CN109977838A - A kind of flame combustion state detection method - Google Patents
A kind of flame combustion state detection method Download PDFInfo
- Publication number
- CN109977838A CN109977838A CN201910211096.9A CN201910211096A CN109977838A CN 109977838 A CN109977838 A CN 109977838A CN 201910211096 A CN201910211096 A CN 201910211096A CN 109977838 A CN109977838 A CN 109977838A
- Authority
- CN
- China
- Prior art keywords
- flame
- area
- image
- neural network
- continuous
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
- G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
- G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
- G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/44—Event detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biodiversity & Conservation Biology (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Photometry And Measurement Of Optical Pulse Characteristics (AREA)
- Image Analysis (AREA)
Abstract
A kind of flame combustion state detection method, using the area of flame feature in a period of time as the foundation of identification;The flame combustion state detection method shown is judged to flame image and state in Qt graphical interfaces;It is identified using video image, in the combustion process of practical boiler flame, flame will appear the interference such as flashing, shake, flue dust, and using image procossing, available more information improve noise resisting ability using area of flame and filtering processing;Flame combustion situation is judged using video acquisition, judgement can be acquired in real time, the reliability of raising system, failure is prevented to greatest extent, it is learned using offline, online recognition improves arithmetic speed and recognition efficiency, and video surveillance can detecte the overall process of flame work, it is identified using Online Video, system real time is good, highly-safe.
Description
Technical field
The invention belongs to flame combustion state technical field is detected in the boiler in Furnace Safeguard Supervisory System, and in particular to
A kind of flame combustion state detection method.
Background technique
Daily industrial practice is widely used in the detection of flame combustion state, and to natural phenomena and certainly
So in the scientific research of rule, such as require to detect the flame combustion state in boiler in power plant, coal works.In pot
In the operational process of furnace, the safe operation of boiler be unable to do without the detection to boiler flame combustion state, and fire defector is burner hearth
Important component part in safety monitoring system.The safe operation of boiler will directly affect the safety and economy benefit of enterprise.It is existing
Boiler flame detection method be mostly the detection based on temperature sensor and light intensity sensor, there is big stagnant for this class method
Afterwards, flame information content is few, can not rule out furnace in interference, the defects such as recognition accuracy is low.
Since boiler is high temperature, high pressure, closed tank body, should not directly observe.There are the factors such as infrared, ultraviolet in furnace
Interference, acutely, leading to the flame of boiler internal, change in shape is complicated in combustion for chemical reaction, and carried out in boiler into
The variation of flame can all be impacted when the operations such as material, exhaust, these problems can all increase the difficulty to flame combustion judgement,
Therefore to a difficult point in the detection of the flame combustion state of boiler internal always Furnace Safeguard Supervisory System.
Summary of the invention
To overcome above-mentioned the deficiencies in the prior art, the object of the present invention is to provide a kind of flame combustion state detection method,
For the flame combustion state for accurately identifying boiler internal;Have the characteristics that real-time is good, characteristic information is continuous, identification is accurate.
The invention patent proposes one kind and pre-processes to collected flame combustion state picture, utilizes mind afterwards later
Flame combustion state is identified through network algorithm, and by the parameter extraction of neural network, carries out off-line learning, online recognition
Flame combustion state detection method.This method using the area of flame feature in a period of time as the input of neural network,
Since input dimension is high, acquisition contains much information, feature identification is accurate.It is identified using Online Video, system real time is good, pacifies
Quan Xinggao.And the flame combustion state detection method shown is judged to flame image and state in Qt graphical interfaces.It is this
Method is identified using video image, and more information can be acquired, and using area of flame and filtering processing, improves anti-noise
Sound ability.Flame combustion situation is judged using one section of flame video, can carry out acquiring judgement in real time, improve the reliable of system
Property and discrimination, due to acquisition contain much information, failure can be detected to greatest extent, utilize off-line learning, online recognition
Arithmetic speed is improved, video surveillance can detecte the overall process of flame work, be identified using Online Video, system real time is good,
It is highly-safe.
To achieve the above object, the technical solution adopted by the present invention:
A kind of flame combustion state detection method, specifically includes the following steps:
Firstly, extracting a flame from lighting to normal combustion again to the continuous flame video for extinguishing overall process, will extract
The image information of view carries out pretreatment and extracts area of flame feature;Known using the area of flame feature as flame combustion state
Other foundation;
The continuous 100 frame area of flame feature for being greater than 10 seconds videos for any one section as the input vector of neural network,
Three-layer neural network is constructed, target is to identify 11 flame status;Using in output state the corresponding state of maximum probability as
Final identification state;
Third, graphical interface of user
Picture collection control, the off-line learning in MATLAB software, according to existing flame video are carried out using Qt platform
Continuous picture extracts a flame from lighting to normal combustion again to the area of flame feature extinguished in overall process, training nerve net
Network parameter constructs neural network Positive Propagation Algorithm in Qt platform, is bringing training neural network parameter into neural network just
Into propagation formula, online recognition.
The neural network Positive Propagation Algorithm, forward-propagating formula are as follows:
Yi=σ (wi,klk+b1,i)W2,k+b2,i
Wherein yiRepresent the i-th dimension output of output vector, w1,kFor every weight of input layer to hidden layer, lkInput is represented,
w1,klkFor matrix operation, b1,iThe items biasing of input layer to hidden layer is represented, σ is activation primitive, w2,kOutput layer is arrived for hidden layer
Every weight, b2,iThe items biasing of hidden layer to output layer is represented, using tansig activation primitive.
The flame image and area of flame change curve at current time are shown in graphical interface of user, and are shown simultaneously
Recognition result, and area data and judging result are saved;Recognition result according to output vector maximum probability principle.
The image information of the view is pre-processed, and color image is converted to gray scale picture;Gray scale is defined later
Threshold value is 0.5, and the pixel point image using gray value in grayscale image greater than 0.5 utilizes expansion-corrosion as the area features of flame
Algorithm is eliminated because of the hollow out caused by shaking with flue dust interference, and the continuous flame area that processing is completed is by sliding mean filter
Processing, using the area value of the continuous 100 frame flame of a period of time as input.
Flame ignition process is divided into five stages by the building three-layer neural network, and extinguishing process is equally divided into five
A stage such as is fed to boiler internal by stationary process and in stationary process, is vented at continuous flame face when operation
Product value is demarcated as plateau, and using 100 input of tansig activation primitive and gradient descent method building, 50 hidden
Layer neuron, the three-layer neural network of 11 outputs.
The beneficial effects of the present invention are:
The present invention is the flame combustion state detected in complicated boiler environment, is identified to the combustion state of flame,
As the foundation changed to flame intensity power and working condition is converted.
The present invention is to be identified by acquiring flame video using the area of flame feature in a period of time as flame status
Foundation, the flame combustion state detection method judged later using intelligent algorithm.It include mainly to flame combustion process
In video acquisition and image procossing, the area of flame feature in a period of time is identified using neural network algorithm, most
Afterwards by acquired image and neural network judgement as the result is shown into Qt graphical interfaces.Utilize the flame front in a period of time
Product feature contains much information compared with the flame characteristic of a picture as the foundation that flame status identifies, characteristic quantity is big, feature letter
The features such as breath is continuous, real-time is good, therefore identify that accurate, accuracy is high.
In the combustion process of flame, by video acquisition, image procossing is carried out, color image is converted into grayscale image
Piece;Defining gray threshold later is 0.5, and the pixel point image using gray value in grayscale image greater than 0.5 is as the area-graph of flame
Picture is eliminated using expansion-erosion algorithm because of the hollow out caused by shaking with flue dust interference, the continuous flame area that processing is completed
Three layers of nerve net are constructed using the area value of the continuous 100 frame flame of a period of time as input by sliding mean filter processing
Flame ignition process is divided into five stages by network structure, and extinguishing process is equally divided into five stages, by stationary process and flat
It continuous flame area value when operation such as fed, be vented during steady to boiler internal as plateau, being demarcated,
Utilize 100 input of tansig activation primitive and gradient descent method building, 50 hidden neurons, three layers of nerve net of 11 outputs
Network structure extracts the weight and biasing of neural network, builds three-layer neural network when recognition correct rate reaches 95% deconditioning
Forward-propagating model carries out online recognition, carries out the judgement of flame combustion state, and this area of flame using a period of time is special
Levy that method dimension as input is high, acquisition contains much information, feature identification is accurate.
Three-layer neural network identifies that discrimination is very high in the case where training set is complete to training set sample, and
With certain associative ability, there is very strong non-linear mapping capability and network structure flexible.Therefore for boiler system
Combustion Flame Recognition Using have very strong resolution capability, the three-layer neural network discrimination built to flame combustion state know
Other accuracy rate is 95% or more, and mistake occurs to can satisfy pot in industry spot in the transition period of adjacent phases
The safe operation requirement of furnace, better than the existing judgment method based on temperature sensor and light intensity sensor.
It calls Qcamera function library to acquire flame video using Qt platform, and the flame image real-time display of acquisition is arrived
In Qt graphical interfaces, the area change trend during flame combustion is shown to Qt figure using Qcustomplot function library
In interface, and by judgment result displays into graphical interfaces.It can be measured in real time, carry out flame combustion state detection, increase
Strong safety and reliability.
This method is contained much information using the video of one section of flame combustion process as basis of characterization, and system real time is good, pacifies
Quan Xinggao, Network Recognition rate is 95% or more, and mistake occurs that it is existing to can satisfy industry in the transition period of adjacent phases
The safe operation requirement of boiler in.
Detailed description of the invention
Fig. 1 is the flow diagram of flame combustion state detection method of the present invention.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
A kind of flame combustion state detection method, specifically includes the following steps:
Firstly, extracting a flame from lighting to normal combustion again to the continuous flame video for extinguishing overall process, will extract
The image information of view carries out pretreatment and extracts area of flame feature;Known using the area of flame feature as flame combustion state
Other foundation constitutes one group of three-dimensional matrice, data volume is much larger than single since flame video is made of one group of continuous flame picture
Width flame picture, therefore information characteristics contain much information compared with the flame characteristic of a picture, characteristic quantity is big, characteristic information connects
Feature continuous, real-time is good, therefore identify that accurate, accuracy is high;
Secondly, being greater than the continuous 100 frame area of flame feature of 10 seconds videos as the input of neural network for any one section
Vector, constructs three-layer neural network, and target is to identify 11 flame status;With the corresponding state of maximum probability in output state
As final identification state;Due to having length requirement to input data, usual camera frequency acquisition is 20 frames or more, to protect
Input sample quantity is demonstrate,proved, input video time requirement is greater than 10 seconds;
Third, graphical interface of user
Picture collection control, the off-line learning in MATLAB software, according to existing flame video are carried out using Qt platform
Continuous picture extracts a flame from lighting to normal combustion again to the area of flame feature extinguished in overall process, training nerve net
Network parameter constructs neural network Positive Propagation Algorithm in Qt platform, is bringing training neural network parameter into neural network just
To propagation formula yi=σ (w1,klk+b1,i)w2,k+b2,iIn, online recognition shows the fire at current time in graphical interface of user
Flame image and area of flame change curve, and recognition result is shown simultaneously, and is saved to area data and judging result;
Recognition result according to output vector maximum probability principle.
The image information of the view is pre-processed, and color image is converted to gray scale picture;Gray scale is defined later
Threshold value is 0.5, and the pixel point image using gray value in grayscale image greater than 0.5 utilizes expansion-corrosion as the area features of flame
Algorithm is eliminated because of the hollow out caused by shaking with flue dust interference, and the continuous flame area that processing is completed is by sliding mean filter
Processing, using the area value of the continuous 100 frame flame of a period of time as input.
Flame ignition process is divided into five stages by the building three-layer neural network, and extinguishing process is equally divided into five
A stage such as is fed to boiler internal by stationary process and in stationary process, is vented at continuous flame face when operation
Product value is demarcated as plateau, and using 100 input of tansig activation primitive and gradient descent method building, 50 hidden
Layer neuron, the three-layer neural network of 11 outputs.
A large amount of fire are acquired using to the video acquisition during flame combustion first against a specific furnace flame
Flame variation of the flame from lighting normal operation, the fire when operation such as in normal course of operation being fed, be vented in boiler
Flame variation and change video from flame into extinguishing process is operated normally, using in MATLAB platform to the flame in video
Image zooming-out area features will be lighted and respectively be divided into five stages with extinguishing process, in addition the even running stage amounts to 11 stages
Carry out the calibration of flame status.Using the continuous area value of the flame of 100 frames after treatment as input, state is demarcated as defeated
Out, training three-layer neural network, extracts neuron weight after the discrimination of network is met the requirements and nerve net is brought in biasing into
The calibration part of Combustion Flame Recognition Using is completed in the forward-propagating of network;
The online recognition that Qt platform is carried out to the flame of the burner hearth carries out flame combustion also with fibre optical sensor and camera
Video acquisition is burnt, in Qt platform, with operation processing flame video image identical with training process, extracts each frame figure of flame
The area of flame of picture carries out after dilation erosion operation, sliding mean filter using the area value of continuous 100 frame image as three layers of mind
Input through network extracts the threshold value and biasing of neural network, brings trained neural network into, obtain recognition result;
The each frame flame image and the area of flame at current time of acquisition are shown in Qt user interface, and will be neural
The recognition result of network is shown in graphical interfaces, completes the real-time flame status identification at moment, and to area of flame with
And judging result is saved in historical data, is finally completed real-time Combustion Flame Recognition Using.
The process of attached drawing 1 is illustrated:
A flame is extracted from lighting to just using according to existing flame video continuous picture in off-line learning part first
The area of flame feature extinguished in overall process is arrived in often burning again, and the flame status of continuous 100 picture is demarcated in sliding, as sample
Data randomly select input vector of the sample data as neural network, carry out the training of three-layer neural network, and with remaining sample
Notebook data is tested as test sample, when accuracy reaches 95% or more, it is believed that network training is completed, and three layers of mind are extracted
Parameter through network completes off-line training part.
The real-time video of practical online acquisition is done same treatment by online recognition part, and sliding updates continuous 100 areas
The parameter of three-layer neural network is brought into the forward-propagating formula y of neural network by input of the feature as neural networki=σ
(w1,klk+b1,i)w2,k+b2,i, calculating identification carried out to the data of actual acquisition, and by the judging result of flame, variation tendency,
Shooting video is shown in the interface Qt, completes flame combustion state detection.
Claims (4)
1. a kind of flame combustion state detection method, which is characterized in that specifically includes the following steps:
Firstly, extracting a flame from lighting to normal combustion again to the continuous flame video for extinguishing overall process, view will be extracted
Image information carry out pretreatment extract area of flame feature;Using the area of flame feature as Combustion Flame Recognition Using
Foundation;
Secondly, the continuous 100 frame area of flame feature for being greater than 10 seconds videos for any one section is as the input vector of neural network,
Three-layer neural network is constructed, target is to identify 11 flame status;Using in output state the corresponding state of maximum probability as
Final identification state;
Third, graphical interface of user, using the progress picture collection control of Qt platform, the off-line learning in MATLAB software, according to
Existing flame video continuous picture extracts a flame from lighting to normal combustion again to the area of flame extinguished in overall process
Feature, training neural network parameter, constructs neural network Positive Propagation Algorithm, online recognition, in graphical user in Qt platform
The flame image and area of flame change curve at current time are shown in interface, and show recognition result simultaneously, and to area
Data are saved with judging result.
2. a kind of flame combustion state detection method according to claim 1, which is characterized in that the image of the view
Information is pre-processed, and color image is converted to gray scale picture;Defining gray threshold later is 0.5, by gray scale in grayscale image
Area features of pixel point image of the value greater than 0.5 as flame, are eliminated using expansion-erosion algorithm because shake and flue dust are interfered
Caused hollow out, the continuous flame area that processing is completed is by sliding mean filter processing, with continuous the 100 of a period of time
The area value of frame flame is as input.
3. a kind of flame combustion state detection method according to claim 1, which is characterized in that the neural network is just
To propagation algorithm, forward-propagating formula are as follows:
yi=σ (w1,klk+b1,i)w2,k+b2,i
Wherein yiRepresent the i-th dimension output of output vector, w1,kFor every weight of input layer to hidden layer, lkRepresent input, w1,klk
For matrix operation, b1,iThe items biasing of input layer to hidden layer is represented, σ is activation primitive, w2,kFor the items of hidden layer to output layer
Weight, b2,iThe items biasing of hidden layer to output layer is represented, using tansig activation primitive.
4. a kind of flame combustion state detection method according to claim 1, which is characterized in that described three layers of mind of building
Through network, flame ignition process is divided into five stages, extinguishing process is equally divided into five stages, by stationary process and flat
It continuous flame area value when operation such as fed, be vented during steady to boiler internal as plateau, being demarcated,
Utilize 100 input of tansig activation primitive and gradient descent method building, 50 hidden neurons, three layers of nerve net of 11 outputs
Network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910211096.9A CN109977838B (en) | 2019-03-20 | 2019-03-20 | Flame combustion state detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910211096.9A CN109977838B (en) | 2019-03-20 | 2019-03-20 | Flame combustion state detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109977838A true CN109977838A (en) | 2019-07-05 |
CN109977838B CN109977838B (en) | 2022-09-30 |
Family
ID=67079543
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910211096.9A Active CN109977838B (en) | 2019-03-20 | 2019-03-20 | Flame combustion state detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109977838B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112699801A (en) * | 2020-12-30 | 2021-04-23 | 上海船舶电子设备研究所(中国船舶重工集团公司第七二六研究所) | Fire identification method and system based on video image |
CN113344909A (en) * | 2021-07-01 | 2021-09-03 | 中国石油大学(北京) | Method and device for identifying and displaying coking of flame-permeable high-temperature filter of thermal power boiler |
CN113551531A (en) * | 2021-07-23 | 2021-10-26 | 中国恩菲工程技术有限公司 | Automatic control system of reaction furnace and state monitoring device of reaction furnace |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090315722A1 (en) * | 2008-06-20 | 2009-12-24 | Billy Hou | Multi-wavelength video image fire detecting system |
CN105678295A (en) * | 2016-01-04 | 2016-06-15 | 武汉科技大学 | Method for real-time monitoring gas heating furnace flame on the basis of ROI average image analysis |
CN105678332A (en) * | 2016-01-08 | 2016-06-15 | 昆明理工大学 | Converter steel-making endpoint determination method and system based on flame image CNN recognizing and modeling process |
-
2019
- 2019-03-20 CN CN201910211096.9A patent/CN109977838B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090315722A1 (en) * | 2008-06-20 | 2009-12-24 | Billy Hou | Multi-wavelength video image fire detecting system |
CN105678295A (en) * | 2016-01-04 | 2016-06-15 | 武汉科技大学 | Method for real-time monitoring gas heating furnace flame on the basis of ROI average image analysis |
CN105678332A (en) * | 2016-01-08 | 2016-06-15 | 昆明理工大学 | Converter steel-making endpoint determination method and system based on flame image CNN recognizing and modeling process |
Non-Patent Citations (2)
Title |
---|
张玉杰等: "基于神经网络的电站锅炉燃烧诊断方法研究", 《西安石油大学学报(自然科学版)》 * |
徐望明等: "燃气加热炉火焰视频监控系统设计", 《中国安全生产科学技术》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112699801A (en) * | 2020-12-30 | 2021-04-23 | 上海船舶电子设备研究所(中国船舶重工集团公司第七二六研究所) | Fire identification method and system based on video image |
CN112699801B (en) * | 2020-12-30 | 2022-11-11 | 上海船舶电子设备研究所(中国船舶重工集团公司第七二六研究所) | Fire identification method and system based on video image |
CN113344909A (en) * | 2021-07-01 | 2021-09-03 | 中国石油大学(北京) | Method and device for identifying and displaying coking of flame-permeable high-temperature filter of thermal power boiler |
CN113344909B (en) * | 2021-07-01 | 2023-12-08 | 中国石油大学(北京) | Method and device for identifying and displaying flame penetration height Wen Lvjing coking of thermal power boiler |
CN113551531A (en) * | 2021-07-23 | 2021-10-26 | 中国恩菲工程技术有限公司 | Automatic control system of reaction furnace and state monitoring device of reaction furnace |
Also Published As
Publication number | Publication date |
---|---|
CN109977838B (en) | 2022-09-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109977838A (en) | A kind of flame combustion state detection method | |
JP6968681B2 (en) | Fire monitoring system | |
CN109522819A (en) | A kind of fire image recognition methods based on deep learning | |
CN107025652B (en) | A kind of flame detecting method based on kinetic characteristic and color space time information | |
TW201224823A (en) | A method for combustion flames diagnosis | |
CN101251942B (en) | Underground space fire intelligent detection early alarming and forecasting method and apparatus | |
CN110378265A (en) | A kind of incipient fire detection method, computer-readable medium and system | |
CN105678332A (en) | Converter steel-making endpoint determination method and system based on flame image CNN recognizing and modeling process | |
CN109614908A (en) | A kind of flame combustion state detection system and detection method | |
CN111007035B (en) | Method, system and device for detecting concentration of high-temperature gas in secondary combustion chamber based on convolutional neural network and storage medium | |
CN108932711A (en) | Image analysis nerve network system | |
CN105975991B (en) | A kind of improved extreme learning machine fire kind recognition methods | |
CN108389359A (en) | A kind of Urban Fires alarm method based on deep learning | |
Hossain et al. | Wildfire flame and smoke detection using static image features and artificial neural network | |
CN102323213B (en) | On-line recognition method for flame free radial based fuel type | |
CN101316371A (en) | Flame detecting method and device | |
CN108897354A (en) | A kind of aluminium fusion process fire box temperature prediction technique based on depth confidence network | |
CN115131937A (en) | Forest fire early warning method based on sensor and deep learning | |
CN110309760A (en) | The method that the driving behavior of driver is detected | |
CN111062350B (en) | Artificial intelligence based firework recognition algorithm | |
CN112881640A (en) | Water quality early warning and testing method and system based on zebra fish stress response | |
CN112733900A (en) | Boiler combustion state stability judgment method based on deep migration learning | |
CN117455828A (en) | Combustion quality detection method based on MLP neural network | |
Yan et al. | Forest Fire Image Intelligent Recognition based on the Neural Network. | |
CN117989559A (en) | Method and system for monitoring flame combustion state of heating furnace |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |