Intelligent fire-fighting early warning system and early warning method

Abstract

The invention relates to the technical field of intelligent fire protection, and discloses an intelligent fire protection early warning system and an early warning method. The method comprises the following steps: acquiring environment data based on various sensors to generate a fire early warning data set, filtering abnormal data by using a preprocessing algorithm, performing multi-level confirmation on fire features through an image analysis module and a sound analysis module, finally generating a fire confirmation signal, triggering an alarm system and activating fire fighting equipment. In the prior art, a fire hazard early warning method only depends on a single sensor or a single data source, and especially under a complex and changeable environment condition, efficient and accurate fire hazard early warning cannot be realized. Due to the fact that multi-dimensional and intelligent fire detection is achieved through multi-source data fusion, convolutional neural network image recognition and fast Fourier transform sound analysis, the problem of false alarm or missing alarm caused by instability of a single data source is effectively avoided, and the accuracy and response speed of fire early warning are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent fire fighting, and in particular relates to an intelligent fire fighting early warning system and an early warning method. Background Art

[0002] At present, traditional fire warning systems are mostly based on a single sensor, such as temperature or smoke sensors for fire detection, but this single detection method often shows shortcomings in complex environments. For example, in high temperature or dusty environments, temperature and smoke sensors are easily interfered, resulting in frequent false alarms and missed alarms. Existing technologies cannot fully meet the needs of fire warnings with high accuracy, low false alarm rate and rapid response. Therefore, there is an urgent need for a method that can still achieve high-precision fire warnings in complex and changing environmental conditions. This method should combine multiple data sources and have intelligent analysis and judgment functions, so as to improve the reliability and response speed of fire detection in different environments. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to propose an intelligent fire warning system and warning method, aiming to solve the problem that the fire warning method in the prior art only relies on a single sensor or a single data source, especially under complex and changeable environmental conditions, and cannot achieve efficient and accurate fire warning.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides an intelligent fire early warning method, comprising:

[0005] Step S10: Generate a fire warning data set based on environmental data collected by multiple sensors, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U;

[0006] Step S20: For the i-th item of data in the fire warning data set, the smoke density value in the smoke density information Y is With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed;

[0007] Step S30: After obtaining the preliminary data set X, the monitoring image data is acquired in real time, and the flame features are recognized by using the pre-trained convolutional neural network CNN. The monitoring image data is input into the pre-trained CNN model, and the CNN model performs image detection based on the flame features.

[0008] When the flame feature is detected, the image confirmation instruction information is pushed to the next step;

[0009] If no flame feature is detected for 10 seconds, the process returns to step S10;

[0010] Step S40: after receiving the image confirmation instruction information, obtaining the environmental sound data and performing fast Fourier transform FFT analysis on the environmental sound data to obtain the sound characteristic frequency signal;

[0011] Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step;

[0012] When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10;

[0013] Step S50: Generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire fighting equipment.

[0014] Preferably, in step S30, the loss function used in the pre-training process of the convolutional neural network CNN is: Loss=- 1 N ∑ j=1 N [y j ∙log( y ̂ j )+ (1-y j )∙log( 1- y ̂ j )] + λ T σ∙+ λ U ∙τ ;

[0015] The sample data in the pre-training process is the historical monitoring image data acquired in advance. is the loss function used in the pre-training process, is the total number of samples, j is the jth sample, is the true label of the jth sample, is the predicted probability of the jth sample, is the normalized outlier of temperature information, is the normalized outlier of air quality information, and are the weight coefficients of temperature anomaly term and air quality anomaly term respectively.

[0016] Preferably, The calculation formula is: ;

[0017] in, is the temperature information of the jth sample, is the reference value corresponding to the temperature information, and are the upper and lower limits of the temperature information, used for normalization processing.

[0018] Preferably, The calculation formula is: ;

[0019] in, is the air quality information of the jth sample, is the reference value corresponding to the air quality information, and are the upper and lower limits of air quality information, used for normalization processing.

[0020] Preferably, in step S10, the frequency of collecting environmental data is once per second.

[0021] Preferably, in step S10, the temperature sensor, the smoke sensor and the air quality sensor are all connected to the main control background via a wireless connection.

[0022] Preferably, in step S20, the smoke concentration deviation value is calculated by smoothing processing.

[0023] The present invention also provides an intelligent fire warning system comprising:

[0024] A data acquisition module, used to collect environmental data based on multiple sensors to generate a fire warning data set, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U;

[0025] The data processing module is used to convert the smoke density value in the smoke density information Y into the smoke density value in the smoke density information Y for the i-th data in the fire warning data set. With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed;

[0026] The image analysis module is used to obtain the monitoring image data in real time after obtaining the preliminary data set X, and use the pre-trained convolutional neural network CNN to identify the flame characteristics of the monitoring image data, and input the monitoring image data into the pre-trained CNN model, and the CNN model performs image detection based on the flame characteristics;

[0027] When the flame feature is detected, the image confirmation instruction information is pushed to the next step;

[0028] If no flame feature is detected for 10 seconds, the process returns to step S10;

[0029] The sound analysis module is used to obtain environmental sound data and perform fast Fourier transform (FFT) analysis on the environmental sound data to obtain a sound characteristic frequency signal after receiving the image confirmation instruction information;

[0030] Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step;

[0031] When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10;

[0032] The alarm control module is used to generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire-fighting equipment.

[0033] The beneficial effect of the present invention is that compared with the fire warning method in the prior art that only relies on a single sensor or a single data source, especially under complex and changeable environmental conditions, it is impossible to achieve efficient and accurate fire warning. Because the present application realizes multi-dimensional and intelligent fire detection through multi-source data fusion, convolutional neural network image recognition and fast Fourier transform sound analysis, it effectively avoids the problem of false alarms or missed alarms caused by the instability of a single data source, and improves the accuracy and response speed of fire warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A schematic flow chart of a first embodiment of an intelligent fire warning method provided by the present invention.

[0036] Figure 2 A system schematic diagram of an intelligent fire warning method provided by the present invention.

[0037] Figure 3 A schematic diagram of equipment for an intelligent fire warning method provided by the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Embodiment 1: Figure 1 , which is a flow chart of the first embodiment of the intelligent fire warning method of the present invention, and proposes the first embodiment of the intelligent fire warning method of the present invention.

[0040] In a first embodiment, the intelligent fire warning method includes:

[0041] Step S10: Generate a fire warning data set based on environmental data collected by multiple sensors, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U;

[0042] It should be noted that temperature information T refers to the real-time temperature data in the environment, which is collected through temperature sensors. When the temperature data rises abnormally, it may mean the occurrence of a fire. Smoke concentration information Y represents the smoke content in the air, which is obtained through smoke sensors. When the smoke concentration exceeds the normal threshold, it may indicate the presence of a fire source. Air quality information U includes the concentration of particulate matter or harmful gases in the air, which is obtained through air quality sensors. Significant changes in air quality may also be a sign of a fire.

[0043] Understandably, in the event of a fire, temperature information often rises rapidly, smoke concentrations may increase significantly, and air quality deteriorates rapidly. Combining this data together can enable early detection of fires and reduce the possibility of false alarms.

[0044] Step S20: For the i-th item of data in the fire warning data set, the smoke density value in the smoke density information Y is With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed;

[0045] It should be noted that the preset smoke concentration threshold It is set according to the normal smoke concentration level of the environment to distinguish between normal conditions and potential fire risks. The threshold setting should take into account the specificity of the environment to reduce the possibility of false alarms.

[0046] It is understandable that the smoke concentration deviation value Indicates the difference between the current smoke density and the normal level. When the deviation value is significantly higher than the preset range, it means that the smoke density may increase due to abnormal factors, indicating that there may be a fire.

[0047] It should be understood that the smoke concentration deviation value The calculation formula is: ;

[0048] in, is the actual smoke concentration value of the i-th data, is the data mean corresponding to the actual smoke concentration value from the i-th item to the i-th item, which is used to smooth the latest n data points and eliminate the influence of accidental noise. is the normal smoke concentration threshold.

[0049] For example, assuming the normal smoke concentration threshold of the environment is is 50 ppm (parts per million), which means that there may be a fire risk when the smoke concentration exceeds 50 ppm. In the i-th data, the smoke concentration detected by the smoke sensor is The value of the data point is 80 ppm, and the five most recent data points are used for smoothing. Assuming that the values ​​of these data points are 48, 50, 52, 49 and 80, their mean is 55.8. The deviation value is calculated according to the formula to obtain 1674. When the deviation value exceeds the system preset deviation range, for example, set to 1000, the data is recorded as abnormal data and used to generate a preliminary data set. By introducing linear interpolation and smoothing, this method can effectively reduce false alarms caused by short-term fluctuations or accidental noise, and improve the accuracy of fire detection and the robustness of the system.

[0050] Step S30: After obtaining the preliminary data set X, the monitoring image data is acquired in real time, and the flame features are recognized by using the pre-trained convolutional neural network CNN. The monitoring image data is input into the pre-trained CNN model, and the CNN model performs image detection based on the flame features.

[0051] When the flame feature is detected, the image confirmation instruction information is pushed to the next step;

[0052] If no flame feature is detected for 10 seconds, the process returns to step S10;

[0053] It should be noted that the training process of the CNN model is based on a large amount of flame and non-flame image data. The model can effectively distinguish flames from other light sources or reflective objects by learning the significant features of flames in terms of shape, color and edges. For example, flame shape features: flames usually have irregular shape edges. The CNN model extracts the features of the irregular edges through multi-layer convolutions. Color features: the color of flames is usually red, orange and yellow under visible light. The model extracts such color features through convolution operations of color channels to distinguish flames from other objects. Edge features: the edge of a flame has a high brightness gradient. The CNN model can extract this gradient information to enhance the recognition of flames.

[0054] The loss function used in the pre-training process of the convolutional neural network CNN is: Loss=- 1 N ∑ j=1 N [y j ∙log( y ̂ j )+ (1-y j )∙log( 1- y ̂ j )] + λ T σ∙+ λ U ∙τ ;

[0055] The sample data in the pre-training process is the historical monitoring image data acquired in advance. is the loss function used in the pre-training process, is the total number of samples, j is the jth sample, is the true label of the jth sample, is the predicted probability of the jth sample, is the normalized outlier of temperature information, is the normalized outlier of air quality information, and are the weight coefficients of temperature anomaly term and air quality anomaly term respectively.

[0056] Step S40: after receiving the image confirmation instruction information, obtaining the environmental sound data and performing fast Fourier transform FFT analysis on the environmental sound data to obtain the sound characteristic frequency signal;

[0057] Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step;

[0058] When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10;

[0059] It should be noted that FFT (Fast Fourier Transform) is an algorithm that converts time domain signals into frequency domain signals. By performing FFT analysis on sound data, the frequency characteristics of the sound can be extracted. This helps to identify abnormal sounds of high frequency or specific frequency that may be generated during a fire.

[0060] It is understandable that when a fire occurs, some specific sound features are usually generated in the environment, such as the crackling sound of burning or the sound of objects collapsing. By establishing a fire sound feature library and comparing the sound feature frequency signals acquired in real time with the feature library, the fire sound and background noise can be effectively distinguished.

[0061] It should be understood that when the abnormal sound wave characteristics meet the preset fire sound characteristic library, the system records it as sound confirmation instruction information and passes it to the next step. This process ensures multiple verification of fire sounds and reduces the possibility of false alarms.

[0062] Step S50: Generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire fighting equipment.

[0063] It should be noted that the image confirmation command information and the sound confirmation command information are obtained from different data sources, and fire detection is performed based on visual features and sound features respectively. When both pieces of information confirm the existence of a fire, the system will generate a final fire confirmation signal. This multiple verification mechanism greatly improves the accuracy of fire detection and reduces the possibility of false alarms.

[0064] It should be understood that the generation of the final fire confirmation signal indicates that the system has fully confirmed the fire risk. On this basis, the system will trigger the alarm system, sound an alarm to notify relevant personnel, and activate fire-fighting equipment (such as sprinklers or fire extinguishers) for emergency treatment, thereby minimizing the losses caused by the fire.

[0065] In addition, the present invention also provides an intelligent fire warning system, please refer to Figure 2 , the system comprising:

[0066] A data acquisition module, used to collect environmental data based on multiple sensors to generate a fire warning data set, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U;

[0067] The data processing module is used to convert the smoke density value in the smoke density information Y into the smoke density value in the smoke density information Y for the i-th data in the fire warning data set. With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed;

[0068] The image analysis module is used to obtain the monitoring image data in real time after obtaining the preliminary data set X, and use the pre-trained convolutional neural network CNN to identify the flame characteristics of the monitoring image data, and input the monitoring image data into the pre-trained CNN model, and the CNN model performs image detection based on the flame characteristics;

[0069] When the flame feature is detected, the image confirmation instruction information is pushed to the next step;

[0070] If no flame feature is detected for 10 seconds, the process returns to step S10;

[0071] The sound analysis module is used to obtain environmental sound data and perform fast Fourier transform (FFT) analysis on the environmental sound data to obtain a sound characteristic frequency signal after receiving the image confirmation instruction information;

[0072] Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step;

[0073] When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10;

[0074] The alarm control module is used to generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire-fighting equipment.

[0075] The intelligent fire warning system provided by the present invention adopts an intelligent fire warning method in the above embodiment to solve the technical problem of intelligent fire warning. Compared with the prior art, the beneficial effect of the intelligent fire warning system provided by the present invention is the same as the beneficial effect of the intelligent fire warning method provided by the above embodiment, and other technical features of the intelligent fire warning system are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0076] The present invention provides an intelligent fire warning device, please refer to Figure 3, an intelligent fire warning device includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute an intelligent fire warning method in the above-mentioned embodiment 1. An intelligent fire warning device in the embodiment of the present invention may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. An intelligent fire warning device is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. An intelligent fire warning device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM) 1004. In RAM1004, various programs and data required for the operation of an intelligent fire warning device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow an intelligent fire warning device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows an intelligent fire warning device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have alternatively.

[0077] The present invention also provides a computer program product, including a computer program, which implements the steps of the above-mentioned intelligent fire warning method when executed by a processor. The computer program product provided by the present invention can solve the technical problem of intelligent fire warning. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as the beneficial effects of the intelligent fire warning method provided by the above-mentioned embodiment, which will not be repeated here.

[0078] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present invention are executed.

[0079] It should be understood that the various parts disclosed in the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0080] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An intelligent fire warning method, characterized in that: include: Step S10: Generate a fire warning data set based on environmental data collected by multiple sensors, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U; Step S20: For the i-th item of data in the fire warning data set, the smoke density value in the smoke density information Y is With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed; Step S30: After obtaining the preliminary data set X, the monitoring image data is acquired in real time, and the flame features of the monitoring image data are recognized using the pre-trained convolutional neural network CNN, and the monitoring image data is input into the pre-trained CNN model, and the CNN model performs image detection based on the flame features; When the flame feature is detected, the image confirmation instruction information is pushed to the next step; If no flame feature is detected for 10 seconds, the process returns to step S10; Step S40: after receiving the image confirmation instruction information, obtaining the environmental sound data and performing fast Fourier transform FFT analysis on the environmental sound data to obtain the sound characteristic frequency signal; Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step; When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10; Step S50: Generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire fighting equipment.

2. An intelligent fire warning method as claimed in claim 1, characterized in that: In step S30, the loss function used in the pre-training process of the convolutional neural network CNN is: ； The sample data in the pre-training process is the historical monitoring image data acquired in advance. is the loss function used in the pre-training process, is the total number of samples, j is the jth sample, is the true label of the jth sample, is the predicted probability of the jth sample, is the normalized outlier of temperature information, is the normalized outlier of air quality information, and are the weight coefficients of temperature anomaly term and air quality anomaly term respectively.

3. An intelligent fire warning method as claimed in claim 2, characterized in that: The calculation formula is: ； in, is the temperature information of the jth sample, is the reference value corresponding to the temperature information, and are the upper and lower limits of the temperature information, used for normalization processing.

4. An intelligent fire warning method as claimed in claim 2, characterized in that: The calculation formula is: ； in, is the air quality information of the jth sample, is the reference value corresponding to the air quality information, and are the upper and lower limits of air quality information, used for normalization processing.

5. The intelligent fire warning method according to claim 1, characterized in that: In step S10, the frequency of collecting environmental data is once per second.

6. The intelligent fire warning method according to claim 1, characterized in that: In step S10, the temperature sensor, smoke sensor and air quality sensor are all connected to the main control background via wireless connection.

7. An intelligent fire warning method as claimed in claim 1, characterized in that: In step S20, the smoke concentration deviation value is calculated by smoothing.

8. An intelligent fire warning system, applied to an intelligent fire warning method according to any one of claims 1 to 7, characterized in that: The system comprises: A data acquisition module, used to collect environmental data based on multiple sensors to generate a fire warning data set, wherein the multiple sensors include a temperature sensor, a smoke sensor, and an air quality sensor, and the fire warning data set includes temperature information T, smoke concentration information Y, and air quality information U; The data processing module is used to convert the smoke density value in the smoke density information Y into the smoke density value in the smoke density information Y for the i-th data in the fire warning data set. With preset smoke concentration threshold Compare and calculate the smoke concentration deviation value , if the smoke concentration deviation value If the preset deviation range is exceeded, the i-th item of data is recorded as abnormal data, and the preliminary data set X is formed after the abnormal data is removed; The image analysis module is used to obtain the monitoring image data in real time after obtaining the preliminary data set X, and use the pre-trained convolutional neural network CNN to identify the flame characteristics of the monitoring image data, and input the monitoring image data into the pre-trained CNN model, and the CNN model performs image detection based on the flame characteristics; When the flame feature is detected, the image confirmation instruction information is pushed to the next step; If no flame feature is detected for 10 seconds, the process returns to step S10; The sound analysis module is used to obtain environmental sound data and perform fast Fourier transform (FFT) analysis on the environmental sound data to obtain a sound characteristic frequency signal after receiving the image confirmation instruction information; Compare the sound characteristic frequency signal with a preset fire sound characteristic library, and when the sound characteristic frequency signal matches the preset fire sound characteristic library, push the sound confirmation instruction information to the next step; When the sound characteristic frequency signal does not conform to the preset fire sound characteristic library, return to step S10; The alarm control module is used to generate a final fire confirmation signal based on the image confirmation instruction information and the sound confirmation instruction information, and trigger the alarm system and activate the fire-fighting equipment.

Images