CN116030591A - Intelligent inspection alarm system and method for mine external fire disaster based on Internet of things - Google Patents
Intelligent inspection alarm system and method for mine external fire disaster based on Internet of things Download PDFInfo
- Publication number
- CN116030591A CN116030591A CN202310031647.XA CN202310031647A CN116030591A CN 116030591 A CN116030591 A CN 116030591A CN 202310031647 A CN202310031647 A CN 202310031647A CN 116030591 A CN116030591 A CN 116030591A
- Authority
- CN
- China
- Prior art keywords
- data
- module
- intelligent
- fire
- internet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Alarm Systems (AREA)
Abstract
The invention discloses an intelligent inspection alarm system and method for mine external fire disaster based on the Internet of things, which belong to the technical field of fire disaster alarm systems.
Description
Technical Field
The invention relates to the technical field of fire alarm systems, in particular to an intelligent inspection alarm system and method for mine external fire based on the Internet of things.
Background
The fire in the belt roadway is taken as one of mine roadway fires, the life safety of underground workers is endangered due to the characteristics of randomness of occurrence places, toxicity of fire products, limitation of disaster relief space, sudden occurrence and development and the like, and the fire in the belt roadway is one of main disasters in mine production, if the fire can be found timely, the development trend of the fire in the belt roadway can be mastered in advance and the fire can be alarmed timely, so that the economic loss in the production process of the coal mine can be effectively reduced. At present, most mine fire alarm systems only configure sensors to monitor fixed places, the processing mode of the collected characteristic data is single, an effective fusion algorithm is lacked, the false alarm probability is high, and the detection accuracy is low. The zigbee wireless transmission technology commonly adopted in mines has the advantages of short transmission distance, low signal strength, weak diffraction capacity, networking resource waste and weak wall penetrating capacity; the bluetooth technology has a limited transmission distance, low data transmission rate and incompatibility of different protocols between devices. If underground monitoring equipment discovers a potential fire source, huge potential safety hazards can be caused when workers go to and look over, personal safety of the workers is endangered, and negative influence is caused to society. The existing fire alarm system cannot intuitively know underground monitoring equipment and distribution conditions of workers, cannot determine potential fire source positions, and cannot provide effective assistance for later fire extinguishment and rescue. Therefore, it is needed to provide an intelligent inspection and reporting method and system for mine external fire based on the internet of things, so as to solve the above problems.
Disclosure of Invention
In order to solve the problems, the invention provides an intelligent mine external fire inspection alarm system and method based on the Internet of things.
In order to achieve the above purpose, the present invention provides the following technical solutions:
mine is because of fire intelligence inspection alarm system outward based on thing networking, its characterized in that includes:
the detection and identification module is used for acquiring current underground data;
the data collection module is used for generating measurement data according to the underground data; the measurement data comprises temperature, resistance, image signals and CAN signals;
the information pre-analysis processing module is used for identifying the measurement data to obtain data quantity and characteristic labels; calculating the number of common statistics and missing values, and carrying out normalization processing on the measured data to obtain metadata; the normalization processing step comprises cleaning, conversion and filling;
the central intelligent processing terminal is used for storing metadata, equipment message data and equipment time sequence data, establishing a detection model based on a preset deep learning algorithm, and training the detection model based on the underground data;
the remote monitoring terminal is used for acquiring a fire alarm signal containing a fire place and output by the detection model in real time by combining a three-dimensional digital map, acquiring an infrared picture and a visible light picture of the fire place based on a preset inspection robot, and generating emergency measures, wherein the emergency measures specifically comprise making an escape route and a fire extinguishing scheme, starting a buzzer alarm and a warning lamp of the inspection robot, and informing workers in an area nearby the fire place to leave as soon as possible according to the escape route through the voice interphone.
As a further aspect of the present invention, the detection and identification module includes:
the temperature sensor is used for reading temperature data;
the smoke sensor is used for reading smoke concentration data;
the track inspection robot is used for reading the image data;
and the personnel ID recognition unit is used for reading the personnel identity information and the position information.
As a further scheme of the invention, the data collection module adopts a multi-channel data acquisition instrument.
As a further aspect of the present invention, the wireless communication module further includes a wireless communication module, where the wireless communication module includes a 5G base station, wi-Fi6 technology, and a wireless radio frequency transceiver.
As a further scheme of the invention, the central intelligent processing terminal comprises a data storage module, an intelligent processor, a grabbing module, a matching module, a feedback module and a power supply module; the data storage module provides a unified data storage platform for equipment metadata, equipment message data and equipment time sequence data in the scene of the Internet of things; the intelligent processor is respectively connected with the grabbing module, the matching module and the feedback module, the information received from the data storage module is conveyed to the grabbing module and the matching module through the intelligent processor to be screened and matched, the feedback module conveys the screened and matched information back to the intelligent processor, the accuracy of the received data is improved, and useless information is screened and removed.
As a further scheme of the invention, the three-dimensional digital map is combined with a geographic information system technology, a database technology and a three-dimensional technology to display the spatial hierarchy and the position of underground equipment.
As a further proposal of the invention, the temperature sensor is a digital temperature sensor of model TMP126, and the smoke sensor is an MQ-2 smoke sensor.
As a further scheme of the invention, the patrol robot uses an ARM+LINUX master control system, cm-level positioning is realized through 3DSLAM+IMU+GPS fusion, the robot can automatically avoid obstacles through the collocation of a laser radar and front and rear cameras, a trolley line is used for charging, the cruising distance is 10km, the emergency parking performance is controlled below 0.2m, and the climbing capacity is less than or equal to 45 degrees.
As a further scheme of the invention, the inspection robot comprises a double-spectrum high-definition camera, the double-spectrum high-definition camera adopts an HB-8501R-F9432-T9310 type intrinsic safety thermal imaging double-spectrum integrated camera to support point, line and frame temperature measurement, and the highest temperature is highlighted through a cross cursor to enhance image details.
As a further scheme of the invention, the escape route established by the inspection robot is based on the D_Star algorithm, and the optimal path calculated by the D_Star algorithm reduces the interference degree.
As a further scheme of the invention, each device in the detection and identification module is provided with a positioning tag, the notes send UWB signals and are communicated with the 5G base station, the 5G base station sends the network original data to the positioning engine in real time, the positioning engine operates a positioning algorithm, the coordinate position with the positioning tag is calculated in real time, and finally the coordinate position is displayed on the real-time display module.
As a further scheme of the invention, the wireless communication module adopts a 5G base station+WIFI 6 cooperative mode to realize wireless full coverage; the 5G base station uses a BBU-AAU architecture and comprises a 5G baseband unit and a 5G radio frequency unit, the wireless communication module further comprises a wireless radio frequency transceiver, an ADRV9008 type is adopted as a receiver, and an ADRV9009 type is adopted as a transmitter.
As a further scheme of the invention, the intelligent mobile monitoring terminal is also provided, cross-system linkage is realized, the intelligent mobile monitoring terminal is close to the equipment identification card on the detection and identification module equipment, acquires and connects the IP address of the detection and identification equipment, realizes that no screen is changed into a screen, and the fixed key operation is changed into mobile operation, so that parameter checking, modification and monitoring of the equipment are realized rapidly and efficiently.
As a further scheme of the invention, the deep learning algorithm adopts YOLO, which is an algorithm model capable of rapidly carrying out target detection, and the main purpose of the deep learning algorithm is to create an S X S unit lattice for an identified image, each unit lattice evaluates objects in the unit lattice, predicts confidence scores of boundary frames and boundary frames of each unit lattice, extracts features of a convolution layer, reduces a feature map of a pooling layer to simplify a complex graph, simultaneously carries out feature compression and extracts main features, and connects all features to a classifier; the algorithm has the advantages of simple and convenient framework, high recognition accuracy and high recognition speed.
As a further scheme of the invention, the algorithm model is trained by using a TensorFlow framework, wherein TensorFlow is a deep learning training framework based on data flow programming and can be compatible with various servers; the system is characterized in that a YOLO network adopts 9 convolution layers for pre-training, 1 full connection layer is added, and a maximum pooling layer and an activation function ReLU are added after each convolution layer;
the deep learning model is trained as follows: the training process is to input the marked flame/smoke pictures into a frame of the YOLO deep learning, iterate for a plurality of times, update common parameters of the model, enhance the characteristic recognition capability, prepare 4820 fire images for a training data set, set each 482 images as a unit and train;
the first 9 convolutional networks output a 9 x 1156 feature vector; by convolving conv pre Converting the vector into a vector of 81 dimensions; the possibility of positioning in the grid center is set to z= (z) 1 ，z 2 ，z 3 ，…，z 81 ) The method comprises the steps of carrying out a first treatment on the surface of the After normalizing it by Softmax classifier, we get t= (t 1 ，t 2 ，t 3 ，…，t 81 )；t i The normalized expression of (2) is as in formula (1),
the loss function of the pre-training stage is shown as (2)
The purpose of the pre-training stage is to use a minimized loss function Q pre Improving the accuracy of prediction, and formally training the loss function Q as formula (3)
Q＝(w-w') 2 +(h-h') 2 +Q pre (3)；
Wherein: w represents the input image pixel width; h represents the input image pixel height; w' represents the real pixel width of the input image; h' represents the true pixel height of the input image;
leading out the detection rate and the precision rate for testing the performance of the learning model;
the detection rate (D) is defined as the ratio of real data to the total number of fire samples accurately marked. Reflecting the ability of the learning model to recognize fire, the expression is as (4)
Wherein: a is that P Positive samples representing correct predictions; b (B) n Negative samples representing mispredictions;
the accuracy rate reflects the reliability of fire alarm, and the expression is shown as (5);
wherein: a is that P Positive samples representing correct predictions; b (B) p Representing a negative sample of the correct prediction.
The invention further provides an intelligent inspection alarm method for mine external fire based on the Internet of things, which comprises the following steps:
acquiring downhole data in real time;
generating measurement data from the downhole data; the measurement data comprises temperature, resistance, image signals and CAN signals;
identifying the measurement data to obtain data quantity and characteristic labels; calculating the number of common statistics and missing values, and carrying out normalization processing on the measured data to obtain metadata; the normalization processing step comprises cleaning, conversion and filling;
storing metadata, equipment message data and equipment time sequence data, establishing a detection model based on a preset deep learning algorithm, and training the detection model based on the underground data;
and acquiring a fire alarm signal containing a fire place output by a detection model in real time by combining a three-dimensional digital map, acquiring an infrared picture and a visible light picture of the fire place based on a preset inspection robot, and generating emergency measures.
Compared with the prior art, the invention has the beneficial effects that: the system can monitor fixed places and mobile places through the combination of the sensor and the inspection robot, the monitoring data is diversified, a fire alarm target detection model is trained by using a deep learning algorithm of machine learning, the false alarm probability is reduced, and the detection accuracy is improved. The inspection robot can reduce the workload of workers, and the inspection robot is used for going to high-risk areas where the workers cannot reach, so that casualties are effectively reduced, and the system adopts a 5G base station and WIFI6 cooperative mode to realize wireless full coverage, so that the problems of complex underground wiring, easy breakage and damage of optical fibers, difficult deployment and the like are solved. The three-dimensional digital map is combined with a Geographic Information System (GIS) technology, a database technology and a three-dimensional technology, the spatial hierarchy and the position of underground equipment are visually displayed, potential fire source points are rapidly determined, the intelligent mobile monitoring terminal can be in cross-system linkage, a worker holds the intelligent mobile monitoring terminal to be connected with the underground equipment through an equipment identification card, parameter checking, modification and monitoring of the equipment are rapidly and efficiently achieved, and real-time diagnosis of fire conditions of a mine is achieved.
Drawings
Fig. 1 is a frame diagram of an intelligent inspection alarm system for mine external fire based on the internet of things, which is provided by the embodiment of the invention.
Fig. 2 is a schematic diagram of a detection and identification module in an intelligent inspection alarm system for mine external fire based on the internet of things according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a track inspection robot in an intelligent inspection alarm system for mine external fire based on the internet of things according to an embodiment of the invention;
fig. 4 is a schematic diagram of a wireless communication module structure in an intelligent inspection alarm system for mine external fire based on the internet of things according to an embodiment of the present invention;
fig. 5 is a schematic flow diagram of a central intelligent processing terminal in an intelligent inspection alarm system for mine external fire based on the internet of things according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a remote monitoring terminal in an intelligent inspection alarm system for mine external fire based on the internet of things according to an embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
As an embodiment of the invention, please refer to fig. 1 to 6, an intelligent inspection alarm system for mine external fire based on internet of things comprises a detection and identification module, a data collection module, an information pre-analysis processing module, a wireless communication module, a central intelligent processing terminal, a remote monitoring terminal and an intelligent mobile monitoring terminal. The detection and identification module, the data collection module, the information pre-analysis processing module, the central intelligent processing terminal and the remote monitoring terminal are respectively connected with the wireless communication module in a two-way mode.
The detection and identification module comprises: temperature sensor, smoke sensor, track inspection robot, personnel ID recognition unit.
The temperature sensor model TMP126 is a digital temperature sensor with an accuracy of 0.25℃and supports an ambient temperature range of-55℃to 175 ℃. TMP126 has 14-bit temperature resolution (0.03125 ℃/LSB), has the characteristics of fast conversion rate, low power supply current, simple 3-wire SPI compatible interface and the like, and is suitable for various applications. To improve reliability in harsh environments, TMP126 also has other advanced features, such as improving data integrity through an optional CRC checksum; a programmable alarm limit; temperature slew rate warning and improved operating temperature range. To ensure accuracy, the device employs a NIST traceable factory scale and also employs a small SOT package that can be placed close to the heat source and has a short response time.
The smoke sensor is used for completing fire prevention according to the concentration of monitoring dense smoke, and when the smoke sensor is contacted with smoke, if the potential barrier at the grain boundary is changed due to the adjustment of the smoke, the change of the surface conductivity is caused. By using this, it is possible to obtain information on the presence of such smoke, and the greater the concentration of smoke, the greater the conductivity, and the lower the output resistance, the greater the analog signal output.
The robot is patrolled and examined to track, uses ARM+LINUX main control system, fuses through 3DSLAM+IMU+GPS, realizes centimeter level location, and the collocation of laser radar+front and back camera can make the robot realize automatic obstacle avoidance, uses the wiping line to charge, and the duration is 10km, and emergent parking performance control is below 0.2m, and climbing ability is less than or equal to 45, the robot is patrolled and examined to track still includes: the double-spectrum high-definition camera, the buzzer alarm, the warning light and the voice intercom.
The double-spectrum high-definition camera is different from a common visible light monitoring and infrared thermal imager, and can be used for simultaneously watching on-site pictures and on-site temperatures, so that the double-spectrum high-definition camera is clearer, more visual and more convenient. The system adopts the An Re imaging double-spectrum integrated camera with the model of HB-8501R-F9432-T9310, supports point, line and frame temperature measurement, supports image detail enhancement functions such as 3D noise reduction, manual AGC and the like by highlighting the highest temperature through a cross cursor, adopts special explosion-proof glass as a window, has high light flux, is free from water adhesion and oil adhesion and repels dust after the glass surface is treated by a nano technology, is manufactured by 304 stainless steel materials, and can be used in a highly corrosive environment.
And the buzzer alarm and the warning lamp start working after receiving the abnormal alarm signal of the central intelligent processing terminal, and remind workers of preparing for danger avoidance.
The voice intercom supports MIC and voice intercom functions, and communication convenience between underground staff and monitoring staff is effectively improved.
The path optimization of the inspection robot is based on a D_Star algorithm, the optimal path calculated by the D_Star algorithm fully considers the interference of the inspection robot on construction activities when collision with a working space cannot be avoided, and the interference degree is reduced, so that the probability and the safety risk of space collision accidents are reduced, and the method has practical application value in improving the safety management level and the safety management efficiency of construction sites.
The personnel ID recognition unit comprises an ID recognizer and a personnel recognition chip, wherein the ID recognizer is arranged at each entrance and exit of the underground roadway and is connected with the server through a wireless communication network.
The personnel identification chip is carried by a worker, the identity information of the worker such as the work type, the age and the like is recorded in the chip, and when the worker passes through one route node and approaches to the ID identifier in distance, the ID identifier determines the identity information of the worker through wireless electromagnetic induction. At the real-time display module, the downhole personnel profile may be determined by the ID identifier.
And each device is provided with a positioning tag, the note sends UWB signals and is communicated with the base station, and the base station sends positioning engines through network original data in real time. And the positioning engine runs a positioning algorithm, calculates the coordinate position of the positioning label in real time, and finally displays the coordinate position on a real-time display module.
The data collection module adopts a multichannel data acquisition instrument with the model of NR-500, and CAN easily measure temperature, resistance, image signals, CAN signals and the like. According to the purpose, the measuring unit is freely combined, and the necessary channel number is used for realizing high-speed and high-precision measurement. The method has the advantages that the steps of connecting an upper server, collecting and setting and the steps are simply and quickly completed, signals of the sensor are directly read through corresponding bus interfaces, and the operation time during data collection and before and after data collection is effectively reduced.
The information pre-analysis processing module has different data generation ways, definition modes, storage media, data quality and the like, so that standardized pre-processing is also needed for different types of data.
The information pre-analysis processing module receives the collected information transmitted by the data collection module, analyzes the size of the data and all the characteristic labels based on the service requirement, and calculates the number of common statistics and missing values. And carrying out normalization processing of cleaning, conversion and filling on the data.
The information pre-analysis processing module performs pre-analysis processing on the acquired information and then transmits the acquired information to the central intelligent processing terminal, and the central intelligent processing terminal transmits the acquired information after analysis processing to the real-time display module for display.
The detection and identification module, the data collection module and the information pre-analysis processing module are connected with the wireless communication module through serial ports, and the GPRS technology is utilized to provide end-to-end and wide-area wireless IP connection for the system, so that the detected data are transmitted in a wireless network manner, and the detected data are transmitted to an upper server through the detection end.
The wireless communication module sends or receives electromagnetic wave signals through a TCP/IP protocol as a bearing layer and converts the electromagnetic wave signals into information which can be understood by people. The wireless communication module is used for connecting objects with each other, so that various terminal devices of the Internet of things can realize information transmission capacity, and various intelligent devices can also be provided with an information interface of the Internet of things.
The wireless communication module has the advantages that the cost of 5G underground coverage is high, the terminal compatibility is weak, the Wi-Fi6 underground coverage has overcome the challenges of large bandwidth, large capacity and low time delay, the key application of large bandwidth and low time delay such as VR/4K/AGV can be supported, the 5G base station and WIFI6 cooperative mode is adopted, the mode has strong signals, ultra-low time delay and lossless roaming, wireless full coverage can be realized, and the problems of complex underground wiring, easy breakage and damage of optical fibers, difficult deployment and the like are solved.
The wireless communication module, the 5G base station uses BBU-AAU architecture, and comprises a 5G baseband unit and a 5G radio frequency unit, which can be connected through CPRI or eCPRI interface. The 5G baseband unit is responsible for NR baseband protocol processing, including overall User Plane (UP) and Control Plane (CP) protocol processing functions, and provides a backhaul interface (NG interface) with the core network and an inter-base station interconnect interface (Xn interface). The 5G radio frequency unit mainly completes the conversion of NR baseband signals and radio frequency signals and the receiving and transmitting processing functions of NR radio frequency signals. In the downlink direction, the baseband signal transmitted from the 5G baseband unit is received, and is transmitted through a switch and an antenna unit after being processed by an up-conversion, digital-to-analog conversion, radio frequency modulation, filtering, signal amplification and other transmitting links (TX). In the uplink direction, the 5G radio frequency unit receives uplink radio frequency signals through the antenna unit, and after being processed by a receiving link (RX) such as low noise amplification, filtering, demodulation and the like, the uplink radio frequency signals are subjected to analog-to-digital conversion and down-conversion, converted into baseband signals and sent to the 5G baseband unit.
The wireless communication module has the technical advantages of WiFi6 including a transmission rate of 9.6Gbit/s at the highest, higher concurrency capability, lower service delay, larger coverage range, lower terminal power consumption and the like. And reasonably distributing the space of the router according to the coverage range of the underground router.
The wireless communication module is configured with a wireless radio frequency transceiver, comprises a receiver and a transmitter,
the receiver, model ADRV9008, consists of two independent wide Bandwidth (BW) direct conversion receivers, and has an advanced dynamic range. The complete receive subsystem includes automatic and manual attenuation control, dc offset correction, quadrature Error Correction (QEC) and digital filtering, eliminating these functions in the digital baseband. The receiver also integrates radio frequency front end control and a plurality of auxiliary functions, such as an analog-to-digital converter (ADC), a digital-to-analog converter (DAC) and general purpose input/output (GPIO) for a Power Amplifier (PA); the transmitter is ADRV9009, and adopts an innovative direct conversion modulator, so that high modulation precision and extremely low noise can be realized. The observation path consists of a wide bandwidth direct conversion receiver with advanced dynamic range. The transmitter adopts an innovative direct conversion modulator, and can realize high modulation precision and extremely low noise.
The central intelligent processing terminal comprises a data storage module, an intelligent processor, a grabbing module, a matching module, a feedback module and a power supply module.
The data storage module provides a unified data storage platform for the device metadata, the device message data and the device time sequence data in the scene of the Internet of things. Wherein device metadata is stored using a wide table model, device message data is stored using a message model, and device timing data is stored using a timing model.
The intelligent processor is respectively connected with the grabbing module, the matching module and the feedback module, the information received from the data storage module is conveyed to the grabbing module and the matching module through the intelligent processor to be screened and matched, the feedback module conveys the screened and matched information back to the intelligent processor, the accuracy of the received data is improved, and useless information is screened and removed.
The intelligent processor introduces a machine learning deep learning algorithm, a fire alarm target detection model is established according to the data measured by the detection and identification module, the deep learning algorithm is high in universality, training is mainly carried out based on real-time data and historical data of the detection and identification module, the algorithm trains itself according to success or error of an output result, and a good prediction can be given out through a large amount of experience training and optimizing algorithm.
The algorithm model is trained by using a TensorFlow framework, wherein TensorFlow is a deep learning training framework developed by google and based on data flow programming, and can be compatible with various servers; the system YOLO network adopts 9 convolution layers for pre-training, and then adds 1 full connection layer. Each convolution layer is followed by a max-pooling layer and an activation function ReLU.
The deep learning model is trained as follows: the training process is to input the marked flame/smoke pictures into a frame of the YOLO deep learning, iterate for a plurality of times, update common parameters of the model and enhance the characteristic recognition capability. For the training dataset, 4820 fire images were prepared, each 482 images were set as a unit and trained.
The first 9 convolutional networks output a 9 x 1156 feature vector; by convolving conv pre Converting the vector into a vector of 81 dimensions; the possibility of positioning in the grid center is set to z= (z) 1 ，z 2 ，z 3 ，…，z 81 ) The method comprises the steps of carrying out a first treatment on the surface of the After normalizing it by Softmax classifier, we get t= (t 1 ，t 2 ，t 3 ，…，t 81 )；t i The normalized expression of (2) is as in formula (1).
The loss function of the pre-training stage is as in equation (2).
The purpose of the pre-training stage is to use a minimized loss function Q pre The prediction accuracy is improved, and the loss function Q of formal training is shown as a formula (3).
Q＝(w-w') 2 +(h-h') 2 +Q pre (3)
Wherein: w—input image pixel width;
h-input image pixel height;
w' —the actual pixel width of the input image;
h' —the input image true pixel height.
In order to test the performance of the learning model, the detection rate and the precision rate are led out.
The detection rate (D) is defined as the ratio of the real data to the total number of fire samples accurately marked (positive samples representing samples judged to have a fire and negative samples representing samples judged not to have a fire). Reflecting the ability of the learning model to recognize a fire.
The expression is shown as (4).
Wherein: a is that P -positive samples of correct prediction;
B n -negative samples of mispredictions;
the accuracy rate reflects the reliability of fire alarm, and the expression is shown as (5).
Wherein: a is that P -positive samples of correct prediction;
B p negative samples of correct predictions;
the power supply module comprises a charging module, a battery module and a regulated power supply module, wherein the charging module provides electric energy for the battery module, and the battery module transmits the electric energy to the regulated power supply module.
The remote monitoring terminal comprises a real-time display module and a remote control module,
the real-time display module comprises a real-time picture shot by the inspection robot and a three-dimensional digital map of the mine.
The inspection robot shoots real-time pictures, including infrared shooting and visible light shooting, can see scene pictures and scene temperature simultaneously, and is clearer, more visual and more convenient, the specific position of the underground fire source point is determined through the inspection robot positioning label, and technical support is provided for later fire rescue.
The three-dimensional digital map is characterized in that a worker enters a mine to count each main road and branch inside the mine, calculates the length, the maximum width and the minimum width and the maximum height of each main road and branch, visually displays the space level and the position of an underground pipeline by combining a Geographic Information System (GIS) technology, a database technology and a three-dimensional technology through CGPMS three-dimensional drawing software, and draws according to scaling, so as to generate a three-dimensional digital map of the internal route of the mine, the equipment of the detection and identification module is embedded into the three-dimensional digital map, the distribution condition of the equipment can be clearly and intuitively known, the detection and identification module reflects data to a three-dimensional digital map interface through a central intelligent processing terminal, if the data is normal, a green light is lighted, if the data is abnormal, an alarm device is triggered, a red light is lighted, and a remote regulation and control module is started.
The remote control module sends a command to the central intelligent processing terminal through the wireless communication module according to the abnormal condition of the real-time display module, the central intelligent processing terminal controls the inspection robot to go to an abnormal point for secondary monitoring and confirmation, if no obvious fire feature exists, the abnormal condition is relieved, and the monitored data can be transmitted to the central intelligent processing terminal to optimize a fire alarm target detection model based on a deep learning algorithm. If the fire disaster is determined to happen, an escape route and a fire extinguishing scheme are formulated at the first time, a buzzer alarm and a warning lamp of the inspection robot are started, and workers in the area nearby the fire disaster are informed to leave as soon as possible according to the escape route through the voice interphone.
The intelligent mobile monitoring terminal realizes cross-system linkage through a soft bus near-field function, is close to the equipment identification card on the detection and identification module equipment, acquires and connects the IP address of the detection and identification equipment, realizes that the screen-free operation is changed into the screen-free operation, and the fixed key operation is changed into the mobile operation, so that the parameter checking, modification and monitoring of the equipment can be realized rapidly and efficiently, and the operation efficiency is greatly improved.
As an embodiment of the invention, the intelligent inspection alarm method for mine external fire based on the Internet of things comprises the following steps:
acquiring downhole data in real time;
generating measurement data from the downhole data; the measurement data comprises temperature, resistance, image signals and CAN signals;
identifying the measurement data to obtain data quantity and characteristic labels; calculating the number of common statistics and missing values, and carrying out normalization processing on the measured data to obtain metadata; the normalization processing step comprises cleaning, conversion and filling;
storing metadata, equipment message data and equipment time sequence data, establishing a detection model based on a preset deep learning algorithm, and training the detection model based on the underground data;
and acquiring a fire alarm signal containing a fire place output by a detection model in real time by combining a three-dimensional digital map, acquiring an infrared picture and a visible light picture of the fire place based on a preset inspection robot, and generating emergency measures.
In an example of the technical scheme of the invention, the hardware architecture comprises a detection and identification module, a data collection module, an information pre-analysis processing module, a wireless communication module, a central intelligent processing terminal, a remote monitoring terminal and an intelligent mobile monitoring terminal. The detection and identification module comprises a temperature sensor, a smoke sensor, a track inspection robot and a personnel ID identification unit; the data collection module adopts a multi-channel data acquisition instrument; the wireless communication module comprises a 5G base station, wi-Fi6 technology and a wireless radio frequency transceiver; the central intelligent processing terminal comprises a data storage module, an intelligent processor, a grabbing module, a matching module, a feedback module and a power supply module. The remote monitoring terminal comprises a real-time display module and a remote regulation and control module, wherein the real-time display module comprises a real-time picture shot by the inspection robot and a three-dimensional digital map of a mine. An intelligent inspection alarm method for mine external fire based on the Internet of things comprises the following specific implementation processes:
step 1, current underground data are collected through a detection and identification module, a sensor respectively reads smoke concentration data and temperature data, a track inspection robot reads image data, and a personnel ID identification unit reads personnel identity information and position information:
and 2, transmitting the original data to a data collection module by the detection and identification module through Wi-Fi6, and measuring temperature, resistance, image signals, CAN signals and the like by the data collection module to finish two steps of connecting an upper server and collecting and setting.
And 3, the information pre-analysis processing module receives the data information transmitted by the data collection module through Wi-Fi6, analyzes the size of the data and all the characteristic labels, calculates the number of common statistics and missing values, and performs normalization processing of cleaning, switching and filling on the data.
And step 4, the information pre-analysis processing module sends the pre-processed data signals to the central intelligent processing terminal through the 5G base station, and stores metadata, equipment message data and equipment time sequence data. And a deep learning algorithm of machine learning is introduced, a fire alarm target detection model is established, and training and optimization are carried out on the model based on real-time data and historical data of the detection and identification module.
And 5, transmitting the processed data to a remote monitoring terminal by the central intelligent processing terminal through a 5G base station, checking an abnormal alarm place of the underground monitoring equipment by the remote monitoring terminal through a three-dimensional digital map, checking a real-time infrared picture and a visible light picture of the abnormal alarm place by using the inspection robot, if a fire disaster occurs, making an escape route and a fire extinguishing scheme at the first time, starting a buzzer alarm and a warning lamp of the inspection robot, and informing workers in an area near the fire disaster to leave as soon as possible according to the escape route by using the voice interphone.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the invention and are described in detail herein without thereby limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (16)
1. Mine is because of fire intelligence inspection alarm system outward based on thing networking, its characterized in that includes:
the detection and identification module is used for acquiring current underground data;
the data collection module is used for generating measurement data according to the underground data; the measurement data comprises temperature, resistance, image signals and CAN signals;
the information pre-analysis processing module is used for identifying the measurement data to obtain data quantity and characteristic labels; calculating the number of common statistics and missing values, and carrying out normalization processing on the measured data to obtain metadata; the normalization processing step comprises data cleaning, data conversion and data filling;
the central intelligent processing terminal is used for storing metadata, equipment message data and equipment time sequence data, establishing a detection model based on a preset deep learning algorithm, and training the detection model based on the underground data;
the remote monitoring terminal is used for acquiring a fire alarm signal containing a fire place and output by the detection model in real time by combining a three-dimensional digital map, acquiring an infrared picture and a visible light picture of the fire place based on a preset inspection robot, and generating emergency measures, wherein the emergency measures specifically comprise making an escape route and a fire extinguishing scheme, starting a buzzer alarm and a warning lamp of the inspection robot, and informing workers in an area nearby the fire place to leave as soon as possible according to the escape route through the voice interphone.
2. The intelligent inspection alarm system for mine external fire based on the internet of things according to claim 1, wherein the detection and identification module comprises:
the temperature sensor is used for reading temperature data;
the smoke sensor is used for reading smoke concentration data;
the track inspection robot is used for reading the image data;
and the personnel ID recognition unit is used for reading the personnel identity information and the position information.
3. The mine external fire intelligent patrol alarm system based on the Internet of things according to claim 1, wherein the data collection module adopts a multichannel data acquisition instrument.
4. The mine external fire intelligent patrol alarm system based on the internet of things according to claim 1, further comprising a wireless communication module, wherein the wireless communication module comprises a 5G base station, wi-Fi6 technology and a wireless radio frequency transceiver.
5. The mine external fire intelligent patrol alarm system based on the Internet of things according to claim 1 is characterized in that the central intelligent processing terminal comprises a data storage module, an intelligent processor, a grabbing module, a matching module, a feedback module and a power supply module; the data storage module provides a unified data storage platform for equipment metadata, equipment message data and equipment time sequence data in the scene of the Internet of things; the intelligent processor is respectively connected with the grabbing module, the matching module and the feedback module, the information received from the data storage module is conveyed to the grabbing module and the matching module through the intelligent processor to be screened and matched, the feedback module conveys the screened and matched information back to the intelligent processor, the accuracy of the received data is improved, and useless information is screened and removed.
6. The mine external fire intelligent patrol alarm system based on the internet of things according to claim 1, wherein the three-dimensional digital map displays the spatial hierarchy and the position of underground equipment by combining a geographic information system technology, a database technology and a three-dimensional technology.
7. The intelligent inspection alarm system for mine external fire disaster based on the Internet of things, which is characterized in that the temperature sensor is a digital temperature sensor of a model TMP126, and the smoke sensor is an MQ-2 type smoke sensor.
8. The mine external fire disaster intelligent inspection alarm system based on the Internet of things, which is disclosed by claim 1, is characterized in that the inspection robot uses an ARM+LINUX main control system, the positioning at the centimeter level is realized through the combination of 3DSLAM+IMU+GPS, the robot can realize automatic obstacle avoidance through the collocation of a laser radar and a front camera and a rear camera, a trolley line is used for charging, the cruising distance is 10km, the emergency parking performance is controlled below 0.2m, and the climbing capacity is less than or equal to 45 degrees.
9. The mine external fire disaster intelligent inspection alarm system based on the Internet of things, which is disclosed by claim 1, is characterized in that the inspection robot comprises a double-spectrum high-definition camera, the double-spectrum high-definition camera adopts an HB-8501R-F9432-T9310 type intrinsic safety thermal imaging double-spectrum integrated camera, the temperature measurement of points, lines and frames is supported, the highest temperature is highlighted through a cross cursor, and the image details are enhanced.
10. The intelligent mine external fire inspection alarm system based on the Internet of things, which is characterized in that the escape route established by the inspection robot is based on a D_Star algorithm, and the optimal path calculated by the D_Star algorithm reduces the interference degree.
11. The intelligent inspection alarm system for mine external fire disaster based on the Internet of things according to claim 4, wherein each device in the detection and identification module is provided with a positioning tag, a note sends UWB signals, the device is communicated with a 5G base station, the 5G base station sends network raw data to a positioning engine in real time, the positioning engine runs a positioning algorithm, the coordinate position with the positioning tag is calculated in real time, and finally the coordinate position is displayed on a real-time display module.
12. The mine external fire disaster intelligent patrol alarm system based on the Internet of things according to claim 4 is characterized in that the wireless communication module adopts a 5G base station+WIFI 6 cooperative mode to realize wireless full coverage; the 5G base station uses a BBU-AAU architecture and comprises a 5G baseband unit and a 5G radio frequency unit, the wireless communication module further comprises a wireless radio frequency transceiver, an ADRV9008 type is adopted as a receiver, and an ADRV9009 type is adopted as a transmitter.
13. The mine external fire disaster intelligent patrol alarm system based on the Internet of things is characterized by further comprising an intelligent mobile monitoring terminal for realizing cross-system linkage, wherein the intelligent mobile monitoring terminal is close to an equipment identification card on the detection and identification module equipment, acquires and connects an IP address of the detection and identification equipment, realizes screen-free and screen-free operation, realizes fixed key operation and mobile operation, and realizes parameter checking, modification and monitoring of the equipment rapidly and efficiently.
14. The intelligent inspection alarm system for mine external fire disaster based on the Internet of things according to claim 1, wherein the deep learning algorithm adopts YOLO, is an algorithm model capable of rapidly carrying out target detection, and is mainly used for creating an S x S unit cell for an identified image, wherein each unit cell is used for evaluating objects in the unit cell, predicting confidence scores of a boundary box and a boundary box of each unit cell, extracting features by a convolution layer, simplifying and calculating complex images by a pooling layer shrinking feature map, simultaneously carrying out feature compression to extract main features, and connecting all features by a full-connection layer and outputting the features to a classifier; the algorithm has the advantages of simple and convenient framework, high recognition accuracy and high recognition speed.
15. The intelligent inspection alarm system for mine external fire disaster based on the Internet of things according to claim 14, wherein the algorithm model is trained by using a TensorFlow framework, wherein TensorFlow is a deep learning training framework based on data flow programming, and is compatible with various servers; the system is characterized in that a YOLO network adopts 9 convolution layers for pre-training, 1 full connection layer is added, and a maximum pooling layer and an activation function ReLU are added after each convolution layer;
the deep learning model is trained as follows: the training process is to input the marked flame/smoke pictures into a frame of the YOLO deep learning, iterate for a plurality of times, update common parameters of the model, enhance the characteristic recognition capability, prepare 4820 fire images for a training data set, set each 482 images as a unit and train;
the first 9 convolutional networks output a 9 x 1156 feature vector; by convolving conv pre Converting the vector into a vector of 81 dimensions; the possibility of positioning in the grid center is set to z= (z) 1 ，z 2 ，z 3 ，…，z 81 ) The method comprises the steps of carrying out a first treatment on the surface of the After normalizing it by Softmax classifier, we get t= (t 1 ，t 2 ，t 3 ，…，t 81 )；t i The normalized expression of (2) is as in formula (1),
the loss function of the pre-training stage is shown as (2)
The purpose of the pre-training stage is to use a minimized loss function Q pre Improving the accuracy of prediction, and formally training the loss function Q as formula (3)
Q＝(w-w') 2 +(h-h') 2 +Q pre (3)；
Wherein: w represents the input image pixel width; h represents the input image pixel height; w' represents the real pixel width of the input image; h' represents the true pixel height of the input image;
leading out the detection rate and the precision rate for testing the performance of the learning model;
the detection rate (D) is defined as the ratio of real data to the total number of fire samples accurately marked, and reflects the ability of the learning model to identify fire, and is expressed as (4)
Wherein: a is that P Positive samples representing correct predictions; b (B) n Negative samples representing mispredictions;
the accuracy rate reflects the reliability of fire alarm, and the expression is shown as (5);
wherein: a is that P Positive samples representing correct predictions; b (B) p Representing a negative sample of the correct prediction.
16. The intelligent inspection alarm method for mine external fire disaster based on the Internet of things is characterized by comprising the following steps of:
acquiring downhole data in real time;
generating measurement data from the downhole data; the measurement data comprises a temperature signal, a resistance signal, an image signal and a CAN signal;
identifying the measurement data to obtain data quantity and characteristic labels; calculating the number of common statistics and missing values, and carrying out normalization processing on the measured data to obtain metadata; the normalization processing step comprises cleaning, conversion and filling;
storing metadata, equipment message data and equipment time sequence data, establishing a detection model based on a preset deep learning algorithm, and training the detection model based on the underground data;
and acquiring a fire alarm signal containing a fire place output by a detection model in real time by combining a three-dimensional digital map, acquiring an infrared picture and a visible light picture of the fire place based on a preset inspection robot, and generating emergency measures.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211729794 | 2022-12-30 | ||
| CN2022117297946 | 2022-12-30 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116030591A true CN116030591A (en) | 2023-04-28 |
Family
ID=86070334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310031647.XA Pending CN116030591A (en) | 2022-12-30 | 2023-01-10 | Intelligent inspection alarm system and method for mine external fire disaster based on Internet of things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116030591A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117132945A (en) * | 2023-10-25 | 2023-11-28 | 四川三思德科技有限公司 | Multi-source data fusion-based key area security method and device |
| CN117809418A (en) * | 2023-08-07 | 2024-04-02 | 安越网络科技(南通)有限公司 | Intelligent dangerous source identification and early warning system based on Internet of things technology |
-
2023
- 2023-01-10 CN CN202310031647.XA patent/CN116030591A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117809418A (en) * | 2023-08-07 | 2024-04-02 | 安越网络科技(南通)有限公司 | Intelligent dangerous source identification and early warning system based on Internet of things technology |
| CN117132945A (en) * | 2023-10-25 | 2023-11-28 | 四川三思德科技有限公司 | Multi-source data fusion-based key area security method and device |
| CN117132945B (en) * | 2023-10-25 | 2024-01-09 | 四川三思德科技有限公司 | Multi-source data fusion-based key area security method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116030591A (en) | Intelligent inspection alarm system and method for mine external fire disaster based on Internet of things | |
| CN109961157B (en) | Inspection method and system of solar photovoltaic power generation system | |
| CN100533506C (en) | Vehicle monitoring apparatus and realizing method thereof | |
| CN103235562A (en) | Patrol-robot-based comprehensive parameter detection system and method for substations | |
| CN110881061B (en) | Arm type terminal comprehensive sensing and intelligent interaction emergency disposal method based on ubiquitous power Internet of things | |
| CN104581087A (en) | Intelligent remote monitoring system of power transmission line on basis of internet of things | |
| CN207268846U (en) | Electric inspection process robot | |
| CN205301613U (en) | But automatic meteorological station of remote access | |
| CN101635078A (en) | Intelligent fire scene detecting device | |
| CN106647813A (en) | Intelligent vehicle-mounted dual-light inspection system expert diagnosing and abnormity processing method | |
| CN110647170A (en) | Navigation mark inspection device and method based on unmanned aerial vehicle | |
| CN114926983A (en) | Traffic accident emergency oriented multi-scale comprehensive sensing method | |
| CN112067137A (en) | Automatic power line temperature measurement method based on unmanned aerial vehicle line patrol | |
| CN113055644A (en) | Novel intelligent voice calling method for unmanned aerial vehicle | |
| CN110933593A (en) | Safety prevention and control system based on UWB technology | |
| CN210835732U (en) | Beacon inspection device based on unmanned aerial vehicle | |
| CN115903973A (en) | Fire detection and emergency linkage control system based on Internet of things | |
| CN106772217B (en) | A kind of pipe network inspection maintaining-managing system | |
| CN113823071A (en) | Emergency comprehensive management application platform | |
| CN105991974A (en) | High altitude monitoring method and system based on 4G mobile communication network | |
| CN116486553A (en) | Sky-ground integrated forest fire monitoring and early warning system | |
| CN110298548A (en) | The long-range real-time diagnosis of Rock Tunnel life cycle management engineering information and feedback system | |
| CN217084853U (en) | Intelligent monitoring device applied to totally-enclosed coal yard | |
| CN111915831A (en) | Intelligent forest grassland fire prevention and rescue method | |
| CN115393971A (en) | Electric power inspection data synchronization system based on Beidou space-time data sensing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |