KR20210085761A - Building collapse prevention system - Google Patents

Abstract

A building collapse prevention system according to one embodiment of the present invention includes: a 3-axis accelerometer attached to a first place of the building to collect acceleration data; a control unit for determining whether the building collapses by analyzing the collected acceleration data; and an alarm generating unit for providing the determination result to a user terminal. The control unit may determine whether the building collapses using an analysis model generated according to a result previously learned based on a deep learning algorithm with respect to a frequency image generated based on a frequency conversion result for the acceleration data.

Description

Building Collapse Prevention System {BUILDING COLLAPSE PREVENTION SYSTEM}

The present invention relates to a building collapse prevention system. In more detail, it is provided in a building or construction site where earthquakes occur frequently, and in order to predict an accident in advance according to an earthquake or a disaster such as shaking or explosion of a building, it can be determined in advance whether a building will collapse according to the result of using a gravity sensor. It relates to a building collapse prevention system.

In general, high-rise buildings are designed to have earthquake-resistant characteristics, and a system for quickly responding to building vibrations caused by earthquakes is essential. In particular, in recent years, the construction of large structures is rapidly increasing along with the trend of building taller and lighter, and there is a risk of building collapse due to vibration of the building.

In addition to earthquakes, vibrations may occur in buildings due to various causes, such as being adjacent to subways or located in areas undergoing large-scale construction. Seriously, the problem of causing collapse of the building structure may occur.

Therefore, in the case of new construction such as most large buildings, the development of technologies that can predict and prevent the risk of building collapse in advance by measuring the vibration of the building continuously or periodically and observing the vibration is required. the current situation. In general, by using vibration monitoring, which refers to a technology that continuously monitors vibration, it is possible to prevent safety accidents such as detecting in advance that a building is collapsing due to vibration so that people in a building in danger of collapsing can evacuate safely in advance. making it possible

Republic of Korea Patent Publication No. 10-1807634 (published date: 2017.12.05)

The present invention is to solve the above problems, and by predicting and judging the prevention of building collapse in real time from the acceleration data obtained using the acceleration sensor provided in the building, a building collapse prevention system capable of predicting the building collapse in advance. The purpose is to provide.

As an embodiment of the present invention, a building collapse prevention system may be provided.

A building collapse prevention system according to an embodiment of the present invention is attached to a first place of a building, a three-axis acceleration sensor for collecting acceleration data, a control unit for determining whether the building collapses by analyzing the collected acceleration data, and a determination and an alarm generating unit that provides a result to the user terminal, wherein the control unit uses an analysis model generated according to a result previously learned based on a deep learning algorithm with respect to a frequency image generated based on a frequency conversion result for acceleration data. It is possible to determine whether a building is collapsing.

In the system for preventing collapse of a building according to an embodiment of the present invention, it further comprises a lidar sensor provided at a second place in the building, and the control unit trains the neural network on the amount of change in the point cloud obtained using the lidar sensor, , it is possible to determine the vibration state of the building based on the learned neural network, generate a warning signal, and provide it to the alarm generator.

In the system for preventing collapse of a building according to an embodiment of the present invention, the second place is a space provided with a column for supporting the building in the basement of the building, and the lidar sensor is located around the column to obtain a point cloud for the column. A plurality may be provided radially.

In the building collapse prevention system according to an embodiment of the present invention, the control unit determines the vibration intensity of the building by deriving the average amount of change in the impact driving force from the acceleration data, and the average change in the shock driving force is to be derived by the following [Equation 1] can

[Equation 1]

: 샘플 인덱스 n에서의 충격 추진력 평균 변화량(IMAV)

: Average change in impact thrust force (IMAV) at sample index n

: 데이터 구간의 길이

: length of data section

: 스케일링 파라미터

: Scaling parameter

: 임의의 시간 구간에 대한 충격량의 변화

: Change in the amount of impulse for an arbitrary time interval

: 시간 n에서의 가가속도

: acceleration at time n

According to the building collapse prevention system of the present invention, there is an effect of predicting the building collapse in advance by predicting and determining the building collapse prevention in real time from the acceleration data obtained using the acceleration sensor provided in the building.

1 is a block diagram illustrating a system for preventing collapse of a building according to an embodiment of the present invention.
2 is an exemplary view illustrating a state in which a three-axis acceleration sensor is attached to a building in a system for preventing collapse of a building according to an embodiment of the present invention.
3 is an exemplary view illustrating a state in which a lidar sensor is provided in a second place in a system for preventing collapse of a building according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art can easily implement them. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Terms used in this specification will be briefly described, and the present invention will be described in detail.

The terms used in the present invention have been selected as currently widely used general terms as possible while considering the functions in the present invention, but these may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technology, and the like. In addition, in a specific case, there is a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the corresponding invention. Therefore, the term used in the present invention should be defined based on the meaning of the term and the overall content of the present invention, rather than the name of a simple term.

When a part "includes" a certain element throughout the specification, this means that other elements may be further included, rather than excluding other elements, unless otherwise stated. In addition, terms such as "...unit" and "module" described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software, or a combination of hardware and software. . In addition, when a part is "connected" to another part throughout the specification, it includes not only a case in which it is "directly connected" but also a case in which it is connected "with another element in the middle".

1 is a block diagram showing a system for preventing collapse of a building according to an embodiment of the present invention, and FIG. 2 is a state in which a 3-axis acceleration sensor is attached to a building in the system for preventing collapse of a building according to an embodiment of the present invention. 3 is an exemplary view showing a state in which a lidar sensor is provided in a second place in the system for preventing collapse of a building according to an embodiment of the present invention. Hereinafter, the present invention will be described in detail with reference to the drawings described above.

As an embodiment of the present invention, a building collapse prevention system may be provided. In the present specification, the building 20 includes not only already completed building structures, but also structures under construction.

The building collapse prevention system according to an embodiment of the present invention is attached to the first place of the building 20, the three-axis acceleration sensor 100 for collecting acceleration data, and analyzing the collected acceleration data of the building 20 It includes a control unit 200 for determining whether collapse and an alarm generation unit 300 for providing the determination result to the user terminal, the control unit 200 with respect to the frequency image generated based on the frequency conversion result for the acceleration data Based on the deep learning algorithm, it is possible to determine whether the building 20 collapses using an analysis model generated according to a result previously learned.

According to one disclosure, the first place may be a location near the edge or vertex of the building 20 , or an area where vibration frequently occurs.

Building collapse prevention system according to an embodiment of the present invention may further include a memory (not shown). The memory may store a program for processing and control of the control unit 200, and may store data input to or output from the building collapse prevention system of the present invention. Programs stored in the memory may be classified into a plurality of modules according to their functions, where the plurality of modules may be modules functionally operated as software rather than hardware. Memory includes flash memory type, hard disk type, multimedia card micro type, card type memory (eg SD or XD memory), RAM (RAM, Random Access Memory SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, optical disk It may include at least one type of storage medium.

The control unit 200, for example, by driving software or a program to control at least one other component (Ex. hardware or software component) of the building collapse prevention system of the present invention connected to the control unit 200, and , various data processing and operations can be performed. The control unit 200 may load a command or data received from another component (eg, a communication unit (not shown)) into a volatile memory for processing, and store the result data in a non-volatile memory. The control unit 200 operates independently of the main processor (Ex. central processing unit or application processor), and additionally or alternatively, uses lower power than the main processor, or a sub-processor specialized for a specified function ( Ex. graphic processing) device, image signal processor, sensor hub processor, or communication processor). Here, the auxiliary processor may be operated separately from or embedded in the main processor.

According to one disclosure, the acceleration data may be time-series analog data. The controller 200 may generate a frequency image according to the frequency conversion result. The controller 200 may use various frequency conversion methods. Preferably, the frequency transformation is performed using a Fast Fourier Transform (FFT) method. That is, as the acceleration data is frequency-converted according to the FFT method, the intensity of each frequency component may be represented as a frequency spectrum, and the frequency image may be generated by converting it into an image form.

According to one disclosure, the controller 200 may perform size conversion with respect to a frequency image to a predetermined size, and may convert the size-converted frequency image into a grayscale image. The grayscale image may be an image converted so that a difference in brightness is set in a gradual step range between black and white.

According to one disclosure, the controller 200 may augment the number of frequency images by shifting the grayscale image. Shifting according to an exemplary embodiment may be performed by a predetermined number of pixels in a horizontal or vertical direction.

According to one disclosure, the controller 200 may determine whether the building 20 collapses using an analysis model with respect to the shifted frequency image.

According to one disclosure, the analysis model may be generated according to a result previously learned using a deep learning algorithm. The deep learning algorithm may include a Convolutional Neural Network (CNN). In addition, the recognition model may be combined with a Recurrent Neural Network (RNN) in addition to the CNN and generated according to a learned result. Hereinafter, it will be described based on the detection model formed according to the CNN-based learning result. First, if we briefly describe CNN, CNN can be a type of artificial neural network used to analyze input data (mainly, visual images), and features are extracted from the input data in the feature extraction layer, and are extracted from the classification layer. Based on the feature, the class to which the input data belongs may be classified. The feature extraction layer may include at least one convolutional layer and a pooling layer, and the classification layer is a fully connected layer including one hidden layer. ) can be The convolutional layer may correspond to a filter that generates a feature map representing a feature of an object through a convolution operation. In the pooling layer, a pooling operation that reduces the size of output data of the convolutional layer or emphasizes specific data may be performed. The pooling layer may include a max pooling layer and an average pooling layer. A fully connected layer may correspond to a classifier for classifying input data based on the extracted feature information.

According to one disclosure, the alarm generating unit 300 may provide a result determined by the control unit 200 to the user terminal. The user terminal includes not only a general desktop computer, but also a mobile phone, a smart phone, a laptop computer, a PDA (Personal Digital Assistants), a PMP (Portable Multimedia Player), a tablet PC, etc. A mobile terminal may also be included. In addition, the user terminal 30 may include home appliances such as a TV and a refrigerator.

In the building collapse prevention system according to an embodiment of the present invention, a lidar sensor 120 provided in the second place 21 in the building 20 is further included, and the control unit 200 includes the lidar sensor 120 . It is possible to train the neural network for the amount of change in the point cloud obtained using , to determine the vibration state of the building 20 based on the learned neural network, and to generate a warning signal and provide it to the alarm generator 300 .

A neural network according to one disclosure is a first CNN for extracting a first feature from input data, a second CNN for extracting a second feature from input data, and extracted first and second features Based on the classification layer for extracting attribution information, and the output of the first CNN and the second CNN are fused to be input to the classification layer may be further included.

The first CNN may correspond to a neural network formed based on a convolutional neural network (CNN) including a single convolutional layer. That is, the first CNN is provided with only the first convolutional layer and the first pooling layer to have a shallow structure, so that the first feature can be extracted from the input image.

Alternatively, the second CNN may correspond to a neural network formed based on a convolutional neural network including a plurality of convolutional layers. That is, the second CNN is provided with second to fourth convolutional layers and a second pooling layer to form a deep structure, so that a second feature can be extracted from input image data.

The neural network including only one convolutional layer may correspond to a neural network having a shallow structure as described above, and a neural network including a plurality of convolutional layers may correspond to a neural network having a deep structure. Accordingly, the first feature extracted from the first CNN that is a shallow structure may be a shallow feature, and the second feature extracted from the second CNN that is a deep structure may be a deep feature.

The fusion layer may be a layer that fuses the outputs of the first CNN and the second CNN to become the input of the classification layer. The fusion layer may be composed of a pooling layer.

The classification layer is the above-described fully connected layer, and the classification layer includes at least one hidden layer and an output layer, as well as a flatten layer for dimension change. may be included. Since the softmax activation function is used in the output layer, the degree of defect occurrence due to welding may be determined.

In more detail, in the first CNN, a single first convolutional layer and a first pooling layer are formed together to form a convolutional neural network having a shallow structure. Contrary to this, in the second CNN, three convolutional layers corresponding to the second convolutional layer, the third convolutional layer, and the fourth convolutional layer are connected to the second pooling layer to form a deep convolutional neural network. have. The number of filters and the filter size of the first to fourth convolutional layers may all be the same or may be set to be different from each other. As described above, when the input data is classified based on the features extracted from the shallow and deep neural networks as described above, more accurate results can be obtained compared to the discrimination results according to the general structure of the convolutional neural network. have. That is, after the feature extraction from the input data was separately performed through two convolutional neural networks, the vibration state of the building 20 was more accurately extracted according to the result of fusion in the fusion layer.

A LiDAR (Light Detection And Ranging) sensor according to one disclosure may be a sensor capable of detecting the shape of an object by measuring a distance from the object by illuminating a laser on the object.

The lidar sensor 120 according to one disclosure may detect the shape of the pillar 22 by acquiring point cloud data for the pillar 22, as described above with respect to the relationship between the change amount of the point cloud and the vibration state. The neural network can be trained together, and the vibration state of the building 20 can be automatically determined from the detection result of the lidar sensor 120 using the neural network according to the learning result.

In the system for preventing collapse of a building according to an embodiment of the present invention, the second place is a space in which a column 22 for supporting the building 20 is provided in the basement of the building 20, and the lidar sensor 120 is In order to obtain a point cloud for the pillar 22, a plurality of radially around the pillar 22 may be provided.

In the building collapse prevention system according to an embodiment of the present invention, the control unit 200 determines the vibration intensity of the building 20 by deriving the average amount of change in the impact driving force from the acceleration data, and the average change in the impact driving force is the following [Equation] 1] can be derived.

[Equation 1]

: 샘플 인덱스 n에서의 충격 추진력 평균 변화량(IMAV)

: Average change in impact thrust force (IMAV) at sample index n

: 데이터 구간의 길이

: length of data section

: 스케일링 파라미터

: Scaling parameter

: 임의의 시간 구간에 대한 충격량의 변화

: Change in the amount of impulse for an arbitrary time interval

: 시간 n에서의 가가속도

: acceleration at time n

Hereinafter, the process of calculating the average amount of change in the impact driving force will be described in more detail.

)은 운동량의 변화량(

)으로, 가해지는 힘과 시간의 곱으로 아래 수학식 4와 같이 표현할 수 있다.In general, the amount of impulse (

) is the change in momentum (

) as the product of the applied force and time, it can be expressed as in Equation 4 below.

[Equation 4]

는 시간 구간

동안 진동에 의해 가해진 힘일 수 있다. 상기 수학식 4에서

는 질량과 가속도의 곱이므로 시간 구간

에서의 충격량은 다음 수학식 5와 같이 표현할 수 있다.In Equation 4 above,

is the time interval

It may be a force applied by vibration during In Equation 4 above

is the product of mass and acceleration, so the time interval

The amount of impact in can be expressed as in Equation 5 below.

[Equation 5]

In addition, the acceleration data collected using the acceleration sensor 100 may be modeled as a combination of sinusoidal functions as shown in Equation 6 below.

[Equation 6]

는 가속도의 진폭을 나타내고,

는 각주파수를 나타내며,

는 위상을 나타낼 수 있다. 임의의 시간 구간에서의 가속도의 시간에 대한 변화를 가가속도(jerk)라 하며 상기 수학식 6으로부터 다음 수학식 7과 같이 표현할 수 있다.In Equation 6 above

represents the amplitude of the acceleration,

represents the angular frequency,

may indicate a phase. A change in acceleration with respect to time in an arbitrary time interval is referred to as an acceleration (jerk) and can be expressed from Equation 6 to Equation 7 below.

[Equation 7]

만큼 증폭 또는 감소되고 위상이 π/2 만큼 변화한 것일 수 있다. 임의의 시간 구간을

라 정의하고 동일한 시간 구간

에 대한 충격량의 변화

를 상기 수학식 5로부터 다음 수학식 8과 같이 정의할 수 있다.The acceleration in Equation 7 is the amplitude of the acceleration in Equation 6

It may be amplified or reduced by the amount and the phase is changed by π/2. any time period

defined as , and the same time interval

change in the amount of impulse to

can be defined as in Equation 8 below from Equation 5 above.

[Equation 8]

는

에서의 충격량일 수 있다. 또한, 상기 수학식 8의 연속시간 충격 추진력 변화량은

(=

)의 간격으로 샘플링한 이산시간 표현을 사용하여 시간 구간(

)에 대한 이산 충격 추진력 변화량을 아래 수학식 9와 같이 표현할 수 있다.In Equation 8 above,

is

may be the amount of impact in In addition, the continuous-time impact driving force change amount of Equation 8 is

(=

) using a discrete-time representation sampled at intervals of

) can be expressed as Equation 9 below.

[Equation 9]

이고

일 수 있다.In Equation 9 above,

ego

can be

)에 대한 이산 시간 표현을 사용하여 상기 수학식 9를 다음 수학식 10과 같이 표현할 수 있다.The time interval of continuous-time acceleration and acceleration shown in Equation 7 above (

Equation 9 can be expressed as Equation 10 below using the discrete time expression for ).

[Equation 10]

) 동안 받는 충격 추진력 변화량의 크기는 가가속도를 사용하여 다음 수학식 11과 같이 표현할 수 있다.From Equation (10), an object having a mass m is a constant unit time (

), the magnitude of the amount of change in the impact thrust received during the period can be expressed by the following Equation 11 using the acceleration.

[Equation 11]

이다. 전술한 과정에 따라 가가속도로부터 유도된 수학식 11의 충격 추진력 변화량은 각 주파수 성분 별 진폭 변화와 해당 주파수 변화(진폭의 증폭 또는 감소)를 동시에 반영할 수 있으며, 상기 수학식 1과 같이 표현될 수 있다.In Equation 11 above, the parameter

to be. The amount of change in the impact driving force of Equation 11 derived from the acceleration according to the above process can simultaneously reflect the amplitude change for each frequency component and the corresponding frequency change (amplification or decrease in amplitude), and can be expressed as in Equation 1 above. can

The above description of the present invention is for illustration, and those of ordinary skill in the art to which the present invention pertains can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and likewise components described as distributed may be implemented in a combined form.

Although the preferred embodiment of the present invention has been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concept of the present invention as defined in the following claims are also provided. is within the scope of the That is, the scope of the present invention is indicated by the following claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be interpreted as being included in the scope of the present invention. do.

10: Building Collapse Prevention System
100: acceleration sensor
200: control unit
300: alarm generator

Claims (4)

In the building collapse prevention system,
a 3-axis accelerometer attached to a first place of a building to collect acceleration data;
a control unit for analyzing the collected acceleration data to determine whether the building has collapsed; and
and an alarm generating unit that provides the determination result to a user terminal,
The control unit determines whether the building collapses using an analysis model generated according to a result previously learned based on a deep learning algorithm with respect to a frequency image generated based on a frequency conversion result for the acceleration data,
Building Collapse Prevention System. The method of claim 1,
Further comprising a lidar sensor provided in a second place in the building,
The control unit trains a neural network for the amount of change in the point cloud obtained using the lidar sensor, detects the vibration state of the building based on the learned neural network, generates a warning signal and provides it to the alarm generator,
Building Collapse Prevention System. 3. The method of claim 2,
The second place is a space provided with a pillar for supporting the building in the basement of the building,
The lidar sensor is that a plurality of radially provided around the pillar to obtain a point cloud for the pillar,
Building Collapse Prevention System. The method of claim 1,
The control unit determines the vibration intensity of the building by deriving the average amount of change in the impact driving force from the acceleration data,
The average amount of change in the impact driving force is to be derived by the following [Equation 1],
Building Collapse Prevention System.
[Equation 1]

: Average change in impact thrust force (IMAV) at sample index n

: length of data section

: Scaling parameter

: Change in the amount of impulse for an arbitrary time interval

: acceleration at time n

Images