Traffic abnormal event detection method based on video stream analysis

Abstract

The invention discloses a traffic abnormal event detection method based on video stream analysis. The method specifically comprises the following steps: obtaining a monitoring video; detecting a preset detection object in the monitoring video by adopting a deep learning target detection algorithm, and outputting a target object according to a category filtering mechanism; tracking the target object by adopting a target tracking algorithm; fitting a lane area based on the running track of the vehicle to realize target area division; and carrying out traffic abnormal event detection on the lane area according to the monitoring state and a preset rule.

Description

Technical Field

[0001] The present invention is applied to the field of abnormal traffic event detection, and specifically is a method for detecting abnormal traffic events based on video stream analysis. Background Art

[0002] With the acceleration of urbanization, highway traffic safety issues are becoming increasingly prominent. Traditional traffic monitoring systems mainly rely on manual monitoring, which is inefficient and prone to omissions. Therefore, it is of great practical significance to develop an automated high-speed traffic abnormal event detection system. The development of deep learning technology, especially the emergence of convolutional neural network (CNN) models, can extract features from a large amount of unstructured data and perform effective pattern recognition, bringing revolutionary progress to the field of traffic abnormal event detection. However, existing deep learning-based technologies often simply detect targets and then identify abnormal events based on event logic. Affected by the complexity of the environment, they have a high misjudgment rate and poor results.

[0003] The present invention combines deep learning technology with video stream collection based on high-position cameras on the roadside to conduct 24-hour automatic computer analysis of abnormal high-speed traffic events, thereby solving the problems faced by traditional traffic monitoring. It also utilizes target category filtering, lane information, and correlation between events to improve the accuracy of detecting abnormal high-speed traffic events. There are many abnormal high-speed traffic events, and here we focus on analyzing the automatic early warning of four aspects: illegal parking, wrong-way driving, congestion, and traffic accidents. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for detecting abnormal traffic events based on video stream analysis in view of the deficiencies of the prior art.

[0005] In order to solve the above technical problems, a method for detecting abnormal traffic events based on video stream analysis of the present invention specifically comprises the following steps:

[0006] Get surveillance video;

[0007] Use deep learning target detection algorithm to detect preset detection objects in surveillance videos, and output target objects based on category filtering mechanism;

[0008] Use target tracking algorithm to track the target object;

[0009] Based on the running trajectory of the vehicle, the lane area is fitted to achieve the target area division;

[0010] Detect abnormal traffic events in the lane area based on monitoring status and preset rules.

[0011] As a possible implementation, further, the step of detecting abnormal traffic events in the lane area according to the monitoring state and preset rules specifically includes: pre-classifying each target vehicle in the lane area according to the abnormal event rules into normal driving, illegal parking, and wrong-way driving;

[0012] Specifically: obtain the movement trajectory of the target vehicle, and based on the movement trajectory of the target vehicle and the judgment logic, pre-determine whether the vehicle is driving normally, illegally parked, or driving in the wrong direction.

[0013] The illegal parking judgment rule is: based on the position information of the front and rear frames of the vehicle, if the overlap frame of the front and rear frames of the target frame is greater than a preset threshold, the target is considered to be in a stationary state; if the duration is greater than a preset threshold, the vehicle is pre-judged to be illegally parked;

[0014] The reverse driving determination rule is: judging whether the vehicle is in reverse driving according to the consistency between the vehicle's driving trajectory and the lane direction; if the vehicle's driving direction is inconsistent with the lane and the vehicle travels more than a preset threshold distance, it is pre-determined to be in reverse driving;

[0015] The normal driving determination rule is: vehicles that are not illegally parked or driving in the wrong direction are judged to be driving normally.

[0016] As a possible implementation, further, the step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules further includes: judging whether the highway is in a congested state according to the driving conditions of the vehicle and the congestion event logic;

[0017] The logic for judging whether the highway is in a congested state is as follows: if the number of vehicles is greater than a preset number threshold, and the average running speed of all vehicles is less than a preset speed threshold, the highway is judged to be in a congested state.

[0018] As a possible implementation, further, the step of detecting abnormal traffic events in the lane area according to the monitoring state and the preset rules further includes: judging whether a traffic accident has occurred according to the state of the traffic accident and in combination with the traffic accident event logic;

[0019] Specifically: If the detection model detects that more than one target vehicle is in a stopped state, and the camera on site detects a non-vehicle target object, it is determined that a suspicious traffic accident has occurred and reported to the traffic management center for further manual confirmation.

[0020] As a possible implementation, further, the step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules further includes: judging whether the vehicle is illegally parked or driving against traffic according to the congestion situation, traffic accident situation, and event correlation;

[0021] Specifically: If the algorithm determines that the highway is congested or a traffic accident occurs, the scene is in an emergency state and the vehicle should not be determined to be illegally parked or driving in the wrong direction. Therefore, the joint events are associated and the warning output of illegal parking and driving in the wrong direction is canceled to reduce false alarms in special circumstances. The background records the video for manual review.

[0022] As a possible implementation method, further, if illegal parking, wrong-way driving, congestion or traffic accident occurs, the system will immediately issue an alarm and send a preset information type to the traffic management center.

[0023] As a possible implementation manner, further, the preset detection objects include vehicles, pedestrians, birds, reflective cones, and tripod warning signs.

[0024] A traffic abnormal event detection system based on video stream analysis, comprising:

[0025] Monitoring video frame acquisition module, used to obtain roadside camera video data;

[0026] A target detection filtering module is used to detect a preset detection object and output a target vehicle;

[0027] The target tracking module is used to track the target category;

[0028] Lane area acquisition module, used to obtain lane area information and reduce false alarms;

[0029] Abnormal individual vehicle event pre-identification module, used to pre-identify and determine whether the target vehicle is illegally parked or driving in the wrong direction;

[0030] A congestion event recognition module is used to determine whether the highway is in a congested state;

[0031] Traffic accident identification module, used to determine whether a traffic accident has occurred;

[0032] The abnormal individual vehicle secondary judgment module combines the event correlation to further determine whether the vehicle is illegally parked or driving in the wrong direction;

[0033] The event reporting module reports abnormal events to the traffic management center.

[0034] The present invention adopts the above technical scheme and has the following beneficial effects: the present invention detects targets based on video streams and combines deep learning technology, and combines multiple factors such as target category filtering, lane information, and event correlation to realize the recognition of high-speed traffic violations, wrong-way driving, congestion, and traffic accidents, thereby improving the recognition accuracy.

[0035] 1. Through the multi-target category detection and filtering mechanism, the misjudgment of other non-target vehicles, such as birds, reflective cones, etc., is eliminated. In addition, the type of engineering vehicles is detected and output separately to eliminate false alarms of high-speed working vehicles.

[0036] 2. By estimating the high-speed lane area through settings or algorithms, false alarms for complex backgrounds (leaves, high poles) are greatly reduced.

[0037] 3. Combine congestion and traffic accidents to make secondary judgments on vehicles parked illegally or driving in the wrong direction, reducing false alarms of abnormal events in special circumstances.

[0038] 4. Propose a set of judgment logic for abnormal traffic event detection to achieve effective judgment of abnormal traffic events. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0040] Example 1

[0041] A method for detecting abnormal traffic events based on video stream analysis specifically comprises the following steps:

[0042] Get surveillance video;

[0043] Use deep learning target detection algorithm to detect preset detection objects in surveillance videos, and output target objects based on category filtering mechanism;

[0044] Use target tracking algorithm to track the target object;

[0045] Based on the running trajectory of the vehicle, the lane area is fitted to achieve the target area division;

[0046] Detect abnormal traffic events in the lane area based on monitoring status and preset rules.

[0047] The steps of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules specifically include: pre-classifying each target vehicle in the lane area according to the abnormal event rules into normal driving, illegal parking, and wrong-way driving; specifically: obtaining the movement trajectory of the target vehicle, and based on the running trajectory of the target vehicle and based on the judgment logic, pre-determining whether the vehicle is normal driving, illegal parking, or wrong-way driving.

[0048] The illegal parking judgment rule is as follows: based on the position information of the front and rear frames of the vehicle, if the overlap frame of the front and rear frames of the target frame is greater than the preset threshold, the target is considered to be stationary; if the duration is greater than the preset threshold, the vehicle is pre-judged to be illegally parked;

[0049] The rule for determining the reverse driving is as follows: judging whether the vehicle is in the reverse driving direction based on the consistency between the vehicle's driving trajectory and the lane direction. If the vehicle's driving direction is inconsistent with the lane and the vehicle travels more than a preset threshold distance, it is pre-determined to be in the reverse driving direction.

[0050] The rule for determining normal driving is: vehicles that are not illegally parked or driving in the wrong direction are judged to be driving normally.

[0051] The step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also includes: judging whether the highway is in a congested state according to the driving conditions of the vehicle and the congestion event logic;

[0052] The logic for judging whether the highway is in a congested state is as follows: if the number of vehicles is greater than a preset number threshold, and the average running speed of all vehicles is less than a preset speed threshold, the highway is judged to be in a congested state.

[0053] The step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also includes: judging whether a traffic accident has occurred according to the status of the traffic accident and the logic of the traffic accident event; specifically: if the detection model detects that more than one target vehicle is in a stopped state, and the camera detects a non-vehicle target object on site, it is determined that a suspicious traffic accident has occurred and reported to the traffic management center for further manual confirmation.

[0054] The steps of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also include: judging whether the vehicle is illegally parked or driving in the wrong direction according to the congestion situation and traffic accident situation and the correlation of events; specifically: if the algorithm determines that the highway is congested or a traffic accident occurs, the scene is in an emergency state at this time, and the vehicle should not be judged as illegally parked or driving in the wrong direction. Therefore, the event correlation is combined, and the warning output of illegal parking and driving in the wrong direction of the vehicle is cancelled at this time to reduce false alarms in special circumstances, and the background records the video for manual review.

[0055] If illegal parking, driving against traffic, congestion or traffic accidents occur, the system will immediately send out an alarm and send the preset information type to the traffic management center. The preset detection objects include vehicles, pedestrians, birds, reflective cones and tripod warning signs.

[0056] A traffic abnormal event detection system based on video stream analysis, comprising:

[0057] Monitoring video frame acquisition module, used to obtain roadside camera video data;

[0058] A target detection filtering module is used to detect a preset detection object and output a target vehicle;

[0059] The target tracking module is used to track the target category;

[0060] Lane area acquisition module, used to obtain lane area information and reduce false alarms;

[0061] Abnormal individual vehicle event pre-identification module, used to pre-identify and determine whether the target vehicle is illegally parked or driving in the wrong direction;

[0062] A congestion event recognition module is used to determine whether the highway is in a congested state;

[0063] Traffic accident identification module, used to determine whether a traffic accident has occurred;

[0064] The abnormal individual vehicle secondary judgment module combines the event correlation to further determine whether the vehicle is illegally parked or driving in the wrong direction;

[0065] The event reporting module reports abnormal events to the traffic management center.

[0066] Example 2

[0067] The present invention provides a method and system for detecting abnormal high-speed traffic events based on video stream analysis. The brief steps are as follows:

[0068] 1. Obtain surveillance video collected by highway roadside cameras

[0069] 2. Use deep learning target detection algorithms to detect vehicles (engineering vehicles, target vehicles), pedestrians, birds, reflective cones, and tripod warning signs in the video, and output target vehicles based on the category filtering mechanism.

[0070] 3. Use target tracking algorithm to track the target.

[0071] 4. Obtain information on high-speed lane areas to achieve detailed division of target areas and reduce false positives of background targets.

[0072] 5. Pre-classify each target vehicle in the lane area according to the abnormal event rules into normal driving, illegal parking, and wrong-way driving.

[0073] 6. Based on the vehicle's driving conditions and the congestion event logic, determine whether the highway is in a congested state.

[0074] 7. Determine whether a traffic accident has occurred based on the state of the traffic accident and the logic of the traffic accident event.

[0075] 8. Based on the congestion situation, traffic accident situation and the correlation between the events, determine whether the vehicle is illegally parked or driving in the wrong direction.

[0076] 9. If illegal parking, wrong-way driving, congestion or traffic accidents occur, the system will immediately issue an alarm and send relevant information (such as location, time, and event type) to the traffic management center so that timely measures can be taken.

[0077] The high-speed traffic abnormal event detection system based on video stream analysis includes the following modules:

[0078] 1. Monitoring video frame acquisition module, mainly used to obtain roadside camera video data

[0079] 2. The target detection filtering module is mainly used to detect vehicles (engineering vehicles, target vehicles), pedestrians, birds, reflective cones, tripods, and output target vehicles.

[0080] 3. Target tracking module, mainly used to track target categories

[0081] 4. High-speed lane area acquisition module, mainly used to obtain lane area information and reduce false alarms

[0082] 5. The abnormal individual vehicle event pre-identification module is mainly used to pre-identify whether the target vehicle is illegally parked or driving in the wrong direction.

[0083] 6. Congestion event recognition module, mainly used to determine whether the highway is in a congested state

[0084] 7. Traffic accident identification module, mainly used to determine whether a traffic accident has occurred.

[0085] 8. The abnormal individual vehicle secondary judgment module mainly combines the event correlation to further judge whether the vehicle is illegally parked or driving against the flow.

[0086] 9. Event reporting module, mainly reports abnormal events to the traffic management center.

[0087] The above process is described in detail below:

[0088] 1. Obtain surveillance video collected by highway roadside cameras

[0089] The system obtains video data from highway roadside cameras through network protocols, such as RTSP, for subsequent algorithm recognition.

[0090] In actual applications, if camera data is obtained at full frame rate, it will generate greater bandwidth pressure, so the video stream will be extracted. In order to achieve better tracking effects, the video stream extraction frequency does not exceed 4 frames. Another way to reduce bandwidth pressure is to use an edge computing box, place the calculation on the edge side, and reduce the pressure of data transmission.

[0091] 2. Use deep learning target detection algorithms to detect video vehicles (engineering vehicles, target vehicles), pedestrians, birds, reflective cones, and tripod warning signs, and output target vehicles based on the target category filtering mechanism.

[0092] Based on the deep learning algorithm, the vehicle is detected. The traditional method of directly detecting the vehicle often leads to the misjudgment of other non-target vehicles, such as birds, reflective cones, pedestrians, etc. In addition, there are often engineering vehicles on the highway to carry out construction and inspections on the road. Such vehicles should not be alarmed. Based on this analysis, the present invention introduces a target category filtering mechanism. At the detection end, target vehicles, engineering vehicles, pedestrians, birds, reflective cones, and tripods are detected at the same time. Finally, only the target vehicle is used as the target vehicle category of abnormal events to reduce the false alarm of non-target vehicle categories. At the same time, the categories of reflective cones, tripods, and pedestrians can provide assistance for the judgment of subsequent event alarm logic.

[0093] The deep learning target detection module here can adopt mainstream single-stage and two-stage deep learning detection network models, such as the YOLO series (yolo v1~v5), SSD or retina and other ready-made open source detection models. The training sample set can be collected by itself, such as high-speed scene data sets of high-altitude scenes, open source vehicle and pedestrian data sets, etc. There is no restriction here.

[0094] 3. Use target tracking algorithm to track the target.

[0095] The target tracking algorithm is used to track the target and output the motion trajectory of each target to facilitate the judgment of subsequent events. The target tracking algorithm here can be tracked using mainstream tracking algorithms in the industry such as optical flow, sort, deepsort, etc., which will not be described in detail here.

[0096] 4. Obtain information on high-speed lane areas to achieve detailed division of target areas and reduce false positives of background targets.

[0097] Highways often contain a large number of poles and tall trees, and full-image detection often contains many background false alarms. The background of the highway lane area is single, and analysis based on the highway lane area can greatly reduce the false alarm rate. Therefore, in the present invention, the target output is limited to the highway lane area to improve the accuracy.

[0098] A commonly used method for obtaining high-speed lane areas is to use manual configuration, that is, before the system is started, the configuration personnel configure the lane area once for the algorithm to use. This method is simple and accurate, but more troublesome; the other is to fit the lane area based on the vehicle's running trajectory. The advantage of this method is that it is free from manual configuration, but the system needs to run stably for a period of time and the running trajectories of all vehicles must be counted. The present invention adopts the second method of automatically generating lane areas.

[0099] 5. Pre-classify each target vehicle in the lane area according to the abnormal event rules into normal driving, illegal parking, and wrong-way driving.

[0100] After the above steps, the present invention has obtained the movement trajectory of the target vehicle. According to the movement trajectory of the target vehicle, based on the judgment logic, it can be pre-determined whether the vehicle is driving normally, illegally parked, or driving in the wrong direction.

[0101] Illegal parking: Based on the position information of the front and rear frames of the vehicle, if the overlap of the front and rear frames of the target frame is greater than a certain threshold T1, the target is considered to be stationary. If the duration is greater than t1, the vehicle is pre-judged to be illegally parked. Here, the overlap threshold T1 is recommended to be set to 0.97, and the illegal parking duration t1 can be set by the management center.

[0102] Reverse driving: Determine whether the vehicle is in reverse driving based on the consistency between the vehicle's driving trajectory and the lane direction. If the vehicle's driving direction is inconsistent with the lane and the vehicle travels more than a certain distance L1, it is pre-determined to be in reverse driving. There are two ways to set the lane direction: one is to manually configure it, and the other is to count the directions of most vehicles when fitting the lane area, and use this direction as the normal driving direction. The present invention adopts the second method.

[0103] Normal driving: Vehicles that are not illegally parked or driving in the wrong direction are considered to be driving normally.

[0104] 6. Based on the vehicle's driving conditions and the congestion event logic, determine whether the highway is in a congested state.

[0105] According to the driving conditions of all vehicles, it is judged whether the highway is in a congested state. The specific judgment logic is: if the number of vehicles is greater than a certain threshold N, and the average running speed of all vehicles is less than a certain threshold V, then the highway is judged to be in a congested state.

[0106] Here, the threshold of the number of vehicles can be set manually according to the size of the lane area;

[0107] A method for setting the speed threshold V of a target is as follows: when the system starts, the time t from the appearance to the disappearance of the lane when the vehicle is driving normally is calculated, and the total number of pixels in the lane is counted as m, and the normal speed is estimated. Therefore, the low speed threshold It is the weight, which can be set by the management center. Here, 0.2 is recommended.

[0108] 7. Determine whether a traffic accident has occurred based on the state of the traffic accident and the logic of the traffic accident event.

[0109] The judgment of traffic accidents can be combined with the changes in vehicle trajectory and speed to predict whether a traffic accident may occur, such as changes in vehicle lane trajectory and a sudden change in speed from high speed to low speed. However, this method is difficult to track and cannot be well identified due to the sudden change in speed caused by the collision of vehicles. In the present invention, the state of the traffic accident is used to determine whether a traffic accident has occurred: if the detection model detects that more than one target vehicle is in a stopped state, and the camera on site detects pedestrians, tripods or reflective cones, it is considered that a suspicious traffic accident has occurred and the accident will be reported to the traffic management center for further manual confirmation.

[0110] 8. Based on the congestion situation, traffic accident situation and the correlation between the events, determine whether the vehicle is illegally parked or driving in the wrong direction.

[0111] If the algorithm determines that the highway is congested or a traffic accident has occurred, the scene is in an emergency state and the vehicle should not be determined to be illegally parked or driving in the wrong direction. Therefore, the joint event association is used to cancel the warning output of illegal parking and wrong driving at this time to reduce false alarms in special circumstances. The background records the video for manual review.

[0112] 9. If illegal parking, wrong-way driving, congestion or traffic accidents occur, the system will immediately send an alarm and relevant information (such as location, time, and event type) to the traffic management center so that timely measures can be taken.

[0113] After detecting an abnormal event, the system automatically alarms and notifies the traffic management center of relevant information (such as location, time, and event type) to take relevant measures.

[0114] The above are embodiments of the present invention. For ordinary technicians in the field, according to the teachings of the present invention, all equivalent changes, modifications, substitutions and variations made within the scope of the patent application of the present invention without departing from the principles and spirit of the present invention should fall within the scope of the present invention.

Claims

1. A method for detecting abnormal traffic events based on video stream analysis, characterized in that: The specific steps include: Get surveillance video; Use deep learning target detection algorithm to detect preset detection objects in surveillance videos, and output target objects based on category filtering mechanism; Use target tracking algorithm to track the target object; Based on the running trajectory of the vehicle, the lane area is fitted to achieve the target area division; Detect abnormal traffic events in the lane area based on monitoring status and preset rules.

2. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that: The step of detecting traffic abnormality events in the lane area according to the monitoring status and preset rules specifically includes: pre-classifying each target vehicle in the lane area according to the abnormal event rules into normal driving, illegal parking, and wrong-way driving; Specifically: the motion trajectory of the target vehicle is obtained, and based on the running trajectory of the target vehicle and the judgment logic, it can be pre-determined whether the vehicle is driving normally, illegally parked, or driving in the wrong direction. The illegal parking judgment rule is: based on the position information of the front and rear frames of the vehicle, if the overlap frame of the front and rear frames of the target frame is greater than a preset threshold, the target is considered to be in a stationary state; if the duration is greater than a preset threshold, the vehicle is pre-judged to be illegally parked; The reverse driving determination rule is: judging whether the vehicle is in reverse driving according to the consistency between the vehicle's driving trajectory and the lane direction; if the vehicle's driving direction is inconsistent with the lane and the vehicle travels more than a preset threshold distance, it is pre-determined to be in reverse driving; The normal driving determination rule is: vehicles that are not illegally parked or driving in the wrong direction are judged to be driving normally.

3. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that: The step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also includes: judging whether the highway is in a congested state according to the driving conditions of the vehicle and the congestion event logic; The logic for judging whether the highway is in a congested state is as follows: if the number of vehicles is greater than a preset number threshold, and the average running speed of all vehicles is less than a preset speed threshold, the highway is judged to be in a congested state.

4. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that: The step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also includes: judging whether a traffic accident has occurred according to the state of the traffic accident and in combination with the traffic accident event logic; Specifically: If the detection model detects that more than one target vehicle is in a stopped state, and the camera on site detects a non-vehicle target object, it is determined that a suspicious traffic accident has occurred and reported to the traffic management center for further manual confirmation.

5. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that: The step of detecting abnormal traffic events in the lane area according to the monitoring status and preset rules also includes: judging whether the vehicle is illegally parked or driving against traffic according to the congestion situation, traffic accident situation, and event correlation; Specifically: If the algorithm determines that the highway is congested or a traffic accident occurs, the scene is in an emergency state and the vehicle should not be determined to be illegally parked or driving in the wrong direction. Therefore, the joint events are associated and the warning output of illegal parking and driving in the wrong direction is canceled to reduce false alarms in special circumstances. The background records the video for manual review.

6. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that , also includes: If illegal parking, wrong-way driving, congestion or traffic accident occurs, the system will immediately issue an alarm and send the preset information type to the traffic management center.

7. The method for detecting abnormal traffic events based on video stream analysis according to claim 1, characterized in that: The preset detection objects include vehicles, pedestrians, birds, reflective cones, and tripod warning signs.

8. A traffic abnormality event detection system based on video stream analysis, characterized in that: include: Monitoring video frame acquisition module, used to obtain roadside camera video data; A target detection filtering module is used to detect a preset detection object and output a target vehicle; The target tracking module is used to track the target category; Lane area acquisition module, used to obtain lane area information and reduce false alarms; Abnormal individual vehicle event pre-identification module, used to pre-identify and determine whether the target vehicle is illegally parked or driving in the wrong direction; A congestion event recognition module is used to determine whether the highway is in a congested state; Traffic accident identification module, used to determine whether a traffic accident has occurred; The abnormal individual vehicle secondary judgment module combines the event correlation to further determine whether the vehicle is illegally parked or driving in the wrong direction; The event reporting module reports abnormal events to the traffic management center.