KR102365578B1 - Intrusion detection system combining high performance rader and machine learning - Google Patents

Abstract

The present invention detects an intrusion through a radar sensor adopting a 24 GHz frequency modulation continuous wave method in facility security, but improves the accuracy of object detection by convergence of big data and rushing machining technology, and enables immediate and intuitive monitoring and response 2D spatial information It relates to an intrusion detection system using high-performance radar and machine learning that can perform real-time monitoring.

Description

Intrusion detection system combining high performance radar and machine learning

The present invention relates to an intrusion detection system, and in detail, detects an intrusion through a radar sensor adopting a 24GHz frequency modulation continuous wave method in facility security, but improves the accuracy of object detection by convergence of big data and running machining technology, and provides an immediate and intuitive It relates to an intrusion detection system using high-performance radar and machine learning that can perform real-time monitoring based on 2D spatial information to enable monitoring and response.

In recent years, various product developments centering on the Internet of Things have been actively developed, and these technologies are gradually expanding into smart cities, public safety and disasters.

The smart city, public safety and disaster industry sectors are one of the sectors where physical security technology is the most important. Physical security is a security technology for the safe operation of major facilities and the prevention of disasters, and representative products include CCTV, IP Camera, alarm/monitoring, and biometric recognition devices.

Demand for physical security products such as CCTV, biometrics, alarm/monitoring, etc. is rapidly increasing for the purpose of preparing for various crimes, terrorist threats, and industrial secret leaks that are increasing worldwide.

Among these, detailed products of alarm/monitoring devices include infrared/laser/vibration/tension sensors, motion detectors/intrusion detection equipment, etc., which are used in various facilities for the purpose of protection from external intrusion, but with technical limitations As crime becomes more intelligent and sophisticated, it is insufficient to prevent evasive intrusion.

In other words, the existing detection sensors allow an intruder to evade intrusion by identifying the sensor characteristics, and the most used cameras are mainly used for identification of criminals after an incident rather than prevention or immediate response.

Accordingly, the security industry is releasing various convergence security-related products, but there are still problems with reliability and accuracy degradation, such as misinformation due to various external environmental changes and various unexpected objects that are not reflected in system design, or alarms are omitted due to missing objects to be monitored. is still in existence.

Registered Patent No. 10-1069406 (2011.09.26) Patent Publication No. 10-2014-0019586 (2014.02.17)

The present invention was created to solve the above problems, and an object of the present invention is to detect an intrusion through a radar sensor adopting a frequency-modulated continuous wave method fused with big data and rushing machining technology, but real-time monitoring based on 2D spatial information The goal is to provide an intrusion detection system using high-performance radar and machine learning that increases the accuracy of object detection and enables immediate and intuitive monitoring and response.

For the above purpose, the present invention provides an intrusion detection system for a facility, comprising: a radar module for outputting a radio wave of a frequency set as a monitoring area, and receiving a reflected wave to generate a detection signal; a storage module for accumulating and storing spatial information of the monitoring area and the detection signal over time; a correction module for collecting and storing an initial value of a detection signal for spatial information for each frequency band; a search module that converts the output frequency of the radar module and analyzes the detection signal, and generates an alarm signal when an object is identified by comparing it with an initial value of the detection signal corresponding to the output frequency; a first learning module for updating the analysis criteria of the search module according to the data accumulated and stored for each frequency and correction information input in response to the alarm signal; It is characterized in that it comprises a.

At this time, the sensor module is composed of a plurality of vibration sensors to detect the vibration in the monitoring area; a vibration analysis module for extracting and transmitting pattern information corresponding to the request from among a plurality of pattern information constituting the detection information of the sensor module in response to a request for pattern information; When the detection signal is received from the radar module, pattern information is requested from the vibration analysis module, and when the received pattern information is analyzed and identified as preset pattern information, an alarm signal is generated through the search module, and the a second learning module for re-requesting another pattern signal associated with the reception time of the detection signal during identification, and transmitting an object identification result through the radar module among the pattern signals received according to the re-request to the vibration analysis module; may further include.

In addition, the image output unit for displaying and outputting the position of the object identified through the search module on a two-dimensional drawing of the monitoring area linked to the spatial information together with the positions of the radar module and the sensor module; may further include.

In addition, the environment collecting unit for collecting weather information including weather and wind direction, wind speed and temperature; Further comprising, the first learning module and the second learning module may be configured to manage the weather information in association with a detection signal and a pattern signal, respectively.

In addition, the radar module and the search module are interlocked through a short-range wireless communication network, but may be configured to be interlocked through transmission and reception of an identification signal including a security code corresponding to a mutual frequency change.

Through the present invention, it is possible to solve the shortcomings of the existing PIR sensor type intrusion detection method, such as short recognition distance, the influence of the environment (strong wind, temperature, etc.) or misrecognition due to the movement of animals or trees, and through high-sensitivity wireless radar technology. It can provide high-performance detection performance even in the distance, environmental change, and the type of sensing object.

By improving the object detection accuracy, an intelligent unmanned/integrated intrusion monitoring system can be built, and for effective monitoring and response of intrusions and anomalies, a realistic 2D visualization technique that considers spatial information such as time, location, path, and range is used. Security situation can be expressed intuitively based on spatial information.

In addition, the present invention collects and integrates logs based on big data and analyzes correlations to improve the monitoring of unknown threats that were previously overlooked and to respond quickly to accidents, and the monitoring target object At the same time, it is possible to reduce the misinformation rate due to non-obvious objects and, at the same time, repeatedly learn a pattern in which a monitoring target object is identified from among a plurality of images constituting an image, so that an optimal determination capable of identifying the monitoring target object can be made.

1 is a conceptual diagram of the present invention;
2 is a block diagram illustrating a configuration and a connection relationship according to an embodiment of the present invention.

Hereinafter, the configuration of the intrusion detection system using the high-performance radar and machine learning of the present invention will be described in detail with reference to the accompanying drawings.

1 is a conceptual diagram of the present invention, FIG. 2 is a block diagram showing a configuration and a connection relationship according to an embodiment of the present invention, and the present invention is a radar module 100 , a management server 200 , and a sensor module 300 . Intrusion in various facilities is detected through the main configuration of is configured to

The radar module 100 applies a 24 GHz FMCW type radar 110 to output a radio wave of a frequency set as a monitoring area, and receives a reflected wave to generate a detection signal.

The FMCW type radar 110 is a type of radar that continuously emits a frequency-modulated signal, and can obtain a distance and a speed from a monitoring target object by using a delay time and a Doppler frequency shift of a received signal.

In the present invention, through the radar 110, regardless of the change in heat (temperature) of the monitoring area, the sensing object distance and size discrimination ability is excellent, so small animals can be distinguished, and the sensing object signal pattern analysis , it is possible to install buried/concealed by detecting non-metal penetration, and it is not affected by fog, etc.

At this time, the radar module 100 may be configured through a single radar or a plurality of radars capable of outputting radio waves of different frequencies according to the area or characteristics of the monitoring area, and the generated detection signal is transmitted to the management server 200 . In order to do this, the first communication unit 120 is provided.

The first communication unit 120 can be configured by applying various communication methods, but in the embodiment of the present invention, it is configured to transmit and receive a detection signal and a control signal of a radar in a short-range wireless communication method using an IoT network, respectively.

The sensor module 300 is configured to improve the accuracy and learning ability of intrusion detection by assisting the radar module. do.

Conventionally, a phenomenon in which a Doppler radar or an IR sensor is misdetected even in a situation where there is no object due to vibration of a building and an external external force occurred. Therefore, in the present invention, by collecting data on the vibration value through the sensor module 300 to find the noise level, the range of the distance displacement value due to the vibration is determined to reduce the misdetection due to the minute vibration.

In addition, it can be used to remove noise caused by vibration when intrusion is detected by applying the vibration value due to the surrounding environment (construction, earthquake-prone area, etc.) during initial installation.

The sensor module 300 also includes a third communication unit 320 configured with the same short-range wireless communication module as the first communication unit, and transmits the vibration pattern signal detected by the short-range wireless communication method using the IoT network to the management server 200 ) will be sent.

The management server 200 is configured to control the radar module and analyze the received detection signal to determine whether there is an intrusion and to make an alert, a second communication unit 210, a storage module 220, and a correction module ( 230), a search module 240, a first learning module 250, a vibration analysis module 260, a second learning module 270, an image output unit 280, and an environment collection unit 290 ), and is located in a place where unauthorized personnel are controlled, such as the central control room of the facility, and the control center 400 of various security companies in charge of security of the facility.

The storage module 220 is a data storage of the management server 200, and the spatial information of the monitoring area, the detection signal transmitted from the radar module 100, and the pattern signal transmitted from the sensor module 300 are transmitted over time. accumulated according to

In this case, the spatial information is various images that can confirm the structure of the monitoring area, and in the case of an outdoor area, a map may be used, or in the case of an indoor area, it may be a variety of structural diagrams.

The correction module 230 is configured to create and store data that is a criterion for intrusion determination, and it is preferable that such reference information is collected before actually being put into monitoring work after the system according to the present invention is built in a facility. Specifically, with respect to the spatial information in which the radar module 100 is installed, the detection signal collected from the radar module is collected and stored for each frequency band in a state where there is no monitoring target, and the initial value of the pattern signal collected from the sensor module is also Save.

The correction module 230 can obtain this initial value setting in the form of big data collection, for example, by installing a radar module in a specific test site to collect detection information, but 4 times a day, 4 times at night, and a total of 8 times at the level of 1 Detected information can be collected for about 5 hours per episode.

The search module 240 converts the output frequency of the radar module 100 through a program and analyzes the detection signal, but compares the detection signal due to the reflected wave corresponding to the output frequency with the initial value of the correction module to the object is identified, an alarm signal is generated.

In this case, the alarm signal may be transmitted to the control center 400 or the designated terminal 500 as well as a separate alarm means such as a siren or a warning light installed in the facility.

The first learning module 250 updates the analysis criteria of the search module according to the data accumulated and stored for each frequency and the correction information input in response to the alarm signal. In other words, the more trial and error there is, the more accurate the judgment can be, and the more reliable the system can be built by using the data obtained through the intrusion detection system that has already been built and operated in the form of big data. there is.

In the case of the correction target, any object or reflector is additionally located within the detection area so that the reflected signal value and distance data can be extracted and classified at the time of object detection or non-detection, and vibration and initial ambient reflection that can cause misdetection In various situations where the target object can be detected from the values, the method of inputting the result by the administrator can be applied.

The vibration analysis module 260 extracts and transmits pattern information corresponding to the request from among a plurality of pattern information constituting the detection information of the sensor module 300 according to a request for pattern information.

In other words, it is a configuration for filtering noise by finding the noise level caused by vibration through machine learning. Various vibration pattern information is set as a reference value in advance, and at the same time, it is continuously cumulatively updated and provided.

As mentioned above, in general, as Doppler radar and IR sensors malfunction in vibration and movement of buildings, ghosts and ground effects can be reduced by detecting and referring to objects due to minute vibrations.

When the second learning module 270 receives a detection signal from the radar module 100, it requests pattern information from the vibration analysis module 260, analyzes the received pattern information, and when it is identified as preset pattern information, the Generate an alarm signal through the search module. At this time, when the pattern information is not identified, another pattern signal associated with the reception time of the detection signal is requested again, and the result of object identification through the radar module among the pattern signals received according to the re-request is the vibration analysis module 260 . will be sent to

That is, the magnitude of the amplitude of the vibration detected through the vibration sensor 310 is compared with the similarity between the pattern information through histogram matching, etc. to confirm that the object is set in response to the similarity between the pattern information. You can find similar objects by analyzing the pattern, or you can input new pattern information.

The image output unit 280 displays the location of the radar module 100 and the sensor module 300 together with the object identified through the search module 240 on the two-dimensional drawing of the monitoring area associated with the spatial information. The position is displayed and printed. Through this, the structure and situation of the detection area can be specifically identified through the control center 400 or the remote terminal 500 .

The environment collecting unit 290 is configured to collect meteorological information including weather, wind direction, wind speed and temperature, and to obtain weather information of the area where the facility is located in connection with the Meteorological Administration DB, including a separate measurement means provided in the facility. can

At the same time, the first learning module 250 and the second learning module 270 manage the detection signal and the pattern signal in connection with the weather information obtained through the environment collection unit 290, respectively, so that more detailed information management is possible. Accuracy can be improved, and the judgment criteria can be changed by reflecting the weather information.

In addition, a plurality of radar modules 100 and a search module 240 may be provided according to the size of the facility or monitoring area, and as mentioned above, the radar module 100 and the search module are short-range wireless communication through the IoT network. It is linked with the management server through the network. In this situation, threats such as hacking exist, so in order to strengthen the security of the detection signal transmitted from the radar module 100 to the management server 200, the first communication unit 120 and the management server 200 of the radar module 100. The second communication unit 210 of the may be configured to be linked through periodic transmission and reception of an identification signal including a security code corresponding to a mutual frequency change. That is, when the frequency band is changed at a set period, information about it is shared in advance to be used as a security authentication method, and a high level of security can be maintained as long as such a set value is not leaked.

The right of the present invention is not limited to the above-described embodiments, but is defined by the claims, and a person of ordinary skill in the art can make various modifications and adaptations within the scope of the claims. it is self-evident

100: radar module 110: radar 120: first communication unit
200: management server 210: second communication unit 220: storage module
230: correction module 240: search module 250: first learning module
260: vibration analysis module 270: second learning module 280: image output unit
290: Environment Collection Department
300: sensor module 310: vibration sensor 320: third communication unit
400: control center
500: terminal

Claims (5)

In the intrusion detection system of a facility,
A radar module 100 having a first communication unit 120 for outputting a radio wave of a frequency set as a monitoring area and receiving a reflected wave to generate a detection signal, and transmitting the detection signal in a short-distance wireless communication method using an IoT network (100) ; class,
In the configuration of the management server 200 having the first communication unit 120 and the second communication unit 210 for transmitting and receiving, respectively, the detection signal and the control signal of the radar,
a storage module 220 in which spatial information of a monitoring area and the detection signal are accumulated and stored over time;
a correction module 230 for collecting and storing an initial value of a detection signal for spatial information for each frequency band;
a search module 240 that converts the output frequency of the radar module and analyzes the detection signal, and generates an alarm signal when an object is identified by comparing it with an initial value of the detection signal corresponding to the output frequency;
a first learning module 250 for updating the analysis criteria of the search module 240 according to the data accumulated and stored for each frequency and correction information input in response to the alarm signal;
a sensor module 300 configured with a plurality of vibration sensors 310 to detect vibration in a monitoring area;
a vibration analysis module 260 for extracting and transmitting pattern information corresponding to the request from among a plurality of pattern information constituting the detection information of the sensor module 300 according to a request for pattern information;
When a detection signal is received from the radar module 100, pattern information is requested from the vibration analysis module 260, and when the received pattern information is analyzed and identified as preset pattern information, an alarm signal is generated through the search module. When the pattern information is not identified, another pattern signal associated with the reception time of the detection signal is requested again, and the result of object identification through the radar module among the pattern signals received according to the re-request is returned to the vibration analysis module 260 ) to the second learning module 270 to transmit;
Image output to display and output the position of the object identified through the search module 240 on a two-dimensional drawing of a monitoring area linked to the spatial information together with the positions of the radar module 100 and the sensor module 300 part 280;
an environment collecting unit 290 for collecting weather information including weather and wind direction and wind speed and temperature; is composed of
The first learning module 250 and the second learning module 270 manage the weather information in association with a detection signal and a pattern signal, respectively,
The first communication unit 120 and the second communication unit 210 change the frequency band at a set period, but share the frequency band information that is changed through periodic transmission and reception of an identification signal including a security code corresponding to the mutual frequency change in advance. An intrusion detection system, characterized in that it is used as a security authentication means and interlocked. delete delete delete delete

Images