CN110378257A - Artificial intelligence model full process automatization system - Google Patents
Artificial intelligence model full process automatization system Download PDFInfo
- Publication number
- CN110378257A CN110378257A CN201910598739.XA CN201910598739A CN110378257A CN 110378257 A CN110378257 A CN 110378257A CN 201910598739 A CN201910598739 A CN 201910598739A CN 110378257 A CN110378257 A CN 110378257A
- Authority
- CN
- China
- Prior art keywords
- artificial intelligence
- image
- module
- calibration block
- scale
- 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
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 82
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000012549 training Methods 0.000 claims abstract description 54
- 230000005540 biological transmission Effects 0.000 claims abstract description 25
- 238000004519 manufacturing process Methods 0.000 claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 230000033001 locomotion Effects 0.000 claims description 12
- 238000004891 communication Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 5
- 238000013459 approach Methods 0.000 claims description 4
- 230000005611 electricity Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 2
- 230000001360 synchronised effect Effects 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 abstract description 7
- 238000013480 data collection Methods 0.000 abstract description 3
- 206010054949 Metaplasia Diseases 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 230000015689 metaplastic ossification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000465 moulding Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B9/00—Simulators for teaching or training purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Educational Administration (AREA)
- Bioinformatics & Computational Biology (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Educational Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Feedback Control In General (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to it is artificial intelligence image recognition, being directed to instrument and meter dial plate, with automatically generated data collection to realize the artificial intelligence model full process automatization systems of intelligent recognition dial plate data.The system includes: calibration block, transmission module, scale feedback module, image capture module, transmission module connects calibration block and scale feedback module, the numerical value of calibration block instruction and the numerical value of scale feedback module detection can establish one-to-one relationship, image capture module is used to obtain the image of calibration block, and the numerical value of scale feedback module detection is used for the image of reference mark module.The system automates integrated management for the overall process training education of artificial intelligence, many instrument and meter dials identification for the enterprise that keeps the safety in production and monitoring data.
Description
Technical field
The present invention relates to a kind of artificial intelligence model full process automatization systems, more particularly to artificial intelligence image recognition
, it is being directed to instrument and meter dial plate, for artificial intelligence overall process training, have automatically generated data collection to realizing intelligence
It can identify the artificial intelligence model full process automatization system of dial plate data.
Background technique
In recent years, artificial intelligence learns to be rapidly developed, and is widely applied among all trades and professions by people, therewith phase
Be that the related education of artificial intelligence study becomes more to popularize.But the intelligence learning of all people's work is all built upon greatly
On the basis of measuring data set, the scale of the effect of the artificial intelligence study data set used to training is directlyed proportional to accuracy.Mesh
Before, the learner after artificial intelligence training can understand program and tuning parameter to improve the accuracy of model, due to directly adjusting
With data with existing collection training, the sampling, processing and foundation of data set cannot be participated in, has lacked a particularly important ring, has caused to train
Learner after instruction lacks the ability for using artificial intelligence solving practical problems, and cannot be to apply.Generate one it is huge and
Accurate data set generally requires a large amount of time and human cost, this expensive time cost and human cost hinder artificial intelligence
The full chain education of energy, so a kind of method for developing quickly and accurately generation data set is artificial intelligence education when business
It is anxious.
Meanwhile many manufacturing enterprises also retain a large amount of off-line type instrument and meter now, these meters can not by it is long-range,
Online acquisition data, it is necessary to acquire data by way of sending special messenger's reading on duty, very big cost will be generated by replacing meter completely.
It is also brought to data integration moreover, miscellaneous meter all passes through communication online acquisition data due to the difference of communication mode
The deuce to pay.Therefore, it goes universally to identify various meter data by the image information that camera acquires, in current practical life
Producing also has substantial worth in environment, investment substantial contribution technological transformation is not needed, without the data format for considering each meter.So
And various meters is various informative, establishing data set training artificial intelligence model requires to spend very big energy, according to different tables
Meter automatically generating data set also plays an important roll actual production to train the artificial intelligence model of identification.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art described above, proposing one kind can automatically generate and demarcate
The method of data set.This quick foundation of acquisition by way of a large amount of and reliable data set, on the one hand advantageously account for
The complicated problem of data set process is generated before, is allowed and is established data set this process and be added to the general of artificial intelligence study
And in education, student is helped comprehensively to understand the process of entire artificial intelligence study, so that study chain is more completely and rich in real
With property, logicality, while learning cost is also reduced, facilitates the popularization of artificial intelligence study education and perfect.
On the other hand for the enterprise that keeps the safety in production, all there are a large amount of instrument and meters, if the reading of these meters depends on
It is artificial to read, it will to expend many human resources, and fault rate when worker handles a large amount of dial plate data at the same time can be big
Increase greatly；And after these enterprises have a kind of this mode for rapidly and accurately generating data set, the reading of these instruments
Pervious manpower can be replaced by reading and monitoring based on artificial intelligence model made of the data set training generated
Resource, so that it may the fault in human cost required for enterprise and production be greatly decreased, to improve production efficiency.Meanwhile
It can be used as on-line monitoring data using the dial plate data that artificial intelligence model is read, convenient for being integrated into automatic management process
In.
A kind of artificial intelligence model full process automatization system, comprising: calibration block, transmission module, scale feedback module,
Image capture module, transmission module connect calibration block and scale feedback module, and the numerical value and scale of calibration block instruction are fed back
The numerical value of module detection can establish one-to-one relationship, and image capture module is used to obtain the image of calibration block, and scale is anti-
The numerical value for presenting module detection is used for the image of reference mark module.
The transmission module includes: motor submodule and motor control module, and motor control module controls motor submodule
Block movement, motor submodule have double output ends, connect respectively with calibration block, scale feedback module.
The image capture module can adjust obtain calibration block image when ambient lighting degree, additionally it is possible to adjustment with
Angle, the distance of calibration block, obtain and mark different illuminances, angle, distance parameter calibration block image, reference mark
The data set of numerical value and the training of the calibration block image construction artificial intelligence model of parameter, the data set are divided into: training set and survey
Examination collection.
The scale feedback module is identical as the method for operation of calibration block and synchronous；Calibration block is instrument and meter table
Count dial；Scale feedback module is referred to using rotation or the potentiometer of linear motion with the resistance value that scale feedback module measures
Show the scale numerical value of calibration block.
The image capture module includes: camera, image acquisition procedure, and image acquisition procedure controls camera and obtains
Image is normalized in image, and is converted into calibration block numerical value according to the detection numerical value that scale feedback module transmits
And tag image.
Transmission module includes: motor submodule and motor control module, and scale feedback module uses rotational potentiometer, scale
Module is meter dial, the state of meter dial is indicated with the resistance value of rotational potentiometer, motor control module passes through electricity
The resistance value of rotational potentiometer is adjusted to a state by loom module, measures the rotational potentiometer actual resistance of the state, according to
The corresponding relationship of the actual resistance and meter dial obtains the meter dial instruction numerical value of this state；Image capture module
Starting takes pictures to meter dial, and size to image and light obtain the unified image of standard according to being set for handling,
And according to the reference mark disk image of the rotational potentiometer measured value corresponding dial numeric indicia state；Meter after label
One element of data set needed for dial image becomes training artificial intelligence model, in this approach, whole scales are covered in generation
The image of disk information state simultaneously automates label, establishes complete data set.
The transmission module includes motor control module and motor submodule, and motor control module controls motor submodule
Movement, the motor submodule of transmission module drive calibration block and the movement of scale feedback module, and motor control module is according to comprehensively
Interval strategy constantly adjustment motor submodule movement, thus image when obtaining each scale state of calibration block；Measurement electricity
The resolution ratio of loom module drive calibration block and scale feedback module；Interval strategy comprehensively, it is comprehensively every including calibration block
One scale state, interval are determined according to motor submodule parameter.
The artificial intelligence overall process Training Methodology of artificial intelligence model full process automatization system, characterized in that artificial intelligence
Energy overall process Training Methodology includes: production data set, building artificial intelligence training pattern, adjusting training Model Parameter Optimization
Energy；It is basis as the data set production for starting link, a kind of calibration block is selected to feed back as image recognition target with scale
The image of the numeric indicia calibration block of module detection, in the way of transmission module building automation generation data set；It can
Select different illuminances, distance, angle parameter and automate and generate different data set；After automated production data set, choosing
Artificial intelligence model training platform is selected, artificial intelligence training pattern is constructed；Data set based on production starts artificial intelligence model
Training；According to training result, continuous adjusting training Model Parameter Optimization performance, until meeting the requirements.
The meter scale recognition methods of artificial intelligence model full process automatization system, characterized in that various meters are shown
Control flow is produced, the dial of meter is as calibration block for reading；The output form of various meters is different, data format
Difference, communication mode is also different, and remote online is replaced to read the communication integration mode of data in a manner of image recognition；Meter is carved
The image recognition of scale is by the way of constructing artificial intelligence training pattern；Data set needed for artificial intelligence training pattern is with certainly
Dynamicization mode generates；With the image of the numeric indicia meter dial of scale feedback module detection, certainly using transmission module building
Metaplasia is moved into the mode of data set；Can select different illuminances, distance, angle parameter and automate and generate different data
Collection；After automated production data set, artificial intelligence model training platform is selected, constructs artificial intelligence training pattern；Based on system
The data set of work starts artificial intelligence model training；According to training result, continuous adjusting training Model Parameter Optimization performance, directly
To the numerical value of correct identification meter dial.
When identifying more than one meter scale, according to meter scale recognition methods, the data that automatic Building day-mark is remembered respectively
Collection, is respectively trained artificial intelligence identification model；When practical application, first the image of acquisition is split, each meter point
It cuts out, trained artificial intelligence model is recycled to identify respectively.
For the training education of artificial intelligence study, need one kind can by it is simple and quick it is accurate in a manner of generate data set
System.If selecting directly although can bring convenience, but to lack using already existing data set in education
The generating process of required data set in education lacks a committed step, and the universal education of such artificial intelligence study is not
Completely.So this requires a kind of relatively advanced methods to automatically generate data set, to keep the process of education more complete.
And for safety in production enterprise, production process, which generally requires, to be monitored various gauge dials to control its life
Produce process.If using traditional artificial meter reading mode, human cost needed for will increase factory, and the mistake of artificial meter reading
There are error and erroneous judgement in journey, production efficiency is reduced；If reading instrument and meter scale using the mode of on-line checking and communication
The data of disk, however the output form of various meters is different, data mode is different, and the integration mode of communication can very complicated, realization
Difficulty is high, and many meters cannot achieve the mode read online；It, then will be according to each table according to general image recognition
The scale disk-form of meter carries out complicated image procossing, and implementation method does not have generality；Artificial intelligence approach carries out instrument
Monitoring, then need to establish a huge and accurate database to train identification model, and establish training dataset and need to spend
Take a large amount of time from manually to be demarcated manually to different instrument status.So this requires a kind of advanced sides
Method automatically generates the training dataset of instrument and meter, to reduce the time, save production cost.
For the universal education of artificial intelligence study, lacks relevant educational suite in the market, be unfavorable for artificial intelligence
Practise the popularization of education.For in industrial production dial reading and monitoring, lack a kind of cheap, method highly effective and versatile to big
It measures different dial plates and is monitored management.
Compared with prior art, the invention has the following advantages: first, for artificial intelligence study overall process training
In instruction education, since problem to be solved, include: acquisition, processing, foundation and the artificial intelligence model training of data set,
The ability that enhancing learner studies in order to practise；The second, a large amount of instrument and meters are solved for manufacturing enterprise online to read and data integration
Problem, convenient for establishing the identification and monitoring of automatic flow.
Detailed description of the invention
Fig. 1 is artificial intelligence model full process automatization system.
Specific embodiment
Elaborate with reference to the accompanying drawing to the embodiment of the present invention: the present embodiment before being with technical solution of the present invention
It puts and is implemented, give detailed embodiment and process, but protection scope of the present invention is not limited to following embodiments.
It is to be appreciated, therefore, that the attached claims are intended to cover fall into all such modifications and changes of true spirit of the invention.
As shown in Figure 1, artificial intelligence model full process automatization system includes: calibration block, transmission module, scale feedback
Module, image capture module, transmission module include: motor submodule and motor control module again, and motor submodule is using configuration
The direct current generator of slowing-down structure, the double output ends of transmission module are realized using gear structure, the motor submodule in transmission module
It is connected in manner of gear wheels with calibration block, scale feedback module, the transmission ratio of calibration block and scale feedback module is 10:1；
The dial of calibration block has 60 scales, and each scale represents numerical value 1；And scale feedback module rotates current potential using 10 circles
Device, resistance value 6k Ω ± 5% of rotational potentiometer, ± 10 Ω of precision, linear ± 0.25%；The 6k Ω resistance value and scale of rotational potentiometer
60 scales of disk correspond.Dial and 45 degree of angles of image capture module, ambient lighting 2lux.
The movement of the motor control module control motor submodule of transmission module, the motor submodule of transmission module, which drives, to be carved
Module and the movement of scale feedback module are spent, motor control module is moved according to comprehensive interval strategy constantly adjustment motor submodule,
To obtain the state that dial indicates each scale；By taking dial shown in FIG. 1 as an example, 60 scales of dial are corresponding
The 6k Ω resistance value of rotational potentiometer, the corresponding 100 Ω resistance values of a scale, ± 10 Ω of precision of rotational potentiometer, so selection 20
Ω as benchmark, a scale set can identified resolution ratio as 5, then each scale will choose five states to obtain
Image, dial 300 states of total setting；Come with the resistance value of rotational potentiometer for the state for referring to dial；If motor control
Molding block give rotational potentiometer resistance value successively are as follows: 20,40,60,80 ..., 6000；Motor submodule as power part,
Rotate the more difficult accurate realization of angle very little, so, the optimal given resistance value embodiment 1 of motor control module are as follows: 100,200 ...,
6000；20,120,220,…,5920；40,140,240,…,5940；60,160,260,…,5960；80,180,280,…,
5980。
Come with the resistance value of rotational potentiometer for the state for referring to dial, motor control module is by motor submodule rotation
The resistance value of potentiometer is adjusted to a state, measures the actual resistance of the state, according to the quarter of the actual resistance and calibration block
The corresponding relationship of scale obtains the dial instruction numerical value of this state；Image capture module starting takes pictures to dial, and compares
The size and light of piece obtain the unified image of standard according to being set for handling, and corresponding according to rotational potentiometer measured value
The dial numeric indicia state dial photo；Dial photo after label just becomes trained artificial intelligence model
One element of required data set generates the image for covering whole dial information states and automates label, just in this approach
Establish complete data set；Based on artificial intelligence training platform PaddlePaddle, the data set is uploaded, according to setting
Artificial intelligence model parameter, it will be able to which training obtains identifying the artificial intelligence identification model of the dial, by debugging artificial intelligence
Energy model parameter, setting model structure, it will be able to the high dial identification model of accuracy is quickly and easily obtained, thus fastly
Victory accurately identifies the meter dial reading.
Embodiment 2, calibration block are divided into two parts, first part 0,1,2,3,4,5,6,7,8,9, and second part is
0,10,20,30,40,50,60,70,80,90 dial, five scales between 0-10 are identified, i.e., each scale represents numerical value
2；Scale feedback module uses the 10 total resistance values of circle for the high-precision rotary potentiometer of 10k Ω；The corresponding 20 Ω resistance values of each scale；It should
Meter dial shares 500 states, is spaced given method according to comprehensive, motor control module gives this 500 states, image
The image of meter dial when acquisition module obtains each state, and carved according to corresponding to scale feedback module detected resistance
Angle value marks the image.
Artificial intelligence model full process automatization system as shown in Figure 1, motor control module use NI myRIO, integrate
PID feedback algorithm controls motor submodule；High-precision rotary potentiometer is using 10 circle range 6k Ω, the electric rotating
Position device can be matched with most of meter dial, to establish the corresponding relationship of resistance value and scale value；Motor submodule packet
Include direct current generator and deceleration device, direct current generator parameter are as follows: voltage rating 5V, rated speed 6000r/m, the deceleration of deceleration device
Than for 10:1；The rotating output shaft of motor submodule can have both ends.
There are three pins: intermediate terminal, one end, the other end for high-precision rotary potentiometer tool as shown in Figure 1, one end access
5V voltage, one end ground connection, intermediate terminal output voltage, the voltage value and its resistance value correspond, and need to only use motor control mould
Block myRIO acquires the voltage value being capable of reference mark disk numerical value；Image capture module is come using USB camera and computer
It realizes, the image that the program write in computer can acquire camera is handled and demarcated；In motor control module myRIO
Control program and image capture module in image acquisition procedure link.
Dial Image Acquisition process are as follows: motor control module myRIO controls motor submodule for high-precision rotary current potential
Device is moved to some position, and acquires the voltage value of moment high-precision rotary potentiometer intermediate terminal output, then by the electricity
Pressure value is converted to dial data and is transmitted in image capture module；This moment, the control program in motor control module myRIO is temporary
Stopping, the program operation in image capture module collects dial image this moment, and normalized by camera, according to
The dial data of motor control module myRIO transmitting demarcate it；The program of image capture module is suspended, motor control module
Control program in myRIO is again started up, and is circuited sequentially, and is constantly generated the dial photo of label, is established data set with this；?
Artificial intelligence training platform AIstudio uploads the data set of foundation, and PaddlePaddle is called easily to establish manually
Model of mind is trained, and can obtain high-precision identification mould according to the subtle adjustment data set of training result and model parameter
Type.
Artificial intelligence model full process automatization system is used for the overall process training education of artificial intelligence, safety in production enterprise
Many instrument and meter dials identification and monitoring data automate integrated management.
Claims (10)
1. a kind of artificial intelligence model full process automatization system, comprising: calibration block, transmission module, scale feedback module, figure
As acquisition module, transmission module connects calibration block and scale feedback module, and the numerical value and scale of calibration block instruction feed back mould
The numerical value of block detection can establish one-to-one relationship, and image capture module is used to obtain the image of calibration block, scale feedback
The numerical value of module detection is used for the image of reference mark module.
2. artificial intelligence model full process automatization system according to claim 1, characterized in that the transmission module
It include: motor submodule and motor control module, motor control module controls the movement of motor submodule, and motor submodule has double
Output end is connect with calibration block, scale feedback module respectively.
3. artificial intelligence model full process automatization system according to claim 1, characterized in that the Image Acquisition
Module can adjust ambient lighting degree when obtaining calibration block image, additionally it is possible to which angle, the distance of adjustment and calibration block obtain
Take and mark different illuminances, angle, distance parameter calibration block image, the calibration block figure of reference mark numerical value and parameter
Data set as constituting artificial intelligence model training, the data set are divided into: training set and test set.
4. artificial intelligence model full process automatization system according to claim 1, characterized in that the scale feedback
Module is identical as the method for operation of calibration block and synchronous；Calibration block is instrument and meter meter dial；Scale feedback module
Using rotation or the potentiometer of linear motion, with the scale numerical value of the resistance value indicating graduation module of scale feedback module measurement.
5. artificial intelligence model full process automatization system according to claim 1, characterized in that the Image Acquisition
Module includes: camera, image acquisition procedure, and image acquisition procedure control camera obtains image, image is normalized
Processing, and calibration block numerical value is converted into and tag image according to the detection numerical value that scale feedback module transmits.
6. artificial intelligence model full process automatization system according to claim 1, characterized in that transmission module includes:
Motor submodule and motor control module, scale feedback module use rotational potentiometer, and calibration block is meter dial, with rotation
Turn the resistance value of potentiometer to indicate the state of meter dial, motor control module is by motor submodule rotational potentiometer
Resistance value is adjusted to a state, the rotational potentiometer actual resistance of the state is measured, according to the actual resistance and meter dial
Corresponding relationship, obtain this state meter dial instruction numerical value；Image capture module starting takes pictures to meter dial, and
Size and light to image obtain the unified image of standard according to being set for handling, and according to rotational potentiometer measured value
The reference mark disk image of the corresponding dial numeric indicia state；It is artificial that meter dial image after label becomes training
One element of data set needed for model of mind generates in this approach and covers the image of whole dial information states and automatic
Change label, establishes complete data set.
7. artificial intelligence model full process automatization system according to claim 1, characterized in that the transmission module packet
Motor control module and motor submodule are included, motor control module controls the movement of motor submodule, motor of transmission module
Module drives calibration block and the movement of scale feedback module, and motor control module is according to comprehensive interval strategy constantly adjustment motor
Block motion, thus image when obtaining each scale state of calibration block；Measure motor submodule drive calibration block and
The resolution ratio of scale feedback module；Interval strategy comprehensively includes each scale state of calibration block, is spaced according to electricity comprehensively
Loom module parameter is determined.
8. a kind of artificial intelligence for the described in any item artificial intelligence model full process automatization systems of claim 2 to 7
Overall process Training Methodology, characterized in that artificial intelligence overall process Training Methodology includes: production data set, building artificial intelligence instruction
Practice model, adjusting training Model Parameter Optimization performance；It is basis as the data set production for starting link, selects a kind of scale mould
Block utilizes transmission module structure as image recognition target with the image of the numeric indicia calibration block of scale feedback module detection
Build the mode that automation generates data set；Can select different illuminances, distance, angle parameter and automate and generate different number
According to collection；After automated production data set, artificial intelligence model training platform is selected, constructs artificial intelligence training pattern；It is based on
The data set of production starts artificial intelligence model training；According to training result, continuous adjusting training Model Parameter Optimization performance,
Until meeting the requirements.
9. a kind of meter scale for the described in any item artificial intelligence model full process automatization systems of claim 2 to 7
Recognition methods, characterized in that various meter display production control flows, the dial of meter is as calibration block for reading；
The output form of various meters is different, and data format is different, and communication mode is also different, replaces remotely existing in a manner of image recognition
The communication integration mode of line reading data；The image recognition of meter dial is by the way of constructing artificial intelligence training pattern；
Data set needed for artificial intelligence training pattern generates in an automated manner；The numeric indicia meter detected with scale feedback module
The image of dial, in the way of transmission module building automation generation data set；Can select different illuminances, distance,
Angle parameter and automate and generate different data set；After automated production data set, select artificial intelligence model training flat
Platform constructs artificial intelligence training pattern；Data set based on production starts artificial intelligence model training；According to training result, no
Disconnected adjusting training Model Parameter Optimization performance, the numerical value until correctly identifying meter dial.
10. meter scale recognition methods according to claim 9, characterized in that when identifying more than one meter scale, root
According to meter scale recognition methods, the data set of automatic Building day-mark note, is respectively trained artificial intelligence identification model respectively；Actually answer
Used time is first split the image of acquisition, each meter is split, and recycles trained artificial intelligence model point
It does not identify.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910598739.XA CN110378257B (en) | 2019-07-04 | 2019-07-04 | Artificial intelligent model whole process automation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910598739.XA CN110378257B (en) | 2019-07-04 | 2019-07-04 | Artificial intelligent model whole process automation system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110378257A true CN110378257A (en) | 2019-10-25 |
| CN110378257B CN110378257B (en) | 2023-12-19 |
Family
ID=68251933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910598739.XA Active CN110378257B (en) | 2019-07-04 | 2019-07-04 | Artificial intelligent model whole process automation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110378257B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101614562A (en) * | 2009-08-05 | 2009-12-30 | 钟胜 | Pointer type reading device with digitizing function |
| CN106194167A (en) * | 2016-09-23 | 2016-12-07 | 上海神开石油设备有限公司 | A kind of device and method of rotation status subscript orientation resistivity |
| RU2604116C1 (en) * | 2015-07-21 | 2016-12-10 | Федеральное государственное унитарное предприятие "Российский Федеральный Ядерный Центр - Всероссийский Научно-Исследовательский Институт Технической Физики имени академика Е.И. Забабахина" (ФГУП "РФЯЦ-ВНИИТФ им. академ. Е.И. Забабахина") | Method for automatic reading from pointer-deflecting instruments |
| WO2017084186A1 (en) * | 2015-11-18 | 2017-05-26 | 华南理工大学 | System and method for automatic monitoring and intelligent analysis of flexible circuit board manufacturing process |
| CN107229930A (en) * | 2017-04-28 | 2017-10-03 | 北京化工大学 | A kind of pointer instrument numerical value intelligent identification Method and device |
| CN107729906A (en) * | 2017-10-24 | 2018-02-23 | 国网江苏省电力公司南京供电公司 | A kind of inspection point ammeter technique for partitioning based on intelligent robot |
| CN107944486A (en) * | 2017-11-20 | 2018-04-20 | 中国电子科技集团公司第四十研究所 | Suitable for the test data identifying processing method and system tested automatically |
| US20190095739A1 (en) * | 2017-09-27 | 2019-03-28 | Harbin Institute Of Technology | Adaptive Auto Meter Detection Method based on Character Segmentation and Cascade Classifier |
| CN109871859A (en) * | 2018-09-28 | 2019-06-11 | 北京矩视智能科技有限公司 | One kind automatically generating training set of images system |
-
2019
- 2019-07-04 CN CN201910598739.XA patent/CN110378257B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101614562A (en) * | 2009-08-05 | 2009-12-30 | 钟胜 | Pointer type reading device with digitizing function |
| RU2604116C1 (en) * | 2015-07-21 | 2016-12-10 | Федеральное государственное унитарное предприятие "Российский Федеральный Ядерный Центр - Всероссийский Научно-Исследовательский Институт Технической Физики имени академика Е.И. Забабахина" (ФГУП "РФЯЦ-ВНИИТФ им. академ. Е.И. Забабахина") | Method for automatic reading from pointer-deflecting instruments |
| WO2017084186A1 (en) * | 2015-11-18 | 2017-05-26 | 华南理工大学 | System and method for automatic monitoring and intelligent analysis of flexible circuit board manufacturing process |
| CN106194167A (en) * | 2016-09-23 | 2016-12-07 | 上海神开石油设备有限公司 | A kind of device and method of rotation status subscript orientation resistivity |
| CN107229930A (en) * | 2017-04-28 | 2017-10-03 | 北京化工大学 | A kind of pointer instrument numerical value intelligent identification Method and device |
| US20190095739A1 (en) * | 2017-09-27 | 2019-03-28 | Harbin Institute Of Technology | Adaptive Auto Meter Detection Method based on Character Segmentation and Cascade Classifier |
| CN107729906A (en) * | 2017-10-24 | 2018-02-23 | 国网江苏省电力公司南京供电公司 | A kind of inspection point ammeter technique for partitioning based on intelligent robot |
| CN107944486A (en) * | 2017-11-20 | 2018-04-20 | 中国电子科技集团公司第四十研究所 | Suitable for the test data identifying processing method and system tested automatically |
| CN109871859A (en) * | 2018-09-28 | 2019-06-11 | 北京矩视智能科技有限公司 | One kind automatically generating training set of images system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110378257B (en) | 2023-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106990384B (en) | Device and method for detecting faulty wiring of three-phase electric energy meter | |
| WO2020233516A1 (en) | Measuring instrument mass transfer/traceability method and measuring instrument management system | |
| CN104089574A (en) | New classroom teaching model based on machine vision technology | |
| CN110154029A (en) | Robot On-line Control and emulation test system and method based on LABVIEW | |
| CN108958212A (en) | Calibrating operation auxiliary device, calibrating operation householder method and recording medium | |
| CN109238111B (en) | Intelligent rotating machinery centering instrument and operation method thereof | |
| CN108984400A (en) | Automation interface test method based on warnlng surveillance system | |
| CN110378257A (en) | Artificial intelligence model full process automatization system | |
| CN202720088U (en) | Device which transmits pressure magnitude through Internet under real-time environment | |
| CN116542469A (en) | Building engineering supervision intelligent integrated cloud platform based on BIM technology and supervision method | |
| CN208637020U (en) | Industry spot sensor test application actual training device | |
| CN104502560B (en) | The chemical examination chamber system checked and accepted for Coal Quality | |
| Qiao et al. | Application research of artificial intelligence technology in error diagnosis of electric energy meter | |
| CN110530409A (en) | A kind of on-line automatic measurement and calibration instrument, calibration method and computer media | |
| CN114374272B (en) | Automatic overhaul method for intelligent substation | |
| CN102706511B (en) | Device for transferring pressure quantity value through internet in real-time environment | |
| CN102200778A (en) | Device for automatically calibrating parameters of regulators | |
| CN201965485U (en) | Automatic verifying device for adjustor parameters | |
| CN106546936B (en) | Virtual detection system and detection method for detecting energy efficiency monitoring equipment | |
| CN112161647A (en) | Measuring instrument, measuring transmission/tracing method thereof, measuring instrument management system and method | |
| CN107729405A (en) | A kind of calibration cycle generation method of ATE | |
| CN114896309A (en) | Method and system for converting and displaying monitoring data of hydropower station | |
| O'Neal et al. | Metering and Calibration in LoanSTAR Buildings | |
| Martín et al. | An electronic instrumentation design project for computer engineering students | |
| JP2004021583A (en) | Maintenance method of plant instrumentation system |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20231122 Address after: 2-1802, Building 1, Longhu Tianpuyuan, Aoti West Road, Lixia District, Jinan City, Shandong Province, 250000 Applicant after: Shandong Qiaosi Intelligent Technology Co.,Ltd. Address before: Room 403, No. 210 Zhongshan South 1st Road, Huangpu District, Shanghai, 200011 Applicant before: SHANGHAI TULING NEW ENERGY TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |