An expressway traffic congestion prediction method and system based on artificial intelligence

Abstract

The application provides a highway traffic congestion prediction method based on artificial intelligence, and belongs to the technical field of highway condition analysis.The method comprises the following steps: S1, a highway is divided into road sections and branches in advance, and cameras and sensors are arranged on the highway; S2, a cloud server collects camera and sensor data of each road section of the highway, and constructs an artificial intelligence algorithm model; S3, the cloud server uses the artificial intelligence algorithm model to predict traffic congestion of each road section of the highway, pushes the prediction result to a user, and corrects the artificial intelligence algorithm model according to the accuracy of the prediction result.The application realizes road condition prediction of the highway according to road sections through the artificial intelligence algorithm model, recommends the road condition to relevant vehicle users, timely identifies congestion prediction, pushes a recommended route to the user before congestion occurs, reduces the influence of congestion as much as possible, and improves the overall traffic efficiency of the highway.

Description

Technical Field

[0001] This invention belongs to the field of highway traffic condition analysis technology, specifically relating to a highway traffic congestion prediction method and system based on artificial intelligence. Background Technology

[0002] As highways are specifically designed for high-speed automobile travel, they provide lanes and directions for vehicles, playing a significant role in improving traffic flow and becoming an important form of transportation. However, with the development of the automotive industry, highway traffic volume is constantly increasing, often leading to congestion. Once congestion occurs, the highway's function of ensuring rapid vehicle movement is compromised, resulting in a significant waste of time and road resources, severely impacting driving efficiency and the driving experience.

[0003] Existing methods for predicting traffic conditions on ordinary roads are relatively mature and can identify and predict congestion in a timely manner. However, highways have different characteristics from ordinary roads. Congestion usually occurs suddenly in a short period of time, and the factors that cause congestion are more numerous and complex. Therefore, the prediction methods for ordinary roads are not applicable to highways.

[0004] In summary, even highways, which currently ensure rapid vehicle travel, can experience congestion. Congestion affects driving efficiency and driving experience, and existing road condition prediction methods for ordinary roads are not applicable to highways.

[0005] This is a shortcoming of the existing technology. Therefore, it is very necessary to provide an artificial intelligence-based method and system for predicting highway traffic congestion in order to address the above-mentioned deficiencies in the existing technology. Summary of the Invention

[0006] In view of the shortcomings of existing technologies, such as the fact that even highways, which are currently designed to ensure the rapid travel of vehicles, can experience congestion, which affects driving efficiency and driving experience, and the fact that existing road condition prediction methods for ordinary roads are not applicable to highways, this invention provides an artificial intelligence-based method and system for predicting highway traffic congestion to solve the above-mentioned technical problems.

[0007] In a first aspect, the present invention provides a method for predicting highway traffic congestion based on artificial intelligence, comprising the following steps:

[0008] S1. Pre-divide the highway into sections and branches, and deploy cameras and sensors on the highway;

[0009] S2. Cloud servers collect data from cameras and sensors along various sections of the highway to build artificial intelligence algorithm models;

[0010] S3. The cloud server uses artificial intelligence algorithm models to predict traffic congestion on various sections of the highway, pushes the prediction results to users, and corrects the artificial intelligence algorithm models based on the accuracy of the prediction results.

[0011] Furthermore, the specific steps of step S1 are as follows:

[0012] S11. The highway is divided into sections in advance, and cameras are installed at the entrances and exits of each section;

[0013] S12. Divide each section of the highway into branches, record the characteristics of the branches, and install cameras at each branch of the road section;

[0014] S13. Weather sensors and road condition sensors are deployed on each section of the highway. Branch characteristics refer to the attribute of whether the current branch is a trunk road or a branch road; when the current branch is a trunk road, the number of corresponding branch roads at the front end is specified.

[0015] Furthermore, the specific steps of step S2 are as follows:

[0016] S21. Each camera will upload the collected video data and the location of its respective road segment and branch to the cloud server;

[0017] S22. Weather sensors upload collected weather information and the location of the road section to the cloud server, and road condition sensors upload collected road condition information to the cloud server;

[0018] S23. The cloud server analyzes the historical data collected by cameras and sensors to construct a dataset, labels the dataset with positive smooth data and negative congestion data, and divides the dataset into training data subset and validation data subset;

[0019] S24. The cloud server creates an artificial intelligence algorithm model and trains the artificial intelligence algorithm model using a subset of training data to obtain model parameters;

[0020] S25. The cloud server uses a subset of validation data to validate the trained artificial intelligence algorithm model, and retrains it if the validation fails, until the artificial intelligence algorithm model meets the requirements.

[0021] Furthermore, the specific steps of step S23 are as follows:

[0022] S231. The cloud server analyzes the video data collected by each camera, identifies the number of vehicles in the road segment and each branch at the sampling time point, and then calculates the vehicle density of the road segment and the vehicle density of the branch by combining the length of the road segment and the branch.

[0023] S232. The cloud server analyzes the video data collected by each camera, identifies the time points when the same vehicle appears in each camera, and calculates the vehicle speed of each vehicle in the corresponding road section and branch.

[0024] S233. The cloud server sets a vehicle speed benchmark based on a preset congestion standard, and classifies vehicle speeds below the benchmark as corresponding to congestion, while classifying vehicle speeds above or equal to the benchmark as corresponding to smooth traffic.

[0025] S234. The cloud server performs forward sampling from the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed is smooth, and marks the forward-sampled data as positive smooth data;

[0026] S235. The cloud server performs reverse sampling on the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed reaches a congested state, and marks the reverse sampled data as reverse congestion data.

[0027] S236. The cloud server collects an equal amount of forward smooth flow data and reverse congestion data as a dataset, and divides the dataset into a training data subset and a test data subset according to a set ratio, wherein the number of forward smooth flow data and reverse congestion data in each data subset is equal.

[0028] Furthermore, the specific steps of step S24 are as follows:

[0029] S241. The cloud server analyzes and calculates the correlation between vehicle density, time attribute data, road surface attribute data, and weather attribute data and the road conditions corresponding to vehicle speed, and initializes parameter weights based on the correlation.

[0030] S242. The cloud server constructs an artificial intelligence algorithm model according to the weights of the initialization parameters;

[0031] S243. The cloud server uses vehicle density, time attribute data, road surface attribute data and weather attribute data in the training data subset as input parameters, and uses the road conditions corresponding to vehicle speed in the training data subset as output parameters to train the artificial intelligence algorithm model and obtain preliminary model parameters.

[0032] S244. Input the initial model parameters into the artificial intelligence algorithm model to complete the model training.

[0033] Furthermore, the specific steps of step S25 are as follows:

[0034] S251. The cloud server inputs the vehicle density, time attribute data, road surface attribute data and weather attribute data from the verification data subset into the trained artificial intelligence algorithm model to obtain the predicted road condition data corresponding to the vehicle speed.

[0035] S252. The cloud server will compare the actual vehicle speeds in the verification dataset with the predicted road conditions to calculate the accuracy.

[0036] S253. The cloud server determines whether the accuracy meets the requirements;

[0037] If so, the artificial intelligence algorithm model is deemed to have passed verification, and the process proceeds to step S3;

[0038] If not, adjust the parameter step size, return to step S24, and retrain.

[0039] Furthermore, the specific steps of step S3 are as follows:

[0040] S 31. The cloud server analyzes the real-time collected data from cameras and sensors, extracts the required input parameters, and inputs the input parameters into the verified artificial intelligence algorithm model to obtain the predicted road conditions;

[0041] S 32. The cloud server will push the predicted road conditions to the relevant vehicles on the corresponding road sections;

[0042] S 33. The cloud server generates suggested routes for vehicles on road sections where traffic conditions are predicted to be congested, and pushes the suggested routes to the corresponding vehicles.

[0043] Furthermore, the specific steps of step S32 are as follows:

[0044] S 321. The cloud server obtains the prediction results of the artificial intelligence algorithm model for each road segment;

[0045] S 322. The cloud server locates each road segment and determines the located road segment and the road segment located at a set distance behind the located road segment as the target road segment;

[0046] S 323. The cloud server pushes the predicted road segment results to vehicles within the target road segment through the base station where the target road segment is located;

[0047] Step S33 is as follows:

[0048] S331. The cloud server identifies road segments predicted to be congested and searches for replacement road segments for the adjacent road segments behind the congested road segments, generating suggested routes based on the replacement road segments;

[0049] S 332. The cloud server uses the road segments connected to the congested road segments as target road segments and pushes suggested routes to vehicles in the target road segments through the base stations where the target road segments are located.

[0050] Secondly, the present invention provides an artificial intelligence-based highway traffic congestion prediction system, comprising:

[0051] The road segmentation module is used to pre-divide highways into road segments and branches, and to deploy cameras and sensors on the highways;

[0052] The model building module is used to collect data from cameras and sensors on various sections of the highway and build artificial intelligence algorithm models.

[0053] The traffic congestion prediction module is used to predict traffic congestion on various sections of highways using artificial intelligence algorithms, push the prediction results to users, and correct the artificial intelligence algorithm model based on the accuracy of the prediction results.

[0054] Furthermore, the road segmentation module includes:

[0055] The segment division unit is used to pre-divide the highway into segments and install cameras at the entrance and exit of each segment;

[0056] Branch division unit is used to divide each segment of the highway according to branches, record the branch characteristics, and deploy cameras at each branch of the segment.

[0057] The sensor deployment unit is used to deploy weather sensors and road condition sensors on each section of the highway.

[0058] The model building module includes:

[0059] The video data uploading unit is used to upload the video data collected by each camera and the location of the road segment and branch to the cloud server;

[0060] The sensor data uploading unit is used to upload the weather information collected by the weather sensor and the location of the road section to the cloud server, and the road condition sensor uploads the road condition information collected to the cloud server.

[0061] The dataset construction, labeling, and partitioning unit is used to analyze and construct a dataset from historical data collected by cameras and sensors. The dataset is labeled with positive smooth data and negative congestion data, and the dataset is divided into training data subsets and validation data subsets.

[0062] The model training unit is used to create artificial intelligence algorithm models and train them using a subset of training data to obtain model parameters.

[0063] The model validation unit is used to validate the trained artificial intelligence algorithm model using a subset of validation data, and to retrain the model if the validation fails, until the artificial intelligence algorithm model meets the requirements.

[0064] The traffic congestion prediction module includes:

[0065] The traffic prediction unit is used to analyze real-time collected camera and sensor data, extract the required input parameters, and input the input parameters into a validated artificial intelligence algorithm model to obtain the predicted traffic conditions.

[0066] The traffic condition push unit is used to push the predicted traffic conditions to relevant vehicles on the corresponding road sections;

[0067] Suggested route push is used to generate suggested routes for vehicles on road sections predicted to be congested, and push the suggested routes to the corresponding vehicles.

[0068] The beneficial effects of this invention are as follows:

[0069] The present invention provides an artificial intelligence-based highway traffic congestion prediction method and system, which uses artificial intelligence algorithm models to predict highway traffic conditions by road segment, recommends traffic conditions to relevant vehicle users, identifies congestion predictions in a timely manner, and pushes suggested routes to users before congestion occurs, thereby minimizing the impact of congestion and improving the overall traffic efficiency of highways.

[0070] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects.

[0071] Therefore, it is evident that the present invention has outstanding substantive features and significant progress compared with the prior art, and the beneficial effects of its implementation are also obvious. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a flowchart illustrating an embodiment of the highway traffic congestion prediction method based on artificial intelligence of the present invention.

[0074] Figure 2 This is a flowchart illustrating another embodiment of the highway traffic congestion prediction method based on artificial intelligence of the present invention.

[0075] Figure 3This is a schematic diagram of the highway traffic congestion prediction system based on artificial intelligence according to the present invention. Detailed Implementation

[0076] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0077] Example 1:

[0078] like Figure 1 As shown, this invention provides a highway traffic congestion prediction method based on artificial intelligence, comprising the following steps:

[0079] S1. Pre-divide the highway into sections and branches, and deploy cameras and sensors on the highway;

[0080] S2. Cloud servers collect data from cameras and sensors along various sections of the highway to build artificial intelligence algorithm models;

[0081] S3. The cloud server uses artificial intelligence algorithm models to predict traffic congestion on various sections of the highway, pushes the prediction results to users, and corrects the artificial intelligence algorithm models based on the accuracy of the prediction results.

[0082] Example 2:

[0083] like Figure 2 As shown, this invention provides a highway traffic congestion prediction method based on artificial intelligence, comprising the following steps:

[0084] S1. Pre-divide the highway into sections and branches, and deploy cameras and sensors along the highway; the specific steps of step S1 are as follows:

[0085] S11. The highway is divided into sections in advance, and cameras are installed at the entrances and exits of each section;

[0086] S12. Divide each section of the highway into branches and record the branch characteristics. Deploy cameras at each branch of the road segment. The branch characteristics are whether the current branch is a trunk road or a branch road. When the current branch is a trunk road, the number of corresponding branches at the front end is recorded. By setting up branch cameras, determine the impact of branch roads on congestion on trunk roads.

[0087] S13. Weather sensors and road condition sensors are deployed on each section of the highway; the road condition sensors detect whether the road surface is icy or covered with snow;

[0088] S2. The cloud server collects data from cameras and sensors along various sections of the highway and constructs an artificial intelligence algorithm model; the specific steps of step S2 are as follows:

[0089] S21. Each camera will upload the collected video data and the location of its respective road segment and branch to the cloud server;

[0090] S22. Weather sensors upload collected weather information and the location of the road section to the cloud server, and road condition sensors upload collected road condition information to the cloud server;

[0091] S23. The cloud server analyzes the historical data collected by cameras and sensors to construct a dataset, labels the dataset with positive smooth traffic data and negative congestion data, and divides the dataset into training data subsets and validation data subsets; the specific steps of step S23 are as follows:

[0092] S231. The cloud server analyzes the video data collected by each camera, identifies the number of vehicles in the road segment and each branch at the sampling time point, and then calculates the vehicle density of the road segment and the vehicle density of the branch by combining the length of the road segment and the branch.

[0093] S232. The cloud server analyzes the video data collected by each camera, identifies the time points when the same vehicle appears in each camera, and calculates the vehicle speed of each vehicle in the corresponding road section and branch.

[0094] S233. The cloud server sets a vehicle speed benchmark based on a preset congestion standard, and classifies vehicle speeds below the benchmark as corresponding to congestion, while classifying vehicle speeds above or equal to the benchmark as corresponding to smooth traffic.

[0095] S234. The cloud server performs forward sampling from the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed reaches a smooth flow condition, and marks the forward sampled data as positive smooth flow data; the data attribute data includes driving time periods, such as 6:00-24:00, idle time periods, such as 0:00-6:00, ordinary weekdays, and holidays; the weather attribute data includes weather conditions that affect traffic conditions, such as sunny days, foggy days, snowy days, and heavy rainy days.

[0096] S235. The cloud server performs reverse sampling on the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed reaches a congested state, and marks the reverse sampled data as reverse congestion data.

[0097] S236. The cloud server collects an equal amount of forward smooth flow data and reverse congestion data as a dataset, and divides the dataset into a training data subset and a test data subset according to a set ratio, wherein the number of forward smooth flow data and reverse congestion data in each data subset is equal.

[0098] S24. The cloud server creates an artificial intelligence algorithm model and trains the model using a subset of training data to obtain model parameters; the specific steps of step S24 are as follows:

[0099] S241. The cloud server analyzes and calculates the correlation between vehicle density, time attribute data, road surface attribute data, and weather attribute data and the road conditions corresponding to vehicle speed, and initializes parameter weights based on the correlation.

[0100] S242. The cloud server constructs an artificial intelligence algorithm model according to the weights of the initialization parameters;

[0101] S243. The cloud server uses vehicle density, time attribute data, road surface attribute data and weather attribute data in the training data subset as input parameters, and uses the road conditions corresponding to vehicle speed in the training data subset as output parameters to train the artificial intelligence algorithm model and obtain preliminary model parameters.

[0102] S244. Input the initial model parameters into the artificial intelligence algorithm model to complete the model training;

[0103] S25. The cloud server uses a subset of validation data to validate the trained AI algorithm model, and retrains it if the validation fails, until the AI ​​algorithm model meets the requirements; the specific steps of step S25 are as follows:

[0104] S251. The cloud server inputs the vehicle density, time attribute data, road surface attribute data and weather attribute data from the verification data subset into the trained artificial intelligence algorithm model to obtain the predicted road condition data corresponding to the vehicle speed.

[0105] S252. The cloud server will compare the actual vehicle speeds in the verification dataset with the predicted road conditions to calculate the accuracy.

[0106] S253. The cloud server determines whether the accuracy meets the requirements;

[0107] If so, the artificial intelligence algorithm model is deemed to have passed verification, and the process proceeds to step S3;

[0108] If not, adjust the parameter step size, return to step S24, and retrain;

[0109] S3. The cloud server uses an artificial intelligence algorithm model to predict traffic congestion on various sections of the highway, pushes the prediction results to users, and corrects the artificial intelligence algorithm model based on the accuracy of the prediction results; the specific steps of step S3 are as follows:

[0110] S 31. The cloud server analyzes the real-time collected data from cameras and sensors, extracts the required input parameters, and inputs the input parameters into the verified artificial intelligence algorithm model to obtain the predicted road conditions;

[0111] S 32. The cloud server pushes the predicted road conditions to relevant vehicles on the corresponding road segments; the specific steps of step S 32 are as follows:

[0112] S 321. The cloud server obtains the prediction results of the artificial intelligence algorithm model for each road segment;

[0113] S 322. The cloud server locates each road segment and determines the located road segment and the road segment located at a set distance behind the located road segment as the target road segment;

[0114] S 323. The cloud server pushes the predicted road segment results to vehicles within the target road segment through the base station where the target road segment is located;

[0115] S33. The cloud server generates suggested routes for vehicles on road sections predicted to be congested, and pushes the suggested routes to the corresponding vehicles; the specific steps of step S33 are as follows:

[0116] S331. The cloud server identifies road segments predicted to be congested and searches for replacement road segments for the adjacent road segments behind the congested road segments, generating suggested routes based on the replacement road segments;

[0117] S 332. The cloud server uses the road segments connected to the congested road segments as target road segments and pushes suggested routes to vehicles in the target road segments through the base stations where the target road segments are located.

[0118] Example 3:

[0119] like Figure 3 As shown, this invention provides an artificial intelligence-based highway traffic congestion prediction system, comprising:

[0120] The road segmentation module is used to pre-divide highways into road segments and branches, and to deploy cameras and sensors on the highways; the road segmentation module includes:

[0121] The segment division unit is used to pre-divide the highway into segments and install cameras at the entrance and exit of each segment;

[0122] Branch division unit is used to divide each segment of the highway according to branches, record the branch characteristics, and deploy cameras at each branch of the segment.

[0123] The sensor deployment unit is used to deploy weather sensors and road condition sensors on each section of the highway.

[0124] The model building module is used to collect data from cameras and sensors along various sections of the highway and construct artificial intelligence algorithm models. The model building module includes:

[0125] The video data uploading unit is used to upload the video data collected by each camera and the location of the road segment and branch to the cloud server;

[0126] The sensor data uploading unit is used to upload the weather information collected by the weather sensor and the location of the road section to the cloud server, and the road condition sensor uploads the road condition information collected to the cloud server.

[0127] The dataset construction, labeling, and partitioning unit is used to analyze and construct a dataset from historical data collected by cameras and sensors. The dataset is labeled with positive smooth data and negative congestion data, and the dataset is divided into training data subsets and validation data subsets.

[0128] The model training unit is used to create artificial intelligence algorithm models and train them using a subset of training data to obtain model parameters.

[0129] The model validation unit is used to validate the trained artificial intelligence algorithm model using a subset of validation data, and to retrain the model if the validation fails, until the artificial intelligence algorithm model meets the requirements.

[0130] The traffic congestion prediction module uses an artificial intelligence algorithm model to predict traffic congestion on various sections of highways, pushes the prediction results to users, and corrects the artificial intelligence algorithm model based on the accuracy of the prediction results. The traffic congestion prediction module includes:

[0131] The traffic prediction unit is used to analyze real-time collected camera and sensor data, extract the required input parameters, and input the input parameters into a validated artificial intelligence algorithm model to obtain the predicted traffic conditions.

[0132] The traffic condition push unit is used to push the predicted traffic conditions to relevant vehicles on the corresponding road sections;

[0133] Suggested route push is used to generate suggested routes for vehicles on road sections predicted to be congested, and push the suggested routes to the corresponding vehicles.

[0134] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the invention should also be covered within the protection scope of the invention. Therefore, the protection scope of the invention should be determined by the scope of the claims.

Claims

1. A method for predicting highway traffic congestion based on artificial intelligence, characterized in that, Includes the following steps: S1. Pre-divide the highway into sections and branches, and deploy cameras and sensors on the highway; S2. Cloud servers collect data from cameras and sensors along various sections of the highway to build artificial intelligence algorithm models; S3. The cloud server uses artificial intelligence algorithm models to predict traffic congestion on various sections of the highway, pushes the prediction results to users, and corrects the artificial intelligence algorithm models based on the accuracy of the prediction results; The specific steps of step S1 are as follows: S11. The highway is divided into sections in advance, and cameras are installed at the entrances and exits of each section; S12. Divide each section of the highway into branches, record the characteristics of the branches, and install cameras at each branch of the road section; S13. Weather sensors and road condition sensors shall be deployed on each section of the highway; The specific steps of step S2 are as follows: S21. Each camera will upload the collected video data and the location of its respective road segment and branch to the cloud server; S22. Weather sensors upload collected weather information and the location of the road section to the cloud server, and road condition sensors upload collected road condition information to the cloud server; S23. The cloud server analyzes the historical data collected by cameras and sensors to construct a dataset, labels the dataset with positive smooth data and negative congestion data, and divides the dataset into training data subset and validation data subset; S24. The cloud server creates an artificial intelligence algorithm model and trains the artificial intelligence algorithm model using a subset of training data to obtain model parameters; S25. The cloud server uses a subset of validation data to validate the trained artificial intelligence algorithm model, and retrains it if the validation fails, until the artificial intelligence algorithm model meets the requirements. The specific steps of step S23 are as follows: S231. The cloud server analyzes the video data collected by each camera, identifies the number of vehicles in the road segment and each branch at the sampling time point, and then calculates the vehicle density of the road segment and the vehicle density of the branch by combining the length of the road segment and the branch. S232. The cloud server analyzes the video data collected by each camera, identifies the time points when the same vehicle appears in each camera, and calculates the vehicle speed of each vehicle in the corresponding road section and branch. S233. The cloud server sets a vehicle speed benchmark based on a preset congestion standard, and classifies vehicle speeds below the benchmark as corresponding to congestion, while classifying vehicle speeds above or equal to the benchmark as corresponding to smooth traffic. S234. The cloud server performs forward sampling from the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed is smooth, and marks the forward-sampled data as positive smooth data; S235. The cloud server performs reverse sampling on the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed reaches a congested state, and marks the reverse sampled data as reverse congestion data. S236. The cloud server collects an equal amount of forward smooth flow data and reverse congestion data as a dataset, and divides the dataset into a training data subset and a test data subset according to a set ratio, wherein the number of forward smooth flow data and reverse congestion data in each data subset is equal.

2. The highway traffic congestion prediction method based on artificial intelligence as described in claim 1, characterized in that, The specific steps of step S24 are as follows: S241. The cloud server analyzes and calculates the correlation between vehicle density, time attribute data, road surface attribute data, and weather attribute data and the road conditions corresponding to vehicle speed, and initializes parameter weights based on the correlation. S242. The cloud server constructs an artificial intelligence algorithm model according to the weights of the initialization parameters; S243. The cloud server uses vehicle density, time attribute data, road surface attribute data and weather attribute data in the training data subset as input parameters, and uses the road conditions corresponding to vehicle speed in the training data subset as output parameters to train the artificial intelligence algorithm model and obtain preliminary model parameters. S244. Input the initial model parameters into the artificial intelligence algorithm model to complete the model training.

3. The highway traffic congestion prediction method based on artificial intelligence as described in claim 2, characterized in that, The specific steps of step S25 are as follows: S251. The cloud server inputs the vehicle density, time attribute data, road surface attribute data and weather attribute data from the verification data subset into the trained artificial intelligence algorithm model to obtain the predicted road condition data corresponding to the vehicle speed. S252. The cloud server will compare the actual vehicle speeds in the verification dataset with the predicted road conditions to calculate the accuracy. S253. The cloud server determines whether the accuracy meets the requirements; If so, the artificial intelligence algorithm model is deemed to have passed verification, and the process proceeds to step S3; If not, adjust the parameter step size, return to step S24, and retrain.

4. The highway traffic congestion prediction method based on artificial intelligence as described in claim 1, characterized in that, The specific steps of step S3 are as follows: S31. The cloud server analyzes the real-time collected data from cameras and sensors, extracts the required input parameters, and inputs the input parameters into the verified artificial intelligence algorithm model to obtain the predicted road conditions; S32. The cloud server will push the predicted road conditions to the relevant vehicles on the corresponding road sections; S33. The cloud server generates suggested routes for vehicles on road sections where traffic conditions are predicted to be congested, and pushes the suggested routes to the corresponding vehicles.

5. The highway traffic congestion prediction method based on artificial intelligence as described in claim 4, characterized in that, The specific steps of step S32 are as follows: S321. The cloud server obtains the prediction results of the artificial intelligence algorithm model for each road segment; S322. The cloud server locates each road segment and determines the located road segment and the road segment located at a set distance behind the located road segment as the target road segment; S323. The cloud server pushes the predicted road segment results to vehicles within the target road segment through the base station where the target road segment is located; The specific steps of step S33 are as follows: S331. The cloud server identifies road segments predicted to be congested and searches for replacement road segments for the adjacent road segments behind the congested road segments, generating suggested routes based on the replacement road segments; S332. The cloud server uses the road segments connected to the congested road segments as target road segments and pushes suggested routes to vehicles in the target road segments through the base stations where the target road segments are located.

6. A highway traffic congestion prediction system based on artificial intelligence, characterized in that, include: The road segmentation module is used to pre-divide highways into road segments and branches, and to deploy cameras and sensors on the highways; The model building module is used to collect data from cameras and sensors on various sections of the highway and build artificial intelligence algorithm models. The traffic congestion prediction module is used to predict traffic congestion on various sections of highways using artificial intelligence algorithm models, push the prediction results to users, and correct the artificial intelligence algorithm models based on the accuracy of the prediction results. The road segmentation module includes: The segment division unit is used to pre-divide the highway into segments and install cameras at the entrance and exit of each segment; Branch division unit is used to divide each segment of the highway according to branches, record the branch characteristics, and deploy cameras at each branch of the segment. The sensor deployment unit is used to deploy weather sensors and road condition sensors on each section of the highway. The model building module includes: The video data uploading unit is used to upload the video data collected by each camera and the location of the road segment and branch to the cloud server; The sensor data uploading unit is used to upload the weather information collected by the weather sensor and the location of the road section to the cloud server, and the road condition sensor uploads the road condition information collected to the cloud server. The dataset construction, labeling, and partitioning unit is used to analyze and construct a dataset from historical data collected by cameras and sensors. The dataset is labeled with positive smooth data and negative congestion data, and the dataset is divided into training data subsets and validation data subsets. The model training unit is used to create artificial intelligence algorithm models and train them using a subset of training data to obtain model parameters. The model validation unit is used to validate the trained artificial intelligence algorithm model using a subset of validation data, and to retrain the model if the validation fails, until the artificial intelligence algorithm model meets the requirements. The dataset construction, annotation, and partitioning unit analyzes historical data collected by cameras and sensors to construct a dataset. The dataset is then labeled with positive (smooth flow) data and negative (congestion) data, and partitioned into training and validation subsets, including: S231. The cloud server analyzes the video data collected by each camera, identifies the number of vehicles in the road segment and each branch at the sampling time point, and then calculates the vehicle density of the road segment and the vehicle density of the branch by combining the length of the road segment and the branch. S232. The cloud server analyzes the video data collected by each camera, identifies the time points when the same vehicle appears in each camera, and calculates the vehicle speed of each vehicle in the corresponding road section and branch. S233. The cloud server sets a vehicle speed benchmark based on a preset congestion standard, and classifies vehicle speeds below the benchmark as corresponding to congestion, while classifying vehicle speeds above or equal to the benchmark as corresponding to smooth traffic. S234. The cloud server performs forward sampling from the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed is smooth, and marks the forward-sampled data as positive smooth data; S235. The cloud server performs reverse sampling on the datasets collected by cameras and sensors, collecting vehicle density, time attribute data, road surface attribute data, and weather attribute data for each road segment or branch at a set time point before the vehicle speed reaches a congested state, and marks the reverse sampled data as reverse congestion data. S236. The cloud server collects an equal amount of forward smooth flow data and reverse congestion data as a dataset, and divides the dataset into a training data subset and a test data subset according to a set ratio, wherein the number of forward smooth flow data and reverse congestion data in each data subset is equal.

7. The artificial intelligence-based highway traffic congestion prediction system as described in claim 6, characterized in that, The traffic congestion prediction module includes: The traffic prediction unit is used to analyze real-time collected camera and sensor data, extract the required input parameters, and input the input parameters into a validated artificial intelligence algorithm model to obtain the predicted traffic conditions. The traffic condition push unit is used to push the predicted traffic conditions to relevant vehicles on the corresponding road sections; Suggested route push is used to generate suggested routes for vehicles on road sections predicted to be congested, and push the suggested routes to the corresponding vehicles.

Images