CN108334902A - A kind of track train equipment room smog fireproof monitoring method based on deep learning - Google Patents
A kind of track train equipment room smog fireproof monitoring method based on deep learning Download PDFInfo
- Publication number
- CN108334902A CN108334902A CN201810107379.4A CN201810107379A CN108334902A CN 108334902 A CN108334902 A CN 108334902A CN 201810107379 A CN201810107379 A CN 201810107379A CN 108334902 A CN108334902 A CN 108334902A
- Authority
- CN
- China
- Prior art keywords
- smoke
- data
- layer
- deep learning
- monitoring method
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013135 deep learning Methods 0.000 title claims abstract description 18
- 239000000779 smoke Substances 0.000 claims abstract description 92
- 230000003137 locomotive effect Effects 0.000 claims abstract description 21
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 230000000644 propagated effect Effects 0.000 claims abstract description 3
- 238000012549 training Methods 0.000 claims description 32
- 238000013527 convolutional neural network Methods 0.000 claims description 29
- 230000002265 prevention Effects 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 17
- 238000011176 pooling Methods 0.000 claims description 14
- 230000009466 transformation Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 6
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 abstract description 5
- 238000004891 communication Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035484 reaction time Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011527 multiparameter analysis Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The track train equipment room smog fireproof monitoring method based on deep learning that the present invention relates to a kind of comprising:Step S40 since the input layer of CNN network models, passes through propagated forward by the test data sample input based on smoke data trained CNN network models;The posterior probability of all kinds of smoke characteristics parameters is obtained in output layer;The posterior probability of obtained all kinds of smoke characteristics parameters is compared with the setting probability threshold value of corresponding class, determines whether rail locomotive equipment room occurs fire according to comparison result by step S50.The present invention can improve the accuracy of fire judging result, reduce from the reaction duration that fire occurs to realization monitoring, and can effectively improve the sensitivity of monitoring.
Description
Technical Field
The invention relates to the technical field of smoke fire prevention monitoring, in particular to a deep learning-based smoke fire prevention monitoring method between rail train equipment.
Background
With the continuous speed increase of railways in China, safe operation becomes more and more important, and therefore higher requirements are put forward on the safety monitoring of locomotives.
The existing method for monitoring the fire prevention between the rail locomotive equipment mainly monitors smoke characteristics, temperature characteristics, video characteristics and the like through corresponding monitoring systems formed by corresponding sensors, and comprises two-bus fire prevention monitoring, multi-parameter real-time fire prevention monitoring and the like based on the rail locomotive.
The fire prevention monitoring based on the two buses of the railway locomotive mainly uses a fire prevention monitoring board card and two buses, wherein the fire prevention monitoring board card is electrically connected with one end of the two buses, the two buses are electrically connected with a plurality of smoke detectors and a plurality of temperature detectors, and the smoke detectors and the temperature detectors are directly connected in parallel for network communication.
The multi-parameter real-time fire prevention monitoring system comprises a smoke and temperature acquisition module, an audio and video acquisition module, a main CPU, a data storage module, a multi-parameter analysis comparison module, a WLAN/3G wireless module, an in-vehicle communication module, a power supply module, a display and input module, a smoke and temperature acquisition module and an audio and video acquisition module which are electrically connected with each other respectively, and analyzes the operation condition of a locomotive through various parameters, multi-communication and wireless communication technologies.
Although the existing method for monitoring the fire between the rail locomotive equipment finally realizes the fire monitoring, the method has the following defects: based on two bus fire prevention monitoring of railway locomotive, real-time fire prevention monitoring of multiparameter etc. all belong to and use corresponding sensor such as smog, temperature, monitor the concentration of smog, the height isoparametric of temperature, data analysis is not meticulous enough, corresponding monitoring result also lacks accuracy and sensitivity, especially when the conflagration takes place in the local range of certain small area between the track locomotive equipment, the method with the help of the sensor of tradition is difficult to detect fast and accurately, therefore it has very big optimization space to follow monitoring reaction duration and sensitivity angle.
Disclosure of Invention
The invention aims to solve the problems in the prior art, and provides a deep learning-based smoke fire prevention monitoring method between rail train equipment, which can reduce the reaction time from fire occurrence to monitoring and effectively improve the monitoring sensitivity.
The purpose of the invention is realized by the following technical scheme:
the invention provides a deep learning-based smoke fire prevention monitoring method between rail train equipment, which comprises the following steps:
step S40, inputting a test data sample based on smoke data into the trained CNN network model, and starting from the input layer of the CNN network model through forward propagation; obtaining posterior probabilities of various smoke characteristic parameters on an output layer;
and step S50, comparing the posterior probability of the obtained various smoke characteristic parameters with the set probability threshold of the corresponding class, and determining whether the fire disaster happens among the rail locomotive equipment according to the comparison result.
More preferably, the method for monitoring smoke fire between rail train equipment further comprises:
step S10, obtaining smoke data including smoke characteristic data and smoke concentration data, wherein the smoke characteristic data includes but is not limited to texture characteristic data, color characteristic data, motion characteristic data and transformation characteristic data;
step S20, grouping the smoke data, wherein one group is used as a training data sample, and the other group is used as a test data sample; performing data completion and arrangement on the training data samples to obtain training data samples which accord with set rules;
and step S30, finishing the training of the CNN network model through the forward propagation and residual backward propagation processes based on the training data samples obtained by sorting to obtain the trained CNN network model.
More preferably, the smoke data comprises:
texture feature data, color feature data, motion feature data, transformation feature data, and smoke density data of smoke.
More preferably, audio and video data including texture feature data, color feature data, motion feature data and transformation feature data of smoke are acquired through a video acquisition unit arranged in the rail locomotive;
smoke concentration data are acquired through a smoke sensor arranged in the rail locomotive.
More preferably, the training data samples account for 80% of the smoke data;
the test data samples accounted for 20% of the smoke data.
More preferably, the forward propagation process in step 30 specifically includes:
starting from an input layer, taking training data samples as input of the input layer; through convolution operation of the convolution layers which are alternately arranged and characteristic integration and classification processing of the pooling layers, an iterative abstract processing process of training data samples is realized, and various integrated and classified smoke local characteristics are obtained; and the local characteristics of various smog obtained in the early stage are integrated through the full connecting layer, and the posterior probability of various smog characteristic parameters is output through the output layer.
More preferably, the back propagation process in step 30 specifically includes:
in the process of backward propagation, the difference between the output layer result obtained after forward propagation and the expected value is calculated through a loss function from the output layer to obtain a corresponding residual error; and (4) the residual error is reversely propagated through a gradient descent method, and trainable parameters of each layer of the convolutional neural network are updated layer by layer.
The technical scheme of the invention can show that the invention has the following technical effects:
according to the invention, the monitored texture characteristics, color characteristics, motion characteristics and transformation characteristics of the smoke are combined with the smoke concentration data of the smoke sensor to be used as the input of a CNN network model (convolutional neural network model), so that the analysis angle is relatively comprehensive, and the obtained analysis result has relatively high accuracy;
in addition, the trained CNN network model has good fault-tolerant capability, parallel processing capability and self-learning capability, has strong adaptability, and can quickly finish the judgment on whether a fire disaster occurs or not under complex environmental information, thereby greatly improving the accuracy of a judgment result, reducing the reaction time from the occurrence of the fire disaster to the realization of monitoring and effectively improving the monitoring sensitivity.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a schematic structural diagram of a CNN network model.
Detailed Description
The technical solution of the present invention will be further described in detail with reference to the accompanying drawings.
Example one
The invention provides a deep learning-based smoke fire prevention monitoring method between rail train equipment, which relies on a convolutional neural network model in a deep learning technology, combines an industrial camera and a sensor, takes multi-dimensional smoke data between rail locomotive equipment acquired by the industrial camera and the smoke sensor as data input of the convolutional neural network model, and finally obtains a judgment result whether fire occurs between the current equipment through a deep learning method. The specific implementation flow of the invention is shown in fig. 1, and comprises the following steps:
step S10, acquiring the collected smoke data, including the feature data and the concentration data of the smoke, wherein the feature data of the smoke includes, but is not limited to, texture feature data, color feature data, motion feature data, and transformation feature data.
Acquiring audio and video data including texture feature data, color feature data, motion feature data, transformation feature data and the like of smoke through a video acquisition unit (such as an industrial camera and a camera) arranged in the rail locomotive; smoke concentration data are acquired through a smoke sensor arranged in the rail locomotive.
Step S20, grouping the collected smoke characteristic data and the collected concentration data, wherein one group is used as training data, and the other group is used as test data; and the training data is subjected to data completion and arrangement to obtain training data samples which accord with set rules.
And dividing the collected smoke characteristic data and the collected concentration data into two parts, wherein 80% of the data is used as a training data sample, and 20% of the data is used as a test data sample for testing the deep learning network model.
And (4) performing data arrangement such as completion on the training data samples, adjusting the training data samples to be uniform in size, and obtaining the regularized training data samples so as to prevent poor-quality data possibly appearing in the training samples from generating adverse effects on the training process and even finally influencing the judgment result of the fire.
And step S30, completing training of the CNN network model (convolutional neural network model) through forward propagation and residual backward propagation processes in sequence based on the training data to obtain the trained CNN network model.
The structure of the CNN network model in step S30 is shown in fig. 2, and includes: input layer, convolution layer, pooling layer, full-link layer and output layer. The convolution layers and the pooling layers are alternately arranged, namely one convolution layer is connected with one pooling layer, the convolution layer is connected with one convolution layer after the pooling layer, and the pooling layer is connected with one pooling layer after the convolution layer. The convolutional layers and the pooling layers are alternately arranged, iterative abstract processing of training data can be achieved, and the CNN network model is stopped when the error reaches a set threshold value.
Each convolution layer extracts various different local features of the previous layer through convolution operation, similar or related features are combined semantically by the pooling layer, various local features extracted earlier are combined by the full connection layer, and the posterior probability of the smoke feature parameters of each category is obtained through the output layer, so that the judgment result of whether a fire disaster occurs is obtained.
In the forward propagation process, starting from an input layer, training data samples are used as input of the input layer; through convolution operation of the convolution layers which are alternately arranged and characteristic integration and classification processing of the pooling layers, an iterative abstract processing process of training data samples is realized, and various integrated and classified smoke local characteristics are obtained; integrating local features of various smog obtained in the early stage through a full connection layer, and classifying the extracted features to obtain input-based probability distribution; and finally, outputting the posterior probability of various smoke characteristic parameters through an output layer.
In the forward propagation process, the convolution operation processing is carried out on the input training data sample by using the following formula:
the meaning of each parameter in formula 1 and formula 2 is:
is the net activation of the mth channel of the convolutional layer by outputting a profile of the smoke characteristics to the previous layerThe result of convolution summation and offset is obtained,is the output of the mth channel of convolutional layer l, f (-) is the activation function,is a bias to the convolved smoke signature,and the connection weight of the channel and the ith channel of the l-1 layer is taken as the connection weight of the channel and the ith channel.
The CNN network model used in the application comprises seven network layers (an input layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, a full-connection layer and an output layer), wherein the first three layers mainly extract low-layer characteristics such as smoke edges and smoke colors, the fourth layer mainly extracts relatively complex local smoke texture characteristics, and the fifth layer starts to extract relatively complete smoke contours, shapes, smoke motion characteristics and smoke transformation characteristics.
In the process of backward propagation, starting from an output layer, calculating the difference between an output layer result obtained after forward propagation and an expected value through a loss function, namely a residual error; and (4) performing back propagation on the residual error by a gradient descent method, and updating trainable parameters of each layer of the convolutional neural network layer by using a weight value updating formula.
The loss function during back propagation is:
wherein the l-th layer is an output layer, tjThe class of the jth smoke data sample is labeled with a true value,the category label of the jth smoke data sample output through the forward propagation network prediction is used for solving the first-order partial derivative of the loss function, and the CNN network weight value updating formula is as follows:
wherein,the connection weight of the channel and the ith channel of the l-1 layer is η, the learning rate is η, and E is the error loss of the current network layer.
And completing the training of the CNN network model through two stages of forward propagation and residual back propagation of the data to obtain the trained CNN network model.
And step S40, inputting the test data into the trained CNN network model, and finally obtaining the posterior probability of each smoke characteristic parameter on an output layer through forward propagation from the input layer of the CNN network model.
The forward propagation process of step S40 specifically includes:
starting from the input layer, taking a test data sample as the input of the input layer; performing iterative abstract processing on the test data sample by the convolution layer and the pooling layer to obtain various integrally classified smoke local characteristics; and the local characteristics of various smog obtained in the early stage are integrated through the full connecting layer, and the posterior probability of various smog characteristic parameters is output through the output layer.
And step S50, comparing the posterior probability of the obtained various smoke characteristic parameters with the set probability threshold of the corresponding class, and determining whether the fire disaster happens among the rail locomotive equipment according to the comparison result.
If the comparison result is less than or equal to zero, determining that no fire disaster occurs among the rail locomotive equipment; and if the comparison result is greater than zero, determining that fire disaster occurs between the rail locomotive equipment, confirming the fire disaster grade according to the specific numerical value of the comparison result, and starting fire disaster rescue of the corresponding grade.
According to the invention, the smoke characteristics between the devices are identified and judged by utilizing the audio and video information acquired by the industrial camera and the smoke concentration data acquired by the smoke sensor and utilizing the deep learning technology aiming at the texture characteristics, the color characteristics, the motion characteristics, the transformation characteristics, the smoke concentration data and the like of smoke in the audio and video information, and finally the smoke fire prevention monitoring between the devices is realized.
Example two
The difference between the second embodiment and the first embodiment is that the second embodiment directly utilizes the trained CNN network model to perform smoke fire monitoring between devices. That is, this second embodiment includes only step S40 and step S50 in the first embodiment described above. The specific implementation of steps S40 and S50 is the same as that described in the first embodiment, and will not be described in detail here.
Although the present invention has been described in terms of the preferred embodiment, it is not intended that the invention be limited to the embodiment. Any equivalent changes or modifications made without departing from the spirit and scope of the present invention also belong to the protection scope of the present invention. The scope of the invention should therefore be determined with reference to the appended claims.
Claims (7)
1. A rail train equipment room smoke fire prevention monitoring method based on deep learning is characterized by comprising the following steps:
step S40, inputting a test data sample based on smoke data into the trained CNN network model, and starting from the input layer of the CNN network model through forward propagation; obtaining posterior probabilities of various smoke characteristic parameters on an output layer;
and step S50, comparing the posterior probability of the obtained various smoke characteristic parameters with the set probability threshold of the corresponding class, and determining whether the fire disaster happens among the rail locomotive equipment according to the comparison result.
2. The deep learning-based smoke and fire monitoring method for the rail train equipment room as claimed in claim 1, wherein the method further comprises:
step S10, obtaining smoke data including smoke characteristic data and smoke concentration data, wherein the smoke characteristic data includes but is not limited to texture characteristic data, color characteristic data, motion characteristic data and transformation characteristic data;
step S20, grouping the smoke data, wherein one group is used as a training data sample, and the other group is used as a test data sample; performing data completion and arrangement on the training data samples to obtain training data samples which accord with set rules;
and step S30, finishing the training of the CNN network model through the forward propagation and residual backward propagation processes based on the training data samples obtained by sorting to obtain the trained CNN network model.
3. The deep learning based rail train equipment room smoke fire monitoring method as claimed in claim 1, wherein the smoke data comprises:
texture feature data, color feature data, motion feature data, transformation feature data, and smoke density data of smoke.
4. The deep learning-based rail train inter-equipment smoke fire monitoring method as claimed in claim 1,
acquiring audio and video data through a video acquisition unit arranged in the rail locomotive, wherein the audio and video data comprise texture characteristic data, color characteristic data, motion characteristic data and transformation characteristic data of smoke;
smoke concentration data are acquired through a smoke sensor arranged in the rail locomotive.
5. The deep learning-based rail train inter-equipment smoke fire monitoring method as claimed in claim 1,
the training data samples account for 80% of the smoke data;
the test data samples accounted for 20% of the smoke data.
6. The deep learning-based smoke fire monitoring method for the rail train equipment room, according to any one of claims 2 to 5, wherein the forward propagation process in the step 30 specifically includes:
starting from an input layer, taking training data samples as input of the input layer; through convolution operation of the convolution layers which are alternately arranged and characteristic integration and classification processing of the pooling layers, an iterative abstract processing process of training data samples is realized, and various integrated and classified smoke local characteristics are obtained; and the local characteristics of various smog obtained in the early stage are integrated through the full connecting layer, and the posterior probability of various smog characteristic parameters is output through the output layer.
7. The deep learning-based smoke fire monitoring method for the rail train equipment room, according to any one of claims 2 to 5, wherein the back propagation process in the step 30 specifically includes:
in the process of backward propagation, the difference between the output layer result obtained after forward propagation and the expected value is calculated through a loss function from the output layer to obtain a corresponding residual error; and (4) the residual error is reversely propagated through a gradient descent method, and trainable parameters of each layer of the convolutional neural network are updated layer by layer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810107379.4A CN108334902A (en) | 2018-02-02 | 2018-02-02 | A kind of track train equipment room smog fireproof monitoring method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810107379.4A CN108334902A (en) | 2018-02-02 | 2018-02-02 | A kind of track train equipment room smog fireproof monitoring method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108334902A true CN108334902A (en) | 2018-07-27 |
Family
ID=62927847
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810107379.4A Pending CN108334902A (en) | 2018-02-02 | 2018-02-02 | A kind of track train equipment room smog fireproof monitoring method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108334902A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109271906A (en) * | 2018-09-03 | 2019-01-25 | 五邑大学 | A kind of smog detection method and its device based on depth convolutional neural networks |
CN109344683A (en) * | 2018-08-03 | 2019-02-15 | 昆明理工大学 | A kind of the fire-smoke detection system and its control method of artificial intelligence |
CN109858456A (en) * | 2019-02-18 | 2019-06-07 | 中国铁路沈阳局集团有限公司科学技术研究所 | A kind of rolling stock status fault analysis system |
CN110119682A (en) * | 2019-04-04 | 2019-08-13 | 北京理工雷科电子信息技术有限公司 | A kind of infrared remote sensing Image Fire point recognition methods |
WO2021212443A1 (en) * | 2020-04-20 | 2021-10-28 | 南京邮电大学 | Smoke video detection method and system based on lightweight 3d-rdnet model |
CN115830787A (en) * | 2022-10-12 | 2023-03-21 | 蒋道莲 | Intelligent fire fighting system and method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070188336A1 (en) * | 2006-02-13 | 2007-08-16 | Axonx, Llc | Smoke detection method and apparatus |
CN106682694A (en) * | 2016-12-27 | 2017-05-17 | 复旦大学 | Sensitive image identification method based on depth learning |
CN107025443A (en) * | 2017-04-06 | 2017-08-08 | 江南大学 | Stockyard smoke monitoring and on-time model update method based on depth convolutional neural networks |
CN107609470A (en) * | 2017-07-31 | 2018-01-19 | 成都信息工程大学 | The method of outdoor fire disaster early-stage smog video detection |
-
2018
- 2018-02-02 CN CN201810107379.4A patent/CN108334902A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070188336A1 (en) * | 2006-02-13 | 2007-08-16 | Axonx, Llc | Smoke detection method and apparatus |
CN106682694A (en) * | 2016-12-27 | 2017-05-17 | 复旦大学 | Sensitive image identification method based on depth learning |
CN107025443A (en) * | 2017-04-06 | 2017-08-08 | 江南大学 | Stockyard smoke monitoring and on-time model update method based on depth convolutional neural networks |
CN107609470A (en) * | 2017-07-31 | 2018-01-19 | 成都信息工程大学 | The method of outdoor fire disaster early-stage smog video detection |
Non-Patent Citations (1)
Title |
---|
周泊龙: "《基于视频图像的火灾烟雾检测算法研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109344683A (en) * | 2018-08-03 | 2019-02-15 | 昆明理工大学 | A kind of the fire-smoke detection system and its control method of artificial intelligence |
CN109271906A (en) * | 2018-09-03 | 2019-01-25 | 五邑大学 | A kind of smog detection method and its device based on depth convolutional neural networks |
CN109858456A (en) * | 2019-02-18 | 2019-06-07 | 中国铁路沈阳局集团有限公司科学技术研究所 | A kind of rolling stock status fault analysis system |
CN110119682A (en) * | 2019-04-04 | 2019-08-13 | 北京理工雷科电子信息技术有限公司 | A kind of infrared remote sensing Image Fire point recognition methods |
WO2021212443A1 (en) * | 2020-04-20 | 2021-10-28 | 南京邮电大学 | Smoke video detection method and system based on lightweight 3d-rdnet model |
CN115830787A (en) * | 2022-10-12 | 2023-03-21 | 蒋道莲 | Intelligent fire fighting system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108334902A (en) | A kind of track train equipment room smog fireproof monitoring method based on deep learning | |
CN110287552B (en) | Motor bearing fault diagnosis method and system based on improved random forest algorithm | |
CN110309886B (en) | Wireless sensor high-dimensional data real-time anomaly detection method based on deep learning | |
CN103914735B (en) | A kind of fault recognition method and system based on Neural Network Self-learning | |
US20200402221A1 (en) | Inspection system, image discrimination system, discrimination system, discriminator generation system, and learning data generation device | |
CN105760835A (en) | Gait segmentation and gait recognition integrated method based on deep learning | |
CN104751110A (en) | Bio-assay detection method and device | |
CN110040594B (en) | Convolutional neural network-based elevator operation detection system and method | |
CN103558519A (en) | GIS partial discharge ultrasonic signal identification method | |
CN110727669B (en) | Electric power system sensor data cleaning device and cleaning method | |
CN111678699B (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
CN111860775A (en) | Ship fault real-time diagnosis method based on CNN and RNN fusion | |
CN106599367A (en) | Method for detecting abnormal state of spacecraft | |
CN116563645A (en) | Model compression method for target-oriented detection by combining iterative pruning and knowledge distillation | |
CN112288700A (en) | Rail defect detection method | |
CN117975372B (en) | Construction site safety detection system and method based on YOLOv and transducer encoder | |
CN112036269A (en) | Fall detection method and device, computer equipment and storage medium | |
CN107490758A (en) | Modularization BP neural network circuit failure diagnosis method based on fault propagation | |
CN118052793A (en) | Real-time monitoring system and method for plush toy production process | |
CN113870199A (en) | Method for identifying defect detection of aircraft skin | |
CN111598807B (en) | Automobile part detection data sharing system and method based on block chain | |
CN112132088A (en) | Inspection point location missing inspection identification method | |
CN104980442A (en) | Network intrusion detection method based on element sample sparse representation | |
CN114842315B (en) | Looseness-prevention identification method and device for lightweight high-speed railway hub gasket | |
CN114781427B (en) | Wind generating set antifriction bearing fault diagnosis system based on information fusion technology |
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 |
Application publication date: 20180727 |
|
RJ01 | Rejection of invention patent application after publication |