Travel intelligent traffic management system and method based on big data

Abstract

The invention discloses a travel intelligent traffic management system and method based on big data, and belongs to the technical field of traffic travel management. The method comprises the following steps: firstly, identifying a vehicle import relationship between road sections, constructing a traffic flow data set in three dimensions of year, month and day, calculating a maximum value, a mean value and a minimum value under each dimension, introducing trend change rate analysis, and performing traffic flow prediction in combination with a road section traffic flow prediction neural network, thereby realizing more accurate traffic flow prediction. By analyzing the flow ratio between the connected road sections and dynamically correcting the prediction result, more accurate monitoring road section traffic flow prediction is finally obtained. According to the invention, the problem of low traffic flow prediction precision in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic management, and particularly relates to an intelligent traffic management system and method for travel based on big data. Background Art

[0002] In modern urban traffic management, accurately predicting traffic flow has always been a complex and challenging problem. Traditional traffic flow prediction methods mainly rely on static statistical models and simple time series analysis, and these methods often fail to effectively capture the dynamics and complexity of the traffic system. Existing technologies usually face multiple limitations: one is the lack of in-depth consideration of the road network, making it difficult to accurately reflect the mutual influence between different road segments; the other is that the prediction model's processing of the time dimension is too single, and it fails to fully consider the multi-scale time characteristics of years, months, and days. Therefore, existing technologies have the problem of low accuracy in predicting traffic flow. Summary of the Invention

[0003] Aiming at the above deficiencies in the prior art, an intelligent traffic management system and method for travel based on big data provided by the present invention solves the problem of low accuracy in predicting traffic flow existing in the prior art.

[0004] To achieve the above invention objective, the technical solution adopted by the present invention is: an intelligent traffic management system for travel based on big data, including: a road segment acquisition subsystem, a set construction subsystem, a change rate acquisition subsystem, a prediction subsystem, and a correction subsystem;

[0005] The road segment acquisition subsystem is used to obtain the connecting road segments having a vehicle import relationship with the monitored road segment according to the position of the monitored road segment in the traffic route;

[0006] The set construction subsystem is used to extract the historical traffic flow of each road segment according to the travel time, and construct sets of the maximum value, average value, and minimum value of the traffic flow in three dimensions of year, month, and day;

[0007] The change rate acquisition subsystem is used to obtain the trend change rate for each set, and obtain the maximum value, average value, and minimum value trend change rates in three dimensions of year, month, and day;

[0008] The prediction subsystem is used to process the sets of the maximum value, average value, and minimum value of the traffic flow in three dimensions by using a road segment traffic flow prediction neural network, perform feature enhancement based on the maximum value, average value, and minimum value trend change rates in three dimensions, and obtain the predicted traffic flow of each road segment;

[0009] The correction subsystem is used to obtain the traffic flow ratio between the monitored road segment and the connecting road segments according to the set of the average value of the traffic flow in the day dimension, and correct the predicted traffic flow of the monitored road segment based on the predicted traffic flow of the connecting road segments to obtain the corrected predicted traffic flow of the monitored road segment.

[0010] Further, the specific process of the set construction subsystem includes:

[0011] Obtain the travel time period according to the travel time;

[0012] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow in the same travel time period of each adjacent historical year for each road segment, and arrange them in chronological order to obtain the annual maximum traffic flow set, annual average traffic flow set, and annual minimum traffic flow set;

[0013] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow in the same travel time period of each adjacent historical month for each road segment, and arrange them in chronological order to obtain the monthly maximum traffic flow set, monthly average traffic flow set, and monthly minimum traffic flow set;

[0014] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow in the same travel time for each adjacent historical day for each road segment, and arrange them in chronological order to obtain the daily maximum traffic flow set, daily average traffic flow set, and daily minimum traffic flow set.

[0015] Further, the specific process of the change rate acquisition subsystem includes:

[0016] Obtain the trend change rates for the annual maximum traffic flow set, annual average traffic flow set, and annual minimum traffic flow set respectively to obtain the annual maximum value trend change rate, annual average value trend change rate, and annual minimum value trend change rate;

[0017] Obtain the trend change rates for the monthly maximum traffic flow set, monthly average traffic flow set, and monthly minimum traffic flow set respectively to obtain the monthly maximum value trend change rate, monthly average value trend change rate, and monthly minimum value trend change rate;

[0018] Obtain the trend change rates for the daily maximum traffic flow set, daily average traffic flow set, and daily minimum traffic flow set respectively to obtain the daily maximum value trend change rate, daily average value trend change rate, and daily minimum value trend change rate.

[0019] Further, the formula for obtaining the trend change rate is: , where θ is the trend change rate, v i is the i-th traffic flow in the set, v i+1 is the (i + 1)-th element in the set, N is the number of traffic flows in the set, and i is a positive integer.

[0020] Further, the road segment traffic flow prediction neural network includes: a first fully connected unit, a second fully connected unit, a third fully connected unit, a first splicing layer, a second splicing layer, a fusion layer E1, a first convolutional layer, a second convolutional layer, and an output layer;

[0021] The first fully connected unit is used to process the set of maximum, average, and minimum values of vehicle flow in the annual dimension to obtain the annual maximum value feature, the annual average value feature, and the annual minimum value feature; the second fully connected unit is used to process the set of maximum, average, and minimum values of vehicle flow in the monthly dimension to obtain the monthly maximum value feature, the monthly average value feature, and the monthly minimum value feature; the third fully connected unit is used to process the set of maximum, average, and minimum values of vehicle flow in the daily dimension to obtain the daily maximum value feature, the daily average value feature, and the daily minimum value feature;

[0022] The first splicing layer is used to splice the output features of the three fully connected units into a feature matrix; the second splicing layer is used to splice the trend change rates of the maximum, average, and minimum values in the three dimensions into a change rate matrix;

[0023] The input end of the fusion layer E1 is respectively connected to the output end of the first splicing layer and the output end of the second splicing layer, and its output end is connected to the input end of the first convolutional layer;

[0024] The output end of the first convolutional layer is connected to the input end of the second convolutional layer;

[0025] The input end of the output layer is connected to the output end of the second convolutional layer, and its output end serves as the output end of the road section traffic flow prediction neural network.

[0026] Furthermore, each of the first fully connected unit, the second fully connected unit, and the third fully connected unit includes: a first fully connected layer, a second fully connected layer, and a third fully connected layer; in the first fully connected unit, the first fully connected layer processes the set of annual vehicle flow maximum values to obtain the annual maximum value feature, the second fully connected layer processes the set of annual vehicle flow average values to obtain the annual average value feature, and the third fully connected layer processes the set of annual vehicle flow minimum values to obtain the annual minimum value feature;

[0027] In the second fully connected unit, the first fully connected layer processes the set of monthly vehicle flow maximum values to obtain the monthly maximum value feature, the second fully connected layer processes the set of monthly vehicle flow average values to obtain the monthly average value feature, and the third fully connected layer processes the set of monthly vehicle flow minimum values to obtain the monthly minimum value feature;

[0028] In the third fully connected unit, the first fully connected layer processes the set of daily vehicle flow maximum values to obtain the daily maximum value feature, the second fully connected layer processes the set of daily vehicle flow average values to obtain the daily average value feature, and the third fully connected layer processes the set of daily vehicle flow minimum values to obtain the daily minimum value feature.

[0029] Furthermore, the fusion layer E1 is used to multiply the feature matrix and the change rate matrix element by element;

[0030] The convolution kernel size of the first convolutional layer is ;

[0031] The convolution kernel size of the second convolutional layer is .

[0032] Furthermore, the specific process of the correction subsystem includes:

[0033] Obtain the flow ratio of the monitoring section and the connecting section according to the traffic flow mean value set in the daily dimension;

[0034] Take the mean of the predicted traffic flows of each connecting section to obtain the mean predicted traffic flow of the connecting section;

[0035] Multiply the mean predicted traffic flow of the connecting section by the flow ratio to obtain the comparison flow;

[0036] Take the mean of the comparison flow and the predicted traffic flow of the monitoring section to obtain the corrected predicted traffic flow of the monitoring section.

[0037] Furthermore, the specific process of obtaining the flow ratio of the monitoring section and the connecting section includes: taking the mean of the traffic flows with the same number in the traffic flow mean value set in the daily dimension of each connecting section to obtain the average traffic flow, taking the ratio of the traffic flow with the same number in the traffic flow mean value set in the daily dimension of the monitoring section to the average traffic flow with the same number as the flow ratio of this number, and taking the mean of the flow ratios of each number as the flow ratio.

[0038] An intelligent traffic management method for travel based on big data includes the following steps:

[0039] S1. According to the position of the monitoring section in the traffic route, obtain the connecting sections that have a vehicle import relationship with the monitoring section;

[0040] S2. According to the travel time, extract the historical traffic flows of each section and construct the maximum, mean, and minimum value sets of traffic flows in the three dimensions of year, month, and day;

[0041] S3. Obtain the trend change rate for each set to obtain the maximum, mean, and minimum trend change rates in the three dimensions of year, month, and day;

[0042] S4. Use the section traffic flow prediction neural network to process the maximum, mean, and minimum value sets of traffic flows in the three dimensions, and perform feature enhancement based on the maximum, mean, and minimum trend change rates in the three dimensions to obtain the predicted traffic flow of each section;

[0043] S5. According to the traffic flow mean value set in the daily dimension, obtain the flow ratio of the monitoring section and the connecting section, and correct the predicted traffic flow of the monitoring section based on the predicted traffic flow of the connecting section to obtain the corrected predicted traffic flow of the monitoring section.

[0044] The beneficial effects of the present invention are as follows:

[0045] 1. Starting from the three dimensions of year, month, and day, the present invention obtains the maximum, average, and minimum value sets of traffic flow in the three dimensions, which can comprehensively capture the periodicity and trend of traffic flow on different time scales. Then, the trend change rates in the three dimensions are extracted to deeply mine the dynamic evolution characteristics of traffic flow. The present invention then processes the maximum, average, and minimum value sets of traffic flow in the three dimensions using a section traffic flow prediction neural network, and improves the accuracy of predicting the traffic flow of each section based on the feature enhancement of the trend change rates in the three dimensions.

[0046] 2. Vehicles will flow between different sections, and the change in traffic flow on the connecting sections will directly or indirectly affect the monitored section. By calculating the flow ratio using the average value set of traffic flow in the daily dimension, this correlation can be quantified, reflecting the proportional law of traffic flow between the monitored section and the connecting section under normal circumstances. Thus, the predicted traffic flow of the monitored section can be corrected based on the predicted traffic flow of the connecting section, further improving the accuracy of traffic flow prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a system block diagram of an intelligent transportation management system for travel based on big data;

[0048] Figure 2 is a relationship diagram between a monitored section and a connecting section;

[0049] Figure 3 is a schematic structural diagram of a section traffic flow prediction neural network;

[0050] Figure 4 is a schematic structural diagram of a first fully connected unit;

[0051] Figure 5 is a schematic structural diagram of a second fully connected unit;

[0052] Figure 6 is a schematic structural diagram of a third fully connected unit. DETAILED DESCRIPTION OF THE INVENTION

[0053] The following describes the specific embodiments of the present invention to facilitate those skilled in the art of the present technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of the present technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions created using the concept of the present invention are within the scope of protection.

[0054] Example 1, as Figure 1As shown in the figure, an intelligent transportation management system for travel based on big data includes: a road section acquisition subsystem, a set construction subsystem, a change rate acquisition subsystem, a prediction subsystem, and a correction subsystem;

[0055] The road section acquisition subsystem is used to obtain the connecting road sections that have a vehicle import relationship with the monitored road section according to the position of the monitored road section in the traffic route;

[0056] The set construction subsystem is used to extract the historical traffic flow of each road section according to the travel time, and construct sets of the maximum, average, and minimum traffic flows in the three dimensions of year, month, and day;

[0057] The change rate acquisition subsystem is used to obtain the trend change rate for each set, and obtain the maximum, average, and minimum trend change rates in the three dimensions of year, month, and day;

[0058] The prediction subsystem is used to process the sets of the maximum, average, and minimum traffic flows in the three dimensions by using a road section traffic flow prediction neural network, and perform feature enhancement based on the maximum, average, and minimum trend change rates in the three dimensions to obtain the predicted traffic flow of each road section;

[0059] The correction subsystem is used to obtain the traffic flow ratio of the monitored road section and the connecting road section according to the set of the average traffic flow in the day dimension, and correct the predicted traffic flow of the monitored road section based on the predicted traffic flow of the connecting road section to obtain the corrected predicted traffic flow of the monitored road section.

[0060] As Figure 2 shown, the vehicles on the connecting road section can drive into the monitored road section.

[0061] In this embodiment, the specific process of the set construction subsystem includes:

[0062] Obtain the travel time period according to the travel time;

[0063] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow of each road section in the same travel time period in each adjacent historical year, and arrange them in chronological order to obtain the annual maximum traffic flow set, annual average traffic flow set, and annual minimum traffic flow set;

[0064] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow of each road section in the same travel time period in each adjacent historical month, and arrange them in chronological order to obtain the monthly maximum traffic flow set, monthly average traffic flow set, and monthly minimum traffic flow set;

[0065] Extract the maximum traffic flow, average traffic flow, and minimum traffic flow of each road section at the same travel time in each adjacent historical day, and arrange them in chronological order to obtain the daily maximum traffic flow set, daily average traffic flow set, and daily minimum traffic flow set.

[0066] The travel time period can be defined as: a continuous time interval intercepted before and after a given travel time, centered around the given travel time.

[0067] For example, the travel time is: 11:00 on May 9, 2025. In this embodiment, the travel time period is set as 10:00 - 12:00 on May 9, 2025. The adjacent historical years extracted are 2024, 2023, 2022, 2021, and 2020. The maximum traffic flow, average traffic flow, and minimum traffic flow are respectively extracted from 10:00 - 12:00 on May 9 of 2024, 2023, 2022, 2021, and 2020. The maximum traffic flows in each annual dimension are arranged in the order of time occurrence to construct a set of annual maximum traffic flows; the average traffic flows in each annual dimension are arranged in the order of time occurrence to construct a set of annual average traffic flows; the minimum traffic flows in each annual dimension are arranged in the order of time occurrence to construct a set of annual minimum traffic flows.

[0068] The adjacent historical months extracted are April 2025, March 2025, February 2025, January 2025, and December 2024. The maximum traffic flow, average traffic flow, and minimum traffic flow of each road section are respectively extracted from the travel time period of these five adjacent historical months: 10:00 - 12:00 on April 9, 2025; 10:00 - 12:00 on March 9, 2025; 10:00 - 12:00 on February 9, 2025; 10:00 - 12:00 on January 9, 2025; and 10:00 - 12:00 on December 9, 2024.

[0069] The adjacent historical days refer to multiple consecutive dates selected forward based on the currently occurred date.

[0070] If the current date is June 5, 2025 (a date that has occurred, and June 7, 2025 and June 8, 2025 have not occurred and are not considered), then the adjacent historical days are: June 6, 2025; June 4, 2025; June 3, 2025; June 2, 2025; and June 1, 2025. The maximum traffic flow, average traffic flow, and minimum traffic flow are respectively extracted from 10:00 - 12:00 of these five dates: June 6, 2025; June 4, 2025; June 3, 2025; June 2, 2025; and June 1, 2025.

[0071] In this embodiment, the specific process of the change rate acquisition subsystem includes:

[0072] Obtain the trend change rates for the annual maximum traffic volume set, the annual average traffic volume set, and the annual minimum traffic volume set respectively, to obtain the annual maximum trend change rate, the annual average trend change rate, and the annual minimum trend change rate;

[0073] Obtain the trend change rates for the monthly maximum traffic volume set, the monthly average traffic volume set, and the monthly minimum traffic volume set respectively, to obtain the monthly maximum trend change rate, the monthly average trend change rate, and the monthly minimum trend change rate;

[0074] Obtain the trend change rates for the daily maximum traffic volume set, the daily average traffic volume set, and the daily minimum traffic volume set respectively, to obtain the daily maximum trend change rate, the daily average trend change rate, and the daily minimum trend change rate.

[0075] In this embodiment, the formula for obtaining the trend change rate is: , where θ is the trend change rate, v i is the i-th traffic volume in the set, v i+1 is the (i + 1)-th element in the set, N is the number of traffic volumes in the set, and i is a positive integer.

[0076] The present invention obtains the trend change rates for the maximum, average, and minimum traffic volume sets at different time scales of year, month, and day respectively. The annual dimension grasps the long-term change law, the monthly dimension reflects the monthly periodicity, and the daily dimension captures the short-term fluctuations, comprehensively covering the analysis requirements of traffic volume change characteristics.

[0077] As Figure 3 shown, the road section traffic volume prediction neural network includes: a first fully connected unit, a second fully connected unit, a third fully connected unit, a first splicing layer, a second splicing layer, a fusion layer E1, a first convolutional layer, a second convolutional layer, and an output layer;

[0078] The first fully connected unit is used to process the maximum, average, and minimum traffic volume sets in the annual dimension to obtain the annual maximum feature, the annual average feature, and the annual minimum feature; the second fully connected unit is used to process the maximum, average, and minimum traffic volume sets in the monthly dimension to obtain the monthly maximum feature, the monthly average feature, and the monthly minimum feature; the third fully connected unit is used to process the maximum, average, and minimum traffic volume sets in the daily dimension to obtain the daily maximum feature, the daily average feature, and the daily minimum feature;

[0079] The first splicing layer is used to splice the output features of the three fully connected units into a feature matrix; the second splicing layer is used to splice the maximum, average, and minimum trend change rates in the three dimensions into a change rate matrix;

[0080] The input end of the fusion layer E1 is respectively connected to the output ends of the first splicing layer and the second splicing layer, and its output end is connected to the input end of the first convolutional layer;

[0081] The output end of the first convolutional layer is connected to the input end of the second convolutional layer;

[0082] The input end of the output layer is connected to the output end of the second convolutional layer, and its output end serves as the output end of the road section traffic flow prediction neural network.

[0083] In the present invention, the first, second, and third fully connected units respectively process the maximum value, average value, and minimum value sets of the traffic flow in the annual, monthly, and daily dimensions, comprehensively capturing the traffic flow characteristics at different time scales. Then, through the first splicing layer, the features of each dimension are integrated into a feature matrix, and the second splicing layer integrates the trend change rate into a change rate matrix, realizing the systematic fusion of multi-dimensional data, enabling the model to comprehensively consider various aspects of traffic flow information and improving the prediction accuracy.

[0084] The first fully connected unit, the second fully connected unit, and the third fully connected unit all include: the first fully connected layer, the second fully connected layer, and the third fully connected layer; as Figure 4 shown, in the first fully connected unit, the first fully connected layer processes the annual traffic flow maximum value set to obtain the annual maximum value feature, the second fully connected layer processes the annual traffic flow average value set to obtain the annual average value feature, and the third fully connected layer processes the annual traffic flow minimum value set to obtain the annual minimum value feature.

[0085] As Figure 5 shown, in the second fully connected unit, the first fully connected layer processes the monthly traffic flow maximum value set to obtain the monthly maximum value feature, the second fully connected layer processes the monthly traffic flow average value set to obtain the monthly average value feature, and the third fully connected layer processes the monthly traffic flow minimum value set to obtain the monthly minimum value feature.

[0086] As Figure 6 shown, in the third fully connected unit, the first fully connected layer processes the daily traffic flow maximum value set to obtain the daily maximum value feature, the second fully connected layer processes the daily traffic flow average value set to obtain the daily average value feature, and the third fully connected layer processes the daily traffic flow minimum value set to obtain the daily minimum value feature.

[0087] In this embodiment, the fusion layer E1 is used to multiply the feature matrix and the change rate matrix element by element;

[0088] The convolution kernel size of the first convolutional layer is ;

[0089] The convolution kernel size of the second convolutional layer is .

[0090] The fusion layer E1 outputs The fusion feature matrix, the convolution kernel size of the first convolutional layer is , therefore, the first convolutional layer performs a convolution operation on each row of the fusion feature matrix.

[0091] In this embodiment, the output layer is implemented using a fully connected layer.

[0092] The feature matrix is , where C is the feature matrix, c y,max is the annual maximum value feature, c y,avg is the annual average value feature, c y,min is the annual minimum value feature, c m,max is the monthly maximum value feature, c m,avg is the monthly average value feature, c m,min is the monthly minimum value feature, c s,max is the daily maximum value feature, c s,avg is the daily average value feature, c s,min is the daily minimum value feature.

[0093] The rate of change matrix is , where R is the rate of change matrix, r y,max is the annual maximum value trend rate of change, r y,avg is the annual average value trend rate of change, r y,min is the annual minimum value trend rate of change, r m,max is the monthly maximum value trend rate of change, r m,avg is the monthly average value trend rate of change, r m,min is the monthly minimum value trend rate of change, r s,max is the daily maximum value trend rate of change, r s,avg is the daily average value trend rate of change, r s,min is the daily minimum value trend rate of change.

[0094] The present invention multiplies the feature matrix and the rate of change matrix element by element to multiply the annual maximum value feature by the annual maximum value trend rate of change, multiply the annual average value feature by the annual average value trend rate of change, multiply the annual minimum value feature by the annual minimum value trend rate of change, and so on, to enhance the corresponding features, enhance the feature expression ability, and improve the prediction accuracy.

[0095] In this embodiment, the specific process of the correction subsystem includes:

[0096] Obtain the traffic flow ratio of the monitoring section and the connecting section according to the traffic flow average value set in the daily dimension;

[0097] Take the average of the predicted traffic flows of each connecting section to obtain the average predicted traffic flow of the connecting section;

[0098] Multiply the predicted traffic flow mean value of the connection section by the flow ratio to obtain the comparison flow.

[0099] Take the mean value of the comparison flow and the predicted traffic flow of the monitoring section to obtain the corrected predicted traffic flow of the monitoring section.

[0100] In this embodiment, the specific process of obtaining the flow ratio of the monitoring section and the connection section includes: taking the mean value of the traffic flow with the same number in the set of daily traffic flow mean values of each connection section to obtain the average traffic flow, and taking the ratio of the traffic flow with the same number in the set of daily traffic flow mean values of the monitoring section to the average traffic flow with the same number as the flow ratio of this number, and taking the mean value of the flow ratios of each number as the flow ratio.

[0101] The calculation formula of the flow ratio is:[[]] , where θ is the flow ratio, I mo,avg,k,n is the traffic flow with number n in the set of daily traffic flow mean values of the kth connection section (i.e., the nth traffic flow), I co,avg,n is the traffic flow with number n in the set of daily traffic flow mean values of the monitoring section, M is the number of traffic flows in the set of daily traffic flow mean values, n and k are positive integers, and K mo is the number of connection sections.

[0102] By obtaining the flow ratio of the monitoring section and the connection section, the present invention takes into account the correlation between the traffic flows of the monitoring section and the surrounding connection sections. Calculating using the set of daily traffic flow mean values can capture the mutual influence of traffic flows between sections in the short term, and using the predicted traffic flow of the connection section to correct the monitoring section can further improve the prediction accuracy.

[0103] Embodiment 2, a big data-based intelligent traffic management method for travel, includes the following steps:

[0104] S1. According to the position of the monitoring section in the traffic route, obtain the connection sections that have a vehicle import relationship with the monitoring section;

[0105] S2. According to the travel time, extract the historical traffic flow of each section, and construct sets of the maximum value, mean value, and minimum value of the traffic flow in three dimensions of year, month, and day;

[0106] S3. Obtain the trend change rate for each set to obtain the maximum value, mean value, and minimum value trend change rates in three dimensions of year, month, and day;

[0107] S4. Use the section traffic flow prediction neural network to process the sets of the maximum value, mean value, and minimum value of the traffic flow in three dimensions, and perform feature enhancement based on the maximum value, mean value, and minimum value trend change rates in three dimensions to obtain the predicted traffic flow of each section;

[0108] S5. Obtain the flow ratio of the monitoring section and the connecting section according to the set of average traffic flow in the daily dimension, and correct the predicted traffic flow of the monitoring section based on the predicted traffic flow of the connecting section to obtain the corrected predicted traffic flow of the monitoring section.

[0109] The specific implementation process of Embodiment 2 is the same as that of Embodiment 1.

[0110] The present invention starts from three dimensions of year, month, and day, obtains the sets of maximum, average, and minimum traffic flows in the three dimensions, can comprehensively capture the periodicity and trend of traffic flow on different time scales, then extracts the trend change rates in the three dimensions, and deeply mines the dynamic evolution characteristics of traffic flow. The present invention then uses a section traffic flow prediction neural network to process the sets of maximum, average, and minimum traffic flows in the three dimensions, and improves the accuracy of predicting the traffic flow of each section based on the enhancement of the characteristics of the trend change rates in the three dimensions.

[0111] Vehicles will flow between different sections, and the change in the traffic flow of the connecting section will directly or indirectly affect the monitoring section. Calculating the flow ratio through the set of average traffic flow in the daily dimension can quantify this correlation relationship and reflect the proportional law of the traffic flow between the monitoring section and the connecting section under normal circumstances. Thus, the predicted traffic flow of the monitoring section is corrected based on the predicted traffic flow of the connecting section, further improving the accuracy of traffic flow prediction.

[0112] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A big data-based intelligent traffic management system, characterized in that: include: Road segment acquisition subsystem, set construction subsystem, change rate acquisition subsystem, prediction subsystem and correction subsystem; The road section acquisition subsystem is used to acquire the connecting road section having a vehicle import relationship with the monitoring road section according to the position of the monitoring road section in the traffic route; The set construction subsystem is used to extract the historical traffic flow of each road section according to the travel time, and construct the maximum, mean and minimum traffic flow sets in the three dimensions of year, month and day; The change rate acquisition subsystem is used to obtain the trend change rate for each set, and obtain the maximum, average and minimum trend change rates in the three dimensions of year, month and day; The prediction subsystem is used to process the maximum, mean and minimum value sets of vehicle flow in three dimensions using the road section flow prediction neural network, and perform feature enhancement based on the maximum, mean and minimum value trend change rates in the three dimensions to obtain the predicted vehicle flow of each road section; The correction subsystem is used to obtain the traffic flow ratio of the monitoring section and the connecting section based on the mean traffic flow set in the daily dimension, and to correct the predicted traffic flow of the monitoring section based on the predicted traffic flow of the connecting section to obtain the corrected predicted traffic flow of the monitoring section.

2. The intelligent traffic management system based on big data according to claim 1 is characterized in that: The specific process of the collection construction subsystem includes: According to the travel time, obtain the travel time period; Extract the maximum traffic flow, average traffic flow and minimum traffic flow in the same travel time period of each adjacent historical year for each road section, arrange them in chronological order, and obtain the annual maximum traffic flow set, annual average traffic flow set and annual minimum traffic flow set; Extract the maximum traffic flow, average traffic flow and minimum traffic flow in the same travel time period of each adjacent historical month for each road section, arrange them in chronological order, and obtain the monthly maximum traffic flow set, monthly average traffic flow set and monthly minimum traffic flow set; The maximum traffic flow, average traffic flow and minimum traffic flow at the same travel time in each adjacent historical day of each road section are extracted and arranged in chronological order to obtain the set of maximum daily traffic flow, the set of average daily traffic flow and the set of minimum daily traffic flow.

3. The intelligent traffic management system based on big data according to claim 1 is characterized in that: The specific process of the change rate acquisition subsystem includes: Obtain trend change rates for the annual maximum value set, the annual average value set and the annual minimum value set of vehicle flow, respectively, and obtain the annual maximum value trend change rate, the annual average value trend change rate and the annual minimum value trend change rate; Obtain trend change rates for the monthly maximum value set, the monthly average value set and the monthly minimum value set of vehicle flow, respectively, and obtain the monthly maximum value trend change rate, the monthly average value trend change rate and the monthly minimum value trend change rate; The trend change rates of the daily maximum traffic flow set, the daily average traffic flow set and the daily minimum traffic flow set are obtained respectively, and the daily maximum value trend change rate, the daily average value trend change rate and the daily minimum value trend change rate are obtained.

4. The intelligent traffic management system based on big data according to claim 1 is characterized in that: The formula for obtaining the trend change rate is: , where θ is the trend change rate, v i is the i-th vehicle flow in the set, v i+1 is the i+1th element in the set, N is the number of vehicle flows in the set, and i is a positive integer.

5. The intelligent traffic management system based on big data according to claim 1 is characterized in that: The road section flow prediction neural network includes: a first fully connected unit, a second fully connected unit, a third fully connected unit, a first splicing layer, a second splicing layer, a fusion layer E1, a first convolutional layer, a second convolutional layer and an output layer; The first fully connected unit is used to process the maximum, mean and minimum value sets of traffic flow in the annual dimension, and obtain the annual maximum value feature, annual average value feature and annual minimum value feature; the second fully connected unit is used to process the maximum, mean and minimum value sets of traffic flow in the monthly dimension, and obtain the monthly maximum value feature, monthly average value feature and monthly minimum value feature; the third fully connected unit is used to process the maximum, mean and minimum value sets of traffic flow in the daily dimension, and obtain the daily maximum value feature, daily average value feature and daily minimum value feature; The first concatenation layer is used to concatenate the output features of the three fully connected units into The second concatenation layer is used to concatenate the maximum, mean, and minimum trend change rates in the three dimensions into The rate of change matrix; The input end of the fusion layer E1 is connected to the output end of the first concatenation layer and the output end of the second concatenation layer respectively, and its output end is connected to the input end of the first convolution layer; The output of the first convolutional layer is connected to the input of the second convolutional layer; The input end of the output layer is connected to the output end of the second convolutional layer, and its output end serves as the output end of the road section flow prediction neural network.

6. The intelligent traffic management system based on big data according to claim 5 is characterized in that: The first fully connected unit, the second fully connected unit and the third fully connected unit all include: a first fully connected layer, a second fully connected layer and a third fully connected layer; in the first fully connected unit, the first fully connected layer processes the annual vehicle flow maximum value set to obtain the annual maximum value feature, the second fully connected layer processes the annual vehicle flow mean value set to obtain the annual mean value feature, and the third fully connected layer processes the annual vehicle flow minimum value set to obtain the annual minimum value feature; In the second fully connected unit, the first fully connected layer processes the maximum value set of monthly traffic flow to obtain the monthly maximum value feature, the second fully connected layer processes the mean value set of monthly traffic flow to obtain the monthly mean value feature, and the third fully connected layer processes the minimum value set of monthly traffic flow to obtain the monthly minimum value feature; In the third fully connected unit, the first fully connected layer processes the maximum value set of daily traffic flow to obtain the daily maximum value feature, the second fully connected layer processes the mean value set of daily traffic flow to obtain the daily mean value feature, and the third fully connected layer processes the minimum value set of daily traffic flow to obtain the daily minimum value feature.

7. The intelligent traffic management system based on big data according to claim 5 is characterized in that: The fusion layer E1 is used to multiply the feature matrix and the change rate matrix element by element; The convolution kernel size of the first convolutional layer is ; The convolution kernel size of the second convolutional layer is .

8. The intelligent traffic management system based on big data according to claim 1 is characterized in that: The specific process of correcting the subsystem includes: According to the average value of vehicle flow in the daily dimension, the flow ratio of the monitored section and the connecting section is obtained; The predicted traffic flow of each connecting section is averaged to obtain the average predicted traffic flow of the connecting section; Multiply the predicted traffic flow mean of the connecting road section by the traffic flow ratio to obtain the comparative traffic flow; The average of the comparison flow and the predicted traffic flow of the monitored section is taken to obtain the revised predicted traffic flow of the monitored section.

9. The intelligent traffic management system based on big data according to claim 8 is characterized in that: The specific process of obtaining the traffic ratio between the monitored section and the connecting section includes: averaging the traffic flow of the same number in the traffic flow mean set of each connecting section in the daily dimension to obtain the average traffic flow, taking the ratio of the traffic flow of the same number in the traffic flow mean set of the monitored section in the daily dimension to the average traffic flow of the same number as the traffic ratio of the number, and taking the average of the traffic ratios of each number as the traffic ratio.

10. A method for intelligent traffic management based on big data, implemented based on the intelligent traffic management system based on big data according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. According to the position of the monitored section in the traffic route, a connecting section having a vehicle import relationship with the monitored section is obtained; S2. Extract the historical traffic volume of each road section according to the travel time, and construct the maximum, mean and minimum traffic volume sets in the three dimensions of year, month and day; S3. Obtain the trend change rate for each set, and obtain the maximum, mean, and minimum trend change rates in the three dimensions of year, month, and day; S4, using a road section traffic flow prediction neural network to process the maximum, mean and minimum value sets of traffic flow in three dimensions, perform feature enhancement based on the maximum, mean and minimum value trend change rates in the three dimensions, and obtain the predicted traffic flow of each road section; S5. According to the set of average traffic flow in the daily dimension, the traffic flow ratio of the monitoring section and the connecting section is obtained, and based on the predicted traffic flow of the connecting section, the predicted traffic flow of the monitoring section is corrected to obtain the corrected predicted traffic flow of the monitoring section.

Images