WO2020151627A1

WO2020151627A1 - System for detecting multiple-vehicle collision and related method

Info

Publication number: WO2020151627A1
Application number: PCT/CN2020/073051
Authority: WO
Inventors: 蔡海文; 李鲁川; 王照勇; 卢斌; 叶青
Original assignee: 中国科学院上海光学精密机械研究所
Priority date: 2019-01-21
Filing date: 2020-01-19
Publication date: 2020-07-30
Also published as: CN111458059B; CN111458059A

Abstract

The present application relates to the technical field of transportation, and discloses a novel system for detecting a multiple-vehicle collision and a related method. The system comprises: an optical fiber disposed at a roadside; a light source providing an optical signal to the optical fiber; a sensing device for detecting a phase change and/or a frequency shift of the optical signal in the optical fiber; and a collision detection device for identifying, according to the detected phase change and/or frequency shift of the optical signal in the optical fiber, whether a vehicle collision event has occurred. Implementation modes of the present application have the advantages of low costs, passive distribution, magnetic-interference prevention, and precise positioning, while also overcoming various shortcomings in conventional vehicle collision sensors such as complex installation, susceptibility to electromagnetic interference, and poor positioning accuracy.

Description

Continuous vehicle collision detection system and method

Technical field

The present invention relates to the field of traffic technology, in particular to vehicle collision detection technology.

Background technique

With the development of transportation, the problem of road traffic safety has increasingly become a major problem that needs to be solved urgently. Among them, vehicle collision accidents are the most serious accidents in road traffic safety, which will not only cause casualties and property losses; if the accident is not handled in time after the accident, it will pose a potential threat to other normal moving vehicles. Most of the current collision detection technologies use stress sensors and acceleration sensors installed in the car to sense whether a moving vehicle has collided, or establish a wireless sensor network through wireless stress sensors installed in key positions on the road. Notify relevant personnel.

Existing technology one [Wang, Yunpeng, et al. "Vehicle collision warning system and collision detection algorithm based on vehicle infrastructure integration." Advanced Forum on Transportation of China IET, 2012.] This article uses sensors installed in the car to provide information to the vehicle. Short-term trajectory changes are detected. By detecting abnormal trajectory changes when the vehicle is running, it can be judged whether the vehicle has a collision. The problem with the prior art is that it is difficult to require all vehicles to be equipped with sensors, and the cost is high. Even if sensors are installed, it is necessary to send information from the vehicle to the server in a wireless manner, and the in-vehicle wireless communication device may be possible when the collision is serious. Can not work normally, so the practicability and reliability are poor.

Existing technology two [Miranda, J., et al. "A Wireless Sensor Network for collision detection on guardrails." IEEE International Symposium on Industrial Electronics IEEE, 2014.] This article establishes wireless sensors by installing wireless mechanical sensors on the road Network, but the data transmission method of this kind of sensor network is based on radio, which is susceptible to environmental interference. Moreover, wireless mechanical sensors are often point-shaped, discretely distributed on both sides of the road, requiring a higher density to effectively monitor all collisions, and the cost is higher. If the distribution density is low, it is possible that the collision may not necessarily occur at the point where the mechanical sensor is installed, so that the collision may be missed.

Summary of the invention

The purpose of the present invention is to provide a new type of continuous vehicle collision detection system that can detect vehicle collision events occurring along the road in real time, including collisions caused by collisions between vehicles and guardrails, and collisions between vehicles and vehicles.

This application discloses a new type of continuous vehicle collision detection system, including: an optical fiber arranged on the side of the road; a light source for providing optical signals to the optical fiber; and a sensor for detecting the phase change and/or frequency shift of the optical signal in the optical fiber Device; a collision detection device for identifying whether a vehicle collision event has occurred based on the phase change and/or frequency shift of the optical signal in the optical fiber detected by the sensing device.

Wherein, the optical fiber is erected in the anti-collision walls on both sides of the road, or the optical fiber is buried in the ground on the side of the road; or the optical fiber is erected on the guardrail on the side of the road.

In another preferred embodiment, the identifying whether a vehicle collision event has occurred based on the phase change and/or frequency shift of the optical signal in the optical fiber detected by the sensing device further includes:

The phase feature and/or frequency shift feature are extracted from the phase of the optical signal in the optical fiber, and pattern recognition is performed according to the extracted phase feature and/or frequency shift feature.

The phase feature and/or frequency shift feature extracted from the phase of the optical signal in the optical fiber include: Mel cepstrum coefficient, linear prediction cepstrum coefficient, short-term zero-crossing rate feature, and short-term energy feature.

In another preferred example, the method of pattern recognition includes: hidden Markov model, vector quantization clustering, Euclidean distance, and machine learning.

In another preferred example, the sensing device includes: a phase sensitive optical time domain reflectometer, an optical frequency domain reflectometer, a Brillouin optical time domain reflectometer, and a Brillouin optical time domain analyzer, Brillouin Dynamic grating.

In another preferred embodiment, the optical fiber is a core in a communication optical cable.

In another preferred embodiment, the light source and the sensing device are arranged at the same end of the optical fiber.

The present application also discloses a vehicle collision detection method, which includes: inputting an optical signal to an optical fiber arranged on the side of the road through a light source; detecting the phase change and/or frequency shift of the optical signal in the optical fiber; and according to the detected optical fiber The phase change and/or frequency shift of the internal light signal can identify whether a vehicle collision event has occurred.

In another preferred example, using the vehicle collision detection method, the optical fiber is erected in the anti-collision wall on both sides of the road, or the optical fiber is buried in the ground on the side of the road; or the optical fiber rests on the side of the road Fence erection.

The implementation of this application has at least the following advantages:

1. Different from the currently widely used point sensors, the implementation of this application has good continuity and can monitor the road 24 hours a day without interruption.

2. Different from the wireless sensor network constructed by wireless mechanical sensors, the embodiment of this application uses passive sensor fibers as sensors to detect vibrations along the road and has the advantage of high positioning accuracy.

3. The implementation of this application has strong anti-electromagnetic interference capability and can work normally in harsh environments.

4. In the embodiment of the application, the light source and the sensing device can be arranged at the same end of the optical fiber, which is convenient for detection and maintenance.

Description of the drawings

Fig. 1 is a schematic diagram of an overall vehicle collision detection system according to an embodiment of the present application;

Figure 2 is a schematic diagram of the optical cable laying method in Embodiment 1 of the present application;

Figure 3 is a functional block diagram of Embodiment 1 of the present application;

Fig. 4 is a schematic diagram of the optical cable laying method in the second embodiment of the present application;

Fig. 5 is a functional block diagram of the second embodiment of the present application;

FIG. 6 is a schematic flowchart of a vehicle collision detection method according to a second embodiment of the present application.

The numbers marked in the drawings are as follows:

2-1 is an anti-collision wall structure; 2-2 is a communication optical cable; 2-3 is an optical cable core; 4-1 is a sensing optical fiber.

detailed description

In the following description, many technical details are proposed for the reader to better understand this application. However, those of ordinary skill in the art can understand that even without these technical details and various changes and modifications based on the following embodiments, the technical solution claimed in this application can be realized.

In order to make the objectives, technical solutions, and advantages of the present application clearer, the implementation manners of the present application will be described in further detail below in conjunction with the accompanying drawings.

The first embodiment of the present application relates to a continuous vehicle collision detection system. As shown in FIG. 1, the continuous vehicle collision detection system includes:

Optical fiber installed on the side of the road. There are many ways to set up the optical fiber. Optionally, the optical fiber is installed in the anti-collision wall on both sides of the road. Optionally, the optical fiber is buried in the ground on the side of the road. Optionally, the optical fiber is erected on the guardrail on the side of the road.

A light source that provides optical signals to optical fibers. The types of light sources can be varied. Optionally, the light source is a laser light source. Optionally, the light source is an LED light source.

A sensing device used to detect the phase change and/or frequency shift of the optical signal in the optical fiber.

The collision detection device is used to identify whether a vehicle collision event has occurred according to the phase change and/or frequency shift of the optical signal in the optical fiber detected by the sensing device. In one embodiment, the phase feature and/or frequency shift feature are extracted from the phase of the optical signal in the optical fiber, and the pattern recognition is performed based on the extracted phase feature and/or frequency shift feature.

The phase feature and/or frequency shift feature extracted from the phase of the optical signal in the optical fiber can be Mel cepstrum coefficient, linear prediction cepstrum coefficient, short-term zero-crossing rate feature, short-term energy feature, etc.

The method of pattern recognition can be hidden Markov model, vector quantization clustering, Euclidean distance, machine learning, etc.

The sensing device may be a phase sensitive optical time domain reflectometer, an optical frequency domain reflectometer, a Brillouin optical time domain reflectometer, a Brillouin optical time domain analyzer, a Brillouin dynamic grating, and so on.

In one embodiment, the optical fiber is a core in a communication cable.

In one embodiment, the light source and the sensing device are arranged at the same end of the optical fiber. In another embodiment, the light source and the sensing device can also be respectively arranged at both ends of the optical fiber.

The second embodiment of the present application relates to a vehicle collision detection method. The flowchart is shown in FIG. 6, and the method includes the following steps:

In step 601, a light source is used to input an optical signal to the optical fiber erected on the side wall of the road.

In step 602, the phase change and/or frequency shift of the optical signal in the optical fiber are detected. The phase change and/or frequency shift can be detected by the sensing device. The sensing device can be a phase sensitive optical time domain reflectometer, an optical frequency domain reflectometer, a Brillouin optical time domain reflectometer, and a Brillouin optical time domain reflectometer. Domain analyzer, Brillouin dynamic grating, etc.

In step 603, according to the detected phase change and/or frequency shift of the optical signal in the optical fiber, it is identified whether a vehicle collision event has occurred. In one embodiment, the phase feature and/or frequency shift feature are extracted from the phase of the optical signal in the optical fiber, and the pattern recognition is performed based on the extracted phase feature and/or frequency shift feature. The method of pattern recognition can be hidden Markov model, vector quantization clustering, Euclidean distance, machine learning, etc.

In one embodiment, the optical fiber is a core in a communication cable.

In one embodiment, the light source and the sensing device are arranged at the same end of the optical fiber. In another embodiment, the light source and the sensing device can also be respectively arranged at the two ends of the optical fiber.

There are many ways to erect the optical fiber. Optionally, the optical fiber is installed in the anti-collision wall on both sides of the road. Optionally, the optical fiber is buried in the ground on the side of the road. Optionally, the optical fiber is erected on the guardrail on the side of the road.

In order to better understand the technical solution of the present application, the following description will be given with two specific examples. The details listed in the examples are mainly for ease of understanding and are not intended to limit the scope of protection of the present application.

Example one:

Use the existing optical cable laid along the road as the sensing optical fiber, as shown in Figure 2. In the actual road, the anti-collision walls on both sides of the road are shown in Figure 2, and 2-1 is the anti-collision wall structure. Optical cables used for communication are laid in the anti-collision wall, and 2-2 are communication optical cables. When the vehicle collides with the wall, the vibration caused by the collision can be transmitted to the optical fiber. A core 2-3 in the optical cable senses the vibration caused by the collision. The phase-sensitive optical time domain reflectometer detects and demodulates the phase information of the light waves to quantitatively reconstruct the vibration information along the road. Vibration information can be reflected by the phase change and/or frequency shift of the optical signal in the optical fiber. Derivation of the obtained phase information, obtain its Mel cepstrum as the information feature:

1) Pre-emphasis. The sampled digital voice signal s(n) is passed through a high-pass filter to enhance the high-frequency part of the signal and make the frequency spectrum flat. Assuming that the original signal s(n), y[n]=s[n]-μ·s[n-1], μ is the pre-emphasis coefficient.

2) Framing and windowing. Divide the collected digital signal into n segments, that is, n frames, and bring each frame into a window function. The window function used in this example is a Hamming window. Suppose each frame signal after framing is S(n), n=0,1,2,...,N-1; then Y(n)=S(n)×W(n) after adding the Hamming window. The form of W(n) is:

3) Fast Fourier transform is performed on each frame signal after frame division and windowing to obtain the frequency spectrum of each frame, X _b (k)=FFT(y _b (n)).

4) Pass the energy spectrum through a set of Mel-scale triangular filter banks, define a filter bank with M filters, the filter used is a triangular filter, the center frequency is f(m), m=1, 2 ,...,M. The interval between each f(m) widens as the value of m increases. The frequency response of the triangle filter is:

among them

5) Calculate the logarithmic energy output by each filter bank,

6) Discrete cosine transform to get Mel cepstrum

After the Mel cepstrum feature of the collision signal is obtained, the signal feature is trained to establish a hidden Markov model. The training method adopts the Baum-Welch algorithm. After the model is established, for the observation model, the steps are similar to the training method. First, the Mel cepstrum is extracted and the Viterbi algorithm is used to match the training model. Example 1 The entire training and recognition process is shown in Figure 2.

Embodiment two:

In the second embodiment, a total of 5 optical fibers were re-laid on both sides of the road, as shown in Fig. 4, and 4-1 is the re-laid 5 sensing fibers.

In the second example, an optical frequency domain reflectometer is used as a sensor device to obtain road vibration information. Phase demodulation is used to quantify the phase change information caused by vibration. Derivation of the obtained phase information, and then feature extraction. Similar to the voice signal, extract the linear prediction cepstrum information of the collision signal:

1) Pre-emphasis, to enhance the high frequency part of the signal, and make the frequency spectrum flat (same example 1)

2) Framing and windowing (same example 1).

3) Autocorrelation and linear prediction analysis. Calculate the autocorrelation between the frame signals after windowing,

P is the order of linear predictive analysis. After linear prediction analysis, the P-order linear prediction coefficients are obtained, and the linear prediction coefficients are obtained according to the Durbin recursive algorithm:

Wherein (i) represents the i-th iteration, each iteration recalculates a1, a2,...,ai until i=P to end the iteration.

4) Linear prediction coefficient conversion, calculation of linear prediction cepstrum coefficient. Convert from P-order linear prediction coefficients to Q-order cepstral coefficients,

Cepstrum weighting (cepstrum boost) is required during the calculation process,

Where Wm is defined as

After obtaining the linear prediction cepstrum feature of the collision signal, the feature vector is trained, and the vector quantization codebook is designed using the LBG algorithm. Specific implementation process:

1. Use the centroid (mean value) of the feature vectors of all the extracted frames as the first codeword vector.

2. Split the current codebook according to the LBG rule to form n codewords.

3. Classify all the feature vectors according to the obtained codebook, and then calculate the sum of the distortion amount and the relative distortion of the training vector quantization. If the relative distortion is less than a certain threshold, the iteration ends, and the current codebook is the designed M The codebook of the codeword, turn 5. Otherwise, go to the next step.

4. Recalculate the new heart of each area to get a new codebook, go to 3.

5. Repeat steps 2, 3 and 4 until a code book with M code words is formed (M is the required number of code words).

Obtain the feature vector of the observed sample, calculate the average quantization distortion of the sample, and set a threshold. If D is less than this threshold, a collision has occurred, otherwise, it is considered that no collision has occurred. The entire training and recognition process of Example 2 is shown in Figure 5.

It should be noted that the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes no Other elements clearly listed, or also include elements inherent to such processes, methods, articles, or equipment. If there are no more restrictions, the element defined by the phrase "including one" does not exclude the existence of other same elements in the process, method, article, or equipment that includes the element. In the application documents of this patent, if it is mentioned that an act is performed according to a certain element, it means that the act is performed at least according to that element, and it includes two situations: performing the act only according to the element, and according to the element and Other elements perform the behavior. Multiple, multiple, multiple, etc. expressions include two, two, two, and two or more, two or more, and two or more expressions.

All documents mentioned in the present invention are considered to be included in the disclosure of this application as a whole, so that they can be used as a basis for modification when necessary. In addition, it should be understood that after reading the above disclosure of this application, those skilled in the art can make various changes or modifications to this application, and these equivalent forms also fall within the scope of protection claimed by this application.

Claims

A continuous vehicle collision detection system is characterized in that it comprises:

Optical fiber installed on the side of the road;

A light source that provides optical signals to the optical fiber;

A sensing device used to detect the phase change and/or frequency shift of the optical signal in the optical fiber;

The collision detection device is used for identifying whether a vehicle collision event has occurred according to the phase change and/or frequency shift of the optical signal in the optical fiber detected by the sensing device.
The system according to claim 1, wherein the optical fiber is erected in the anti-collision wall on both sides of the road, or,

The optical fiber is buried in the ground on the side of the road; or,

The optical fiber is erected on the guardrail on the side of the road.
The system according to claim 1, wherein:

The identifying whether a vehicle collision event has occurred according to the phase change and/or frequency shift of the optical signal in the optical fiber detected by the sensing device further includes:

The phase feature and/or frequency shift feature are extracted from the phase of the optical signal in the optical fiber, and pattern recognition is performed according to the extracted phase feature and/or frequency shift feature.
The system according to claim 1, wherein the phase feature and/or frequency shift feature extracted from the phase of the optical signal in the optical fiber include: Mel cepstrum coefficient, linear prediction cepstrum coefficient, short-term Zero-crossing rate characteristics, short-term energy characteristics.
The system according to claim 1, wherein the method of pattern recognition includes: hidden Markov model, vector quantization clustering, Euclidean distance, and machine learning.
The system according to claim 1, wherein the sensing device comprises: a phase sensitive optical time domain reflectometer, an optical frequency domain reflectometer, a Brillouin optical time domain reflectometer, and a Brillouin optical time domain reflectometer Domain analyzer, Brillouin dynamic grating.
The system according to any one of claims 1-6, wherein the optical fiber is a core in a communication optical cable.
The system according to any one of claims 1-6, wherein the light source and the sensing device are arranged at the same end of the optical fiber.
A vehicle collision detection method is characterized in that it comprises:

Input the optical signal to the optical fiber arranged on the side of the road through the light source;

Detecting the phase change and/or frequency shift of the optical signal in the optical fiber;

According to the detected phase change and/or frequency shift of the optical signal in the optical fiber, it is recognized whether a vehicle collision event has occurred.
The method according to claim 9, wherein the optical fiber is erected in the anti-collision walls on both sides of the road, or,

The optical fiber is buried in the ground on the side of the road; or,

The optical fiber is erected on the guardrail on the side of the road.