CN109614647A - A kind of high-speed rail trailer system failure prediction method based on Bayesian network - Google Patents
A kind of high-speed rail trailer system failure prediction method based on Bayesian network Download PDFInfo
- Publication number
- CN109614647A CN109614647A CN201811330679.5A CN201811330679A CN109614647A CN 109614647 A CN109614647 A CN 109614647A CN 201811330679 A CN201811330679 A CN 201811330679A CN 109614647 A CN109614647 A CN 109614647A
- Authority
- CN
- China
- Prior art keywords
- bayesian network
- feature
- node
- closed loop
- value
- 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
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
-
- 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"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention discloses a kind of the high-speed rail trailer system failure prediction method based on Bayesian network, step are as follows: 1, build closed loop Bond Graph Model: Direct Torque Control module is added in original Bond Graph Model；2, it builds Bayesian network: first determining the node of bayesian network structure, the causality between node provided according to closed loop Bond Graph Model determines the structure of Bayesian network, finally opens the closed loop in Bayesian network；3, data prediction: being first filtered denoising to original signal, then carry out feature extraction, then carries out monotonicity screening and carries out further feature extraction, finally carries out k-means cluster to feature；4, failure predication: supervised learning is carried out with EM algorithm, then obtains the state of on-line measurement value with above-mentioned data preprocessing method, is classified using Bayesian network, equipment remaining life range is obtained.Equipment fault prediction may be implemented in the present invention, has more practicability and reliability.
Description
Technical field
The invention belongs to equipment fault electric powder prediction, in particular to a kind of high railway traction based on Bayesian network
System failure prediction technique.
Background technique
Failure predication is use state currently to equip as starting point, in conjunction with architectural characteristic, the ginseng of known prediction object
Number, environmental condition and historical data, to equipment, following failure is predicted, analyzed and is judged.Failure predication has very big
Uncertainty because itself is a random processes for object outages mechanism, and predicts that process itself can also generate error.
Since China's high-speed rail range of operation is very big, weather conditions complicated and changeable increase the complexity of its working environment, further plus
The big uncertainty of failure predication, brings great challenge for failure predication.
Bayesian network is one of the most effective model of current processing uncertain problem, is artificial intelligence, probability reason
The product combined by, graph theory and Analysis of Policy Making.It intuitively expresses the joint probability point of each variable in a manner of patterned
Cloth, and conditional independence assumption is utilized, probability inference calculation amount is greatly reduced, is mentioned for complicated uncertain inference problem
Good solution is supplied.Bond Graph Model accurately can clearly describe the causality between each variable, be conducive to pattra leaves
This network topology structure is built, and keeps the prediction result of Bayesian network more accurate.The Bond Graph Model of open cycle system at present
It has built and has finished, but trailer system will receive the interference of many environmental factors in actual moving process, open cycle system can become
It is extremely unstable, so closed-loop control link is essential in real system.To make the physical model built and subsequent
Result of study more be actually consistent, the present invention joined closed-loop control link on the basis of original open loop models, forefathers'
On the basis of further sophisticated model.
Summary of the invention
Goal of the invention: it is directed to problem of the prior art, the causal closed-loop system of trailer system can be established by providing one kind
Model, the high-speed rail trailer system failure prediction method based on Bayesian network for realizing equipment fault prediction.
Technical solution: in order to solve the above technical problems, the present invention provides a kind of high railway traction system based on Bayesian network
System failure prediction method, includes the following steps:
(1) it builds closed loop Bond Graph Model: adding Direct Torque Control module in original Bond Graph Model；
(2) it builds Bayesian network: bayesian network structure is determined by the physical quantity measured in practical high-speed rail trailer system
Node, the causality between node provided according to closed loop Bond Graph Model determines the structure of Bayesian network, finally beats
The closed loop in Bayesian network is opened, the causality conflict between node is avoided；Under causality between its interior joint
Literary abbreviation causality；
(3) data prediction: denoising first is filtered to original signal, then carries out feature extraction, then carries out monotonicity
It screens and carries out further feature extraction, k-means cluster finally is carried out to feature；
(4) failure predication: using the state of measured value as the input of Bayesian network, the corresponding three classes of feature are remaining
Output of the life span as Bayesian network carries out supervised learning with EM algorithm, then is obtained with above-mentioned data preprocessing method
To the state of on-line measurement value, classified using Bayesian network, obtains equipment remaining life range.
Further, closed loop Bond Graph Model is built in the step (1) specific step is as follows:
(1.1) difference of rotary speed setting value and system revolving speed is subjected to PI adjusting, obtains torque instruction value T*；It is true to turn
Square T is calculated by following formula:
Wherein T is electromagnetic torque, npFor number of pole-pairs,For rotor flux, irFor rotor current, α, β represent motor in α-
Two-phase under β-o coordinate system；Defined variable HT, to indicate the size relation of torque calculation value Yu its given value:
Wherein εTFor the direct torque error of permission；
(1.2) instruction value of stator magnetic linkage amplitude is set | ψs *|；True magnetic linkage amplitude | ψs| it is calculated by following formula:
Wherein | ψsα| and | ψsβ| it is the magnetic linkage amplitude of motor α, β phase；Defined variable Hψ, indicate magnetic linkage amplitude calculated value with
The size relation of its given value:
Wherein εψError is controlled for the magnetic linkage amplitude of permission；
(1.3) plane space is divided into 6 sector S16；To ψsαAnd ψsβSymbol carry out Taxonomic discussion, in conjunction with
Arctan function acquires stator magnetic linkage phase theta；Magnetic linkage is divided to corresponding sector according to phase theta；
(1.4) H acquired according to optimized switching table and step (1.1) (1.2) (1.3)T、HψAnd sector value, to inverter
Control instruction is issued, corresponding IGBT switching tube is connected.
Further, Bayesian network is built in the step (2) specific step is as follows:
(2.1) by the node of practical high-speed rail trailer system surveyed physical quantity and determine bayesian network structure, including motor
Voltage, torque, revolving speed, voltage and lower voltage on intermediate circuit, amount to 5 nodes；
(2.2) it is drawn and is directed toward according to the cause and effect of closed loop Bond Graph Model, determine the causality between Bayesian network node,
Thereby determine that bayesian network structure；
(2.3) to avoid causality conflict, the closed loop in Bayesian network need to be opened.
Further, specific step is as follows for data prediction in the step (3):
(3.1) to initial data filtering and noise reduction, frequency domain character is extracted with wavelet transformation, and calculate its temporal signatures, including
Mentioned feature normalization is removed dimension impact by mean value, very poor, variance, standard deviation, the degree of bias, kurtosis；
(3.2) will treated data input-bound Boltzmann machine, further extract further feature；
(3.3) it is individually clustered with feature of the k-means algorithm to 5 physical quantitys in step (2.1), obtains Bayes
The state of each node in network, input data when as training；Again with k-means to all features of 5 physical quantitys together
Feature is divided into long-term, mid-term, short-term three classes according to remaining life by cluster, and the training label as data is Bayesian network
The output data of network.
Further, specific step is as follows for failure predication in the step (4):
(4.1) it by the input data and the corresponding node of output data input Bayesian network in step (3), is calculated with EM
Method carries out supervised learning；
(4.2) online data is handled with step (3), obtains the state of current measurement value；Each node state is inputted into pattra leaves
This network is classified, and equipment remaining life range is obtained.
Compared with the prior art, the advantages of the present invention are as follows:
(1) closed loop control module is added in para-linkage graph model of the present invention, more meets the actual conditions of high-speed rail trailer system,
Result of study based on the closed loop model has higher practicability and reliability；
(2) present invention derives bayesian network structure by the causality that Bond Graph Model provides, compared to based on data
Derivation method, result is more acurrate, is not influenced by data volume size and data reliability；
(3) for traditional Bayesian network mostly using original signal as input data, the present invention extracts time-frequency to original signal
Characteristic of field, and further feature further is extracted with limited Boltzmann machine, the deep information of data is excavated, is more clear straight
The degenerative process for showing equipment seen, greatly improves the learning effect of Bayesian network；
(4) remaining life of prediction is divided into three ranges by the present invention, and certain redundancy is provided for plant maintenance
Degree, reduces the operation difficulty and maintenance cost of enterprise.
Detailed description of the invention
Fig. 1 is flow chart of the method for the present invention；
Fig. 2 is direct torque control theory block diagram in specific embodiment；
Fig. 3 is that voltage vector distribution and sector divide schematic diagram in specific embodiment；
Fig. 4 is inverter equivalent circuit diagram in specific embodiment；
Fig. 5 is closed loop Bond Graph Model structure chart in specific embodiment；
Fig. 6 is bond graph causality schematic diagram in specific embodiment；
Fig. 7 is Bayesian network loop abbreviation schematic diagram in specific embodiment；
Fig. 8 is bayesian network structure figure in specific embodiment.
Specific embodiment
With reference to the accompanying drawings and detailed description, the present invention is furture elucidated.Embodiments described herein are only
A part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
Personnel's obtained other embodiments without making creative work belong to the range that the present invention is protected.
Shown in referring to Fig.1, a kind of high-speed rail trailer system failure prediction method based on Bayesian network of the invention is first
The high-speed rail trailer system Bond Graph Model for first building closed loop builds Bayesian network according to the causality that Bond Graph Model provides
Network structure, then further feature extraction is carried out to original signal, with k-means by its according to remaining life be divided into long-term, mid-term,
Short-term three classes.Online data is processed similarly, is classified with Bayesian network, realizes the predicting residual useful life of equipment.
Whole process includes the following steps:
1) closed loop Bond Graph Model is built:
The bond graph open loop models of high-speed rail trailer system include inverter, motor and gear-box, and wherein inverter is motor
Power is provided.The present invention is controlled inverse according to the output signal of motor and the adjusting parameter of setting using Direct Torque Control algorithm
Become the IGBT switching tube of device, to realize the closed-loop control to motor.Fig. 2 is direct torque control theory block diagram, specific steps
It is as follows:
11) difference of rotary speed setting value and system revolving speed is subjected to PI adjusting, obtains torque instruction value T*；True torque T
It is calculated by formula (1):
Wherein T is electromagnetic torque, npFor number of pole-pairs,For rotor flux, irFor rotor current, α, β represent motor in α-
Two-phase under β-o coordinate system.Defined variable HT, to indicate the size relation of torque calculation value Yu its given value:
Wherein εTFor the direct torque error of permission.
12) instruction value of stator magnetic linkage amplitude is set | ψs *|；True magnetic linkage amplitude | ψs| it is calculated by formula (3):
Wherein | ψsα| and | ψsβ| it is the magnetic linkage amplitude of motor α, β phase.Defined variable Hψ, to indicate magnetic linkage amplitude calculated value
With the size relation of its given value:
Wherein εψError is controlled for the magnetic linkage amplitude of permission.
13) plane space is divided into 6 sector S16, to ψsαAnd ψsβSymbol carry out Taxonomic discussion, in conjunction with arctan
Function acquires stator magnetic linkage phase theta:
Magnetic linkage is divided to corresponding sector according to phase theta.
14) H acquired according to optimized switching table and step (11) (12) (13)T、HψAnd sector value, inverter is issued and is controlled
System instruction, is connected corresponding IGBT switching tube.
1 optimized switching table of table
In table 1, V16For 6 kinds of voltage vectors of inverter output, the relative space position of voltage vector and sector distribution is such as
Shown in Fig. 3.6 voltage vectors correspond to 6 kinds of working conditions of inverter, and the switching tube be connected in every kind of working condition is successively
Are as follows: T1T2T3 → T2T3T4 → T3T4T5 → T4T5T6 → T5T6T1 → T6T1T2.Switching tube distribution number is as shown in Figure 4.Figure
5 be the high-speed rail trailer system closed loop Bond Graph Model with Direct Torque Control module.
2) bayesian network structure is built:
Bayesian network structure belongs to directed acyclic graph, is made of node and side with the arrow, its physical node of arrow is
Cause, direction node are fruit, illustrate the causality between node.The cause and effect indicated between variable is drawn in Bond Graph Model with cause and effect
Relationship, in Fig. 6, A and B are two elements in bond graph, and between the two with key connection, cause and effect, which is drawn, to be used positioned at key end simultaneously
It is indicated perpendicular to the short-term of key.Cause and effect draws the guide direction for illustrating gesture signal, and for A, it is fruit that stream, which is because of gesture,；To B
Speech, it is fruit that gesture, which is because of stream,.Causality between variable is derived by closed loop Bond Graph Model, obtains bayesian network structure.Finally
To avoid causality conflict, the closed loop in Bayesian network is opened.Specific step is as follows:
21) by the node of practical high-speed rail trailer system surveyed physical quantity and determine bayesian network structure, including torque T,
Voltage Ucd1 and lower voltage Ucd2 on revolving speed w, intermediate circuit amounts to 4 nodes.
22) causality being directed toward between determining Bayesian network node is drawn according to the cause and effect of closed loop Bond Graph Model, thus
Determine bayesian network structure.
23) to avoid causality conflict, the closed loop in Bayesian network need to be opened.By taking Fig. 7 as an example, node
1,2,3 be closed loop configuration, and open here node 2 is directed toward the side of node 3, and replica node 2 is node 2 ', and node 2 ' is directed toward
Node 3 is as node 3 because thus unlocking the contradiction of 3 reciprocal causation of node 2 and node.
24) regard remaining life node life as equipment degree of degeneration, therefore node life is that other nodes is caused to change
The node is added to Bayesian network and as because of node by reason.Fig. 8 is final bayesian network structure figure.
3) data prediction:
Original signal is denoised with traditional method, feature extraction, after standardization, the present invention is on this basis
Further depth extraction is carried out to the feature of extraction with limited Boltzmann machine again, feature is used for Bayes by treated
Network training.Specific step is as follows:
31) to initial data filtering and noise reduction, frequency domain character is extracted with wavelet transformation, and calculate its temporal signatures, including equal
Value, very poor, variance, standard deviation, the degree of bias, kurtosis.It chooses the preferable feature of monotonicity and it is standardized, remove dimension
It influences.
32) will treated data input-bound Boltzmann machine, further extract further feature.
33) it is individually clustered with feature of the k-means algorithm to 4 physical quantitys chosen in step 2, obtains Bayesian network
The three state of each node in network, input data when as training.Again with k-means to all features one of 4 physical quantitys
Cluster is played, feature is divided into long-term, mid-term, short-term three classes according to remaining life, the training label as data is Bayes
The output data of network.
4) failure prediction method:
Online data is extracted into feature according to step 3, and obtains the state of each node with k-means.Node state is defeated
Enter trained Bayesian network, i.e. predictable equipment remaining life range (long-term, mid-term, short-term).Specific step is as follows:
41) by the input data and the corresponding node of output data input Bayesian network in step 3, with EM algorithm into
Row supervised learning.
42) online data is handled with step 3, obtains the state of current measurement value.Each node state is inputted into Bayesian network
Network is classified, and equipment remaining life range is obtained.
Claims (5)
1. a kind of high-speed rail trailer system failure prediction method based on Bayesian network, which comprises the steps of:
(1) it builds closed loop Bond Graph Model: adding Direct Torque Control module in original Bond Graph Model；
(2) it builds Bayesian network: determining the section of bayesian network structure by the physical quantity measured in practical high-speed rail trailer system
Point, the causality between node provided according to closed loop Bond Graph Model determine the structure of Bayesian network, finally open shellfish
Closed loop in this network of leaf avoids the causality conflict between node；
(3) data prediction: being first filtered denoising to original signal, then carry out feature extraction, then carries out monotonicity screening
And further feature extraction is carried out, k-means cluster finally is carried out to feature；
(4) failure predication: using the state of measured value as the input of Bayesian network, by the corresponding three classes remaining life model of feature
The output as Bayesian network is enclosed, carries out supervised learning with EM algorithm, then obtained online with above-mentioned data preprocessing method
The state of measured value, is classified using Bayesian network, obtains equipment remaining life range.
2. a kind of high-speed rail trailer system failure prediction method based on Bayesian network according to claim 1, feature
It is, building closed loop Bond Graph Model in the step (1), specific step is as follows:
(1.1) difference of rotary speed setting value and system revolving speed is subjected to PI adjusting, obtains torque instruction value T*；True torque T by
Following formula is calculated:
Wherein T is electromagnetic torque, npFor number of pole-pairs,For rotor flux, irFor rotor current, α, β represent motor and sit in alpha-beta-o
Two-phase under mark system；Defined variable HT, to indicate the size relation of torque calculation value Yu its given value:
Wherein εTFor the direct torque error of permission；
(1.2) instruction value of stator magnetic linkage amplitude is set | ψs *|；True magnetic linkage amplitude | ψs| it is calculated by following formula:
Wherein | ψsα| and | ψsβ| it is the magnetic linkage amplitude of motor α, β phase；Defined variable Hψ, indicate that magnetic linkage amplitude calculated value is given with it
The size relation of value:
Wherein εψError is controlled for the magnetic linkage amplitude of permission；
(1.3) plane space is divided into 6 sector S16；To ψsαAnd ψsβSymbol carry out Taxonomic discussion, in conjunction with arctan letter
Number, acquires stator magnetic linkage phase theta；Magnetic linkage is divided to corresponding sector according to phase theta；
(1.4) H acquired according to optimized switching table and step (1.1) (1.2) (1.3)T、HψAnd sector value, inverter is issued and is controlled
System instruction, is connected corresponding IGBT switching tube.
3. a kind of high-speed rail trailer system failure prediction method based on Bayesian network according to claim 1, feature
It is, building Bayesian network in the step (2), specific step is as follows:
(2.1) by the node of practical high-speed rail trailer system surveyed physical quantity and determine bayesian network structure, including electric moter voltage,
Torque, revolving speed, voltage and lower voltage on intermediate circuit, amount to 5 nodes；
(2.2) it is drawn and is directed toward according to the cause and effect of closed loop Bond Graph Model, determine the causality between Bayesian network node, thus really
Determine bayesian network structure；
(2.3) to avoid causality conflict, the closed loop in Bayesian network need to be opened.
4. a kind of high-speed rail trailer system failure prediction method based on Bayesian network according to claim 3, feature
It is, specific step is as follows for data prediction in the step (3):
(3.1) to initial data filtering and noise reduction, frequency domain character is extracted with wavelet transformation, and calculate its temporal signatures, including mean value,
Mentioned feature normalization is removed dimension impact by very poor, variance, standard deviation, the degree of bias, kurtosis；
(3.2) will treated data input-bound Boltzmann machine, further extract further feature；
(3.3) it is individually clustered, is obtained in Bayesian network with feature of the k-means algorithm to 5 physical quantitys in step (2.1)
The state of each node, input data when as training；It is clustered, is pressed together with all features of the k-means to 5 physical quantitys again
Feature is divided into long-term, mid-term, short-term three classes according to remaining life, the training label as data is the output of Bayesian network
Data.
5. a kind of high-speed rail trailer system failure prediction method based on Bayesian network according to claim 4, feature
It is, specific step is as follows for failure predication in the step (4):
(4.1) it by the input data and the corresponding node of output data input Bayesian network in step (3), is carried out with EM algorithm
Supervised learning；
(4.2) online data is handled with step (3), obtains the state of current measurement value；Each node state is inputted into Bayesian network
Network is classified, and equipment remaining life range is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811330679.5A CN109614647B (en) | 2018-11-09 | 2018-11-09 | Bayesian network-based high-speed rail traction system fault prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811330679.5A CN109614647B (en) | 2018-11-09 | 2018-11-09 | Bayesian network-based high-speed rail traction system fault prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109614647A true CN109614647A (en) | 2019-04-12 |
| CN109614647B CN109614647B (en) | 2023-05-26 |
Family
ID=66003702
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811330679.5A Active CN109614647B (en) | 2018-11-09 | 2018-11-09 | Bayesian network-based high-speed rail traction system fault prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109614647B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110503217A (en) * | 2019-08-29 | 2019-11-26 | 新誉轨道交通科技有限公司 | A kind of slow leakage failure prediction technique of air conditioner coolant, device, equipment and system |
| CN110866315A (en) * | 2019-11-20 | 2020-03-06 | 重庆大学 | Electric drive system multi-field coupling optimization method based on bond diagram modeling |
| CN110866316A (en) * | 2019-11-20 | 2020-03-06 | 重庆大学 | Speed reducer bonding diagram model optimization method based on six-degree-of-freedom gear meshing model |
| CN112036051A (en) * | 2020-11-05 | 2020-12-04 | 中国人民解放军国防科技大学 | Method, device, equipment and medium for predicting residual service life of magnetic suspension system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106124175A (en) * | 2016-06-14 | 2016-11-16 | 电子科技大学 | A kind of compressor valve method for diagnosing faults based on Bayesian network |
| CN108285071A (en) * | 2018-01-25 | 2018-07-17 | 暨南大学 | A kind of elevator Gernral Check-up method based on Bayesian network |
-
2018
- 2018-11-09 CN CN201811330679.5A patent/CN109614647B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106124175A (en) * | 2016-06-14 | 2016-11-16 | 电子科技大学 | A kind of compressor valve method for diagnosing faults based on Bayesian network |
| CN108285071A (en) * | 2018-01-25 | 2018-07-17 | 暨南大学 | A kind of elevator Gernral Check-up method based on Bayesian network |
Non-Patent Citations (2)
| Title |
|---|
| 姜斌等: "高速列车牵引***故障诊断与预测技术综述", 《控制与决策》 * |
| 张歆炀等: "基于故障树与键合图的贝叶斯网络故障诊断", 《电测与仪表》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110503217A (en) * | 2019-08-29 | 2019-11-26 | 新誉轨道交通科技有限公司 | A kind of slow leakage failure prediction technique of air conditioner coolant, device, equipment and system |
| CN110866315A (en) * | 2019-11-20 | 2020-03-06 | 重庆大学 | Electric drive system multi-field coupling optimization method based on bond diagram modeling |
| CN110866316A (en) * | 2019-11-20 | 2020-03-06 | 重庆大学 | Speed reducer bonding diagram model optimization method based on six-degree-of-freedom gear meshing model |
| CN112036051A (en) * | 2020-11-05 | 2020-12-04 | 中国人民解放军国防科技大学 | Method, device, equipment and medium for predicting residual service life of magnetic suspension system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109614647B (en) | 2023-05-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109614647A (en) | A kind of high-speed rail trailer system failure prediction method based on Bayesian network | |
| CN109102005B (en) | Small sample deep learning method based on shallow model knowledge migration | |
| CN109635928B (en) | Voltage sag reason identification method based on deep learning model fusion | |
| CN109297689B (en) | Large-scale hydraulic machinery intelligent diagnosis method introducing weight factors | |
| CN109633369B (en) | Power grid fault diagnosis method based on multi-dimensional data similarity matching | |
| CN111427266B (en) | Nonlinear system identification method aiming at disturbance | |
| Gouriveau et al. | Strategies to face imbalanced and unlabelled data in PHM applications. | |
| Razavi-Far et al. | A multiple observers and dynamic weighting ensembles scheme for diagnosing new class faults in wind turbines | |
| CN111795819B (en) | Gear box fault diagnosis method integrating vibration and current signal collaborative learning | |
| CN113219328B (en) | Intelligent fault diagnosis method for circuit breaker operating mechanism based on information fusion | |
| Kezunovic | Translational knowledge: From collecting data to making decisions in a smart grid | |
| CN114034486A (en) | Unsupervised transfer learning-based bearing fault diagnosis method for pump mechanical equipment | |
| Xue et al. | LSTM-based intelligent fault detection for fuzzy Markov jump systems and its application to tunnel diode circuits | |
| Mansouri et al. | Reduced Gaussian process regression based random forest approach for fault diagnosis of wind energy conversion systems | |
| CN110221167B (en) | Power system transmission line short-circuit fault diagnosis method based on determination learning | |
| Zhang et al. | Directed-graph-learning-based diagnosis of multiple faults for high speed train with switched dynamics | |
| Guo et al. | An equipment multiple failure causes intelligent identification method based on integrated strategy for subway sliding plug door system under variable working condition | |
| CN110779988A (en) | Bolt life prediction method based on deep learning | |
| Eissa et al. | Observer-based fault detection approach using fuzzy adaptive poles placement system with real-time implementation | |
| Bessam et al. | A novel method for induction motors stator inter-turn short circuit fault diagnosis based on wavelet energy and neural network | |
| Bi et al. | Research on Fault Diagnosis for Pumping Station Based on T‐S Fuzzy Fault Tree and Bayesian Network | |
| CN116304649A (en) | Motor fault signal feature extraction method, system, computer and storage medium | |
| Tojeiro et al. | Fault detection based on Neuro-Fuzzy models and residual evaluation with fuzzy thresholds applied to a photovoltaic system. | |
| Opara | Information theoretic state estimation in power systems | |
| Qian et al. | Identification of conductive leakage signal in power cable based on multi-classification PSO-SVM |
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 |