CN-122024423-A - Air-ground integrated monitoring and early warning method and system for natural protection area

CN122024423ACN 122024423 ACN122024423 ACN 122024423ACN-122024423-A

Abstract

The application relates to the field of ecological monitoring, in particular to an air-to-ground integrated monitoring and early warning method, system and medium for a natural protection area, wherein the method comprises the steps of collecting multisource heterogeneous monitoring data of the natural protection area through integrated deployment of satellite remote sensing equipment, unmanned aerial vehicle aerial photographing equipment and various ground sensors; the method comprises the steps of carrying out structural enhancement on multi-source heterogeneous monitoring data to construct a unified ecological database, inputting the data subjected to structural enhancement into a trained deep learning model for reasoning to identify monitoring targets or abnormal events of preset categories, wherein the deep learning model is based on a convolutional neural network architecture, and generating early warning information and pushing the early warning information to a management terminal according to an output result of the deep learning model. According to the application, through systematic integration of deep learning technology, the scattered data of the natural protection area are converted into linkage information, and the artificial low-efficiency processing is updated into intelligent active decision, so that the application value of the space-earth integrated data in the management of the protection area can be fully released.

Inventors

WANG SHUI
ZHANG BING
WAN YUMEI
LU YUJIE
LU GUANGSHI

Assignees

安徽江北美图信息科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260211

Claims (10)

1. An air-ground integrated monitoring and early warning method for a natural protection area is characterized by comprising the following steps of: the method comprises the steps of collecting multisource heterogeneous monitoring data of a natural protection area through integrated deployment of satellite remote sensing equipment, unmanned aerial vehicle aerial photographing equipment and various ground sensors; Carrying out structural enhancement on the acquired multi-source heterogeneous monitoring data to construct a unified ecological database; Inputting the data subjected to structural reinforcement into a trained deep learning model for reasoning so as to identify a monitoring target or an abnormal event of a preset category, wherein the deep learning model is based on a convolutional neural network architecture; And generating early warning information according to the output result of the deep learning model and pushing the early warning information to a management terminal.
2. The method for integrated space-time monitoring and early warning in a natural protection area according to claim 1, wherein the structural reinforcement comprises: identifying and removing abnormal values through a statistical method, and filling missing values by using an interpolation method or a statistical value; Performing regular field calibration on the ground sensor, and performing cross verification and calibration by using monitoring data of satellite remote sensing equipment, unmanned aerial vehicle aerial photographing equipment and the ground sensor; And converting the data from different sources into a unified coordinate system, data precision and file format.
3. The method for integrated space-time monitoring and early warning in a natural protection area according to claim 2, wherein the structural reinforcement further comprises: And classifying and storing the processed data according to at least one of monitoring elements, data sources, time sequences and confidentiality levels.
4. The method for integrated space-time monitoring and early warning of a natural protection zone according to claim 1, further comprising a model training process before inputting data into the deep learning model for reasoning, the model training process comprising: Constructing a training data set by using the marked natural protection area scene data; Training an initial model based on a convolutional neural network architecture by using the training data set, and adjusting super parameters by using a verification set; And evaluating the performance of the trained model by using the test set to select a final model for reasoning.
5. The method of claim 4, wherein the process of constructing the training data set includes a data enhancement operation that includes at least one of spatial enhancement, color enhancement, and timing enhancement of the original image or video data.
6. The method for integrated space-earth monitoring and early warning in natural protection area according to claim 1, characterized in that, The monitoring targets or abnormal events of the preset categories comprise at least one of wild animal species, forest fires, abnormal human behaviors, vehicles and map spot changes of the protected areas; the output result of the deep learning model comprises the category, the position, the confidence coefficient and the time interval information of the occurrence of the abnormal event of the target.
7. The method for space-to-ground integrated monitoring and early warning of a natural protection zone according to claim 1, further comprising a model iterative optimization process comprising: And collecting misidentification cases or new scene data generated in the reasoning process of the deep learning model, and updating and fine-tuning the deep learning model by utilizing the collected data.
8. An aerospace-ground integrated monitoring and early warning system for a natural protection area, for implementing the aerospace-ground integrated monitoring and early warning method for the natural protection area according to any one of claims 1 to 7, characterized in that the system comprises: The data acquisition module is configured to control the satellite remote sensing equipment, the unmanned aerial vehicle aerial photographing equipment and various ground sensors which are connected to acquire multi-source heterogeneous monitoring data of a natural protection area; the data processing module is configured to perform structural enhancement on the collected multi-source heterogeneous monitoring data so as to construct a unified ecological database; the intelligent analysis module is configured to run a deep learning model trained based on a convolutional neural network architecture so as to infer data output by the data processing module, so that monitoring targets or abnormal events of preset categories are identified; and the early warning decision module is configured to generate early warning information according to the output result of the intelligent analysis module and push the early warning information to the management terminal.
9. The system of claim 8, wherein the ground sensor comprises at least three of an infrared camera, a surveillance camera, a water quality monitoring station, a weather monitoring station, an electronic fence, an insect condition warning light, and a geological disaster monitor.
10. A computer-readable storage medium storing one or more programs, characterized in that, When the one or more programs are executed, the space-to-ground integrated monitoring and early warning method for the natural protection area according to any one of claims 1 to 7 is realized.

Description

Air-ground integrated monitoring and early warning method and system for natural protection area Technical Field The application relates to the technical field of ecological monitoring, in particular to an air-to-ground integrated monitoring and early warning method, an air-to-ground integrated monitoring and early warning system and a storage medium for a natural protection area. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. Along with the deep advancement of ecological civilization construction, the refinement and intelligent supervision of natural protection areas become increasingly important. The space, the sky and the ground integrated monitoring technology provides massive multi-source space-time data for a natural protection area through integrating satellite remote sensing, unmanned aerial vehicle aerial photography and a ground sensor network. Geographic Information Systems (GIS) are widely used as a traditional spatial data management platform in natural protection area management, and are mainly used for displaying and analyzing historical and current data. However, when the existing technical scheme is used for coping with the complex management requirement of the natural protection area, a plurality of limitations still exist, namely, firstly, data from different sources such as space, day and ground have differences in format, standard and space-time reference, so that a data island is formed, and an effective fusion tool is lacking, so that the collaborative analysis of the data is difficult, and the whole value is difficult to develop. Second, data processing and analysis is highly dependent on manual labor, e.g., manual screening and identification of wild animal species from tens of thousands of infrared camera photographs is inefficient, and the associated analysis capability for unstructured data such as sound, video, etc. is inadequate. Furthermore, the existing system mainly uses static display and post analysis, lacks space-time dynamic prediction and intelligent early warning capability for ecological processes and disaster events (such as forest fires and illegal invasion), and is difficult to realize the transition from passive response to active decision. In addition, the traditional machine learning model has poor generalization capability, is difficult to adapt to the unique environmental characteristics of different protection areas and the rapidly-changing monitoring requirements, and has high model updating and maintenance cost. Therefore, a new technical scheme capable of effectively integrating multi-source heterogeneous data, realizing automatic intelligent analysis and providing accurate space-time prediction and collaborative decision support is urgently needed in the field so as to promote scientificity, timeliness and initiative of natural protection area management. Disclosure of Invention In order to overcome the defects of the background technology, the application provides an air-to-ground integrated monitoring and early warning method, an air-to-ground integrated monitoring and early warning system and a storage medium for a natural protection area, wherein scattered data of the natural protection area are converted into linkage information through systematic integration of a deep learning technology, manual low-efficiency processing is updated into intelligent active decision, and the application value of the air-to-ground integrated data in the management of the protection area can be fully released. In order to achieve the above purpose, the present application provides the following technical solutions: In a first aspect, an air-space integrated monitoring and early warning method for a natural protection area is provided, and the method comprises the following steps: the method comprises the steps of collecting multisource heterogeneous monitoring data of a natural protection area through integrated deployment of satellite remote sensing equipment, unmanned aerial vehicle aerial photographing equipment and various ground sensors; Carrying out structural enhancement on the acquired multi-source heterogeneous monitoring data to construct a unified ecological database; Inputting the data subjected to structural reinforcement into a trained deep learning model for reasoning so as to identify a monitoring target or an abnormal event of a preset category, wherein the deep learning model is based on a convolutional neural network architecture; And generating early warning information according to the output result of the deep learning model and pushing the early warning information to a management terminal. Further, the structured enhancement includes: identifying and removing abnormal values through a statistical method, and filling missing values by using an interpolation method or a statistical value; Performing regular field calibration on the groun