CN-121789406-B - Intelligent supervision and safety early warning system for water area

CN121789406BCN 121789406 BCN121789406 BCN 121789406BCN-121789406-B

Abstract

The invention discloses an intelligent water area supervision and safety early warning system, which relates to the technical field of water area monitoring and comprises a data acquisition and preprocessing unit, an image data extraction unit, a warning line data recognition unit and an acousto-optic data recognition unit, wherein the data acquisition and preprocessing unit is used for acquiring image data, warning line data and acousto-optic data of a target water area, preprocessing the image data, the warning line data and the acousto-optic data, the image data extraction unit is used for extracting building outlines from the preprocessed image data by utilizing an improved illegal building recognition model and carrying out illegal occupation judgment by combining with geofence data to obtain a judgment result, the warning line data recognition unit is used for recognizing the state of a warning line from the preprocessed warning line data by utilizing an improved warning line state detection model to generate a safety evaluation report, and the acousto-optic data recognition unit is used for recognizing the preprocessed acousto-optic data by utilizing audio analysis software to obtain the voiceprint data. The invention constructs a full-chain treatment system from intelligent identification to emergency response to evidence solidification, thereby improving the safety protection level.

Inventors

TANG CHUNLEI
Li Lule
XU DONGYI
ZHOU ZHIYUAN

Assignees

江苏迈鼎科技(集团)有限公司

Dates

Publication Date: 20260508
Application Date: 20260306

Claims (8)

1. Water area intelligent supervision and safety precaution system, its characterized in that includes: the data acquisition and preprocessing unit is used for acquiring image data, warning line data and acousto-optic data of the target water area and preprocessing the image data, the warning line data and the acousto-optic data; The image data extraction unit is used for extracting the outline of the building from the preprocessed image data by utilizing the improved illegal building identification model, and carrying out illegal occupation judgment by combining with the geofence data to obtain a judgment result; A warning line data identification unit for identifying the state of a warning line from the preprocessed warning line data by using an improved warning line state detection model, and generating a security evaluation report, comprising: The model optimization module is used for marking physical guard lines and electronic fences based on a target detection model frame, introducing a strip-shaped anchor frame design, dynamically adjusting the size of the anchor frame by utilizing a data driving method through clustering analysis of the scale of guard line data, and obtaining an improved guard line detection model so as to adapt to guard lines and fences with different sizes; The error correction module is used for detecting the preprocessed warning line data by using the improved warning line state detection model, identifying the abnormal state of the warning line data, correcting the improved warning line state detection model by using an error correction mechanism and finally outputting a safety evaluation report; the sound-light data identification unit is used for identifying the preprocessed sound-light data by utilizing audio analysis software to obtain voiceprint data; the data fusion unit is used for fusing the judging result, the security assessment report and the voiceprint data to construct a backtracking association chain; The monitoring and early warning unit is used for dynamically adjusting the acquisition frequency of the image data and the warning line data by monitoring the voiceprint data in real time, generating a evidence obtaining package of the illegal building, detecting the evidence obtaining package of the illegal building and generating early warning information of different levels, and comprises the following steps: the voiceprint monitoring module is used for collecting the acousto-optic data of the target water area in real time through the acousto-optic warning device, and identifying the acousto-optic data by utilizing audio analysis software to generate voiceprint data; The liveness judging module is used for judging the liveness of the voiceprint data of the target water area by analyzing the voiceprint data frequency acquired in real time, triggering a dynamic adjustment mechanism when the liveness exceeds the preset liveness, dynamically adjusting the acquisition frequency of the image data and the warning line data, and generating a evidence obtaining packet of the illegal building; And the safety early warning module is used for detecting the evidence obtaining package of the illegal building through the lightweight model, confirming whether the illegal occupying building accords with the illegal building standard, generating a detection result, and generating early warning information of different levels according to the detection result.
2. The intelligent water supervision and safety pre-warning system according to claim 1, wherein the data acquisition and pre-processing unit comprises: The image acquisition module is used for deploying the pan-tilt camera array to a high point along the line of the target water area, acquiring image data of the target water area in real time, and eliminating water surface reflection interference of the image data by utilizing a multispectral fusion algorithm to obtain preprocessed image data; the warning line acquisition module is used for carrying a triaxial stability augmentation cradle head and a real-time dynamic carrier phase difference technology on the unmanned aerial vehicle cluster to carry out inspection on a target water area to generate warning line data, and carrying out enhancement processing on the definition of the warning line data by utilizing an image processing and analyzing technology of a deep learning technology to obtain preprocessed warning line data; And the acousto-optic acquisition module is used for deploying the acousto-optic warning device to the ground of the target water area, identifying the acousto-optic data of the target water area, and denoising the acousto-optic data by utilizing a denoising technology to obtain preprocessed acousto-optic data.
3. The intelligent water area supervision and safety early warning system according to claim 2, wherein the carrying the tri-axial stability augmentation cradle head and the real-time dynamic carrier phase difference technology through the unmanned aerial vehicle cluster patrols and examines the target water area to generate warning line data, and the enhancing the definition of the warning line data by using the image processing and analysis technology of the deep learning technology, the obtaining the preprocessed warning line data comprises: the Thiessen polygon algorithm is adopted to conduct dynamic path planning and airspace distribution on the unmanned aerial vehicle clusters so as to ensure that the unmanned aerial vehicle clusters can cooperatively work, automatically avoid no-fly zones and mutual conflicts, and achieve maximum inspection range and no-blind area coverage; part of unmanned aerial vehicle clusters are responsible for executing oblique photography, acquiring images from different angles, and constructing a three-dimensional live-action model by combining synchronous positioning and map construction technologies; And the other part of unmanned aerial vehicle clusters concentrate on carrying out inspection along the river boundary of the target water area, and high-definition images of the physical guard line area and the electronic fence area are collected to generate guard line data.
4. The intelligent water supervision and safety precaution system according to claim 1, wherein the image data extraction unit comprises: the standardized processing module is used for carrying out standardized processing on the preprocessed image data to obtain standardized image data; The model improvement module is used for increasing the weight of the building in the target water area through a weighted cross entropy loss function based on the illegal building identification model, and weighting the building in the target water area through the cross ratio weighting loss to obtain an improved illegal building identification model so as to improve the accuracy of identifying the building by the improved illegal building identification model; The building detection module is used for carrying out object detection on the standardized image data by utilizing the improved illegal building identification model and extracting building contour data; the boundary judging module is used for carrying out overlapping degree calculation on the specified boundary of each building contour data and the geofence data based on the building contour data and combining the geofence data to judge whether the building exceeds the specified boundary; The non-exceeding boundary module is used for regarding as a normal building if the building does not exceed the specified boundary and storing the building into a pre-constructed database; And the exceeding boundary module is used for considering the illegal occupying building if the building exceeds the specified boundary, comparing the historically acquired image data set with the illegal occupying building by utilizing a time sequence image analysis technology, detecting the appearance of a newly added building or the area change of an original building, and generating a judging result.
5. The intelligent water supervision and safety precaution system according to claim 1, wherein the detecting the preprocessed warning line data by using the improved warning line detection model, identifying an abnormal state of the warning line data, correcting the improved warning line state detection model by an error correction mechanism, and finally outputting a safety assessment report includes: detecting the preprocessed warning line data by using an improved warning line state detection model, identifying the abnormal state of each warning line data, and comparing the abnormal state of the warning line data with the marked state label; if the abnormal state of the warning line data does not exceed the marked state label, directly generating a security assessment report; if the abnormal state of the warning line data exceeds the marked state label, triggering an error correction mechanism, calculating positioning errors, classification errors and confidence errors in the improved warning line detection model through a loss function, correcting and optimizing, and generating a safety evaluation report.
6. The intelligent water supervision and safety precaution system according to claim 5, wherein the calculation formula of the loss function is: ; in the formula, Representing a loss function value; A weight coefficient representing a positioning error; A grid number representing the feature map; Indicating a function, wherein if the ith grid is used, the jth frame is responsible for a certain target; A weight coefficient representing a target confidence error; A weight coefficient representing no target confidence error; A weight coefficient representing the classification error; indicating that if the box is not responsible for detecting any targets; representing the target confidence of the ith grid and the jth prediction frame; All represent abnormal states of the warning line data; all represent labeled status labels; conditional category probabilities representing abnormal states of the guard line data; the conditional category probability of the marked state label is represented, and c represents the category index.
7. The intelligent water supervision and safety precaution system according to claim 1, wherein the data fusion unit comprises: The data storage module is used for carrying out standardized processing on the judging result, the security assessment report and the voiceprint data and storing the standardized processing result, the security assessment report and the voiceprint data into a pre-constructed database; And the data tracing module is used for taking the standardized judging result, the security evaluation report and the voiceprint data in the database as references and matching the time stamps through space coordinates to construct a tracing association chain.
8. The intelligent water area monitoring and safety pre-warning system according to claim 7, wherein the step of determining the activity of the target water area by analyzing the frequency of voiceprint data acquired in real time, and triggering a dynamic adjustment mechanism when the activity exceeds a preset activity, and dynamically adjusting the acquisition frequency of image data and warning line data, and the step of generating the evidence collection package for the illegal building comprises: Analyzing the voice print data frequency obtained in real time, and judging the liveness of voice print data of a target water area; When the activity level does not exceed the preset activity level, continuing to monitor the activity level of the target water area; When the activity exceeds the preset activity, triggering a dynamic adjustment mechanism, and calculating and outputting the optimal acquisition frequency of the pan-tilt camera array and the unmanned aerial vehicle cluster according to the activity change rate of the voiceprint data; Synchronously acquiring image data and warning line data based on the adjusted frequency control holder camera array and the unmanned aerial vehicle cluster, respectively inputting the acquired image data and warning line data into an improved violation building identification model and an improved warning line state detection model for processing, and simultaneously extracting a judging result and a safety evaluation report; And aligning the judging result with the security assessment report by taking the timestamp as a reference, and performing null fusion processing to generate the violation building evidence obtaining packet.

Description

Intelligent supervision and safety early warning system for water area Technical Field The invention relates to the technical field of water area monitoring, in particular to an intelligent water area supervision and safety early warning system. Background Along with the acceleration of the urban process and the increase of the utilization of water area resources, how to effectively monitor the water area environment and ensure the safety of water activities becomes an important issue. The intelligent water area monitoring and safety early warning system aims to realize real-time monitoring and safety management of water areas such as rivers, lakes and seas through integrating modern information technologies such as Internet of things, big data analysis, artificial intelligence and the like. Not only can natural parameters such as water quality, water flow speed and the like be monitored, but also emergencies such as illegal invasion, drowning accidents and the like can be identified, and scientific basis and technical support are provided for water area management. Especially around the dense or important facilities of tourist area, port, reservoir etc., the application of the intelligent water area supervision and safety early warning system is important. In the existing water area periphery supervision faces the identification process of building encroaching flood-passing areas and exceeding the blue line range, the environment factors such as illumination change, shooting angle difference, object shielding and the like are easy to interfere, precise space fusion means of GIS geofence data and real-time images are lacking, space positioning precision is insufficient, and accurate judgment of illegal encroaching behaviors is difficult to realize under the constraint of refined space. For slender targets such as warning lines, the problems of breakage, offset, shielding and the like often occur under a complex background, the existing detection model is low in detection recall rate of the targets due to lack of targeted network structure design and loss function optimization, and meanwhile, the detection delay is high, so that high-real-time support cannot be provided for safety early warning. The whole flow from discovery to disposal of water area supervision events lacks a closed-loop association cooperative mechanism, each link works relatively independently, monitoring records of law enforcement behaviors are incoherent and incomplete, event tracing capability is weak, law enforcement compliance and disposal efficiency are difficult to be effectively guaranteed, construction activity judgment lacks intelligent dynamic analysis basis, illegal activity intensity cannot be accurately identified through voiceprint characteristics and other data, unmanned aerial vehicle inspection frequency and image acquisition rate are fixed, and self-adaptive adjustment according to risk grades is difficult to cause untimely coverage of high-risk areas, and a quick response system cannot be formed. The existing supervision model is insufficient in generalization capability, precision fluctuation is easy to occur under the condition of terrain differences of different regions and environmental changes of various scenes, model update depends on a complex cloud training process, efficient online fine adjustment and quick iteration are difficult to achieve at the edge end, complex variable requirements of water domain supervision scenes cannot be flexibly adapted, and the overall improvement of supervision intelligence level is restricted. For the problems in the related art, no effective solution has been proposed at present. Disclosure of Invention Aiming at the problems in the related art, the invention provides an intelligent water area supervision and safety early warning system for overcoming the technical problems existing in the prior related art. For this purpose, the invention adopts the following specific technical scheme: The invention provides an intelligent water area supervision and safety early warning system, which comprises: the data acquisition and preprocessing unit is used for acquiring image data, warning line data and acousto-optic data of the target water area and preprocessing the image data, the warning line data and the acousto-optic data; The image data extraction unit is used for extracting the outline of the building from the preprocessed image data by utilizing the improved illegal building identification model, and carrying out illegal occupation judgment by combining with the geofence data to obtain a judgment result; A warning line data identification unit for identifying the state of the warning line from the preprocessed warning line data by using the improved warning line state detection model, and generating a security evaluation report; the sound-light data identification unit is used for identifying the preprocessed sound-light data by utilizing audio analysis software to obtain voicep