KR20190046569A - Acoustic Tunnel Accident Detection System - Google Patents

Abstract

An embodiment of the present invention relates to an acoustic-based tunnel accident detection system. Specifically, according to one embodiment of the present invention, the tunnel accident detection system comprises: an analysis unit collecting acoustic data in a tunnel and analyzing the collected acoustic data to detect whether an accident occurs; and an operation unit controlling the tunnel system operation based on the detection result of the analysis unit. The analysis unit may include an acoustic accident sound separation unit separating background noises and accident sounds from the collected acoustic data, and an acoustic accident detection unit detecting whether an accident occurs from the separated acoustic signal.

Description

{Acoustic Tunnel Accident Detection System}

The present invention relates to an acoustic-based incident detection system. More particularly, the present invention relates to a system for classifying and detecting an accident sound in a tunnel environment where various background noises are mixed.

An acoustic-based event detection method means to detect the occurrence of an acoustic event from a recorded sound source and to classify the event according to the characteristics of the sound. In this paper, we propose a new method for detecting acoustic event based on a Gaussian mixed model and a non-singular nonlinear system. Is a method of detecting an acoustic event using a recorded signal.

Generally, there are many image-based accident detection systems in the tunnel. However, it is a reality that many tunnels are erroneously detected due to irregular light coming from the vehicle light with different brightness. Therefore, in the field, though there is an image-based accident detection system, there are cases where it is not actually used.

Hereinafter, the present invention proposes a robust tunnel accident detection method through acoustic analysis to supplement this image-based accident detection system.

The present invention proposes a system capable of accurately detecting whether an accident has occurred in a tunnel by analyzing an acoustic signal.

The tunnel accident detection system according to an embodiment of the present invention includes an analyzer for collecting acoustic data in a tunnel and analyzing the collected acoustic data to detect occurrence of an accident; And an operation unit for controlling the tunnel system operation based on the detection result of the analysis unit. The analysis unit includes an acoustic noise / noise separator for separating the background noise and the accident sound from the collected acoustic data, And an acoustic abnormality detecting unit for detecting an acoustic abnormality.

The results of the acoustic signal analysis and the existing image information can be combined to more accurately determine whether an accident occurred in the tunnel.

1 is a flowchart showing a sound detection method in a general environment.
Fig. 2 shows each step of the sound detection method of Fig. 1 in detail.
3 is a diagram illustrating a step S40 of determining whether an event sound is generated based on a channel gain value in the sound detection method according to the embodiment of the present invention.
4 is a diagram illustrating a step S50 of extracting a feature value for signals discriminated in the sound detection method according to the embodiment of the present invention.
5 is a diagram illustrating a step S60 of verifying likelihood by performing HMM classification on the feature values extracted in the sound detection method according to the embodiment of the present invention.
6 is a block diagram illustrating an acoustic-based tunnel accident detection system according to an embodiment of the present invention.
FIG. 7 is a block diagram illustrating a concrete operation of an acoustic accident sound analysis unit according to an embodiment of the present invention.
8 is a flowchart of an acoustic-based tunnel accident detection method according to an embodiment of the present invention.

The embodiments of the present invention will be described in detail with reference to the accompanying drawings, but the present invention is not limited to these embodiments. In describing the present invention, a detailed description of well-known functions or constructions may be omitted for the sake of clarity of the present invention.

Acoustic signal processing and machine learning algorithms for tunnel acoustic analysis can include high-frequency separation of interest in acoustic in tunnels based on non-sound tensor decomposition and high-frequency classification of acoustic interest in tunnels based on GMM-HMM. The sound detection method will be described in a general environment.

1 is a flowchart showing a sound detection method in a general environment.

Referring to FIG. 1, a method for detecting an acoustic event according to an embodiment includes receiving a stereo input acoustic signal S1, converting a received input signal to a mel-amplitude spectrum signal S2, (S3), a step (S4) of discriminating whether an event sound is generated based on the channel gain value, a step (S5) of extracting a feature value for the discriminated signals, a HMM likelihood verification (S6), and detecting the occurrence of an abnormal event (S7) according to the verification result.

FIG. 2 is a diagram illustrating each step of the sound detection method according to the embodiment of the present invention in detail.

Referring to FIGS. 1 and 2 together, it is assumed that an acoustic event detection method of the present invention is such that, in separating a sound source, an input signal is a signal obtained from a stereo channel. The stereo input signal obtained in step S1 may be expressed by the following equation.

[Equation 1]

는 c번 채널의 e번 분류 음향 사건 신호,

는 c번 채널의 배경 잡음 신호를 나타낸다. 스테레오 입력 신호(

(n))는 E개의 음향 사건이 혼재되어 있는 것으로 가정하였다. i is the frame number, c is the channel number, e is the event classification number, E is the number of event classifications,

Is the e-classification sound event signal of channel c,

Represents the background noise signal of channel c. Stereo input signal (

(n)) are assumed to have a mixture of E acoustic events.

In step S2, the input signal spectrum is obtained from the received input signal y _i ^c (n) as described above. The input signal is subjected to a short-term Fourier transform (STFT) _i ^e (k) |) and this signal is converted to a Mel amplitude spectrum signal (Hz _i ^c (m)).

The mel-amplitude spectrum signal can be expressed by the following equation.

&Quot; (2) "

Where m and M represent the order of mel-amplitude spectrum.

Then, in step S3, a step of decomposing mel-amplitude spectrum into a non-sound tensor is performed. The mel-amplitude spectrum by non-sound hydrostatic resolution can consist of a channel gain, a frequency basis, and a time activation matrix called a tensor. This can be expressed by the following equation.

&Quot; (3) "

는 텐서 프로덕트이며, J는 NTF 분해에서 기저의 순위를 나타낸다. 채널 이득(C), 주파수 기저(B) 및 시간 활성화(A) 매트릭스를 나타내면 다음과 같다. here,

Is the tensor product, and J is the rank of the basis in the NTF decomposition. The channel gain (C), frequency base (B) and time activation (A) matrix are as follows.

: 채널 이득 매트릭스 (2ㅧJ)

: Channel gain matrix (2 ㅧ J)

C _i ^{e, S} , C _i ^D represent the channel gain matrix of the acoustic event and background noise.

: 주파수 이득 매트릭스 (2ㅧ┶)

: Frequency Gain Matrix (2 ㅧ ┶)

┶ represents the base rank of each acoustic event and background noise.

B _i ^{e, S} , and B _i ^D represent the frequency gain matrix of the acoustic event and the background noise.

: 시간 이득 매트릭스 (1ㅧJ)

: Time gain matrix (1 ㅧ J)

A _i ^{e, S} , and A _i ^D represent the time gain matrix of the acoustic event and the background noise.

The channel gain and the time gain may be updated by successive updating rules. The channel gain and time gain can be updated by repeatedly performing the following equation.

&Quot; (4) "

Where h is a repetition factor, and ㅀ represents a multiplication operation. And P ^h _{c, k, m} represent the ratio of Y _i to Hz _i . Can be a percentage value of Y _i for in the update rule of the channel gain as shown in the channel gain, frequency gain and ㎐ _i consideration, Y on the In channel gains, frequency gain and ㎐ _i to update the rules of the time gain _i Can be considered.

Equation (4) can be performed until the relative decrease of the KL divergence becomes smaller than a predetermined threshold value.

The Y _i may be expressed as a sum of each event sound and background sound as follows.

&Quot; (5) "

Then, Y _i can be expressed as a background noise and an abnormal sound event signal by applying a tensor product composed of the respective factors as follows.

&Quot; (6) "

,

,

As shown in Equation (6), the abnormal acoustic event signal and background noise can be decomposed into a combination of channel gain, frequency gain, and time gain tensor, respectively.

3 is a diagram illustrating a step S40 of determining whether an event sound is generated based on a channel gain value in the sound detection method according to the embodiment of the present invention. The acoustic detection method of the present invention divides an input signal transformed into a time-frequency domain into a combination of a channel gain, a time gain, and a tensor with respect to a frequency gain, and in particular, detects an abnormal sound event This is primarily used for

In operation S40, it is determined whether an event sound is generated based on the channel gain value. In operation S40, an average of channel gains for each event sound may be determined. The channel gain of the e-th event can be calculated as:

&Quot; (7) "

는 각 사건 분류마다 대응되는 기저의 숫자를 나타내며, 모든 채널 이득의 평균(C_i)은 다음의 수식으로 나타낼 수 있다. here

Represents the number of bases corresponding to each event classification, and the average (C _i ) of all channel gains can be expressed by the following equation.

&Quot; (8) "

Where J represents the number of all Basis.

In addition, the embodiment first derives the channel gain value and the average of all the channel gains for each sound, and applies the mean-to-max threshold value to the average of the channel gains of the e-th (arbitrary) event . And calculating a ratio value of an e-th channel gain with respect to an average of all channel gains and comparing the ratio value with a predetermined threshold value.

If, in the case larger than the threshold (thr _c) the ratio value is a predetermined acoustic signal is carried over the case of the e-th HMM classification step. If the ratio value is smaller than the predetermined threshold value, the corresponding signal is determined to be a background noise signal, and does not consider the event occurrence decision. The above-described judgment process can be expressed by the following equation.

&Quot; (9) "

If the channel gain ratio of the e-th signal is greater than a predetermined value as shown in Equation (9), it is classified as an acoustic having the event information of the first degree, and this signal can be denoted as Flag _c ^e .

4 is a diagram illustrating a step S50 of extracting a feature value for signals discriminated in the sound detection method according to the embodiment of the present invention.

Referring to FIG. 4, a process of converting signals filtered by channel gain using NTF decomposition to a feature vector composed of MFCCs may be performed to detect an event sound using HMM classification.

That is, an MFCC extraction is performed with reference to a signal (Flag _c ^e ) in which the NTF is performed considering a channel gain, and extracted feature values can be expressed as follows.

&Quot; (10) "

와 같이 구한 후에는, 특징값의 변화값을 구한다. 상기 변화값은 이전 특징값과의 차이 및 이전 특징값의 이전값일 수 있다. Then, the average of the feature values

, The change value of the feature value is obtained. The change value may be a difference from the previous feature value and a previous value of the previous feature value.

이라 할 때, 메인 특징값은

와 같이 표현될 수 있다.That is, the feature value of the i-th sound

, The main feature value is

Can be expressed as

)에 대해 HMM 우도를 검출하는 단계(S60)가 수행될 수 있다. 도 5는 본 발명의 실시예에 따른 음향 검출 방법에서 추출된 특징값에 대한 HMM 분류를 수행하여 우도를 검증하는 S60 단계를 나타낸 도면이다. Subsequently, the characteristic value (

(Step S60) of detecting the HMM likelihood with respect to the HMM may be performed. 5 is a diagram illustrating a step S60 of verifying likelihood by performing HMM classification on the feature values extracted in the sound detection method according to the embodiment of the present invention.

…

) 및 배경 잡음 HMM(

)를 사용할 수 있다. 상기 HMM들을 사용하여 검출된 사건 음향에 대한 우도가 도출될 수 있다. Referring to FIG. 5, in order to classify an HMM, a pre-trained acoustic event HMM

...

) And background noise HMM (

) Can be used. A likelihood for the event sound detected using the HMMs can be derived.

) 및 배경 잡음 HMM(

)는 초기 상태 분배(π), 상태 전이 확률 매트릭스(T), 가우시안 혼합 관찰(θ)을 인자로 포함할 수 있으며,

,

와 같이 나타낼 수 있다.Acoustic event HMM (

) And background noise HMM (

) May include an initial state distribution (?), A state transition probability matrix (T), a Gaussian mixed observation (?) As a factor,

,

As shown in Fig.

배경 잡음에 대한 우도는

과 같이 나타내질 수 있다. A likelihood (L _i ^Flag ) for the event sound and a likelihood (L _i ^D ) for the background noise can be derived by calculating probability values for the HMMs and the feature values. Likelihood of event sound

The likelihood for background noise is

As shown in FIG.

)을 구하고, 이 값을 정규화(normalization)한 값(

)을 구한다. 이 과정을 거치면 가중된 우도 확률은

과 같이 나타내어지며, 이 값을 사용하여 사건 음향의 판별을 수행할 수 있다. Then, the average of the channel gains of the event sounds derived in step S50 (

), And a value obtained by normalizing this value (

). Through this process, the weighted likelihood probability is

And this value can be used to perform the determination of the event sound.

In step S60 of performing HMM classification on the feature values extracted in step S50 to verify the likelihood, step S61 of performing maximum likelihood classification and step S60 of performing a maximum likelihood classification are performed. In step S61, the ratio of likelihoods for event sound to likelihood for background noise And determining whether the event sound is an event sound according to the step S62.

)과 S40 단계에서 얻어진 정규화(normalization)된 채널 이득의 평균값(

)을 받아, 미리 훈련된 복수개의(E개) 사건 음향 HMM으로부터의 데이터(

~

) 및 배경 잡음 HMM으로부터의 데이터(

)를 참조하여 최종적인 사건 음향에 대한 우도 및 배경 잡음에 대한 우도를 구할 수 있다. In step S61, the feature value obtained in step S50

) And the average value of the normalized channel gain obtained in step S40

), Data from a plurality of (E) event sound HMMs previously trained (

~

) And background noise HMM data (

), The likelihood for the final event sound and the likelihood for the background noise can be obtained.

In step S62, the ratio of the likelihood of the event sound to the likelihood (L _i ^D ) of the background noise HMM is calculated. If the value of the likelihood is greater than the predetermined value, it can be determined that the sound signal is a singular event. The expression is as follows.

&Quot; (11) "

)은 e 또는 0을 나타내며 이를 통해 현재 프레임에 이상 상황이 있는지의 여부를 판단할 수 있다. The final resulting value (

) Represents e or 0, and it is possible to judge whether or not there is an abnormal situation in the current frame.

}이 e_i를 나타낼 경우에는 i번째 프레임이 이상 음향을 포함하고 있는 것으로 판단할 수 있다. 그리고 결과값이 0을 나타낼 경우에는 i번째 프레임에 배경 잡음만이 존재하는 것으로 판단할 수 있다. Here, if the reference value thr ^Flagec is a predetermined threshold value, the detected result value {

} In this case represent the _i e, it can be determined that the i-th frame and including at least acoustic. If the result value is 0, it can be determined that only the background noise exists in the i-th frame.

If it is determined that the i-th frame includes the abnormal sound through the comparison of the likelihood with the reference value as described above, it can be determined that an abnormal sound exists in the currently input signal.

6 is a block diagram illustrating an acoustic-based tunnel accident detection system according to an embodiment of the present invention.

6, the acoustic-based tunnel accident detection system 100 according to an exemplary embodiment of the present invention may include an analysis unit 110, a system operation unit 120, and a DB server 130. As shown in FIG. In addition, it may further include an in-tunnel sound collecting unit 200 and an external link system 300.

The analysis unit 110 acquires sound data from the in-tunnel sound collecting unit 220 and analyzes the sound data to detect whether an accident has occurred. The analysis unit 110 may detect occurrence of an accident in the tunnel by using the acoustic event occurrence determination algorithm described with reference to FIG. 1, and a specific configuration will be described below.

The analysis unit 110 may include a sound / accident sound separating unit 111, an acoustic / accident detection unit 112, and a post-processing unit 113.

The acoustic accident sound analysis unit 111 separates the background noise and the accident sound from the acquired sound data. Acoustic data collected in tunnels are mostly acoustics that occur on a daily basis. For example, the sound data collected by the in-tunnel sound collecting unit 200 may include a vehicle engine sound, an accident-related brake sound, a tire friction sound, or a tunnel resonance sound, It can be seen as a background sound. Therefore, it is easy to determine the sound caused by the accident by removing the background noise that occurs on a daily basis.

The acoustic-noise separating unit 111 separates accident sounds by removing background noise from the collected acoustic data using the NTF algorithm in one embodiment. Here, the accident sound is a sound signal having a characteristic different from that of the background noise, and is an acoustic signal that is estimated to be generated by an accident in a tunnel.

The detailed description of the NTF algorithm is as described above and will not be described in detail here. However, in the above description, the background noise becomes a general sound in the tunnel, and in the case of an acoustic event, it becomes an accident sound.

The acoustic noise / noise analysis unit 111 of the present invention performs a real-time tunnel noise model update through an adaptive training method using a tunnel noise model used for separating a background noise and an accident sound. do.

FIG. 7 is a block diagram showing a specific operation of the acoustic noise / sound separator according to the embodiment of the present invention.

As shown in FIG. 7, the acoustic-noise separating unit 111 can separate accident sounds included in the acoustic data using two models, an accident sound model and a tunnel noise model.

Here, the accident sound model is a model of the sound caused by an accident. For example, the accident sound model may be about a vehicle crash source, a tire drag source, an explosion source, or a horn source. In addition, the tunnel noise model may be related to the tunnel background sound in the general situation described above.

In case of accident sound, it has remarkable characteristics, its occurrence frequency is low, and it does not show different characteristics for each tunnel. In general, it is not necessary to update through model adaptation training. On the other hand, in the case of tunnel noise, tunnel characteristics including the shape and position of the tunnel may be changed by the usual sound that occupies most of the acoustic data collected.

Therefore, the noise model used by the acoustic noise / noise separating unit 111 according to the embodiment of the present invention is a noise model that uses the noise spectrum separated by the acoustic noise / noise separating unit 111 as an input value of the noise model, Training. By repeating this operation, a noise model suitable for the background noise that varies depending on the tunnel can be generated.

Returning to FIG. 6 again.

The acoustic accident detector 112 detects the occurrence of an accident from the acoustic signal separated by the acoustic accident sound separator 111. Based on the machine learning, the acoustic accident detection unit 112 detects an accident from an acoustic signal separated into a background sound and an accident sound. In one embodiment, the machine learning performed by the acoustic accident detection unit 112 may be the GMM-HMM based machine learning described above.

Specifically, the acoustic-accident detection unit 112 extracts a feature vector from a separated sound signal, and calculates a maximum likelihood method for comparing the feature value with a previously-trained acoustic model or a tunnel accident sound detected through a network, And verifies the occurrence.

Specifically, the acoustic-accident detection unit 112 compares the ratio of the likelihood of the background noise with the threshold of the accident sound to a threshold value, and when the result is greater than the threshold value of the comparison, it is determined that the sound is generated due to an actual accident.

The post-processing unit 113 synthesizes one or more accident sounds detected by the acoustic-accident detection unit 112 and post-processes the detection result based on the reliability of the detection result.

The system operation unit 120 controls the tunnel system operation based on the analysis result of the analysis unit 110.

Specifically, the system operation unit 120 confirms an accident in connection with the external link system 300, and controls the system operation based on the incident. At this time, the system operation unit 120 can receive the traffic information from the external link system 120, and further can receive the image information on the tunnel situation. The system operation unit can collect the accident detection result, image information and traffic information judged by the sound to finally confirm whether or not the accident has been issued.

Therefore, since the analysis result of sound is added to the system which detected the accident only by the image, it is possible to detect the accident more accurately than the conventional method. In a specific embodiment, if an abnormal vehicle light is detected on the image, but no accident sound is detected in the acoustic signal, the system may determine that no accident has occurred, and vice versa.

In addition, the system operation unit 120 controls the system in the tunnel when an accident is confirmed. In a specific embodiment, the system operation unit 120 may sound an alarm, block entry of a tunnel, or perform an operation to notify a central control center of an accident.

The DB server 130 stores data generated during operation of the system. For example, the DB server 130 may store data related to mapping information, sound data, status information, and accident information. The data stored in the DB server 130 may be used for additional model updates.

Specifically, data of the DB server 130 may be used to design an acoustic model optimized for each tunnel. The data stored in the DB server is data related to the acoustic signals analyzed and determined by the system up to now, and the acoustic accident detection unit 112 collects and refines the acoustic data from the DB server. Then, the acoustic features are extracted based on the refined database and used as input for adaptive training and retraining for the optimized acoustic model design for each tunnel. As a result, the acoustic model of the tunnel accident sound used by the acoustic accident detection unit 112 is optimized for the corresponding tunnel, so that the accident can be detected more accurately.

8 is a flowchart of an acoustic-based tunnel accident detection method according to an embodiment of the present invention.

The tunnel accident detection system collects acoustic data (S101). The tunnel accident detection system collects acoustic data from one or more acoustic collectors. Acoustic data may be for all sounds occurring in the tunnel.

The tunnel accident detection system separates acoustic noise from the collected acoustic data (S103). At this time, the tunnel accident detection system can separate the acoustic data obtained by the NNT algorithm into background noise and accident sound. At this time, a tunnel noise model that is updated in real time can be used.

The tunnel accident detection system confirms whether an accident has occurred based on the separated acoustic signal (S105). Specifically, the tunnel accident detection system extracts feature values from the separated acoustic signals, and applies the maximum likelihood method to the extracted feature values to determine whether the separated accident sounds are acoustic signals collected from actual incidents. In addition, the tunnel accident inspection system considers at least one of traffic information and image information when the acoustic signal is determined to have been collected from an actual accident, and finally confirms whether an accident has occurred.

If it is determined that an accident has occurred, the tunnel accident detection system controls the operation of the system in the tunnel (S107). For example, a tunnel accident detection system can activate an alarm system in a tunnel or block the entry of a tunnel. In addition, the tunnel accident detection system can inform the central control center of the accident occurrence.

The present invention described above can be embodied as computer-readable codes on a medium on which a program is recorded. The computer readable medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of the computer readable medium include a hard disk drive (HDD), a solid state disk (SSD), a silicon disk drive (SDD), a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, , And may also be implemented in the form of a carrier wave (e.g., transmission over the Internet).

Accordingly, the above description should not be construed in a limiting sense in all respects and should be considered illustrative. The scope of the present invention should be determined by rational interpretation of the appended claims, and all changes within the scope of equivalents of the present invention are included in the scope of the present invention.

Claims (7)

An analyzer for collecting acoustic data in the tunnel and analyzing the collected acoustic data to detect occurrence of an accident; And
And an operation unit for controlling the tunnel system operation based on the detection result of the analysis unit,
The analysis unit may further include an acoustic / acoustic noise separating unit for separating background noise and accident sound from the collected acoustic data, and an acoustic / accident detection unit for detecting occurrence of an accident from the separated acoustic signal
Acoustic - based tunnel accident detection system. The method according to claim 1,
The acoustic-noise separating unit separates accident sounds through an accident sound model and a tunnel noise model,
The tunnel noise model receives the noise spectrum separated by the acoustic noise analyzer and receives real-time noise model adaptive training
Acoustic - based tunnel accident detection system. 3. The method of claim 2,
The acoustic-acoustic-noise separating unit separates accident sounds through a non-sound-tensor-based algorithm
Acoustic - based tunnel accident detection system. The method according to claim 1,
The acoustic accident detection unit detects the occurrence of an accident based on the GMM-HMM-based machine learning
Acoustic - based tunnel accident detection system. 5. The method of claim 4,
Detecting the occurrence of an accident from acoustic signals using a maximum likelihood method that extracts a feature vector from a separated acoustic signal and compares the extracted feature value with a previously trained acoustic model or a tunnel accident sound detected through a network
Acoustic - based tunnel accident detection system. 6. The method of claim 5,
The pre-trained acoustic model may be an adaptively trained acoustic model using data previously analyzed and determined for the acoustic signal as the re-
Acoustic - based tunnel accident detection system. The method according to claim 1,
The operating unit
Collects external information including traffic information and image information from the outside, collects the collected external information and the analysis result of the analysis unit,
Acoustic - based tunnel accident detection system.

Images