KR20240000831A - And filter mounted on the master controller memory server of machine learning based perimeter intrusion detection sensor system - Google Patents

Abstract

The present invention relates to a condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system that reduces false alarms in response to the surrounding environment and weather of major national protection facilities. More specifically, it is divided into sections. They are installed in large numbers at predetermined intervals along the length of each fence, and generate an alarm signal by detecting changes in acceleration due to shock and vibration generated on the fence, changes in inclination of the fence, and motion of objects approaching the fence. Intelligent complex sensor that An environmental collection device installed on each fence to measure and collect wind speed using an environmental sensor; It is placed in response to each fence divided into sections, and receives acceleration and slope change data, motion data, alarm signals, and wind speed data from the intelligent complex sensor and environmental collection device installed on each fence and transmits them to the main server. Including the main controller, the intelligent complex sensor includes an acceleration and tilt sensor unit that detects acceleration due to shock and vibration generated on the fence and changes in the tilt of the fence; A motion sensor unit that detects the motion of a target object approaching the fence; A communication unit that communicates with neighboring intelligent complex sensors and main controllers, respectively; A central processing unit that generates an alarm signal through acceleration change data, tilt change data, and motion data and transmits it to the main controller through a communication unit; a memory unit storing acceleration and tilt change data and motion data; It includes a power unit that supplies power to the central processing unit, and the main controller includes a communication unit that communicates with an intelligent complex sensor, an environmental collection device, and a main server, respectively. A central processing unit that receives acceleration and slope change data, motion data, alarm signals, and wind speed data through the communication unit and transmits them to the main server; A memory unit that stores acceleration and slope change data, motion data, and wind speed data; In a machine learning-based fence intrusion detection system characterized by including a power unit that supplies power to the central processing unit, the memory unit of the main controller has a built-in main controller memory server that communicates with the user PC, and the main controller memory server has a built-in main controller memory server that communicates with the user PC. When the condition filter is installed and the condition filter is activated in the main controller memory server, the condition filter enters the primary detection state when one of the acceleration and tilt sensor units and the motion sensor unit detects it, and the preset time in the primary detection state If the rest of the acceleration and tilt sensor unit and the motion sensor unit detect while, it enters a secondary detection state and transmits an alarm signal to the main controller, and in the primary detection state, the acceleration and tilt sensor unit and the motion sensor operate for a preset time. This relates to a condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system, which has the function of returning to the previous state of the first detection state if the remaining parts do not detect it.

Description

Condition filter mounted on the master controller memory server of machine learning based perimeter intrusion detection sensor system}

The present invention relates to a condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system. In particular, machine learning-based fence intrusion detection that reduces false alarms in response to the surrounding environment and weather of major national protection facilities. This relates to the condition filter mounted on the main controller memory server for the system.

Most of the conventional fence mounted sensors used in major national protection facilities apply only one detection technology and convert the vibration transmitted to the sensor into an electrical signal to maintain a preset threshold level. An alarm is determined when it exceeds the time limit.

However, the conventional technology is a widely used technology in domestic and foreign industries, and is a technology that is learned and exposed to a lot of virtual hostile intruders, so the detection ability is greatly reduced and many false alarms occur.

In addition, the fence mount cable sensor is installed on the security fence fabric and does not distinguish between naturally occurring vibration and vibration from actual intrusion activities such as cutting/bending/wolting/lifting, and detects changes in electrical energy or light. Because only a single technology was applied to detect the rise value, maintenance time, and number of times compared to the threshold value by converting it into an electrical signal, the probability of false alarms was high and reliability was low.

In addition, fence-mounted cable sensors often use Co-axial Cable, Magnetic Polymer Cable, Ceramic Magnetic Cable, and Fiber Optic cable, and change electrical or optical signals proportional to the magnitude of vibration transmitted directly to the cable sensor due to artificial mechanical shock. occurs, and the generated analog signal is converted into digital data and the alarm is analyzed and determined in a single microprocessor device.

However, a method of analyzing the magnitude of vibration transmitted to a cable sensor installed in a section of approximately 50 to 200 m with a single microprocessor unit (MPU) and determining an alarm is used, and the variable value of the entire section where the cable is installed is set to a fixed standard threshold. Judging the alarm based on the value alone cannot reflect the effects of the aging condition of the approximately 50-200m fence, installation sturdiness, ground conditions, wind speed, season, etc., so the probability of false alarms is high, and a fundamental solution cannot be proposed. .

In this regard, Patent Document 1 includes a sensing cable that senses an impact generated from the outside by having a central conductor inside and a sensor conductor provided on the outside of the central conductor; a sensing module connected to an end of the sensing cable, transmitting and receiving a communication signal using the central conductor, and calculating shock location information through a sensing signal generated from the sensor conductor when an external shock occurs; And a server unit for receiving and outputting impact location information through a signal processing unit, wherein the sensing cable includes a central conductor; a polyethylene core surrounding the central conductor; a pair of sensor spaces spaced apart from the central conductor and provided at corresponding positions inside the polyethylene core; and a pair of sensor conductors provided in the sensor space and flowing by an impact from the outside, wherein an intermediate insulating material, a copper wire and an outer shell are sequentially wrapped around the outside of the polyethylene core, and the sensor conductors are connected to the central conductor. A fence monitoring device was provided that generates a detection signal from a sensor conductor using electromotive force generated during an external impact within a magnetic field generated as a current flows.

In addition, Patent Document 2 includes a sensor unit that detects an abnormality or intrusion in the fence and generates an alarm signal; an alarm unit that outputs an alarm signal to the outside when the alarm signal occurs; and an inspection robot that moves along the net of the fence and outputs a test signal to the sensor unit, wherein the inspection robot includes: a body unit; At least one moving part connected to the body part; A fixed part connected to the moving part and detachable from a part of the net; and a control unit that controls the moving part and the fixed part. An inspection system for a fence detection sensor was performed.

In addition, Patent Document 3 discloses a fence comprising pillars vertically arranged at spaced intervals, and a surface supported by the pillars and forming the boundary of the fence; A sensor unit coupled to and disposed on the surface of the fence; A cable connected to the sensor unit; and a control center that receives the sensing value input from the cable, wherein the sensor unit detects a change in the position of the surface at a specific location and generates a signal to the control center through the cable. Provided a fence monitoring system using surface-mounted hierarchical structure 3D sensors.

In addition, Patent Document 4 includes a plurality of piezoelectric accelerometer detection sensors installed on different types of continuously connected fences; Among the different types of fences, piezoelectric accelerometer sensors installed on the same type of fence are grouped, and sensitivity is set for the grouped piezoelectric accelerometer sensors. When an event outside the set sensitivity range occurs, the piezoelectric accelerometer sensor where the event occurred is set. A sensor detection analyzer that receives location information from an accelerometer detection sensor; The sensor detection analyzer is set to group piezoelectric accelerometer detection sensors installed on the same type of fence, sensitivity is set to the piezoelectric accelerometer detection sensors corresponding to each set group, and the location of each grouped piezoelectric accelerometer detection sensor is set. Setting terminal for setting information; And a control system that receives location information from the at least one piezoelectric accelerometer sensor from the sensor detection analyzer and displays it on a map, wherein the setting terminal tells the sensor detection analyzer that a fence is installed even if it is the same type of fence. Piezoelectric accelerometer detection sensors are set to be grouped according to environmental conditions, sensitivity is set to grouped piezoelectric accelerometer detection sensors according to environmental conditions, and the sensor detection analyzer detects different types of fence sizes or piezoelectric accelerometers. A smart exterior intrusion detection system is provided, wherein the sensor detection analyzer is additionally installed when the design of the plurality of piezoelectric accelerometer detection sensors is exceeded according to the installation interval of the sensors.

However, there is currently a need for a fence intrusion detection system different from patent documents 1, 2, 3, and 4 that can reduce false alarms by actively responding to the surrounding environment and weather of major national protection facilities.

Patent Document 1: Domestic Patent No. 10-1646825 (registered on August 2, 2016) Patent Document 2: Domestic Patent No. 10-1871634 (registered on June 20, 2018) Patent Document 3: Domestic Patent No. 10-1287079 (registered on July 11, 2013) Patent Document 4: Domestic Patent No. 10-2203541 (registered on January 11, 2021)

Therefore, unlike the prior art, the present invention has a function of transmitting an alarm signal only when both the acceleration and tilt sensor units and the motion sensor unit are detected, thereby preventing false alarms by optimally responding to the surrounding environment and weather of major national protection facilities. The purpose is to provide a condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system.

In order to achieve the above object, the present invention is installed at predetermined intervals along the length of each fence divided into sections, and changes in acceleration due to shock and vibration generated on the fence, change in inclination of the fence, and , an intelligent complex sensor that detects the motion of a target object approaching the fence and generates an alarm signal; An environmental collection device installed on each fence to measure and collect wind speed using an environmental sensor; It is placed in response to each fence divided into sections, and receives acceleration and slope change data, motion data, alarm signals, and wind speed data from the intelligent complex sensor and environmental collection device installed on each fence and transmits them to the main server. Including the main controller, the intelligent complex sensor includes an acceleration and tilt sensor unit that detects acceleration due to shock and vibration generated on the fence and changes in the tilt of the fence; A motion sensor unit that detects the motion of a target object approaching the fence; A communication unit that communicates with neighboring intelligent complex sensors and main controllers, respectively; A central processing unit that generates an alarm signal through acceleration change data, tilt change data, and motion data and transmits it to the main controller through a communication unit; a memory unit storing acceleration and tilt change data and motion data; It includes a power unit that supplies power to the central processing unit, and the main controller includes a communication unit that communicates with an intelligent complex sensor, an environmental collection device, and a main server, respectively. A central processing unit that receives acceleration and slope change data, motion data, alarm signals, and wind speed data through the communication unit and transmits them to the main server; A memory unit that stores acceleration and slope change data, motion data, and wind speed data; In a machine learning-based fence intrusion detection system characterized by including a power unit that supplies power to the central processing unit, the memory unit of the main controller has a built-in main controller memory server that communicates with the user PC, and the main controller memory server has a built-in main controller memory server that communicates with the user PC. When the condition filter is installed and the condition filter is activated in the main controller memory server, the condition filter enters the primary detection state when one of the acceleration and tilt sensor units and the motion sensor unit detects it, and the preset time in the primary detection state If the rest of the acceleration and tilt sensor unit and the motion sensor unit detect while, it enters a secondary detection state and transmits an alarm signal to the main controller, and in the primary detection state, the acceleration and tilt sensor unit and the motion sensor operate for a preset time. If the rest of the units do not detect, a condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system is provided, which has the function of returning to the previous state of the first detection state.

The present invention performs primary detection in one of the acceleration and tilt sensor units and the motion sensor unit through a condition filter set in the main controller memory server, and the remainder performs secondary detection of a different type from the primary detection for a preset time. It has the effect of transmitting an alarm signal only when detected.

In other words, the condition filter is a situation in which naturally occurring vibration and shock occurs, such as wind, which may be judged as an alarm on the fence, or an intruder's artificial act, such as an object blown by the wind hitting the fence, but may be judged as an alarm on the fence. This has a useful effect in situations where sufficient vibration and shock occurs, that is, in false alarm situations.

In other words, it has the effect of reducing false alarms by optimally responding to the surrounding environment and weather of major national protection facilities.

The present invention has the effect of more effectively detecting an intruder's intrusion because the acceleration and tilt sensor unit and the motion sensor unit are physically integrated into an intelligent complex sensor.

In the present invention, since a waterproof rubber ring is installed at the part where the first and second main bodies and the cable are connected, it is effective in preventing electrical short circuit due to water inflow.

The present invention has the effect of recognizing through a tamper switch that the first and second main bodies have been forcibly disassembled by an intruder.

The present invention has the effect of allowing an alarm to be known audibly or visually from outside through an alarm device or a light emitting device.

In the present invention, since multiple intelligent complex sensors and the main controller communicate in a bus or ring type, installation is simple while using less cable.

1 and 2 are diagrams of the installation state of a machine learning-based fence intrusion detection system according to an embodiment of the present invention;
3 to 8 are full and partial configuration diagrams of a machine learning-based fence intrusion detection system according to an embodiment of the present invention;
Figure 9 is a machine learning flowchart of the intelligent complex sensor and main controller of the machine learning-based fence intrusion detection system according to an embodiment of the present invention;
Figure 10 is a flowchart of the condition filter operation of the machine learning-based fence intrusion detection system according to an embodiment of the present invention.

Hereinafter, the configuration of a specific embodiment of the present invention will be examined with reference to the accompanying drawings.

As shown in Figures 1 to 8, the machine learning-based fence intrusion detection system 1000 according to an embodiment of the present invention is a plurality of fences at predetermined intervals along the longitudinal direction of each fence 1 divided into sections. It is installed, and when an acceleration change due to shock and vibration generated in the fence 1 is detected, an alarm signal is generated by comparing the detected acceleration change data with various false alarm pattern data of the pre-stored false alarm pattern collection, and the fence (1) It includes an intelligent complex sensor 100 that generates an alarm signal when it detects a change in inclination of the fence 1) and generates an alarm signal when it detects the motion of a target object approaching the fence 1.

Here, the fence 1 is installed on the ground to surround and protect the country's main protection facilities.

In addition, the fence 1 is composed of posts installed on the ground at predetermined intervals, and barbed wire on which an intelligent complex sensor 100 is installed while connecting the posts.

In addition, the fence intrusion detection system 1000 includes an environmental collection device 200 installed on each fence 1 to measure and collect wind speed with an environmental sensor 210.

Here, the environmental collection device 200 is installed on the pillar of the fence 1, and, unlike the environmental sensor 210, another environmental sensor that measures weather changes can be connected.

In addition, the fence intrusion detection system 1000 is disposed in response to each fence 1 divided into sections, and receives acceleration information from the intelligent complex sensor 100 and the environment collection device 200 installed on each fence 1. And tilt change data, motion data, warning signal, and wind speed data are received and transmitted to the main server 400, and the stored acceleration change data is converted into various false alarm pattern data and machine learning from the pre-stored false alarm pattern collection. If it is analyzed and determined to be a false alarm, a main controller 300 is included that generates new false alarm pattern data and transmits it to the intelligent complex sensor 100.

Here, the false alarm pattern collection unit generates shock and vibration on the fence 1 in various cases such as artificially throwing or shaking an object at the fence 1, and measures it in advance with the acceleration and tilt sensor unit 110. A variety of false alarm pattern data is collected in advance by dividing the obtained data into patterns, and the false alarm pattern collection unit is stored in advance in the intelligent complex sensor 100 and the main controller 300.

In addition, a plurality of the intelligent complex sensors 100 and the main controller 300 are connected to communicate in a bus type or ring type.

Additionally, the main controller 300 manages multiple analysis algorithms.

In addition, the intelligent complex sensor 100 includes an acceleration and tilt sensor unit 110 that detects acceleration due to shock and vibration generated in the fence 1 and a change in inclination of the fence 1; A motion sensor unit 120 that detects the motion of a target object approaching the fence (1); A communication unit 130 that communicates with the neighboring intelligent complex sensor 100 and the main controller 300, respectively; Generates an alarm signal by comparing the detected acceleration change data with various false alarm pattern data from the pre-stored false alarm pattern collection, generates an alarm signal through the slope change data, and generates an alarm signal through the motion data to communicate with the communication unit 130. A central processing unit 140 that transmits data to the main controller 300 through; a memory unit 150 that stores a false alarm pattern collection in which the acceleration and slope change data, motion data, and various false alarm pattern data are collected; A power supply unit 160 that supplies power to the central processing unit 140 is included.

Here, the acceleration and tilt sensor unit 110 includes a three-axis accelerometer that measures changes in the acceleration of the fence (1) and a digital gyroscope (tilt machine) that measures changes in the tilt of the fence (1).

In addition, the memory unit 150 stores data exchanged with the intelligent complex sensor 100 and the main controller 300.

In addition, the intelligent complex sensor 100 is coupled to each other and includes a first, acceleration and tilt sensor unit 110, a communication unit 130, a central processing unit 140, a memory unit 150, and a power unit 160. 2 Main body portions 101 and 102 are included.

Here, the first and second body parts 101 and 102 are fastened to the fence 1 with a fixing bracket and fastening member.

In addition, on one side of the first and second main bodies 101 and 102, first installation holes 101a and 102a are provided, respectively, facing each other when the first and second main bodies 101 and 102 are coupled to each other. It is formed as a pair, and a pair of second installation holes 102b are formed on the outer surface of the second main body 102 that does not face the first main body 101.

In addition, the intelligent composite sensor 100 is installed by inserting between the first installation holes 101a and 102a, and includes a waterproof rubber ring 103 having a pair of through holes 103a on the outer surface; A pair of cables 104 that pass through the through hole 103a and are connected to the communication unit 130 and the power supply unit 160 are included.

Then, the motion sensor unit 120 is inserted and installed into each of the pair of second installation holes 102b.

Here, the motion sensor unit 120 is composed of a PIR element that detects infrared heat of a target object, such as a person, when the size and speed of the target object change.

In addition, one of the detection functions of the motion sensor unit 120 consisting of the PIR element is a method of detecting signals by accumulating them over a certain period of time, and is a function of detecting the continuity of small signals that may move slowly or occur at a long distance, and is a reference value. When exceeded, the intelligent complex sensor 100 generates an alarm signal.

In addition, the remaining one of the detection functions of the motion sensor unit 120 consisting of the PIR element is a function to quickly detect when a large level signal is briefly detected, and when the reference value is exceeded, the intelligent complex sensor 100 generates an alarm signal. I do it.

In addition, the motion sensor unit 120 is not limited to a PIR element and can be replaced with an ultrasonic element or a lidar element that detects the approach of a target object such as a person.

In addition, the intelligent complex sensor 100 is built into the first and second main bodies 101 and 102 that are coupled to each other, and when the first and second main bodies 101 and 102 are disconnected from each other, an alarm signal is sent. A tamper switch 170 that generates and transmits to the main controller 300 is included.

Here, the alarm signal generated by the tamper switch 170 is transmitted to the main controller 300 as a high-priority alarm signal.

In addition, the tamper switch 170, acceleration and tilt sensor unit 110, communication unit 130, central processing unit 140, memory unit 150, and power unit 160 are mounted on the PCB.

In addition, the main controller 300 includes a communication unit 310 that communicates with the intelligent complex sensor 100, the environment collection device 200, and the main server 400, respectively; Acceleration and tilt change data, motion data, warning signals, and wind speed data are transmitted through the communication unit 310 and transmitted to the main server 400, and the stored acceleration change data is stored in a pre-stored false alarm pattern collection unit. A central processing unit 320 that analyzes various false alarm pattern data and machine learning and, if it is determined to be a false alarm, generates new false alarm pattern data and transmits it to the intelligent complex sensor 100; A memory unit 330 that stores a false alarm pattern collection in which the acceleration and slope change data, motion data, wind speed data, and various false alarm pattern data are collected; A power supply unit 340 that supplies power to the central processing unit 320 is included.

Here, each of the intelligent complex sensors 100 and the main controller 300 perform serial communication through the communication unit 130 and 310, and the main controller 300 and the main server 400 communicate with the communication unit 310. ) is used for Ethernet communication.

In addition, the main controller 300 includes an input/output unit 350 that communicates with the alarm device 600 or the light emitting device 610.

Here, the alarm device 600 is made of a siren, etc., the light emitting device 610 is made of a warning light, etc., and the input/output unit 350 sends an alarm signal in digital form to the alarm device 600 or the light emitting device. It is output through (610).

And, the environmental collection device 200 that transmits wind speed data to the main controller 300 includes an input/output unit 220 that receives analog wind speed data measured by the environmental sensor 210; A/D converter 230 that converts the analog wind speed data into digital wind speed data; A communication unit 240 that communicates with the main controller 300; A central processing unit 250 that transmits the digital wind speed data to the main controller 300 through the communication unit 240; A memory unit 260 that stores the digital wind speed data; A power supply unit 270 that supplies power to the central processing unit 250 is included.

Here, the environment collection device 200 is connected to at least one main controller 300, and the environment collection device 200 and the main controller 300 communicate serially through the communication units 240 and 310.

Additionally, the memory unit 260 stores data exchanged with the main controller 300.

In addition, the memory unit 330 of the main controller 300 is equipped with a main controller memory server 331 that communicates with the fixed or mobile user PC 500.

Here, the main controller memory server 331 and the user PC 500 communicate over Ethernet.

In addition, the memory unit 330 stores data exchanged with the intelligent complex sensor 100, the environment collection device 200, the main server 400, and the user PC 500.

And, the main controller memory server 331 is equipped with a condition filter 331a.

And, Figure 9 shows an intelligent complex sensor of the fence intrusion detection system 1000 according to an embodiment of the present invention, which collects various patterns that may occur in the fence 1 and then classifies them into false alarm patterns to reduce the false alarm rate. This is a flowchart of the machine learning process of (100) and the main controller (300).

To explain, the intelligent composite thin line 100 compares acceleration change data detected by the acceleration and tilt sensor unit 110 with various false alarm pattern data from a pre-stored false alarm pattern collection unit.

Here, the false alarm pattern data goes through data noise removal, monitoring, and analysis processes in the user PC 500 and is then stored in the false alarm pattern collection unit.

Therefore, if the acceleration change data does not correspond to false alarm pattern data, the intelligent composite thin line 100 generates an alarm signal and transmits it to the main controller 300 along with the acceleration change data.

Then, the main controller 300 analyzes the stored acceleration change data with various false alarm pattern data from the pre-stored false alarm pattern collection and machine learning, and if it is determined to be a false alarm, generates new false alarm pattern data and uses the intelligent complex sensor 100 will be transmitted to.

In addition, the main controller 300 and the plurality of intelligent complex sensors 100 connected to the main controller 300 collect and store new false alarm pattern data in the false alarm pattern collection unit.

However, if the acceleration change data is not close to a false alarm when analyzed using the machine learning, it is determined to be a final alarm and an alarm signal is transmitted to the main server 400 and the main controller memory server 331.

And, the user PC 500 checks the alarm signal through the main controller memory server 331.

And, Figure 10 is a flowchart of the operation of the condition filter 331a of the fence intrusion detection system 1000 according to an embodiment of the present invention, which reduces false alarms of the intelligent complex sensor 100 and increases the efficiency of intrusion detection at the same time.

To explain, the condition filter 331a is activated in the main controller memory server 331.

Therefore, when one of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 detects the first detection state.

That is, when the acceleration and tilt sensor unit 110 detects a change in acceleration or tilt, or the motion sensor unit 120 detects motion, the primary detection state is entered.

And, in the first detection state, if the rest of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 detect a type different from that in the first detection for a period of time preset by the main controller memory server 331. , When the secondary detection state is reached, an alarm signal is transmitted from the intelligent complex sensor 100 to the main controller 300.

Then, the alarm signal transmitted to the main controller 300 is determined to be the final alarm, and the alarm signal is transmitted to the main server 400 and the main controller memory server 331.

And, the user PC 500 checks the alarm signal through the main controller memory server 331.

Here, if the rest of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 do not detect, the primary detection state is maintained for a preset time.

However, if the rest of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 do not detect during the preset time in the first detection state, it returns to the previous state of the first detection state.

Then, the condition filter 331a is deactivated in the main controller memory server 331, and when one of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 detects it, the main control is performed in the intelligent composite sensor 100. An alarm signal is transmitted to the machine 300.

Then, the alarm signal transmitted to the main controller 300 is determined to be the final alarm, and the alarm signal is transmitted to the main server 400 and the main controller memory server 331.

And, the user PC 500 checks the alarm signal through the main controller memory server 331.

In the above, the present invention has been shown and described by taking specific preferred embodiments as examples, but the present invention is not limited to the above-described embodiments and is within the scope of the spirit of the present invention and is within the scope of common knowledge in the technical field to which the invention pertains. Various changes and modifications will be possible by those who have it.

100: Intelligent composite sensor 101: First main body
101a: first installation hole 102: second main body
102a: first installation hole 102b: second installation hole
103: Waterproof rubber ring 103a: Through hole
104: Cable 110: Acceleration and tilt sensor unit
120: motion sensor unit 130: communication unit
140: central processing unit 150: memory unit
160: Power unit 170: Tamper switch
200: environmental collection device 210: environmental sensor
220: input/output unit 230: A/D converter
240: Communication Department 250: Central Processing Department
260: memory unit 270: power unit
300: main controller 310: communication department
320: central processing unit 330: memory unit
331: Main controller memory server 331a: Condition filter
340: power unit 350: input/output unit
400: Main server 500: User PC
600: Alarm device 610: Light emitting device
1000: Fence intrusion detection system

Claims (5)

A plurality of them are installed at predetermined intervals along the longitudinal direction of each fence (1) divided into sections, and include changes in acceleration due to shock and vibration generated in the fence (1), changes in inclination of the fence (1), An intelligent complex sensor (100) that detects the motion of a target object approaching the fence (1) and generates an alarm signal;
An environmental collection device (200) installed on each fence (1) to measure and collect wind speed with an environmental sensor (210);
Acceleration and tilt change data, motion data, and Includes a main controller 300 that receives warning signals and wind speed data and transmits them to the main server 400,
The intelligent complex sensor 100 includes an acceleration and tilt sensor unit 110 that detects acceleration due to shock and vibration generated in the fence 1 and a change in inclination of the fence 1; A motion sensor unit 120 that detects the motion of a target object approaching the fence (1); A communication unit 130 that communicates with the neighboring intelligent complex sensor 100 and the main controller 300, respectively; A central processing unit 140 that generates an alarm signal through the acceleration change data, tilt change data, and motion data and transmits it to the main controller 300 through the communication unit 130; a memory unit 150 that stores the acceleration and tilt change data and motion data; It includes a power supply unit 160 that supplies power to the central processing unit 140,
The main controller 300 includes a communication unit 310 that communicates with the intelligent complex sensor 100, the environment collection device 200, and the main server 400, respectively; A central processing unit 320 that receives acceleration and slope change data, motion data, warning signals, and wind speed data through the communication unit 310 and transmits them to the main server 400; A memory unit 330 that stores the acceleration and slope change data, motion data, and wind speed data; In the machine learning-based fence intrusion detection system comprising a power supply unit 340 that supplies power to the central processing unit 320,
The memory unit 330 of the main controller 300 is equipped with a main controller memory server 331 that communicates with the user PC 500,
The main controller memory server 331 is equipped with a condition filter 331a,
When the condition filter 331a is activated in the main controller memory server 331,
The condition filter 331a enters the primary detection state when one of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 detects,
If the rest of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 detect for a preset period of time in the primary detection state, the secondary detection state is entered and an alarm signal is transmitted to the main controller 300. ,
If the rest of the acceleration and tilt sensor unit 110 and the motion sensor unit 120 do not detect for a preset time in the first detection state, the function is to return to the previous state of the first detection state. A condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system. According to paragraph 1,
The intelligent complex sensor 100 is coupled to each other, and the acceleration and tilt sensor unit 110, the communication unit 130, the central processing unit 140, the memory unit 150, and the power unit 160 are built into the first and second main bodies. Including parts (101) (102),
When the first and second body parts 101 and 102 are coupled to each other, first installation holes 101a and 102a are formed on one side of the first and second body parts 101 and 102, respectively, and face each other. , a pair of second installation holes 102b are formed on the outer surface of the second body 102 that does not face the first body 101,
The intelligent composite sensor 100 is installed by inserting between the first installation holes 101a and 102a, and includes a waterproof rubber ring 103 having a through hole 103a on the outer surface; It includes a cable 104 that passes through the through hole 103a and is connected to the communication unit 130 and the power supply unit 160,
A condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system, characterized in that the motion sensor units 120 are respectively inserted and installed in the pair of second installation holes (102b). According to paragraph 2,
The intelligent complex sensor 100 is built into the first and second main bodies 101 and 102 that are coupled to each other, and generates an alarm signal when the first and second main bodies 101 and 102 are disconnected from each other. A condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system, characterized in that it includes a tamper switch 170 that transmits the data to the main controller 300. According to paragraph 1,
The main controller 300 is a main controller memory server for a machine learning-based fence intrusion detection system, characterized in that it includes an input/output unit 350 that communicates with the alarm device 600 or the light emitting device 610. Built-in conditional filters. According to paragraph 1,
A condition filter mounted on the main controller memory server for a machine learning-based fence intrusion detection system, characterized in that the plurality of intelligent complex sensors 100 and the main controller 300 communicate in a bus type or ring type.

Images