NL2032669B1 - Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning - Google Patents
Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning Download PDFInfo
- Publication number
- NL2032669B1 NL2032669B1 NL2032669A NL2032669A NL2032669B1 NL 2032669 B1 NL2032669 B1 NL 2032669B1 NL 2032669 A NL2032669 A NL 2032669A NL 2032669 A NL2032669 A NL 2032669A NL 2032669 B1 NL2032669 B1 NL 2032669B1
- Authority
- NL
- Netherlands
- Prior art keywords
- isolating switch
- voltage isolating
- high voltage
- model
- mechanical
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/327—Testing of circuit interrupters, switches or circuit-breakers
- G01R31/3271—Testing of circuit interrupters, switches or circuit-breakers of high voltage or medium voltage devices
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01H—ELECTRIC SWITCHES; RELAYS; SELECTORS; EMERGENCY PROTECTIVE DEVICES
- H01H11/00—Apparatus or processes specially adapted for the manufacture of electric switches
- H01H11/0062—Testing or measuring non-electrical properties of switches, e.g. contact velocity
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01H—ELECTRIC SWITCHES; RELAYS; SELECTORS; EMERGENCY PROTECTIVE DEVICES
- H01H33/00—High-tension or heavy-current switches with arc-extinguishing or arc-preventing means
- H01H33/02—Details
- H01H33/28—Power arrangements internal to the switch for operating the driving mechanism
- H01H33/36—Power arrangements internal to the switch for operating the driving mechanism using dynamo-electric motor
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Manufacturing & Machinery (AREA)
- Control Of Electric Motors In General (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The present disclosure relates to the technical field of intelligent diagnosis for a high-voltage isolating switch, and in particular, to an intelligent method for diagnosing a 5 mechanical fault of a high-voltage isolating switch based on transfer learning. The method includes: obtaining a large number of power-time signals of a driving motor under various mechanical states of the high-voltage isolating switch by using an electromechanical co-simulation technology, training a multilayer neural network model by using the obtained simulation results as training samples, and further training last 10 several layers of the multilayer neural network by using experimental or field measurement data as small samples, to achieve optimal parameters for a single or composite fault diagnosis model. The present disclosure can solve the problem of low accuracy of an intelligent diagnosis algorithm of the high-voltage isolating switch due to insufficient training samples, and can improve the accuracy of the technology for 15 mechanical state diagnosis of the high-voltage isolating switch based on the deep neural network, thus making an intelligent diagnosis result of the high-voltage isolating switch recognized by industry.
Description
Ie
INTELLIGENT METHOD FOR DIAGNOSING MECHANICAL FAULT OF
HIGH-VOLTAGE ISOLATING SWITCH BASED ON TRANSFER LEARNING
The present disclosure relates to the technical field of intelligent diagnosis for a high-voltage isolating switch, and in particular, to an intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning.
High-voltage isolating switches are widely used as high-voltage switchgear in a power system, and the normal operation of the high-voltage isolating switches is crucial to the safety of maintenance personnel and the normal operation of a power grid.
A fault causes serious bazards to the safety of the maintenance personnel and the normal operation of the power grid. With the increasingly high land utibization £53 requirements in China, gas<insulated high-voltage switchgear (GIS) type isolating switches, with advantages such as a small occupied area and high degree of integration, are increasingly used to ensure the safety of high-voltage power transmission.
Meanwhile, mechanical faults of the high-voltage isolating switch occur frequently, severely threatening the safety of power system operations, :
In the online diagnosis for a mechanical state of a high-voltage isolating switch, the intelligent diagnosis techoology based on a motor driving power in the opening/closing process of the high-voltage isolating switch 15 one of the technologies with clear prospects at present. However, an intelligent algorithm involved in the intelligent diagnosis technology often requires a large number of training samples 10 achieve high accuracy, and it 1s difficult to obtain tens of thousands of training samples through industrial field and actual measurements, which makes it difficult for 2 mechanical state result of the high-voltage isolating switch obtained by intelligent algorithms of techniques such as a deep neural network to be recognized by industry.
In view of this, to improve the accuracy of the mechanical state diagnosis technology for the high-voltage isolating switch based on the deep neural network, an intelligent technology for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning is proposed.
I
An objective of the present disclosure is to provide an intelligent method for diagnosing a mechanical fault of a high-voliage isolating switch based on transfer learning, to solve the problems mentioned in the background.
To solve the foregoing technical problems, one objective of the present disclosure is to provide an intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning, which includes the following steps:
St: constructing a physical high-vollage isolating switch model by using engineering drawing software, performing fine modeling on parts related to a kinetic {U analysis process, and performing equivalent simplification on parts unrelated to the kinetic analysis process; 82 constructing an asynchronous motor model by using power simulation software, and setting motor parameters as actoal operating parameters of a driving motor of the high~voltage isolating switch; £5 83: debugging an glectromechanical co-simuiation model by using co-sumplation modules of the two pisces of software in steps 81 and 82;
Sá: setting various mechanical states in the co~simulation model established in step
S3, performing repeated simulations and obiaining sample results of a power-time signal of the driving motor;
SS: training a deep neural network model by using a large number of samples obtained in step 34,
Só: obtaining finite samples of the power-time signal of the driving motor under different mechanical states of the high-voltage isolating switch through experiments or held measurements; and
ST: training, by using small samples obtained in step 86, last several layers of the model trained in step S5, to obtain a high-accuracy intelligent model for mechanical state diagnosis of the high-voltage isolating switch.
As a further unprovement of the technical solution, In step 81, the physical high-voltage isolating switch model is constructed through a modeling function of commercial engineering drawing software based on a fine ratio of 1:1,
As a further improvement of the technical solution, in step S2, the asynchronous
<3 - motor model is established through a modeling fonction of commercial power simulation software based on a fine ratio of 1:1.
As a further improvement of the technical solution, in step S3, the physical high-voltage isolating switch model and the asynchronous motor model are jointed through co-simulation interfaces of the two pieces of software.
As a further improvement of the technical solution, in step S4, simulated mechanical states of the high-voltage isolating switch are set in the co-simulation model through a constraint condition, and the simulated mechanical states include, but are not limited to, opening, closing, reversing (operation, standby, and maintenance) {0 and various pulling states.
As a further improvement of the technical solution, in step S6, a specific method for obtaining the finite samples of the power-fime signal of the driving motor under different mechanical states of the high-voltage isolating switch through experiments or field measurements includes the following steps: : £3 measuring, by building a high-voltage isolating switch experimental platform or in a transformer substation, a power-time curve of the driving motor during the opening/closing process of the high-voltage isolating switch; and sampling mechanical states of the high-voltage isolating switch during a maintenance process, to obtain small samples of the power-time curve of the driving motor during the actual opening/closing process of the high-voltage isolating switch, where the actual mechanical states of the high-voltage isolating switch correspond to the simulated mechanical states in step S4. :
As a further improvement of the technical solution, in step S7, when the last several layers of the model trained in step 55 are trained by using the small samples obtained in step S6, the number of the trained model layers depends on an accuracy requirement of the high-accuracy intelligent model for mechanical state diagnosis of the high-voltage isolating switch.
A second objective of the present disclosure is to provide a platform device for running an intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning, The device includes a processor, a memory, and a computer program stored in the memory and running on the processor, and when executing the computer program, the processor is configured to implement some steps of the foregoing intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning.
A third objective of the present disclosure is to provide a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, implements some steps of the foregoing intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning.
Compared with the prior ant, the present disclosure has the following beneficial effects:
I. In the intelligent method for diagnosing a mechanical faalt of a high-voltage isolating switch based on transfer learning, a large number of samples are obtained through electromechanical co-simulation] a deep veural network model is trained using the large mumber of samples; then, finite real sample data is obtained through experiments or field measurements, and the last few {3 layers of the deep neural network model are trained using finite samples based on transfer learning, to achieve optimal parameters for a single or composite fault diagnosis model, and solve the problem of low accuracy of the intelligent diagnosis algonthm of the high-voltage isolating switch due to insufficient trajming samples. 2. In the intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer leaming, intelligent diagnosis of a mechanical fault of the high-voltage isolating switch is implemented by training a high-accuracy intelligent model for mechanical state diagnosis of the high-voltage isolating switch, which can improve the accuracy of the technology for mechanical state diagnosis of the high-voltage isolating switch based on the deep neural network, and thus make an intelligent diagnosis result of the high-voltage isolating switch recognized by industry.
FIG. 1 is a Howchart of electromechanical co-simulation according to the present disclosure;
FIG. 2 is a schematic structural diagram of deep learning according to the present
~ 5. disclosure;
FIG. 3 is a flowchart of training based on transfer learning according to the present disclosure; and
FIG. 4 is a structural diagram of an example of an electronic computer platform device $5 according to the present disclosure.
Embodiment 1
As shown in FIG. 1 to FIG, 4, this embodiment provides an intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on wansfer {earning including the following steps:
S51: Construct a physical high-voltage isolating switch model by using engineering drawing software, perform fine modeling on parts related te a kinetic analysis process, and perform equivalent simplification on parts unrelated to the kinetic analysis process. 13 52: Construct an asynchronous motor model by using power simulation software, and set motor parameters as actual operating parameters of a driving motor of the high-voltage isolating switch. 53: Debug an electromechanical co~simulation model by using co-simulation modules of the two pieces of software in steps $1 and S2.
S4: Set various mechanical states in the vo-simulation model established in step 83, perform repeated simulations and obtain sample results of a power-time signal of the driving motor.
SS: Train a deep neural network model by using a large number of samples obtained in step 84.
S6: Obtain finite samples of the power-time signal of the driving motor under different mechanical states of the high-voltage isolating switch through experiments or field measurements. 87: Train, by using small samples obtained in step S6, last several layers of the model trained in step SS, to obtain a high-accuracy intelligent model for mechanical state diagnosis of the high-voltage isolating switch,
In this embodiment, step $1 to step 54 are an electromechanical co-sinmlation process, as shown in FIG, 1,
“=
In step Sl, because an output power of the driving motor of the high-voltage isolating switch depends directly on forque acting on a spindle of the driving motor, and the torque acting on the spindle of the motor is directly related to a mechanical state of the switch, Therefore, to analyze power variations of the driving motor of the high-voltage isolating switch under various mechanical states by simulation, it is necessary to construct a mechanical model of the high-voltage 1solating switch first.
During construction of the high-voltage isolating switch model by commercial modeling software based on a fine ratio of 1:1, fing modeling is performed on parts closely related to the mechanical analysis, and parts unrelated to the mechanical analysis are simplified or cunitted to reduce the running time.
In step 82, the power of the driving motor during an opening/closing process of the high-voltage isolating switch is not only related to the mechanical structure connected to the spindle of the motor, but also related to motor parameters. Therefore, by building a refined model of the driving motor of the high-voliage isolating switch in commercial £5 software, the actual situation can be better reproduced.
The physical high-voltage isolating switch model is constructed through a modeling function of commercial engineering drawing software based on a fine ratio of 1.1, and the asynchronous motor model is established through a modeling function of commercial power simulation software based on a fine ratio of 121.
In step 83, the physical high-voltage isolating switch model and the asynchronous motor model built tn steps SI and 52 are jointed through co-simulation interfaces of the twa pieces of software, to constitute the electromechanical simulation model of the high-voltage isolating switch.
In step $4, various mechanical states of the high-voltage isolating switch are set in the co-simulation model through a constraint condition, and a large number of sample signals of a power-time curve of the driving motor under various mechanical states are derived through electromechanical co-simudation, :
Specifically, in step $4, simulated mechanical states of the high-voltage isolating switch are set in the co-simulation model through a constraint condition, and the simulated mechanical states include, but are not limited to, opening, closing, reversing {operation, standby, and maintenance) and various pulling states.
In this embodiment, step 85 is a deep learning process, as shown in FIG. 2.
a
In step 85, a multilayer deep neural network is trained preliminarily by using the finally obtained samples of the power-iime curve of the driving motor under various mechanical states of the high-voltage isolating switch in FIG. 1, to optimize parameters of the multilayer neural network, '
In this embodiment, step 856 to step $7 are a training process based on transfer learning, as shown in FIG 3. in step S6, a power-time curve of the driving motor during the opening/closing process of the high-voltage isolating switch is measured by building a high-voltage isolating switch experimental platform or in a transformer substation; and mechanical states of the high-voltage isolating switch during a maintenance process are sampled, to obtain small samples of the power-time curve of the driving motor during the actual opening/closing process of the high-voltage isolating switch
The actual mechanical states of the high-voltage isolating switch correspond to the simulated mechanical states in step 34. £3 In step 87, the last several layers of the deep neural network mods! trained in FIG. 2 are trained by using the small sample data obtained in step 86, to obtain the high-aceuragy intelligent model for mechanical state diagnosis of the high-voltage isolating switch.
In this case, in step 87, the number of the trained model layers depends on an accuracy requirement of the high-accuracy yutelligent model for mechanical state diagnosis of the high-voltage isolating switch,
As shown in FIG. 4, this embodiment further provides a platform device for running an intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning. The device includes a processor, a memory, and a computer program stored in the memory and running on the processor.
The processor includes one or more processing cores. The processor is connected to the memory through a bus, The memory is configured to store program instructions.
The processor executes the program instructions in the memory to implement some steps of the foregoing intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learaing
Optionally, the memory may be implemented by any type of volatile or non-volatile storage device or a combination thereof, for example, a static random og. access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disc. 3 In addition, the present disclosure further provides a computer-readable storage mediom, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, implements some steps of the foregoing intelligent method for diagnosing a mechanical fault of a high-voltage isolating switch based on transfer learning. 19 Optionally, the present disclosure further provides a computer program product including instructions. When run on a computer, the computer program product causes the computer to perform some steps of the foregoing intelligent method for diagnosing a mechanical fault of a high-voliage isolating switch based on transfer learning.
Those of ordinary skill in the art can understand that all or some of the steps in the is foregoing embodiments may be implemented by hardware, or by instructing related hardware by using a program. The program may be stored in a computer-readable storage medium, The storage medium may be a read-only memory, a magnetic disk, an optical dier, or the like.
Claims (1)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
NL2032669A NL2032669B1 (en) | 2022-08-03 | 2022-08-03 | Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
NL2032669A NL2032669B1 (en) | 2022-08-03 | 2022-08-03 | Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning |
Publications (2)
Publication Number | Publication Date |
---|---|
NL2032669A NL2032669A (en) | 2022-09-20 |
NL2032669B1 true NL2032669B1 (en) | 2023-05-23 |
Family
ID=83744620
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
NL2032669A NL2032669B1 (en) | 2022-08-03 | 2022-08-03 | Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning |
Country Status (1)
Country | Link |
---|---|
NL (1) | NL2032669B1 (en) |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110084148A (en) * | 2019-04-09 | 2019-08-02 | 东南大学 | A kind of Mechanical Failure of HV Circuit Breaker diagnostic method |
CN110068759B (en) * | 2019-05-22 | 2021-11-09 | 华雁智能科技(集团)股份有限公司 | Fault type obtaining method and device |
-
2022
- 2022-08-03 NL NL2032669A patent/NL2032669B1/en active
Also Published As
Publication number | Publication date |
---|---|
NL2032669A (en) | 2022-09-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021232655A1 (en) | Method for evaluating mechanical state of high-voltage shunt reactor on the basis of vibration feature | |
CN112051481B (en) | Alternating current-direct current hybrid power grid fault area diagnosis method and system based on LSTM | |
CN103983453B (en) | A kind of executing agency of aero-engine and the differentiating method of sensor fault diagnosis | |
CN114065667B (en) | SF6 equipment gas pressure prediction method based on Prophet-LSTM model | |
CN109324601A (en) | The test platform of robot controller or control system based on hardware in loop | |
CN107944193A (en) | Avionics semi-matter simulating system | |
CN116610998A (en) | Switch cabinet fault diagnosis method and system based on multi-mode data fusion | |
US20210364481A1 (en) | Method for evaluating mechanical state of high-voltage shunt reactor based on vibration characteristics | |
NL2032669B1 (en) | Intelligent method for diagnosing mechanical fault of high-voltage isolating switch based on transfer learning | |
CN112462208A (en) | Multi-parameter-based direct current cable insulation diagnosis and performance test system | |
CN115392118A (en) | Debugging system for intelligent diagnosis of mechanical fault of high-voltage isolating switch | |
CN111953657B (en) | Sequence-data joint driven CPS network attack identification method for power distribution network | |
CN108233356B (en) | Consistency evaluation method and evaluation platform for photovoltaic inverter controller | |
CN117192369A (en) | Traction motor monitoring and diagnosing method based on digital twin technology | |
CN111458634A (en) | Handheld power equipment relay on-line tester and testing method | |
CN116307773A (en) | Reliability estimation method for secondary equipment of transformer substation | |
CN103197162A (en) | Multiple switch parameter fault diagnosis method based on template matching | |
CN115409059A (en) | Online fault early warning method for high-voltage electronic switch equipment and terminal equipment | |
CN114487643A (en) | On-spot handing-over of extra-high voltage GIL equipment is accepted and is synthesized test platform | |
Niang et al. | Formal Verification for Validation of PSEEL's PLC Program | |
CN111474905B (en) | Parameter drift fault diagnosis method in manufacturing process of electromechanical product | |
CN113848410A (en) | Detection method and device of control and protection device | |
Winkel et al. | Run-to-failure relay dataset for predictive maintenance research with machine learning | |
CN108107294B (en) | Microcomputer relay protection test system based on measurement modularization | |
CN111506998B (en) | Method for constructing parameter drift fault feature sample library in manufacturing process of electromechanical product |