Intelligent cone barrel night traffic early warning method based on Bayesian network

Abstract

The invention discloses an intelligent cone barrel night traffic early warning method based on a Bayesian network, and the method comprises the following steps: carrying out the recognition of variables needed in the network, enabling the variables to serve as network nodes, constructing the Bayesian network through the network nodes, and carrying out the training, adjustment and determination of the Bayesian network through the collected data; a sensor module and equipment in the intelligent cone barrel are used for collecting vehicle track data and surrounding environment data; real-time data collected by the intelligent cone barrels are input into the Bayesian network, and the Bayesian network issues each item of data to a corresponding node for monitoring, so that early warning and prediction of traffic are realized; according to a prediction result of the Bayesian network, the intelligent cone barrel starts a corresponding processing strategy, the detected road is predicted in real time by using the Bayesian network, when an abnormal condition is detected, early warning and reporting are performed in time, and compared with other traffic event detection algorithms of multiple devices, the detection efficiency is improved. The traffic incident detection method designed by the patent is better in accuracy and stronger in pertinence.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic incident detection and early warning, and specifically is a smart cone-barrel nighttime traffic early warning method based on a Bayesian network. Background Art

[0002] In recent years, the continuous advancement of Internet of Things technology has led to the intelligentization and efficiency of science and technology in all walks of life. Among them, Internet of Things technology has played a vital role in promoting the rapid development of smart transportation systems. The combination of traditional transportation facilities or equipment with Internet of Things technology has made the transportation system more visual, safer and more convenient. The update of technology and equipment of smart transportation systems will also provide more powerful support for the accurate monitoring of various traffic events.

[0003] However, monitoring of various traffic incidents in construction scenarios at night faces multiple difficulties. These difficulties mainly stem from the particularity of the night environment and the potential impact of construction activities on traffic. The particularity of the night environment is reflected in the significant reduction of natural light in night construction scenarios, resulting in reduced visibility. This affects the imaging quality and monitoring accuracy of traffic monitoring equipment (such as cameras), increasing the difficulty of traffic monitoring. At the same time, night construction may cause surrounding roads to become the main channels for traffic detours, thereby increasing traffic flow. When facing relatively low traffic flow, construction roads may also have potential traffic accidents due to various factors. Reflective cones are commonly used warning reflective equipment in night construction scenarios. Smart cones formed by integrating Internet of Things technology can be combined with other equipment to detect vehicles and the surrounding environment, which is of great significance for ensuring safety warnings on night construction roads.

[0004] At present, research on road traffic safety by multi-device systems is relatively common, but most of them focus on the safety protection of highway maintenance operations, and lack safety warnings in specific night construction scenarios. After searching, the patent with announcement number CN202323015417.9 discloses a collaborative safety protection system for highway maintenance operations. This patent uses relevant machine learning algorithms to monitor the violations of operators in highway maintenance operations, and uses the VFD-YOLOv4 algorithm model to track vehicles in the close-range AI video provided by the smart maintenance pole, overcoming the shortcomings of the existing technology of low detection accuracy and slow detection rate. However, due to the weak light conditions in night construction scenes, video technology cannot detect and track vehicles well, so the performance of the safety protection system will be significantly reduced in night construction scenes.

[0005] The patent with announcement number CN202210262363.7 discloses a real-time sharing system for road construction conditions. The patent designs a real-time sharing system for road construction conditions including smart cones, positioning satellites, cloud servers, an Internet publishing system and a safety management system. The smart cones in the system have the function of reporting tilt and fall, which can remind road management personnel to promptly prop them up or automatically restore them. At the same time, the smart cones can also detect traffic accidents by determining whether they have been hit, and report them to the safety management system. Through the Internet publishing system, road management personnel can remind drivers on various navigation maps that traffic accidents have occurred on the road ahead, and please slow down. The patent is complete in system design, but lacks specific vehicle detection and accurate judgment methods for traffic accidents. For this reason, this application proposes a multi-device night traffic warning method based on a Bayesian network. Summary of the invention

[0006] The present invention proposes a multi-device nighttime traffic warning method based on a Bayesian network. This method is targeted at nighttime construction scenes. The Bayesian network obtained by historical data and real-time sensor data enable the intelligent warning device of the traffic monitoring system to have higher real-time and accuracy, and can quickly identify possible traffic accidents, issue corresponding warnings to remind equipment managers to respond to sudden traffic conditions, and ensure traffic safety in nighttime construction scene sections.

[0007] In order to solve the above problems, the present invention provides a technical solution:

[0008] A smart cone-barrel nighttime traffic warning method based on a Bayesian network comprises the following steps:

[0009] S1. Constructing a Bayesian network: Identify the variables needed in the network. The variables will be used as network nodes to construct a Bayesian network. The collected data will be used to train, adjust and determine the Bayesian network.

[0010] S2. Obtaining traffic environment data: Using the sensor modules and devices inside the smart cone to collect vehicle trajectory data and surrounding environment data;

[0011] S3. Traffic warning prediction: The real-time data collected by the smart cones is input into the Bayesian network, which sends each data item to the corresponding node for monitoring to achieve traffic warning prediction;

[0012] S4. Nighttime road condition reporting: Based on the prediction results of the Bayesian network, the smart cone barrel initiates the response strategy.

[0013] Preferably, the devices inside the smart cone include a magnetic sensor and a light intensity sensor;

[0014] The magnetic sensor is used to detect passing vehicles within the detection range;

[0015] The light intensity sensor is used to detect the light intensity of the surrounding environment of the smart cone.

[0016] Preferably, a wireless communication module and a computing module are integrated inside the smart cone;

[0017] The wireless communication module is used to push road information and traffic vehicle information to the cloud platform;

[0018] The calculation module tracks the trajectory of the vehicle through a multi-target tracking algorithm and Kalman filtering, estimates the vehicle's driving speed and acceleration data, and is also used to fuse the above data.

[0019] Preferably, the smart cone barrel is provided with reflective material on the outside, and the reflective material is used to warn people when the traffic on the construction road is normal at night.

[0020] Preferably, the variables in step S1 include light intensity, vehicle speed, vehicle acceleration, smart cone posture and smart cone position.

[0021] Preferably, the implementation steps of the S3 traffic warning prediction are:

[0022] S31. Place smart cones at equal intervals around the nighttime construction scene without affecting the normal passage of vehicles, and collect various data of the vehicles;

[0023] S32, inputting data such as the speed data of the tracked vehicle, the ambient light, and whether the vehicle is detected into the Bayesian network, and calculating the probability prediction of a traffic accident through the Bayesian network.

[0024] The beneficial effects of the present invention are as follows: in response to the problem of traffic incident detection on construction roads at night, based on the intelligent transportation system, a nighttime construction road traffic detection method using intelligent multi-devices combined with Bayesian networks is proposed, filling the gap in the field of traffic incident detection in nighttime construction scenarios. In response to the problem that vehicles breaking into construction areas or nighttime roads with potential safety hazards are difficult to detect, the Bayesian network is used to make real-time predictions on the detected roads, and when abnormal situations are detected, early warnings are reported in a timely manner. Compared with other multi-device traffic incident detection algorithms, the traffic incident detection method designed in this patent has better accuracy and is more targeted. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] For ease of explanation, the present invention is described in detail with reference to the following specific implementations and the accompanying drawings.

[0026] Figure 1is a structural diagram of the data acquisition device of the present invention;

[0027] Figure 2 It is a Bayesian network structure diagram of nighttime traffic detection of the present invention;

[0028] Figure 3 It is a Bayesian network prediction process diagram of the present invention;

[0029] Figure 4 It is a flow chart of real-time detection of nighttime road conditions of the present invention. DETAILED DESCRIPTION

[0030] like Figure 1-4 As shown, the specific implementation adopts the following technical solutions:

[0031] Embodiment 1:

[0032] A smart cone-barrel nighttime traffic warning method based on a Bayesian network comprises the following steps:

[0033] S1. Constructing a Bayesian network: Identify the variables needed in the network. The variables will be used as network nodes to construct a Bayesian network. The collected data will be used to train, adjust and determine the Bayesian network.

[0034] S2. Obtaining traffic environment data: Using the sensor modules and devices inside the smart cone to collect vehicle trajectory data and surrounding environment data;

[0035] S3. Traffic warning prediction: The real-time data collected by the smart cones is input into the Bayesian network, which sends each data item to the corresponding node for monitoring to achieve traffic warning prediction;

[0036] S4. Nighttime road condition reporting: Based on the prediction results of the Bayesian network, the smart cone barrel initiates the response strategy.

[0037] Among them, the equipment inside the smart cone includes a magnetic sensor and a light intensity sensor; the magnetic sensor is used to detect passing vehicles within the detectable range; the light intensity sensor is used to detect the light intensity of the environment around the smart cone; the smart cone is integrated with a wireless communication module and a computing module; the wireless communication module is used to push road information and traffic vehicle information to the cloud platform; the computing module tracks the trajectory of the vehicle through a multi-target tracking algorithm and Kalman filtering, and estimates the vehicle's driving speed and acceleration data, and is also used to fuse the above data; the outside of the smart cone is provided with reflective material, and the reflective material is used to warn people when the traffic on the construction road is normal at night.

[0038] The variables in step S1 include light intensity, vehicle speed, vehicle acceleration, smart cone posture and smart cone position.

[0039] The implementation steps of the S3 traffic warning forecast are as follows:

[0040] S31. Place smart cones at equal intervals around the nighttime construction scene without affecting the normal passage of vehicles, and collect various data of the vehicles;

[0041] S32, inputting data such as the speed data of the tracked vehicle, the ambient light, and whether the vehicle is detected into the Bayesian network, and calculating the probability prediction of a traffic accident through the Bayesian network.

[0042] Embodiment 2:

[0043] The Bayesian network is drawn according to the dependency relationship. For roads where traffic accidents occur or where there are major safety hazards, the number of vehicles on the road, vehicle speed, lighting conditions, vehicle detection, and whether the cone barrel is hit are used to construct a Bayesian network. The structure is as follows: Figure 2 As shown in the figure, after obtaining the Bayesian network structure diagram, the conditional probability of each node is calculated, and the events of poor lighting conditions, abnormal vehicle speed, abnormal vehicle missed detection, cone barrel self-detected collision, and traffic accidents or traffic hazards are A, A1, A2, A3, and T respectively; since the probability of poor lighting conditions in night construction scenes is relatively high, P(A) can be estimated to be a larger value, and the probabilities of events A1, A2, and A3 are all conditional probabilities calculated under poor lighting conditions, recorded as independent probabilities: P(A1|A), P(A2|A), and P(A3|A), which can also be It can be obtained by analyzing historical road traffic data. For example, by counting the distribution of vehicle speeds under poor lighting conditions, P(A1|A) can be obtained, and by counting the distribution of each variable in cases of traffic accidents, P(AA1A2A3|T) can be obtained. The conditional probability reflects the distribution characteristics of each variable under poor lighting conditions, which helps the model to more accurately predict the probability of traffic accidents. The marginal probability P(T) refers to the probability of a nighttime construction road traffic event itself, without considering the influence of other variables. Therefore, according to the Bayesian network, the probability of a vehicle accidentally entering the construction area or a hidden danger event T can be predicted. The specific formula is:

[0044]

[0045] In actual application, without affecting the normal traffic of vehicles, the smart cones are placed at equal intervals around the night construction scene. The magnetic sensors and light intensity sensors inside the smart cones collect vehicle trajectory data such as vehicle speed, acceleration, light intensity, and detection of whether the vehicle has been hit. The collected data is input into the calculation module inside the smart cones for data processing, and combined with other smart cones to achieve vehicle tracking and vehicle number statistics. The processed data is input into the Bayesian network for event prediction. The probability prediction of traffic accidents is obtained through Bayesian network calculation. For example, if a vehicle accidentally breaks into the construction area, if the prediction results show that the vehicle on this part of the night construction road has If there is a high probability of accidentally breaking into the construction area, the smart cone will actively turn on the periodic flashing and alarm system to remind distant night-time vehicles. At the same time, the internal communication module of the smart cone will upload the detected vehicle data, environmental data, decision-making information to the cloud platform, and finally pass it to the management system and transportation department of the night construction site, prompting the construction site management personnel to deal with potential safety hazards in a timely manner and ensure the safety of the lives of people on the night construction roads; if the prediction results show that the traffic on this part of the night construction road is normal, the reflective material on the surface of the cone can be used to remind night personnel, saving energy inside the cone, and reporting normal vehicle data and surrounding environmental data at the same time.

[0046] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart cone-bucket nighttime traffic warning method based on Bayesian network, characterized in that: The following steps are involved: S1. Constructing a Bayesian network: Identify the variables needed in the network. The variables will be used as network nodes to construct a Bayesian network. The collected data will be used to train, adjust and determine the Bayesian network. S2. Obtaining traffic environment data: Using the sensor modules and devices inside the smart cone to collect vehicle trajectory data and surrounding environment data; S3. Traffic warning prediction: The real-time data collected by the smart cones is input into the Bayesian network, which sends each data item to the corresponding node for monitoring to achieve traffic warning prediction; S4. Nighttime road condition reporting: Based on the prediction results of the Bayesian network, the smart cone barrel initiates the response strategy.

2. According to the Bayesian network-based smart cone nighttime traffic warning method of claim 1, it is characterized in that: The devices inside the smart cone include a magnetic sensor and a light intensity sensor; The magnetic sensor is used to detect passing vehicles within the detection range; The light intensity sensor is used to detect the light intensity of the surrounding environment of the smart cone.

3. According to the Bayesian network-based smart cone nighttime traffic warning method of claim 2, it is characterized in that: The smart cone is internally integrated with a wireless communication module and a computing module; The wireless communication module is used to push road information and traffic vehicle information to the cloud platform; The calculation module tracks the trajectory of the vehicle through a multi-target tracking algorithm and Kalman filtering, estimates the vehicle's driving speed and acceleration data, and is also used to fuse the above data.

4. According to the Bayesian network-based smart cone nighttime traffic warning method of claim 1, it is characterized in that: The outside of the smart cone is provided with reflective material, and the reflective material is used to warn people when the traffic on the construction road is normal at night.

5. According to the Bayesian network-based smart cone nighttime traffic warning method of claim 1, it is characterized in that: The variables in step S1 include light intensity, vehicle speed, vehicle acceleration, smart cone posture and smart cone position.

6. According to the Bayesian network-based smart cone nighttime traffic warning method of claim 1, it is characterized in that: The implementation steps of the S3 traffic warning prediction are: S31. Place smart cones at equal intervals around the nighttime construction scene without affecting the normal passage of vehicles, and collect various data of the vehicles; S32, inputting data such as the speed data of the tracked vehicle, the ambient light, and whether the vehicle is detected into the Bayesian network, and calculating the probability prediction of a traffic accident through the Bayesian network.

Images