KR-102961785-B1 - Drones and abnormal noise detection alarm systems for monitoring military guard areas and critical dangerous facilities, and their methods

KR102961785B1KR 102961785 B1KR102961785 B1KR 102961785B1KR-102961785-B1

Abstract

The present invention relates to a drone and abnormal noise detection and alarm system and method for monitoring military border security areas and critical hazardous facilities. By having a border guard carry a portable acoustic detection and alarm device on their person while performing border duty, accurately detecting drones and abnormal noise, automatically issuing an alarm, and transmitting detection information of the drones and abnormal noise to the control room in real time, threats caused by drone intrusion can be blocked and neutralized at an early stage. The method for detecting and warning about a drone and abnormal noise for monitoring military border security areas and critical dangerous facilities according to the present invention detects the drone and abnormal noise source by detecting the frequency spectrum, time, and amplification of the drone and abnormal noise source through a portable acoustic detection alarm device (110) carried by a border guard during border duty, and automatically provides an alarm for the drone and abnormal noise source, wherein a) when a drone and an abnormal noise source are input through the microphone (253) of the on-device AI noise sensor (200) of the portable acoustic detection alarm device (110), the AI of the on-device AI noise sensor (200) operates to scan the noise state of the drone and abnormal noise source through machine learning of an ANN algorithm and performs data labeling to provide context by adding one or more meaningful and informative labels indicating whether there is a change in amplitude and frequency of the noise source, and calculates the reference noise data of the normal state and the changed noise data through machine learning to distinguish between normal and abnormal conditions for the drone and abnormal noise source. a) a step of detecting drones and abnormal noise sources by deriving normal and abnormal results; and b) a step of distinguishing and displaying the machine learning operation and the normal and abnormal results by color through the indicator LED display of the on-device AI noise sensor (200), and when an abnormal result of the drone and abnormal noise source is detected, displaying a red LED on the indicator LED display as a blinking red LED, providing guidance through a voice message via a speaker (254), and transmitting the detection data of the drone and abnormal noise source in real time to an edge gateway device (300) through any one of the wireless LAN, Bluetooth, and USB provided in the on-device AI noise sensor (200).

Inventors

강언욱

Assignees

주식회사 레스코

Dates

Publication Date: 20260507
Application Date: 20250708

Claims (8)

A method for detecting and automatically alarming drones and abnormal noise sources for monitoring military border guard areas and critical dangerous facilities, wherein a portable acoustic detection alarm device (110) carried by a border guard soldier during border duty detects the frequency spectrum, time, and amplification of the drone and abnormal noise source, thereby detecting the drone and abnormal noise source, and, a) When a drone and an abnormal noise source are input through the microphone (253) of the on-device AI noise sensor (200) of the portable acoustic detection alarm device (110), the AI of the on-device AI noise sensor (200) operates to scan the noise status of the drone and the abnormal noise source and filter data through machine learning of an ANN algorithm, identify the raw data of the noise source, and perform data labeling that provides context by adding one or more meaningful and informative labels indicating whether there is a change in amplitude and frequency of the noise source, and calculates the normal state reference noise data and the changed noise data through machine learning to distinguish normal and abnormal conditions for the drone and the abnormal noise source and derive normal and abnormal results to detect the drone and the abnormal noise source; and b) displaying the machine learning operation and the normal and abnormal results by distinguishing them by color through the indicator LED display of the on-device AI noise sensor (200), and when an abnormal result is detected regarding the drone and abnormal noise source, displaying a red LED on the indicator LED display as a blinking signal, providing guidance through a voice message via the speaker (254), and transmitting the detection data of the drone and abnormal noise source in real time to the edge gateway device (300) through any one of the wireless LAN, Bluetooth, and USB provided in the on-device AI noise sensor (200); The above portable acoustic detection alarm device (110) is, A waterproof and explosion-proof case (111) having a sound source detection microphone (112) and an abnormality notification speaker (113) positioned on the front, a wireless transmission antenna (114) for situation propagation, an abnormality detection notification LED (115), and an alarm sound adjustment volume (116) positioned on the upper side, an on/off switch (118) positioned on one side, and a USB charging and interface terminal (119) positioned on the rear; A fixing clip (117) positioned on the rear of the above waterproof and explosion-proof case (111); and An on-device AI noise sensor (200) placed inside the above waterproof and explosion-proof case (111); A drone and abnormal noise detection and alarm method including
In claim 1, The above-mentioned on-device AI noise sensor (200) is, Using on-device artificial intelligence, noise is transformed into a Fast Fourier Transform (FFT) in the frequency domain without relying on output intensity, but specifically into a Short Time Fourier Transform (STFT) technique. Through lightweight supervised learning of ANN, QNN, and CNN neural networks, it detects drones and abnormal noise sources, and detects not only the frequency spectrum of specific noise sources but also three factors: time and amplification. The application structure of the aforementioned ANN, QNN, and CNN neural networks is processed through quantization, pruning, and knowledge distillation via model structure optimization, thereby performing supervised learning on various causes of abnormal sound sources, generating labels appropriate to the class, and finally porting it to on-device AI for operation. Drone and abnormal noise detection and alarm method.
In claim 1, The edge gateway device (300) above is, a) receiving normal and abnormal result data regarding the drone and abnormal noise source from the on-device AI noise sensor (200) and performing a data preprocessing process; b) A step of performing a data post-processing step after performing a data pre-processing step in the on-device AI processor of the edge gateway device (300); c) a step of automatically correcting changed codes and data by analyzing and deriving multiple error types and anomaly occurrences, including changes in communication data, communication traffic, communication failures, and path settings, by learning the data preprocessing and postprocessing processes in the above-described on-device AI processor through machine learning; and d) a step of transmitting normal and abnormal result data regarding the drone and abnormal noise source, post-processed at the edge gateway device (300), to a remote control room server (400) in real time; A drone and abnormal noise detection and alarm method including
In a drone and abnormal noise detection and alarm system for monitoring military border guard areas and critical dangerous facilities, which detects drones and abnormal noise sources by detecting the frequency spectrum, time, and amplification of drones and abnormal noise sources through a portable acoustic detection and alarm device (110) carried by a border guard during border duty, The above portable acoustic detection alarm device (110) is, A waterproof and explosion-proof case (111) having a sound source detection microphone (112) and an abnormality notification speaker (113) positioned on the front, a wireless transmission antenna (114) for situation propagation, an abnormality detection notification LED (115), and an alarm sound adjustment volume (116) positioned on the upper side, an on/off switch (118) positioned on one side, and a USB charging and interface terminal (119) positioned on the rear; A fixing clip (117) positioned on the rear of the above waterproof and explosion-proof case (111); and The on-device AI noise sensor (200) is positioned inside the above-mentioned waterproof and explosion-proof case (111), and when a drone and an abnormal noise source are input through the sound source detection microphone (112), the AI operates to scan the noise status of the drone and the abnormal noise source and filter data through machine learning of an ANN algorithm, identify the raw data of the noise source, perform data labeling to provide context by adding one or more meaningful and informative labels indicating whether the amplitude and frequency of the noise source have changed, calculate the reference noise data of the normal state and the changed noise data of the drone and the abnormal noise source through machine learning to distinguish normal and abnormal conditions of the noise source and derive normal and abnormal results to detect the drone and the abnormal noise source, and transmits the detection data of the drone and the abnormal noise source to an edge gateway device via wireless LAN, Bluetooth, or USB. An edge gateway device (300) having an RDBMS embedded therein that receives data from the on-device AI noise sensor (200), performs data preprocessing and postprocessing in an on-device AI processor, and learns the data preprocessing and postprocessing in the on-device AI processor through machine learning to analyze and derive multiple error types and anomalies including changes in communication data, communication traffic, communication failures, and path settings, automatically corrects changed codes and data, connects the network to enable compatibility with heterogeneous protocols through a routing function, and transmits the postprocessed data to a remote control room server via standardized Modebus RTU or Modebus TCP, and includes an intelligent data router function capable of routing to multiple network paths and autonomous data aggregation processing; and A remote control room server (400) that stores and monitors normal and abnormal result data regarding the drone and abnormal noise received from the edge gateway device (300), and remotely controls the on-device AI noise sensor (200); A drone and abnormal noise detection and alarm system including
delete
In claim 4, The above-mentioned on-device AI noise sensor (200) is, A body (210) formed in a rectangular frame shape, having a transmitting/receiving antenna hole (211), an ON/OFF button hole (212), and a menu selection button hole (213) formed on the upper surface, a USB terminal hole (214) formed on one side, and a speaker hole (215) formed on the lower surface; A front plate (220) positioned on the front of the body (210), having a microphone hole and an indicator display window (222) formed on the front; A rear plate (230) disposed on the rear of the body (210) and having a battery installation groove (231) formed on its upper surface; and A PCB board (260) disposed inside the body (210), wherein a transmitting/receiving antenna (240), an ON/OFF button (251), a menu selection button (252), a USB terminal (255), a microphone (253), a speaker (254), and an indicator LED display (223) are electrically connected and disposed on the upper part of the board, and a battery (280) is disposed on the lower part of the board; A drone and abnormal noise detection and alarm system including
In claim 4, The above drone and abnormal noise detection and alarm system is, A vibration and shock sensor that detects vibration and shock to detect factor value variables with simultaneity accompanied by a noise source, an RF scanner that detects and identifies a drone by detecting RF radio frequencies and RF radio frequency bands, and a spectrum analyzer that detects radio interference in real time, comprising one or more of Drone and abnormal noise detection and alarm system.
In claim 4, The above drone and abnormal noise detection and alarm system is, Including one or more of a night thermal imaging camera and a video camera, Drone and abnormal noise detection and alarm system.

Description

Drones and abnormal noise detection alarm systems for monitoring military guard areas and critical dangerous facilities, and their methods The present invention relates to a drone for monitoring military border security areas and critical hazardous facilities, an abnormal noise detection and alarm system, and a method thereof. More specifically, it relates to a drone for monitoring military border security areas and critical hazardous facilities and an abnormal noise detection and alarm system and method thereof that uses an on-device AI noise sensor to extract the acoustic noise and frequency band of the drone, thereby accurately detecting the drone and automatically issuing an alarm. Military personnel are on guard duty at the armistice line fences and military security zones to ensure the safety of the nation and its people. In the case of GP guard duty at the armistice line, reliance on binoculars and cameras for visual surveillance and patrolling is a concern, so intelligence and reconnaissance using drones pose a security threat not only to the vicinity of the armistice line but also to military units and critical national facilities. Although additional defense reinforcement projects, such as the use of CCTV, are currently underway due to the recent advancement of defense science, countermeasures for drone detection and recognition are urgently required to further strengthen drone security and prepare for early intelligence activities and threats. With the proliferation of state-of-the-art intelligent drones, there is a need to establish drone defense systems for national infrastructure and military facilities; however, there is currently no integrated system capable of detecting, identifying, and responding to drone threats in real time. In addition, as drones have recently become smaller and their flight speeds have increased, it is necessary to improve the detection performance of detection systems to defend against them; however, to effectively respond to the threat of micro-drones, an environment is required that enables real-time detection and identification from a long distance as much as possible, followed by real-time response. To defend against drone threats, many authorities are striving to find solutions for drone surveillance and countermeasures against drone attacks. While various detection technologies have been proposed to detect, identify, and respond to intrusive drones, complete detection is impossible with individual technological elements alone. Consequently, technologies are being introduced to build anti-drone systems by combining detection equipment such as radar, lidar, RF scanners, acoustic sensors, and electro-optical (EO)/infrared (IR) sensors. The aforementioned radar detection technology determines the location (distance/altitude/direction) and speed of an object by transmitting radio signals in a specific band and analyzing the energy and frequency of the returning waves. However, detecting drones with a small radar cross-section (RCS) requires very high radar transmission power, resulting in very high implementation costs and issues such as the need for government support, including frequency band allocation. The above RF detection technology detects the presence of signals in the 2.4GHz band used for transmitting and receiving control signals of a drone and the 5.8GHz band used for transmitting and receiving video data. Since Wi-Fi in the ISM (Industrial Scientific and Medical) band is widely used as a communication method, it is difficult to distinguish signals in urban areas, and when 3G/4G and satellite communication are used, it is difficult to determine whether the frequency used by the drone is the one being detected. The aforementioned optical detection technology analyzes images utilizing both visible light and thermal imaging ranges. However, using a high-magnification zoom lens for long-range detection narrows the field of view, allowing observation only at specific angles and requiring a precision control system. Consequently, it is generally integrated with radar detection equipment and utilized for visual identification; yet, thermal imaging detection equipment is expensive and has limitations in long-range detection of drones, which emit minimal heat. If the average movement speed of a drone is estimated at 40 km/h, it can travel 3 km within the 5-minute emergency response time, so detection is generally sought from 3 km away from major protected facilities. In this case, there is a problem in that introducing one set each of radar detection equipment, RF detection equipment, and optical detection equipment is expensive, requiring a budget of hundreds of millions of won for each, but the detection rate is only in the 70% range. In the case of optical equipment, in order to film and detect a small drone measuring 35 cm at a distance of 3 km, a zoom lens of 1,000 mm or more must be mounted, and at this time, the camera's field of view (FOV) becomes 0.3 deg