CN110942670A - Expressway fog area induction method - Google Patents
Expressway fog area induction method Download PDFInfo
- Publication number
- CN110942670A CN110942670A CN201911153267.3A CN201911153267A CN110942670A CN 110942670 A CN110942670 A CN 110942670A CN 201911153267 A CN201911153267 A CN 201911153267A CN 110942670 A CN110942670 A CN 110942670A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- audio
- target
- audio signal
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/16—Anti-collision systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a highway fog zone guidance method, which adopts an audio monitoring unit to collect sound from a monitoring zone and send the sound to a target identification unit, the target identification unit detects whether an audio signal comprises an audio target to be detected, if the identification result of the target identification unit comprises alarm content, an alarm signal unit sends out alarm information, an event identification unit obtains the moving speed and the speed change of the target by comprehensively analyzing identification data transmitted by each target identification unit, and then the vehicle driving condition is detected and identified, and the alarm signal is sent to the alarm signal unit to send out a corresponding alarm signal. The invention can overcome the difficulty of poor visibility in rainy and foggy weather, correctly identify the vehicle target and dangerous events, prevent the equipment from being influenced by natural phenomena such as rain, fog, sand and dust and the like, and avoid the functional failure of the equipment caused by the deviation of the installation position of the equipment due to environmental vibration, windy weather and the like.
Description
Technical Field
The invention discloses a highway fog zone induction method, and belongs to the field of intelligent traffic.
Background
The visibility of the highway is low due to rain and fog weather, and traffic accidents are easily caused because the front vehicles cannot be seen clearly.
The invention patent of a group fog monitoring and early warning system based on a Beidou satellite navigation system (publication number: 109523811A) discloses a method, which comprises six subsystems and four modules; the Beidou subsystem is connected with the ground monitoring module through the information transmission module, the Beidou subsystem and the ground monitoring module are connected with the local data storage subsystem through the information transmission module, the local data storage subsystem is connected with the manual reinspection subsystem through the information transmission module, the manual reinspection subsystem is respectively connected with the highway traffic management platform and the self-regulation subsystem through the information transmission module, the highway traffic management platform is respectively connected with the vehicle induction module and the local data storage subsystem through the information transmission module, the self-regulation subsystem is connected with the local data storage subsystem through the information transmission module, and the early warning subsystem is connected with the receiving module through the information transmission module. The system effectively reduces traffic hazards caused by the mist, has the characteristics of stability, accuracy, timeliness and reliability, and provides guarantee for the safety of vehicles running in the mist area. The method needs a complex vehicle networking system, and has huge construction, implementation and maintenance cost.
The invention discloses a laser ranging-based fog area induction early warning system (publication number: 107680411A), which comprises a core control module and a plurality of terminal guide unit modules, wherein the core control module is respectively connected with each terminal guide unit module through a transmission optical fiber, the terminal guide unit modules measure visibility information near the terminal guide unit modules, vehicle position information near the terminal guide unit modules is obtained through a high-frequency scanning test, and when the distance between two vehicles is smaller than a threshold value L, a flashing light group is triggered to carry out early warning. According to the technical scheme provided by the invention, a road section is subjected to information control, information sending and information processing through a core control module, the vehicle positions of the road section and the distances between vehicles are monitored in real time, and the optimal triggering time, the optimal flashing frequency and the optimal flashing brightness are calculated according to the visibility value. Therefore, the technical scheme provided by the invention has higher practicability, can improve the utilization rate of resources and simultaneously reduces the attenuation of the lamp group. The method has the advantages of complex networking, expensive laser ranging cost of the core technology, low target identification accuracy and higher difficulty in popularization, application and implementation.
The invention discloses an intelligent recognition warning and safety warning system for a vehicle running track in a low visibility environment (an authorization notice number: 104008663B), which comprises a vehicle track sensing system, a color lamp induction system, an abnormal event monitoring system, an emergency rescue help-seeking system and the like; the invention arranges a red and yellow light induction system at the roadside, adopts laser, infrared and microwave sensing, can intelligently track and identify the position of passing vehicles in a low-visibility environment through comprehensive sensing and studying, calculates the distance between vehicles in front through information delivery and regional linkage modes, starts the red and yellow light induction system according to the calculation result, warns and reminds following vehicles in a mode of combining multicolor stroboscopic and electronic guideboards, and achieves the purpose of reducing the occurrence of malignant interlinked rear-end collisions. The traditional laser, infrared and microwave sensors are adopted in the patent, the implementation and maintenance cost is high, the equipment reliability is insufficient under the long-term field working condition, and the device is not suitable for large-scale deployment and application.
The invention patent of a traffic danger early warning method and system of a smart phone based on vehicle noise (application publication No. CN 106448050A) discloses a method, based on the smart phone, comprising a data acquisition module, a feature extraction module, a central processing unit, a background database, a control module and an early warning module, establishing a matching relation and a model by utilizing the collected training samples in a background database, collecting a vehicle noise data set in real time through a smart phone, extracting the characteristics of the vehicle noise data set, calling the matching relation and the model stored in the background database to output a result, and the result is analyzed and judged, the smart phone is controlled to trigger the primary early warning and the emergency early warning, the traveling crowd with hearing disorder and drowning while walking and listening to mobile phones with earphones is timely reminded, the safety of road travelers is improved, and the accident rate on the road is effectively reduced.
The patent uses a smart phone as a hardware carrier, and moves along with a holder of the smart phone, so that the following main problems exist: and can not be used for vehicle-mounted early warning. This patent is based on the smart mobile phone and is done the hardware carrier, under the on-vehicle condition, the sound of other vehicles has been covered to car self noise, can't solve the induced problem in highway fog zone at other vehicles of perception under the highway environment.
Pickup technology is not feasible. This patent adopts the cell-phone to gather environment vehicle sound, and cell-phone pickup microphone possesses the ambient noise elimination mechanism, under the conventional conditions, can't gather the car noise beyond ten meters for the vehicle early warning that normally traveles is not feasible.
Long-distance early warning cannot be realized. This patent is based on smart mobile phone and is done the hardware carrier, and sound collection unit and warning display unit integration are on same cell-phone, see that the alarm signal unit on the cell-phone is too close with pickup unit distance, can't realize remote early warning.
The network technology can not be utilized to identify and early warn the behaviors of slow running, parking, reverse running, traffic accidents and the like. Since the population of cell phone users is substantially distributed in two dimensions, the patent does not provide the ability to analyze the dangerous events using the information provided by the tandem location smart phones via network technology.
The method needs the support of a background database, has large time lag in vehicle target identification, and is not suitable for large-scale deployment along the highway as a real-time early warning system.
At present, in order to solve the problem of highway fog zone induction, a fog zone induction device based on an infrared correlation technology is widely adopted for a highway at present. After the infrared correlation sensor finds the passing vehicle, the LED warning lamp corresponding to the roadside is triggered to warn the vehicle behind to decelerate. Because the infrared correlation device easily pollutes and shelters from the light path because of factors such as weather, passing vehicle under the field condition, also easily because the operation such as clean maintenance causes the light path skew, causes the induction system and can not normally work, the actual use effect is not ideal.
Some new techniques have also been evaluated experimentally. Based on thermal imaging camera device, usable video target detection technique detects the passing vehicle, replaces infrared correlation device. Thermal imaging devices suffer from two major drawbacks: (1) the penetration distance of the thermal imaging device is three times that of a common visible light imaging device, and in heavy fog weather, the target in the working distance range still cannot be seen clearly; (2) the thermal imaging device is expensive in cost, and is deployed along the full coverage of the highway, so that the cost is high. The method detects the passing vehicles based on the radar technology, and aims at the identification of the road surface target, and has the defects of short detection distance, high false alarm rate, high cost and the like.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a highway fog area induction method, which can overcome the defects that the visibility in rainy and foggy weather is poor, the vehicle target and dangerous events are difficult to correctly identify, equipment is not easily influenced by natural phenomena such as rain, fog, sand and dust, and the like, and the function failure of the equipment caused by the deviation of the installation position of the equipment due to environmental vibration, strong wind weather and the like.
In order to solve the technical problem, the technical scheme adopted by the invention is as follows: an expressway fog area induction method comprises the following steps: s01), taking a section of highway as a monitoring interval, installing more than one audio monitoring unit at one side of the highway in the monitoring interval, wherein the audio monitoring units are responsible for collecting the sound from the monitoring interval and sending the collected audio signals to a target identification unit; s02), the target identification unit detects whether the audio signal includes an audio target to be detected, the audio target to be detected is sound related to road traffic, and the type of the audio target is determined by extracting sound frequency spectrum characteristics; s03), if the target identification unit identification result contains the warning content, sending an instruction to the warning signal unit, and sending warning information by the warning signal unit according to the warning type; s04), the target recognition unit sends the recognized target information to the event recognition unit, the moving speed and the speed change of the target are obtained by comprehensively analyzing the recognition data sent by each target recognition unit, and then the vehicle driving condition is detected and recognized, and the warning signal is sent to the warning signal unit to send out the corresponding warning signal.
Further, the target identification unit extracts audio signal features through the audio signal zero crossing rate to determine the audio target type, and the audio signal zero crossing rate calculation formula is as follows:
wherein N is1Is the number of sampling points of a frame, m is a sampling point, x (m) is the audio signal amplitude of the current sampling point, x (m-1) is the audio signal amplitude of the previous sampling point, sgn [, ]]In order to be a function of the sign,
furthermore, the target identification unit extracts the audio signal characteristics through the short-time energy of the audio signal to determine the type of the audio signal, and the short-time energy calculation formula of the audio signal is as follows:
N2is the window length, m is the sampling point, n is the current pixel point, x (m) is the amplitude of the audio signal, w (n-m) is the weight of the audio signal, the visible short-time energy is the weighted sum of squares of the sample values of a frame, when the window function is a rectangular window,
further, the target identification unit extracts audio signal features through an audio signal short-time autocorrelation function and determines the type of the audio target, and the specific steps are as follows:
x (m) is the audio signal amplitude at the current sample point, and x (m + k) is the audio signal amplitude after k sample points.
Further, audio signal features are extracted through a convolutional neural network, and the type of the audio target is determined, which specifically comprises the following steps: s21), collecting training samples, wherein the training samples comprise vehicle noise data, vehicle types, vehicle speeds and distances between the vehicles and the audio monitoring units, the training samples are vehicle noises corresponding to k vehicle types under the conditions of p vehicle speeds and g distances between the vehicles and the audio monitoring units, and the vehicle noises comprise vehicle noise intensity values and vehicle noise audio values; s22), under the condition that the vehicle is moving, extracting and marking common vehicle noise audio frequency spectrum characteristics respectively corresponding to K vehicle types at different driving speeds and at different distances from the audio monitoring equipment; s23), classifying different vehicle noises and environmental noises after the collection marking is completed, wherein the input layer is different types of noises after the marking is completed, and the output layer is the probability of different types of noises.
Further, the vehicle speed sample value, the distance sample value between the vehicle and the audio monitoring unit and the collected corresponding vehicle noise data are stored in a classification matrix form in the subfile of the vehicle type.
Further, the step of identifying the vehicle driving condition by the event identification unit is as follows: s41), identifying the vehicle type; s42), c is the propagation speed of sound in air, v is the speed of automobile, f1Is the fundamental frequency of the incoming noise of the vehicle, f2Is the noise fundamental frequency of the vehicle leaving
S43), if v<If 0 or v is 0, the vehicle is judged to be stopped or driven backwards; s44), let t be S/v, S be the distance between two audio monitoring units, v be a speed value set for detecting the running state of the vehicle, d be half the length of the detection time interval, if the event identification unit receives the time interval [ t-d, t + d ] after the k-th target identification unit signal]And receiving the signal of the (k + 1) th target identification unit, and judging that the vehicle stops or slowly runs between the (k) and the (k + 1) th target units.
Further, the audio frequency target detected by the target identification unit comprises automobile passing, automobile static idling, automobile brake, traffic accident, falling object, natural disaster, invading person and animal.
Further, the audio monitoring unit adopts a directional sound pickup to collect sound from a specific direction of a road.
Further, the audio monitoring unit adopts a ring-shaped directional sound pickup to collect the sound from the road.
The invention has the beneficial effects that: by using the audio monitoring and identifying technology, the method is not influenced by rain and fog weather, and can provide more accurate identifying effect than infrared, laser, microwave and other methods;
the equipment design and production cost based on the audio monitoring and recognition technology are cheaper in various optical technology-based schemes in the industry at present, and the method is suitable for large-scale popularization and use;
the equipment can independently work in a non-networking state, and is convenient and quick to install, maintain and disassemble.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a schematic structural diagram of the constraint in the Convolutional neural network described in embodiment 1.
Detailed Description
The invention is further described with reference to the following figures and specific embodiments.
Example 1
The embodiment discloses a highway fog zone induction method, which is used for detecting and judging whether vehicles pass through according to a field audio signal, and further sending out a warning to the following vehicles through a warning device at a position of a warning distance behind a monitoring device, so that the problems of poor reliability and high implementation and maintenance cost of an identification technology based on an optical path when visibility is insufficient in rainy and foggy weather are solved.
As shown in the flowchart of fig. 1, the method of this embodiment is implemented based on an audio monitoring unit, a target identification unit, an alert signal unit, a network bus, and an event identification unit, and specifically includes the following steps:
s01), taking a section of highway as a monitoring interval, installing more than one audio monitoring unit at one side of the highway in the monitoring interval, wherein the audio monitoring units are responsible for collecting the sound from the monitoring interval and sending the collected audio signals to a target identification unit;
in this embodiment, the audio monitoring unit adopts a directional sound pickup to collect sounds from a specific direction of a road, and in other embodiments, an annular sound pickup microphone array can be used to collect sounds from various directions of the road according to actual conditions, or other audio collecting components are used to collect live audio.
S02), the target identification unit detects whether the audio signal includes an audio target to be detected, the audio target to be detected is sound related to road traffic, and the type of the audio target is determined by extracting sound frequency spectrum characteristics;
in this embodiment, sounds related to road traffic, such as passing of an automobile, static idling of the automobile, braking of the automobile, traffic accidents, falling objects, natural disasters, intruders, animals, and the like, are used as detection targets, and specifically, the audio frequency is determined in the following modes:
mode 1: extracting the spectral characteristics of the automobile sound through the zero crossing rate of the audio signal, determining the type of the audio target, and calculating the zero crossing rate of the audio signal according to the following formula:
wherein N is1Is the number of sampling points of a frame, m is a sampling point, x (m) is the audio signal amplitude of the current sampling point, x (m-1) is the audio signal amplitude of the previous sampling point, sgn [, ]]In order to be a function of the sign,
when the automobile passes through the audio monitoring unit, the zero-crossing rate of the audio signal is subjected to a process of rapidly reducing from a high point to a low point, and the characteristic can be used as a judgment basis for the passing of the automobile.
Mode 2: extracting the spectral characteristics of the automobile sound by using the short-time energy of the audio signal, determining the type of the audio target, and calculating the short-time energy of the audio signal according to the following formula:
wherein N is2Is the window length, m is the sampling point, n is the current pixel point, x (m) is the amplitude of the audio signal, w (n-m) is the weight of the audio signal, the visible short-time energy is the weighted sum of squares of the sample values of a frame, when the window function is a rectangular window,
when the automobile passes through the audio monitoring unit, the short-time energy of the audio signal can quickly reach the peak and then quickly attenuate, and the characteristic is used as the judgment basis for the passing of the automobile.
Mode 3: extracting the spectral characteristics of the automobile sound by using the short-time autocorrelation function of the audio signal, determining the type of the audio target, and calculating the short-time autocorrelation function of the audio signal according to the formula:
When the automobile passes through the audio monitoring unit, the short-time autocorrelation function of the audio signal can quickly reach a peak value and then quickly attenuate, and the characteristic can be used as a judgment basis for passing of the automobile.
Mode 4: the method comprises the following steps of extracting audio signal characteristics by using convolutional neural network models such as CNN, RNN and CRNN, and determining the type of an audio target, wherein the method comprises the following specific steps:
s21), collecting training samples, collecting the training samples and storing the training samples in a background database, wherein the training samples comprise vehicle noise data, vehicle types, vehicle speeds and distances between the vehicles and audio monitoring equipment, the training samples are vehicle noises corresponding to k vehicle types under the conditions of p vehicle speeds and g vehicle and audio monitoring unit distances, the vehicle noises comprise vehicle noise intensity values and vehicle noise audio values, and the vehicle types comprise passenger cars, trucks and special operation cars.
The storage mode of the training samples in the background database is as follows: in subfiles of three vehicle types, vehicle speed sample values, vehicle-audio monitoring unit distance sample values and collected corresponding vehicle noise data are stored in a classification matrix form, the content of the matrix is shown in table 1, wherein Vi (i is 1, 2.. multidot., p) is p vehicle speed training sample values, Sj (j is 1, 2.. multidot., g) is vehicle-audio monitoring unit distance training sample values, Fij is a vehicle noise intensity value of a vehicle at a distance value Sj with the vehicle speed Vi, Qij is a vehicle noise audio value of the vehicle at the distance value Sj with the vehicle speed Vi, and specific actual vehicle sample noise collection is shown in table 1:
TABLE 1
In the embodiment, the types of the vehicles comprise a passenger car, a truck and a special work vehicle, wherein the passenger car comprises a mini-type passenger car, a medium-sized passenger car, a passenger train, a large-sized passenger car and the like; the trucks comprise 6 types of trucks, such as trucks, truck trains, semi-trailer trains and the like; the special operation vehicle comprises 6 types of special operation vehicles, such as trailers and engineering vehicles. Training samples are stored for each vehicle type.
In the embodiment, the speed sample values are obtained by respectively and uniformly selecting p speeds as training samples in a vehicle speed interval (0, 200km/h) according to empirical values, and the speed intervals can be 5km/h, 10km/h and 20 km/h.
S22), under the condition that the vehicle travels, extracting and marking common vehicle noise audio frequency spectrum characteristics respectively corresponding to the three types of vehicles under different driving speeds and different distances from the acquisition equipment;
s23), constructing a model, classifying different vehicle noises and environmental noises after the acquisition and the marking are completed, extracting the characteristics by adopting a deep convolutional neural network, wherein the input layer of the deep convolutional neural network is different types of noises (vehicle noises and environmental noises) which are already marked, and the output layer is the probability of the different types of noises;
in this embodiment, the input layer data are marked vehicle noise, environmental noise and the like, and N types of noise are assumed to exist and are respectively marked as 0, 1,2, · and N-1;
the design network structure of the convolutional neural network used in this embodiment is shown in table 2:
TABLE 2
In table 2, the volume indicates the Convolutional layer (Conv), normalization (BN), and activation layer (LeakyRelu), and the structure is shown in fig. 2.
Residual represents the Residual structure, and is expressed by the formula y ═ x + f (x); n is the number of noise classes, AvgPool is the global pooling layer, FC represents the full connection layer, Softmax is the classifier, the specific formula is as follows,
the final output through the network training is the probabilities of the various classes.
S03), if the target identification unit identification result contains the warning content, sending an instruction to the warning signal unit, and sending warning information by the warning signal unit according to the warning type;
after the warning signal unit receives the warning signal, the LED array warning lamp is changed from the yellow constant to the red to flash, so that the follow-up vehicle is reminded of slowing down and slowly moving, or other display modes are adopted according to the types and the numbers of the warning signals.
S04), the target recognition unit sends the recognized target information to the event recognition unit through the network bus, the moving speed and the speed change of the target are obtained by comprehensively analyzing the recognition data sent by each target recognition unit, and then the vehicle driving condition is detected and recognized, and the warning signal is sent to the warning signal unit to send out the corresponding warning signal.
In this embodiment, the event recognition unit may recognize the target type, and the target type mainly includes a vehicle type, a large vehicle, a medium vehicle, a small vehicle, and a special vehicle.
In this embodiment, the target identification unit calculates the vehicle passing speed by using the change of the noise signal spectrum when the vehicle passes through according to the doppler effect, and the calculation method includes: c is the speed of sound in air, v is the speed of the vehicle, f1Is the fundamental frequency of the incoming noise of the vehicle, f2Is the fundamental frequency of the automobile departure
Judging the size of V, if V<And 0 or v is 0, and the vehicle is judged to be stopped or driven in the reverse direction.
In this embodiment, the event recognition unit may detect a driving state of the vehicle between two target units, and the determination process is as follows: if the event recognition unit receives a k +1 th target recognition unit signal in a time interval [ t-d, t + d ] after receiving the k-th target recognition unit signal, the vehicle is judged to stop between the k and k +1 target units or to run slowly.
In the embodiment, the target recognition unit can also train and construct a dangerous audio signal early warning neural network model, and abnormal sound signals are recognized by collecting a large amount of training materials such as braking, collision, traffic jam and human voice.
The method of the embodiment utilizes the audio monitoring and identification technology, is not influenced by rain and fog weather, and can provide more accurate identification effect than methods such as infrared, laser, microwave and the like; the device is cheaper in design and generation cost than various schemes based on optical technology in the industry at present by adopting the technology based on audio monitoring and recognition, and is suitable for large-scale popularization and use; the modules for realizing the method can independently work in a non-networking state, and are convenient and quick to install, maintain and disassemble.
The foregoing description is only for the purpose of illustrating the general principles and preferred embodiments of the present invention, and it is within the scope of the present invention for modifications and alterations by those skilled in the art in light of the present disclosure.
Claims (10)
1. A highway fog zone induction method is characterized in that: the method comprises the following steps: s01), taking a section of highway as a monitoring interval, installing more than one audio monitoring unit at one side of the highway in the monitoring interval, wherein the audio monitoring units are responsible for collecting the sound from the monitoring interval and sending the collected audio signals to a target identification unit; s02), the target identification unit detects whether the audio signal includes an audio target to be detected, the audio target to be detected is sound related to road traffic, and the type of the audio target is determined by extracting sound frequency spectrum characteristics; s03), if the target identification unit identification result contains the warning content, sending an instruction to the warning signal unit, and sending warning information by the warning signal unit according to the warning type; s04), the target recognition unit sends the recognized target information to the event recognition unit, the moving speed and the speed change of the target are obtained by comprehensively analyzing the recognition data sent by each target recognition unit, and then the vehicle driving condition is detected and recognized, and the warning signal is sent to the warning signal unit to send out the corresponding warning signal.
2. The highway mist induction method according to claim 1, wherein: the target identification unit extracts audio signal characteristics through the zero crossing rate of the audio signal and determines the type of the audio target, wherein the zero crossing rate of the audio signal is calculated according to the formula:
wherein N is1Is the number of sampling points of a frame, m is a sampling point, x (m) is the audio signal amplitude of the sampling point, x (m-1) is the audio signal amplitude of the previous sampling point, sgn [ [ MEANS ] ]]In order to be a function of the sign,
3. the highway mist induction method according to claim 1, wherein: the target identification unit extracts the audio signal characteristics through the short-time energy of the audio signal, determines the type of the audio signal, and calculates the short-time energy of the audio signalThe formula is as follows:
N2is the window length, m is the sampling point, n is the current pixel point, x (m) is the amplitude of the audio signal, w (n-m) is the weight of the audio signal, the visible short-time energy is the weighted sum of squares of the sample values of a frame, when the window function is a rectangular window,
4. the highway mist induction method according to claim 1, wherein: the target identification unit extracts audio signal characteristics through an audio signal short-time autocorrelation function and determines the type of an audio target, and the specific steps are as follows:
m is the sampling point, x (m) is the audio signal amplitude of the current sampling point, and x (m + k) is the audio signal amplitude after k sampling points.
5. The highway mist induction method according to claim 1, wherein: extracting audio signal features through a convolutional neural network, and determining the type of an audio target, wherein the method specifically comprises the following steps: s21), collecting training samples, wherein the training samples comprise vehicle noise data, vehicle types, vehicle speeds and distances between the vehicles and the audio monitoring units, the training samples are vehicle noises corresponding to k vehicle types under the conditions of p vehicle speeds and g distances between the vehicles and the audio monitoring units, and the vehicle noises comprise vehicle noise intensity values and vehicle noise audio values; s22), under the condition that the vehicle is moving, extracting and marking common vehicle noise audio frequency spectrum characteristics respectively corresponding to K vehicle types at different driving speeds and at different distances from the audio monitoring equipment; s23), classifying different vehicle noises and environmental noises after the collection marking is completed, wherein the input layer is different types of noises after the marking is completed, and the output layer is the probability of different types of noises.
6. The highway mist induction method according to claim 5, wherein: and storing the vehicle speed sample value, the distance sample value between the vehicle and the audio monitoring unit and the collected corresponding vehicle noise data in a classification matrix form in the subfile of the vehicle type.
7. The highway mist induction method according to claim 1, wherein: the steps of the event recognition unit for recognizing the vehicle running condition are as follows: s41), identifying the vehicle type; s42), c is the propagation speed of sound in air, v is the speed of automobile, f1Is the fundamental frequency of the incoming noise of the vehicle, f2Is the noise fundamental frequency of the vehicle leaving
S43), if v<If 0 or v is 0, the vehicle is judged to be stopped or driven backwards; s44), let t be S/v, S be the distance between two audio monitoring units, v be a speed value set for detecting the running state of the vehicle, d be half the length of the detection time interval, if the event identification unit receives the time interval [ t-d, t + d ] after the k-th target identification unit signal]And receiving the signal of the (k + 1) th target identification unit, and judging that the vehicle stops or slowly runs between the (k) and the (k + 1) th target units.
8. The highway mist induction method according to claim 1, wherein: the audio frequency target detected by the target identification unit comprises automobile passing, automobile static idle speed, automobile brake, traffic accidents, falling objects, natural disasters, intruders and animals.
9. The highway mist induction method according to claim 1, wherein: the audio monitoring unit adopts a directional sound pickup to collect sound from a specific direction of a road.
10. The highway mist induction method according to claim 1, wherein: the audio monitoring unit adopts a ring-shaped directional sound pickup to collect sound from a road.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911153267.3A CN110942670A (en) | 2019-11-20 | 2019-11-20 | Expressway fog area induction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911153267.3A CN110942670A (en) | 2019-11-20 | 2019-11-20 | Expressway fog area induction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110942670A true CN110942670A (en) | 2020-03-31 |
Family
ID=69907878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911153267.3A Pending CN110942670A (en) | 2019-11-20 | 2019-11-20 | Expressway fog area induction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110942670A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112706691A (en) * | 2020-12-25 | 2021-04-27 | 奇瑞汽车股份有限公司 | Vehicle reminding method and device |
| CN113643540A (en) * | 2021-08-16 | 2021-11-12 | 李振宇 | Rapid detection and early warning system and method for highway accidents |
| CN113888873A (en) * | 2021-10-27 | 2022-01-04 | 李振宇 | Highway accident detection and early warning system and method based on short-term traffic flow |
| CN114067612A (en) * | 2021-11-26 | 2022-02-18 | 山东科技大学 | Vehicle perception and danger early warning method and system based on environmental sound analysis |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10162283A (en) * | 1996-11-28 | 1998-06-19 | Hitachi Ltd | Road condition monitoring device |
| CN102163427A (en) * | 2010-12-20 | 2011-08-24 | 北京邮电大学 | Method for detecting audio exceptional event based on environmental model |
| CN104851301A (en) * | 2015-05-22 | 2015-08-19 | 重庆交通大学 | Vehicle parameter identification method based on deceleration strip sound analysis |
| CN105489019A (en) * | 2015-12-18 | 2016-04-13 | 中山大学 | Model-based traffic flow monitoring system based on dual audio signal acquisition |
| CN106960581A (en) * | 2017-04-25 | 2017-07-18 | 中国计量大学 | Speed measurer for motor vehicle based on voice signal |
| CN107545890A (en) * | 2017-08-31 | 2018-01-05 | 桂林电子科技大学 | A kind of sound event recognition method |
| CN107610715A (en) * | 2017-10-10 | 2018-01-19 | 昆明理工大学 | A kind of similarity calculating method based on muli-sounds feature |
| CN108648507A (en) * | 2018-06-29 | 2018-10-12 | 中国矿业大学 | Vehicle and pedestrian mutually knows and method for early warning for a kind of zebra stripes " ghost probe " |
-
2019
- 2019-11-20 CN CN201911153267.3A patent/CN110942670A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10162283A (en) * | 1996-11-28 | 1998-06-19 | Hitachi Ltd | Road condition monitoring device |
| CN102163427A (en) * | 2010-12-20 | 2011-08-24 | 北京邮电大学 | Method for detecting audio exceptional event based on environmental model |
| CN104851301A (en) * | 2015-05-22 | 2015-08-19 | 重庆交通大学 | Vehicle parameter identification method based on deceleration strip sound analysis |
| CN105489019A (en) * | 2015-12-18 | 2016-04-13 | 中山大学 | Model-based traffic flow monitoring system based on dual audio signal acquisition |
| CN106960581A (en) * | 2017-04-25 | 2017-07-18 | 中国计量大学 | Speed measurer for motor vehicle based on voice signal |
| CN107545890A (en) * | 2017-08-31 | 2018-01-05 | 桂林电子科技大学 | A kind of sound event recognition method |
| CN107610715A (en) * | 2017-10-10 | 2018-01-19 | 昆明理工大学 | A kind of similarity calculating method based on muli-sounds feature |
| CN108648507A (en) * | 2018-06-29 | 2018-10-12 | 中国矿业大学 | Vehicle and pedestrian mutually knows and method for early warning for a kind of zebra stripes " ghost probe " |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112706691A (en) * | 2020-12-25 | 2021-04-27 | 奇瑞汽车股份有限公司 | Vehicle reminding method and device |
| CN113643540A (en) * | 2021-08-16 | 2021-11-12 | 李振宇 | Rapid detection and early warning system and method for highway accidents |
| CN113643540B (en) * | 2021-08-16 | 2022-06-14 | 李振宇 | Rapid detection and early warning system and method for highway accidents |
| CN113888873A (en) * | 2021-10-27 | 2022-01-04 | 李振宇 | Highway accident detection and early warning system and method based on short-term traffic flow |
| CN114067612A (en) * | 2021-11-26 | 2022-02-18 | 山东科技大学 | Vehicle perception and danger early warning method and system based on environmental sound analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108510797B (en) | Forewarning System of Freeway and method based on radar detecting | |
| CN110942670A (en) | Expressway fog area induction method | |
| CN108919256B (en) | Four-dimensional real-scene traffic simulation vehicle overspeed whole-course tracking detection alarm system and method | |
| CN112750326B (en) | Generalized vehicle-road cooperation method | |
| CN106652558B (en) | Vehicle-road cooperative intelligent traffic control system | |
| US8576069B2 (en) | Mobile sensing for road safety, traffic management, and road maintenance | |
| WO2018058958A1 (en) | Road vehicle traffic alarm system and method therefor | |
| CN105390027B (en) | A kind of road safety monitoring and early warning device and method | |
| CN102632839A (en) | Back sight image cognition based on-vehicle blind area early warning system and method | |
| CN102592457A (en) | Compound type interval velocity-measuring system based on internet of things technology and method thereof | |
| CN216110866U (en) | Tunnel vehicle accident detection and early warning device based on RSSI | |
| CN113147733A (en) | Intelligent speed limiting system and method for automobile in rain, fog and sand-dust weather | |
| CN114387785A (en) | Safety management and control method and system based on intelligent highway and storable medium | |
| CN113112830B (en) | Signal control intersection emptying method and system based on laser radar and track prediction | |
| CN107146423A (en) | A kind of crossing inlet track vehicle guiding signal control device and method | |
| CN111477011A (en) | Detection device and detection method for road intersection early warning | |
| CN112466144A (en) | Road network traffic situation management and control system | |
| US10930145B2 (en) | Traffic system for predicting and providing traffic signal switching timing | |
| CN110807929A (en) | Traffic police auxiliary command information analysis and research and judgment system | |
| CN114655260A (en) | Control system of unmanned tourist coach | |
| CN113870551A (en) | Roadside monitoring system capable of identifying dangerous and non-dangerous driving behaviors | |
| CN116165655A (en) | Global vehicle track tracking system based on millimeter wave radar group | |
| CN116798222A (en) | Method, system and device for detecting visibility of expressway and intelligently inducing visibility of expressway | |
| CN100419460C (en) | Highway climate monitoring and alarm system | |
| CN114299715A (en) | Expressway information detection system based on videos, laser radar and DSRC |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200331 |
|
| WD01 | Invention patent application deemed withdrawn after publication |