Early warning and forecasting method for typhoon rainstorm type geological disasters

Abstract

The invention discloses a typhoon rainstorm type geological disaster early warning and forecasting method, and belongs to the technical field of geological monitoring, and the method comprises the steps: building a hidden danger point full-attribute three-dimensional real scene situation map based on a point cloud through an unmanned plane oblique photography technology in combination with geological disaster prevention and control data; arranging panoramic intelligent probes at the hidden danger points to form an electronic fence for early warning of accidental entry of people; constructing an early warning model, dynamically analyzing a hidden danger point multi-angle time sequence picture captured by a panoramic intelligent probe through a neural network algorithm, carrying out intelligent monitoring interpretation and real-time early warning on risk factors of the hidden danger point, comprehensively reflecting the risk factors into a full-attribute three-dimensional real scene situation map, and constructing a geological disaster eagle eye system; issuing early-warning information, feeding back the early-warning information to the early-warning model according to the disaster and dangerous situation, and updating the early-warning model. According to the typhoon rainstorm type geological disaster early warning and forecasting method, the manpower and material resource investment of management and control is reduced to the maximum extent, and the accuracy of pre-judgment and early warning is improved to the maximum degree.

Description

Technical Field

[0001] The present invention relates to the technical field of geological monitoring, and in particular to a method for early warning and forecasting of typhoon and rainstorm type geological disasters. Background Art

[0002] At present, with the deployment of monitoring equipment in geological disaster risk prevention areas, the scientific and intelligent level of "technical defense + human defense" management and control of geological disaster hazards has been greatly improved. It has also continuously accumulated first-hand monitoring data for in-depth analysis and research on the induction and early warning mechanisms of geological disaster hazards.

[0003] When proactively investigating the operation and maintenance of geological disaster monitoring equipment in a certain city, we gained an in-depth understanding of the actual application of the Geological Disaster Intelligent Prevention APP at the grassroots level, and widely collected opinions and suggestions on the functions of current monitoring equipment from grassroots resource planning departments and group monitoring and prevention personnel, mainly including the following points: (1) The comprehensive situation in the hazard area is only available in the form of "paper data". The satellite remote sensing images are not very timely and lack a comprehensive prevention and control full-attribute three-dimensional real-life situation map. As a result, group monitoring and prevention personnel can only go to the point and compare the remote sensing images on the spot to gradually understand the dynamic situation in the grid; (2) During the main flood season, when warning information is issued, it is impossible to determine whether anyone has accidentally entered the hidden danger "red line" area, which invisibly increases the difficulty of control; (3) The study of the triggering mechanism of geological disasters requires long-term accumulation of various factors and comprehensive comparative analysis and judgment. Current monitoring methods lack real-time dynamic image data before and during the occurrence of hidden dangers. Summary of the invention

[0004] The purpose of the present invention is to provide a typhoon and rainstorm type geological disaster early warning and forecasting method to solve the problems existing in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides a typhoon and rainstorm type geological disaster early warning and forecasting method, comprising the following steps: Combined with geological disaster prevention and control data, a three-dimensional real-life situation map of hidden danger points with full attributes based on point cloud is established through drone oblique photography technology, which facilitates management and control decision makers, technical personnel and group monitoring and prevention personnel to intuitively and comprehensively grasp the actual situation of hidden danger points anytime and anywhere; Panoramic intelligent probes are deployed at potential danger points to form an electronic fence for early warning of accidental entry of personnel; Build an early warning model, dynamically analyze the multi-angle time-series images of hidden danger points captured by panoramic intelligent probes through neural network algorithms, carry out intelligent monitoring and interpretation of dangerous factors such as dangerous rock rolling, soil layer movement, water level surge, and personnel entry at hidden danger points, and issue real-time early warnings, which are comprehensively reflected in the three-dimensional real-life situation map of all attributes of hidden danger points, thereby building a geological disaster eagle eye system; The warning information is released and reflected on the large screen of smart management and control and the small screen of mobile management and control terminals. It is also fed back to the warning model based on the disaster risk situation and the warning model is continuously updated.

[0006] Preferably, the deployment of panoramic intelligent probes includes the triangulated deployment method on mountains without vegetation coverage to achieve omnidirectional coverage and the time-sequential deployment method on impact ditch paths to gain the last "minute" for disaster avoidance; the panoramic intelligent probes use wide-angle cameras or linked cameras, and their number is precisely planned by experts from the Geological Bureau, which can effectively improve equipment utilization.

[0007] Preferably, a neural network algorithm is used to dynamically analyze the multi-angle time-series images of the hidden danger points captured by the panoramic intelligent probe, and intelligent monitoring and interpretation of the dangerous factors of the hidden danger points are carried out to provide real-time warnings, specifically: The manually marked geological disasters and normal images are input into the early warning model as classification labels; The images collected in real time by the panoramic intelligent probe are input into the early warning model as input images and compared with classification labels (such as landslide, mud-rock flow, landslide, etc.) for classification; Determine whether there are any errors in the comparison classification. If there are errors, modify the early warning model parameters; if there are no errors, release the early warning information interpreted by the early warning model.

[0008] Preferably, comparative classification includes single image classification and multi-image classification; single image classification is to find a classification label that matches the input image from a fixed set of classification labels and assign it to the input image; multi-image classification is to find several classification labels that match the input image from a fixed set of classification labels and assign them to the input image.

[0009] Preferably, the geological disaster Eagle Eye system includes an observation mode and an early warning mode; during the non-flood season, the geological disaster Eagle Eye system enters the observation mode; during the flood season, the geological disaster Eagle Eye system enters the early warning mode.

[0010] Preferably, the observation mode includes monitoring and recording potential danger points and monitoring unstable factors of potential danger points through artificial intelligence, and timely reflecting the early warning information on the smart management and control large screen; the early warning mode includes persuading people who accidentally enter the electronic fence to leave, and reflecting the information on the smart management and control large screen. At the same time, when a disaster occurs, the disaster risk is perceived in real time, warnings are issued and reflected on the smart management and control large screen.

[0011] Preferably, the display of warning information includes launching warning information and broadcasting warnings; the language of the broadcast warning includes a Mandarin version and a dialect version.

[0012] Preferably, the full-attribute three-dimensional real-scene situation map is a map that includes high-altitude remote sensing satellite images of hidden danger points, three-dimensional real-scene situation of hidden danger points’ centrally controlled drone oblique camera point clouds, and digital control situation of long-term monitoring video images. Each hidden danger point is coded, and the overall situation and control information of the hidden danger points can be obtained by scanning the codes on site.

[0013] Therefore, the present invention adopts the above-mentioned typhoon and rainstorm type geological disaster early warning and forecasting method, which has the following beneficial effects: (1) Automatic early warning: After automatically identifying geological disasters, early warning information is automatically issued without human intervention; (2) Backend algorithm, low hardware cost, and more convenient subsequent upgrades; (3) Supports reuse of old cameras, can use most cameras on the market, and supports 4G / 5G networks plus solar energy equipment; (4) Voice dissuasion and broadcast warnings support loop playback in Mandarin and multiple languages ​​(local dialects).

[0014] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a typhoon and rainstorm type geological disaster early warning and forecasting method of the present invention; Figure 2 The figure is a schematic diagram of the process of interpreting the monitored image data by the early warning model of the present invention. DETAILED DESCRIPTION

[0016] Example In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0017] like Figure 1 As shown, a typhoon and rainstorm type geological disaster early warning and forecasting method is provided, which obtains the weather forecast of the area, obtains the existing rain gauge data through the interface, and then compares the rainfall threshold of the area to switch the early warning state to the two states of opening the early warning and closing the early warning, including the following steps: Combined with geological disaster prevention and control data, a three-dimensional real-life situation map of hidden danger points with full attributes based on point cloud is established through drone oblique photography technology, which facilitates management and control decision makers, technical personnel and group monitoring and prevention personnel to intuitively and comprehensively grasp the actual situation of hidden danger points anytime and anywhere; Panoramic intelligent probes are deployed at potential danger points to form electronic fences for early warning of accidental entry of personnel; the deployment of panoramic intelligent probes includes the triangular deployment method on mountains without vegetation coverage to achieve omnidirectional coverage and the time-series deployment method on impact ditch paths to gain the last "minute" for disaster avoidance; panoramic intelligent probes use wide-angle cameras or linked cameras, and their number is precisely planned by experts from the Geological Bureau to effectively improve equipment utilization.

[0018] Supports most platform cameras, supports direct acquisition via rtsp, supports h264, h265 encoding, has low traffic, high definition and fast speed. It is recommended to use a wide-angle camera and support night fill light. Taking the CMOS camera as an example, its parameters are shown in Table 1: Table 1 CMOS parameter details ;

[0019] The linked camera can continuously rotate 360 ​​degrees horizontally and -20 to 90 degrees vertically, and can automatically flip 180 degrees for continuous monitoring without blind spots.

[0020] Supports built-in GPU chip and deep learning algorithm to effectively improve detection accuracy; Support face detection; support face optimization capture; support face enhancement; support face attribute extraction; support 6 attributes and 8 expressions; support multiple face cutout scheme settings; Support dual-channel video output: panoramic 180° wide-angle 4 million W stitching, F1.O, 1 / 2.8-inch CMOS sensor; detail 2 million W, 23x optical zoom, 16x digital zoom, 1 / 2.8-inch CMOS sensor; Support starlight-level ultra-low illumination, color: 0.001Lux@F1.0 black and white: 0.0001Lux@F1.0; Support H.265 encoding to achieve ultra-low bit rate transmission; Built-in 150-meter infrared light fill light, using the magnification and infrared light power matching algorithm, the fill light effect is more uniform; The detail road can rotate 360° in horizontal direction and can be automatically flipped 180° in vertical direction from -20° to 90° for continuous monitoring without blind spots. The detailed route supports 300 preset positions, 8 cruise paths, and 5 patrol paths; The detail road supports electronic fog penetration, which can output clear and thorough images even in foggy days; Support 1-channel audio input and 1-channel audio output; Built-in 2-way alarm input and 1-way alarm output, supporting alarm linkage function; Support IP66 protection level, 8000V lightning protection, surge protection and surge protection; Support DC36V±25% wide voltage input; Support national encryption algorithms SM1, SM2, SM3, SM4, and GB35114A level; It comes with a speaker module, which supports automatic reminders after perimeter detection and broadcasts relevant warning information at the same time.

[0021] Build an early warning model, dynamically analyze the multi-angle time series images of hidden danger points captured by the panoramic intelligent probe through the neural network algorithm, carry out intelligent monitoring and interpretation of dangerous factors such as rock rolling, soil movement, water level surge, and personnel entry at the hidden danger points, and provide real-time early warning, such as Figure 2 As shown, specifically: The manually marked geological disasters and normal images are input into the early warning model as classification labels; The images collected in real time by the panoramic intelligent probe are input into the early warning model as input images and compared with classification labels (such as landslide, mud-rock flow, landslide, etc.) for classification; Determine whether there are any errors in the comparison classification. If there are errors, modify the early warning model parameters; if there are no errors, release the early warning information interpreted by the early warning model.

[0022] Comparative classification includes single image classification and multi-image classification; single image classification is to find a classification label that matches the input image from a fixed set of classification labels and assign it to the input image; multi-image classification is to find several classification labels that match the input image from a fixed set of classification labels and assign them to the input image. And it is comprehensively reflected in the three-dimensional real-life situation map of all attributes of the hidden danger points, so as to construct a geological disaster Eagle Eye system; the geological disaster Eagle Eye system includes observation mode and early warning mode; during the non-flood season, the geological disaster Eagle Eye system enters the observation mode, one is to conduct long-term monitoring and recording of hidden danger points (areas), to provide first-hand research materials for the mechanism of geological disaster formation, and the other is to monitor the unstable factors such as abnormal human excavation at hidden danger points through artificial intelligence, and reflect the early warning information on the smart management and control screen in time; during the flood season, according to the hidden danger warning level issued by the relevant government departments, at the yellow, orange and red levels, the system can automatically switch to the early warning mode, and scan the situation around the hidden danger point (area) through the formed electronic fence. Once it is found that someone accidentally enters the early warning range of the hidden danger point, a voice (Mandarin + dialect) warning and persuasion can be issued, and the early warning information will be reflected on the smart management and control screen in time.

[0023] When a landslide occurs at a potential risk point, the panoramic intelligent probe can sense the displacement of the landslide body in real time, and cross-check with the data of the monitoring equipment currently deployed at the potential risk point, issue a voice (Mandarin + dialect) warning on the spot, and promptly reflect it on the smart control screen to provide a reference for the final decision. At the same time, when the emergency response is initiated and personnel in the potential risk point are required to evacuate, the panoramic intelligent probe can monitor the evacuation of personnel in real time in combination with the location heat map.

[0024] At the same time, a "one-picture" digital control situation will be established, including high-altitude remote sensing satellite images of hidden danger areas, three-dimensional real-life models of hidden danger points taken by mid-air drones, and long-term monitoring video images. Each hidden danger point will be coded so that control technicians and the public can obtain the overall situation and related control information of the hidden danger point (area) by scanning the code on site. Neural network algorithm training includes: 1. Select the framework. 2. Build a data set and read data. Image classification is a data-driven task, and a high-quality data set is the guarantee of accuracy. Therefore, in the early stage of clarifying the task requirements, it is necessary to build a data set with clear content and in line with reality, and divide it into training set, validation set and test set; 3. Network model selection: screening based on existing research content and results; 4. Training and parameter debugging, selecting appropriate learning rate, optimization method, and loss function for training, and using different data enhancement methods to further improve the model's sensitivity to data; 5. Testing: After the accuracy requirements are met on the test set, the model is further deployed.

[0025] Release warning information, and the display of warning information includes launching warning information and broadcasting warning; sending warning information gives priority to sending message templates bound to managers. Contains the content and effective range of abnormal pictures. Send SMS reminders at the same time. The sending time range and sending frequency can be set according to actual conditions. Support docking with the geological disaster intelligent prevention APP interface to push messages to specific administrators; broadcast warnings can record multiple templates, such as automatic persuasion broadcasts after abnormalities are found. After finding that automatic persuasion is ineffective. Broadcasts manually triggered by personnel. The broadcast content can be recorded in Mandarin and local dialects; administrators bound to the mobile terminal can directly receive messages about abnormal situations on the mobile phone. The message content includes time, camera location, pictures, effective range and other data. After clicking the task, you can directly see the screen of the current camera. If you have been persuaded to leave, click the task to complete directly. If you do not click, the system will automatically complete it after persuading you to leave. If you have not been persuaded to leave, the manager can directly click to play the manual persuasion broadcast. The warning information is reflected on the large screen of smart management and control and the small screen of the mobile management and control terminal at the same time, and is fed back to the warning model according to the dangerous situation, and the warning model is continuously updated.

[0026] It also includes automatic equipment detection: automatic detection robot function. Timed self-checking equipment: panoramic intelligent probe and sound column. Automatic notification to management personnel and manufacturers when abnormalities are found.

[0027] Therefore, the present invention adopts the above-mentioned typhoon and rainstorm type geological disaster early warning and forecasting method to minimize the human and material resources input in management and control, and improve the accuracy of prediction and warning with the highest precision. The development and deployment cycle of the geological disaster Eagle Eye system is short, the effect is quick, and it is economically feasible.

[0028] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A typhoon and rainstorm type geological disaster early warning and forecasting method, characterized in that: The following steps are involved: Combined with geological disaster prevention and control data, a three-dimensional real-life situation map of hidden danger points with full attributes based on point clouds is established through drone tilt photography technology; Panoramic intelligent probes are deployed at potential danger points to form an electronic fence for early warning of accidental entry of personnel; Build an early warning model, dynamically analyze the multi-angle time series images of hidden danger points captured by panoramic intelligent probes through neural network algorithms, carry out intelligent monitoring and interpretation of dangerous factors of hidden danger points, give real-time early warnings, and comprehensively reflect them in the three-dimensional real-life situation map of all attributes of hidden danger points, thereby building a geological disaster eagle eye system; The warning information is released and reflected on the large screen of smart management and control and the small screen of mobile management and control terminals. It is also fed back to the warning model based on the disaster risk situation and the warning model is continuously updated.

2. A typhoon and rainstorm type geological disaster early warning and forecasting method according to claim 1, characterized in that: The deployment of panoramic intelligent probes includes the triangulation method for mountains without vegetation coverage and the time-series deployment method for impact ditch paths; the panoramic intelligent probes use wide-angle cameras or linkage cameras.

3. The method for early warning and forecasting of typhoon and rainstorm type geological disasters according to claim 1, characterized in that: Through the neural network algorithm, the multi-angle time series images of the hidden danger points captured by the panoramic intelligent probe are dynamically analyzed, and the dangerous factors of the hidden danger points are intelligently monitored and interpreted, and real-time warnings are issued. Specifically: The manually marked geological disasters and normal images are input into the early warning model as classification labels; The images collected in real time by the panoramic intelligent probe are input into the early warning model as input images and compared with the classification labels for classification; Determine whether there is an error in the comparison classification. If there is an error, modify the early warning model parameters; If there is no error, the warning information interpreted by the warning model will be released.

4. A typhoon and rainstorm type geological disaster early warning and forecasting method according to claim 3, characterized in that: Comparative classification includes single image classification and multi-image classification; single image classification is to find a classification label that matches the input image from a fixed set of classification labels and assign it to the input image; multi-image classification is to find several classification labels that match the input image from a fixed set of classification labels and assign them to the input image.

5. The method for early warning and forecasting of typhoon and rainstorm type geological disasters according to claim 1, characterized in that: The geological disaster Eagle Eye system includes observation mode and early warning mode; during the non-flood season, the geological disaster Eagle Eye system enters the observation mode; during the flood season, the geological disaster Eagle Eye system enters the early warning mode.

6. A typhoon and rainstorm type geological disaster early warning and forecasting method according to claim 5, characterized in that: The observation mode includes monitoring and recording potential danger points and monitoring unstable factors at potential danger points through artificial intelligence, and timely reflecting the early warning information on the smart management and control screen; the early warning mode includes persuading people who accidentally enter the electronic fence to leave, and reflecting the information on the smart management and control screen. At the same time, when a disaster occurs, the disaster situation is perceived in real time, warnings are issued and reflected on the smart management and control screen.

7. The method for early warning and forecasting of typhoon and rainstorm type geological disasters according to claim 1, characterized in that: The display of warning information includes launching warning information and broadcasting warnings; the languages ​​of broadcasting warnings include Mandarin version and dialect version.

8. The method for early warning and forecasting of typhoon and rainstorm type geological disasters according to claim 1, characterized in that: The full-attribute three-dimensional real-scene situation map is a map of the digital control situation that includes high-altitude remote sensing satellite images of hidden danger points, three-dimensional real-scene situation of hidden danger point cloud taken by central-controlled drones, and long-term monitoring video images. Each hidden danger point is coded, and the overall situation and control information of the hidden danger point can be obtained by scanning the code on site.

Images