CN115083103A - Multi-band infrared pyroelectric flame detector based on deep learning and detection method - Google Patents
Multi-band infrared pyroelectric flame detector based on deep learning and detection method Download PDFInfo
- Publication number
- CN115083103A CN115083103A CN202210656259.6A CN202210656259A CN115083103A CN 115083103 A CN115083103 A CN 115083103A CN 202210656259 A CN202210656259 A CN 202210656259A CN 115083103 A CN115083103 A CN 115083103A
- Authority
- CN
- China
- Prior art keywords
- data
- deep learning
- unit
- model
- flame
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 29
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 238000012544 monitoring process Methods 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 35
- 238000013136 deep learning model Methods 0.000 claims abstract description 21
- 230000007613 environmental effect Effects 0.000 claims abstract description 21
- 238000002372 labelling Methods 0.000 claims abstract description 18
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 238000000034 method Methods 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 7
- 230000036760 body temperature Effects 0.000 claims description 2
- 238000013527 convolutional neural network Methods 0.000 abstract description 9
- 230000008901 benefit Effects 0.000 abstract description 4
- 238000012549 training Methods 0.000 description 19
- 239000000779 smoke Substances 0.000 description 11
- 238000004422 calculation algorithm Methods 0.000 description 7
- 230000005670 electromagnetic radiation Effects 0.000 description 6
- 239000002245 particle Substances 0.000 description 6
- 230000005855 radiation Effects 0.000 description 6
- 238000005070 sampling Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000000825 ultraviolet detection Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Fire Alarms (AREA)
- Photometry And Measurement Of Optical Pulse Characteristics (AREA)
- Fire-Detection Mechanisms (AREA)
Abstract
The invention discloses a multiband infrared pyroelectric flame detector based on deep learning and a detection method, wherein the detection method comprises the following steps: the system comprises a data acquisition unit, a signal processing unit and a model monitoring unit; the data acquisition unit is used for acquiring and marking environmental data to obtain marked environmental data, wherein the environmental data comprises sunlight data, human body data and flame data; the signal processing unit is used for preprocessing the labeling environment data and constructing a deep learning model; and the model monitoring unit is used for carrying out fire monitoring based on the deep learning model to obtain a monitoring result. Through the technical scheme, the advantages of the convolutional neural network in the aspect of image processing are fully utilized, and the accuracy rate of flame identification can be improved.
Description
Technical Field
The invention belongs to the technical field of fire detectors, and particularly relates to a multiband infrared pyroelectric flame detector based on deep learning and a detection method.
Background
The fire detector is one of the most important components of a fire detection system, and the most common components comprise a photoelectric induction smoke detector, a particle smoke detector, a temperature-sensitive detector, an image type fire detector and the like. The photoelectric and particle type smoke detector mainly detects smoke particles and realizes alarm, the temperature detector mainly carries out fire alarm by sensing temperature change, and the three methods have certain delay, so that the smoke particles or the temperature change caused by fire can respond only when reaching the detector. The image type fire detector is not mature enough as a novel special fire detector, and a certain false alarm rate and a certain missing report rate exist at present, so that the image type fire detector cannot well meet the requirements of users.
In the prior art, an infrared fire detection system records temperature changes in different vision fields of a field within a period of time to detect early or late stage fires, and judges whether the system is a prompt for impending fires or other related conditions based on the temperature changes. However, the detection system simply judges whether a fire occurs according to temperature changes, does not perform judgment and analysis under various interference conditions such as ambient temperature and other heat sources, and may have a certain false alarm.
Disclosure of Invention
The invention aims to provide a multiband infrared pyroelectric flame detector based on deep learning, which aims to solve the problems in the prior art.
In order to achieve the above object, the present invention provides a multiband infrared pyroelectric flame detector based on deep learning, comprising: the system comprises a data acquisition unit, a signal processing unit and a model monitoring unit;
the data acquisition unit is used for acquiring and marking environmental data to obtain marked environmental data, wherein the environmental data comprises sunlight data, human body data and flame data;
the signal processing unit is used for analyzing the labeling environment data based on a deep learning model;
and the model monitoring unit is used for carrying out fire monitoring based on the deep learning model to obtain a monitoring result.
Preferably, the data acquisition unit includes: the system comprises a plurality of infrared pyroelectric sensors and data marking units;
the infrared pyroelectric sensor is used for detecting electromagnetic wave bands in the environment to obtain environment data;
and the data labeling unit is used for labeling the environment data to obtain labeled environment data.
Preferably, the electromagnetic band comprises: the electromagnetic band of daylight data, the electromagnetic band of human body data, and the electromagnetic band of flame data.
Preferably, the signal processing unit includes: a signal preprocessing unit and a signal analyzing unit;
the signal preprocessing unit is used for preprocessing the labeling environment data to obtain a two-dimensional data frame;
and the signal analysis unit is used for analyzing the two-dimensional data frame based on the trained deep learning model.
Preferably, the model monitoring unit includes: the system comprises an alarm output unit and an environmental data monitoring unit;
the alarm output unit is used for outputting an alarm signal;
and the environment data monitoring unit is used for monitoring the environment data.
Preferably, the device also comprises an alarm display unit and a power supply unit;
the alarm display unit is used for displaying the alarm signal;
and the power supply unit is used for supplying power to the data acquisition unit, the signal processing unit, the model monitoring unit and the alarm display unit.
On the other hand, in order to achieve the technical purpose, the invention provides a detection method of a multiband infrared pyroelectric flame detector based on deep learning, which comprises the following steps:
collecting and marking environmental data to obtain marked environmental data, wherein the environmental data comprises sunlight data, human body temperature data and flame data;
analyzing the labeling environment data based on a deep learning model;
and carrying out fire monitoring based on the deep learning model to obtain a monitoring result.
Preferably, the process of collecting and labeling the environmental data includes:
detecting the electromagnetic wave band in the environment to obtain environment data; labeling the environment data to obtain labeled environment data; wherein the electromagnetic band comprises: the electromagnetic band of the daylight data, the electromagnetic band of the human body data, and the electromagnetic band of the flame data.
Preferably, the process of analyzing the annotation environment data based on the deep learning model comprises:
preprocessing the labeling environment data to obtain a two-dimensional data frame; and analyzing the two-dimensional data frame based on the trained deep learning model.
The invention has the technical effects that:
according to the invention, the data acquisition unit is used for acquiring and marking environmental data, so that the electromagnetic radiation information of fire flames can be acquired more comprehensively, and the most common human body and sunlight interference in detection can be effectively eliminated;
according to the method, the marked environment data are preprocessed through the signal processing unit and a deep learning model is built, compared with a common threshold value method, the flame and other interferences in a complex environment can be more accurately and effectively distinguished, compared with the traditional one-dimensional signal analysis, time domain information is added, the huge advantage of a convolutional neural network in the aspect of processing images is utilized, and the accuracy of recognition can be improved by further training the flame recognition model based on a powerful pre-training model;
according to the invention, the model monitoring unit is used for carrying out fire monitoring based on the deep learning model to obtain a monitoring result, and compared with a smoke detector, once open fire occurs, the flame radiation can be monitored more rapidly, and the non-contact detection mode can give an alarm more rapidly than a photoelectric smoke detector, a particle smoke detector and a temperature detector, so that more time is won for later-stage fire early warning, and the fire loss is reduced.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the application and, together with the description, serve to explain the application and are not intended to limit the application. In the drawings:
FIG. 1 is a diagram of a multi-band pyroelectric sensor fire flame detector in accordance with an embodiment of the present invention;
FIG. 2 is an external view of a multi-band pyroelectric sensor fire flame detector in an embodiment of the present invention;
FIG. 3 is a flow chart of a multi-band pyroelectric sensor fire flame detection algorithm in an embodiment of the present invention;
FIG. 4 is a flow chart of the operation of a multiband pyroelectric sensor fire flame detector in an embodiment of the present invention;
FIG. 5 is a schematic diagram of model input data for a multiband pyroelectric sensor fire flame detector in an embodiment of the present invention;
201-a digital signal processor chip, 202-a reserved serial signal input interface, 203-a digital signal processing unit power interface, 204-a pyroelectric sensor unit, 205-a signal output serial port, 206-an LED alarm lamp, 207-a buzzer, 208-a host switch key and 209-a detector panel.
Detailed Description
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer-executable instructions and that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different than presented herein.
Example one
As shown in fig. 1, the present embodiment provides a multiband infrared pyroelectric flame detector based on deep learning, which includes: the device can detect infrared radiation signals of 3 different wave bands, outputs 3 paths of infrared pyroelectric sensor monitoring signals and one path of ultraviolet signals, and the size of a signal value represents whether a radiation source of a corresponding wave band exists in a monitored environment.
In some embodiments, the power circuit unit is externally connected with 5V voltage input, and is mainly used for the infrared pyroelectric sensor unit, the digital signal processing unit, the serial interface unit and the alarm unit after voltage division;
in some embodiments, the infrared pyroelectric sensor unit comprises 3 pyroelectric sensors, and the judgment is performed by detecting a unique electromagnetic wave band emitted by flame during combustion, so that the purpose of detecting the flame is achieved. The detection of the electromagnetic wave with the wavelength of 5.0 mu m by the sensor is mainly used for detecting whether sunlight exists; the detection of 4.4 μm wavelength is mainly used for detecting whether flame exists or not so as to fundamentally judge whether fire occurs or not; detecting electromagnetic waves with the wavelength of 3.8 mu m, and mainly detecting whether a human body exists or not; in addition, the sensor unit also provides data collected by one path of ultraviolet detection tube to assist flame identification;
in some embodiments, the signal processing part of the digital signal processing unit comprises a digital signal processor, can process multi-channel input data in real time, can perform real-time signal preprocessing and signal analysis based on a deep learning algorithm, and has multi-channel serial port data output;
in some embodiments, the alarm unit comprises a buzzer for outputting the alarm signal and an LED lamp alarm module.
As shown in fig. 1, the specific structure of the detector is as follows:
the fire flame detector of the multiband pyroelectric sensor based on deep learning comprises five parts, namely a power circuit unit, a pyroelectric sensor unit, a digital signal processing unit, a first serial interface unit, a second serial interface unit and an alarm unit. The power supply circuit unit is used for supplying power to the digital signal processing unit, the pyroelectric sensor unit and the alarm unit; the pyroelectric sensor unit comprises four pyroelectric sensors with different wave bands, and is mainly used for sensing electromagnetic radiation in the environment, converting the electromagnetic radiation into an electric signal and outputting the electric signal to the digital signal processing unit; the signal input interface is used for receiving multi-channel signals of the pyroelectric sensor unit, reserving a serial interface for receiving data of other detection equipment and the like; the digital signal processing unit is used for analyzing the pyroelectric sensor unit and other detection signals; the signal output interface is a serial port used for outputting a fire flame detection result after processing; the alarm unit comprises an LED lamp and a buzzer.
As shown in fig. 2, the appearance of the fire flame detector of the multiband pyroelectric sensor based on deep learning specifically includes: a digital signal processor chip 201; a reserved serial signal input interface 202; the power interface 203 of the digital signal processing unit is used for supplying power to the digital signal processing unit, the pyroelectric sensor unit, the buzzer and the LED alarm lamp; a pyroelectric sensor unit 204 including a plurality of pyroelectric sensors therein; the signal output serial port 205 outputs data such as system state, fault information, alarm information and the like; an LED warning light 206; a buzzer 207; a host switch button 208; a detector panel 209 on which all components are integrated.
Examples of practical applications are as follows:
the embodiment is suitable for places where fire flame detectors need to be installed, such as server rooms, hospitals, prisons, detention houses, nuclear power stations and the like.
1. Hardware installation and system establishment: the multiband pyroelectric sensor fire flame detector shown in fig. 1 and fig. 2 is integrated according to a common packaging mode of a point type fire detector, the detector is connected to a power supply on the ceiling of a building according to the installation mode of the common point type fire detector, the detector is started to test, a signal output serial port 205 outputs the working state of the detector, the detector can be connected to a notebook computer through a serial line, and whether the hardware of the detector works normally or not is checked on the computer. The invention can also be connected with a network management system to upload alarm data to form an intelligent security system. The detector can also be connected with a notebook computer in a wireless mode, so that the test and data access are facilitated.
2. Device initialization: after the system is started, all components are initialized, each device of the system performs self-checking, and if a fault occurs, fault information is output through the serial port 205. The alarm signal is sent by the interrupt program, the system controls the on-off of the corresponding alarm serial port, if the alarm is given, the corresponding interrupt service program is called, and the sound-light alarm is sent; and if the self-checking is normal, the monitoring system starts to work.
3. The working process of the device is as follows: when the digital signal processing unit detects that fire flames appear, audible and visual alarm information is sent out to inform related personnel to check monitoring information of the fire flame detector, and if the fire flames do exist, the related personnel can take corresponding measures.
4. And (3) a fire detection algorithm: first, model training is performed. Collecting electromagnetic radiation data with flame and without flame under different working conditions, marking the data in a digital signal processing unit, then coding and normalizing the data, inputting the training data into a designed image recognition data classification model based on a convolutional neural network for model training, testing the performance of the model, stopping training if the performance meets the actual use requirement, and storing the model; if the performance does not meet the actual use requirement, training sample data is added and model parameters are corrected until the performance of the model meets the requirement. When the fire disaster classification method is used, the trained classification model is called to identify, and finally, whether the fire disaster classification result is obtained. If fire flame appears, alarm information is sent out through the serial port, and sound and light alarm is sent out.
Example two
The embodiment provides a method for identifying flame of a multiband pyroelectric sensor based on deep learning, which comprises the following steps:
step one, data acquisition, namely acquiring sample data X under different working conditions and adding marking information with flame or without flame;
secondly, preprocessing data, namely decoding the original data, carrying out data normalization, and constructing a two-dimensional data frame required by a model by using data output by a multi-path sensor within a period of time;
step three, model design and training, wherein a deep learning algorithm is designed and a model is trained by using prepared data;
and step four, calling the model, detecting fire flames, and giving an alarm when the flames are detected.
In some embodiments, step one collects raw data X ═ X under different operating conditions through multi-band pyroelectric sensor hardware 1 ,x 2 ,x 3 ,x 4 },x i Respectively corresponding to four different signals provided by the sensor unit. Collecting sample data under different set conditionsIncluding the 6 types: the data of the sensor are the datum data of the sensor under the condition of no interference, the data of the flame under the condition of human body radiation interference, the data of the flame under the condition of sunlight interference, the data of the flame under the condition of no interference, the data of the flame under the condition of human body radiation interference and the data of the flame under the condition of sunlight interference, wherein the datum of the flame is marked as 0, and the datum of the flame is marked as 1.
In some embodiments, the step of preprocessing the raw data, the data obtained from the sensor is 16-system codes transmitted back by a serial port, and multiple data are needed to decode, the raw data is transmitted in units of frames, wherein one frame of data contains 10 bytes, such as AA 0904080307020100 EE. AA and EE respectively represent a frame head and a frame tail of the frame data, 09 multiplied by 256+04 represents an infrared 5.0um wave band value, 08 multiplied by 256+03 represents an infrared 4.4um wave band value, 07 multiplied by 256+02 represents an infrared 3.8um wave band value, 01 multiplied by 256+00 represents an ultraviolet counting value, the decoded data is normalized to 0-1 by using a normalization method, then a 4-channel multiplied by 60 frame two-dimensional array is constructed as a monitoring data frame, and the monitoring data frame is supplied for deep learning model training;
in some embodiments, in step three, a deep learning model is trained, in the present invention, a convolutional neural network algorithm is used for model training, after the model training is completed, a model prediction ═ k (X) is obtained, k (·) is a trained flame recognition model based on a convolutional neural network, and X is an input of the model, that is, X ═ { X ═ is input into the model 1 ,x 2 ,x 3 ,x 4 },x i For 60 data collected by one sensor, prediction is the prediction value of the model, the prediction value is 1 or 0, 1 represents fire, and 0 represents no fire.
In some embodiments, step four calls the model to perform flame detection when the sensor newly returns a data X ═ { X ═ X 1 ',x 2 ',x 3 ',x 4 '} calling the model k (-) and judging whether a fire disaster occurs according to the value of k (X') according to the prediction result.
As shown in FIG. 3, a flow chart of a multi-band pyroelectric sensor fire flame detection algorithm based on deep learning is divided into two parts, namely model training and model calling. In the model training process, firstly, the differences are collectedPyroelectric sensor data X ═ { X under operating conditions 1 ,x 2 ,x 3 ,x 4 },x i Representing the output data of a pyroelectric sensor in a certain wave band, and then marking the data, wherein the data with flame is marked as 1, and the data without flame is marked as 0; decoding data acquired by a sensor, and preparing the data for model training; and designing an image recognition model algorithm based on a convolutional neural network, training the model by using prepared data, testing the model after the model is trained, and if the accuracy is low, improving the performance of the model by increasing training data, correcting model parameters and the like. And (3) storing the model after training the model, calling the trained model after the pyroelectric sensor acquires new data, inputting the data after decoding and normalization for prediction, if the predicted value is 1, indicating that flame exists, giving out an acousto-optic alarm, informing relevant personnel to process, and if the predicted value is 0, continuously acquiring the data for fire monitoring.
As shown in fig. 4, in the operation flow chart of the multiband pyroelectric sensor fire flame detector based on deep learning, after the device is powered on and started, a pyroelectric sensor unit is initialized, data acquisition is performed according to a certain frequency, then the data is transmitted to a digital signal processing unit through a serial port, the data is preprocessed in the digital signal processing unit, a trained K-nearest neighbor data classification algorithm model is called to perform fire flame identification, if flame is judged, an audible and visual alarm is given out to prompt related personnel to process, then fire monitoring is continuously performed, and if flame is not generated, the data is continuously acquired without any processing to perform fire monitoring.
As shown in fig. 5, a model input data schematic diagram of a deep learning based multiband pyroelectric sensor fire flame detector is shown, wherein the horizontal axis represents four different sensors, the vertical axis represents sampling points, and the vertical axis represents 60 sampling points in the diagram, that is, the sampling frequency is 60Hz, and the sampling frequency can also be adjusted according to the actual situation.
EXAMPLE III
The embodiment provides an automatic flame identification method, which utilizes the multiband pyroelectric sensor fire flame detector and a flame identification method based on deep learning, and comprises the following steps:
the method comprises the following steps that firstly, a pyroelectric sensor unit collects environmental data, the sampling frequency can be set according to actual requirements, the data are sent to a digital signal processing unit, and the pyroelectric sensor unit is connected with the digital signal processing unit through a serial interface;
decoding the acquired serial port data in a digital signal processing unit, normalizing the decoded data, and constructing a two-dimensional data frame required by a model by using data output by a multi-channel sensor within a period of time;
calling a trained flame recognition model based on a convolutional neural network, judging whether a fire disaster occurs according to the output of the model, if the output of the model is 1, indicating that the fire disaster occurs, driving an alarm model, and giving out sound and flashing alarm of an LED lamp; if the output of the model is 0, the fire disaster does not happen, and the serial port data is continuously analyzed to monitor the fire disaster.
In some embodiments, more band monitoring data are acquired through the pyroelectric sensors in different bands, more abundant environment monitoring information can be acquired, electromagnetic radiation information of flames can be effectively acquired, whether sunlight interference and personnel interference exist can also be monitored, sensor output data under different working conditions can be collected, marked and analyzed, and more guarantees are provided for improving fire flame identification accuracy under complex environments.
In some embodiments, the method for judging whether a fire disaster occurs by simply setting a threshold value is improved to a method for identifying the fire disaster by relying on a large amount of sample data analysis, constructing a two-dimensional data frame by using multiple paths of acquired one-dimensional data, processing a two-dimensional array similar to an image by using a convolutional neural network, and through a deep learning model.
In some embodiments, the point-type detector structure is used, the point-type fire detector can be arranged and installed according to the most common national standard and related specifications of the point-type fire detector, the current application scene can be better adapted, and the building fire prevention design cost is reduced.
In some embodiments, all units are connected through interfaces, the functional modules are relatively independent, product integration and equipment transformation and upgrading are facilitated, fire incidents can be found rapidly through a deep learning method, alarming is carried out through sound and light, the alarming mode is optional, and the intelligent level of fire alarming is improved.
Compared with the prior art, the embodiment has the advantages that:
compared with the prior art and the method, the invention carries out fire detection based on the multiband pyroelectric sensor, can more comprehensively acquire the electromagnetic radiation information of fire flames, and can effectively eliminate the most common human body and sunlight interference in detection; the method adopts a deep learning method to analyze the multi-channel signals, and can more accurately determine flame and other interferences in a complex environment more effectively compared with a common threshold value method; by constructing a two-dimensional data frame as input data of the model, compared with the traditional one-dimensional signal analysis, time domain information is added, the huge advantage of a convolutional neural network in the aspect of processing images is utilized, and the accuracy of the identification can be improved by further training a flame identification model based on a powerful pre-training model; compared with a smoke detector, once open fire occurs, the flame radiation can be detected more rapidly, the non-contact detection mode can give an alarm faster than a photoelectric smoke detector, a particle smoke detector and a temperature detector, more time is strived for later-stage fire early warning, and fire loss is reduced.
Due to the design of independent coupling of multiple interfaces and functional modules, the practicability of the detection system is improved, and later equipment transformation is facilitated. In the signal processing part, a deep learning method is used for information analysis, so that fire flame detection and identification are realized, and the intellectualization of fire-fighting equipment is promoted.
The above description is only for the preferred embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (9)
1. Multiband infrared pyroelectric flame detector based on deep learning, its characterized in that includes: the system comprises a data acquisition unit, a signal processing unit and a model monitoring unit;
the data acquisition unit is used for acquiring and labeling environmental data to obtain labeled environmental data, wherein the environmental data comprises sunlight data, human body data and flame data;
the signal processing unit is used for analyzing the labeling environment data based on a deep learning model;
and the model monitoring unit is used for carrying out fire monitoring based on the deep learning model to obtain a monitoring result.
2. The deep learning based multiband pyroelectric flame detector of claim 1, wherein the data acquisition unit comprises: the system comprises a plurality of infrared pyroelectric sensors and data marking units;
the infrared pyroelectric sensor is used for detecting electromagnetic wave bands in the environment to obtain environment data;
and the data labeling unit is used for labeling the environment data to obtain labeled environment data.
3. The deep learning based multiband pyroelectric flame detector of claim 2,
the electromagnetic band includes: the electromagnetic band of the daylight data, the electromagnetic band of the human body data, and the electromagnetic band of the flame data.
4. The deep learning based multiband pyroelectric flame detector of claim 1,
the signal processing unit includes: a signal preprocessing unit and a signal analyzing unit;
the signal preprocessing unit is used for preprocessing the labeling environment data to obtain a two-dimensional data frame;
and the signal analysis unit is used for analyzing the two-dimensional data frame based on the trained deep learning model.
5. The deep learning based multiband pyroelectric flame detector of claim 1,
the model monitoring unit includes: the system comprises an alarm output unit and an environmental data monitoring unit;
the alarm output unit is used for outputting an alarm signal;
and the environment data monitoring unit is used for monitoring the environment data.
6. The deep learning based multiband infrared pyroelectric flame detector according to claim 5, characterized by further comprising an alarm display unit and a power supply unit;
the alarm display unit is used for displaying the alarm signal;
and the power supply unit is used for supplying power to the data acquisition unit, the signal processing unit, the model monitoring unit and the alarm display unit.
7. The detection method of the multiband infrared pyroelectric flame detector based on deep learning is characterized by comprising the following steps of:
collecting and marking environmental data to obtain marked environmental data, wherein the environmental data comprises sunlight data, human body temperature data and flame data;
analyzing the labeling environment data based on a deep learning model;
and carrying out fire monitoring based on the deep learning model to obtain a monitoring result.
8. The detection method of the multiband pyroelectric flame detector based on deep learning of claim 7,
the process of collecting and labeling the environmental data comprises the following steps:
detecting the electromagnetic wave band in the environment to obtain environment data; labeling the environment data to obtain labeled environment data; wherein the electromagnetic band comprises: the electromagnetic band of the daylight data, the electromagnetic band of the human body data, and the electromagnetic band of the flame data.
9. The detection method of the multiband infrared pyroelectric flame detector based on deep learning of claim 7,
the process of analyzing the annotation environment data based on the deep learning model comprises the following steps:
preprocessing the labeling environment data to obtain a two-dimensional data frame; and analyzing the two-dimensional data frame based on the trained deep learning model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210656259.6A CN115083103A (en) | 2022-06-10 | 2022-06-10 | Multi-band infrared pyroelectric flame detector based on deep learning and detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210656259.6A CN115083103A (en) | 2022-06-10 | 2022-06-10 | Multi-band infrared pyroelectric flame detector based on deep learning and detection method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115083103A true CN115083103A (en) | 2022-09-20 |
Family
ID=83251194
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210656259.6A Pending CN115083103A (en) | 2022-06-10 | 2022-06-10 | Multi-band infrared pyroelectric flame detector based on deep learning and detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115083103A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102034329A (en) * | 2010-12-29 | 2011-04-27 | 上海大学 | Infrared fire detection method based on multiband and multi-feature |
CN102426753A (en) * | 2011-11-21 | 2012-04-25 | 上海大学 | Infrared multiparameter fire and gas detector |
CN108010254A (en) * | 2017-11-28 | 2018-05-08 | 无锡职业技术学院 | One kind is based on four wave band infrared flame detectors and its flame identification algorithm |
CN109165670A (en) * | 2018-07-12 | 2019-01-08 | 江南大学 | A kind of TS-RBF fuzzy neural network robust fusion algorithm applied to infra red flame identification |
CN109816925A (en) * | 2017-11-20 | 2019-05-28 | 成都宇卓时代科技有限公司 | Infrared fire detection system based on STM32 |
CN212256533U (en) * | 2020-06-29 | 2020-12-29 | 西安光控电子科技有限公司 | High-performance ultraviolet-infrared double-spectrum composite optical fire detector |
CN212256534U (en) * | 2020-06-29 | 2020-12-29 | 西安光控电子科技有限公司 | High-speed false alarm prevention three-infrared flame detector |
CN113160513A (en) * | 2021-04-19 | 2021-07-23 | 杭州舜程科技有限公司 | Flame detection device based on multisensor |
-
2022
- 2022-06-10 CN CN202210656259.6A patent/CN115083103A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102034329A (en) * | 2010-12-29 | 2011-04-27 | 上海大学 | Infrared fire detection method based on multiband and multi-feature |
CN102426753A (en) * | 2011-11-21 | 2012-04-25 | 上海大学 | Infrared multiparameter fire and gas detector |
CN109816925A (en) * | 2017-11-20 | 2019-05-28 | 成都宇卓时代科技有限公司 | Infrared fire detection system based on STM32 |
CN108010254A (en) * | 2017-11-28 | 2018-05-08 | 无锡职业技术学院 | One kind is based on four wave band infrared flame detectors and its flame identification algorithm |
CN109165670A (en) * | 2018-07-12 | 2019-01-08 | 江南大学 | A kind of TS-RBF fuzzy neural network robust fusion algorithm applied to infra red flame identification |
CN212256533U (en) * | 2020-06-29 | 2020-12-29 | 西安光控电子科技有限公司 | High-performance ultraviolet-infrared double-spectrum composite optical fire detector |
CN212256534U (en) * | 2020-06-29 | 2020-12-29 | 西安光控电子科技有限公司 | High-speed false alarm prevention three-infrared flame detector |
CN113160513A (en) * | 2021-04-19 | 2021-07-23 | 杭州舜程科技有限公司 | Flame detection device based on multisensor |
Non-Patent Citations (1)
Title |
---|
张永超: "红外多频火焰探测器设计", 计测技术, no. 2010, pages 135 - 137 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109068099B (en) | Virtual electronic fence monitoring method and system based on video monitoring | |
CN102708647B (en) | Image and multi-band infrared-ultraviolet compound fire disaster detection system and method | |
CN106251568A (en) | A kind of fire alarm safety-protection system based on ultraviolet and image detecting technique and method | |
CN103839373A (en) | Sudden abnormal event intelligent identification alarm device and system | |
CN111190065A (en) | Non-contact on-line monitoring system for power equipment | |
CN104994334A (en) | Automatic substation monitoring method based on real-time video | |
CN209690533U (en) | It is a kind of with data double-way transfer function and remote controlled detector gate | |
CN210603602U (en) | Temperature measurement door capable of identifying identity and system thereof | |
CN203630902U (en) | Improved image fire-hazard detector | |
CN113160513A (en) | Flame detection device based on multisensor | |
CN201091014Y (en) | Fire detecting device | |
CN108986420A (en) | A kind of power distribution cabinet monitoring alarm system | |
CN106887105A (en) | A kind of advance notice of novel road tunnel fire hazard and warning system based on disaster-stricken people's feature | |
CN110930632B (en) | Early warning system based on artificial intelligence | |
CN104594754A (en) | Intelligent quick automatic screening system for infectious disease at public place | |
CN201516026U (en) | Infrared body temperature detecting instrument | |
CN202306757U (en) | Embedded flame detection device based on double-waveband | |
CN111063358A (en) | Early fire early warning and escape indicating system with life recognition function | |
CN209433517U (en) | It is a kind of based on more flame images and the fire identification warning device for combining criterion | |
CN113362545A (en) | Sound and video combined residential community monitoring system | |
CN115083103A (en) | Multi-band infrared pyroelectric flame detector based on deep learning and detection method | |
CN204291227U (en) | Video information process device | |
CN115393340A (en) | AI vision product quality detection system based on 5G algorithm | |
CN114155483A (en) | Monitoring alarm method, device, storage medium and computer equipment | |
CN113570808A (en) | Wireless smoke detector based on ZYNQ7020 |
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: 20220920 |