Composite fire alarm system using photoelectric smoke detection and image transmission recognition

Abstract

The present invention relates to the field of fire alarm monitoring, and discloses a composite fire alarm system using photoelectric smoke detection and image transmission recognition, including: a dust cover, a dust net is installed inside the dust net, a convex lens is arranged inside the dust net, a transparent sheet is arranged on one side of the convex lens, and a buzzer sheet is installed between the convex lens and the transparent sheet; an antenna, which is connected to an upper cover through a buzzer sheet, a QR code sticker is arranged on the side of the upper cover close to the antenna, a camera module is installed on the upper cover through double-sided sticker 1, a flame module is installed on the upper cover through double-sided sticker 2, and one side of the upper cover is connected to a maze cover. Through the photoelectric smoke detection module, the temperature sensor module and the image transmission recognition module, multi-level fire detection and alarm are realized, and then through the intelligent data fusion module, the data of multiple sensors are integrated and analyzed to automatically adapt to different environmental conditions and improve the accuracy and timeliness of fire detection.

Description

Technical Field

[0001] The present invention relates to the technical field of fire alarm monitoring, in particular to a composite fire alarm system using photoelectric smoke detection and image transmission recognition. Background Art

[0002] Existing fire alarm systems usually rely on a series of fixed parameters and thresholds for fire detection. These systems mainly use traditional equipment such as photoelectric smoke detectors and temperature sensors to identify fire signals by detecting changes in smoke concentration and ambient temperature. Photoelectric smoke detectors use the scattering principle of light beams to detect smoke particles in the air. When the smoke particles reach a certain set concentration, the system will trigger an alarm. The temperature sensor determines whether a fire has occurred by monitoring the rising trend of the ambient temperature. When the temperature exceeds a preset threshold, the alarm will be activated. This type of system has good detection performance under specific environmental conditions and can provide basic fire alarm functions for many occasions. In addition, these systems are usually integrated into the safety infrastructure of buildings to provide a cost-effective fire warning solution. The design of traditional fire alarm systems is usually simple, easy to install and maintain, and can adapt to general indoor environments. However, these systems also face technical limitations.

[0003] First, in existing fire alarm systems, a single type of sensor (such as photoelectric smoke detectors or temperature sensors) usually cannot provide sufficient fire detection accuracy and timeliness in complex environments. These systems may be interfered by environmental factors such as dust, humidity changes and temperature fluctuations, resulting in false alarms or missed alarms. At the same time, the signal of a single sensor may not be sufficient to accurately identify the occurrence of a fire in the early stages. Traditional systems often rely on a single sensing parameter and lack a comprehensive analysis of multiple fire indicators, resulting in limited detection performance.

[0004] Secondly, traditional fire alarm systems usually operate under set fixed parameters and lack the ability to adapt to environmental changes. When environmental conditions such as temperature, humidity, air quality, lighting conditions and noise levels change, the sensitivity and response characteristics of the sensor may be affected, resulting in an increase in false alarm rates. These changes make it difficult for the system to maintain a high level of detection accuracy and reliability.

[0005] Finally, in many existing fire alarm systems, fault detection and maintenance usually rely on regular manual inspections, or repairs are performed only after a fault occurs. This passive maintenance method may result in failure to detect faults in a timely manner, thereby affecting the normal operation of the system and detection reliability. In addition, the lack of the ability to predict potential faults also makes preventive maintenance impossible to perform effectively, increasing the maintenance cost and complexity of the system. Finally, when installing the hardware of the composite fire alarm system, the composite fire alarm system hardware is often directly installed on the wall with bolts. After long-term use, it is difficult to quickly remove the composite fire alarm system hardware due to the rust of the bolts. Summary of the invention

[0006] In view of the deficiencies of the prior art, the present invention provides a composite fire alarm system using photoelectric smoke detection and image transmission and recognition, which solves the problem that detection by a single sensor is easily limited.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a composite fire alarm system using photoelectric smoke detection and image transmission recognition, comprising:

[0008] A dust cover, wherein a dust net is integrally formed inside the dust cover, an upper cover is integrally formed below the dust cover, a dust net blocking quality inspection transmitting tube and a dust net blocking quality inspection receiving tube are provided on the upper cover, and the dust net blocking quality inspection transmitting tube and the dust net blocking quality inspection receiving tube are both located between the dust nets, a convex lens is provided inside the upper cover, a filter is provided on one side of the convex lens, and a buzzer is installed between the convex lens and the filter;

[0009] The antenna is connected to the upper cover through a buzzer sheet, a QR code sticker is arranged on one side of the upper cover close to the antenna, a camera module is installed on the upper cover through a double-sided sticker 1, a flame module is installed on the upper cover through a double-sided sticker 2, and one side of the upper cover is connected to the maze cover;

[0010] A labyrinth is installed inside the upper cover through a labyrinth cover, wherein a transmitting tube 1 and a transmitting tube 2 are equidistantly distributed on one side of the labyrinth, a receiving tube is arranged between the transmitting tube 1 and the transmitting tube 2, and a shielding cover is arranged on one side of the receiving tube, and the shielding cover is used to shield the receiving tube position;

[0011] A PCBA mainboard is installed on a side of the maze away from the maze cover, and a positive electrode of a battery spring and a negative electrode of a battery spring are equidistantly distributed on one side of the PCBA mainboard;

[0012] The base is connected to the upper cover, a support is fixed inside the base, a battery is installed inside the support, and a foot pad is fixed on the side of the base away from the support.

[0013] Preferably, a first locking plate is detachably arranged on the base, a locking member is detachably mounted on the first locking plate, a second locking plate is detachably mounted on the bottom of the locking member, and the locking member comprises:

[0014] A locking frame, which is detachably mounted on the first locking plate, and in which an embedded block is detachably mounted, forming a clamping cavity between the locking frame and the embedded block;

[0015] An external plug-in plate, which is slidably arranged in the locking frame, a first roller is rotatably arranged on one end thereof, an ear plate is detachably arranged on it, a side plug rod is detachably arranged on the ear plate, and the side plug rod passes through the locking frame, a connecting spring is removably sleeved on the side plug rod, and the connecting spring is respectively in contact with the ear plate and the locking frame, and a limit plate is arranged on the end of the side plug rod passing through the locking frame;

[0016] A limiting insert ring is detachably mounted on the second locking plate, and a blocking block is detachably mounted on the limiting insert ring. The blocking block has a first inclined surface, and a through hole is formed on the blocking block, and the through hole can penetrate into the limiting insert ring, and the through hole can allow the outer insert plate to pass through.

[0017] A reinforcing plug plate is slidably inserted into the limiting plug ring, and one end of the reinforcing plug plate is initially located in the through hole, an inclined plate is integrally formed on the reinforcing plug plate, a fitting pad is integrally formed on the inclined plate, a first connecting base plate is detachably provided on the reinforcing plug plate, and a hanging spring is removably hung on the first connecting base plate;

[0018] A second connecting substrate is detachably mounted in the limiting insert ring and is hung by the hanging spring;

[0019] The unlocking seat has an unlocking pad formed integrally on its side surface, a second inclined surface inside it, and a cross locking shaft detachably disposed on its bottom surface, and the cross locking shaft can pass through the locking frame, the limiting insert ring, and the embedded block in sequence;

[0020] An unlocking lever, one end of which is detachably mounted on one end of the outer plug-in plate, and a second roller is rotatably arranged on the other end of the unlocking lever, and the second roller is located above the unlocking pad;

[0021] A limiting collar is detachably mounted on the locking frame and can allow the unlocking rod to pass through.

[0022] Preferably, the PCBA mainboard includes a photoelectric smoke detection module, an image transmission recognition module, a temperature sensor module, an intelligent data fusion module, a self-diagnosis and calibration mechanism module, an edge computing and distributed processing module, an environmental compensation technology module and a data transmission module;

[0023] The photoelectric smoke detection module is used to detect smoke particles in the air;

[0024] The image transmission and recognition module is used to capture fire scene images and perform recognition analysis;

[0025] The temperature sensing module is used to monitor changes in ambient temperature;

[0026] The intelligent data fusion module is used to integrate multiple sensor data;

[0027] The self-diagnosis and calibration mechanism module is used to maintain the accuracy and reliability of the system sensor;

[0028] The edge computing and distributed processing modules are used to improve the real-time response capability of the system;

[0029] The environmental compensation technology module is used to reduce the interference of environmental factors on system detection;

[0030] The data transmission module is used to realize efficient transmission of images and data.

[0031] Preferably, the photoelectric smoke detection module includes a light source and a receiving unit and a sensitivity adjustment unit:

[0032] The light source and receiving unit are used to detect smoke scattering through the light beam and determine the smoke concentration;

[0033] The sensitivity adjustment unit is used to automatically adjust the sensitivity to adapt to different environmental conditions.

[0034] Preferably, the image transmission identification module includes a camera capture unit, an image analysis unit and an image compression and transmission unit;

[0035] The camera capture unit uses a camera to capture fire scene images in real time;

[0036] The image analysis unit uses a convolutional neural network algorithm to identify flame and smoke features;

[0037] The convolutional neural network algorithm formula is specifically:

[0038] ;

[0039] in is the activation function output, is the convolution kernel weight, is the input feature, is the bias term, is the activation function;

[0040] The image compression and transmission unit is used to compress the image data into JPEG format and transmit it to the central processing unit through the network.

[0041] Preferably, the temperature sensing module includes a temperature detection unit and a threshold alarm unit;

[0042] The temperature detection unit uses a digital temperature sensor to detect and record changes in ambient temperature in real time;

[0043] The threshold alarm unit is used to trigger a buzzer to sound an alarm when an abnormal temperature change is detected. The threshold is set to an absolute temperature higher than 70°C, and the abnormal temperature change value exceeds the set threshold.

[0044] Preferably, the intelligent data fusion module includes a Bayesian network fusion unit, a Kalman filter processing unit and a dynamic weight adjustment unit;

[0045] Bayesian network fusion unit:

[0046] Function: Probabilistic analysis of different sensor data through Bayesian network;

[0047] The Bayesian network algorithm formula is:

[0048] ;

[0049] For the given evidence Assumption The posterior probability of Assumption Evidence The likelihood, Assumption The prior probability of For evidence The marginal probability of

[0050] The Kalman filter processing unit uses a Kalman filter algorithm to dynamically adjust and optimize sensor data;

[0051] The specific formula of the Kalman filter algorithm is:

[0052] ;

[0053] in, To predict the state estimate, is the prediction covariance matrix, is the Kalman gain, is the observed quantity, , , are the state transfer, control and observation matrices respectively, is the observation matrix, , are the process noise covariance matrix and the observation noise covariance matrix, respectively. is the control input matrix, is the control vector, is the identity matrix, For time;

[0054] The dynamic weight adjustment unit automatically adjusts the fusion weight according to the sensor data characteristics.

[0055] Preferably, the self-diagnosis and calibration mechanism module includes a self-diagnosis detection unit, an automatic calibration module and a fault feedback and alarm unit;

[0056] The self-diagnosis detection unit is used to continuously monitor the working status of the sensor and identify faults and abnormalities;

[0057] The working state means that the sensor output is stable and within the expected range, data acquisition and processing are normal, and self-test signal feedback is normal;

[0058] The fault feedback includes hardware fault and software fault;

[0059] The abnormalities include but are not limited to environmental abnormalities, signal abnormalities and sensitivity abnormalities;

[0060] The automatic calibration module is used to automatically perform a sensor calibration process;

[0061] The fault feedback and alarm unit is used to issue an alarm to the user when a fault is detected.

[0062] Preferably, the edge computing and distributed processing module includes an edge computing node unit, a distributed processing architecture and a real-time response module;

[0063] The edge computing node unit is used to perform local data processing at the sensor node;

[0064] The distributed processing architecture is used to implement multi-node collaborative data processing;

[0065] The real-time response module is used to optimize data processing and decision-making speed.

[0066] Preferably, the environmental compensation technology module includes an environmental feature learning unit, an automatic compensation unit and an environmental monitoring and adjustment unit;

[0067] The environmental feature learning unit uses a random forest model to analyze and store environmental feature data;

[0068] The random forest model establishment and training includes the following steps:

[0069] S1. Data collection and preparation: Continuously collect data from various sensors (temperature, humidity, air quality, light, noise), process missing data, outliers and noise to ensure data quality, and finally determine the key environmental characteristics related to fire detection;

[0070] S2. Model building: Select random forest as the environmental feature learning model, divide the data set into training set, validation set and test set (such as 70% training, 15% validation, 15% test), and then set the parameters of random forest, such as the number of trees and maximum depth;

[0071] S3. Model training: Use the training set data to train the random forest model, and then use cross-validation methods (such as K-fold cross-validation) to optimize model parameters to prevent overfitting. Finally, during the training process, feature importance analysis is used to identify the features that have the greatest impact on the output.

[0072] S4, Model evaluation and deployment: Evaluate the model performance on the test set, using indicators such as accuracy, recall, and F1-score, and then adjust the model parameters based on the evaluation results to further improve the model accuracy. Finally, integrate the trained model into the environmental feature learning unit for real-time analysis and update of environmental feature data;

[0073] The environmental characteristics include temperature, humidity, air quality, lighting conditions, air flow rate, and noise level;

[0074] The automatic compensation unit is used to dynamically adjust the sensor output according to the environmental feature library;

[0075] The environmental monitoring and adjustment unit is used to monitor environmental changes in real time and automatically adjust system parameters as needed;

[0076] The data transmission module includes a wired transmission unit, a wireless transmission unit and a data compression and encryption unit;

[0077] The wired transmission unit transmits data via optical fiber or Ethernet;

[0078] The wireless transmission module uses Wi-Fi, Bluetooth or Zigbee to perform wireless data transmission;

[0079] The data compression and encryption unit is used to compress and encrypt data.

[0080] The present invention provides a composite fire alarm system using photoelectric smoke detection and image transmission recognition, which has the following beneficial effects:

[0081] 1. Through the overall arrangement of the locking components, it is first necessary to lift the unlocking seat to move upward so that the cross locking shaft can pass through the embedded block, the limiting plug ring, and the locking frame in sequence. In the process of the unlocking seat moving upward, the second inclined surface is moved upward as a whole. Since the second inclined surface is in contact with the second roller, the second roller can be pushed to the right in a synchronous manner during the upward movement of the second inclined surface, and the unlocking rod can slide in the limiting collar to prevent the unlocking rod from shaking left and right and being unstable when being pushed by the second inclined surface. At the same time, after being blocked by the unlocking pad, the unlocking seat can be effectively prevented from being completely separated from the second roller. Since the unlocking rod is connected to the external plug-in plate, When the cross locking shaft is released from the positioning in the up and down directions, it can simultaneously drive the outer plug-in plate to move to the right, so that the first roller as a whole withdraws from the limiting plug-in ring, and at the same time, the fitting pad releases the positioning of the embedded block, and also releases the positioning of the limiting plug-in ring by the outer plug-in plate, and the limiting plug-in ring is pulled to disengage the limiting plug-in ring from the locking frame. The locking components arranged as a whole can make the base quickly removed from the wall when it is connected to the wall, so as to facilitate the overall installation and disassembly of the hardware of the composite fire alarm system using photoelectric smoke detection and image transmission recognition, and facilitate the disassembly and maintenance of the hardware of the composite fire alarm system using photoelectric smoke detection and image transmission recognition.

[0082] 2. The present invention realizes multi-level fire detection and alarm through photoelectric smoke detection module, temperature sensor module and image transmission recognition module, and then integrates and analyzes the data of various sensors through intelligent data fusion module, so as to automatically adapt to different environmental conditions and improve the accuracy and timeliness of fire detection.

[0083] 3. The present invention dynamically adjusts the sensor output according to real-time environmental data through the automatic compensation unit, and the environmental monitoring and adjustment unit automatically adjusts the system parameters according to the monitored environmental changes, thereby achieving adaptive compensation for environmental changes and improving detection accuracy and reliability.

[0084] 4. The present invention analyzes historical fault data and predicts potential faults through a self-diagnosis and calibration mechanism module. This mechanism can not only quickly alarm when a fault occurs, but also predict the time and location where the fault may occur in advance, so as to arrange preventive maintenance, thereby improving the reliability and maintenance efficiency of the entire fire alarm system. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 This is an exploded view of the hardware structure of the composite fire alarm system for photoelectric smoke detection and image transmission and recognition of the present invention;

[0086] Figure 2It is a schematic diagram of the structure of the composite fire alarm system of photoelectric smoke detection and image transmission recognition of the present invention when the hardware is combined;

[0087] Figure 3 This is a schematic diagram of the structure of the composite fire alarm system hardware combination of photoelectric smoke detection and image transmission recognition from another perspective of the present invention;

[0088] Figure 4 It is a schematic diagram of the overall structure of the locking component of the present invention;

[0089] Figure 5 It is a schematic structural diagram of the combination of the limit insert ring and the stop block of the present invention;

[0090] Figure 6 This is a schematic structural diagram of the combination of the limit insert ring and the blocking block of the present invention from another angle;

[0091] Figure 7 It is a structural schematic diagram of the combination of the locking frame and the external plug-in board of the present invention;

[0092] Figure 8 It is a structural exploded diagram of the present invention when the locking frame is connected to the external plug-in board;

[0093] Fig. 9 It is a schematic diagram of the PCBA board system module of the present invention;

[0094] Fig.10 This is a schematic diagram of the photoelectric smoke detection module of the present invention;

[0095] Fig.11 It is a schematic diagram of the image transmission recognition module of the present invention;

[0096] Fig.12 It is a schematic diagram of a temperature sensing module of the present invention;

[0097] Fig.13 It is a schematic diagram of the intelligent data fusion module of the present invention;

[0098] Fig.14 It is a schematic diagram of the self-diagnosis and calibration mechanism module of the present invention;

[0099] Fig.15 It is a schematic diagram of the edge computing and distributed processing module of the present invention;

[0100] Fig.16 It is a schematic diagram of the environmental compensation technology module of the present invention;

[0101] Fig.17 It is a schematic diagram of a data transmission module of the present invention.

[0102] Among them, 1. dust cover; 2. dust net; 3. convex lens; 4. filter; 5. buzzer; 6. antenna; 7. QR code sticker; 8. top cover; 9. camera module; 10. flame module; 11. double-sided sticker one; 12. double-sided sticker two; 13. labyrinth cover; 14. labyrinth; 15. transmitting tube one; 16. transmitting tube two; 17. receiving tube; 18. shielding cover; 19. battery spring positive pole; 20. battery spring negative pole; 21. PCBA mainboard; 22. battery; 23. support; 24. base; 25. foot pad; 26. first locking plate; 27. second locking plate; 28. locking component; 281. locking machine frame; 282 , embedded block; 283, external plug-in plate; 284, first roller; 285, ear plate; 286, side plug-in rod; 287, connecting spring; 288, limit plate; 289, limit plug ring; 2810, blocking block; 2811, first inclined surface; 2812, through-hole; 2813, reinforcement plug-in plate; 2814, inclined plate; 2815, fitting pad; 2816, first connecting substrate; 2817, second connecting substrate; 2818, hanging spring; 2819, unlocking seat; 2820, unlocking pad; 2821, second inclined surface; 2822, cross locking shaft; 2823, unlocking rod; 2824, second roller; 2825, limit collar. DETAILED DESCRIPTION

[0103] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings 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.

[0104] Example:

[0105] Please see attached Figure 1 -Attached Figure 8 The embodiment of the present invention provides a composite fire alarm system using photoelectric smoke detection and image transmission recognition, including:

[0106] A dust cover 1 is provided with a dust net 2 integrally formed inside, an upper cover 8 is integrally formed below the dust cover 1, a dust net blocking quality inspection transmitting tube and a dust net blocking quality inspection receiving tube are provided on the upper cover 8, and the dust net blocking quality inspection transmitting tube and the dust net blocking quality inspection receiving tube are both located between the dust net 2, a convex lens 3 is provided inside the upper cover 8, a filter 4 is provided on one side of the convex lens 3, and a buzzer 5 is installed between the convex lens 3 and the filter 4;

[0107] The antenna 6 is connected to the upper cover 8 through the buzzer sheet 5. A QR code sticker 7 is provided on the upper cover 8 near the antenna 6. The upper cover 8 is installed with a camera module 9 through a double-sided sticker 11. The upper cover 8 is installed with a flame module 10 through a double-sided sticker 2 12. One side of the upper cover 8 is connected to the labyrinth cover 13;

[0108] The labyrinth 14 is installed inside the upper cover 8 through the labyrinth cover 13. A transmitting tube 15 and a transmitting tube 2 16 are evenly distributed on one side of the labyrinth 14. A receiving tube 17 is arranged between the transmitting tube 15 and the transmitting tube 2 16. A shielding cover 18 is arranged on one side of the receiving tube 17. The shielding cover 18 is used to shield the position of the receiving tube 17.

[0109] A PCBA mainboard 21 is mounted on the side of the maze 14 away from the maze cover 13, and a battery spring positive electrode 19 and a battery spring negative electrode 20 are evenly distributed on one side of the PCBA mainboard 21;

[0110] A base 24 is connected to the upper cover 8, a support 23 is fixed inside the base 24, a battery 22 is installed inside the support 23, and a foot pad 25 is fixed on the side of the base 24 away from the support 23;

[0111] Next, a more specific explanation of the filter 4 is given. First of all, it should be explained that the filter 4 is a model of 850nm. The visible light below 850nm is shielded and filtered, and the privacy of the photo will not be leaked. When the flame light above 850nm passes through the filter 4, it will be captured and photographed;

[0112] When the hardware of the composite fire alarm system of photoelectric smoke detection and image transmission recognition is installed on site, an initial pass rate value of the dust screen is formed. When the device sends heartbeat packets regularly, it is performed every two weeks. The light waves are emitted from the dust screen blockage quality inspection transmitting tube, which pass through the dust screen and are finally received by the dust screen blockage quality inspection receiving tube. The light waves are compared with the initial pass rate value. If the pass rate value is lower than 70%, an alarm will be triggered.

[0113] Of course, in other embodiments, the pass rate value may also be set according to actual conditions, such as between 80% and 60%, and of course other pass rate values ​​may also be used.

[0114] In order to further facilitate the rapid installation and disassembly of the entire base 24, the locking component 28 is given a more specific structure and construction as a whole for further explanation. A first locking plate 26 is detachably arranged on the base 24, a locking component 28 is detachably installed on the first locking plate 26, and a second locking plate 27 is detachably installed at the bottom of the locking component 28. The locking component 28 includes a locking frame 281, which is detachably installed on the first locking plate 26, and an embedded block 282 is detachably installed therein, forming a clamping cavity between the outer plug-in plate 283 and the outer plug-in plate 283, which is slidably arranged on the locking frame 28 1, a first roller 284 is rotatably arranged on one end thereof, an ear plate 285 is detachably arranged on it, a side plug rod 286 is detachably arranged on the ear plate 285, and the side plug rod 286 passes through the locking machine frame 281, a connecting spring 287 is also removably sleeved on the side plug rod 286, and the connecting spring 287 is respectively in contact with the ear plate 285 and the locking machine frame 281, and a limiting plate 288 is arranged on the end of the side plug rod 286 passing through the locking machine frame 281; a limiting insert ring 289 is detachably mounted on the second locking plate 27, and a blocking block 2810 is detachably mounted on it, and the blocking block 2810 has a first inclined surface 2811 , a through hole 2812 is also provided on the blocking block 2810, and the through hole 2812 can penetrate into the limiting plug ring 289, and the through hole 2812 can allow the outer plug plate 283 to pass through; the reinforcing plug plate 2813 is slidably arranged in the limiting plug ring 289, and one end of it is initially located in the through hole 2812, and an inclined plate 2814 is integrally formed on it, and a fitting pad 2815 is integrally formed on the inclined plate 2814, and a first connecting substrate 2816 is also detachably arranged on the reinforcing plug plate 2813, and a hanging spring 2818 is removably hung on the first connecting substrate 2816; the second connecting substrate 2817 is detachably installed in the limiting plug ring 289 , which is hung by a hanging spring 2818; an unlocking seat 2819, whose side surface is integrally formed with an unlocking pad 2820, and has a second inclined surface 2821 therein, and a cross locking shaft 2822 is detachably arranged on its bottom surface, and the cross locking shaft 2822 can pass through the locking frame 281, the limiting insert ring 289, and the embedded block 282 in sequence; an unlocking rod 2823, one end of which is detachably mounted on one end of the outer insert plate 283, and a second roller 2824 is rotatably arranged on the other end, and the second roller 2824 is located above the unlocking pad 2820; a limiting collar 2825, which is detachably mounted on the locking frame 281, and can allow the unlocking rod 2823 to pass through;

[0115] To further explain the principle of rapid installation and disassembly of the base 24 as a whole and the wall, the staff first installs the second locking plate 27 on the pre-installed plane. It should be noted that the limiting insert ring 289 is connected to the second locking plate 27. Therefore, in the process of pushing the limiting insert ring 289 into the locking frame 281, the limiting insert ring 289 will first penetrate into the clamping cavity formed by the locking frame 281 and the embedded block 282, and at the same time, the embedded block 282 can also penetrate into the limiting insert ring 289. When the limiting insert ring 289 is pushed toward the locking frame 281, the outer plug plate 283 and the first roller 284 will first move on the first inclined surface 2811. When the first roller 284 slides to the limit position of the first inclined surface 2811, it will come to the plane of the blocking block 2810. When the first roller 284 and the outer plug plate 283 slide into the through-hole 2812, the outer plug plate 283 will quickly enter into the through-hole 2812 to complete the positioning of the limiting insert ring 289 and the blocking block 2810.

[0116] At the same time, when the outer plug plate 283 quickly enters the through-hole 2812 to position the limit plug ring 289, the reinforcing plug plate 2813 is triggered to work. When the outer plug plate 283 enters the through-hole 2812, it also enters the limit plug ring 289. After pushing the reinforcing plug plate 2813, the reinforcing plug plate 2813 is driven to continue to move forward, and the hanging spring 2818 is stretched at the same time, so that the fitting pad 2815 can clamp the embedded block 282 stably. At the same time, the outer plug plate 283 can position the limit plug ring 289 and the blocking block 2810 as a whole, and The embedded block 282 can be positioned synchronously, and the limiting insert ring 289 can be pushed into the clamping cavity, and the embedded block 282 can be inserted into the limiting insert ring 289 to achieve internal and external clamping positioning, and at the same time, the unlocking seat 2819 drives the cross locking shaft 2822 to penetrate into the locking frame 281, the limiting insert ring 289, and the embedded block 282 in sequence. When the cross locking shaft 2822 is in a cross shape, the contact between the cross locking shaft 2822 and the locking frame 281, the limiting insert ring 289 and the embedded block 282 can be further increased to achieve the positioning of the limiting insert ring 289 as a whole;

[0117] When unlocking the limiting insert ring 289, it is first necessary to lift the unlocking seat 2819 to move upward so that the cross locking shaft 2822 can pass through the embedded block 282, the limiting insert ring 289, and the locking frame 281 in sequence. In the process of the unlocking seat 2819 moving upward, the second inclined surface 2821 is moved upward as a whole. Since the second inclined surface 2821 is in contact with the second roller 2824, the second roller 2824 can be pushed to the right in a synchronous manner during the upward movement of the second inclined surface 2821, and the unlocking rod 2823 is made to slide in the limiting ring 2825 to prevent the unlocking rod 2823 from shaking left and right and being unstable when being pushed by the second inclined surface 2821. 20 blocking, it can effectively prevent the unlocking seat 2819 from being completely separated from the second roller 2824. Since the unlocking rod 2823 is connected to the outer plug-in plate 283, when the cross locking shaft 2822 is released from the positioning in the up and down directions, it can simultaneously drive the outer plug-in plate 283 to move to the right, so that the first roller 284 is withdrawn from the limiting plug ring 289 as a whole, and at the same time, the fitting pad 2815 releases the positioning of the embedded block 282, and also releases the positioning of the limiting plug ring 289 by the outer plug-in plate 283. When the limiting plug ring 289 is pulled to separate from the locking frame 281, the locking component 28 can be arranged as a whole so that the base 24 can be quickly removed from the wall when it is connected to the wall.

[0118] In some other embodiments, in order to further solve the dust environmental pollution problem in the maze darkroom of the smoke detector, a micro motor or other micro vibration device can be installed in the base. By generating tiny vibrations inside the smoke detector, the dust particles attached to the surface of the photoelectric transmitting and receiving tubes are loosened and fall off. The micro vibration can shake the dust off the surface and discharge it to the outside of the detector through air flow. The micro vibration by a vibration motor or other micro vibrator is relatively simple and does not require complex equipment or technology, and is easy to implement. Compared with vibration methods such as ion generators, the energy required for micro vibration is lower and will not significantly increase the overall power consumption of the equipment. Micro vibration does not produce ozone or other harmful substances and is harmless to the environment and human body.

[0119] The PCBA mainboard 21 includes a photoelectric smoke detection module, an image transmission recognition module, a temperature sensing module, an intelligent data fusion module, a self-diagnosis and calibration mechanism module, an edge computing and distributed processing module, an environmental compensation technology module, and a data transmission module;

[0120] The photoelectric smoke detection module is used to detect smoke particles in the air;

[0121] The image transmission recognition module is used to capture fire scene images and perform recognition analysis;

[0122] The temperature sensor module is used to monitor changes in ambient temperature;

[0123] Intelligent data fusion module is used to integrate multiple sensor data;

[0124] The self-diagnosis and calibration mechanism module is used to maintain the accuracy and reliability of the system sensors;

[0125] Edge computing and distributed processing modules are used to improve the real-time response capabilities of the system;

[0126] The environmental compensation technology module is used to reduce the interference of environmental factors on system detection;

[0127] The data transmission module is used to achieve efficient transmission of images and data.

[0128] Specifically, the dust cover 1 and the dust net 2 protect the internal components from dust and other particulate matter, while ensuring the clarity of the optical components. The convex lens 3 is used to focus the optical signal from the fire scene and transmit the light through the transparent sheet 4. The buzzer sheet 5 is located between the convex lens 3 and the transparent sheet 4. When an abnormal situation is detected, the buzzer sheet 5 will sound an alarm. The antenna 6 is connected to the upper cover 8 through the buzzer sheet 5 for sending and receiving wireless signals to achieve remote monitoring and data transmission. The upper cover 8 is equipped with a camera module 9 and a flame module 10, which are fixed by double-sided tape 1 11 and double-sided tape 2 12. These modules are respectively used to capture and analyze fire images and monitor flame signals, and transmit data to the PCBA main board 21 for further processing. The maze 14 structure is installed inside the upper cover 8 through the maze cover 13. The transmitting tube 1 15 and the transmitting tube 2 16 arranged inside alternately transmit optical signals, and the receiving tube 17 receives and detects signal changes caused by smoke or obstacles. The shielding cover 18 positions and shields the receiving tube 17 to prevent ambient light interference. When smoke passes through the maze At 14 o'clock, the photoelectric smoke detection module detects smoke particles in the air and activates the alarm process. At the same time, the image transmission and recognition module captures the fire scene image and performs real-time recognition and analysis. If the fire image or smoke concentration exceeds the set threshold, the temperature change information monitored by the temperature sensor module will be transmitted to the intelligent data fusion module, which integrates smoke, image and temperature data for comprehensive judgment. The self-diagnosis and calibration mechanism module continuously monitors and calibrates the sensor in the whole process to ensure data accuracy and reliability. The edge computing and distributed processing module processes the data in real time to improve the system response speed, and reduces the interference of environmental factors through the environmental compensation technology module. Finally, the data transmission module transmits the processed image and data efficiently and wirelessly through the antenna 6 and sends it to the remote monitoring center or cloud platform for further processing and response. The whole system is powered by the battery 22 in the base 24. The battery spring positive electrode 19 and the battery spring negative electrode 20 ensure the continuous supply of power and are fixed by the support 23. The foot pad 25 provides stability and anti-slip effect of the equipment.

[0129] Please see attached Fig.10 -Attached Fig.12 The photoelectric smoke detection module includes a light source, a receiving unit and a sensitivity adjustment unit:

[0130] The light source and receiving unit are used to detect smoke scattering through the light beam and determine the smoke concentration;

[0131] The sensitivity adjustment unit is used to automatically adjust the sensitivity to adapt to different environmental conditions;

[0132] The image transmission recognition module includes a camera capture unit, an image analysis unit and an image compression and transmission unit;

[0133] The video capture unit uses a camera to capture images of the fire scene in real time;

[0134] The image analysis unit uses a convolutional neural network algorithm to identify flame and smoke features;

[0135] The specific formula of the convolutional neural network algorithm is:

[0136] ;

[0137] in is the activation function output, is the convolution kernel weight, is the input feature, is the bias term, It is an activation function, which directly outputs when the input is greater than 0, otherwise it outputs 0;

[0138] The image compression and transmission unit is used for compressing the image data into JPEG format and transmitting it to the central processing unit through the network;

[0139] The temperature sensing module includes a temperature detection unit and a threshold alarm unit;

[0140] The temperature detection unit uses a digital temperature sensor to monitor and record changes in ambient temperature in real time;

[0141] The threshold alarm unit is used to trigger the buzzer 5 to sound an alarm when abnormal temperature changes are detected. The threshold is set to an absolute temperature higher than 70° C., and the abnormal temperature change value exceeds the set threshold.

[0142] Specifically, the light source and receiving unit use LED light beams and photodiodes to determine the smoke concentration in the air by detecting light scattering, and the sensitivity adjustment unit dynamically adjusts the detection sensitivity through an adjustable gain amplifier to adapt to different environmental conditions; the image transmission and recognition module consists of a camera capture unit, an image analysis unit, and an image compression and transmission unit. The camera is responsible for capturing images of the fire scene in real time. The image analysis unit uses a convolutional neural network algorithm to identify flame and smoke characteristics. The analyzed image is compressed in JPEG format and transmitted to the core processing. The temperature sensing module includes a temperature detection unit and a threshold alarm unit. A digital temperature sensor such as DS18B20 is used to monitor and record the ambient temperature in real time. When the absolute temperature is detected to be higher than 70°C, the threshold alarm unit triggers an alarm, photoelectric smoke detection provides a preliminary smoke alarm, image recognition verifies the accuracy of the alarm, and temperature monitoring provides additional safety protection. Finally, multiple detection results are sent to the distributed processing architecture through the compression and transmission unit, so as to determine whether a fault has occurred through the dual protection of image recognition and temperature recognition.

[0143] Please see attached Fig.13 -Attached Fig.15,The intelligent data fusion module includes a Bayesian network fusion unit, a Kalman filter processing unit and a dynamic weight adjustment unit;

[0144] Bayesian network fusion unit:

[0145] Function: Probabilistic analysis of different sensor data through Bayesian network;

[0146] The Bayesian network algorithm formula is:

[0147] ;

[0148] For the given evidence Assumption The posterior probability of Assumption Evidence The likelihood, Assumption The prior probability of For evidence The marginal probability of

[0149] The Kalman filter processing unit uses the Kalman filter algorithm to dynamically adjust and optimize the sensor data;

[0150] The specific formula of the Kalman filter algorithm is:

[0151] ;

[0152] in, To predict the state estimate, is the prediction covariance matrix, is the Kalman gain, is the observed quantity, , , are the state transfer, control and observation matrices respectively, is the observation matrix, , are the process noise covariance matrix and the observation noise covariance matrix, respectively. is the control input matrix, is the control vector, is the identity matrix, For time;

[0153] The dynamic weight adjustment unit automatically adjusts the fusion weight according to the sensor data characteristics;

[0154] The self-diagnosis and calibration mechanism module includes a self-diagnosis detection unit, an automatic calibration module and a fault feedback and alarm unit;

[0155] The self-diagnosis detection unit is used to continuously monitor the working status of the sensor and identify faults and abnormalities;

[0156] Working status means that the sensor output is stable and within the expected range, data acquisition and processing are normal, and self-test signal feedback is normal;

[0157] Fault feedback includes hardware failure and software failure;

[0158] Abnormalities include but are not limited to environmental abnormalities, signal abnormalities, and sensitivity abnormalities;

[0159] The automatic calibration module is used to automatically perform the sensor calibration process;

[0160] The fault feedback and alarm unit is used to alert the user when a fault is detected;

[0161] The edge computing and distributed processing module includes edge computing node units, distributed processing architecture and real-time response modules;

[0162] The edge computing node unit is used to perform local data processing at the sensor node;

[0163] Distributed processing architecture is used to achieve multi-node collaborative data processing;

[0164] The real-time response module is used to optimize data processing and decision-making speed.

[0165] Specifically, the Bayesian network fusion unit determines the possibility of fire by probabilistically analyzing the data from different sensors, thereby improving the accuracy of data fusion; the Kalman filter processing unit dynamically adjusts and optimizes the sensor data, filters out noise and abnormal fluctuations, and ensures the stability and accuracy of the data; the dynamic weight adjustment unit automatically adjusts the fusion weight according to the changes in the characteristics of the sensor data, so that the contribution of each sensor's data to the overall judgment can change dynamically according to the actual situation; the self-diagnosis detection unit continuously monitors the working status of the sensor to ensure that the sensor output is stable, data collection and processing are normal, and self-checks that the signal feedback is correct, and promptly identifies hardware faults (such as signal loss or circuit problems) and software faults (such as data processing errors); at the same time, it identifies environmental anomalies (such as temperature changes, electromagnetic interference), signal anomalies (such as noise The automatic calibration module automatically performs the calibration process according to the predetermined cycle or the detected abnormality to ensure the accuracy and consistency of the sensor. When a fault is detected, the fault feedback and alarm unit immediately issues an alarm to notify the user to perform maintenance. The edge computing node unit is responsible for performing preliminary data processing on each sensor node, reducing the delay in data transmission and reducing the burden on the central processor. The distributed processing architecture improves the processing efficiency and reliability of the entire system by coordinating data processing of multiple nodes. The real-time response module ensures that the system can respond quickly to fire events by optimizing data processing and decision-making processes. These units work together to achieve efficient operation and accurate monitoring of the fire alarm system, ensuring stable and reliable fire detection and alarm services in a changing and complex environment.

[0166] Please see attached Fig.16 -Attached Fig.17 ,The environmental compensation technology module includes an environmental feature learning unit, an ,automatic compensation unit, and an environmental monitoring and adjustment unit;

[0167] The environmental feature learning unit uses a random forest model to analyze and store environmental feature data;

[0168] Random forest model building and training includes the following steps:

[0169] S1. Data collection and preparation: Continuously collect data from various sensors (temperature, humidity, air quality, light, noise), process missing data, outliers and noise to ensure data quality, and finally determine the key environmental characteristics related to fire detection;

[0170] S2. Model building: Select random forest as the environmental feature learning model, divide the data set into training set, validation set and test set (such as 70% training, 15% validation, 15% test), and then set the parameters of random forest, such as the number of trees and maximum depth;

[0171] S3. Model training: Use the training set data to train the random forest model, and then use the cross-validation method (such as K-fold cross-validation) to optimize the model parameters to prevent overfitting. Finally, during the training process, feature importance analysis is used to identify the features that have the greatest impact on the output.

[0172] S4, Model evaluation and deployment: Evaluate the model performance on the test set, using indicators such as accuracy, recall, and F1-score, and then adjust the model parameters based on the evaluation results to further improve the model accuracy. Finally, integrate the trained model into the environmental feature learning unit for real-time analysis and update of environmental feature data;

[0173] Environmental characteristics include temperature, humidity, air quality, lighting conditions, airflow rates, and noise levels;

[0174] The automatic compensation unit is used to dynamically adjust the sensor output according to the environmental feature library;

[0175] The environmental monitoring and adjustment unit is used to monitor environmental changes in real time and automatically adjust system parameters as needed;

[0176] The data transmission module includes a wired transmission unit, a wireless transmission unit and a data compression and encryption unit;

[0177] The wired transmission unit transmits data via optical fiber or Ethernet;

[0178] The wireless transmission module uses Wi-Fi, Bluetooth or Zigbee for wireless data transmission;

[0179] The data compression and encryption unit is used to compress and encrypt data.

[0180] Specifically, the environmental feature learning unit uses the random forest model to analyze and store key environmental feature data, including temperature, humidity, air quality, lighting conditions, airflow rate and noise level. The establishment and training of the random forest model includes four steps: first, data collection and preparation to ensure that the data obtained from the sensor is of reliable quality after processing and to identify key features related to fire detection; second, the model establishment link divides the data set into training, validation and test sets, and sets the model parameters; then the model is trained, and the parameters are optimized through cross-validation to ensure the generalization ability of the model and identify the features that have the greatest impact on the output; finally, the model is evaluated on the test set, and its performance is verified using indicators such as accuracy, recall and F1-score. After adjustment, it is integrated into the system for real-time analysis and updating of the environment The automatic compensation unit dynamically adjusts the output of the sensor according to the real-time updated environmental feature library to improve the detection accuracy, while the environmental monitoring and adjustment unit is responsible for real-time monitoring of environmental changes and automatically adjusting system parameters accordingly; the wired transmission unit performs stable data transmission through optical fiber and Ethernet, which is suitable for occasions with high requirements on transmission speed and security; the wireless transmission module uses Wi-Fi, Bluetooth or Zigbee for flexible wireless data transmission, which is suitable for scenarios requiring rapid deployment and mobility requirements. The data compression and encryption unit is responsible for effectively compressing and encrypting the data to ensure high efficiency and security during the data transmission process. These units work closely together to ensure the sensitivity and accuracy of the system under different environmental conditions, while ensuring the efficiency and security of data transmission, providing comprehensive support for fire detection and alarm.

[0181] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A composite fire alarm system using photoelectric smoke detection and image transmission recognition, characterized in that: include: A dust cover (1) is provided with a dust net (2) integrally formed inside the dust cover (1); an upper cover (8) is integrally formed below the dust cover (1); a dust net-blocked quality inspection transmitting tube and a dust net-blocked quality inspection receiving tube are provided on the upper cover (8); the dust net-blocked quality inspection transmitting tube and the dust net-blocked quality inspection receiving tube are both located between the dust net (2); a convex lens (3) is provided inside the upper cover (8); a filter (4) is provided on one side of the convex lens (3); and a buzzer (5) is installed between the convex lens (3) and the filter (4); The antenna (6) is connected to the upper cover (8) via a buzzer sheet (5); a QR code sticker (7) is provided on a side of the upper cover (8) close to the antenna (6); a camera module (9) is installed on the upper cover (8) via a double-sided sticker (11); a flame module (10) is installed on the upper cover (8) via a double-sided sticker (12); and one side of the upper cover (8) is connected to a maze cover (13); A labyrinth (14) is installed inside the upper cover (8) via a labyrinth cover (13); a transmitting tube 1 (15) and a transmitting tube 2 (16) are equidistantly distributed on one side of the labyrinth (14); a receiving tube (17) is arranged between the transmitting tube 1 (15) and the transmitting tube 2 (16); a shielding cover (18) is arranged on one side of the receiving tube (17); the shielding cover (18) is used to shield the position of the receiving tube (17); A PCBA mainboard (21) is mounted on a side of the maze (14) away from the maze cover (13), and a battery spring positive electrode (19) and a battery spring negative electrode (20) are equidistantly distributed on one side of the PCBA mainboard (21); A base (24) connected to the upper cover (8), a support (23) being fixed inside the base (24), a battery (22) being installed inside the support (23), and a foot pad (25) being fixed on a side of the base (24) away from the support (23); A first locking plate (26) is detachably arranged on the base (24), a locking component (28) is detachably mounted on the first locking plate (26), a second locking plate (27) is detachably mounted on the bottom of the locking component (28), and the locking component (28) comprises: A locking frame (281) is detachably mounted on the first locking plate (26), and an embedded block (282) is detachably mounted therein, forming a clamping cavity between the frame and the embedded block (282); An external plug-in plate (283) is slidably arranged in the locking frame (281), a first roller (284) is rotatably arranged on one rear end thereof, an ear plate (285) is detachably arranged on the external plug-in plate (283), a side plug-in rod (286) is detachably arranged on the ear plate (285), and the side plug-in rod (286) passes through the locking frame (281), a connecting spring (287) is removably sleeved on the side plug-in rod (286), and the connecting spring (287) is in contact with the ear plate (285) and the locking frame (281), respectively, and a limiting plate (288) is arranged on the end of the side plug-in rod (286) passing through the locking frame (281); A limiting insert ring (289) is detachably mounted on the second locking plate (27), and a blocking block (2810) is detachably mounted on the limiting insert ring (289). The blocking block (2810) has a first inclined surface (2811), and a through hole (2812) is formed on the blocking block (2810). The through hole (2812) can penetrate into the limiting insert ring (289), and the through hole (2812) can allow the external insert plate (283) to pass through. An unlocking seat (2819) has an unlocking pad (2820) integrally formed on its side surface, a second inclined surface (2821) inside, and a cross locking shaft (2822) detachably disposed on its bottom surface, and the cross locking shaft (2822) can pass through the locking frame (281), the limiting insert ring (289), and the embedded block (282) in sequence; An unlocking rod (2823), one end of which is detachably mounted on one end of the external plug-in plate (283), and a second roller (2824) is rotatably arranged on the other end of the unlocking rod (2823), and the second roller (2824) is located above the unlocking pad (2820); A limiting collar (2825) is detachably mounted on the locking frame (281) and can allow the unlocking rod (2823) to pass through.

2. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 1 is characterized in that: The locking component (28) further comprises: a reinforcing plug plate (2813) which is slidably inserted into the limiting plug ring (289), and one end of which is initially located in the through hole (2812), an inclined plate (2814) is integrally formed on the reinforcing plug plate, a fitting pad (2815) is integrally formed on the inclined plate (2814), a first connecting base plate (2816) is detachably provided on the reinforcing plug plate (2813), and a hanging spring (2818) is removably hung on the first connecting base plate (2816); a second connection substrate (2817) which is detachably mounted in the limit insert ring (289) and is hung by the hanging spring (2818); the PCBA mainboard (21) comprises a photoelectric smoke detection module, an image transmission recognition module, a temperature sensor module, an intelligent data fusion module, a self-diagnosis and calibration mechanism module, an edge computing and distributed processing module, an environmental compensation technology module and a data transmission module; The photoelectric smoke detection module is used to detect smoke particles in the air; The image transmission and recognition module is used to capture fire scene images and perform recognition analysis; The temperature sensing module is used to monitor changes in ambient temperature; The intelligent data fusion module is used to integrate multiple sensor data; The self-diagnosis and calibration mechanism module is used to maintain the accuracy and reliability of the system sensor; The edge computing and distributed processing modules are used to improve the real-time response capability of the system; The environmental compensation technology module is used to reduce the interference of environmental factors on system detection; The data transmission module is used to realize efficient transmission of images and data.

3. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 2 is characterized in that: The photoelectric smoke detection module includes a light source and a receiving unit and a sensitivity adjustment unit: The light source and receiving unit are used to detect smoke scattering through the light beam and determine the smoke concentration; The sensitivity adjustment unit is used to automatically adjust the sensitivity to adapt to different environmental conditions.

4. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 3 is characterized in that: The image transmission recognition module includes a camera capture unit, an image analysis unit and an image compression and transmission unit; The camera capture unit uses a camera to capture fire scene images in real time; The image analysis unit uses a convolutional neural network algorithm to identify flame and smoke features; The convolutional neural network algorithm formula is specifically: ； in is the activation function output, is the convolution kernel weight, is the input feature, is the bias term, is the activation function; The image compression and transmission unit is used to compress the image data into JPEG format and transmit it to the central processing unit through the network.

5. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 4 is characterized in that: The temperature sensing module includes a temperature detection unit and a threshold alarm unit; The temperature detection unit uses a digital temperature sensor to detect and record changes in ambient temperature in real time; The threshold alarm unit is used to trigger the buzzer (5) to sound an alarm when an abnormal temperature change is detected. The threshold is set to an absolute temperature higher than 70° C., and the abnormal temperature change value exceeds the set threshold.

6. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 5 is characterized in that: The intelligent data fusion module includes a Bayesian network fusion unit, a Kalman filter processing unit and a dynamic weight adjustment unit; wherein, The Bayesian network fusion unit is used to perform probability analysis on different sensor data through the Bayesian network; The Bayesian network algorithm formula is: ； For the given evidence Assumption The posterior probability of Assumption Evidence The likelihood, Assumption The prior probability of For evidence The marginal probability of The Kalman filter processing unit uses a Kalman filter algorithm to dynamically adjust and optimize sensor data; The specific formula of the Kalman filter algorithm is: ； in, To predict the state estimate, is the prediction covariance matrix, is the Kalman gain, is the observed quantity, , , are the state transfer, control and observation matrices respectively, is the observation matrix, , are the process noise covariance matrix and the observation noise covariance matrix, respectively. is the control input matrix, is the control vector, is the identity matrix, For time; The dynamic weight adjustment unit automatically adjusts the fusion weight according to the sensor data characteristics.

7. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 6 is characterized in that: The self-diagnosis and calibration mechanism module includes a self-diagnosis detection unit, an automatic calibration module and a fault feedback and alarm unit; The self-diagnosis detection unit is used to continuously monitor the working status of the sensor and identify faults and abnormalities; The working state means that the sensor output is stable and within the expected range, data acquisition and processing are normal, and self-test signal feedback is normal; The fault feedback includes hardware fault and software fault; The abnormalities include environmental abnormalities, signal abnormalities and sensitivity abnormalities; The automatic calibration module is used to automatically perform a sensor calibration process; The fault feedback and alarm unit is used to issue an alarm to the user when a fault is detected.

8. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 7 is characterized in that: The edge computing and distributed processing module includes an edge computing node unit, a distributed processing architecture and a real-time response module; The edge computing node unit is used to perform local data processing at the sensor node; The distributed processing architecture is used to implement multi-node collaborative data processing; The real-time response module is used to optimize data processing and decision-making speed.

9. The composite fire alarm system using photoelectric smoke detection and image transmission recognition according to claim 8 is characterized in that: The environmental compensation technology module includes an environmental feature learning unit, an automatic compensation unit and an environmental monitoring and adjustment unit; The environmental feature learning unit uses a random forest model to analyze and store environmental feature data; The random forest model establishment and training includes the following steps: S1. Data collection and preparation: Continuously collect data from various sensors, process missing data, outliers and noise, ensure data quality, and finally determine the key environmental features related to fire detection; S2. Model building: Select random forest as the environmental feature learning model, divide the data set into training set, validation set and test set, and then set the parameters of random forest; S3. Model training: Use the training set data to train the random forest model, and then use the cross-validation method to optimize the model parameters. Finally, during the training process, identify the features that affect the output through feature importance analysis; S4, model evaluation and deployment: Evaluate the model performance on the test set to obtain the evaluation results, adjust the model parameters according to the evaluation results, and finally integrate the trained model into the environmental feature learning unit for real-time analysis and update of environmental feature data; The environmental characteristics include temperature, humidity, air quality, lighting conditions, air flow rate, and noise level; The automatic compensation unit is used to dynamically adjust the sensor output according to the environmental feature library; The environmental monitoring and adjustment unit is used to monitor environmental changes in real time and automatically adjust system parameters as needed; The data transmission module includes a wired transmission unit, a wireless transmission unit and a data compression and encryption unit; The wired transmission unit transmits data via optical fiber or Ethernet; The wireless transmission module uses Wi-Fi, Bluetooth or Zigbee to perform wireless data transmission; The data compression and encryption unit is used to compress and encrypt data.

Images