CN-122024448-A - Vehicle accident event monitoring and early warning system and method based on star flashover

CN122024448ACN 122024448 ACN122024448 ACN 122024448ACN-122024448-A

Abstract

The invention discloses a traffic accident event monitoring and early warning system and method based on star flashover, the system comprises a traffic accident monitoring host, an audio acquisition device, a deceleration warning lamp and an alarm broadcast, wherein all the devices are connected through a star flash communication module. When the system works, the audio acquisition device continuously acquires road environment sound and sends the road environment sound to the host, the host carries out framing, feature extraction and classification recognition on the audio, judges whether braking or collision sound occurs or not, further judges that traffic accidents occur according to continuous or combined acoustic events, and controls the deceleration warning lamp to flash and broadcasts voice to warn the rear vehicle once the traffic accidents are confirmed. The invention gets rid of dependence on visual conditions by utilizing acoustic monitoring, realizes all-weather traffic accident sensing, realizes low power consumption, long distance and reliable networking of multiple devices by star-flash communication, supports road scale deployment, effectively prevents secondary accidents by means of real-time acousto-optic early warning and improves the active protection capability of road traffic safety.

Inventors

CHEN JINQUAN
WANG YUEDONG
GAO XIANG
LI QIUYI
LIU HONGXIN

Assignees

三峡基地发展有限公司通信网络分公司
中国长江三峡集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260226

Claims (10)

1. The traffic accident event monitoring and early warning system based on star flash communication is characterized by comprising a traffic accident monitoring host, an audio acquisition device, a deceleration warning lamp and an alarm broadcast, wherein the traffic accident monitoring host comprises a star flash communication module, an alarm linkage module and a traffic accident identification module; The audio acquisition device is used for acquiring environmental audio data and transmitting the environmental audio data to the traffic accident monitoring host through the star flash communication module; The traffic accident identification module is used for processing and analyzing the received audio data and judging whether traffic accident events occur or not; The alarm linkage module is used for controlling the deceleration warning lamp to flash through the star flash communication module and controlling the warning broadcast to play warning voice when the car accident recognition module judges that a car accident event occurs.
2. The traffic accident event monitoring and early warning system based on star flashover according to claim 1, wherein the star flashover communication module adopts a star flashover SLE protocol to realize communication connection and supports low power consumption, low time delay, multi-device access and long-distance communication.
3. The star-flash-based car accident event monitoring and early warning system of claim 1, wherein the car accident identification module comprises: the audio preprocessing and feature extracting unit is used for framing, filtering, windowing, normalizing and feature extracting the audio data; The audio classifier is used for judging whether the audio is braking sound or vehicle collision sound based on the extracted audio characteristics; And the traffic accident event judging unit is used for judging whether a traffic accident event occurs or not according to the output result of the audio classifier and combining with preset logic.
4. A star-flash-based car accident event monitoring and early warning system according to claim 3, wherein the audio classifier is a classifier trained based on a K nearest neighbor algorithm, and the input is an audio feature vector subjected to dimension reduction processing.
5. A star-flash-based car accident event monitoring and early warning system according to claim 3, wherein the audio preprocessing and feature extraction unit performs the steps of: Carrying out framing treatment on the audio signal, wherein the frame length is 150-300ms, and adjacent frames are overlapped by 50% -75%; Performing median filtering and windowing on each frame of audio; Calculating short-time energy of each frame of audio, and triggering further feature extraction when the short-time energy exceeds the average energy setting multiple of the historical frames; And extracting the Mel frequency cepstrum coefficient or Mel frequency spectrum characteristic of each frame of audio, and carrying out normalization processing.
6. The star-flash-based car accident event monitoring and early warning system according to claim 3, wherein the car accident event judgment unit performs at least one of the following judgment logics: if the braking sound is identified and the braking sound is continuously detected for a plurality of times within the set threshold time, judging that the traffic accident is likely to occur; if the vehicle collision sound is detected within the set threshold time after the brake sound is identified, judging that the vehicle accident is likely to happen; If the vehicle collision sound is recognized and the short-time energy exceeds the set threshold, it is determined that a vehicle accident is likely to occur.
7. The traffic accident event monitoring and early warning system based on star flash as set forth in claim 1, wherein the traffic accident monitoring host is further configured to report traffic accident event information to the traffic management platform through the network after judging that the traffic accident event occurs, and to receive a warning command from the traffic management platform to control the deceleration warning lamp and the warning broadcast to stop working.
8. A method for a star-flash communication based traffic accident monitoring and early warning system according to any one of claims 1 to 7, comprising the steps of: Continuously acquiring road environment audio through an audio acquisition device, and transmitting the road environment audio to a traffic accident monitoring host through star flash communication; framing, filtering and extracting features of the audio data to obtain audio features; classifying the audio characteristics by using a trained audio classifier, and judging whether braking sound or vehicle collision sound is contained; If the classification result is that the brake sound or the vehicle collision sound is contained, determining whether the vehicle accident event occurs or not based on preset vehicle accident event judging logic; if the traffic accident event is confirmed, the deceleration warning lamp is controlled to flash through star flash communication, and the warning broadcast is controlled to play warning voice.
9. The method for a traffic accident event monitoring and early warning system based on star flash communication of claim 8, the method is characterized in that the extracting of the audio features comprises the following steps: Calculating short-time energy of each frame of audio, and triggering Mel frequency spectrum feature extraction when short-time energy mutation is detected; converting each frame of audio into Mel frequency spectrum coefficients, and performing standardization processing; And reducing the Mel frequency spectrum coefficient to a low-dimensional feature vector by using a preset dimension reduction model, and taking the low-dimensional feature vector as the input of the classifier.
10. The method of a star-flash communication based traffic accident event monitoring and early warning system of claim 8, further comprising: after judging that the traffic accident event occurs, reporting event information to a remote traffic management center; And receiving an alarm command from the traffic management center, stopping early warning output and recovering the normal monitoring state.

Description

Vehicle accident event monitoring and early warning system and method based on star flashover Technical Field The invention belongs to the technical field of traffic accident early warning, and particularly relates to a traffic accident event monitoring and early warning system and method based on star flashover. Background In the existing road traffic accident monitoring method, firstly, a collision sensing device is added through a road fence to monitor traffic accident events, secondly, an acceleration sensor is arranged on a vehicle to monitor the traffic accident events, thirdly, the video monitoring is utilized to monitor the traffic accident events, and the image recognition technology is utilized to monitor the traffic accident events, and the three methods have the following defects: The traffic accident event is monitored based on road fence collision, when vehicles collide and do not collide with a fence, the traffic accident event cannot be monitored, an acceleration sensor is arranged on the vehicles to monitor the traffic accident event, part of the vehicles do not have response sensors, the vehicles behind cannot be effectively warned after the traffic accident event occurs, the traffic accident event is monitored through an image recognition technology, video monitoring is required to be carried out on road sections, road monitoring is high in full coverage cost, great in calculation force requirement, not suitable for large-scale deployment and incapable of coping with the traffic accident monitoring requirement under the condition of heavy fog weather, traffic accidents on the front road cannot be timely warned when the traffic accident occurs on the current expressway, and the vehicles behind cannot be timely braked, so that the traffic accident event monitoring and warning system and method based on star flashover are very necessary to solve the technical problems. Disclosure of Invention The invention aims to solve the technical problem of providing a traffic accident event monitoring and early warning system and method based on star flashover, which aims to solve the defects existing in the prior art, so that the monitoring and early warning can be effectively realized under severe weather conditions, and the early warning level is improved. In order to solve the technical problems, the invention adopts the following technical scheme: the traffic accident event monitoring and early warning system based on star flash communication comprises a traffic accident monitoring host, an audio acquisition device, a deceleration warning lamp and an alarm broadcast, wherein the traffic accident monitoring host comprises a star flash communication module, an alarm linkage module and a traffic accident identification module; The audio acquisition device is used for acquiring environmental audio data and transmitting the environmental audio data to the traffic accident monitoring host through the star flash communication module; The traffic accident identification module is used for processing and analyzing the received audio data and judging whether traffic accident events occur or not; The alarm linkage module is used for controlling the deceleration warning lamp to flash through the star flash communication module and controlling the warning broadcast to play warning voice when the car accident recognition module judges that a car accident event occurs. Preferably, the star flashover communication module adopts a star flashover SLE protocol to realize communication connection and supports low power consumption, low time delay, multi-device access and long-distance communication. Preferably, the car accident recognition module includes: the audio preprocessing and feature extracting unit is used for framing, filtering, windowing, normalizing and feature extracting the audio data; The audio classifier is used for judging whether the audio is braking sound or vehicle collision sound based on the extracted audio characteristics; And the traffic accident event judging unit is used for judging whether a traffic accident event occurs or not according to the output result of the audio classifier and combining with preset logic. Preferably, the audio classifier is a classifier trained based on a K nearest neighbor algorithm, and the input of the audio classifier is an audio feature vector subjected to dimension reduction processing. Preferably, the audio preprocessing and feature extraction unit performs the steps of: Carrying out framing treatment on the audio signal, wherein the frame length is 150-300ms, and adjacent frames are overlapped by 50% -75%; Performing median filtering and windowing on each frame of audio; Calculating short-time energy of each frame of audio, and triggering further feature extraction when the short-time energy exceeds the average energy setting multiple of the historical frames; And extracting the Mel frequency cepstrum coefficient or Mel frequency spectrum characteristic of e