Multifunctional intelligent fire fighting robot system based on BP network and EKF

Abstract

The invention discloses a multifunctional intelligent fire fighting robot system based on a BP network and an EKF, and relates to the technical field of intelligent fire fighting, a fire monitoring and early warning module is a wireless data acquisition node which is identified through a three-light composite fire source detection sensor and an FIS intelligent image and takes an STM32F103 single-chip microcomputer as a core; the fire behavior identification module realizes a fire source identification function through RFID; the control module achieves real-time positioning in an outdoor environment through an RTK navigation positioning system, meanwhile, mapping is conducted on an area through an SLAM technology, and the method is achieved through an EKF multi-sensor information fusion technology and a BP neural network algorithm. Risks caused by sudden, contingency and rapid development of fire disasters are solved, fire source information is obtained through a smoke and fire detection early warning technology and a three-light composite fire source detection sensor, the fire source information is accurately recognized, and a proper fire extinguishing mode is adopted; the fire fighting capacity is improved in a higher-efficiency fire extinguishing mode by processing fire scenes under different conditions.

Description

Technical Field

[0001] The present invention relates to the field of intelligent firefighting technology, and in particular to a multifunctional intelligent firefighting robot system based on BP network and EKF. Background Art

[0002] With the acceleration of urban construction in my country, the increase in the number of large buildings and the increase in the flow of people, the number of fires in my country has gradually increased in recent years. Complex environment, interrupted information exchange and uneven facility configuration are several major factors that make fire difficult to control. Prosperous commercial areas, busy enterprises and factories with complex environments are all severely affected by fires, and once a fire occurs in these areas, it will bring huge property losses. Faced with new problems and hidden dangers in the new era, fire prevention and fire fighting can no longer follow the traditional model, and we must seek safer and more effective ways. Faced with irreversible losses after the disaster, we should consider how to reduce the frequency of fires and how to reduce casualties and losses. Summary of the invention

[0003] The purpose of the present invention is to address the defects and shortcomings of the prior art and provide a multifunctional intelligent firefighting robot system based on BP network and EKF to solve the risks brought by the suddenness, contingency and rapid development of fires. Fire source information is obtained through smoke and fire detection early warning technology and three-light composite fire source detection sensor, and the fire source information is accurately identified to adopt appropriate fire extinguishing methods. By handling fire scenarios under different circumstances, the firefighting capability is improved with a more efficient fire extinguishing method, while the loss of property caused by fire is effectively reduced. In addition, the early treatment of the initial fire can effectively save firefighting resources and reduce the risk of firefighting.

[0004] To achieve the above-mentioned purpose, the present invention adopts the following technical scheme: it comprises a fire monitoring and early warning module, a fire condition identification module and a control module; the fire monitoring and early warning module is connected to the control module, and the fire condition identification module is connected to the control module; the fire monitoring and early warning module is a wireless data acquisition node which uses a three-light composite fire source detection sensor and FIS intelligent image recognition, and takes the STM32F103 single-chip microcomputer as the core; the fire condition identification module realizes the fire source identification function through RFID; the control module realizes real-time positioning in an outdoor environment through an RTK navigation and positioning system, and uses SLAM technology to map the area, which is realized through EKF's multi-sensor information fusion technology and BP neural network algorithm.

[0005] Preferably, the STM32F103 single chip microcomputer in the fire monitoring and early warning module integrates modules such as 485 bus communication module circuit and LoRa wireless communication module circuit.

[0006] Preferably, the fire monitoring and early warning module uploads the fire flame data of the images taken by the CDD camera to the cloud, and uses technologies such as DNN and BP neural network algorithms to integrate various characteristic information of the fire flame, and designs a BP neural network recognition system for early fire flame detection in the cloud.

[0007] Preferably, the fire monitoring and early warning module constructs a YOLO model to obtain flame data, substitutes it into the DNN for output, and intelligent real-time judgment improves the fire source identification effect.

[0008] Preferably, the fire identification module uses RFID as identification technology to read and transmit information stored in electronic tags using wireless radio frequency signals without direct contact, so as to achieve the purpose of identifying targets and exchanging data.

[0009] Preferably, the control module adopts the EFK operation process of estimating the optimal posture of the robot. By performing a series of iterative replacement updates on the posture estimation, the optimal estimation of the posture of the fire-fighting robot can be obtained, and after training, it can become a fuzzy neural network.

[0010] Compared with the prior art, the present invention has the following beneficial effects:

[0011] 1. Use SLAM technology to map the area, improve the robot's automatic navigation and obstacle avoidance technology performance, improve the robot's learning ability, and achieve the adaptability and stability of the multifunctional intelligent firefighting robot system that integrates fire warning and fire control;

[0012] 2. Based on the RGB criteria, HSI constraints are added, and the conversion of RGB images to HSI images can be realized through MATLAB software, which improves the accuracy of fire source identification and the predictability of initial flames;

[0013] 3. Optimizing fire protection construction by leveraging technologies such as the Internet of Things will help promote the development of smart cities, accelerate the integration of fire protection informatization towards digitalization and intelligence, and comprehensively upgrade the fire protection information network structure to meet the needs of social security and stable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a module block diagram of the present invention.

[0015] Figure 2 It is an operating principle diagram of the fire monitoring and early warning module in the present invention.

[0016] Figure 3 It is a flow chart of the fire identification module in the present invention.

[0017] Description of reference numerals:

[0018] Fire monitoring and early warning module 1, fire situation identification module 2, control module 3. DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. The preferred embodiments in the description are only used as examples, and all other embodiments obtained by those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, this specific implementation adopts the following technical scheme: it includes a fire monitoring and early warning module 1, a fire identification module 2 and a control module 3; the fire monitoring and early warning module 1 is connected to the control module 3, and the fire identification module 2 is connected to the control module 3; the whole system is composed of STM32 single-chip microcomputer, CDD camera, three-light composite fire source detection sensor and RFID tag as hardware equipment, and BP neural network algorithm, DNN network algorithm, RGB grayscale threshold segmentation method, RTK navigation technology and EKF multi-sensor information fusion technology are used as software technology to realize system functions; wherein, through the radio acquisition node with STM32F103 single-chip microcomputer as the core, combined with FIS intelligent image recognition, the smoke and fire warning function is realized, and the fire source information acquisition function is realized through the three-light composite fire source detection sensor; the fire source identification function is realized through RFID; the mapping and navigation positioning function is realized through RTK technology; the robot obstacle avoidance function is realized through EKF multi-sensor information fusion technology and BP neural network algorithm.

[0021] See also Figure 2 As shown, the fire monitoring and early warning module 1 first adopts FIS intelligent image recognition, a wireless data acquisition node with STM32F103 single chip microcomputer as the core, and combines 485 bus communication module circuit, LoRa wireless communication module circuit and other technologies to upload the fire flame data of the CDD camera image to the cloud; then, using DNN, BP neural network algorithm and other technologies, various characteristic information of fire flames are integrated, and a BP neural network recognition system for early fire flame detection is designed in the cloud; the CDD camera multi-angle shooting technology in the building is used to record the scene of the flame, and YOLO modeling is performed on the terminal. The BP neural network judges whether it is a flame of fire nature based on the fire source, and records the location of the fire source. If a fire actually occurs, the robot will be dispatched and use the three-light composite fire source detection sensor to find the root of the fire. At the same time, the CDD camera will provide the distribution of fire sources inside the burning building to determine whether it is a fire source, providing assistance for fire rescue work. It solves the dilemma caused by the complexity of the fire sources of a fire, the fire situation can only be understood by the firefighters on the scene and the guidance outside the scene, and the firefighters cannot know the specific location of the fire source in advance before arriving at the fire scene. Figure 2 x1-x5 are the different flame characteristics summarized; Wij are different pairing methods; y is the output flame type; a j =W ij (x i ); a=v j (a j ); y = f(a).

[0022] See also Figure 3 As shown, the fire identification module 2 converts the RGB image uploaded to the Web page platform into the HSV format, and uses the Ret-inex algorithm to filter the image information, and uploads it to the cloud database for image comparison, so as to identify the burning object, and then decide on a specific fire extinguishing plan; the specific working principle of the fire identification module 2: in 3D computer graphics, the weighted average method is used for the RGB color image obtained by the visible light camera, and the target area of ​​the suspected fire source is segmented on the grayscale image based on the threshold segmentation method. The threshold segmentation method mainly determines whether each pixel is in the target area by judging the threshold. The Depth Map is obtained by the above grayscale processing. The Depth Map is an image or image channel containing information related to the distance of the surface of the scene object of the viewpoint. Each pixel value is the actual distance between the sensor and the object. Usually, the RGB image and the Depth image are registered, so there is a one-to-one correspondence between the pixels, which can display the depth position relationship of the detected fire source.

[0023] As one of the supporting technologies of the Internet of Things, the RFID is an identification technology that does not require direct contact. It uses wireless radio frequency signals to read and transmit the information stored in the electronic tag, so as to achieve the purpose of identifying the target and exchanging data. A fire may have a single or multiple ignition points, and the source of the ignition point will be affected by factors such as the burning material and the amount. If the appropriate and correct amount of fire extinguishing agent is not used, it will not only reduce the efficiency of fire extinguishing, but may even increase the fire. RFID tags can be used to mark each item in the current scene. If a tagged item catches fire, the RFID tag will be fed back to the terminal based on the temperature sensor, so that the fire-fighting robot can correctly use the corresponding fire-fighting agent when extinguishing the fire, and quickly and effectively extinguish the fire point. For foreign objects that are not RFID-tagged, RGB image recognition and database analysis technology are used to supplement. Current technology does not take into account the combustible materials of the fire source. The combination of RFID tags and RGB image recognition effectively solves the impact of the ignition material on the fire. When using it, the RFID tag is first covered with high-temperature resistant material, and then the tag is placed on each device in the area to be monitored one by one and the corresponding device information is entered. By entering the RFID tag information, the system can easily identify the fire source, so that different fire extinguishing materials can be used to extinguish the fire in the fire area.

[0024] The control module 3 adopts high-precision positioning technology based on the RTK positioning system to achieve real-time positioning in outdoor environments. Before autonomous navigation, SLAM technology needs to be used to map the area. Mapping is implemented by the GMapping function package, which performs positioning and mapping based on the particle filter algorithm. Specifically: manually control the robot to walk on the road, obtain RTK positioning system data in real time and calculate the robot odometer information to preliminarily determine the robot's posture. At the same time, the distance information of the surrounding environment and obstacles obtained by the laser radar scan is obtained in real time as observation data, and the robot odometer motion model and the laser radar observation data are added to the proposal distribution function at the same time to calculate the weight of the particles in the particle filter, and the robot's posture in the map is corrected. The local map is constructed based on the corrected robot posture and the global map update is completed, and finally the global map of the entire area is completed.

[0025] The multi-sensor information fusion technology of EKF and BP neural network algorithm, the operation flow of estimating the optimal posture of the robot using EFK can be divided into 5 parts; the state equation of the system can be used to estimate the current posture of the fire-fighting robot from the optimal posture estimation system at the previous time point; the optimal estimation of the posture of the fire-fighting robot can be obtained by performing a series of iterative replacement updates on the posture estimation; when the sensor measures that the distance between the vehicle body and the obstacle is greater than the set value, the distance information in the control system is marked as the set value, and the system input is "far"; at this time, the robot continues to drive; when the sensor measures that the distance between the vehicle body and the obstacle is less than the set value, the system input is "near"; at this time, the robot turns according to the measured position of the obstacle to avoid the obstacle; the system fuzzy variables [Z, N, P] = [the target position is on the left side of the robot's central axis, the angle is negative; the target position is at the intersection of the two central axes of the robot body, the angle is 0; the target position is on the right side of the robot's central axis, the angle is positive]. The value range of the speed membership function of the left and right driving wheels output by the system is [0,1], indicating that the robot's driving speed is 0-1m / s. The angle between the robot body and the target position is 0 when it is 30°, negative when it is less than 30°, and positive when it is greater than 30°; 0.5m / s is the dividing point of the robot's left and right driving wheel speeds. After training, it becomes a fuzzy neural network.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] 1. Use SLAM technology to map the area, improve the robot's automatic navigation and obstacle avoidance technology performance, improve the robot's learning ability, and achieve the adaptability and stability of the multifunctional intelligent firefighting robot system that integrates fire warning and fire control;

[0028] 2. Based on the RGB criteria, HSI constraints are added, and the conversion of RGB images to HSI images can be realized through MATLAB software, which improves the accuracy of fire source identification and the predictability of initial flames;

[0029] 3. Optimizing fire protection construction by leveraging technologies such as the Internet of Things will help promote the development of smart cities, accelerate the integration of fire protection informatization towards digitalization and intelligence, and comprehensively upgrade the fire protection information network structure to meet the needs of social security and stable development.

[0030] For those skilled in the art, they can modify the technical solutions described in the aforementioned embodiments and make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multifunctional intelligent firefighting robot system based on BP network and EKF, characterized by: It comprises a fire monitoring and early warning module (1), a fire identification module (2) and a control module (3); the fire monitoring and early warning module (1) is connected to the control module (3), and the fire identification module (2) is connected to the control module (3); the fire monitoring and early warning module (1) is a wireless data acquisition node using a three-light composite fire source detection sensor and FIS intelligent image recognition, and taking an STM32F103 single-chip microcomputer as the core; the fire identification module (2) realizes a fire source identification function through RFID; the control module (3) realizes real-time positioning in an outdoor environment through an RTK navigation and positioning system, and simultaneously uses SLAM technology to map the area, which is realized through EKF multi-sensor information fusion technology and BP neural network algorithm.

2. The multifunctional intelligent firefighting robot system based on BP network and EKF according to claim 1 is characterized in that: The STM32F103 single chip microcomputer in the fire monitoring and early warning module (1) is integrated with a 485 bus communication module circuit and a LoRa wireless communication module.

3. The multifunctional intelligent firefighting robot system based on BP network and EKF according to claim 2 is characterized in that: The fire monitoring and early warning module (1) uploads the fire flame data of the images captured by the CDD camera to the cloud, and uses the DNN and BP neural network algorithms to integrate the characteristic information of the fire flame, and designs a BP neural network recognition system for early fire flame detection in the cloud.

4. The multifunctional intelligent firefighting robot system based on BP network and EKF according to claim 3 is characterized by: The fire monitoring and early warning module (1) constructs a YOLO model to obtain flame data, substitutes it into the DNN for output, and intelligent real-time judgment improves the fire source identification effect.

5. The multifunctional intelligent firefighting robot system based on BP network and EKF according to claim 4 is characterized in that: The fire identification module (2) uses a wireless radio frequency signal without direct contact to read and transmit the information stored in the electronic tag RFID as an identification technology, thereby achieving the purpose of identifying the target and exchanging data.

6. The multifunctional intelligent firefighting robot system based on BP network and EKF according to claim 5 is characterized by: The control module (3) adopts the EFK operation process of estimating the optimal posture of the robot. By performing a series of iterative replacement and updating of the posture estimation, the optimal estimation of the posture of the fire-fighting robot can be obtained. After training, it can become a fuzzy neural network.

Images