CN117935561B - Intelligent traffic flow analysis method based on Beidou data - Google Patents
Intelligent traffic flow analysis method based on Beidou data Download PDFInfo
- Publication number
- CN117935561B CN117935561B CN202410317217.9A CN202410317217A CN117935561B CN 117935561 B CN117935561 B CN 117935561B CN 202410317217 A CN202410317217 A CN 202410317217A CN 117935561 B CN117935561 B CN 117935561B
- Authority
- CN
- China
- Prior art keywords
- traffic
- data
- time
- traffic flow
- regulation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000005206 flow analysis Methods 0.000 title claims abstract description 20
- 238000010586 diagram Methods 0.000 claims abstract description 45
- 230000008859 change Effects 0.000 claims abstract description 34
- 238000009826 distribution Methods 0.000 claims abstract description 31
- 238000004458 analytical method Methods 0.000 claims abstract description 30
- 238000005457 optimization Methods 0.000 claims abstract description 25
- 238000004140 cleaning Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 36
- 230000000694 effects Effects 0.000 claims description 35
- 230000006870 function Effects 0.000 claims description 29
- 238000004422 calculation algorithm Methods 0.000 claims description 24
- 238000011156 evaluation Methods 0.000 claims description 18
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 16
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 16
- 238000007726 management method Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 9
- 230000001105 regulatory effect Effects 0.000 claims description 7
- 230000002123 temporal effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 5
- 206010039203 Road traffic accident Diseases 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 230000006872 improvement Effects 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000007619 statistical method Methods 0.000 claims description 4
- 238000012300 Sequence Analysis Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 2
- 241001235534 Graphis <ascomycete fungus> Species 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007789 sealing Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Traffic Control Systems (AREA)
Abstract
The invention relates to the technical field of traffic flow analysis, in particular to an intelligent traffic flow analysis method based on Beidou data, which comprises the following steps of: s1: collecting vehicle position data in a preset area through a Beidou satellite navigation system; s2: cleaning and formatting the original data collected in the step S1; s3: calculating a vehicle density distribution in each time period; s4: constructing a space-time diagram convolutional network prediction model, and predicting traffic flow change trend of a preset area in a short period; s5, introducing traffic influence factors for analysis, and optimizing the prediction accuracy of the traffic flow change trend; s6: generating real-time traffic regulation and control suggestions for traffic management departments; s7: and adjusting the prediction model and the traffic regulation suggestion. According to the invention, the traffic flow change trend is accurately predicted in real time, the traffic regulation suggestion is dynamically generated, and the efficiency and accuracy of urban traffic management are obviously improved by combining with continuous optimization prediction and regulation strategies.
Description
Technical Field
The invention relates to the technical field of traffic flow analysis, in particular to an intelligent traffic flow analysis method based on Beidou data.
Background
In modern urban life, along with the rapid increase of the number of vehicles, traffic jam has become a major problem affecting urban operation efficiency and resident life quality, effective traffic flow management and planning are important for relieving traffic jam, improving road use efficiency and guaranteeing traffic safety, traditional traffic management methods mainly rely on historical traffic flow data and artificial experience to make decisions, and the methods are difficult to respond to dynamic changes of traffic flows in real time, especially when emergencies or special conditions (such as holidays, bad weather and the like) are encountered, and the effect of the traditional methods is greatly reduced.
With the development of intelligent traffic systems and big data technologies, it is possible to monitor and predict traffic flows in real time by using advanced data analysis and prediction models, however, how to accurately predict traffic flow trends and how to formulate effective traffic control strategies based on prediction results is still a technical problem, on one hand, the existing traffic flow prediction models often ignore the complexity of traffic flows affected by various factors, such as weather changes, holidays, urban activities, etc., resulting in low prediction accuracy; on the other hand, even if traffic flow can be predicted more accurately, how to dynamically optimize traffic regulation measures according to the prediction result and how to evaluate and adjust the actual effect of the regulation measures is a problem to be solved.
Disclosure of Invention
Based on the above purpose, the invention provides an intelligent traffic flow analysis method based on Beidou data.
An intelligent traffic flow analysis method based on Beidou data comprises the following steps:
s1: collecting vehicle position data in a preset area through a Beidou satellite navigation system;
s2: cleaning and formatting the original data collected in the step S1 by utilizing a data preprocessing algorithm;
s3: according to the data processed in the step S2, a space analysis algorithm is applied to calculate the vehicle density distribution in each time period;
S4: constructing a space-time diagram convolutional network prediction model, analyzing the vehicle density distribution obtained in the step S3, and predicting the traffic flow change trend of a preset area in a short period;
s5: on the basis of S4, introducing traffic influence factors for analysis, and optimizing the prediction accuracy of the traffic flow change trend;
s6: based on the prediction results of the steps S4 and S5, combining historical traffic data, and generating real-time traffic regulation and control suggestions for a traffic management department through a preset dynamic optimization algorithm;
s7: and comparing the actual traffic flow data with the prediction result and the regulation suggestion, and regulating the prediction model and the traffic regulation suggestion according to the comparison result.
Further, the S1 specifically includes:
S11: firstly, determining a predetermined area range, including a main urban traffic network, important traffic hubs and high-density commercial areas, and specifically defining the area by an urban traffic planning map and a geographic information system interface of a Beidou satellite navigation system;
s12: in the preset area determined in the step S1, traffic monitoring equipment with a Beidou satellite receiving module is deployed, wherein the traffic monitoring equipment comprises Beidou positioning equipment on a mobile vehicle and a Beidou receiving station fixed on a preset road section, and the Beidou positioning equipment and the Beidou receiving station can receive signals from Beidou satellites and are used for capturing real-time position information of the vehicle;
s13: collecting position data by a Beidou satellite navigation system, wherein the position data comprises longitude, latitude, time stamp, speed and running direction of each vehicle;
s14: the accuracy of the position data is improved by utilizing the differential positioning function of the Beidou satellite navigation system, and the differential positioning realizes higher-accuracy position information measurement by comparing the difference between satellite signals received by the fixed Beidou receiving stations in the preset area and signals received by the Beidou receiving modules on the mobile vehicle;
S15: and transmitting the collected position data to a data processing center for standby through a wireless communication network.
Further, the step S2 specifically includes:
S21: receiving the original position data from the S1, and performing data cleaning to remove invalid data and abnormal values, wherein the specific operations comprise identifying and eliminating obvious errors in the position data and filtering stationary vehicle data;
s22: formatting the cleaned data to enable the data to conform to a standard format of subsequent analysis, specifically comprising a time stamp format of unified data, and standardizing longitude and latitude, speed and direction data into a preset numerical format to ensure consistency and comparability of the data;
s23: performing data deduplication processing, checking whether repeated records exist in the data set processed in the steps S22 and S23, and replacing repeated items with the latest data records so as to avoid data redundancy in the analysis process;
s25: and data segmentation is carried out, and the data set is divided into a plurality of time periods according to a preset time interval, so that time sequence analysis and traffic flow trend prediction are conveniently carried out.
Further, the step S3 specifically includes:
S31: presetting a grid system, namely determining a geographic boundary of a preset area through a geographic information system tool; uniformly dividing a preset area into square grid units with fixed sizes according to the total area of the area and the expected analysis precision, wherein the side length of each grid unit is set to be 100 meters, and each grid unit is provided with a unique identifier and an accurate geographic coordinate range;
S32: calculating the vehicle density of grid units in each time period, determining the total number of vehicles passing through each grid unit in unit time by adopting a direct counting method, wherein the vehicle density is defined as the number of vehicles in unit grid area, and the specific calculation formula is as follows: Wherein, the method comprises the steps of, wherein, Representing a total number of vehicles passing through the grid cell over a predetermined period of time; is the area of the grid cell, i.e. 10000 square meters;
s33: integrating and creating a vehicle density distribution map, representing the vehicle density value of each grid cell calculated in the step S32 by using color codes, mapping the color codes to geographic positions of the corresponding grid cells, gradually changing the color of the vehicle density from green to red, and visually displaying the vehicle distribution condition in each grid cell.
Further, the constructing a space-time diagram convolutional network prediction model in S4 specifically includes:
s41: initializing a map structure, constructing a map based on the vehicle density distribution map obtained in S33 Wherein the picture isBy node setsSum edge setEach node represents a grid unit, the connection relation between the nodes is determined based on the adjacency of the geographic position, and particularly when two grid units are adjacent, the nodes corresponding to the two grid units are in the graphThe middle parts are connected through edges;
s42: defining a space-time feature matrix, regarding the vehicle density of each node in different time periods as a feature, and forming the space-time feature matrix Feature matrixIs of the size of，Is the total number of nodes that are to be counted,Is the length of the observation period;
S43: spatial features are extracted by applying graph convolution, and graph convolution network is specifically adopted for graph comparison The spatial feature learning is carried out, and the graph rolling operation is specifically expressed as follows: Wherein, the method comprises the steps of, wherein, Is the addition of a self-connecting adjacency matrix,Is a graphIs used for the adjacent matrix of (a),Is an identity matrix of the unit cell,Is a diagonal matrix of the node degree matrix,Is the firstThe node characteristic matrix of the layer,Is the weight matrix of the layer,A nonlinear activation function;
S44: integrating one-dimensional convolution processing time sequences, and applying one-dimensional convolution to the time sequence characteristics of each node to capture a change mode in a time dimension and provide time characteristics for vehicle density prediction;
S45: constructing a final space-time diagram convolutional network prediction model, integrating the spatial features and the temporal features processed in the steps S43 and S44 to form a complete space-time diagram convolutional network, wherein the function of the space-time diagram convolutional network prediction model is as follows: Wherein, the method comprises the steps of, wherein, Representing a prediction of vehicle density for a future time period,Is the space-time feature matrix of the input,Including the parameters of all spatial and temporal convolution layers,Representing a function of the overall space-time diagram convolutional network.
Further, in S4, analyzing the vehicle density distribution obtained in S3, and predicting a traffic flow trend of the predetermined area in a short period specifically includes:
S46: processing the vehicle density distribution data obtained in the step S3 to adapt to the input requirements of a space-time diagram convolution network prediction model, wherein the specific processing comprises time sequence standardization and space feature coding; the time sequence normalization is to normalize the vehicle density time sequence of each grid unit to eliminate the influence of different orders, and the normalization formula is as follows: Wherein, the method comprises the steps of, wherein, Representing the original vehicle density value of the vehicle,AndRepresenting the mean and standard deviation of the time series respectively,Is a normalized value; the space feature codes are used for converting the geographic position information of the grid cells into space feature codes for the space-time diagram convolutional network prediction model to learn, so that the space-time diagram convolutional network prediction model can identify and utilize the spatial interrelationship;
S47: training STGCN models by using the data processed in the step S46, wherein the training aim is to minimize the difference between the predicted vehicle density and the actual vehicle density, and specifically, the mean square error is adopted as a loss function, and the formula is as follows: Wherein, the method comprises the steps of, wherein, Is the loss function of the device,Is the number of samples that are to be taken,Is the model pairThe predicted value of the individual samples is calculated,Is the corresponding true value;
S48: predicting the vehicle density in a future period of time by using a space-time diagram convolution network prediction model after training is completed, and outputting a predicted vehicle density value of each grid unit at a future time point, wherein the future time point comprises a peak time in the morning and evening, a class time in the school, a working day or a holiday;
s49: according to the vehicle density distribution in S48, a future traffic flow variation trend in the predetermined area is analyzed, the variation trend including identifying a congestion area, and evaluating a variation trend of the traffic volume.
Further, the step S5 specifically includes:
s51: selecting key influencing factors of traffic flow change, wherein the key influencing factors comprise weather conditions and holidays;
S52: for each determined influencing factor, collecting related data, integrating the collected data into a format which can be identified by a space-time diagram convolution network prediction model, specifically encoding weather conditions into numerical variables, and expressing holidays through Boolean variables;
S53: the input of the expansion model, namely the traffic influence factor data integrated in the step S52 is used as additional characteristics to be input into a space-time diagram convolution network prediction model constructed in the step S4;
S54: on the basis of new input, retraining a space-time diagram convolutional network prediction model, and optimizing model parameters by supervision and learning by using an extended data set comprising traffic influence factors so as to improve the prediction accuracy of future traffic flow change trend;
S55: the optimized space-time diagram convolution network prediction model formula is as follows: Wherein, the method comprises the steps of, wherein, Is the model predicted optimized vehicle density distribution,Is an original space-time characteristic matrix,A feature set representing a traffic influencing factor,Is an optimized model parameter set.
Further, the step S6 specifically includes:
S61: integrating the vehicle density prediction result of the preset area in the future period obtained in the steps S4 and S5 with historical traffic flow data of the past year in the same period to form a comprehensive data set;
S62: defining an optimization target and a constraint, wherein the specific optimization target is to minimize traffic jam and improve road traffic efficiency, the constraint condition comprises road capacity, traffic signal control parameters, road closure or limiting measures, and the specific function of the optimization problem is as follows:
，
subjectto Wherein, the method comprises the steps of, wherein, Represents the firstCongestion cost function for each grid cell,AndRepresenting the predicted traffic flow and traffic control parameters within the grid cell respectively,AndRespectively isAndIs a feasible region of (2);
S63: the optimization problem in the step S62 is solved by adopting a dynamic programming algorithm, the dynamic programming algorithm is used for solving a complex problem into smaller sub-problems, and the solution of the original problem is constructed by utilizing the solution of the sub-problems, so that an optimal traffic regulation strategy is found, and a specific algorithm formula is as follows:
Wherein, the method comprises the steps of, wherein, Is represented in a given current stateDown, from the firstMinimum congestion cost achieved by starting from the grid cells;
Represents the first Current traffic states of the individual grid cells; Is directed to the first Traffic control measures to be taken by the grid cells; Representing a set of all possible traffic regulation measures; Is at the first Grid cells, given the current traffic stateAnd take regulation measuresThe direct congestion cost at that time; Is the first to A set of adjacent grid cells, represented in the traffic network, the firstNeighboring grid cells directly affected by the individual grid cells; representing adjacent grid cells In its current stateMinimum congestion cost;
S64: according to the result of the dynamic planning algorithm, specific traffic regulation suggestions are generated for each grid unit, wherein the suggestions comprise adjustment of signal lamps, temporary closing or opening of road sections and change of traffic guidance, and the generated suggestions are summarized and transmitted to a traffic management department for real-time traffic regulation.
Further, the step S7 specifically includes:
S71: firstly, collecting actual traffic flow data, including indexes of traffic flow density, speed and passing time, and sorting out traffic flow data predicted by a model and traffic regulation suggestions formulated based on the prediction;
s72: the statistical analysis method of absolute percentage error is adopted to quantitatively compare the actual traffic flow data with the model prediction result, and the calculation formula of the absolute percentage error is as follows:
Wherein, the method comprises the steps of, wherein, Is the average of the absolute percentage error, is used to quantify the prediction accuracy,Is the total number of data points considered in the evaluation,Is the firstThe actual traffic flow value of the data point,The corresponding model predicts the traffic flow value; by calculating MAPE, the average level of the prediction error can be intuitively known, and the lower MAPE value represents the higher prediction accuracy;
S73: the effect of the implemented traffic regulation measures is evaluated, the influence of the traffic regulation measures on the actual traffic flow is analyzed, the specific evaluation comprises the steps of comparing traffic flow data before and after regulation, and the specific regulation effect evaluation formula is as follows:
Effectiveness Wherein EFFECTIVENESS denotes the percentage of effect of the regulatory measures, Is the average value of traffic flow or congestion index before the implementation of the regulation measures,Is the corresponding value after the implementation of the regulation measures; the regulation and control effect evaluation formula is used for quantitatively measuring the change degree of the traffic condition before and after the regulation and control measures are implemented, and positive values indicate that the traffic condition is improved;
S74: based on the comparison analysis results of S72 and S73, the defects of the prediction model and the improvement space of the regulation strategy are specifically identified.
Further, the step S74 specifically includes:
S741: firstly calculating an integral MAPE value predicted by a model, determining integral prediction accuracy, and then carrying out deep analysis on the MAPE value to distinguish MAPE values of different time periods and different areas so as to identify prediction deviation of the model under different conditions;
S742: calculating the change of traffic flow before and after implementing the regulation measures, quantitatively evaluating the actual influence of each regulation measure by using the regulation effect evaluation formula, and comparing the indexes of traffic flow and speed before and after regulation to evaluate the regulation effect; analyzing the effect of each regulation and control measure, and identifying measures which fail to reach the expected target;
S743: providing model optimization suggestions based on the analysis results of S741 and S742, including introducing new data features to enhance the response capability of the model to road construction, traffic accidents and temporary road closure, adjusting the model structure to improve the prediction accuracy of a predetermined area or time period, or retraining the model to better fit actual traffic flow data;
S744: according to the evaluation result of the regulation strategy, the optimization direction is provided for different regulation measures, including the measures of adjusting the implementation period of the regulation measures to adapt to the peak value of traffic flow, enhancing or reducing the target area of the regulation measures to achieve the expected regulation effect or completely replacing the measures with unobvious effect.
The invention has the beneficial effects that:
According to the invention, by utilizing the high-precision positioning data of the Beidou satellite navigation system and combining with advanced data processing and prediction algorithms, the traffic flow trend can be accurately predicted in real time. Compared with the traditional prediction method based on historical data, the method can respond to the dynamic change of traffic flow more effectively, and can provide more accurate traffic flow information especially when dealing with emergencies or special situations (such as severe weather, large-scale activities and the like), so that powerful data support is provided for traffic management departments to formulate regulation and control measures.
According to the invention, by introducing a preset dynamic optimization algorithm, real-time traffic regulation and control suggestions can be generated according to the prediction result, and regulation and control measures can be evaluated and adjusted in real time according to actual traffic flow data. The dynamic optimization and feedback mechanism enables traffic regulation measures to be more flexible and effective, can ensure that traffic management measures are closely matched with actual traffic conditions, effectively reduces traffic jams, and improves road use efficiency.
According to the invention, through accurately predicting traffic flow and formulating an effective regulation strategy, traffic jam phenomenon can be obviously reduced, commute time is shortened, the occurrence rate of traffic accidents is reduced, scientific decision basis is provided for public traffic planning and urban development, and the overall operation efficiency and safety of an urban traffic system are improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only of the invention and that other drawings can be obtained from them without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of an intelligent traffic flow analysis method based on Beidou data according to an embodiment of the invention;
fig. 2 is a schematic flow chart of a prediction model of a convolution network for constructing a space-time diagram according to an embodiment of the present invention.
Detailed Description
The present invention will be further described in detail with reference to specific embodiments in order to make the objects, technical solutions and advantages of the present invention more apparent.
It is to be noted that unless otherwise defined, technical or scientific terms used herein should be taken in a general sense as understood by one of ordinary skill in the art to which the present invention belongs. The terms "first," "second," and the like, as used herein, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items.
As shown in fig. 1-2, an intelligent traffic flow analysis method based on Beidou data comprises the following steps:
s1: collecting vehicle position data in a preset area through a Beidou satellite navigation system, wherein the vehicle position data comprises real-time positions, speeds and running directions of vehicles, and taking the real-time positions, speeds and running directions as data bases;
s2: cleaning and formatting the original data collected in the step S1 by utilizing a data preprocessing algorithm, wherein the steps comprise removing abnormal values, filling data missing items and normalizing the data so as to ensure the accuracy and reliability of subsequent analysis;
s3: according to the data processed in the step S2, a space analysis algorithm is applied to calculate the vehicle density distribution in each time period so as to reveal the traffic flow characteristics of different areas and different time periods;
s4: constructing a space-time diagram convolutional network prediction model, analyzing the vehicle density distribution obtained in the step S3, predicting the traffic flow change trend of a preset area in a short period, and combining Beidou data characteristics and an urban traffic flow mode;
S5: on the basis of S4, introducing traffic influence factors to analyze, and optimizing the prediction accuracy of the traffic flow change trend, wherein the traffic influence factors comprise weather conditions, holidays, large events and the like;
S6: based on the prediction results of the steps S4 and S5, combining historical traffic data, generating real-time traffic regulation and control suggestions for a traffic management department through a preset dynamic optimization algorithm, and aiming at preventing potential traffic jam points and optimizing traffic flow;
s7: and comparing the actual traffic flow data with the prediction result and the regulation suggestion, and regulating the prediction model and the traffic regulation suggestion according to the comparison result.
S1 specifically comprises:
S11: firstly, determining a predetermined area range, including a main urban traffic network, important traffic hubs (such as railway stations and airports) and high-density commercial areas, and defining the area by specifically using an urban traffic planning map and a geographic information system interface of a Beidou satellite navigation system;
s12: in the preset area determined in the step S1, traffic monitoring equipment with a Beidou satellite receiving module is deployed, wherein the traffic monitoring equipment comprises Beidou positioning equipment on a mobile vehicle and a Beidou receiving station fixed on a preset road section, and the Beidou positioning equipment and the Beidou receiving station can receive signals from Beidou satellites and are used for capturing real-time position information of the vehicle;
s13: collecting position data by a Beidou satellite navigation system, wherein the position data comprise longitude, latitude, time stamp, speed and running direction of each vehicle, the data are obtained by analyzing Beidou satellite signals, the longitude and the latitude are determined by satellite positioning, the time stamp is provided by a time synchronization function of the Beidou system, and the speed and the running direction are calculated by comparing the change of continuous two-time position information;
s14: the accuracy of the position data is improved by utilizing the differential positioning function of the Beidou satellite navigation system, and the differential positioning realizes higher-accuracy position information measurement by comparing the difference between satellite signals received by the fixed Beidou receiving stations in the preset area and signals received by the Beidou receiving modules on the mobile vehicle;
S15: and transmitting the collected position data to a data processing center for standby through a wireless communication network, carrying out subsequent preprocessing and analysis, and adopting an encryption technology to ensure the safety and the integrity of the data in the data transmission process.
S2 specifically comprises:
s21: receiving the original position data from S1, and performing data cleaning to remove invalid data and abnormal values, wherein specific operations comprise identifying and eliminating obvious errors (such as abnormal coordinate points caused by signal interference) in the position data, and filtering out stationary (speed is zero and duration exceeds a preset threshold) vehicle data to ensure the quality of data of subsequent analysis;
S22: formatting the cleaned data to enable the data to conform to a standard format of subsequent analysis, wherein the standard format specifically comprises a time stamp format of unified data (for example, converting the time stamp format into UTC standard time), and standardizing longitude, latitude, speed and direction data into a preset numerical format so as to ensure consistency and comparability of the data;
s23: performing data deduplication processing, checking whether repeated records exist in the data set processed in the steps S22 and S23, and replacing repeated items with the latest data records so as to avoid data redundancy in the analysis process;
S25: dividing a data set into a plurality of time periods according to preset time intervals (such as every minute or every hour), facilitating time sequence analysis and traffic flow trend prediction, storing formatted data in a special database of a data processing center, and providing a prepared data source for a subsequent space analysis algorithm and a traffic flow prediction model;
Through the substep S21S25, the original traffic position data collected from the Beidou satellite navigation system is ensured to be subjected to effective cleaning and formatting treatment, an accurate and consistent data basis is provided for high-precision traffic flow analysis and management, and meanwhile, the continuity and the high efficiency of a data processing flow are ensured.
S3 specifically comprises:
S31: presetting a grid system in a research area, and specifically determining a geographic boundary of the preset area through a geographic information system tool; then, according to the total area of the area and the expected analysis precision, uniformly dividing the preset area into square grid units with fixed sizes, wherein the side length of each grid unit is set to be 100 meters, sufficient resolution is ensured to reflect the spatial variation of the vehicle density, the process is automatically completed by GIS software, and each grid unit is ensured to be provided with a unique identifier and an accurate geographic coordinate range;
S32: calculating the vehicle density of grid units in each time period, determining the total number of vehicles passing through each grid unit in unit time by adopting a direct counting method, wherein the vehicle density is defined as the number of vehicles in unit grid area, and the specific calculation formula is as follows: Wherein, the method comprises the steps of, wherein, Representing a total number of vehicles passing through the grid cell over a predetermined period of time; for the area of the grid cells, namely 10000 square meters, the process is executed in GIS software through an automatic script to ensure that the vehicle density of each grid cell can be accurately calculated;
S33: integrating and creating a vehicle density distribution map, representing the vehicle density value of each grid unit calculated in the step S32 by using color codes, mapping the color codes to geographic positions of the corresponding grid units, and visually displaying the vehicle distribution condition in each grid unit by gradually changing the color of the vehicle density from green (representing low density) to red (representing high density), wherein the density distribution map is generated by data visualization software, so that an visual traffic flow characteristic display mode is provided for subsequent analysis;
the above-described methods S31 to S33 describe how to define the grid system, calculate the vehicle density, and create the process of the vehicle density distribution map, ensuring the clarity of the operation and the accuracy of the result.
The construction of the space-time diagram convolutional network prediction model in the S4 specifically comprises the following steps:
s41: initializing a map structure, constructing a map based on the vehicle density distribution map obtained in S33 Wherein the picture isBy node setsSum edge setEach node represents a grid unit, the connection relation (i.e. edge) between the nodes is determined based on the adjacency of the geographic position, and particularly when two grid units are adjacent (i.e. share a boundary), then the nodes corresponding to the two grid units are in the graphThe middle parts are connected through edges;
s42: defining a space-time feature matrix, regarding the vehicle density of each node in different time periods as a feature, and forming the space-time feature matrix Feature matrixIs of the size of，Is the total number of nodes (grid cells),Is the length of the observation period;
s43: spatial features are extracted using graph convolution, in particular using Graph Convolution Network (GCN) versus graph The spatial feature learning is carried out, and the graph rolling operation is specifically expressed as follows: Wherein, the method comprises the steps of, wherein, Is the addition of a self-connecting adjacency matrix,Is a graphIs used for the adjacent matrix of (a),Is an identity matrix of the unit cell,Is a diagonal matrix of the node degree matrix,Is the firstThe node characteristic matrix of the layer,Is the weight matrix of the layer,Nonlinear activation functions, such as ReLU;
s44: integrating one-dimensional convolution processing time sequences, applying one-dimensional convolution to the time sequence features of each node to capture the change mode in the time dimension, providing time features for vehicle density prediction, and aiming at the graph Has a time-varying vehicle density signature sequence, assuming for the nodeThe time sequence is characterized in that: Wherein Is the length of the time series, our goal is to process this sequence with a one-dimensional convolution network (1 DCNN) to learn and capture the pattern of changes over time, the one-dimensional convolution operation being formulated as: Wherein, the method comprises the steps of, wherein, Is a nodeAt the point of timeThe output characteristics of the above-mentioned device,Is an activation function, such as a ReLU function,Is a convolution kernel (or filter) of lengthRepresenting the window size applied over the time series, this convolution kernel is combined with a local region of the input time seriesA consecutive point in time) performs a dot product operation,Is a bias term that is used to determine,Is a time sliding index from 0 to; The essence of this one-dimensional convolution operation is to slide the convolution kernel over the time sequenceAnd calculating a weighted sum of the kernel and the input features at each location, and then applying an activation function, in such a way that the 1DCNN can extract useful local temporal features from the time series, which capture the pattern of change in vehicle density over a short period of time;
S45: constructing a final space-time diagram convolutional network prediction model, integrating the spatial features and the temporal features processed in the steps S43 and S44 to form a complete space-time diagram convolutional network, wherein the function of the space-time diagram convolutional network prediction model is as follows: Wherein, the method comprises the steps of, wherein, Representing a prediction of vehicle density for a future time period,Is the space-time feature matrix of the input,Including the parameters of all spatial and temporal convolution layers,Representing a function of the overall space-time diagram convolution network, the output of the model is a prediction of the vehicle density for each grid cell over a period of time in the future.
S4, analyzing the vehicle density distribution obtained in the step S3, and predicting the traffic flow change trend of the preset area in a short period specifically comprises the following steps:
S46: processing the vehicle density distribution data obtained in the step S3 to adapt to the input requirements of a space-time diagram convolution network prediction model, wherein the specific processing comprises time sequence standardization and space feature coding; the time sequence normalization is to normalize the time sequence of the vehicle density of each grid cell to eliminate the influence of different orders, and the normalization formula is: Wherein, the method comprises the steps of, wherein, Representing the original vehicle density value of the vehicle,AndRepresenting the mean and standard deviation of the time series respectively,Is a normalized value; the space feature coding is used for converting the geographic position information of the grid unit into space feature coding for the space-time diagram convolutional network prediction model to learn, so that the space-time diagram convolutional network prediction model can identify and utilize the spatial interrelation;
S47: training STGCN models by using the data processed in the step S46, wherein the training aim is to minimize the difference between the predicted vehicle density and the actual vehicle density, and specifically, the mean square error is adopted as a loss function, and the formula is as follows: Wherein, the method comprises the steps of, wherein, Is the loss function of the device,Is the number of samples that are to be taken,Is the model pairThe predicted value of the individual samples is calculated,Is the corresponding true value;
s48: predicting the vehicle density in a future period of time by using a space-time diagram convolution network prediction model after training is finished, and outputting a predicted vehicle density value of each grid unit at a future time point according to the latest vehicle density data and a historical trend in a prediction process, wherein the future time point comprises a peak time in the morning and evening, a school lesson time, a working day or a holiday;
s49: according to the vehicle density distribution in S48, a future traffic flow variation trend in the predetermined area is analyzed, the variation trend including identifying a congestion area, and evaluating a variation trend of the traffic volume.
S5 specifically comprises the following steps:
s51: selecting key influencing factors of traffic flow changes, wherein the key influencing factors comprise weather conditions (such as rain, snow, fog and the like), holidays and the like, and the factors directly influence vehicle density and traffic flow changes;
S52: for each determined influencing factor, collecting related data, integrating the collected data into a format which can be identified by a space-time diagram convolution network prediction model, specifically encoding weather conditions into numerical variables, and expressing holidays through Boolean variables;
S53: the input of the expansion model, namely, the traffic influence factor data integrated in the step S52 is used as additional characteristics to be input into a space-time diagram convolution network prediction model constructed in the step S4, which requires the input layer of the model to be adjusted so as to contain the new characteristic vectors;
S54: on the basis of new input, retraining a space-time diagram convolutional network prediction model, and optimizing model parameters by supervision and learning by using an extended data set comprising traffic influence factors so as to improve the prediction accuracy of future traffic flow change trend;
S55: the optimized space-time diagram convolution network prediction model formula is as follows: Wherein, the method comprises the steps of, wherein, Is the model predicted optimized vehicle density distribution,Is an original space-time characteristic matrix,A feature set representing a traffic influencing factor,Is an optimized model parameter set;
By the above step S51S55, it is explained in detail how the prediction accuracy of the traffic flow trend is optimized by introducing traffic influencing factors, and by incorporating key traffic influencing factors into the prediction model, the traffic flow changes under different conditions can be predicted more accurately, thereby improving the effectiveness of traffic management and planning.
Application scenario example: urban holiday traffic management optimization
Background: urban A faces the challenge of holiday traffic congestion, especially around commercial areas and major transportation hubs, and traditional traffic management systems fail to effectively predict holiday traffic flow changes, resulting in traffic congestion and inefficiency of management.
The object is: by utilizing the technical schemes of the steps S51 to S55, traffic influence factors are introduced for analysis so as to optimize traffic flow change trend prediction of holidays, thereby providing effective traffic regulation measures, relieving congestion and improving traffic efficiency.
The implementation steps are as follows:
determining traffic influencing factors: determining holidays, special activities (such as shopping malls and urban marathons) and weather conditions (such as rainy days and snowy days) as key traffic influencing factors;
Data collection and integration: collecting traffic flow data, weather records and special event calendars of past holidays, and encoding the information into a format which can be processed by the model;
expansion model input: integrating the encoded traffic impact factor data into existing vehicle density time series data to form an extended data set;
model training and optimizing: retraining STGCN the model by using the extended data set, and optimizing model parameters to improve prediction accuracy by considering the influence of holidays, special activities and weather conditions;
predicting and planning: and predicting the traffic flow change trend of the upcoming holiday by using the optimized model, and planning traffic regulation and control measures in advance by a traffic management department according to the prediction result, wherein the traffic regulation and control measures comprise adjusting signal lamp timing, setting temporary traffic control areas, issuing traffic guidance and the like.
The application effect is as follows: by accurately predicting the traffic flow change of holidays, effective traffic regulation measures are implemented in advance, so that traffic jams of main road sections and business areas are obviously relieved; the traffic signal control is optimized, the waiting time of vehicles is reduced, and the road use efficiency is improved; for the predicted potential congestion point, the traffic management department can quickly respond, adjust the traffic flow direction and avoid the worsening of the congestion condition.
S6 specifically comprises the following steps:
S61: integrating the vehicle density prediction result of the preset area in the future period obtained in the steps S4 and S5 with historical traffic flow data of the past year in the same period to form a comprehensive data set;
S62: defining an optimization target and a constraint, wherein the specific optimization target is to minimize traffic jam and improve road traffic efficiency, the constraint condition comprises road capacity, traffic signal control parameters, road closure or limiting measures, and the specific function of the optimization problem is as follows:
，
subjectto Wherein, the method comprises the steps of, wherein, Represents the firstCongestion cost function for each grid cell,AndRepresenting the predicted traffic flow and traffic control parameters within the grid cell respectively,AndRespectively isAndIs a feasible region of (2);
S63: the optimization problem in the step S62 is solved by adopting a dynamic programming algorithm, the dynamic programming algorithm is used for solving a complex problem into smaller sub-problems, and the solution of the original problem is constructed by utilizing the solution of the sub-problems, so that an optimal traffic regulation strategy is found, and a specific algorithm formula is as follows:
Wherein, the method comprises the steps of, wherein, Is represented in a given current stateDown, from the firstThe minimum congestion cost that can be achieved by starting from the grid cell, which is a cost function in dynamic planning, represents the optimal cost under the specific traffic regulation strategy from the current grid cell;
Represents the first The current traffic state of the individual grid cells, such as traffic density or speed, is a state variable that affects the decision process; Is directed to the first Traffic regulation measures to be taken by the grid units, such as signal lamp adjustment, road section sealing and the like; Representing a set of all possible traffic regulation measures; Is at the first Grid cells, given the current traffic stateAnd take regulation measuresThe immediate congestion cost at the time, which reflects the immediate effect of implementing a particular regulatory measure; Is the first to A set of adjacent grid cells, represented in the traffic network, the firstNeighboring grid cells directly affected by the individual grid cells; representing adjacent grid cells In its current stateThe minimum congestion cost below, this part accounts for the secondary grid cellsThe influence of the regulation decision of (a) on the surrounding grid cells;
S64: according to the result of the dynamic planning algorithm, specific traffic regulation suggestions are generated for each grid unit, wherein the suggestions comprise adjustment of signal lamps, temporary closing or opening of road sections and change of traffic guidance, and the generated suggestions are summarized and transmitted to a traffic management department for real-time traffic regulation.
S7 specifically comprises the following steps:
S71: firstly, collecting actual traffic flow data, including indexes of traffic flow density, speed and passing time, and sorting out traffic flow data predicted by a model and traffic regulation suggestions formulated based on the prediction;
s72: the statistical analysis method of absolute percentage error is adopted to quantitatively compare the actual traffic flow data with the model prediction result, and the calculation formula of the absolute percentage error is as follows:
Wherein, the method comprises the steps of, wherein, Is the average of the absolute percentage error, is used to quantify the prediction accuracy,Is the total number of data points considered in the evaluation,Is the firstThe actual traffic flow value of the data point,The corresponding model predicts the traffic flow value; by calculating MAPE, the average level of the prediction error can be intuitively known, and the lower MAPE value represents the higher prediction accuracy;
s73: the effect of the implemented traffic regulation measures is evaluated, the influence of the traffic regulation measures on the actual traffic flow is analyzed, the specific evaluation comprises the steps of comparing traffic flow data before and after regulation and the accuracy and the effectiveness of the implementation of regulation suggestion, and the specific regulation effect evaluation formula is as follows:
Effectiveness Wherein EFFECTIVENESS denotes the percentage of effect of the regulatory measures, Is the average value of traffic flow or congestion index before the implementation of the regulation measures,Is the corresponding value after the implementation of the regulation measures; the regulation and control effect evaluation formula is used for quantitatively measuring the change degree of the traffic condition before and after the regulation and control measures are implemented, and positive values indicate that the traffic condition is improved;
S74: based on the comparison analysis results of S72 and S73, the defects of the prediction model and the improvement space of the regulation strategy are specifically identified.
S74 specifically includes:
S741: calculating the overall MAPE value predicted by the model to determine the overall prediction accuracy, then carrying out deep analysis on the MAPE value to distinguish MAPE values of different time periods (such as early and late peak periods) and different areas (such as business areas and residential areas) so as to identify the prediction deviation of the model under different conditions, wherein the step may reveal that the model is insensitive to certain traffic mode changes or the prediction accuracy is insufficient, analyzing the distribution characteristics of prediction errors by using statistical analysis tools such as histograms and box charts, finding evidence of systematic deviation, for example, the errors are concentrated in a certain direction (too high or too low), or the errors are abnormally increased under certain conditions, which is helpful for identifying the deficiency of the model on specific events or condition reactions;
s742: calculating the change of traffic flow before and after implementing the regulation measures, quantitatively evaluating the actual influence of each regulation measure by using the regulation effect evaluation formula, and comparing the indexes of traffic flow and speed before and after regulation to evaluate the regulation effect; analyzing the effect of each regulation and control measure, identifying measures which fail to reach the expected target, particularly paying attention to those measures which lead to traffic flow deterioration (such as congestion aggravation and traffic time extension), and analyzing possible reasons of failure, such as improper time arrangement, too small or too large influence range and the like;
S743: providing model optimization suggestions based on the analysis results of S741 and S742, including introducing new data features to enhance the response capability of the model to road construction, traffic accidents and temporary road closure, adjusting the model structure to improve the prediction accuracy of a predetermined area or time period, or retraining the model to better fit actual traffic flow data;
S744: according to the evaluation result of the regulation strategy, the optimization direction is provided for different regulation measures, including the measures of adjusting the implementation period of the regulation measures to adapt to the peak value of traffic flow, enhancing or reducing the target area of the regulation measures to achieve the expected regulation effect or completely replacing the measures with unobvious effect.
The present invention is intended to embrace all such alternatives, modifications and variances which fall within the broad scope of the appended claims. Therefore, any omission, modification, equivalent replacement, improvement, etc. of the present invention should be included in the scope of the present invention.
Claims (9)
1. The intelligent traffic flow analysis method based on the Beidou data is characterized by comprising the following steps of:
s1: collecting vehicle position data in a preset area through a Beidou satellite navigation system;
s2: cleaning and formatting the original data collected in the step S1 by utilizing a data preprocessing algorithm;
s3: according to the data processed in the step S2, a space analysis algorithm is applied to calculate the vehicle density distribution in each time period;
S4: constructing a space-time diagram convolutional network prediction model, analyzing the vehicle density distribution obtained in the step S3, and predicting the traffic flow change trend of a preset area in a short period;
s5: on the basis of S4, introducing traffic influence factors for analysis, and optimizing the prediction accuracy of the traffic flow change trend;
s6: based on the prediction results of the steps S4 and S5, combining historical traffic data, and generating real-time traffic regulation and control suggestions for a traffic management department through a preset dynamic optimization algorithm, wherein the method specifically comprises the following steps:
S61: integrating the vehicle density prediction result of the preset area in the future period obtained in the steps S4 and S5 with historical traffic flow data of the past year in the same period to form a comprehensive data set;
S62: defining an optimization target and a constraint, wherein the specific optimization target is to minimize traffic jam and improve road traffic efficiency, the constraint condition comprises road capacity, traffic signal control parameters, road closure or limiting measures, and the specific function of the optimization problem is as follows:
Wherein C i represents the congestion cost function of the ith grid cell, X i and Y i represent the predicted traffic flow and traffic control parameters within the grid cell,/>, respectively And/>Feasible regions of X i and Y i, respectively;
S63: the optimization problem in the step S62 is solved by adopting a dynamic programming algorithm, the dynamic programming algorithm is used for solving a complex problem into smaller sub-problems, and the solution of the original problem is constructed by utilizing the solution of the sub-problems, so that an optimal traffic regulation strategy is found, and a specific algorithm formula is as follows:
Wherein V i(Xi) represents the minimum congestion cost achievable starting from the ith grid cell given the current state X i; x i represents the current traffic state of the ith grid cell; y i is the traffic control measure to be taken for the ith grid cell; /(I) Representing a set of all possible traffic regulation measures; c i(Xi,Yi) is the direct congestion cost at the ith grid cell given the current traffic state X i and taking regulatory measure Y i; /(I)Is a set of grid cells adjacent to the ith grid cell, representing adjacent grid cells directly affected by the ith grid cell in the traffic network; v j(Xj) represents the minimum congestion cost of the neighboring grid cell j at its current state X j;
S64: according to the result of the dynamic planning algorithm, specific traffic regulation suggestions are generated for each grid unit, wherein the suggestions comprise adjustment of signal lamps, temporary closing or opening of road sections and change of traffic guidance, and the generated suggestions are summarized and transmitted to a traffic management department for real-time traffic regulation;
s7: and comparing the actual traffic flow data with the prediction result and the regulation suggestion, and regulating the prediction model and the traffic regulation suggestion according to the comparison result.
2. The intelligent traffic flow analysis method based on Beidou data according to claim 1, wherein the S1 specifically comprises:
S11: firstly, determining a predetermined area range, including a main urban traffic network, important traffic hubs and high-density commercial areas, and specifically defining the area by an urban traffic planning map and a geographic information system interface of a Beidou satellite navigation system;
s12: in the preset area determined in the step S1, traffic monitoring equipment with a Beidou satellite receiving module is deployed, wherein the traffic monitoring equipment comprises Beidou positioning equipment on a mobile vehicle and a Beidou receiving station fixed on a preset road section, and the Beidou positioning equipment and the Beidou receiving station can receive signals from Beidou satellites and are used for capturing real-time position information of the vehicle;
s13: collecting position data by a Beidou satellite navigation system, wherein the position data comprises longitude, latitude, time stamp, speed and running direction of each vehicle;
s14: the accuracy of the position data is improved by utilizing the differential positioning function of the Beidou satellite navigation system, and the differential positioning realizes higher-accuracy position information measurement by comparing the difference between satellite signals received by the fixed Beidou receiving stations in the preset area and signals received by the Beidou receiving modules on the mobile vehicle;
S15: and transmitting the collected position data to a data processing center for standby through a wireless communication network.
3. The intelligent traffic flow analysis method based on Beidou data according to claim 2, wherein the step S2 specifically comprises:
S21: receiving the original position data from the S1, and performing data cleaning to remove invalid data and abnormal values, wherein the specific operations comprise identifying and eliminating obvious errors in the position data and filtering stationary vehicle data;
s22: formatting the cleaned data to enable the data to conform to a standard format of subsequent analysis, specifically comprising a time stamp format of unified data, and standardizing longitude and latitude, speed and direction data into a preset numerical format to ensure consistency and comparability of the data;
s23: performing data deduplication processing, checking whether repeated records exist in the data set processed in the steps S22 and S23, and replacing repeated items with the latest data records so as to avoid data redundancy in the analysis process;
s25: and data segmentation is carried out, and the data set is divided into a plurality of time periods according to a preset time interval, so that time sequence analysis and traffic flow trend prediction are conveniently carried out.
4. The intelligent traffic flow analysis method based on Beidou data according to claim 3, wherein the step S3 specifically comprises:
S31: presetting a grid system, namely determining a geographic boundary of a preset area through a geographic information system tool; uniformly dividing a preset area into square grid units with fixed sizes according to the total area of the area and the expected analysis precision, wherein the side length of each grid unit is set to be 100 meters, and each grid unit is provided with a unique identifier and an accurate geographic coordinate range;
S32: calculating the vehicle density of grid units in each time period, determining the total number of vehicles passing through each grid unit in unit time by adopting a direct counting method, wherein the vehicle density is defined as the number of vehicles in unit grid area, and the specific calculation formula is as follows: Wherein N represents the total number of vehicles passing through the grid cell within a predetermined period of time; a is the area of the grid unit, namely 10000 square meters;
s33: integrating and creating a vehicle density distribution map, representing the vehicle density value of each grid cell calculated in the step S32 by using color codes, mapping the color codes to geographic positions of the corresponding grid cells, gradually changing the color of the vehicle density from green to red, and visually displaying the vehicle distribution condition in each grid cell.
5. The intelligent traffic flow analysis method based on Beidou data according to claim 4, wherein the constructing of the space-time diagram convolutional network prediction model in the S4 specifically comprises:
s41: initializing a graph structure, and constructing a graph G based on the vehicle density distribution diagram obtained in the step S33, wherein the graph G consists of a node set V and an edge set E, each node represents a grid unit, the connection relation between the nodes is determined based on the adjacency of geographic positions, and particularly when two grid units are adjacent, the nodes corresponding to the two grid units are connected through edges in the graph G;
S42: defining a space-time feature matrix, regarding the vehicle density of each node in different time periods as a feature, and forming a space-time feature matrix X, wherein the size of the feature matrix X is NxT, N is the total number of nodes, and T is the length of an observation time period;
s43: the spatial features are extracted by applying graph convolution, and the graph G is subjected to spatial feature learning by adopting a graph convolution network, wherein the graph convolution operation is specifically expressed as follows: wherein/> Is a contiguous matrix added with self-connection, A is a contiguous matrix of the graph G, I N is an identity matrix,/>Is a diagonal matrix of the node degree matrix, H (l) is a node characteristic matrix of the first layer, W (l) is a weight matrix of the layer, and sigma is a nonlinear activation function;
S44: integrating one-dimensional convolution processing time sequences, and applying one-dimensional convolution to the time sequence characteristics of each node to capture a change mode in a time dimension and provide time characteristics for vehicle density prediction;
S45: constructing a final space-time diagram convolutional network prediction model, integrating the spatial features and the temporal features processed in the steps S43 and S44 to form a complete space-time diagram convolutional network, wherein the function of the space-time diagram convolutional network prediction model is as follows: y=f ST-GCN (X; Θ), where Y represents the vehicle density prediction for the future time period, X is the input spatio-temporal feature matrix, Θ includes parameters of all spatial and temporal convolution layers, and f ST-GCN represents a function of the whole spatio-temporal graph convolution network.
6. The intelligent traffic flow analysis method based on Beidou data according to claim 5, wherein the step S4 of analyzing the vehicle density distribution obtained in the step S3, and predicting traffic flow change trend of a preset area in a short period specifically comprises:
S46: processing the vehicle density distribution data obtained in the step S3 to adapt to the input requirements of a space-time diagram convolution network prediction model, wherein the specific processing comprises time sequence standardization and space feature coding; the time sequence normalization is to normalize the vehicle density time sequence of each grid unit to eliminate the influence of different orders, and the normalization formula is as follows: Wherein X represents an original vehicle density value, mu and sigma represent an average value and a standard deviation of a time sequence respectively, and X std is a normalized value; the space feature codes are used for converting the geographic position information of the grid cells into space feature codes for the space-time diagram convolutional network prediction model to learn, so that the space-time diagram convolutional network prediction model can identify and utilize the spatial interrelationship;
S47: training STGCN models by using the data processed in the step S46, wherein the training aim is to minimize the difference between the predicted vehicle density and the actual vehicle density, and specifically, the mean square error is adopted as a loss function, and the formula is as follows: Where L is the loss function, N is the number of samples, Y pred,i is the model's predicted value for the ith sample, and Y true,i is the corresponding true value;
S48: predicting the vehicle density in a future period of time by using a space-time diagram convolution network prediction model after training is completed, and outputting a predicted vehicle density value of each grid unit at a future time point, wherein the future time point comprises a peak time in the morning and evening, a class time in the school, a working day or a holiday;
s49: according to the vehicle density distribution in S48, a future traffic flow variation trend in the predetermined area is analyzed, the variation trend including identifying a congestion area, and evaluating a variation trend of the traffic volume.
7. The intelligent traffic flow analysis method based on Beidou data according to claim 6, wherein the step S5 specifically comprises:
s51: selecting key influencing factors of traffic flow change, wherein the key influencing factors comprise weather conditions and holidays;
S52: for each determined influencing factor, collecting related data, integrating the collected data into a format which can be identified by a space-time diagram convolution network prediction model, specifically encoding weather conditions into numerical variables, and expressing holidays through Boolean variables;
S53: the input of the expansion model, namely the traffic influence factor data integrated in the step S52 is used as additional characteristics to be input into a space-time diagram convolution network prediction model constructed in the step S4;
S54: on the basis of new input, retraining a space-time diagram convolutional network prediction model, and optimizing model parameters by supervision and learning by using an extended data set comprising traffic influence factors so as to improve the prediction accuracy of future traffic flow change trend;
S55: the optimized space-time diagram convolution network prediction model formula is as follows: y opt＝fST-GCN(X,F;Θopt), wherein Y opt is the model predicted optimized vehicle density distribution, X is the original space-time feature matrix, F represents the feature set of the traffic influencing factors, and Θ opt is the optimized model parameter set.
8. The intelligent traffic flow analysis method based on Beidou data according to claim 7, wherein the step S7 specifically comprises:
S71: firstly, collecting actual traffic flow data, including indexes of traffic flow density, speed and passing time, and sorting out traffic flow data predicted by a model and traffic regulation suggestions formulated based on the prediction;
s72: the statistical analysis method of absolute percentage error is adopted to quantitatively compare the actual traffic flow data with the model prediction result, and the calculation formula of the absolute percentage error is as follows:
Wherein MAPE is the average value of absolute percentage errors, used for quantifying prediction accuracy, N is the total number of data points considered in the evaluation, Y actual,i is the actual traffic flow value of the ith data point, Y predict,i is the corresponding model predicted traffic flow value; by calculating MAPE, the average level of the prediction error can be intuitively known, and the lower MAPE value represents the higher prediction accuracy;
S73: the effect of the implemented traffic regulation measures is evaluated, the influence of the traffic regulation measures on the actual traffic flow is analyzed, the specific evaluation comprises the steps of comparing traffic flow data before and after regulation, and the specific regulation effect evaluation formula is as follows:
Wherein EFFECTIVENESS represents the effect percentage of the regulation measures, trafficbefore is the average value of traffic flow or congestion indexes before the implementation of the regulation measures, and TRAFFICAFTER is the corresponding value after the implementation of the regulation measures; the regulation and control effect evaluation formula is used for quantitatively measuring the change degree of the traffic condition before and after the regulation and control measures are implemented, and positive values indicate that the traffic condition is improved;
S74: based on the comparison analysis results of S72 and S73, the defects of the prediction model and the improvement space of the regulation strategy are specifically identified.
9. The intelligent traffic flow analysis method based on the Beidou data according to claim 8, wherein the step S74 specifically includes:
S741: firstly calculating an integral MAPE value predicted by a model, determining integral prediction accuracy, and then carrying out deep analysis on the MAPE value to distinguish MAPE values of different time periods and different areas so as to identify prediction deviation of the model under different conditions;
S742: calculating the change of traffic flow before and after implementing the regulation measures, quantitatively evaluating the actual influence of each regulation measure by using the regulation effect evaluation formula, and comparing the indexes of traffic flow and speed before and after regulation to evaluate the regulation effect; analyzing the effect of each regulation and control measure, and identifying measures which fail to reach the expected target;
S743: providing model optimization suggestions based on the analysis results of S741 and S742, including introducing new data features to enhance the response capability of the model to road construction, traffic accidents and temporary road closure, adjusting the model structure to improve the prediction accuracy of a predetermined area or time period, or retraining the model to better fit actual traffic flow data;
S744: according to the evaluation result of the regulation strategy, the optimization direction is provided for different regulation measures, including the measures of adjusting the implementation period of the regulation measures to adapt to the peak value of traffic flow, enhancing or reducing the target area of the regulation measures to achieve the expected regulation effect or completely replacing the measures with unobvious effect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410317217.9A CN117935561B (en) | 2024-03-20 | 2024-03-20 | Intelligent traffic flow analysis method based on Beidou data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410317217.9A CN117935561B (en) | 2024-03-20 | 2024-03-20 | Intelligent traffic flow analysis method based on Beidou data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117935561A CN117935561A (en) | 2024-04-26 |
| CN117935561B true CN117935561B (en) | 2024-05-31 |
Family
ID=90752184
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410317217.9A Active CN117935561B (en) | 2024-03-20 | 2024-03-20 | Intelligent traffic flow analysis method based on Beidou data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117935561B (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107293133A (en) * | 2017-08-14 | 2017-10-24 | 清华大学深圳研究生院 | A kind of method for controlling traffic signal lights |
| CN110264717A (en) * | 2019-06-25 | 2019-09-20 | 牡丹江师范学院 | A kind of municipal intelligent traffic regulator control system |
| CN111127879A (en) * | 2019-12-09 | 2020-05-08 | 湖南大学 | Urban traffic flow prediction method based on generative countermeasure network |
| CN112183862A (en) * | 2020-09-29 | 2021-01-05 | 长春理工大学 | Traffic flow prediction method and system for urban road network |
| CN112950924A (en) * | 2019-12-10 | 2021-06-11 | 东北大学秦皇岛分校 | Complex traffic network traffic speed prediction method based on deep learning |
| CN113450561A (en) * | 2021-05-06 | 2021-09-28 | 浙江工业大学 | Traffic speed prediction method based on space-time graph convolution-generation countermeasure network |
| WO2021232387A1 (en) * | 2020-05-22 | 2021-11-25 | 南京云创大数据科技股份有限公司 | Multifunctional intelligent signal control system |
| CN113792929A (en) * | 2021-04-26 | 2021-12-14 | 青岛大学 | Traffic flow prediction method, electronic device, and storage medium |
| CN114049764A (en) * | 2021-10-27 | 2022-02-15 | 东南大学 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
| CN114202917A (en) * | 2021-12-02 | 2022-03-18 | 安徽庐峰交通科技有限公司 | Construction area traffic control and induction method based on dynamic traffic flow short-time prediction |
| CN116258258A (en) * | 2023-02-20 | 2023-06-13 | 内蒙古工业大学 | Multi-information fusion space-time diagram convolution traffic flow prediction method |
| CN117496726A (en) * | 2023-10-20 | 2024-02-02 | 华东交通大学 | Traffic control coordination method and system based on intelligent traffic system |
| CN117575349A (en) * | 2023-11-13 | 2024-02-20 | 上海营邑城市规划设计股份有限公司 | Urban data model-based traffic influence evaluation method and system |
| CN117671992A (en) * | 2024-01-30 | 2024-03-08 | 山东恒宇电子有限公司 | Intelligent bus dispatching method and system |
-
2024
- 2024-03-20 CN CN202410317217.9A patent/CN117935561B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107293133A (en) * | 2017-08-14 | 2017-10-24 | 清华大学深圳研究生院 | A kind of method for controlling traffic signal lights |
| CN110264717A (en) * | 2019-06-25 | 2019-09-20 | 牡丹江师范学院 | A kind of municipal intelligent traffic regulator control system |
| CN111127879A (en) * | 2019-12-09 | 2020-05-08 | 湖南大学 | Urban traffic flow prediction method based on generative countermeasure network |
| CN112950924A (en) * | 2019-12-10 | 2021-06-11 | 东北大学秦皇岛分校 | Complex traffic network traffic speed prediction method based on deep learning |
| WO2021232387A1 (en) * | 2020-05-22 | 2021-11-25 | 南京云创大数据科技股份有限公司 | Multifunctional intelligent signal control system |
| CN112183862A (en) * | 2020-09-29 | 2021-01-05 | 长春理工大学 | Traffic flow prediction method and system for urban road network |
| CN113792929A (en) * | 2021-04-26 | 2021-12-14 | 青岛大学 | Traffic flow prediction method, electronic device, and storage medium |
| CN113450561A (en) * | 2021-05-06 | 2021-09-28 | 浙江工业大学 | Traffic speed prediction method based on space-time graph convolution-generation countermeasure network |
| CN114049764A (en) * | 2021-10-27 | 2022-02-15 | 东南大学 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
| CN114202917A (en) * | 2021-12-02 | 2022-03-18 | 安徽庐峰交通科技有限公司 | Construction area traffic control and induction method based on dynamic traffic flow short-time prediction |
| CN116258258A (en) * | 2023-02-20 | 2023-06-13 | 内蒙古工业大学 | Multi-information fusion space-time diagram convolution traffic flow prediction method |
| CN117496726A (en) * | 2023-10-20 | 2024-02-02 | 华东交通大学 | Traffic control coordination method and system based on intelligent traffic system |
| CN117575349A (en) * | 2023-11-13 | 2024-02-20 | 上海营邑城市规划设计股份有限公司 | Urban data model-based traffic influence evaluation method and system |
| CN117671992A (en) * | 2024-01-30 | 2024-03-08 | 山东恒宇电子有限公司 | Intelligent bus dispatching method and system |
Non-Patent Citations (4)
| Title |
|---|
| Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting;Yu, B (Yu, Bing) 等;《 PROCEEDINGS OF THE TWENTY-SEVENTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE》;20180101;全文 * |
| 基于深度强化学习的单路口交通信号控制;刘皓;吕宜生;;交通工程;20200415(第02期);全文 * |
| 基于自适应动态规划的城市交通信号优化控制方法综述;赵冬斌;刘德荣;易建强;;自动化学报;20090615(第06期);全文 * |
| 封闭路拥堵信号灯系统;庞婷;;电子制作;20130715(第13期);正文全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117935561A (en) | 2024-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| GB2582532A (en) | Method for detecting traffic anomaly of urban road based on peak hour traffic flow rate | |
| US9111442B2 (en) | Estimating incident duration | |
| US9454902B2 (en) | Performing-time-series based predictions with projection thresholds using secondary time-series-based information stream | |
| Li et al. | Identifying important variables for predicting travel time of freeway with non-recurrent congestion with neural networks | |
| CN104778837A (en) | Multi-time scale forecasting method for road traffic running situation | |
| González-Vidal et al. | IoT for water management: Towards intelligent anomaly detection | |
| CN113496314B (en) | Method for predicting road traffic flow by neural network model | |
| CN109979193B (en) | Data anomaly diagnosis method based on Markov model | |
| CN106887141A (en) | Continuous Traffic node congestion degree forecast model based on queueing theory, system and method | |
| CN111815098A (en) | Traffic information processing method and device based on extreme weather, storage medium and electronic equipment | |
| CN114463972A (en) | Road section interval traffic analysis and prediction method based on ETC portal communication data | |
| Mihaita et al. | Traffic congestion anomaly detection and prediction using deep learning | |
| CN116631186B (en) | Expressway traffic accident risk assessment method and system based on dangerous driving event data | |
| CN115331425A (en) | Traffic early warning method, device and system | |
| CN111179592A (en) | Urban traffic prediction method and system based on spatio-temporal data flow fusion analysis | |
| CN116092294A (en) | Expressway traffic flow prediction method based on Transformer and graph meaning network | |
| CN114283590B (en) | Traffic flow peak prediction method and device and electronic equipment | |
| CN111985731B (en) | Method and system for predicting number of people at urban public transport station | |
| CN116863723B (en) | Use method of digital twin base | |
| CN117935561B (en) | Intelligent traffic flow analysis method based on Beidou data | |
| CN116758762A (en) | Control method based on big data | |
| CN113191568B (en) | Meteorological-based urban operation management big data analysis and prediction method and system | |
| Salamanis et al. | Evaluating the effect of time series segmentation on STARIMA-based traffic prediction model | |
| Wu et al. | Location–allocation strategies for traffic counters—a citywide deployment | |
| Hazeghi et al. | Designing and developing a model for detection of unusual traffic condition at intelligent signalized intersection equipped with SCATS |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |