CN110850837A - System life analysis and fault diagnosis method based on long-time and short-time memory neural network - Google Patents
System life analysis and fault diagnosis method based on long-time and short-time memory neural network Download PDFInfo
- Publication number
- CN110850837A CN110850837A CN201810947872.7A CN201810947872A CN110850837A CN 110850837 A CN110850837 A CN 110850837A CN 201810947872 A CN201810947872 A CN 201810947872A CN 110850837 A CN110850837 A CN 110850837A
- Authority
- CN
- China
- Prior art keywords
- neural network
- time
- long
- training
- fault diagnosis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Abstract
The invention provides a system life analysis and fault diagnosis method based on a long-time and short-time memory neural network, which comprises the steps of selecting environment variables, operation variables and sensor data which have important influence on system work to carry out data acquisition and data preprocessing; extracting the data characteristics by using a characteristic engineering means; training a long-term memory neural network by using the originally acquired data and the extracted new features; and training the life prediction accuracy and the fault diagnosis accuracy of the neural network on the test sample. The method can accurately evaluate the residual life and the fault location by using the historical record and the current state of the system, has good model universality, and can be applied to various systems by less modification.
Description
Technical Field
The invention belongs to the technical field of complex system prediction and health management, and particularly relates to a Long-Short Term Memory (LSTM) neural network-based system life analysis and fault diagnosis method.
Background
With the rapid development of sensors and big data technologies, the informatization level of an industrial system is continuously improved, faults are difficult to predict, position and repair accurately and rapidly by means of traditional maintenance concepts, modes and means, and the maintenance efficiency and benefits cannot be guaranteed. In order to reduce maintenance guarantee cost and improve the economy of a system and components, an engine health management technology taking diagnosis and prediction as marks becomes the most main technical realization way, and the machine learning technology is adopted to carry out real-time state monitoring and residual life analysis on key components, so that the guarantee management capability of the whole system and the key components can be greatly improved, and the aims of reducing maintenance manpower, spare parts and maintenance cost and maximizing maintenance and component purchasing interval time are fulfilled. Machine learning techniques have become the primary support for implementing new maintenance and security modes based on performance, autonomous security, and intelligent maintenance and security.
Fig. 1 shows the complete process of the Remaining Life (Remaining Useful Life, RUL) prediction task. The generalized residual life prediction means that under a certain operating condition, aiming at a specific system object, the current health state of the system is judged in real time by monitoring main sensor data, trend analysis is carried out on the subsequent model degradation process, estimation of the residual life is given, and a decision is made in time and a corresponding action is taken according to the estimation. The broad predictive task thus includes not only the time to life estimation but also the diagnosis of the state of health. By utilizing a machine learning technology based on data driving, the latest research result in the field of artificial intelligence science and the increasingly abundant computing resources at present can be fully used for feature extraction and self-learning modeling of a complex system and design of a diagnostor and a predictor with robustness, anti-noise capability, reliability and accuracy, so that fault diagnosis and residual life estimation of the complex system are realized.
Disclosure of Invention
The invention aims to provide a system life analysis and fault diagnosis method based on a long-time and short-time memory neural network.
The technical scheme of the invention is as follows:
a system life analysis and fault diagnosis method based on a long-time and short-time memory neural network is characterized by comprising the following steps:
(1) analyzing the working mechanism of a target system to a certain extent, determining environmental variables influencing the service life of the system when the system works, all manual operation modes of the system work and physical quantities of main parts, and then setting sensors meeting detection requirements aiming at various variables to collect all variables and carry out certain data preprocessing;
(2) performing characteristic engineering processing on the acquired original characteristics, wherein the characteristic engineering processing mainly comprises principal component analysis, operation mode clustering, interframe dynamic information differential calculation under different environments/operation conditions and the like;
(3) and constructing a long-time memory neural network, and training the neural network by using the acquired data.
Further, in the method for analyzing system life and diagnosing faults based on the long-time and short-time memory neural network, the step (3) further includes:
(31) the neural network structure design comprises a network type, a cascade mode, a network layer number, a network scale and a regularization scheme;
(32) sample preparation, namely performing sequence marking of residual life and fault category marking on all collected samples, then uniformly mixing the samples and dividing the samples into a training set, a verification set and a test set according to the proportion;
(33) selecting an optimization algorithm, selecting a network initialization scheme and carrying out hyper-parameter optimization training;
(34) and (4) training the neural network model according to the optimal network structure and parameter configuration obtained in the steps (31) to (33) on the verification set, and testing the residual life accuracy and the fault diagnosis accuracy on the test set.
The method can accurately evaluate the residual life and the fault location by using the historical record and the current state of the system, has good model universality, and can be applied to various systems by less modification.
Drawings
FIG. 1 is a diagram of a remaining life prediction and fault diagnosis process;
fig. 2 is a flow chart of the implementation of the present invention.
Detailed Description
Fig. 2 is a flowchart illustrating the system life analysis and fault diagnosis based on the long-and-short-term memory neural network according to the present invention, which includes the following steps:
the method comprises the following steps: analyzing the working mechanism of a target system to a certain extent, determining environmental variables influencing the service life of the system when the system works, all manual operation modes of the system work and physical quantities of main parts, and then setting sensors meeting detection requirements aiming at various variables to collect all variables and carry out certain data preprocessing;
step two: performing characteristic engineering processing on the acquired original characteristics, wherein the characteristic engineering processing mainly comprises principal component analysis, operation mode clustering, interframe dynamic information differential calculation under different environments/operation conditions and the like;
step three: and constructing a long-time memory neural network, and training the neural network by using the acquired data. The method comprises the following specific steps:
(1) the neural network structure design comprises network types, cascade modes, network layer numbers, network scales, regularization schemes and the like;
(2) sample preparation, namely performing sequence marking of residual life and fault category marking on all collected samples, then uniformly mixing the samples and dividing the samples into a training set, a verification set (development set) and a test set according to the proportion;
(3) selecting an optimization algorithm, selecting a network initialization scheme and carrying out hyper-parameter optimization training;
(4) and (4) training a neural network model according to the optimal network structure and parameter configuration obtained on the verification set in the steps (1) to (3), and testing the residual life accuracy and the fault diagnosis accuracy on the test set.
Claims (2)
1. A system life analysis and fault diagnosis method based on a long-time and short-time memory neural network is characterized by comprising the following steps:
(1) analyzing the working mechanism of a target system to a certain extent, determining environmental variables influencing the service life of the system when the system works, all manual operation modes of the system work and physical quantities of main parts, and then setting sensors meeting detection requirements aiming at various variables to collect all variables and carry out certain data preprocessing;
(2) performing characteristic engineering processing on the acquired original characteristics, wherein the characteristic engineering processing mainly comprises principal component analysis, operation mode clustering, interframe dynamic information differential calculation under different environments/operation conditions and the like;
(3) and constructing a long-time memory neural network, and training the neural network by using the acquired data.
2. The method for analyzing the service life and diagnosing the faults of the system based on the long-time and short-time memory neural network as claimed in claim 1, wherein the step (3) further comprises the following steps:
(31) the neural network structure design comprises a network type, a cascade mode, a network layer number, a network scale and a regularization scheme;
(32) sample preparation, namely performing sequence marking of residual life and fault category marking on all collected samples, then uniformly mixing the samples and dividing the samples into a training set, a verification set and a test set according to the proportion;
(33) selecting an optimization algorithm, selecting a network initialization scheme and carrying out hyper-parameter optimization training;
(34) and (4) training the neural network model according to the optimal network structure and parameter configuration obtained in the steps (31) to (33) on the verification set, and testing the residual life accuracy and the fault diagnosis accuracy on the test set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810947872.7A CN110850837A (en) | 2018-08-20 | 2018-08-20 | System life analysis and fault diagnosis method based on long-time and short-time memory neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810947872.7A CN110850837A (en) | 2018-08-20 | 2018-08-20 | System life analysis and fault diagnosis method based on long-time and short-time memory neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110850837A true CN110850837A (en) | 2020-02-28 |
Family
ID=69594426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810947872.7A Pending CN110850837A (en) | 2018-08-20 | 2018-08-20 | System life analysis and fault diagnosis method based on long-time and short-time memory neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110850837A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111581888A (en) * | 2020-05-18 | 2020-08-25 | 中车永济电机有限公司 | Construction method of residual service life prediction model of wind turbine bearing |
| CN112395806A (en) * | 2020-11-11 | 2021-02-23 | 北京京航计算通讯研究所 | Method and device for predicting residual life of comprehensive transmission hydraulic system |
| CN113075498A (en) * | 2021-03-09 | 2021-07-06 | 华中科技大学 | Power distribution network traveling wave fault positioning method and system based on residual error clustering |
-
2018
- 2018-08-20 CN CN201810947872.7A patent/CN110850837A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111581888A (en) * | 2020-05-18 | 2020-08-25 | 中车永济电机有限公司 | Construction method of residual service life prediction model of wind turbine bearing |
| CN112395806A (en) * | 2020-11-11 | 2021-02-23 | 北京京航计算通讯研究所 | Method and device for predicting residual life of comprehensive transmission hydraulic system |
| CN112395806B (en) * | 2020-11-11 | 2021-11-09 | 北京京航计算通讯研究所 | Method and device for predicting residual life of comprehensive transmission hydraulic system |
| CN113075498A (en) * | 2021-03-09 | 2021-07-06 | 华中科技大学 | Power distribution network traveling wave fault positioning method and system based on residual error clustering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110361176B (en) | Intelligent fault diagnosis method based on multitask feature sharing neural network | |
| EP3809220A1 (en) | Method and system for semi-supervised deep anomaly detection for large-scale industrial monitoring systems based on time-series data utilizing digital twin simulation data | |
| Lindemann et al. | Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks | |
| CN116625438B (en) | Gas pipe network safety on-line monitoring system and method thereof | |
| CN111737909B (en) | Structural health monitoring data anomaly identification method based on space-time graph convolutional network | |
| CN114048600A (en) | Digital twin-driven multi-model fusion industrial system anomaly detection method | |
| CN110850837A (en) | System life analysis and fault diagnosis method based on long-time and short-time memory neural network | |
| CN105759201A (en) | High voltage circuit breaker self-diagnosis method based on abnormal sample identification | |
| Mathew et al. | Regression kernel for prognostics with support vector machines | |
| CN110757510A (en) | Method and system for predicting remaining life of robot | |
| CN116822115A (en) | Environment management method and system for intelligent park based on digital twin technology | |
| CN110320802B (en) | Complex system signal time sequence identification method based on data visualization | |
| CN117057668A (en) | Industrial pollutant emission prediction method based on deep learning model | |
| CN117216647A (en) | Predictive maintenance and fault diagnosis method and system for vehicle body welding equipment | |
| CN112257224A (en) | Method, system and terminal for overhauling state of steam turbine generator | |
| Gu et al. | Research on prediction of remaining useful life of underwater turntable based on slow feature analysis | |
| CN115600695A (en) | Fault diagnosis method of metering equipment | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN112560252A (en) | Prediction method for residual life of aircraft engine | |
| Du et al. | A method for predictive maintenance of mechanical equipment based on Transferable CNN-LSTM | |
| Zheng et al. | Research on Predicting Remaining Useful Life of Equipment Based on Health Index | |
| Hong et al. | Drainage network flow anomaly classification based on XGBoost | |
| CN111832832B (en) | District self-inspection system based on thing networking | |
| CN114330515A (en) | Bridge monitoring data abnormity diagnosis and repair method | |
| CN117407757A (en) | Intelligent fault diagnosis method and system for production line lifting appliance based on PCA-SVDD |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200228 |
|
| WD01 | Invention patent application deemed withdrawn after publication |