CN111967618A - Online diagnosis method for voltage regulator based on deep learning - Google Patents
Online diagnosis method for voltage regulator based on deep learning Download PDFInfo
- Publication number
- CN111967618A CN111967618A CN201910420646.8A CN201910420646A CN111967618A CN 111967618 A CN111967618 A CN 111967618A CN 201910420646 A CN201910420646 A CN 201910420646A CN 111967618 A CN111967618 A CN 111967618A
- Authority
- CN
- China
- Prior art keywords
- pressure
- flow
- model
- encoder
- gas
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000013135 deep learning Methods 0.000 title claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000012544 monitoring process Methods 0.000 claims abstract description 14
- 238000009530 blood pressure measurement Methods 0.000 claims abstract description 13
- 238000010586 diagram Methods 0.000 claims abstract description 10
- 239000000284 extract Substances 0.000 claims abstract description 10
- 230000005856 abnormality Effects 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 14
- 238000005259 measurement Methods 0.000 claims description 14
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000013500 data storage Methods 0.000 claims description 5
- 230000002787 reinforcement Effects 0.000 claims description 5
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000004806 packaging method and process Methods 0.000 claims description 3
- 238000007726 management method Methods 0.000 abstract description 4
- 238000013024 troubleshooting Methods 0.000 abstract description 2
- 230000005540 biological transmission Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000010295 mobile communication Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000001595 flow curve Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q10/20—Administration of product repair or maintenance
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F17—STORING OR DISTRIBUTING GASES OR LIQUIDS
- F17D—PIPE-LINE SYSTEMS; PIPE-LINES
- F17D3/00—Arrangements for supervising or controlling working operations
- F17D3/01—Arrangements for supervising or controlling working operations for controlling, signalling, or supervising the conveyance of a product
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F17—STORING OR DISTRIBUTING GASES OR LIQUIDS
- F17D—PIPE-LINE SYSTEMS; PIPE-LINES
- F17D5/00—Protection or supervision of installations
- F17D5/005—Protection or supervision of installations of gas pipelines, e.g. alarm
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Business, Economics & Management (AREA)
- Mechanical Engineering (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a pressure regulator online diagnosis method based on deep learning.A pipeline monitoring end collects a pressure measurement message, extracts and generates a pressure value variable quantity, compares the pressure value variable quantity with a preset standard pressure variable quantity, generates a task number after an abnormality occurs, and sequentially sends the task number to a fault approval end and a pressure regulator online diagnosis module. According to the invention, whether the pressure regulator has a problem is judged by collecting the pressure values of the inlet and the outlet of the gas management, the position of the problem is judged according to specific flow data in the pipeline, the system maps the fault position into the pipeline diagram and then sends the pipeline diagram to the mobile terminal for alarming, the device is convenient and intelligent to integrate, fault troubleshooting is completely completed by the system, manual operation is reduced, time and labor are saved, online diagnosis of the pressure regulator is divided into two steps of model offline training and model online updating, and higher fault diagnosis accuracy and generalization can be achieved.
Description
Technical Field
The invention relates to the technical field of voltage regulators, in particular to a voltage regulator online diagnosis method based on deep learning.
Background
With the development of economy, the demand of society on energy is increasing, the gas supply becomes an important guarantee for the rapid development of the socioeconomic performance, and the gas pressure regulator occupies a very important position in the urban gas supply system. The safe and stable operation of the gas pressure regulator is related to the normal operation of the whole supply system, and is closely related to the safety of lives and properties of downstream users. In actual use, the medium-low pressure gas pressure regulator fails and generally shows the pressure value at the inlet and the outlet of the medium-low pressure gas pressure regulator, and in order to ensure the operation safety of the medium-low pressure gas pressure regulator, the data needs to be acquired in real time and processed periodically.
Disclosure of Invention
The invention aims to: the on-line diagnosis method for the pressure regulator based on deep learning is provided for solving the problem of the fault of the gas pipeline pressure regulator.
In order to achieve the purpose, the invention adopts the following technical scheme:
a voltage regulator online diagnosis method based on deep learning comprises the following steps:
s1, respectively acquiring pressure values of the input end and the output end of the gas pipeline by the input end pressure transmitter and the output end pressure transmitter and transmitting the pressure values to the central control module;
s2, the pipeline monitoring end collects the pressure measurement message sent by the pressure measurement module to the central control module, extracts and generates a pressure value variable quantity, compares the pressure value variable quantity with a preset standard pressure variable quantity, generates a task number after an abnormality occurs, and sequentially sends the task number to the fault approval end and the overhaul position approval end;
s3, collecting flow information inside the gas pipeline by the gas sensor, packaging the flow information to generate message information, and sending the message information to the central control module;
s4, the fault approval end collects flow measurement messages sent by the gas flow measurement module to the central control module in real time, extracts the flow measurement messages to generate flow variation values, compares the flow variation values with preset standard flow variation values, generates flow comparison results and stores the flow comparison results into the data storage end;
s5, calling the flow comparison result by the position approval end, calling the equipment position of the gas sensor according to the equipment serial number, and mapping the equipment position into a gas pipeline diagram;
and S6, the information transmission terminal packs the gas management graph mapped with the fault position into an alarm message and transmits the alarm message to the user mobile equipment.
As a further description of the above technical solution:
the pressure measurement module comprises an input end pressure transmitter and an output end pressure transmitter which are respectively arranged on an input port and an output port of the gas pipeline.
As a further description of the above technical solution:
and the pipeline monitoring end compares the pressure value variation with a preset standard pressure variation, and if the comparison deviation exceeds a threshold value, a task number is generated.
As a further description of the above technical solution:
the pipeline monitoring end generates task numbers, then synchronously generates sequence tasks related to the task numbers, marks priorities for one or more sequence tasks, and sends sequence task requests to the fault approval end and the overhaul position approval end according to the priority sequence.
As a further description of the above technical solution:
the gas flow measuring module comprises a plurality of gas sensors arranged inside the gas pipeline.
As a further description of the above technical solution:
the message information contains the occurrence timestamp of the flow information and the equipment serial number for collecting the flow information.
As a further description of the above technical solution:
the mobile communication equipment comprises intelligent mobile equipment such as a mobile phone, a tablet computer and a notebook computer.
As a further description of the above technical solution:
and the fault approval end collects flow message information, sorts the flow message information according to the equipment serial number, and sequentially generates a plurality of flow change values, wherein the flow change values are gas flow difference values collected by two adjacent gas sensors.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
according to the invention, whether the pressure regulator has a problem is judged by collecting the pressure values of the inlet and the outlet of the gas management, the position of the problem is judged according to specific flow data in the pipeline, the system maps the fault position into the pipeline diagram and then sends the pipeline diagram to the mobile terminal for alarming, the device is convenient and intelligent to integrate, fault troubleshooting is completely completed by the system, manual operation is reduced, time and labor are saved, online diagnosis of the pressure regulator is divided into two steps of model offline training and model online updating, and higher fault diagnosis accuracy and generalization can be achieved.
Drawings
Fig. 1 is a schematic diagram of a system structure of a voltage regulator online diagnosis method based on deep learning according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
A pressure regulator online diagnosis system based on deep learning comprises a central control module, a pressure measurement module, a gas flow measurement module and a pressure regulator online diagnosis module;
the central control module comprises a pipeline monitoring end, a fault approval end, a position approval end, an information transmission end and a data storage end;
the pipeline monitoring end collects a pressure measurement message sent by the pressure measurement module to the central control module, extracts and generates a pressure value variable quantity, compares the pressure value variable quantity with a preset standard pressure variable quantity, generates a task number after an abnormality occurs, and sequentially sends the task number to the fault approval end and the pressure regulator online diagnosis module;
specifically, the pipeline monitoring end compares the pressure value variation with a preset standard pressure variation, and if the comparison deviation exceeds a threshold value, a task number is generated;
specifically, the pipeline monitoring end generates a task number and then synchronously generates sequence tasks related to the task number, marks priorities for one or more sequence tasks, and sends a sequence task request to the fault approval end and the overhaul position approval end according to the priority sequence;
the fault approval end acquires a flow measurement message sent by the gas flow measurement module to the central control module in real time, extracts the flow measurement message to generate a flow variation value, compares the flow variation value with a preset standard flow variation value, generates a flow comparison result and stores the flow comparison result into the data storage end;
specifically, after collecting flow message information, the fault approval end sorts the flow message information according to the equipment serial number, and sequentially generates a plurality of flow change values, wherein the flow change values are gas flow difference values collected by two adjacent gas sensors;
the position approval end calls the flow comparison result, searches and calls the equipment position of the gas sensor according to the equipment serial number, and maps the equipment position into a gas pipeline diagram;
the information transmission end packs the gas management graph mapped with the fault position into an alarm message and then transmits the alarm message to the user mobile equipment;
preferably, the mobile communication device comprises an intelligent mobile device such as a mobile phone, a tablet computer, a notebook computer and the like;
the pressure measuring module comprises an input end pressure transmitter and an output end pressure transmitter which are respectively arranged at an input port and an output port of the gas pipeline, and the input end pressure transmitter and the output end pressure transmitter respectively collect pressure values of the input end and the output end of the gas pipeline and transmit the pressure values to the central control module;
the gas flow measuring module comprises a plurality of gas sensors arranged in a gas pipeline, and the gas sensors collect flow information in the gas pipeline, package the flow information to generate message information and send the message information to the central control module;
specifically, the message information includes an occurrence timestamp of the traffic information and an equipment serial number for acquiring the traffic information;
the online diagnosis module of the voltage regulator comprises a model offline training end and a model online updating end, wherein the offline training end carries out modeling based on the existing data sample, and the model online updating end updates the model according to the rechecking result of field workers after the system is online;
the specific work flow of the off-line training end comprises the following steps:
s1, establishing a basic information base by taking the data obtained by cleaning as training samples;
s2, enhancing data on the basis of the basic information base, and adding local noise;
s3, randomly weighting and fusing the data in the basic information base according to the time period similar data;
s4, the data obtained by enhancement and the basic information base are combined to form a reinforcement learning offline model information base;
it should be noted that, in the pressure-flow curve diagram of the pressure regulator, the pressure regulator can stabilize the outlet pressure at a set value in the actual operation process, the set value of the outlet pressure can be determined by the initial compression amount of the spring of the pressure regulator, the operation condition of the pressure regulator can be inferred to a certain extent by judging the position of the stable point of the outlet pressure of the current pressure regulator, and when the stable value of the outlet pressure deviates from the set value to a large extent, the abnormality of the pressure regulator is indicated;
it should be noted that the core algorithm of the on-line diagnosis module of the pressure regulator is to obtain a time sequence characteristic model of the pressure regulator based on a deep learning method, extract an operation characteristic value of the pressure regulator, construct a pressure prediction model by using the characteristic value, give an outlet pressure stable point by the prediction model, and consider that an abnormality occurs when a predicted value is too large in difference with a true value;
the off-line training end is provided with a self-encoder, the self-encoder performs self-supervision feature extraction, the self-encoder comprises an encoder and a decoder, and the specific operation mode of the self-encoder comprises the following steps:
s1, taking the model output of the encoder as a feature extraction layer to obtain the model features of the voltage regulator;
s2, reducing the dimension of the model through sparse coding, and compressing the dimension of the model to the dimension of the model;
s3, introducing KL divergence into a loss function in the sparse coding feature extractor, wherein the KL divergence is used for representing the sparsity of the model;
s4, introducing a Loss function, setting preset parameters to represent the activity of the hidden layer unit, and adding the preset parameters into the Loss function to enable h to tend to be sparse;
it should be noted that the self-encoder simultaneously uses the original sample as the model input and the model output to achieve the goal of self-supervision learning;
it should be noted that the sparse autoencoder can be obtained by combining with the self-encoder of the sparse coding, and the sparse autoencoder can be used for extracting the model characteristics of the voltage regulator;
further, the training process of the sparse self-coding feature extractor comprises the following steps:
s1, constructing an encoder by adopting a deep neural network;
s2, constructing a decoder according to the structural symmetry of the encoder;
s3 adding KL divergence in the penalty function of the deep neural network to make the model tend to be sparse;
s4, compressing the daily pressure data and flow data to one dimension, and transmitting the one dimension as a model input and model output label to the model;
s5 iterative training model;
after the characteristics of the voltage regulator are obtained, analyzing according to the characteristics, and obtaining state classification through a clustering sensor;
it should be noted that the voltage regulator is divided into five state types: the method comprises the following steps of (1) normally running, internal leakage, transfinite, high closing and abnormal restarting pressure, a sensor can be constructed, and the current state of the voltage regulator can be obtained by threshold value demarcation and the result of the clustered sensor;
it should be noted that the structure of the sensing machine is as follows: the input layer is 1024-dimensional nodes, hidden layer dimensions are 2048-dimensional nodes, the output layer is 5-dimensional nodes, the loss function is trained by adopting a sigmoid function, the output of the 5-dimensional nodes represents the probability of five states, and the output range is 0-1;
a voltage regulator online diagnosis method based on deep learning comprises the following steps:
s1, respectively acquiring pressure values of the input end and the output end of the gas pipeline by the input end pressure transmitter and the output end pressure transmitter and transmitting the pressure values to the central control module;
specifically, the pressure measurement module comprises an input end pressure transmitter and an output end pressure transmitter which are respectively arranged at an input port and an output port of the gas pipeline;
s2, the pipeline monitoring end collects the pressure measurement message sent by the pressure measurement module to the central control module, extracts and generates a pressure value variable quantity, compares the pressure value variable quantity with a preset standard pressure variable quantity, generates a task number after an abnormality occurs, and sequentially sends the task number to the fault approval end and the pressure regulator online diagnosis module;
specifically, the pipeline monitoring end compares the pressure value variation with a preset standard pressure variation, and if the comparison deviation exceeds a threshold value, a task number is generated;
specifically, the pipeline monitoring end generates a task number and then synchronously generates sequence tasks related to the task number, marks priorities for one or more sequence tasks, and sends a sequence task request to the fault approval end and the overhaul position approval end according to the priority sequence;
s3-1, collecting flow information inside a gas pipeline by a gas sensor, packaging the flow information to generate message information, and sending the message information to a central control module;
specifically, the gas flow measuring module comprises a plurality of gas sensors arranged in a gas pipeline;
specifically, the message information includes an occurrence timestamp of the traffic information and an equipment serial number for acquiring the traffic information;
s4-1, the fault approval end collects flow measurement messages sent by the gas flow measurement module to the central control module in real time, extracts the flow measurement messages to generate flow variation values, compares the flow variation values with preset standard flow variation values, generates flow comparison results and stores the flow comparison results into the data storage end;
specifically, after collecting flow message information, the fault approval end sorts the flow message information according to the equipment serial number, and sequentially generates a plurality of flow change values, wherein the flow change values are gas flow difference values collected by two adjacent gas sensors;
s5-1, calling the flow comparison result by a position approval end, calling the equipment position of the gas sensor according to the equipment serial number, and mapping the equipment position into a gas pipeline diagram;
s3-2, the voltage regulator online diagnosis module carries out modeling based on the existing data sample, and the specific work flow comprises the following steps:
s3-2-1, establishing a basic information base by taking the data obtained by cleaning as a training sample;
s3-2-2, enhancing data on the basis of the basic information base, and adding local noise;
s3-2-3, randomly weighting and fusing the data in the basic information base according to the time period similar data;
s3-2-4, establishing a reinforcement learning offline model information base by the data obtained by reinforcement and a basic information base;
s4-2, an off-line training end is provided with a self-encoder, the self-encoder performs self-supervision feature extraction, the self-encoder comprises an encoder and a decoder, and the specific operation mode of the self-encoder comprises the following steps:
s4-2-1, taking the model output of the encoder as a feature extraction layer to obtain the model features of the voltage regulator;
s4-2-2, reducing the dimension of the model through sparse coding, and compressing the dimension of the model to the dimension of the model;
s4-2-3 introduces KL divergence in the loss function of the sparse coding feature extractor, wherein the KL divergence is used for representing the sparsity of the model;
s4-2-4, introducing a Loss function, setting preset parameters to represent the activity of a hidden layer unit, and adding the preset parameters into a Loss function to enable h to tend to be sparse;
specifically, a sparse self-encoder is obtained by combining a self-encoder of sparse coding, the sparse self-encoder is used for extracting the characteristics of a voltage regulator model, and the training process of the sparse self-encoding characteristic extractor comprises the following steps:
1, constructing an encoder by adopting a deep neural network;
2, symmetrically constructing a decoder according to the structure of the encoder;
adding KL divergence into a penalty function of the deep neural network to enable the model to tend to be sparse;
4, compressing the daily pressure data and flow data to one dimension, and transmitting the compressed data and flow data to the model as model input and model output labels;
5, iteratively training the model;
and S5-2, analyzing according to the characteristics after the characteristics of the voltage regulator are obtained, and obtaining state classification through a clustering sensor.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (8)
1. A voltage regulator online diagnosis method based on deep learning is characterized by comprising the following steps:
s1, respectively acquiring pressure values of the input end and the output end of the gas pipeline by the input end pressure transmitter and the output end pressure transmitter and transmitting the pressure values to the central control module;
s2, the pipeline monitoring end collects the pressure measurement message sent by the pressure measurement module to the central control module, extracts and generates a pressure value variable quantity, compares the pressure value variable quantity with a preset standard pressure variable quantity, generates a task number after an abnormality occurs, and sequentially sends the task number to the fault approval end and the pressure regulator online diagnosis module;
s3-1, collecting flow information inside a gas pipeline by a gas sensor, packaging the flow information to generate message information, and sending the message information to a central control module;
s4-1, the fault approval end collects flow measurement messages sent by the gas flow measurement module to the central control module in real time, extracts the flow measurement messages to generate flow variation values, compares the flow variation values with preset standard flow variation values, generates flow comparison results and stores the flow comparison results into the data storage end;
s5-1, calling the flow comparison result by a position approval end, calling the equipment position of the gas sensor according to the equipment serial number, and mapping the equipment position into a gas pipeline diagram;
s3-2, the voltage regulator online diagnosis module carries out modeling based on the existing data sample, and the specific work flow comprises the following steps:
s3-2-1, establishing a basic information base by taking the data obtained by cleaning as a training sample;
s3-2-2, enhancing data on the basis of the basic information base, and adding local noise;
s3-2-3, randomly weighting and fusing the data in the basic information base according to the time period similar data;
s3-2-4, establishing a reinforcement learning offline model information base by the data obtained by reinforcement and a basic information base;
s4-2, an off-line training end is provided with a self-encoder, the self-encoder performs self-supervision feature extraction, the self-encoder comprises an encoder and a decoder, and the specific operation mode of the self-encoder comprises the following steps:
s4-2-1, taking the model output of the encoder as a feature extraction layer to obtain the model features of the voltage regulator;
s4-2-2, reducing the dimension of the model through sparse coding, and compressing the dimension of the model to the dimension of the model;
s4-2-3 introduces KL divergence in the loss function of the sparse coding feature extractor, wherein the KL divergence is used for representing the sparsity of the model;
s4-2-4, introducing a Loss function, setting preset parameters to represent the activity of a hidden layer unit, and adding the preset parameters into a Loss function to enable h to tend to be sparse;
and S5-2, analyzing according to the characteristics after the characteristics of the voltage regulator are obtained, and obtaining state classification through a clustering sensor.
2. The on-line diagnosis method for the pressure regulator based on the deep learning of claim 1, wherein the pressure measurement module comprises an input end pressure transmitter and an output end pressure transmitter which are respectively arranged at an input port and an output port of the gas pipeline.
3. The on-line diagnosis method for the pressure regulator based on the deep learning of claim 1, wherein the pipeline monitoring end compares the pressure value variation with a preset standard pressure variation, and if the comparison deviation exceeds a threshold, a task number is generated.
4. The on-line diagnosis method for the voltage regulator based on the deep learning of claim 1, wherein the pipeline monitoring end generates a task number and then synchronously generates a sequence task associated with the task number, marks priority on one or more sequence tasks, and sends a sequence task request to the fault approval end and the overhaul position approval end according to the priority order.
5. The on-line diagnosis method for the pressure regulator based on the deep learning of claim 1, wherein the gas flow measurement module comprises a plurality of gas sensors arranged inside a gas pipeline.
6. The on-line diagnosis method for the voltage regulator based on the deep learning of claim 1, wherein the message information includes an occurrence timestamp of the traffic information and a device serial number for collecting the traffic information.
7. The on-line diagnosis method for the pressure regulator based on the deep learning of claim 1, wherein the fault approval end collects flow message information and then sorts the flow message information according to the equipment serial number, and a plurality of flow variation values are sequentially generated, wherein the flow variation values are gas flow difference values collected by two adjacent gas sensors.
8. The on-line diagnosis method for the voltage regulator based on the deep learning of claim 1, wherein a sparse self-encoder is obtained by combining a self-encoder of sparse coding, the sparse self-encoder is used for extracting model features of the voltage regulator, and a training process of the sparse self-encoding feature extractor comprises the following steps:
1, constructing an encoder by adopting a deep neural network;
2, symmetrically constructing a decoder according to the structure of the encoder;
adding KL divergence into a penalty function of the deep neural network to enable the model to tend to be sparse;
4, compressing the daily pressure data and flow data to one dimension, and transmitting the compressed data and flow data to the model as model input and model output labels;
and 5, iteratively training the model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910420646.8A CN111967618A (en) | 2019-05-20 | 2019-05-20 | Online diagnosis method for voltage regulator based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910420646.8A CN111967618A (en) | 2019-05-20 | 2019-05-20 | Online diagnosis method for voltage regulator based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111967618A true CN111967618A (en) | 2020-11-20 |
Family
ID=73357843
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910420646.8A Pending CN111967618A (en) | 2019-05-20 | 2019-05-20 | Online diagnosis method for voltage regulator based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111967618A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819107A (en) * | 2021-04-16 | 2021-05-18 | 四川九门科技股份有限公司 | Artificial intelligence-based fault prediction method for gas pressure regulating equipment |
| CN113987907A (en) * | 2021-09-09 | 2022-01-28 | 浙江浙能天然气运行有限公司 | Natural gas station transmission and distribution flow abnormity detection system based on machine learning and detection method thereof |
| CN114139648A (en) * | 2021-12-07 | 2022-03-04 | 北京科技大学 | Intelligent detection method and system for abnormity of tailing filling pipeline |
| CN116578889A (en) * | 2023-06-30 | 2023-08-11 | 国恒能元（天津）电力科技发展有限公司 | Power generation fault diagnosis method |
| CN117056734A (en) * | 2023-10-12 | 2023-11-14 | 山东能源数智云科技有限公司 | Method and device for constructing equipment fault diagnosis model based on data driving |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003016851A1 (en) * | 2001-08-07 | 2003-02-27 | Kunze, Silvia | Method for monitoring the function of pressure medium lines and corresponding device |
| US20040236472A1 (en) * | 2002-05-03 | 2004-11-25 | Junk Kenneth W. | Methods and apparatus for operating and performing diagnostics in a control loop of a control valve |
| CN103307447A (en) * | 2013-06-03 | 2013-09-18 | 清华大学 | Technical failure information monitoring and early warning system for urban gas pipe network |
| CN106899664A (en) * | 2017-02-15 | 2017-06-27 | 东北大学 | Oil pipeline distributed collaboration leak detection system and method based on multiple agent |
-
2019
- 2019-05-20 CN CN201910420646.8A patent/CN111967618A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003016851A1 (en) * | 2001-08-07 | 2003-02-27 | Kunze, Silvia | Method for monitoring the function of pressure medium lines and corresponding device |
| US20040236472A1 (en) * | 2002-05-03 | 2004-11-25 | Junk Kenneth W. | Methods and apparatus for operating and performing diagnostics in a control loop of a control valve |
| CN103307447A (en) * | 2013-06-03 | 2013-09-18 | 清华大学 | Technical failure information monitoring and early warning system for urban gas pipe network |
| CN106899664A (en) * | 2017-02-15 | 2017-06-27 | 东北大学 | Oil pipeline distributed collaboration leak detection system and method based on multiple agent |
Non-Patent Citations (4)
| Title |
|---|
| 姜勇 等: "《城市天然气管道网络SCADA系统应用技术研究》", 31 August 2016, 吉林人民出版社 * |
| 张海天 等: "基于稀疏自编码的深度故障诊断方法与研究", 《通信技术》 * |
| 桑运水 等: "《输送管道在线检测安全评价及修复技术》", 31 December 2005, 中国石油大学出版社 * |
| 郝学军 等: "供热锅炉用燃气调压器故障诊断系统的设计与实现", 《区域供热》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819107A (en) * | 2021-04-16 | 2021-05-18 | 四川九门科技股份有限公司 | Artificial intelligence-based fault prediction method for gas pressure regulating equipment |
| CN112819107B (en) * | 2021-04-16 | 2021-07-02 | 四川九门科技股份有限公司 | Artificial intelligence-based fault prediction method for gas pressure regulating equipment |
| CN113987907A (en) * | 2021-09-09 | 2022-01-28 | 浙江浙能天然气运行有限公司 | Natural gas station transmission and distribution flow abnormity detection system based on machine learning and detection method thereof |
| CN114139648A (en) * | 2021-12-07 | 2022-03-04 | 北京科技大学 | Intelligent detection method and system for abnormity of tailing filling pipeline |
| CN116578889A (en) * | 2023-06-30 | 2023-08-11 | 国恒能元（天津）电力科技发展有限公司 | Power generation fault diagnosis method |
| CN116578889B (en) * | 2023-06-30 | 2023-11-10 | 国网甘肃省电力公司经济技术研究院 | Power generation fault diagnosis method |
| CN117056734A (en) * | 2023-10-12 | 2023-11-14 | 山东能源数智云科技有限公司 | Method and device for constructing equipment fault diagnosis model based on data driving |
| CN117056734B (en) * | 2023-10-12 | 2024-02-06 | 山东能源数智云科技有限公司 | Method and device for constructing equipment fault diagnosis model based on data driving |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111967618A (en) | Online diagnosis method for voltage regulator based on deep learning | |
| CN102130783B (en) | Intelligent alarm monitoring method of neural network | |
| CN113361663B (en) | Artificial intelligence-based power grid event diagnosis autonomous learning method and system | |
| CN115640915B (en) | Intelligent gas pipe network compressor safety management method and Internet of things system | |
| CN112492567B (en) | Fault analysis and solution method and device in emergency command communication | |
| CN114201374A (en) | Operation and maintenance time sequence data anomaly detection method and system based on hybrid machine learning | |
| CN117560300B (en) | Intelligent internet of things flow prediction and optimization system | |
| CN116827764B (en) | Internet of things fault detection control method and system based on neural network | |
| CN109638318A (en) | Fuel cell remote monitoring system and method | |
| CN115730749B (en) | Power dispatching risk early warning method and device based on fusion power data | |
| CN115222303B (en) | Industry risk data analysis method and system based on big data and storage medium | |
| CN111143835A (en) | Non-invasive protection method for business logic of electric power metering system based on machine learning | |
| CN113378887B (en) | Emulsion pump fault grading diagnosis method | |
| CN106709659A (en) | LNG gas station equipment management system | |
| CN107065605A (en) | A kind of fault diagnosis and alarm method | |
| CN114138601A (en) | Service alarm method, device, equipment and storage medium | |
| CN112947364A (en) | System and method for early warning of equipment faults of power distribution station based on big data | |
| CN115273394B (en) | Intelligent circuit breaker under special application scene | |
| CN116863664A (en) | Real-time monitoring method and system for gas equipment | |
| CN115908082A (en) | Enterprise pollution discharge monitoring method and device based on electricity utilization characteristic indexes | |
| CN113129160A (en) | Electric power communication network inspection method based on equipment state perception and intellectualization | |
| CN112364088A (en) | Visual configuration system based on factory digital manufacturing resources | |
| CN117892212B (en) | Distributed heterogeneous energy station situation awareness monitoring method | |
| CN118506553B (en) | AIoT anomaly identification method, disaster early warning system and road safety system | |
| CN114969456A (en) | Method for judging comprehensive data information of electric intelligent well lid |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201120 |