Traffic incident detection method suitable for low visibility of first-grade road

Through the Fourier transform and mutual correlation coefficient calculation method, the problem of traffic event detection in the existing technology under low visibility is solved, and traffic event detection under low visibility and low traffic flow conditions is realized, which improves the reliability of the detection and Accuracy reduces safety risks.

CN118942235APending Publication Date: 2024-11-12SHANGHAI NORMAL UNIVERSITY
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410735900.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-07
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing traffic event detection algorithm is difficult to effectively detect traffic events under low visibility conditions, especially when the traffic flow is not high, which leads to high safety hazards and is limited by visual observation, making it difficult to play a role in low visibility.

Method used

Fourier transform and mutual correlation coefficient calculation method are used to calculate the mutual correlation coefficient and autocorrelation function of the traffic flow signal through the analysis of upstream and downstream traffic sampling signals, and combine time parameter changes to detect traffic events, independent of visual observation and traffic flow Saturation.

Benefits of technology

The traffic event detection under low visibility and low traffic flow conditions is realized, which improves the reliability and accuracy of detection, is not limited to traffic volume, provides traffic event detection methods that are not limited to traffic saturation rate, and reduces safety hazards. .

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942235A_ABST
    Figure CN118942235A_ABST
Patent Text Reader

Abstract

The invention relates to a traffic incident detection method suitable for low visibility of a first-grade road, and the method comprises the steps: determining the positions of upstream and downstream collection points, collecting the speed information of a single vehicle, and recording the upstream and downstream traffic sampling signals; performing Fourier transform on the upstream and downstream traffic sampling signals to obtain upstream and downstream traffic signals; and performing normalization processing on the upstream and downstream traffic signals, calculating a cross correlation coefficient of the upstream and downstream traffic flow signals, verifying the cross correlation coefficient based on the upstream and downstream traffic flow signals, calculating a correlation coefficient at each moment according to the traffic flow signals with a fixed time length, and synthesizing the traffic flow size to obtain the traffic flow. And detecting the traffic event according to the change of the rho m (t) or the tau max. According to the method, the reasonability of converting the traffic flow signals into the discrete time signals is fully demonstrated, traffic flow positioning is carried out by applying correlation, the reliability of correlation calculation in the aspect of representing traffic flow similarity is proved, and the method is still applicable when the traffic volume is too low or too high.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of road section traffic incident detection, and in particular to a traffic incident detection method adapted to low visibility of a first-class highway. Background Art

[0002] Traffic events include accidents, vehicle breakdowns, construction, bad weather and other traffic conditions that affect road capacity, and are the main cause of congestion. Timely detection of traffic events is of great significance for taking measures to alleviate road congestion; at present, with the full coverage of high-definition cameras, domestic and foreign scholars have mainly developed automatic traffic event detection algorithms (AID), such as the classic California algorithm and McMaster algorithm.

[0003] Changes in visibility will have a direct impact on the driver's driving behavior. For example, under the influence of sudden low visibility weather, the driver may not be able to adjust the driving status in time, which may easily cause traffic accidents. Drivers who are in low visibility for a long time are prone to visual fatigue and may engage in dangerous driving behavior due to long-term high mental concentration. At the same time, low visibility will affect the recognition of traffic signs and the maintenance of vehicle speed and distance, affect the identification of line shape and entrances and exits, and lead to reduced traffic capacity and the occurrence of traffic accidents.

[0004] Current typical traffic incident detection algorithms, such as the California algorithm, work on the principle of determining whether a traffic incident has occurred by using the difference in traffic occupancy rates between upstream and downstream sections. However, under low visibility conditions, the field of view of high-definition cameras is greatly limited. The application of such algorithms is also limited by traffic flow saturation. When a traffic incident occurs under low traffic flow conditions and the traffic volume is lower than the capacity after the traffic incident, algorithms such as the California algorithm are difficult to function. Especially under low visibility conditions, the road sections where traffic incidents occur will pose huge safety hazards. Summary of the invention

[0005] The purpose of the present invention is to provide a traffic incident detection method suitable for low visibility on a first-class highway, so as to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A traffic incident detection method adapted to low visibility on a first-class highway, comprising:

[0008] S1. Set the upstream and downstream collection point locations, collect bicycle speed information, and record the upstream and downstream traffic sampling signals;

[0009] S2, performing Fourier transform on the upstream and downstream traffic sampling signals to obtain the upstream and downstream traffic signals;

[0010] S3, normalizing the upstream and downstream traffic signals, calculating the mutual correlation coefficients of the upstream and downstream traffic flow signals and verifying them;

[0011] S4, based on the mutual correlation coefficient of the upstream and downstream traffic flow signals, according to the traffic flow signals of fixed time length, calculate the correlation coefficient at each moment, comprehensively consider the traffic flow size, and calculate the correlation coefficient according to the first function ρ m (t) or time parameter τ max Traffic incidents are detected based on changes in traffic flow.

[0012] Optionally, the traffic sampling signal includes: the speed at each moment of the sampling point, the traffic volume and the corresponding time.

[0013] Optionally, performing Fourier transform on the upstream and downstream traffic sampling signals includes:

[0014] Define the time period and space period, perform period extension on the upstream and downstream traffic flow signals to obtain discrete periodic time signals, perform Fourier transform on the discrete periodic time signals to obtain the upstream and downstream traffic signals, and if the preset threshold is not met, fill the length by zero padding; wherein the traffic sampling signal is a non-periodic discrete time signal.

[0015] Optionally, defining the time period and the space period includes:

[0016] The speed and distance of the entire road section are calculated to obtain the time period; the speed and distance of the upstream and downstream sections are calculated to obtain the spatial period.

[0017] Optionally, calculating the mutual correlation coefficient of the upstream and downstream traffic flow signals includes:

[0018] According to the definition of energy signal, it is determined whether the discrete periodic time signal, i.e., the discrete non-periodic time signal, belongs to the energy signal. The mutual correlation coefficient of the upstream and downstream traffic flow signals is obtained according to the combination of the first property of the energy signal and the second property of the energy signal, as follows:

[0019] R xy (τ)=FFT(W xy (k)) = FFT(X(k)·Y * (k))

[0020] Among them, τ is the time-shift variable, the leading τ in the time-shift direction is positive, and the lagging τ is negative;

[0021] k is the number of the frequency domain signal, ranging from 0 to N-1;

[0022] R xy (τ) is the expression of the mutual correlation coefficient between traffic flow signals x(n) and y(n);

[0023] W xy (k) is the expression of the cross-energy spectral density of traffic flow signals x(n) and y(n);

[0024] Y * (k) represents the conjugate of Y(k).

[0025] Optionally, the first property of the energy signal is:

[0026] W xy (k) = X(k)·Y*(k)

[0027] Among them, Y * (k) represents the conjugate of Y(k); X(k) is the traffic flow signal; Y(k) is the discrete periodic time signal.

[0028] Optionally, the second property of the energy signal is:

[0029] R xy (τ)=FFT(W xy (k))

[0030] Among them, FFT (W xy (k)) is the Fourier transform of the mutual energy spectral density of the traffic flow signals x(n) and y(n).

[0031] Optionally, verifying the mutual correlation coefficients of the upstream and downstream traffic flow signals includes:

[0032] The autocorrelation function expression R of the normalized traffic flow signal x(n) x (τ), and obtain the autocorrelation function denoted by ρ x (τ), based on the first property of the energy signal and the second property of the energy signal, τ=0 obtains the highest correlation.

[0033] Optionally, according to the first function ρ m (t) or time parameter τ max The changes to traffic incident detection include:

[0034] Let natural time be t, and calculate R(τ) at each moment. xy |t, and extract the maximum value R(τ) xy,max , for the maximum value R(τ) xy,max Perform normalization processing to obtain the processing result ρ xy,max , record the corresponding parameter τ max , based on the processing result ρ xy,max and the corresponding parameter τ max , respectively, to obtain the first function ρ m (t) and the second function τm (t), when the traffic flow is lower than the set threshold, the corresponding parameter τ is determined max , whether fluctuation occurs, if fluctuation occurs, it is considered that a traffic incident has occurred, if not, it is considered that no traffic incident has occurred, when the traffic flow is greater than the threshold, the first function ρ is judged m (t), is it lower than the first function ρ at each moment? m (t) is the average value within the natural time t If yes, a traffic incident has occurred; if no, no traffic incident has occurred.

[0035] Optionally, for the maximum value R(τ) xy,max The method for normalization is:

[0036]

[0037] Among them, R(τ) xy,max R(τ) at each moment xy,max The maximum value of .

[0038] The beneficial effects of the present invention are:

[0039] The present invention demonstrates the rationality of converting traffic flow signals into discrete time signals, applies correlation to locate traffic flow, proves the reliability of correlation calculation in characterizing traffic flow similarity, and finally proposes a traffic incident detection method that is not limited to traffic saturation rate and low visibility conditions, providing a reference for studying traffic incident detection methods in local road networks under low visibility conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0041] Figure 1 A flow chart of a traffic incident detection method adapted to low visibility on a first-class highway according to an embodiment of the present invention;

[0042] Figure 2 A four-section traffic flow signal time domain diagram of a traffic incident detection method adapted to low visibility on a first-class highway according to an embodiment of the present invention; wherein: Figure 2 (a) is the velocity amplitude of section number 1, Figure 2 (b) is the velocity amplitude of section number 2, Figure 2 (c) is the velocity amplitude of section number 3, Figure 2(d) is the velocity amplitude of section number 4;

[0043] Figure 3 is the ρ of the traffic flow signal of section 1 and section 4 of a traffic incident detection method adapted to low visibility of a first-class highway in an embodiment of the present invention. 14 (τ) Schematic diagram;

[0044] Figure 4 A schematic diagram of an x(n) updating method of a traffic incident detection method adapted to low visibility on a first-class highway according to an embodiment of the present invention;

[0045] Figure 5 ρ is a traffic incident detection method adapted to low visibility on a first-class highway under different traffic flow conditions according to an embodiment of the present invention. m (t) and τ m (t) time development diagram; where, Figure 5 (a) is the ρ with or without traffic incidents under the condition of 500 vehicles per hour. m (t) development, Figure 5 (b) is the τ with or without traffic incidents under the condition of 500 vehicles per hour. m (t) development, Figure 5 (c) is the ρ with or without traffic incidents under the condition of 1500 vehicles per hour. m (t) development, Figure 5 (d) is the τ with or without traffic incidents under the condition of 1500 vehicles per hour. m (t) development, Figure 5 (e) is the ρ with or without traffic incidents under the condition of 2500 vehicles per hour. m (t) development, Figure 5 (f) is the τ with or without traffic incidents under the condition of 2500 vehicles per hour. m (t) development, Figure 5 (g) is the ρ with or without traffic incidents under the condition of 3500 vehicles per hour. m (t) development, Figure 5 (h) is the τ with or without traffic incidents under the condition of 3500 vehicles per hour m (t) Development. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] The present invention aims at the fact that classic automatic incident detection (AID) algorithms, such as the California algorithm, are difficult to function in low visibility and have limitations in low traffic flow saturation. A traffic incident detection method adapted to low visibility on first-class highways is proposed. The method is not limited by traffic volume and locates traffic flow based on correlation. The reliability of correlation calculation in characterizing traffic flow similarity is proved. Traffic incidents are detected using the cross-correlation properties of discrete time signals. The detection method of the present invention is not restricted by traffic volume and does not rely on visual observation in low visibility.

[0049] like Figure 1 As shown, the method of the present invention first analyzes and processes the traffic flow sampling signal, which is a discrete periodic time signal, that is, a discrete non-periodic time signal. The processed discrete time signal is determined to be an energy signal according to the signal characteristics, and the cross-correlation of the energy signal is used to locate the traffic flow, and the average speed of the traffic flow section is verified to prove the reliability of the correlation calculation in characterizing the similarity of the traffic flow. Finally, a traffic event detection method is proposed. It includes the following steps:

[0050] Step 1: Use coils, speed radars and other speed measuring devices to collect single-vehicle speed information in the upstream and downstream of a closed traffic flow, and take a certain value τ v As a cycle, its amplitude is the average speed of the vehicle in each cycle, and the amplitude is 0 if there is no vehicle in the cycle; the upstream and downstream position and speed information are acquired through the above method, and the upstream and downstream traffic sampling signals are recorded, wherein the upstream and downstream traffic sampling signals include the speed at each moment, the traffic volume and the corresponding time;

[0051] Step 2 performs Fourier transform according to the independent variable form of the time signal x(n) and the signal periodicity. Since the sampling is discrete during traffic event detection, it is easy to judge that the traffic flow sampling signal is a non-periodic discrete time signal.

[0052] Perform Fourier analysis (DFT) on the non-periodic discrete time signal to obtain the upstream and downstream traffic flow signals. The upstream and downstream traffic flow signals themselves and the inverse transformation form are as follows: Figure 4 As shown, where X() represents the frequency domain expression of x(n):

[0053]

[0054] Wherein, k is the number of the frequency domain signal (the value range is 0 to N-1);

[0055] n represents the number of the discrete time-domain signal (the value range is 0 to N-1);

[0056] M represents the number of points of discrete signals in the time domain and signals in the frequency domain;

[0057] However, the DFT algorithm has a large amount of calculation. In order to simplify the algorithm, the FFT method (fast Fourier transform) is used for Fourier transform. In order to make the samples meet the applicable conditions of the FFT method, the processing method is to periodically extend x(n). The FFT requirement is that the signal length of x(n) is a power of 2. If the signal length is not a power of 2, it cannot be used. Therefore, when the signal length of x(n) does not meet the power of 2, it is padded with zeros.

[0058] Step 3 uses the correlation function to quantify the correlation between the two sets of signals x(n) and y(n), and derives the cross-correlation expression of the discrete non-periodic time signal belonging to the energy signal;

[0059] According to the nature of energy signals, the correlation function is the Fourier transform of energy spectrum density; the cross-correlation expression of traffic flow signals is derived as follows:

[0060] The upstream and downstream traffic flow signals are energy signals. One of the properties of energy signals is that for energy signals, the signal energy calculated in the time domain is equal to the signal energy calculated in the frequency domain, that is:

[0061] W x (k)=|X(k)| 2

[0062] The above formula is called the energy spectral density expression, which is also valid when extended to two different groups of signals. That is, the mutual energy spectral density expression corresponding to the two groups of signals is obtained as follows:

[0063] W xy (k) = X(k)·Y * (k)

[0064] where Y * (k) represents the conjugate of Y(k).

[0065] The correlation function is the Fourier transform of the energy spectral density, that is, the cross-correlation calculation method is as follows:

[0066] R(τ) xy =∑x(n)y(n+τ)

[0067] Substitute the cross-correlation calculation formula to obtain the autocorrelation function expression R of the traffic flow signal x(n): x The cross-correlation expression R between (τ) and traffic flow signal xy (τ):

[0068] Rx (τ)=FFT(W x (k)) = FFT(|X(k)| 2 )

[0069] R xy (τ)=FFT(W xy (k)) = FFT(X(k)·Y * (k))

[0070] Among them, τ is the time-shift variable, the leading τ in the time-shift direction is positive, and the lagging τ is negative;

[0071] n is the number of the discrete time domain signal, ranging from 0 to N-1;

[0072] k is the number of the frequency domain signal, ranging from 0 to N-1;

[0073] R x (τ) is the autocorrelation function expression of the traffic flow signal x(n);

[0074] R xy (τ) is the expression of the mutual correlation coefficient between traffic flow signals x(n) and y(n);

[0075] W x (k) is the energy spectrum density expression of the traffic flow signal x(n);

[0076] W xy (k) is the expression of the cross-energy spectral density of traffic flow signals x(n) and y(n);

[0077] Y*(k) represents the conjugate of Y(k).

[0078] Normalization R x (τ), the processed autocorrelation function is recorded as ρ x (τ);

[0079] Using the autocorrelation function properties of the energy signal, ρ x (τ) reaches its maximum value at τ = 0, that is, when the signal completely overlaps with itself, the highest correlation can be obtained. When the two signals are exactly the same, ρ x (0) = 1, ρ x (τ) reaches its maximum value when τ = 0, indicating that the highest correlation can be obtained when the signal completely overlaps itself.

[0080] like Figure 5 As shown in (a)-(h), the basis for judging the occurrence of an accident is the cross-correlation of traffic flow signals on two sections with natural time t. The judgment method needs to calculate the size of the cross-correlation coefficient of the two sets of traffic flow signals. The specific judgment process is as follows:

[0081] Calculation preparation: Let natural time be t, and calculate the correlation coefficient R(τ) at each moment. xy |t, and extract the maximum value R(τ) xy,max , normalized, and obtained ρ xy,max , record the corresponding parameter τ max , based on the ρ xy,max and the τ max , respectively, to obtain the first function ρ m (t) and the second function τ m (t);

[0082] Judgment basis: When the AC flow rate is lower than the set threshold, the τ max , whether fluctuation occurs, if fluctuation occurs, it is considered that a traffic incident has occurred, if not, it is considered that no traffic incident has occurred, when the traffic flow is greater than the threshold, it is determined that the ρ m (t) is lower than the average value. If yes, a traffic incident has occurred; if no, no traffic incident has occurred.

[0083] like Figure 2 As shown in (a)-(d), in order to more clearly prove the reliability of the above method, four cross-section traffic flow data on a circular curve of G109 are taken, and the vehicle speed of each section is counted. The traffic flow speed of the four sections is shown in Table 1.

[0084] Table 1

[0085] Section number 1 2 3 4 Vehicle speed / (km / h) 64.1 63.9 54.0 55.8

[0086] According to the traffic flow size, the sampling period t is taken as 8.6s, and it is converted into four groups of traffic flow signals x1(n)~x4(n) according to the method described in step 2; the time domain diagrams of the traffic flow signals of the four sections are drawn for comparison, and the results show that the four groups of signals are indeed similar.

[0087] τ when the maximum correlation coefficient is obtained by statistically analyzing different cross-section combinations. Figure 3 Analysis, at τ = -3 ρ 14 (τ) obtains an obvious maximum value, indicating that when x(n) is shifted back by 3 places, x1(n) and x4(n) are most similar, and it can be determined that the same group of traffic flows arrives at section 4 from section 1.

[0088] Calculate the estimated speed of the road section and compare it with the collected vehicle speeds of each section. Count the τ values ​​when the maximum correlation coefficient is obtained for different section combinations, and calculate the estimated speed of the road section based on time and distance. The estimated traffic flow speed between the four sections is shown in Table 2.

[0089] Table 2

[0090] Section number 1 2 3 4 1 - 55.5~110.9 59.3~89.0 59.1~78.8 2 55.5~110.9 - 33.5~67.0 41.9~628 3 59.3~89.0 33.5~67.0 - 29.3~58.6 4 59.1~78.8 41.9~62.8 29.3~58.6 -

[0091] Compared with Table 1, although the calculated speed estimation range is large, it is basically consistent with the fixed-point vehicle speed of the cross section, which confirms the following conclusions:

[0092] Calculating the maximum value of the mutual correlation coefficient can indicate how similar the arrangement of the same group of traffic flows is when they pass through the upstream and downstream sections.

[0093] τ corresponding to the maximum value of the mutual correlation coefficient max Characterizes the time it takes for the same group of traffic flows to go from the upstream section to the downstream section.

[0094] Step 4 calculates the cross-correlation of traffic flow signals on the two sections over natural time t.

[0095] Since the traffic flow signal may be unstable, a fixed length of traffic flow signal x(n) is taken for analysis in each section;

[0096] The update method of x(n) is to develop over time, adding the latest signal and replacing the oldest signal.

[0097] Let natural time be t, and calculate R(τ) at each moment. xy |t, the maximum value of the calculation result is recorded as R(τ) xy,max , normalize it and record it as ρ xy,max , the corresponding parameter is denoted as τ max .

[0098] According to the nature of traffic flow signals, under normal driving conditions, R(τ) xy,max Will be able to maintain a certain level in the time dimension, and τ max The fluctuation will not be big.

[0099] When a traffic incident occurs between the upstream and downstream sections, the similarity of the two sets of traffic flow signals will decrease, R(τ) xy,max will decrease rapidly, τ max is stretched, that is, R(τ) xy,max and τ max There will be varying degrees of vibration.

[0100] To obtain regular results, normalize ρ xy,max and τ max , and ρ xy,max and τ max The function over time is denoted as ρ m (t) and τ m (t);

[0101] By comprehensively considering ρ m (t) and τ m(t) The development on the time axis determines whether a traffic incident occurs between the two corresponding sections.

[0102] In order to more clearly illustrate how the above method detects traffic incidents, Vissim is used to simulate and collect traffic data. The simulation conditions are set to a one-way two-lane highway. The upstream and downstream sections of the collected data are 1 km apart, and the traffic development is simulated for 6 hours each time. Four traffic flows are set, namely 500veh / h, 1500veh / h, 2500veh / h and 3500veh / h. Each traffic flow is divided into two situations: no traffic incident and traffic incident. The traffic incident occurs around the 2.5th hour.

[0103] Follow steps 2 to 4 to calculate and draw ρ m (t) and τ m (t) Development graph at time t.

[0104] The results show that when the traffic volume is low, ρ m (t) does not change significantly, but the reduction in traffic capacity has a significant impact on τ m (t) significant impact;

[0105] As the traffic volume increases, traffic events have an impact on ρ m (t) has an increasingly significant impact, but τ m The change of (t) becomes vague.

[0106] Therefore, the traffic flow size is comprehensively considered and different discrimination indicators are used to detect traffic incidents;

[0107] The following table is used as the basis for dividing high and low flow rates; under low flow rates, when τ max Fluctuations are considered to have occurred; under high traffic conditions, when ρ m When (t) is lower than the average value, it is considered that a traffic incident has occurred. The classification of high and low traffic in different situations is shown in Table 3. Table 3 is the classification standard of high and low traffic states.

[0108] Table 3

[0109] Traffic status Traffic Low Flow ≤1500veh / h High flow >1500veh / h

[0110] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A traffic incident detection method adapted to low visibility on a first-class highway, characterized in that: include: S1. Set the upstream and downstream collection point locations, collect bicycle speed information, and record the upstream and downstream traffic sampling signals; S2, performing Fourier transform on the upstream and downstream traffic sampling signals to obtain the upstream and downstream traffic signals; S3, normalizing the upstream and downstream traffic signals, calculating the mutual correlation coefficients of the upstream and downstream traffic flow signals and verifying them; S4, based on the mutual correlation coefficient of the upstream and downstream traffic flow signals, according to the traffic flow signals of fixed time length, calculate the correlation coefficient at each moment, comprehensively consider the traffic flow size, and calculate the correlation coefficient according to the first function ρ m (t) or time parameter τ max Traffic incidents are detected based on changes in traffic flow.

2. A traffic incident detection method adapted to low visibility on a first-class highway according to claim 1, characterized in that: The traffic sampling signal includes: the speed at each sampling point, the traffic volume and the corresponding time.

3. The method for detecting traffic incidents in low visibility conditions on a first-class highway according to claim 1, characterized in that: Performing Fourier transform on the upstream and downstream traffic sampling signals, including: Define the time period and space period, perform period extension on the upstream and downstream traffic flow signals to obtain discrete periodic time signals, perform Fourier transform on the discrete periodic time signals to obtain the upstream and downstream traffic signals, and if the preset threshold is not met, fill the length by zero padding; wherein the traffic sampling signal is a non-periodic discrete time signal.

4. A traffic incident detection method adapted to low visibility on a first-class highway according to claim 3, characterized in that: Defining the time period and the space period includes: The speed and distance of the entire road section are calculated to obtain the time period; the speed and distance of the upstream and downstream sections are calculated to obtain the spatial period.

5. The method for detecting traffic incidents in low visibility conditions on a first-class highway according to claim 1, characterized in that: Calculating the mutual correlation coefficients of the upstream and downstream traffic flow signals, including: According to the definition of energy signal, it is determined whether the discrete periodic time signal, i.e., the discrete non-periodic time signal, belongs to the energy signal. The mutual correlation coefficient of the upstream and downstream traffic flow signals is obtained according to the combination of the first property of the energy signal and the second property of the energy signal, as follows: R xy (τ)=FFT(W xy (k))=FFT(X(k)·Y * (k)) Among them, τ is the time-shift variable, the leading τ in the time-shift direction is positive, and the lagging τ is negative; k is the number of the frequency domain signal, ranging from 0 to N-1; R xy (τ) is the expression of the mutual correlation coefficient between traffic flow signals x(n) and y(n); W xy (k) is the expression of the cross-energy spectral density of traffic flow signals x(n) and y(n); Y * (k) represents the conjugate of Y(k).

6. According to the method for detecting traffic incidents in low visibility on a first-class highway according to claim 5, the first property of the energy signal is: W xy (k) = X(k)·Y * (k) in, Y * (k) represents the conjugate of Y(k); X(k) is the traffic flow signal; Y(k) is the discrete periodic time signal.

7. According to the method for detecting traffic incidents in low visibility on a first-class highway according to claim 6, the second property of the energy signal is: R xy (τ)=FFT(W xy (k)) in, FFT(W xy (k)) is the Fourier transform of the mutual energy spectral density of the traffic flow signals x(n) and y(n).

8. The method for detecting traffic incidents in low visibility conditions on a first-class highway according to claim 5, characterized in that: Verifying the mutual correlation coefficients of the upstream and downstream traffic flow signals, including: The autocorrelation function expression R of the normalized traffic flow signal x(n) x (τ), and obtain the autocorrelation function denoted by ρ x (τ), based on the first property of the energy signal and the second property of the energy signal, τ=0 obtains the highest correlation.

9. The method for detecting traffic incidents in low visibility conditions on a first-class highway according to claim 1, characterized in that: According to the first function ρ m (t) or time parameter τ max The changes to traffic incident detection include: Let natural time be t, and calculate R(τ) at each moment. xy |t, and extract the maximum value R(τ) xy,max , for the maximum value R(τ) xy,max Perform normalization processing to obtain the processing result ρ xy,max , record the corresponding parameter τ max , based on the processing result ρ xy,max and the corresponding parameter τ max , respectively, to obtain the first function ρ m (t) and the second function τ m (t), when the traffic flow is lower than the set threshold, the corresponding parameter τ is determined max , whether fluctuation occurs, if fluctuation occurs, it is considered that a traffic incident has occurred, if not, it is considered that no traffic incident has occurred, when the traffic flow is greater than the threshold, the first function ρ is judged m (t), is it lower than the first function ρ at each moment? m (t) is the average value within the natural time t If yes, a traffic incident has occurred; if no, no traffic incident has occurred.

10. The method for detecting traffic incidents in low visibility conditions on a first-class highway according to claim 8, characterized in that: For the maximum value R(τ) xy,max The method for normalization is: Among them, R(τ) xy,max R(τ) at each moment xy,max The maximum value of .

Citation Information

Cited By

  • Discriminating method for judging weak vital signs based on infrared pair tube illumination technology

    CN121971062A