracy of predicting the degradation trend of the aero-engine is improved.
Description
Technical Field
The invention belongs to the technical field of aero-engine performance degradation, and particularly relates to a method for predicting aero-engine degradation trend based on a sequence reconstruction residual stacked convolution network.
Background
Prediction of degradation trend of an aero-engine is one of the key points of research in the field of aviation industry, and efficient and accurate prediction of degradation performance trend can help managers to maintain the health state of the engine and save maintenance cost. Based on the monitoring data of the aircraft engine, the performance degradation state index of the engine can be obtained, the performance degradation index can effectively reflect the degradation trend of the engine, if the engine is in a degradation state, the engine degradation state index is predicted, the future change condition of the running state of the engine can be effectively monitored, and an effective maintenance strategy is formulated.
At present, the prediction method of the degradation trend of the aircraft engine mainly comprises a method based on a traditional model and a method based on an intelligent learning algorithm. The prediction method based on the traditional model generally comprises an autoregressive model, an autoregressive moving average model and the like, the traditional model has strict requirements on the regularity of the change of the degradation trend, the accuracy is low, and the application range is small. The prediction method based on the intelligent learning algorithm mainly comprises an artificial neural network model, a support vector regression model and the like. The traditional time series prediction model is generally only suitable for stable signals, but large fluctuation exists in the signals of the aero-engine, so that researchers have strong nonlinear approximation capability based on an intelligent learning algorithm to predict the performance degradation trend of the engine. The following problems still exist in the prediction of the performance degradation trend of the aircraft engine: (1) the future performance state of the engine is related to the performance degradation state of a plurality of previous flight cycles, so that when the performance degradation state of the engine is predicted, not only the state of the historical flight cycles but also the time accumulation effect of the performance degradation process are considered. (2) The degradation process of the engine has high randomness, and a degradation state prediction model established based on some classical distributions, such as a Gaussian process, a gamma process and the like, cannot effectively monitor the severe change of the performance degradation state caused by an emergency.
In order to solve the problems, a new method for predicting the performance degradation trend of the aircraft engine is provided, namely a prediction method based on a residual stacking convolution network of sequence reconstruction.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a method for predicting the degradation trend of an aircraft engine based on a residual stacked convolutional network with sequence reconstruction, aiming at the defects in the prior art. Information in the degraded sequence is fully mined and utilized by a residual stacking convolution network method based on sequence reconstruction. The plurality of basic unit blocks are integrated into one stack, residual errors of the plurality of basic unit blocks are input into the next stack, and the problem that the residual errors are too small and the gradient disappears when the residual errors are directly input into the next unit is effectively solved. And finally, integrating a plurality of stack structures to form a final model. The residual error of the last stack learned by each stack is summed to obtain the final output result, and the multi-module integration idea improves the accuracy of the overall prediction of the model.
In order to solve the technical problems, the invention adopts the technical scheme that: the method for predicting the degradation trend of the aero-engine based on the residual stacked convolutional network of the sequence reconstruction is characterized by comprising the following steps of:
step one, data preprocessing:
step 101: let the state sequence of the whole degradation process of the engine be N, and the degradation state sequence X ═ X1,x2,…,xN]The performance degradation state of the ith sensor is denoted as xiI is 1,2, …, N, and the performance degradation state sequence is sequence data which comprehensively represents the performance degradation state in the aircraft engine degradation process;
step 102: dividing samples of the whole degradation state sequence by a sliding window method, selecting data at continuous k moments as model input data, and constructing a training sample { X) by taking data at the k +1 moment as a model predicted valuein＝[x1,x2,…,xk]；Yin＝[xk+1]},k＜N；
Step two, constructing a basic unit block:
step 201: according to the formula hL＝fr(Xk·wL+bL) Establishing a linear layer of elementary blocks, processing the input data at k instants, where Xin＝[x1,x2,…,xk]For k input data, wLAnd bLWeight and offset, h, representing the linear layerLRepresenting linear layersOutput of (f)rActivating a function for the ReLU;
step 202: according to the formula hC＝fr(Conv1d(hL) Build up a convolutional layer of basic unit blocks, output h to the linear layerLProcessing is performed, where Conv1d denotes a one-dimensional convolution operation, hCRepresenting the output of the convolutional layer, frFor the ReLU activation function, the convolution layer comprises z layers of one-dimensional convolution operation together for extracting the characteristics of input data;
step 203: according to formula Xf＝fr(FCf(hC) A reconstructed sequence portion of a basic cell block, which is a portion for performing a re-prediction on input sequence data, wherein Xf＝[x1',x2',…,xk']Represents the output of the re-prediction, hCRepresenting the output of the convolutional layer, FCfDenotes the fully-connected layer, frActivating a function for the ReLU;
step 204: according to formula Xb＝fr(FCb(hC) Building a prediction part of a basic unit block, which is a part for predicting data at time k +1, where Xb＝[xk+1']Represents the output of the prediction at time k +1, hCIndicating the output of the convolutional layer, FCbDenotes the fully-connected layer, frActivating a function for the ReLU;
step three, constructing an integral model and training:
step 301: according to the formula
Combining h basic cell blocks into a stack, where XrRepresenting the sum of residuals for each basic unit block in the stack,
for the output of the re-prediction sequence of the jth basic unit block, the input of the current basic unit block in the stack is the output X of the previous basic unit block re-predictionf＝[x1',x2',…,xk']；
Step 302:according to the formula
A prediction output part of the calculation stack, wherein YsThe output of the stack is represented by,
the prediction output of the j basic unit block k +1 in the stack is represented;
step 303: according to the formula
Calculating a prediction output of an integral model consisting of m stacks, wherein
Representing the predicted output of the p-th stack, and Y representing the overall predicted output of the model;
step 304: according to the formula
Calculating a training loss of the model, wherein YtRepresenting the predicted output of the model, ytRepresenting the real output of the samples, and S represents the number of the samples; the model activation function selects a ReLU function, and the optimizer selects an Adam optimizer;
fourthly, predicting the degradation trend of the aircraft engine by using the model:
step 401: let the sequence to be predicted be Xpre＝[x1,x2,…,xT]Intercepting k continuous time sequences at the end of the sequence to be predicted according to the step 102 to form a sample X 'to be predicted'pre＝[xT-k,xT-k+1,…,xT]Wherein T represents the sequence length of the sample to be predicted;
step 402: sample X 'to be detected'pre＝[xT-k,xT-k+1,…,xT]Inputting the data into the model trained in the third step, and calculating to obtain the state Y of predicting the degradation trend of the aircraft engine at the moment T +1pre＝xT+1；
Compared with the prior art, the invention has the following advantages:
1. the invention provides a basic unit block for sequence reconstruction based on a one-dimensional convolutional neural network, wherein the basic unit block not only outputs a future predicted value, but also outputs a reconstructed sequence for predicting an output sequence example again, and fully considers the time accumulation effect and effectively excavates information in a degraded sequence.
2. The invention provides a sequence reconstruction-based method for predicting the degradation trend of an aeroengine of a residual stacked convolutional network. And finally, integrating a plurality of stack structures to form a final model. The method provided by the invention fully excavates time accumulated information in the degradation process, effectively avoids the problem of gradient disappearance by adopting an indirect residual sum connection mode, and improves the accuracy of model overall prediction by utilizing the idea of multi-module integration.
In conclusion, the method for predicting the degradation trend of the aero-engine based on the residual stacked convolutional network with the sequence reconstruction can better solve the problems existing in the current degradation trend prediction, and can accurately predict the degradation trend of the aero-engine in the future short term.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a block diagram of the process of the present invention
FIG. 2 is a block diagram of basic units
FIG. 3 is a diagram of a stack model
FIG. 4 is a diagram of an overall model of a residual stacking convolution network based on sequence reconstruction
FIG. 5 shows the results of different experiments
Detailed Description
The method of the present invention will be described in further detail with reference to examples.
As shown in fig. 1, the present invention comprises the steps of:
step one, data preprocessing:
to verify the validity of the proposed method, a C-MAPSS dataset published by NASA was used for validation. The FD001 dataset of the C-MAPSS dataset was used to generate a degradation trend sequence.
Step 101: let the state sequence of the whole degradation process of the engine be N, and the degradation state sequence X ═ X1,x2,…,xN]The performance degradation state of the ith sensor is denoted as xiI is 1,2, …, N, and the performance degradation state sequence is sequence data which comprehensively represents the performance degradation state in the aircraft engine degradation process;
step 102: dividing samples of the whole degradation state sequence by a sliding window method, selecting data at continuous k moments as model input data, and constructing a training sample { X) by taking data at the k +1 moment as a model predicted valuein＝[x1,x2,…,xk]；Yin＝[xk+1]},k＜N；
In practical use, taking an engine in the FD001 data set as an example, the complete performance degradation cycle includes 198 flight cycles, and 188 training data can be constructed by constructing input data according to k-10 flight cycles. The corresponding test set contains 74 flight cycles, i.e., 64 test data can be constructed.
Step two, constructing a basic unit block:
step 201: according to the formula hL＝fr(Xk·wL+bL) Establishing a linear layer of elementary blocks, processing the input data at k instants, where Xin＝[x1,x2,…,xk]For k input data, wLAnd bLWeight and offset, h, representing the linear layerLRepresenting the output of the linear layer, frActivating a function for the ReLU;
step 202: according to the formula hC＝fr(Conv1d(hL) Build up a convolutional layer of basic unit blocks, output h to the linear layerLProcessing is performed, where Conv1d denotes a one-dimensional convolution operation, hCRepresenting the output of the convolutional layer, frFor ReLU activation functions, volumesThe packed layer comprises z layers of one-dimensional convolution operation for extracting the characteristics of input data;
step 203: according to formula Xf＝fr(FCf(hC) A reconstructed sequence portion of a basic cell block, which is a portion for performing a re-prediction on input sequence data, wherein Xf＝[x1',x2',…,xk']Represents the output of the re-prediction, hCRepresenting the output of the convolutional layer, FCfDenotes a fully connected layer, frActivating a function for the ReLU;
step 204: according to formula Xb＝fr(FCb(hC) A prediction part for creating a basic unit block which is a part for predicting data at the time k + 1, wherein X isb＝[xk+1']Represents the output of the prediction at time k + 1, hCRepresenting the output of the convolutional layer, FCbDenotes a fully connected layer, frActivating a function for the ReLU;
in practical use, the basic unit blocks are as shown in fig. 2, samples first pass through a linear transform layer in each basic unit block, and the size of the output samples is changed from (10,1) to (128, 1); then, feature extraction is carried out through the one-dimensional convolutional layers, the size of a convolutional kernel arranged on each layer of one-dimensional convolutional layer is 1 multiplied by 30, the step length is 1, and finally the size of a sample output through the three layers of one-dimensional convolutional layers is changed into (41, 1); finally, two different outputs are generated by two different linear layers, where the FC1 output is the reconstructed sequence with the same size as the input sequence (10,1), and the FC2 output is the predicted value for the 11 th time instant with size (1, 1).
Step three, constructing an integral model and training:
step 301: according to the formula
Combining h basic cell blocks into a stack, where XrRepresenting the sum of residuals for each basic unit block in the stack,
is the jth basic unitOutput of a re-prediction sequence of blocks, the input of the current basic unit block in the stack being the output X of the previous basic unit block re-predictionf＝[x1',x2',…,xk']；
Step 302: according to the formula
A prediction output part of the calculation stack, wherein YsThe output of the stack is represented by,
the prediction output of the j basic unit block k +1 in the stack is represented;
step 303: according to the formula
Calculating a prediction output of an integral model consisting of m stacks, wherein
Representing the predicted output of the p-th stack, and Y representing the overall predicted output of the model;
step 304: according to the formula
Calculating a training loss of the model, wherein YtRepresenting the predicted output of the model, ytRepresenting the real output of the samples, and S represents the number of the samples; the model activation function selects a ReLU function, and the optimizer selects an Adam optimizer;
in actual use, a plurality of basic unit blocks form a stack as shown in fig. 3, the output of the current stack includes a residual sum and a prediction output, the residual sum is input into the next stack for training, the prediction output and the output of other stacks are integrated to form the output of the final model, and the overall model is as shown in fig. 4.
Fourthly, predicting the degradation trend of the aircraft engine by using the model:
step 401: let the sequence to be predicted be Xpre＝[x1,x2,…,xT]Intercepting k continuous time sequences at the end of the sequence to be predicted according to the step 102 to form a sample X 'to be detected'pre＝[xT-k,xT-k+1,…,xT]Wherein T represents the sequence length of the sample to be predicted;
step 402: sample X 'to be detected'pre＝[xT-k,xT-k+1,…,xT]Inputting the data into the model trained in the third step, and calculating to obtain the state Y of predicting the degradation trend of the aircraft engine at the moment T +1pre＝xT+1；
In the actual use process, 100 engines in an FD001 data set are used for carrying out experiments, in order to verify the effectiveness of the method, a long-time memory network (LSTM), an interpretable time sequence prediction network (N-BEATS) and a one-dimensional convolutional neural network (1D-CNN) are used for carrying out experimental comparison, and an evaluation index selects a mean absolute percentage error MAPE and a root mean square error RMSE. Figure 5 demonstrates the average accuracy and stability of the different methods on the test set in the FD001 dataset. In tables 4-3, the proposed method achieved the minimum MAPE of 15.09% and the minimum RMSE0.0653, and it can be seen from the experimental results that the proposed method has good performance in the mission of prediction of the degradation trend of the aircraft engine.
The above embodiments are only examples of the present invention, and are not intended to limit the present invention, and all simple modifications, changes and equivalent structural changes made to the above embodiments according to the technical spirit of the present invention still fall within the protection scope of the technical solution of the present invention.
Claims (1)
1. The method for predicting the degradation trend of the aero-engine based on the residual stacked convolutional network of the sequence reconstruction is characterized by comprising the following steps of:
step one, data preprocessing:
step 101: let the state sequence of the whole degradation process of the engine be N, and the degradation state sequence X ═ X1,x2,…,xN]The performance degradation state of the ith sensor is denoted as xiI ═ 1,2, …, N, sequence of performance degrading statesSequence data which comprehensively represents performance degradation state in the degradation process of the aircraft engine;
step 102: dividing samples of the whole degradation state sequence by a sliding window method, selecting data at continuous k moments as model input data, and constructing a training sample { X) by taking data at the k +1 moment as a model predicted valuein＝[x1,x2,…,xk]；Yin＝[xk+1]},k＜N；
Step two, constructing a basic unit block:
step 201: according to the formula hL＝fr(Xk·wL+bL) Establishing a linear layer of elementary blocks, processing the input data at k instants, where Xin＝[x1,x2,…,xk]For k input data, wLAnd bLWeight and offset, h, representing the linear layerLRepresenting the output of the linear layer, frActivating a function for the ReLU;
step 202: according to the formula hC＝fr(Conv1d(hL) Build up convolutional layers of basic unit blocks, output h to linear layersLProcessing is performed, where Conv1d denotes a one-dimensional convolution operation, hCRepresenting the output of the convolutional layer, frFor the ReLU activation function, the convolution layer comprises z layers of one-dimensional convolution operation together for extracting the characteristics of input data;
step 203: according to formula Xf＝fr(FCf(hC) A reconstructed sequence portion of a basic cell block, which is a portion for performing a re-prediction on input sequence data, wherein Xf＝[x1',x2',…,xk']Represents the output of the re-prediction, hCRepresenting the output of the convolutional layer, FCfDenotes the fully-connected layer, frActivating a function for the ReLU;
step 204: according to formula Xb＝fr(FCb(hC) A prediction part for creating a basic unit block which is a part for predicting data at the time k +1, wherein X isb＝[xk+1']Represents the output of the prediction at time k +1, hCRepresenting the output of the convolutional layer, FCbDenotes the fully-connected layer, frActivating a function for the ReLU;
step three, constructing an integral model and training:
step 301: according to the formula
Combining h basic cell blocks into a stack, where XrRepresenting the sum of residuals for each basic unit block in the stack,
for the output of the re-prediction sequence of the jth basic cell block, the input of the current basic cell block in the stack is the output X of the previous basic cell block re-predictionf＝[x1',x2',…,xk']；
Step 302: according to the formula
The predicted output portion of the computation stack, where YsThe output of the stack is represented by,
the prediction output of the j basic unit block k +1 in the stack is represented;
step 303: according to the formula
Calculating a prediction output of an integral model consisting of m stacks, wherein
Representing the predicted output of the p-th stack, and Y representing the overall predicted output of the model;
step 304: according to the formula
Of computational modelsLoss of training wherein YtRepresenting the predicted output of the model, ytRepresenting the real output of the samples, and S represents the number of the samples; the model activation function selects a ReLU function, and the optimizer selects an Adam optimizer;
fourthly, predicting the degradation trend of the aircraft engine by using the model:
step 401: let the sequence to be predicted be Xpre＝[x1,x2,…,xT]Intercepting k continuous time sequences at the end of the sequence to be predicted according to the step 102 to form a sample X 'to be predicted'pre＝[xT-k,xT-k+1,…,xT]Wherein T represents the sequence length of the sample to be predicted;
step 402: sample X 'to be detected'pre＝[xT-k,xT-k+1,…,xT]Inputting the state Y into the model trained in the third step, and calculating to obtain the state Y of predicting the degradation trend of the aeroengine at the moment T +1pre＝xT+1。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210223893.0A CN114741948B (en) | 2022-03-09 | 2022-03-09 | Aero-engine degradation trend prediction method based on residual stacking convolution network of sequence reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210223893.0A CN114741948B (en) | 2022-03-09 | 2022-03-09 | Aero-engine degradation trend prediction method based on residual stacking convolution network of sequence reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114741948A true CN114741948A (en) | 2022-07-12 |
| CN114741948B CN114741948B (en) | 2024-03-12 |
Family
ID=82274386
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210223893.0A Active CN114741948B (en) | 2022-03-09 | 2022-03-09 | Aero-engine degradation trend prediction method based on residual stacking convolution network of sequence reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114741948B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10852439B1 (en) * | 2020-04-30 | 2020-12-01 | Beihang University | Global ionospheric total electron content prediction system |
| CN112016251A (en) * | 2020-09-02 | 2020-12-01 | 哈尔滨工程大学 | Nuclear power device fault diagnosis method and system |
| CN112132264A (en) * | 2020-09-11 | 2020-12-25 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Regional exhaust emission prediction method and system based on space-time residual perception network |
| JP6980958B1 (en) * | 2021-06-23 | 2021-12-15 | 中国科学院西北生態環境資源研究院 | Rural area classification garbage identification method based on deep learning |
| CN114091525A (en) * | 2021-10-28 | 2022-02-25 | 国家能源集团宿迁发电有限公司 | Rolling bearing degradation trend prediction method |
-
2022
- 2022-03-09 CN CN202210223893.0A patent/CN114741948B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10852439B1 (en) * | 2020-04-30 | 2020-12-01 | Beihang University | Global ionospheric total electron content prediction system |
| CN112016251A (en) * | 2020-09-02 | 2020-12-01 | 哈尔滨工程大学 | Nuclear power device fault diagnosis method and system |
| CN112132264A (en) * | 2020-09-11 | 2020-12-25 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Regional exhaust emission prediction method and system based on space-time residual perception network |
| JP6980958B1 (en) * | 2021-06-23 | 2021-12-15 | 中国科学院西北生態環境資源研究院 | Rural area classification garbage identification method based on deep learning |
| CN114091525A (en) * | 2021-10-28 | 2022-02-25 | 国家能源集团宿迁发电有限公司 | Rolling bearing degradation trend prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 彭鸿博;蒋雄伟;: "基于相关向量机的发动机剩余寿命预测", 科学技术与工程, no. 18, 28 June 2020 (2020-06-28) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114741948B (en) | 2024-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | Data-driven remaining useful life prediction via multiple sensor signals and deep long short-term memory neural network | |
| Ke et al. | Short-term electrical load forecasting method based on stacked auto-encoding and GRU neural network | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN111340282B (en) | DA-TCN-based method and system for estimating residual service life of equipment | |
| CN113777496A (en) | Lithium ion battery residual life prediction method based on time convolution neural network | |
| CN116316591A (en) | Short-term photovoltaic power prediction method and system based on hybrid bidirectional gating cycle | |
| CN114169091A (en) | Method for establishing prediction model of residual life of engineering mechanical part and prediction method | |
| CN114707712A (en) | Method for predicting requirement of generator set spare parts | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN113673768A (en) | Wind power probability prediction model based on VMD and CNN-LSTM | |
| CN115618732A (en) | Nuclear reactor digital twin key parameter autonomous optimization data inversion method | |
| CN115017826A (en) | Method for predicting residual service life of equipment | |
| Pan et al. | A meta network pruning framework for remaining useful life prediction of rocket engine bearings with temporal distribution discrepancy | |
| CN110782083B (en) | Aero-engine standby demand prediction method based on deep Croston method | |
| CN115221793A (en) | Tunnel surrounding rock deformation prediction method and device | |
| CN113673774A (en) | Aero-engine remaining life prediction method based on self-encoder and time sequence convolution network | |
| Guo et al. | MHT: A multiscale hourglass-transformer for remaining useful life prediction of aircraft engine | |
| CN114880767B (en) | Aero-engine residual service life prediction method based on attention mechanism Dense-GRU network | |
| CN114741948A (en) | Aero-engine degradation trend prediction method based on residual stacked convolution network of sequence reconstruction | |
| CN116432359A (en) | Variable topology network tide calculation method based on meta transfer learning | |
| CN114254828B (en) | Power load prediction method based on mixed convolution feature extractor and GRU | |
| CN114336792B (en) | Power grid operation state prediction method and system | |
| CN114841063A (en) | Aero-engine residual life prediction method based on deep learning | |
| Khumprom et al. | Neural networks based feature selection approaches for prognostics of aircraft engines | |
| CN112685933B (en) | Method for predicting residual service life of roller screw pair |
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 |