Traffic road surface vehicle stopping method and vehicle stopping roadblock based on danger situation awareness algorithm

Abstract

According to the traffic road surface vehicle stopping method based on the danger situation awareness algorithm and the vehicle stopping roadblock, the situation awareness intelligent algorithm combining the warning column and the portable vehicle stopping roadblock is adopted, and the warning column serves as a control center of a system, can automatically recognize approaching vehicles and send out warning signals and is in synchronous linkage with roadblock equipment, so that the traffic road surface vehicle stopping efficiency is improved. Automatic auxiliary operation in the traffic road surface vehicle stopping process is achieved, the vehicle is recognized through the YOLOv8 target detection model, and then the potential danger of the vehicle is predicted through the random forest algorithm. When the vehicle overspeeds or enters a dangerous area, the controller sends an unfolding signal through the ZigBee wireless module to trigger the roadblock system. The servo motor receives the signal and then unlocks the limiting plate, the cover plate is automatically unfolded, and a tire breaking chain is arranged. According to the design, the response speed and deployment precision of the roadblock system are greatly improved, the misjudgment rate is reduced, the safety and effectiveness of operation are enhanced, and the method is suitable for emergency blocking tasks in various complex scenes.

Description

Technical Field

[0001] The present invention relates to the application field of artificial intelligence situation awareness technology in traffic identification, and in particular to a traffic road blocking method and a roadblock based on a dangerous situation awareness algorithm. Background Art

[0002] In the current equipment field, especially in traffic control and public safety maintenance, portable roadblocks are widely used as an important security tool in various temporary or emergency road guidance, vehicle interception and other scenes, including tire breaker, also known as road thorn roadblock, barbed roadblock, etc., which is a tire breaker type roadblock driven by hydraulic power device, remote control or wire control. The thorns of road thorn roadblock are relatively sharp. After the vehicle tires are rolled over, they will be pierced and the gas in the tires will be vented through the deflation holes, causing the vehicle to be unable to move forward. Therefore, it is a necessary anti-terrorist roadblock in some key security places.

[0003] Although the existing portable roadblocks can meet the needs of rapid deployment and withdrawal to a certain extent, there are still some shortcomings in actual use. First, the traditional roadblocks are less portable and inconvenient to carry around. In actual use, a large amount of manpower is required for deployment and recovery, which not only increases the workload of traffic control personnel, but also reduces the efficiency of traffic control. Secondly, the deployment and withdrawal process of existing roadblocks is relatively complicated and takes a long time, which reduces the work efficiency of traffic control personnel. Finally, the roadblock is opened by a snap connection. When opening, the snap needs to be opened, then the cover is flipped open, and then the internal chain is thrown out. However, in actual use, the vehicle to be blocked may cause damage to the outer shell, making the outer shell and the cover inconvenient to install and repair.

[0004] In view of the above problems, the design and performance of portable vehicle-blocking barriers in the existing technology need to be further optimized and improved to meet the use needs of traffic control personnel in various complex environments, especially to improve their portability, ease of operation and stability, which is a key issue that needs to be urgently solved in the current equipment technology field. Summary of the invention

[0005] In order to overcome the existing problems, the embodiments of the present application provide a traffic road blocking method and a vehicle blocking roadblock based on a dangerous situation perception algorithm, an intelligent system that combines a warning post with a portable vehicle blocking roadblock. The warning post serves as the control center of the system, which can automatically identify approaching vehicles, issue a warning signal, and synchronize with the roadblock equipment to achieve automated auxiliary operation of the traffic road blocking process.

[0006] The technical solution adopted by the embodiment of the present application to solve the technical problem is:

[0007] A method for blocking vehicles on a traffic road based on a dangerous situation awareness algorithm, a situation awareness intelligent algorithm that combines a warning column with a portable blocking roadblock. The warning column serves as the control center of the system, can automatically identify approaching vehicles, send out warning signals, and synchronize with the roadblock equipment, thereby realizing the automated auxiliary operation of the traffic road blocking process and enhancing the efficiency and safety of the overall operation. The specific steps are as follows:

[0008] Step 1: Data collection and analysis of sensors in warning columns

[0009] 1. Sensor detection

[0010] Camera: Use the camera installed on the warning column to collect the image and video data of the vehicle in real time; the camera can capture the specific image information of the approaching vehicle;

[0011] LiDAR: LiDAR can scan the surrounding environment with laser beams, accurately measure the distance between the vehicle and the warning column, provide vehicle location information, and calculate the vehicle's speed and acceleration based on the returned light signal;

[0012] 2. Types of data collected

[0013] Vehicle image and video data: visual data acquired through cameras;

[0014] Vehicle location: The real-time location of the vehicle determined by LiDAR;

[0015] Speed ​​and acceleration: Calculate the speed and acceleration of the vehicle based on time series data, and provide vehicle movement trend information;

[0016] Distance data: The distance between the vehicle and the warning column is estimated by LiDAR;

[0017] 3. Data Annotation

[0018] Behavior annotation: Automatically annotate the collected data and record the behavior of each vehicle at different time points; typical annotation information includes:

[0019] Whether to slow down: whether the vehicle slows down significantly when approaching the roadblock;

[0020] Speeding: Mark whether the vehicle is speeding according to the speed limit stipulated on the road;

[0021] Distance marking: real-time distance between the vehicle and the warning column or roadblock;

[0022] 4. Data Preprocessing

[0023] Image size normalization: When training a model, the size of the image data needs to be unified, usually scaling the image to a fixed size;

[0024] Color standardization: Normalize the RGB color channels in the image and map the pixel values ​​from the original range (usually 0 to 255) to between 0 and 1 for normalization;

[0025] Data enhancement: In order to improve the generalization ability of the model and avoid overfitting of the model, various forms of data enhancement can be performed on the image data;

[0026] Step 2: Model selection and training

[0027] 1. Vehicle Detection Model

[0028] First, use the target detection algorithm YOLOv8 (You Only Look Once) to detect and identify vehicles;

[0029] Input data: The input data is an image or video frame with a size of W×H×3 (i.e., width, height, and three RGB color channels). YOLOv8 divides the entire input image into multiple grid cells, each of which is responsible for detecting the target belonging to its area in the image;

[0030] Grid division and prediction: YOLOv8 predicts multiple bounding boxes (anchor boxes) for each grid cell, and predicts the following values ​​for each bounding box: the center coordinates of the bounding box (x, y), the width and height of the bounding box (w, h), the confidence score c of each bounding box (i.e., the probability that the bounding box contains an object), and the category probability distribution p (i.e., the probability of the category to which the target belongs);

[0031] The bounding box of the YOLO model uses the following parameterization:

[0032] Center coordinates (b x ,b y ):

[0033] b x =σ(t x )+c x ,b y =σ(t y )+c y

[0034] Among them, t x and t y is the offset predicted by the network, σ is the Sigmoid function, c x and c y is the coordinate of the upper left corner of the grid cell, b x and b y are the center coordinates;

[0035] Width and height (b w ,bh ):

[0036]

[0037] Among them, p w and p h is the width and height of the anchor box, t w and t h is the scaling parameter predicted by the network, e is an exponential function, and b w and b h is the width and height;

[0038] Loss function: YOLOv8 uses a multi-task loss function to simultaneously optimize classification, positioning, and confidence scores. The loss function can be expressed as:

[0039] L=L bbox +L conf +L cls

[0040]

[0041] Among them, L bbox is the bounding box localization loss, is the bounding box predicted by the model, is the true bounding box, L conf is the confidence loss, y (i) is the true label, p (i) is the confidence of the model prediction, L cls is the classification loss, C is the number of categories, yc(i) is the category label, is the probability that the model predicts the category, i is the index label, log is the logarithmic function, x, y, w, h are the horizontal coordinate, vertical coordinate, width and height of the center of the bounding box, and true is the true label;

[0042] Reasoning process: After training, the YOLOv8 model can directly output the bounding box, category, and confidence of the object in each grid cell. The steps of the reasoning process include:

[0043] Forward propagation: forward propagates the input image through the model to obtain the prediction results for each grid cell;

[0044] Confidence threshold screening: By setting the confidence threshold, high-confidence prediction boxes are screened out;

[0045] Non-maximum suppression (NMS): For predicted overlapping boxes, the NMS algorithm is used to remove duplicate boxes and retain the prediction results with the highest confidence;

[0046] Step 3: Build a behavior prediction model

[0047] After the vehicle detection is completed, the system needs to further predict the vehicle's behavior to determine whether the vehicle will slow down, exceed the speed limit, or enter a dangerous area. This is to more accurately control the timing of the roadblock deployment and ensure that the vehicle is stopped at the right moment.

[0048] 1. Input features

[0049] Multiple features X = (x1, x2, ..., x n ), x1: vehicle speed, x2: vehicle distance, x3: vehicle acceleration, x n : Other sensor data;

[0050] Output: Classification result y: y = 1: The vehicle is about to enter the dangerous area and the roadblock needs to be deployed; y = 0: The vehicle is in normal condition and no roadblock needs to be deployed;

[0051] 2. Construction of decision tree

[0052] The random forest consists of multiple decision trees, each of which performs classification tasks independently. The construction process of each decision tree is as follows:

[0053] Segmentation and conditional judgment: Each decision tree is segmented according to the data features, and the best features and segmentation points are selected. For each feature x j , the decision tree divides the data into two parts according to a certain threshold value, forming left and right branches. For a certain feature x j , there is a split point s, select the optimal feature x j and the split point s is achieved by maximizing the information gain or minimizing the Gini index;

[0054] Gini Index: The Gini Index is a commonly used splitting metric used to measure the purity of a node. The calculation formula for the Gini Index G is:

[0055]

[0056] Among them, p k is the probability of category k, C is the number of categories,

[0057] In vehicle behavior prediction, the category is binary (0 or 1), indicating whether the vehicle will enter a dangerous area; Information Gain: Information gain measures the degree of improvement in data purity before and after splitting under a certain feature. The calculation formula of information gain IG is:

[0058]

[0059] Among them, H(D) is the entropy of the data set D, which represents the uncertainty in the data set, |D i∣ is the size of the subset, D parent is the dataset before splitting, H(D i ) is the entropy of the ith child node;

[0060] Random forest achieves better prediction results by integrating multiple decision trees. The specific steps are as follows: Bootstrap sampling: When training random forest, random forest randomly samples (with replacement) from the original training data set to generate multiple different subsets, each of which is used to train a decision tree; Random feature selection: During the construction of each decision tree, the random forest algorithm randomly selects a part of all features for segmentation; This random feature selection mechanism can prevent the correlation between decision trees from being too high, thereby improving the generalization ability of the model;

[0061] Growth of decision trees: Each decision tree is trained based on the subset it samples and grows until there is no further partitioning space (i.e. there is only one category in the node), without pruning; each decision tree makes classification judgments independently;

[0062] Prediction of random forest: In the prediction stage, random forest integrates the prediction results of multiple decision trees to make the final classification judgment;

[0063] Prediction of a single decision tree: Each decision tree is based on the input feature X = (x1, x2, ..., x n ), make a prediction and get a classification result y t , where t represents the tth decision tree;

[0064] Majority voting mechanism: Random forest determines the final prediction output by performing majority voting on the classification results of all decision trees. It is determined by the voting results of all decision trees:

[0065]

[0066] Where T is the total number of decision trees, and majority_vote represents the voting mechanism;

[0067] Model loss function:

[0068]

[0069] in, represents the indicator function, is the prediction result of the model, y i is the true label, N is the total number of samples, and i is the label index;

[0070] Step 4: Decision rule setting

[0071] In the vehicle behavior prediction based on the random forest model, the purpose of the algorithm is to make a classification judgment by combining the input features, so as to decide whether to deploy the vehicle blocking roadblock; in the intelligent vehicle blocking roadblock system, reasonable threshold setting is crucial to ensure the safety and accuracy of the system. By setting the speed threshold of the vehicle to Q km / h and the distance threshold to P meters, in the urban section, the system can effectively judge the vehicles driving at high speed and approaching the dangerous area, and deploy the roadblock in time to block them;

[0072] The decision rule process is as follows:

[0073] Input feature collection: Real-time collect features such as the speed v(t), distance d(t), and acceleration a(t) of the vehicle to form an input feature vector X(t) = (v(t), d(t), a(t));

[0074] Random forest classification: Input the feature vector X(t) into the trained random forest model. The random forest model outputs the classification result y^(t) through the majority voting mechanism, that is, whether the vehicle enters the dangerous area;

[0075] Decision rule application: According to the set speed and distance thresholds, check the current state of the vehicle: If v(t) > Q and d(t) < P, it is judged that the vehicle enters the dangerous area, and the system issues an instruction to deploy the roadblock; otherwise, maintain the monitoring state and wait for further input data update;

[0076] Step Five: Intelligent Vehicle Blocking Roadblock System

[0077] Signal sending (control part)

[0078] Signal generation:

[0079] Controller decision: The controller decides to send an instruction to deploy the roadblock according to the output result of the algorithm;

[0080] Generate a control signal: The controller generates a signal for the deployment instruction; this signal includes: Deployment command: Notify the vehicle blocking roadblock system to start; Execution parameters: The servo motor follows the received execution instruction;

[0081] Signal transmission

[0082] In the intelligent vehicle blocking roadblock system, signal transmission is a key link for the controller to send the deployment instruction to the vehicle blocking roadblock;

[0083] Wireless communication module: The controller of the warning column has the characteristics of low power consumption, short-distance communication, low latency and self-organizing network capability through the built-in wireless communication module ZigBee; ZigBee has the ability of self-organizing network, which can form a stable mesh network between the warning column and the roadblock system to ensure the stability and anti-interference of signal transmission. The warning column controller sends the generated control signal to the vehicle blocking roadblock system wirelessly through its built-in ZigBee module. The ZigBee module will encode the control signal and transmit it to the receiving device through wireless frequency;

[0084] Multi-node connection and stable transmission: ZigBee supports multi-node connection, which means that there are multiple warning posts and roadblocks on the same road section. They can form a self-organizing mesh network through ZigBee modules. ZigBee can ensure that the control signal can be reliably transmitted to the roadblock system even when the signal is interfered with through its mesh network and self-recovery capabilities;

[0085] Detailed process of signal transmission:

[0086] The warning column controller generates a signal: When the warning column detects that the approaching vehicle exceeds the speed threshold or distance threshold, the intelligent algorithm in the controller generates a command signal to deploy the roadblock;

[0087] ZigBee module encoding and transmission: The ZigBee module of the warning column encodes the control signal into a radio signal and sends the signal to the receiving module of the vehicle barrier system using a 2.4GHz wireless frequency;

[0088] The barrier receives and decodes the signal: After the ZigBee receiving module of the barrier receives the signal from the warning post, it decodes it and transmits it to the main controller of the barrier; the controller interprets the signal, identifies the operation to be performed, and activates the corresponding mechanical device;

[0089] Signal reception

[0090] The roadblock receives the signal, and the controller of the roadblock system verifies the received signal to ensure the integrity and validity of the signal; once the signal is confirmed to be correct, the controller enters the execution preparation stage, notifies the system to start the servo motor and start physical operation;

[0091] Servo motor start

[0092] The controller issues an execution command: After receiving a valid deployment command, the controller of the vehicle barrier system converts the wireless signal into a physical operation command. The controller drives the servo motor through the internal circuit and instructs it to start working and unlock the limit plate.

[0093] Servo motor start: The servo motor starts running according to the command of the controller. The servo motor usually performs a specific mechanical action by rotating;

[0094] Unlocking the limit plate: The task of the servo motor is to control the locking mechanism of the limit plate by driving the rotating connecting bolt (at this time, the motor shaft of the servo motor is connected to the connecting bolt); the rotation of the servo motor will loosen the bolt and release the buckle, and the rotating action of the servo motor will release the locking relationship between the limit plate and the storage box, so that the limit plate can rotate freely. Once the locking mechanism of the limit plate is released, the limit plate will be ready to rotate and unfold, and the chain layout operation will begin;

[0095] The intelligent roadblock system can effectively convert wireless signals into mechanical actions, ensuring that the roadblock system can quickly deploy and lay tire-breaking chains according to the instructions received, and quickly complete the blocking task.

[0096] A portable vehicle blocking barrier of the present invention comprises a storage box and a cover plate arranged at one end of the storage box, wherein the size of the cover plate is the same as the size of the placement slot, so that the cover plate can be built into the placement slot to protect and resist the tire-breaking chain inside the storage box, the cover plate and the storage box are rotatable structures, and a tire-breaking chain is arranged inside the storage box;

[0097] Among them, the storage box is provided with a limit plate above the cover plate, and a connecting bolt is provided at the intersection of the limit plate and the storage box, and the limit plate is rotatable around the connecting bolt as the axis, and the tire-breaking chain inside the storage box squeezes the cover plate, so the tire-breaking chain can be quickly deployed, and the storage box and the cover plate adopt a flip structure, so that the overall design is compact and easy to carry, and a limit plate is set at the intersection of the storage box and the cover plate, and one end of the limit plate is connected to the storage box and can be rotated around the axis of the connection to achieve quick unlocking of the cover plate and the storage box, and only a simple action is needed to unlock the limit plate and throw out the internal tire-breaking chain, which will impact the cover plate, and the cover plate will automatically unfold without additional operation, so as to achieve quick deployment.

[0098] Preferably, the storage box is provided with a fixed plate at one end of the cover plate, and a limiting column is provided on the surface wall of one end of the cover plate. When the cover plate rotates, the limiting column is clamped with the clamping pad, the fixed plate is fixed to the storage box, a clamping pad is embedded in the fixed plate, and a fixing plate is provided at one end of the cover plate of the storage box, and a clamping pad is provided inside the fixed plate. When the cover plate is impacted, the limiting column on the surface wall of the cover plate is clamped with the clamping pad inside the fixed plate, so as to avoid the cover plate from rebounding when subjected to a large impact, resulting in the inconvenience of laying the internal tire chain.

[0099] Preferably, the tire-breaking chain comprises a first support rod and a second support rod, the first support rod and the second support rod are connected in sequence, and when the tire-breaking chain is unfolded, the first support rod and the second support rod are separated in sequence, and a connecting shaft is provided at the intersection of the first support rod and the second support rod;

[0100] Wherein, a plurality of tire-breaking spikes are arranged on the upper surface of the second support rod, and the plurality of tire-breaking spikes are arranged on the upper surface of the second support rod in sequence and at equal intervals, and the vehicle can be intercepted by the plurality of tire-breaking spikes.

[0101] Preferably, the cover plate is provided with a rotating shaft at one end of the limiting column, and the storage box is provided with a connecting hole at a position corresponding to the rotating shaft, the rotating shaft is connected to the connecting hole, the cover plate is connected to the storage box through the rotating shaft, and the cover plate rotates around the rotating shaft, the length of the rotating shaft is greater than the height of the cover plate, the rotating shaft passes through the cover plate, and the rotating shaft located on the outer wall of the cover plate and the connecting hole inside the storage box are connected to each other.

[0102] Preferably, a placement groove is provided inside the storage box, and the tire chain is placed inside the placement groove, and the tire chain is stored through the placement groove. A handle is provided on one side of the storage box, and the storage box is easy to carry through the setting of the handle.

[0103] The advantages of the embodiments of the present application are:

[0104] 1. The storage box and the cover adopt a flip structure, which makes the overall design compact and easy to carry. A limit plate is set at the intersection of the storage box and the cover. One end of the limit plate is connected to the storage box and can rotate around the axis of the connection to achieve quick unlocking of the cover and the storage box. Only a simple action is needed to unlock the limit plate and throw out the internal tire-breaking chain. The tire-breaking chain impacts the cover, and the cover automatically unfolds without additional operation, so that quick deployment can be achieved.

[0105] 2. A fixed plate is provided at one end of the cover of the storage box, and a clamping pad is provided inside the fixed plate. When the cover is impacted, the limit column on the surface wall of the cover is clamped with the clamping pad inside the fixed plate to prevent the cover from rebounding when subjected to a large impact, resulting in inconvenience in laying the internal tire chain.

[0106] 3. The high-strength, lightweight storage box material and the built-in tire-breaking chain design make the roadblock extremely portable and easy to carry. At the same time, the efficient tire-breaking effect of the tire-breaking chain provides effective support for traffic control, improves the success rate and safety of traffic control, and adapts to the rapid response needs of various traffic controls.

[0107] 4. The present invention proposes a method for blocking vehicles on a traffic road based on a dangerous situation perception algorithm, and realizes a highly automated blocking operation through the intelligent linkage of warning columns and blocking roadblocks. The warning column serves as the control center of the system, and uses sensors such as cameras and laser radars to detect the speed, distance and behavior of approaching vehicles in real time. The vehicle is identified by the YOLOv8 target detection model, and the potential danger of the vehicle is predicted by the random forest algorithm. When the vehicle exceeds the speed limit or enters a dangerous area (at this time, there is a certain risk of being hit by laying roadblocks manually. Roadblocks can be laid manually when the risk is controllable. In the case of uncontrollable risks or emergencies, the blocking roadblock system will cause the tire chain to automatically pop open, which is convenient for placing roadblocks on the road), the controller sends a deployment signal through the ZigBee wireless module to trigger the blocking roadblock system. After receiving the signal, the servo motor unlocks the limit plate, and the cover plate can be automatically deployed and the tire chain can be laid. This design greatly improves the response speed and deployment accuracy of the roadblock system, reduces the misjudgment rate, enhances the safety and effectiveness of the operation, and is suitable for emergency blocking tasks in a variety of complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0108] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0109] Figure 1 This is a schematic diagram of the overall storage structure of the portable vehicle-blocking roadblock of the present invention;

[0110] Figure 2 This is a schematic diagram of the overall structure of the portable vehicle-blocking roadblock of the present invention;

[0111] Figure 3 This is a schematic diagram of the overall structure of the portable vehicle blocking barrier of the present invention with the cover removed;

[0112] Figure 4 This is a schematic diagram of the overall structure of the cover plate in the portable vehicle-blocking roadblock of the present invention;

[0113] Figure 5 This is a schematic diagram of the overall structure of the limit plate in the portable vehicle-blocking roadblock of the present invention;

[0114] Figure 6 This is a schematic diagram of the overall structure of the tire-breaking chain in the portable vehicle-blocking roadblock of the present invention;

[0115] Figure 7 The present invention is a flow chart of an intelligent algorithm for deploying a tire-breaking chain in a portable vehicle-blocking roadblock.

[0116] Description of main reference numerals:

[0117] 1. Storage box; 2. Cover plate; 3. Tire-breaking chain; 31. First support rod; 32. Second support rod; 33. Connecting shaft; 34. Tire-breaking nail; 4. Limiting plate; 5. Handle; 6. Fixing plate; 7. Placement slot; 8. Connecting hole; 9. Limiting column; 10. Rotating shaft; 11. Clamping pad. DETAILED DESCRIPTION

[0118] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, for the convenience of description below, the referenced "upper", "lower", "left", and "right" are the same as the upper, lower, left, and right directions of the drawings themselves. The "first", "second", etc. below are distinguished for the description and have no other special meanings.

[0119] like Figure 1-6 The embodiment of the present application solves the problems in the prior art by providing a portable vehicle blocking roadblock and a method for using the same. The storage box and the cover plate adopt a flip structure, which makes the overall design compact and easy to carry. A limit plate is arranged at the intersection of the storage box and the cover plate. One end of the limit plate is connected to the storage box and can be rotated around the axis of the connection to realize quick unlocking of the cover plate and the storage box. Only a simple action is required to unlock the limit plate, and the internal tire-breaking chain is thrown out. The tire-breaking chain impacts the cover plate, and the cover plate is automatically unfolded, and no additional operation is required to realize quick deployment. There is a fixed plate with a clip-on pad inside the fixed plate. When the cover plate is impacted, the limit column on the surface wall of the cover plate is clipped with the clip-on pad inside the fixed plate to avoid the cover plate from rebounding when it is subjected to a large impact, which would cause the inconvenience in laying the internal tire-breaking chain. The use of high-strength and lightweight storage box materials and a built-in tire-breaking chain design makes the roadblock extremely portable and easy to carry. At the same time, the efficient tire-breaking effect of the tire-breaking chain provides effective support for traffic control, improves the success rate and safety of traffic control, and adapts to the rapid response needs of various traffic controls.

[0120] The technical solution in the embodiment of the present application is to solve the above problems, and the overall idea is as follows:

[0121] Embodiment 1: This embodiment provides a specific structure of a portable roadblock, such as Figure 1-6 As shown, it includes a storage box 1 and a cover plate 2 arranged at one end of the storage box 1. The size of the cover plate 2 is the same as the size of the placement slot 7, so that the cover plate 2 can be built into the placement slot 7 to protect the tire-breaking chain 3 inside the storage box 1. The cover plate 2 and the storage box 1 are rotatable structures, and the storage box 1 is provided with a tire-breaking chain 3;

[0122] Among them, the storage box 1 is provided with a limit plate 4 above the cover plate 2, and a connecting bolt is provided at the intersection of the limit plate 4 and the storage box 1 (as an embodiment, the connecting bolt can be rotated manually or driven to rotate by a servo motor controlled by a situation awareness algorithm), the limit plate 4 is rotatable around the connecting bolt as the axis, the tire breaking chain 3 inside the storage box 1 squeezes the cover plate 2, and the tire breaking chain 3 can be quickly deployed, the storage box 1 and the cover plate 2 adopt a flip structure, so that the overall design is compact and easy to carry, a limit plate 4 is set at the intersection of the storage box 1 and the cover plate 2, one end of the limit plate 4 is connected to the storage box 1, and can rotate around the axis of the connection to achieve quick unlocking of the cover plate 2 and the storage box 1, only a simple action is needed to unlock the limit plate 4, and the internal tire breaking chain 3 is thrown out, the tire breaking chain 3 impacts the cover plate 2, and the cover plate 2 is automatically unfolded, and no additional operation is required, so that quick deployment is achieved.

[0123] A fixing plate 6 is provided at one end of the storage box 1 at the cover 2, and a limiting column 9 is provided on the surface wall of one end of the cover 2. When the cover 2 rotates, the limiting column 9 is clamped with the clamping pad 11, and the fixing plate 6 is fixed to the storage box 1. A clamping pad 11 is embedded inside the fixing plate 6. A fixing plate 6 is provided at one end of the storage box 1 at the cover 2, and a clamping pad 11 is provided inside the fixing plate 6. When the cover 2 is impacted, the limiting column 9 on the surface wall of the cover 2 is clamped with the clamping pad 11 inside the fixing plate 6, so as to avoid rebound of the cover 2 when subjected to a large impact, which makes it inconvenient to lay the internal tire chain 3.

[0124] The tire-breaking chain 3 includes a first support rod 31 and a second support rod 32. The first support rod 31 and the second support rod 32 are connected in sequence. When the tire-breaking chain 3 is unfolded, the first support rod 31 and the second support rod 32 are separated in sequence. A connecting shaft 33 is provided at the intersection of the first support rod 31 and the second support rod 32.

[0125] A plurality of tire-breaking spikes 34 are disposed on the upper surface of the second support rod 32 . The plurality of tire-breaking spikes 34 are arranged on the upper surface of the second support rod 32 at equal intervals. The plurality of tire-breaking spikes 34 can intercept a vehicle.

[0126] A rotating shaft 10 is provided at one end of the cover plate 2 at the limiting column 9, and a connecting hole 8 is opened at the corresponding position of the rotating shaft 10 in the storage box 1. The rotating shaft 10 is connected to the connecting hole 8. The cover plate 2 is connected to the storage box 1 through the rotating shaft 10, and the cover plate 2 rotates around the rotating shaft 10 as the axis. The length of the rotating shaft 10 is greater than the height of the cover plate 2. The rotating shaft 10 passes through the cover plate 2, and the rotating shaft 10 located on the outer wall of the cover plate 2 and the internal connecting hole 8 of the storage box 1 are connected to each other.

[0127] A placement groove 7 is provided inside the storage box 1, and the tire chain 3 is placed inside the placement groove 7. The tire chain 3 is stored through the placement groove 7. A handle 5 is provided on one side of the storage box 1. The setting of the handle 5 makes it easy to carry the storage box 1.

[0128] By adopting the above technical solution:

[0129] The storage box 1 and the cover plate 2 adopt a flip structure, which makes the overall design compact and easy to carry. A limit plate 4 is set at the intersection of the storage box 1 and the cover plate 2. One end of the limit plate 4 is connected to the storage box 1 and can rotate around the axis of the connection to realize quick unlocking of the cover plate 2 and the storage box 1. Only a simple action is needed to unlock the limit plate 4 and throw out the internal tire-breaking chain 3. The tire-breaking chain 3 impacts the cover plate 2, and the cover plate 2 automatically unfolds without additional operation, so that rapid deployment can be achieved. In addition, a fixing plate 6 is provided at one end of the cover plate 2 of the storage box 1, and a clamping pad 11 is provided inside the fixing plate 6. When the cover plate 2 is impacted, the limit column 9 on the surface wall of the cover plate 2 is clamped with the clamping pad 11 inside the fixing plate 6 to avoid rebound of the cover plate 2 when it is subjected to a large impact, which leads to the problem of inconvenience in laying the internal tire-breaking chain 3.

[0130] Embodiment 2: This embodiment provides a specific structure of a portable roadblock, such as Figure 1-6 As shown, it includes a storage box 1 and a cover plate 2 arranged at one end of the storage box 1. The size of the cover plate 2 is the same as the size of the placement slot 7, so that the cover plate 2 can be built into the placement slot 7 to protect the tire-breaking chain 3 inside the storage box 1. The cover plate 2 and the storage box 1 are rotatable structures, and the storage box 1 is provided with a tire-breaking chain 3;

[0131] Among them, the storage box 1 is provided with a limit plate 4 above the cover 2, and a connecting bolt is provided at the intersection of the limit plate 4 and the storage box 1. The limit plate 4 can rotate around the connecting bolt as the axis, and the tire-breaking chain 3 inside the storage box 1 squeezes the cover 2, so the tire-breaking chain 3 can be quickly deployed. The storage box 1 and the cover 2 adopt a flip structure, which makes the overall design compact and easy to carry. A limit plate 4 is set at the intersection of the storage box 1 and the cover 2, and one end of the limit plate 4 is connected to the storage box 1 and can rotate around the axis of the connection to achieve quick unlocking of the cover 2 and the storage box 1. Only a simple action is needed to unlock the limit plate 4, and the internal tire-breaking chain 3 is thrown out. The tire-breaking chain 3 impacts the cover 2, and the cover 2 is automatically unfolded. No additional operation is required, and quick deployment is achieved.

[0132] The tire chain 3 includes a first support rod 31 and a second support rod 32. The ends of the multiple first support rods 31 and the ends of the second support rods 32 are connected in sequence at a certain angle. When the tire chain 3 is unfolded, the first support rod 31 and the second support rod 32 are separated in sequence. A connecting shaft 33 is provided at the intersection of the first support rod 31 and the second support rod 32 (the intersection point of the staggered arrangement). As an embodiment, the first end of the placement groove 7 is one end of the first support rod 31 and the second support rod 32; the end of the first support rod 31 is The tail end of the second support rod 32 is connected to the two ends of the V-shaped rod (the two sections of the V-shaped rod are adjustable along the common fulcrum angle). The above-mentioned tire chain 3 is required to maintain outward pressure on the cover plate 2 when in a compressed and folded state (when stored in the storage box 1). As a further improved embodiment, a spring device (multiple springs are evenly distributed in the same direction toward the cover plate 2) can be set on one side of the inside of the storage box 1. After the limit plate 4 is rotated and unlocked, the spring releases pressure in the direction of the cover plate 2, thereby pushing the tire chain 3 out of the storage box 1. As an embodiment, at this time, one end of the spring is connected to the inner side of the storage box 1, and the other end of the spring is connected to the internal cover plate. The internal cover plate is hinged to the first support rod 31 and the second support rod 32 at the very beginning. This embodiment facilitates the algorithm to realize the automatic operation of the traffic road blocking process.

[0133] A plurality of tire-breaking spikes 34 are disposed on the upper surface of the second support rod 32 . The plurality of tire-breaking spikes 34 are arranged on the upper surface of the second support rod 32 at equal intervals. The plurality of tire-breaking spikes 34 can intercept a vehicle.

[0134] A rotating shaft 10 is provided at one end of the cover plate 2, and a connecting hole 8 is opened at the position corresponding to the rotating shaft 10 in the storage box 1. The rotating shaft 10 is connected to the connecting hole 8. The cover plate 2 is connected to the storage box 1 through the rotating shaft 10, and the cover plate 2 rotates around the rotating shaft 10. The length of the rotating shaft 10 is greater than the height of the cover plate 2. The rotating shaft 10 passes through the cover plate 2, and the rotating shaft 10 located on the outer wall of the cover plate 2 and the internal connecting hole 8 of the storage box 1 are connected to each other.

[0135] A placement groove 7 is provided inside the storage box 1, and the tire chain 3 is placed inside the placement groove 7. The tire chain 3 is stored through the placement groove 7. A handle 5 is provided on one side of the storage box 1. The setting of the handle 5 makes it easy to carry the storage box 1.

[0136] By adopting the above technical solution:

[0137] The storage box 1 and the cover 2 adopt a flip structure, which makes the overall design compact and easy to carry. A limit plate 4 is set at the intersection of the storage box 1 and the cover 2. One end of the limit plate 4 is connected to the storage box 1 and can rotate around the axis of the connection to achieve quick unlocking of the cover 2 and the storage box 1. Only a simple action is needed to unlock the limit plate 4 and throw out the internal tire-breaking chain 3. The tire-breaking chain 3 impacts the cover 2, and the cover 2 is automatically unfolded without additional operation, thereby achieving quick deployment.

[0138] Embodiment 3: This embodiment provides a method for using a portable roadblock, such as Figure 1-6 As shown, the steps of using the portable vehicle barrier are as follows:

[0139] Step 1: Rotate the limit plate 4 (as an embodiment, manually rotate according to the urgency of the vehicle blocking or drive the servo motor to rotate the limit plate 4 through the controller), the limit plate 4 rotates around the storage box 1 as the axis, the operator swings the storage box 1, and the tire chain 3 inside the storage box 1 presses the cover plate 2;

[0140] Step 2: The cover plate 2 is impacted, and the limiting columns 9 on the surface wall of the cover plate 2 are plugged into the clamping pads 11 inside the fixing plate 6 to connect the cover plate 2 with the fixing plate 6.

[0141] Step 3: Throw out the internal tire-breaking chain 3, which impacts the cover plate 2, and the cover plate 2 is automatically unfolded without additional operation, so that rapid deployment can be achieved.

[0142] Step 4: Through reverse operation, the tire chain 3 is stored in the storage box 1, and then the limit plate 4 is locked to return the tire chain 3 to the storage box 1 to achieve rapid recovery.

[0143] By adopting the above technical solution:

[0144] The limiting plate 4 is rotated, and the limiting plate 4 rotates about the axis of the storage box 1, and the storage box 1 is swung. Since the tire-breaking chain 3 inside the storage box 1 squeezes the cover plate 2, the cover plate 2 is impacted, and the limiting column 9 on the surface wall of the cover plate 2 is plugged into the clamping pad 11 inside the fixed plate 6, which is used to connect the cover plate 2 with the fixed plate 6, and the internal tire-breaking chain 3 is swung out. The tire-breaking chain 3 impacts the cover plate 2, and the cover plate 2 is automatically unfolded without additional operation, so that rapid deployment can be achieved. Through reverse operation, the tire-breaking chain 3 is stored inside the storage box 1, and then the limiting plate 4 is locked, so that the tire-breaking chain 3 is returned to the storage box 1 to achieve rapid recovery.

[0145] like Figure 7 , a method for blocking vehicles on a traffic road based on a dangerous situation awareness algorithm is proposed. The situation awareness intelligent algorithm combines warning columns with portable blocking roadblocks. The warning column, as the control center of the system, can automatically identify approaching vehicles, send out warning signals, and synchronize with the roadblock equipment to realize the automated auxiliary operation of the traffic road blocking process, enhancing the efficiency and safety of the overall operation; the specific steps are as follows:

[0146] Step 1: Data collection and analysis of sensors in warning columns

[0147] 1. Sensor detection

[0148] Camera: Use the camera installed on the warning column to collect the image and video data of the vehicle in real time; the camera can capture the specific image information of the approaching vehicle;

[0149] LiDAR: LiDAR can scan the surrounding environment with laser beams, accurately measure the distance between the vehicle and the warning column, provide vehicle location information, and calculate the vehicle's speed and acceleration based on the returned light signal;

[0150] 2. Types of data collected

[0151] Vehicle image and video data: visual data acquired by the camera, image resolution: 1080p (1920×1080 pixels), frame rate: 30 frames per second, capturing the initial position of the vehicle: 100 meters from the warning column, timestamp: t=0 seconds;

[0152] Vehicle location: The real-time location of the vehicle determined by LiDAR;

[0153] Speed ​​and acceleration: Calculate the speed and acceleration of the vehicle based on time series data, and provide vehicle movement trend information. The initial distance d0 = 100 meters, the initial speed of the vehicle v0 = 60km / h, and the vehicle acceleration a = 2m / s 2 ;

[0154] Distance data: The distance between the vehicle and the warning column is estimated by LiDAR;

[0155] 3. Data Annotation

[0156] Behavior annotation: Automatically annotate the collected data and record the behavior of each vehicle at different time points; typical annotation information includes:

[0157] Whether to slow down: whether the vehicle slows down significantly when approaching the roadblock;

[0158] Speeding: Mark whether the vehicle is speeding according to the speed limit stipulated on the road;

[0159] Distance marking: real-time distance between the vehicle and the warning column or roadblock;

[0160] 4. Data Preprocessing

[0161] Image size normalization: When training the model, the size of the image data needs to be unified. Usually, the image is scaled to a fixed size, and the image is scaled to 416×416 pixels (the standard size of YOLOv8 input);

[0162] Color standardization: Normalize the RGB color channels in the image and map the pixel values ​​from the original range (usually 0 to 255) to between 0 and 1 for normalization;

[0163] Data enhancement: In order to improve the generalization ability of the model and avoid overfitting of the model, various forms of data enhancement can be performed on the image data;

[0164] Step 2: Model selection and training

[0165] 1. Vehicle Detection Model

[0166] First, use the target detection algorithm YOLOv8 (You Only Look Once) to detect and identify vehicles;

[0167] Input data: The input data is an image or video frame with a size of W×H×3 (i.e., width, height, and three RGB color channels). YOLOv8 divides the entire input image into multiple grid cells, each of which is responsible for detecting the target belonging to its area in the image;

[0168] Grid division and prediction: YOLOv8 predicts multiple bounding boxes (anchor boxes) for each grid cell, and predicts the following values ​​for each bounding box: the center coordinates of the bounding box (x, y), the width and height of the bounding box (w, h), the confidence score c of each bounding box (i.e., the probability that the bounding box contains an object), and the category probability distribution p (i.e., the probability of the category to which the target belongs);

[0169] The bounding box of the YOLO model uses the following parameterization:

[0170] Center coordinates (b x ,b y ):

[0171] b x =σ(t x )+c x ,b y =σ(t y )+c y

[0172] Among them, t x and t y is the offset predicted by the network, σ is the Sigmoid function, c x and c y is the coordinate of the upper left corner of the grid cell, b x and b y are the center coordinates;

[0173] Width and height (b w ,b h ):

[0174]

[0175] Among them, p wand p h is the width and height of the anchor box, t w and t h is the scaling parameter predicted by the network, e is an exponential function, and b w and b h is the width and height;

[0176] Loss function: YOLOv8 uses a multi-task loss function to simultaneously optimize classification, positioning, and confidence scores. The loss function can be expressed as:

[0177] L=L bbox +L conf +L cls

[0178]

[0179] Among them, L bbox is the bounding box localization loss, is the bounding box predicted by the model, is the true bounding box, L conf is the confidence loss, y (i) is the true label, p (i) is the confidence of the model prediction, L cls is the classification loss, C is the number of categories, yc(i) is the category label, is the probability that the model predicts the category, i is the index label, log is the logarithmic function, x, y, w, h are the horizontal coordinate, vertical coordinate, width and height of the center of the bounding box, and true is the true label;

[0180] Reasoning process: After training, the YOLOv8 model can directly output the bounding box, category, and confidence of the object in each grid cell. The steps of the reasoning process include:

[0181] Forward propagation: forward propagates the input image through the model to obtain the prediction results for each grid cell;

[0182] Confidence threshold screening: By setting the confidence threshold, high-confidence prediction boxes are screened out;

[0183] Non-maximum suppression (NMS): For predicted overlapping boxes, the NMS algorithm is used to remove duplicate boxes and retain the prediction results with the highest confidence;

[0184] After detection, the vehicle bounding box is obtained as follows: x =200,b y =150,b w =50,b h =40, confidence level c = 0.95, the predicted category is "car";

[0185] Step 3: Build a behavior prediction model

[0186] After the vehicle detection is completed, the system needs to further predict the vehicle's behavior to determine whether the vehicle will slow down, exceed the speed limit, or enter a dangerous area. This is to more accurately control the timing of the roadblock deployment and ensure that the vehicle is stopped at the right moment.

[0187] 1. Input features

[0188] Multiple features X = (x1, x2, ..., x n ), x1: vehicle speed, x2: vehicle distance, x3: vehicle acceleration, x n : Other sensor data;

[0189] X=(v(t),d(t),a)=(16.67m / s, 62.66 meters, 2m / s 2 )

[0190] Output: Classification result y: y = 1: The vehicle is about to enter the dangerous area and the roadblock needs to be deployed; y = 0: The vehicle is in normal condition and no roadblock needs to be deployed;

[0191] 2. Construction of decision tree

[0192] The random forest consists of 10 decision trees, each of which performs classification tasks independently. The construction process of each decision tree is as follows:

[0193] Segmentation and conditional judgment: Each decision tree is segmented according to the data features, and the best features and segmentation points are selected. For each feature x j , the decision tree divides the data into two parts according to a certain threshold value, forming left and right branches. For a certain feature x j , there is a split point s, select the optimal feature x j and the split point s is achieved by maximizing the information gain or minimizing the Gini index;

[0194] Gini Index: The Gini Index is a commonly used splitting metric used to measure the purity of a node. The calculation formula for the Gini Index G is:

[0195]

[0196] Among them, p k is the probability of category k, C is the number of categories;

[0197] G=1-(0.8 2 +0.2 2 )=1-(0.64+0.04)=0.32

[0198] In vehicle behavior prediction, the category is binary (0 or 1), indicating whether the vehicle will enter a dangerous area; Information Gain: Information gain measures the degree of improvement in data purity before and after splitting under a certain feature. The calculation formula of information gain IG is:

[0199]

[0200] Among them, H(D) is the entropy of the data set D, which represents the uncertainty in the data set, |D i ∣ is the size of the subset, D parent is the dataset before splitting, H(D i ) is the entropy of the ith child node;

[0201] Random forest achieves better prediction results by integrating multiple decision trees. The specific steps are as follows: Bootstrap sampling: When training random forest, random forest randomly samples (with replacement) from the original training data set to generate multiple different subsets, each of which is used to train a decision tree;

[0202] Random feature selection: During the construction of each decision tree, the random forest algorithm randomly selects a portion of all features for segmentation; this random feature selection mechanism can prevent the correlation between decision trees from being too high, thereby improving the generalization ability of the model;

[0203] Growth of decision trees: Each decision tree is trained based on the subset it samples and grows until there is no further partitioning space (i.e. there is only one category in the node), without pruning; each decision tree makes classification judgments independently;

[0204] Prediction of random forest: In the prediction stage, random forest integrates the prediction results of multiple decision trees to make the final classification judgment;

[0205] Prediction of a single decision tree: Each decision tree is based on the input feature X = (x1, x2, ..., x n ), make a prediction and get a classification result y t , where t represents the tth decision tree;

[0206] Majority voting mechanism: Random forest determines the final prediction output by performing majority voting on the classification results of all decision trees. It is determined by the voting results of all decision trees:

[0207]

[0208] Where T is the total number of decision trees, and majority_vote represents the voting mechanism;

[0209] Model loss function:

[0210]

[0211] in, represents the indicator function, is the prediction result of the model, y i is the true label, N is the total number of samples, and i is the label index;

[0212] Step 4: Decision rule setting

[0213] In the vehicle behavior prediction based on the random forest model, the purpose of the algorithm is to make a classification judgment by combining the input features, so as to decide whether to deploy a roadblock. In the intelligent roadblock system, reasonable threshold setting is crucial to ensure the safety and accuracy of the system. By setting the vehicle speed threshold to 80km / h and the distance threshold to 50 meters, the system can effectively judge vehicles traveling at high speed and approaching dangerous areas on urban roads, and deploy roadblocks in time to block them.

[0214] The decision rule process is as follows:

[0215] 1. Input feature collection: collect the vehicle's speed v(t), distance d(t) and acceleration a(t) in real time to form an input feature vector X(t) = (v(t), d(t), a(t));

[0216] 2. Random forest classification: Input the feature vector X(t) into the trained random forest model. The random forest model outputs the classification result y^(t) through the majority voting mechanism, that is, whether the vehicle enters the dangerous area;

[0217] 3. Decision rule application: Check the current state of the vehicle according to the set speed and distance thresholds: If v(t)>80km / h and d(t)<50m, the vehicle is judged to have entered the danger zone and the system issues a command to deploy the roadblock; otherwise, the monitoring state is maintained and further input data is updated;

[0218] At this time, v(t) = 60.01 m / s, d(t) = 62.66 m, which does not meet the deployment conditions, so the system continues to monitor;

[0219] Step 5: Intelligent roadblock system

[0220] 1. Signal sending (control part)

[0221] Signal Generation:

[0222] Controller decision: The controller decides to send a command to deploy the roadblock based on the output of the algorithm;

[0223] Generate a control signal: the controller generates a signal of an expansion instruction; the signal includes: an expansion command: notifying the vehicle blocking roadblock system to start; execution parameters: the servo motor executes the received execution instruction;

[0224] 2. Signal transmission

[0225] In the intelligent roadblock system, signal transmission is the key link for the controller to send deployment instructions to the roadblock;

[0226] Wireless communication module: The controller of the warning column has the characteristics of low power consumption, short-distance communication, low latency and self-organizing network capability through the built-in wireless communication module ZigBee; ZigBee has the ability of self-organizing network, which can form a stable mesh network between the warning column and the roadblock system to ensure the stability and anti-interference of signal transmission. The warning column controller sends the generated control signal to the vehicle blocking roadblock system wirelessly through its built-in ZigBee module. The ZigBee module will encode the control signal and transmit it to the receiving device through wireless frequency;

[0227] Multi-node connection and stable transmission: ZigBee supports multi-node connection, which means that there are multiple warning posts and roadblocks on the same road section. They can form a self-organizing mesh network through ZigBee modules. ZigBee can ensure that the control signal can be reliably transmitted to the roadblock system even when the signal is interfered with through its mesh network and self-recovery capabilities;

[0228] 3. Detailed process of signal transmission:

[0229] The warning column controller generates a signal: When the warning column detects that the approaching vehicle exceeds the speed threshold or distance threshold, the intelligent algorithm in the controller generates a command signal to deploy the roadblock;

[0230] ZigBee module encoding and transmission: The ZigBee module of the warning column encodes the control signal into a radio signal and sends the signal to the receiving module of the vehicle barrier system using a 2.4GHz wireless frequency;

[0231] The barrier receives and decodes the signal: After the ZigBee receiving module of the barrier receives the signal from the warning post, it decodes it and transmits it to the main controller of the barrier; the controller interprets the signal, identifies the operation to be performed, and activates the corresponding mechanical device;

[0232] 4. Signal reception

[0233] The roadblock receives the signal, and the controller of the roadblock system verifies the received signal to ensure the integrity and validity of the signal; once the signal is confirmed to be correct, the controller enters the execution preparation stage, notifies the system to start the servo motor and start physical operation;

[0234] 5. Servo motor start

[0235] The controller issues an execution command: After receiving a valid deployment command, the controller of the vehicle barrier system converts the wireless signal into a physical operation command. The controller drives the servo motor through the internal circuit and instructs it to start working and unlock the limit plate; (The vehicle barrier system consists of a connected receiver, a controller and a driving servo motor)

[0236] Servo motor start: The servo motor starts running according to the command of the controller. The servo motor usually performs specific mechanical actions by rotating. Torque: 10Nm, speed: 300rpm;

[0237] Unlocking the limit plate: The task of the servo motor is to control the locking mechanism of the limit plate by driving the rotating connecting bolts; the rotation of the servo motor will loosen the bolts and release the buckle, and the rotation of the servo motor will release the locking relationship between the limit plate and the storage box, so that the limit plate can rotate freely. Once the locking mechanism of the limit plate is released, the limit plate will be ready to rotate and unfold, and the chain layout operation will begin;

[0238] The intelligent roadblock system can effectively convert wireless signals into mechanical actions, ensuring that the roadblock system can quickly deploy and lay tire-breaking chains according to the instructions received, and quickly complete the blocking task.

[0239] Finally, it should be noted that: Obviously, the above embodiments are only examples for clearly explaining the present invention, and are not intended to limit the implementation methods. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from this are still within the scope of protection of the present invention.

Claims

1. A method for blocking vehicles on a traffic road based on a dangerous situation awareness algorithm, characterized in that: The intelligent situational awareness algorithm combines warning posts with portable roadblocks. The warning posts serve as the control center of the system. They can automatically identify approaching vehicles, send out warning signals, and synchronize with roadblock equipment to achieve automated auxiliary operations in the process of blocking vehicles on the traffic road.

2. A method for blocking vehicles on a traffic road based on a dangerous situation awareness algorithm according to claim 1, characterized in that: The specific steps are as follows: Step 1: Data collection and analysis of sensors in warning columns (1) Sensor detection Camera: The camera installed on the warning column collects the image and video data of the vehicle in real time; the camera captures the specific image information of the approaching vehicle; LiDAR: LiDAR installed on the warning pole can scan the surrounding environment through laser beams, accurately measure the distance between the vehicle and the warning pole, and provide the vehicle's position information. It can also calculate the vehicle's speed and acceleration based on the returned light signal. (2) Data Types Collected Vehicle image and video data: visual data acquired through cameras; Vehicle location: The real-time location of the vehicle determined by LiDAR; Speed ​​and acceleration: Calculate the speed and acceleration of the vehicle based on time series data, and provide vehicle movement trend information; Distance data: The distance between the vehicle and the warning column is estimated by LiDAR; (3) Data labeling Behavior annotation: Automatically annotate the collected data and record the behavior of each vehicle at different time points; Typical annotation information includes: Whether to slow down: whether the vehicle slows down significantly when approaching the roadblock; Speeding: Mark whether the vehicle is speeding according to the speed limit stipulated on the road; Distance marking: real-time distance between the vehicle and the warning column or roadblock; (4) Data preprocessing Image size normalization: When training a model, the size of the image data needs to be unified, usually scaling the image to a fixed size; Color standardization: Normalize the RGB color channels in the image and map the pixel values ​​from the original range of 0 to 255 to between 0 and 1 for normalization; Data enhancement: In order to improve the generalization ability of the model and avoid overfitting of the model, image data can be enhanced; Step 2: Model selection and training (1) Vehicle detection model First, the target detection algorithm YOLOv8 is used to detect and identify vehicles; Input data: The input data is an image or video frame with a size of W×H×3, i.e., width, height, and three RGB color channels. YOLOv8 divides the entire input image into multiple grid cells, each of which is responsible for detecting the target belonging to its area in the image; Grid division and prediction: YOLOv8 predicts multiple bounding boxes for each grid cell and predicts the following values ​​for each bounding box: the center coordinates of the bounding box (x, y), the width and height of the bounding box (w, h), the confidence score c of each bounding box, and the category probability distribution p; The bounding box of the YOLO model uses the following parameterization: Center coordinates (b x ,b y ): b x =σ(t x )+c x ,b y =σ(t y )+c y Among them, t x and t y is the offset predicted by the network, σ is the Sigmoid function, c x and c y is the coordinate of the upper left corner of the grid cell, b x and b y are the center coordinates; Width and height (b w ,b h ): Among them, p w and p h is the width and height of the anchor box, t w and t h is the scaling parameter of the network prediction, and e is an exponential function; Loss function: YOLOv8 uses a multi-task loss function to simultaneously optimize classification, positioning, and confidence scores. The loss function can be expressed as: L=L bbox +L conf +L cls Among them, L bbox is the bounding box localization loss, is the bounding box predicted by the model, is the true bounding box, L conf is the confidence loss, y (i) is the true label, p (i) is the confidence of the model prediction, L cls is the classification loss, C is the number of categories, y c (i) is the category label, is the probability that the model predicts the category, i is the index label, log is the logarithmic function, x, y, w, h are the horizontal coordinate, vertical coordinate, width and height of the center of the bounding box, and true is the true label; Reasoning process: After training, the YOLOv8 model can directly output the bounding box, category, and confidence of the object in each grid cell. The steps of the reasoning process include: Forward propagation: forward propagates the input image through the model to obtain the prediction results for each grid cell; Confidence threshold screening: By setting the confidence threshold, high-confidence prediction boxes are screened out; Non-maximum suppression (NMS): For predicted overlapping boxes, the NMS algorithm is used to remove duplicate boxes and retain the prediction results with the highest confidence. Step 3: Build a behavior prediction model After the vehicle detection is completed, the system needs to further predict the vehicle's behavior to determine whether the vehicle will slow down, exceed the speed limit, or enter a dangerous area. This is to more accurately control the timing of the roadblock deployment and ensure that the vehicle is stopped at the right moment. (1) Input features Multiple features X = (x1, x2, ..., x n ), x1: vehicle speed, x2: vehicle distance, x3: vehicle acceleration, x n : Other sensor data; Output: Classification result y: y = 1: The vehicle is about to enter the dangerous area and the roadblock needs to be deployed; y = 0: The vehicle is in normal condition and no roadblock needs to be deployed; (2) Construction of decision tree The random forest consists of multiple decision trees, each of which performs classification tasks independently. The construction process of each decision tree is as follows: Segmentation and conditional judgment: Each decision tree is segmented according to the data features, and the best features and segmentation points are selected. For each feature x j , the decision tree divides the data into two parts according to a certain threshold value, forming left and right branches. For a certain feature x j , there is a split point s, select the optimal feature x j and the split point s is achieved by maximizing the information gain or minimizing the Gini index; Gini index G: The Gini index is a commonly used splitting metric used to measure the purity of a node. The calculation formula for the Gini index G is: Among them, p k is the probability of category k, C is the number of categories; In vehicle behavior prediction, the category is binary (0 or 1), indicating whether the vehicle will enter a dangerous area; Information gain IG: Information gain measures the degree of improvement in data purity before and after splitting under a certain feature. The calculation formula of information gain IG is: Among them, H(D) is the entropy of the data set D, which represents the uncertainty in the data set, |D i ∣ is the size of the subset, D parent is the dataset before splitting, H(D i ) is the entropy of the ith child node; Random forest achieves better prediction results by integrating multiple decision trees. The specific steps are as follows: Bootstrap sampling: When training random forest, random forest randomly samples from the original training data set to generate multiple different subsets, each of which is used to train a decision tree; Random feature selection: During the construction of each decision tree, the random forest algorithm randomly selects a portion of all features for segmentation; this random feature selection mechanism can prevent the correlation between decision trees from being too high, thereby improving the generalization ability of the model; Growth of decision trees: Each decision tree is trained based on the subset it samples and grows until there is no space for further division, without pruning. Each decision tree makes classification decisions independently. Prediction of random forest: In the prediction stage, random forest integrates the prediction results of multiple decision trees to make the final classification judgment; Prediction of a single decision tree: Each decision tree is based on the input feature X = (x1, x2, ..., x n ), make a prediction and get a classification result y t , where t represents the tth decision tree; Majority voting mechanism: Random forest determines the final prediction output by performing majority voting on the classification results of all decision trees. It is determined by the voting results of all decision trees: Where T is the total number of decision trees, and majority_vote represents the voting mechanism; Model loss function: in, represents the indicator function, is the prediction result of the model, y i is the true label, N is the total number of samples, and i is the label index; Step 4: Decision rule setting In the vehicle behavior prediction based on the random forest model, the purpose of the algorithm is to make a classification judgment by combining the input features, so as to decide whether to deploy a roadblock. In the intelligent roadblock system, reasonable threshold setting is crucial to ensure the safety and accuracy of the system. By setting the vehicle speed threshold to Qkm / h and the distance threshold to P meters, the system can effectively judge vehicles traveling at high speed and approaching dangerous areas in urban roads, and deploy roadblocks in time to block them. The decision rule process is as follows: (1) Input feature collection: Real-time collect features such as the vehicle speed v(t), distance d(t), and acceleration a(t) to form an input feature vector X(t) = (v(t), d(t), a(t)); (2) Forest classification: Input the feature vector X(t) into the trained random forest model. The random forest model outputs the classification result y^(t) through the majority voting mechanism, that is, whether the vehicle enters the dangerous area; (3) Decision rule application: According to the set speed and distance thresholds, check the current state of the vehicle: If v(t) > Q and d(t) < P, it is determined that the vehicle enters the dangerous area, and the system issues an instruction to prompt the operator to deploy the roadblock; otherwise, maintain the monitoring state and wait for further input data updates; Step Five: Intelligent vehicle blocking roadblock system (1) Signal sending Signal generation: Controller decision-making: The controller decides to send an instruction to deploy the roadblock according to the output result of the algorithm; Generate a control signal: The controller generates a signal for the deployment instruction; this signal includes: Deployment command: Notify the vehicle blocking roadblock system to start; Execution parameter: The servo motor follows the received execution instruction; (2) Signal transmission In the intelligent vehicle blocking roadblock system, signal transmission is a key link for the controller to send a deployment instruction to the vehicle blocking roadblock; Wireless communication module: The controller of the warning post uses the built-in wireless communication module ZigBee; ZigBee has the ability to self-organize a network, which can form a stable mesh network between the warning post and the roadblock system to ensure the stability and anti-interference of signal transmission. The controller of the warning post wirelessly sends the generated control signal to the vehicle blocking roadblock system through its built-in ZigBee module. The ZigBee module will encode the control signal and transmit it to the receiving device through the wireless frequency; Multi-node connection and stable transmission: ZigBee supports multi-node connection, which means that there are multiple warning posts and roadblock devices on the same road section. They can form a self-organized mesh network through the ZigBee module; ZigBee can ensure that even when the signal is interfered, the control signal can still be reliably transmitted to the vehicle blocking roadblock system through its mesh network and self-recovery ability; 1) Detailed process of signal transmission: The warning post controller generates a signal: When the warning post detects that the approaching vehicle exceeds the speed threshold or distance threshold, the intelligent algorithm in the controller will generate an instruction signal to deploy the roadblock; ZigBee module encoding and transmission: The ZigBee module of the warning post encodes the control signal into a radio signal and sends the signal to the receiving module of the vehicle blocking roadblock system using a 2.4GHz wireless frequency; The vehicle blocking roadblock receives and decodes the signal: After the ZigBee receiving module of the vehicle blocking roadblock system receives the signal sent by the warning post, it will decode it and transmit it to the main controller of the roadblock; The controller will analyze the signal, identify the operation to be performed, and start the corresponding mechanical device; (2) Signal reception The roadblock receives the signal, and the controller of the roadblock system verifies the received signal to ensure the integrity and validity of the signal; once the signal is confirmed to be correct, the controller enters the execution preparation stage, notifies the system to start the servo motor and start physical operation; 3) Servo motor start The controller issues an execution command: After receiving a valid deployment command, the controller of the vehicle barrier system converts the wireless signal into a physical operation command. The controller drives the servo motor through the internal circuit and instructs it to start working and unlock the limit plate. Servo motor start: The servo motor starts running according to the command of the controller. The servo motor usually performs a specific mechanical action by rotating; Unlocking the limit plate: The task of the servo motor is to control the locking mechanism of the limit plate by driving the rotating connecting bolts; the rotation of the servo motor will loosen the bolts and release the buckle, and the rotation of the servo motor will release the locking relationship between the limit plate and the storage box, so that the limit plate can rotate freely. Once the locking mechanism of the limit plate is released, the limit plate will be ready to rotate and unfold, and the chain layout operation will begin; The intelligent roadblock system can effectively convert wireless signals into mechanical actions, ensuring that the roadblock system can quickly deploy and lay tire-breaking chains according to the instructions received, and quickly complete the blocking task.

3. A portable roadblock, characterized in that: It comprises a storage box (1) and a cover plate (2) arranged at one end of the storage box (1), the cover plate (2) and the storage box (1) are rotatable structures, and a tire-breaking chain (3) is arranged inside the storage box (1); The storage box (1) is provided with a limit plate (4) located above the cover plate (2), and a connecting bolt is provided at the intersection of the limit plate (4) and the storage box (1). The limit plate (4) is rotatable around the connecting bolt as the axis. The tire-breaking chain (3) inside the storage box (1) presses the cover plate (2), so that the tire-breaking chain (3) can be quickly laid out.

4. A portable vehicle barrier as claimed in claim 1, characterized in that: The storage box (1) is provided with a fixing plate (6) at one end of the cover plate (2), the fixing plate (6) is fixed to the storage box (1), and a clamping pad (11) is embedded inside the fixing plate (6); A limiting column (9) is provided on a surface wall at one end of the cover plate (2); when the cover plate (2) rotates, the limiting column (9) is clamped with the clamping pad (11).

5. A portable vehicle barrier as claimed in claim 1, characterized in that: The tire-breaking chain (3) comprises a first support rod (31) and a second support rod (32), and a connecting shaft (33) is provided at the intersection of the first support rod (31) and the second support rod (32); Wherein, a plurality of tire-breaking spikes (34) are provided on the upper surface of the second support rod (32).

6. A portable vehicle barrier as claimed in claim 4, characterized in that: The cover plate (2) is provided with a rotating shaft (10) at one end of the limiting column (9), and the length of the rotating shaft (10) is greater than the height of the cover plate (2).

7. A portable vehicle barrier as claimed in claim 4, characterized in that: The storage box (1) is provided with a connection hole (8) at a position corresponding to the rotating shaft (10), and the rotating shaft (10) is connected to the connection hole (8); a placement groove (7) is provided inside the storage box (1), and a handle (5) is provided on one side of the storage box (1).

8. A portable vehicle barrier as claimed in claim 7, characterized in that: The size of the cover plate (2) is the same as the size of the placement groove (7).

9. A portable vehicle barrier as claimed in claim 5, characterized in that: The first support rod (31) and the second support rod (32) are connected in sequence, and when the tire-breaking chain (3) is unfolded, the first support rod (31) and the second support rod (32) are separated in sequence.

Images