CN114186710A - Method and system for predicting average speed of road section during short-time large-scale activity - Google Patents
Method and system for predicting average speed of road section during short-time large-scale activity Download PDFInfo
- Publication number
- CN114186710A CN114186710A CN202111246104.7A CN202111246104A CN114186710A CN 114186710 A CN114186710 A CN 114186710A CN 202111246104 A CN202111246104 A CN 202111246104A CN 114186710 A CN114186710 A CN 114186710A
- Authority
- CN
- China
- Prior art keywords
- influence
- data
- road section
- scale activity
- short
- 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.)
- Pending
Links
- 230000000694 effects Effects 0.000 title claims abstract description 177
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000007637 random forest analysis Methods 0.000 claims abstract description 51
- 238000007667 floating Methods 0.000 claims abstract description 46
- 230000002093 peripheral effect Effects 0.000 claims abstract description 37
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 230000008859 change Effects 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims description 54
- 238000004458 analytical method Methods 0.000 claims description 28
- 230000008569 process Effects 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 8
- 238000003066 decision tree Methods 0.000 description 27
- 238000012545 processing Methods 0.000 description 9
- 230000009467 reduction Effects 0.000 description 8
- 238000011156 evaluation Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 208000025174 PANDAS Diseases 0.000 description 3
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 3
- 240000004718 Panda Species 0.000 description 3
- 235000016496 Panda oleosa Nutrition 0.000 description 3
- 238000010835 comparative analysis Methods 0.000 description 3
- 238000007418 data mining Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000012535 impurity Substances 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000010420 art technique Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 235000012489 doughnuts Nutrition 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Health & Medical Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a method and a system for predicting the average speed of a road section during a short-time large-scale activity, which comprises the following steps: acquiring original data of a floating car, original data of a detector and large-scale activity data; analyzing the running speed and section traffic volume of the peripheral road section of the large-scale activity data to determine influence characteristic information; the influence characteristic information comprises an influence time period, an influence degree and a speed change; determining influence factors according to the large-scale activity data and the influence characteristic information; obtaining an average speed prediction result by utilizing a road section average speed prediction model in a short-time large-scale activity period according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm. The invention improves the efficiency and the precision of the vehicle speed prediction by considering the influence of short-time large-scale activities on traffic.
Description
Technical Field
The invention relates to the field of public transport data mining application and service evaluation, in particular to a method and a system for predicting the average speed of a road section during a short-time large-scale activity.
Background
With the improvement of national economic level, more and more large-scale activities are held in each big city, and the development of the activities leads to the concentration of people in the area, thereby not only causing great pressure on peripheral infrastructure, but also being easy to have group events. The method has the advantages that the accurate prediction of the average vehicle speed reduction of the roads around the venue caused by the large-scale activities and the adoption of effective control measures for dispersion are important problems to be solved by relevant departments, so that the influence of the short-time large-scale activities on the surrounding roads is necessarily researched, and intrinsic characteristic rules are searched, so that bases are provided for making and implementing countermeasures, means and measures in a targeted manner, and the method has important significance for guaranteeing good traffic environment during the large-scale activities and ensuring the smooth development of the activities.
Because the daily dimension or traffic demand prediction research in China is relatively more and complete, the main research objects are continuous large-scale activities such as the Olympic Games and the world Expo Games, and the research on the traffic operation characteristics of roads around a large-scale activity venue is less and has a single angle. In the prior art, a time convolution neural network is used for replacing a recurrent neural network, so that the calculation speed is increased, and the prediction result is more accurate. Although the multitask learning model can predict the vehicle speeds of a plurality of roads simultaneously by using one model, the multitask learning model cannot reflect the characteristics of the vehicle speeds of different road sections around a venue holding a large event under different factors. The existing technology collects the video information of the traffic crossing; analyzing the traffic comprehensive information according to the collected video information, and loading the traffic comprehensive information into historical traffic comprehensive information; constructing a traffic model according to the historical traffic comprehensive information, and predicting the predicted traffic comprehensive information of the traffic crossing at a node of a predetermined time in the future; generating a traffic light control signal of the traffic crossing according to the predicted traffic comprehensive information; and controlling the execution of the traffic lights at the traffic crossing according to the traffic light control signals. However, the method only obtains traffic data through the existing monitoring system, if the traffic data cannot be corrected in time, a good traffic light control effect cannot be guaranteed, and the real-time matching degree is not high. Some prior art techniques calculate a recent predicted vehicle speed; calculating a long-term predicted vehicle speed; calculating a hybrid predicted vehicle speed using the recent predicted vehicle speed and the future predicted vehicle speed; and calculating the required travel time of the whole route according to the predicted vehicle speed of each section in the route. However, this vehicle speed prediction method only considers the near-term correlation and the far-term correlation, and many influence factors are not taken into consideration.
Therefore, there is a need for a vehicle speed prediction method that can improve the efficiency and accuracy of vehicle speed prediction while taking into account the impact of short-term heavy events on traffic.
Disclosure of Invention
The invention aims to provide a method and a system for predicting the average speed of a road section during a short-time large-scale activity, which improve the efficiency and the accuracy of speed prediction under the condition of considering the influence of the short-time large-scale activity on traffic.
In order to achieve the purpose, the invention provides the following scheme:
a method for predicting average vehicle speed of a road section during a short-time large-scale activity comprises the following steps:
acquiring original data of a floating car, original data of a detector and large-scale activity data;
analyzing the running speed and section traffic volume of the peripheral road section of the large-scale activity data to determine influence characteristic information; the influence characteristic information comprises an influence time period, an influence degree and a speed change;
determining influence factors according to the large-scale activity data and the influence characteristic information;
obtaining an average speed prediction result by utilizing a road section average speed prediction model in a short-time large-scale activity period according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm.
Optionally, after the acquiring raw data of the floating car, raw data of the detector, and large-scale activity data, the method further includes:
and removing and screening the floating car original data and the detector original data.
Optionally, the analyzing the operation speed and the section traffic volume of the peripheral road section on the large-scale activity data to determine the influence characteristic information specifically includes:
analyzing the running speed of the peripheral road sections of the large-scale activity data, and determining the peripheral road section influence time period in the influence time period, the peripheral road section influence degree in the influence degree and the speed change;
and analyzing the cross section traffic volume of the large-scale activity data, and determining the cross section influence time interval in the influence time interval and the cross section influence degree in the influence degree.
Optionally, the construction process of the road segment average vehicle speed prediction model during the short-time large-scale activity period specifically includes:
adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model;
constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers;
and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
A system for predicting average vehicle speed of a road segment during a short-term large-scale activity, comprising:
the acquisition module is used for acquiring original data of the floating car, original data of the detector and large-scale activity data;
the influence characteristic information determining module is used for analyzing the running speed and the section traffic volume of the peripheral road section of the large-scale activity data and determining influence characteristic information; the influence characteristic information comprises an influence time period, an influence degree and a speed change;
the influence factor determining module is used for determining influence factors according to the large-scale activity data and the influence characteristic information;
the average speed prediction result determining module is used for obtaining an average speed prediction result by utilizing a short-time large-scale activity period road section average speed prediction model according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm.
Optionally, the method further includes:
and the removing and screening module is used for removing and screening the original data of the floating car and the original data of the detector.
Optionally, the influence characteristic information determining module specifically includes:
the peripheral road section operation speed analysis unit is used for carrying out peripheral road section operation speed analysis on the large-scale activity data and determining the peripheral road section influence time interval in the influence time interval, the peripheral road section influence degree in the influence degree and the speed change;
and the section traffic volume analysis unit is used for carrying out section traffic volume analysis on the large-scale activity data and determining the section influence time interval in the influence time interval and the section influence degree in the influence degree.
Optionally, the construction process of the road segment average vehicle speed prediction model during the short-time large-scale activity period specifically includes:
adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model;
constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers;
and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a method and a system for predicting the average speed of a road section during a short-time large-scale activity, which are used for analyzing the running speed and section traffic volume of a peripheral road section of large-scale activity data and determining influence characteristic information; the influence characteristic information comprises influence time intervals, influence degrees and speed changes; determining influence factors according to the large-scale activity data and the influence characteristic information; obtaining an average speed prediction result by utilizing a road section average speed prediction model in a short-time large-scale activity period according to the original data of the floating car, the original data of the detector and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm. The method is characterized in that a random forest-based prediction model of the average speed of the road section in the short-time large-scale activity period is constructed, the random forest can process high-dimensional data, and the method has the characteristics of strong generalization capability, high training speed and the like. Therefore, the method provided by the invention has the advantages of high prediction speed and high precision.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flowchart of a method for predicting average speed of a road section during a short-time large-scale activity according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method and a system for predicting the average speed of a road section during a short-time large-scale activity, which improve the efficiency and the accuracy of speed prediction under the condition of considering the influence of the short-time large-scale activity on traffic.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
As shown in FIG. 1, the invention provides a method for predicting average vehicle speed of a road section during a short-time large-scale activity, which comprises the following steps:
step 101: raw data of the floating car, raw data of the detector and large-scale activity data are obtained.
Step 102: analyzing the running speed and section traffic volume of the peripheral road section of the large-scale activity data to determine influence characteristic information; the influence characteristic information includes an influence period, an influence degree and a speed change. The analysis of the running speed and the section traffic volume of the peripheral road section is carried out on the large-scale activity data, and influence characteristic information is determined, and the analysis specifically comprises the following steps: analyzing the running speed of the peripheral road sections of the large-scale activity data, and determining the peripheral road section influence time period in the influence time period, the peripheral road section influence degree in the influence degree and the speed change; and analyzing the cross section traffic volume of the large-scale activity data, and determining the cross section influence time interval in the influence time interval and the cross section influence degree in the influence degree.
Step 103: and determining influence factors according to the large-scale activity data and the influence characteristic information.
Step 104: obtaining an average speed prediction result by utilizing a road section average speed prediction model in a short-time large-scale activity period according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm.
The construction process of the road section average speed prediction model in the short-time large-scale activity period specifically comprises the following steps: adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model; constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers; and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
Wherein, in the step 101: after acquiring the raw data of the floating car, the raw data of the detector and the large-scale activity data, the method further comprises the following steps: and removing and screening the floating car original data and the detector original data.
The invention also provides a prediction method of the average speed of the road section during the short-time large-scale activity period in practical application, which comprises the following steps:
step 1, analysis data processing:
step 1.1, extracting the main field contents of the original data of speed and traffic volume:
the basic principle of the floating car information acquisition technology is as follows: the GPS receiving device records the position coordinates, the speed, the time data and the like of the vehicle at certain time intervals, and the vehicle-mounted intelligent equipment transmits the data to the control center by utilizing the communication equipment after acquiring the GPS data. The control center uses related algorithms such as map matching, path conjecture and the like to associate the floating car data with the road network data to obtain traffic parameters such as the interval speed, the travel time and the like of each road section. According to the running state of road traffic, the data needs to have stronger stability, and can reflect the running conditions of the road traffic of different road sections and time periods, and the data of 5min time interval granularity can well reflect the running state of vehicles on the road, and has better stability.
The floating car data of the time granularity comprises the following main fields: link name, link direction, link start and end point, link length, travel time, average travel speed, time, and date, as shown in table 1. Table 1 is a floating car data main field table.
TABLE 1 Floating car data Primary field
The fields mainly contained in the detector data are: detector number, date, time of day, traffic flow, speed, link type, and detection occurrence time, as shown in table 2. The detector number, date, time of day and traffic flow data are mainly used in these fields, as shown in table 2. The detector numbers are used to match the road segments and directions to be analyzed, the time instants are used to determine the time intervals, and the traffic flow data are used to analyze the status of the road segments.
TABLE 2 Detector data Primary field
Inputting original data of the floating car and original data of the detector, analyzing and processing the data through a Pandas module in Python, extracting the original data of the floating car to obtain a road section name, a road section direction and an average driving speed field, obtaining the number, date, time and traffic flow data of the detector by the original data of the detector, applying the field data to the step 2, and cleaning the data.
Step 1.2, the extracted data of the floating car and the detector are cleaned:
the rules for eliminating error data and screening valid data are as follows:
(1) and eliminating records with the same detector number and time in the data records, and performing deduplication processing on the data by using a pandas drop duplicates () function in python.
(2) Deleting 'V', namely the speed field is '0 or-1', and representing that the record is in the detector speed abnormal data; and applying the cleaned field data to step 5, generation of a training set.
Step 2, analyzing the space-time characteristics of the influence of large activities on the operation of surrounding roads:
step 2.1, determining an influence range and a road section set;
the large event data mainly comprises the week of the large event, the holding date, the weather condition (whether it is raining or not, haze or air temperature, etc.), the name of the event, the type of the event, the holding venue, the number of the participants of the event and the starting and ending time.
The different types, contents and sizes of the events affect different space-time ranges of the large events according to the difference between the holding place and the nature of the large events. For the study on the influence of short-term events on surrounding roads, the specific range and road affected by the event at the place where a large event is held are first clarified.
Step 2.2, analyzing the influence characteristics of short-term large activities on the operation of surrounding roads;
influence characteristic analysis is carried out on road sections with different grades and different distances from the activity hall, and the influence degree of the large-scale activity on the road sections is quantitatively shown through analyzing the running speed and the section traffic volume of the surrounding road sections holding the large-scale activity, so that the influence time period is clear. Specific aspects of the comparison from the section average vehicle speed include: (1) influence on the time period; (2) influence degree; (3) speed variation aspects; the analysis of traffic volume from the road section comprises the following specific aspects: (1) influence on the time period; (2) influence degree. Different road sections are affected differently, the road section closer to the venue is affected by activities in advance to cause speed reduction of the road section, and traffic flow is increased; the express way is affected to a greater extent by activity.
Step 3, researching the influence factors of the average speed of the road section:
the establishment of the random forest model needs to consider factors influencing the change rule of the data volume so as to achieve the purpose of accurately learning and predicting the data, construct a training sample set, and analyze the influencing factors so as to determine the maximum characteristic number of the model.
(1) Different roads
The speed of the vehicle may be affected to varying degrees due to differences in road grade, distance to the event venue, and the like. The road section close to the activity venue is influenced by large activities in advance, so that the speed of the road section is reduced; due to the fact that the designed speeds of roads in different levels are different, the average speed of the express way road sections is reduced by a larger range than that of other levels.
(2) Date attribute
For large-scale activities, legal holidays and double-holiday road sections are generally influenced to a great extent, and under the condition of different date attributes (working days, double-holidays and holidays), the change characteristics and rules of the average speed of the road sections around the large-scale activities show different characteristics and have great differences.
(3) Time period
The average speed of the road sections in each time period is different, and the time distribution is not balanced. The average speed of the early and late peak road sections is low compared with other time sections, so that the time sections need to be included in the prediction model parameters when specific parameters are considered.
(4) Week
The average speed of the road sections in each day of the week can be different, so that the week needs to be considered in the subsequent prediction model as a parameter in the influence factors.
(5) Month of the year
The average vehicle speed of each month road section in one year can have difference, so the month needs to be considered in the subsequent prediction model as a parameter in the influence factors.
(6) Nature of movement
The short-time large-scale activities are various, including literary and artistic activities such as a concert, sports activities such as a football game, commercial activities such as product release and an exhibition, and the like. Due to the different audiences facing large activities of different nature, the choice of transportation means may be different.
(7) Size of activity
The activity scale refers to the number of watching people attracted by a short-time large-scale activity, the size of the activity scale influences the total number of the number of people participating in the activity, and the increase of traffic flow directly influences the running condition of a road section. Therefore, the influence of large-scale activities of different sizes on the average speed of a road section is direct and is an important parameter in prediction.
(8) Weather conditions
The weather has great influence on the ground traffic operation conditions, and the selection of the travel modes of people is also influenced. The influence of the difference of the precipitation on the average speed of the vehicle can be different, so that the weather condition is one of factors to be considered in modeling.
(9) Time from beginning to end of activity
Through the analysis of the average speed data, the road sections are influenced 1-2 hours before the start of the activity, and the road sections are scattered within 1 hour after the end of the activity, so that extremely high traffic demands are generated in the two time periods, and great pressure is caused on the peripheral road network.
(10) Restriction of traffic
The restricted range of the working day Beijing is five-loop (including inner roads), the license plate tail number alternation mode is executed according to the license plate tail number alternation mode of motor vehicles in Beijing, and the license plate tail number alternation mode is changed once every three months. Generally, under the condition of limiting different license plate tail numbers, the congestion degrees of road sections are different, and under the general condition, the speed of the road sections is the lowest when the roads are limited by 4 and 9, and the roads are the most congested compared with the roads limited by other tail numbers.
And inputting a training sample influence factor set, performing attribute selection as shown in table 3, applying the attribute selection to the step 4, and performing standardization processing.
Table 3 data attribute selection
Step 4, quantifying and optimizing model parameters:
(1) in the method for predicting the average vehicle speed of the road section during the specific short-time large-scale activity, the physical meanings represented by the parameters are different, so that the parameters have dimensional difference. The heterogeneity is a main factor influencing the overall evaluation of the object, so that the parameters are standardized uniformly before the evaluation, the washed field data and the training sample influence factor set are input, and the data processing parameters of the Pandas in Python are standardized.
(2) Optimal iteration number of weak learner: n _ estimators represents the maximum number of iterations of the weak learner, i.e., the number of the largest weak learners. Too little n _ estimators will result in poor prediction accuracy, too much will result in data overfitting, and the optimal value of n _ estimators is explored using the scimit-leran random forest class library GridSearchCV method in Python.
The parameter adjustment algorithm is as follows:
gsearch1＝GridSearchCV(estimator＝RandomForestClassifier(min_samples_split＝100,min_samples_leaf＝20,max_depth＝8,max_features＝'sqrt',random_state＝10),param_grid＝param_test1,scoring＝'roc_auc')
gsearch1.fit(x,y)
wherein: min _ samples _ split is the minimum sample number required by internal node subdivision, min _ samples _ leaf is the minimum sample number of leaf nodes, max _ depth is the maximum depth of the decision tree, max _ features is the maximum feature number, random _ state is a random state, sorting is a score attribute, and fit is the result of each row of samples in the training set.
And performing grid search on n _ estimators, so that the output scoring indexes are optimally scored, and finally determining to use { n _ estimators-10 } as the optimal iteration number of the model. Table 4 is a table of the best iteration number scoring result, as shown in table 4.
TABLE 4 best iteration number scoring results table
(3) Maximum feature number: the maximum number of features is denoted by max _ features, None by default, representing that all features are considered in the partition, and besides, log2 denotes that at most log is considered2N features, sqrt/auto, represent the maximum consideration
The feature, int (n), represents an integer which can be freely selected from any number greater than 1 and less than the number of features. In the present document, the influence factors of the average speed of the road section include weather, week, restriction, and the like, which all have a non-negligible influence on the average speed of the vehicle on the road section at that time, and { max _ features ═ 13} is selected and used as the optimal feature number of the model.
(4) Decision tree maximum depth and minimum number of samples required: the maximum depth of the decision tree is represented by max _ depth, and in order to reduce the training time of the model, prevent the decision tree from generating an overfitting phenomenon after being branched all the time and improve the overall generalization capability of the model, the decision tree is subjected to depth limitation; min _ samples _ split represents the minimum number of samples required for the inner node to subdivide, and the subtree can be restricted from continuing to divide. The minimum number of samples required for the subdivision scoring results are shown in table 5.
TABLE 5 scoring results table for minimum number of samples required for subdivision
The parameter adjustment algorithm is as follows:
gsearch2＝GridSearchCV(estimator＝RandomForestClassifier(n_estimators＝10,min_samples_leaf＝20,max_features＝'sqrt',oob_score＝True,random_state＝10),param_grid＝param_test2,scoring＝'roc_auc',iid＝False)
gsearch2.fit(x,y)
wherein: the optimal number of iterations for n _ estimators is 10, and oob _ score is whether to use an out-of-bag sample.
And carrying out grid search on the maximum depth max _ depth of the decision tree and the minimum sample number min _ samples _ split required by the internal node subdivision. And finally, adjusting parameters through the model, wherein { max _ depth ═ 20} is used as the optimal iteration number of the model, and { min _ samples _ split ═ 2} is used as the optimal characteristic number of the model.
Step 5, constructing a short-time large-scale activity period road section average speed prediction model based on the random forest:
step 5.1, construction of random forest model
The construction of the random forest model comprises three main steps: firstly, sampling in the existing data set to generate a training set; then, a decision tree is constructed by using the training set; and finally, generating a random forest, and executing a classification and regression algorithm.
(1) Random sampling to generate training set
Each decision tree has a corresponding training set to train, so the same number of data sets must be generated from the original total number set. Random sampling is replaced in the process of generating the data set by using a random sampling (bootstrap) method, namely after single sample is extracted, the extracted sample is still replaced in the original data set.
(2) Building decision trees
Decision trees are unit classifiers that constitute a random forest, similar to tree models in data structures. The decision tree first generates relevant rules by training the classifier with a training set of known classes, and then applies this part of the rules to classify and data mine the data set of unknown classes.
The random forest algorithm selected by the invention utilizes a classification regression tree (CART) as a meta classifier. The CART algorithm adopts a binary recursive mode to construct a tree, each division is a binary division, and a current sample set is divided into two subsets, so that subtrees of a left branch and a right branch are generated. The CART algorithm attribute index is a Gini coefficient. The kini coefficient itself reflects the uncertainty of the sample. When the kini coefficient is smaller, the difference between samples is small, and the uncertainty is low. The process of classification is itself a process of uncertainty reduction, i.e. an increase in purity. Therefore, when the CART algorithm is used for node splitting, the splitting principle is used for minimizing the Gini coefficient.
The kini coefficients for sample set D are defined as follows:
wherein P isiAnd m is the probability of the ith class of the sample in D and the total number of types.
A weighted sum of the impurity levels of each partitioned result partition is calculated. If D is based on the binary division result of A (A is a certain type of attribute) as two subsets D1And D2Then the kini coefficient of this division D is:
the impurity level reduction caused by the division of D according to the attribute A is as follows:
ΔGINI(A)＝GINI(D)-GINIA(D) (3)
the feature attributes participating in splitting are feature variables, a group of attributes are obtained through a method of putting back random sampling at splitting nodes of each tree, an optimal attribute is selected from the group of attributes to segment data to generate an independent decision tree, and finally the category attribute of an input sample is determined according to voting results of all the decision trees.
(3) Generating random forests
Because the decision tree is attributed to a single classifier, a relatively single decision method is used, and thus there are many problems that cannot be avoided. For example, the classification rule is complicated, and problems such as replacement of the entire optimal solution by a partial optimal solution, overfitting, and the like may occur. The random forest algorithm is formed by combining a plurality of decision trees, the decision trees of the components do not need high classification accuracy, and the final result of the random forest operation is generated by voting of a plurality of decision trees and has higher precision.
Assuming that y is an output variable, the sample data set consisting of (x, y) is called an original sample data set. The classification result needs to make a comprehensive decision on all decision tree classification results, and the category of the input variable x is the category with the most votes. The final classification results were as follows:
wherein h (x) is a combined classifier model representing the classification result of the random forest, hi(x) Is a model of decision tree classification, representing a single decision tree classification result, I () is used as an indicator function (representing function to make the value of the set 1, not 0), and y is the classification target.
Step 5.2, model algorithm process
And importing the data needing regression into each weak learner after training, performing regression analysis, outputting results, and performing weighting and normalization processing on the judgment result of each weak classifier to obtain the final judgment result of the whole random forest model. The specific algorithm flow is as follows:
assuming that there are M features in the original data set O, the algorithm flow of the random forest is as follows:
(1) resampling with feedback from original data set O to generate n training subsets O ═ O { O } with same sample size as original data set1,O2,…,On}。
(2) When each decision tree is constructed, one of training subsets is selected as a training set of the decision tree, M (M < M) features are randomly selected from all the features, a CART splitting algorithm is used for splitting the nodes of the decision tree based on the M features, the process is continuously continued until a preset condition is reached, and each decision tree is not pruned.
(3) And combining the generated n completely grown decision trees to form a random forest.
(4) When the test samples are input into the random forest model, the output result of the random forest is simple majority voting decision or averaging. When the random forest model solves the classification problem, the final result is determined by the mode of each decision tree result; when the random forest model solves the regression problem, the final result is the average value of each decision tree result.
Step 5.3, model evaluation index
The accurate and reasonable evaluation index plays a great role in optimizing model parameters, evaluating the reasonability of model selection and checking the accuracy of a prediction result. For regression model prediction, the corresponding indexes are selected as follows:
(1) mean Absolute Error (Mean _ Absolute _ Error, MAE)
Wherein n issamplesTo predict the quantity, yiIn the form of an actual value of the value,
is a predicted value.
(2) Mean square Error (Mean _ Squared _ Error, MSE)
(3) Determination coefficient (r2_ score)
Wherein,
is the average of the actual values.
The invention selects Beijing worker gym as a calculation object, and carries out comparative analysis research on the average speed of a road and the section traffic volume under the condition of whether short-time large-scale activities are held, comprising the following steps:
step 1, analysis data processing:
acquiring floating car data with granularity of 5min time intervals according to the running state of road traffic comprises the following steps: road section name, road section direction, road section starting and ending point, road section length, travel time, average driving speed, time and date; the detector data comprises a detector number, date, time, traffic flow, speed, road section type and detection occurrence time, and the large-scale activity data mainly comprises a large-scale activity week, a holding date, weather conditions (whether rainfall, haze, air temperature and the like), an activity name, an activity type, a holding venue, activity participation numbers and start and end time.
Step 2, analyzing the space-time characteristics of the influence of large activities on the operation of surrounding roads:
step 2.1, determining an influence range and a road section set:
for the research on the influence of the short-time activities on the surrounding roads, the specific range and the road influenced by the activities in the stadium of workers are firstly determined. Specific road segments and names are shown in table 6.
TABLE 6 surrounding influence road section of short-term large-scale activities in worker stadium
Step 2.2, analyzing the influence characteristics of short-time large activities on the operation of surrounding roads:
through the analysis of the running speed and the section traffic volume of the peripheral road sections for holding the large-scale activities in the stadium of Beijing workers, the influence degree of the large-scale activities on the road sections is quantitatively shown, the influence time period is clear, and the summary is as follows:
from comparative analysis of the average speed of the road segment, it can be obtained:
(1) aspect of influence period
The time period of influence of activities held at night in Beijing workshops on the peripheral road network is mainly concentrated on 1-2 hours before starting. The distance from each road section to the Beijing worker stadium is two dong rings, a new east road, a worker stadium north road and a worker stadium east road from far to near in sequence (the distance from the worker stadium north road and the worker stadium east road to the activity place is the same). The time of the influence of the activities on the two road sections of the east road of the worker stadium and the north road of the worker stadium is about 2 hours before the activity starts, the time of the influence on the new east road is about 1.5 hours, and the speed of the influence on the two road sections of the east road and the east road section of the two ring road is reduced 1 hour before the activity starts. The affected time of the four road sections within the activity starting time range lasts for about 2.5 hours. At the end of the large campaign, the duration of the impact on the surrounding road segments was about 1 hour.
(2) Aspect of degree of influence
The pressure on the northeast two-ring section as an express way section connected to a Beijing worker is larger, and in the activity starting stage, compared with the normal state, the maximum congestion time period has about 54.73% of amplitude reduction, 40.34% of new eastern road amplitude reduction, 37.99% of worker stadium northeast road amplitude reduction and 26.66% of worker stadium eastern road amplitude reduction. The donut loop segment is affected the most. At the end of the activity, the degree of influence has the same characteristic law.
(3) Aspect of speed variation
The driving speed of the four road sections is gradually increased after the end of the late peak in the period of no large-scale activity on non-working days. In the time range of the beginning of the activity, the speed of the east dicyclo is kept between 50km/h and 70km/h during the period without large-scale activity, and the speed is kept between 20km/h and 30km/h during the period with large-scale activity; the speed fluctuates around 30km/h when no large-scale activities exist in the new east road, and the speed is reduced to 10 km/h-20 km/h when large-scale activities are held; the speed of the worker on the north road of the stadium fluctuates at 25km/h when no large-scale activity exists, and the speed is about 15km/h when the large-scale activity exists; the speed of the east road of the worker stadium is between 20km/h and 30km/h when no large-scale activity exists, and the speed is reduced to be below 20km/h under the influence of the large-scale activity.
From the analysis of the traffic volume of the road section, it can be obtained:
(1) aspect of influence period
The section traffic of the dong-bi-ring road section from north to south, the dong-bi-ring road section from south to north and the dong-bi-ring road section 1 hour before the start of the activity gradually rises, and the influence duration is about 2.5 hours. The duration of the influence after the end of the activity is about 1 hour, and is matched with the analysis result of the average vehicle speed of the dong-dicyclo road section.
(2) Aspect of degree of influence
The traffic flow of the road sections in different directions of the Dong-dicyclo is increased approximately. In the activity starting stage, the influence degree of large-scale activities on the east two rings is greater than that of the east three rings, and the activity ending stage shows the opposite law, namely the east three rings are influenced by the activities and the traffic flow is increased more than that of the east two rings. Overall consideration shows that the influence degree of the large-scale activity starting stage on the road section is larger than that of the large-scale activity ending stage.
Step 3, analyzing influence factors of the average speed of the road section:
the establishment selection factor set of the random forest model comprises the following steps: different roads, date attributes, time periods, weeks, months, nature of the event, size of the event, weather conditions, distance to the start of the event, end time, and restrictions.
Step 4, quantifying and optimizing model parameters:
(1) quantification of parameters
In the prediction of speed of the road sections around the Beijing worker gym, the parameters need to be standardized, see Table 3, and see Table 7 according to the standardized processing rules of the parameters and the final processed results.
TABLE 7 parameter normalization processing reference table
(2) Model parameter optimization
In the present invention, the r2_ score evaluation index is used to perform the optimal scoring, and finally it is determined to use { n _ estimators ═ 10} as the optimal number of iterations of the present model, { max _ features ═ 13} as the optimal feature number, { min _ samples _ split ═ 2} as the optimal feature number, and { max _ depth ═ 20} as the optimal number of iterations.
And 5, analyzing a prediction result:
the average speed data of vehicles of peripheral road sections and short-time large-scale activity data (sports games) in the period from 2017 to 2018 of the Beijing worker stadium are constructed into a data set, see a table 8, the constructed model is used for predicting the average speed data of the road sections in the period of large-scale activities of the Dongbi, the new east road, the east road of the worker stadium and the north road of the worker stadium, and the actual data is used for perusal, inspection and evaluation.
TABLE 8 test set Activities
(1) Dongxi mean vehicle speed predicted value precision analysis
As shown in table 9, the predicted value of the average vehicle speed of the east-second-ring link is substantially consistent with the trend of the true value, the average accuracy is 94.09%, and the prediction accuracy is high.
TABLE 9 DOWNHOMORE average velocity accuracy
(2) Precision analysis of new east road average vehicle speed predicted value
As shown in table 10, the predicted value of the average vehicle speed of the new east road segment is relatively consistent with the trend of the true value, the average accuracy is 88.90%, the prediction accuracy in the activity starting stage is relatively low, the lowest value is 70.15%, the overall prediction accuracy is over 80%, and the accuracy is relatively high.
TABLE 10 New east road average speed accuracy
(3) Precision analysis of predicted value of average speed of north road of Beijing worker stadium
As shown in table 11, the predicted value of the average speed of the north road of the stadium of beijing workers is substantially consistent with the trend of the true value, the average accuracy is 78.11%, and the prediction accuracy is relatively low due to the fact that the speed is low and the true value is too small in part of time intervals.
TABLE 11 worker stadium north road average speed accuracy
(4) Precision analysis of predicted value of mean vehicle speed of east road of Beijing worker stadium
As shown in Table 12, the predicted value of the average speed of the Beijing worker stadium east road is basically consistent with the trend of the true value, the average precision is 84.67%, and the prediction precision is high.
TABLE 12 worker stadium east road average speed accuracy
Through comparative analysis, the predicted values and the actual values of the average vehicle speeds of the affected road sections of the east-two ring, the new east road, the north road and the east road of the worker stadium are predicted, the average accuracy is 94.09%, 88.90%, 78.11% and 86.44%, the total prediction accuracy is high, and the method is shown in table 13. The prediction precision of the dong-dicyclo and the new east road is higher than that of the north road and the east road of the worker stadium.
TABLE 13 prediction accuracy summary
The invention provides a short-time large-scale activity period road section average vehicle speed prediction system, which comprises:
and the acquisition module is used for acquiring the original data of the floating car, the original data of the detector and the large-scale activity data.
The influence characteristic information determining module is used for analyzing the running speed and the section traffic volume of the peripheral road section of the large-scale activity data and determining influence characteristic information; the influence characteristic information includes an influence period, an influence degree and a speed change. The influence characteristic information determining module specifically includes: the peripheral road section operation speed analysis unit is used for carrying out peripheral road section operation speed analysis on the large-scale activity data and determining the peripheral road section influence time interval in the influence time interval, the peripheral road section influence degree in the influence degree and the speed change; and the section traffic volume analysis unit is used for carrying out section traffic volume analysis on the large-scale activity data and determining the section influence time interval in the influence time interval and the section influence degree in the influence degree.
And the influence factor determining module is used for determining influence factors according to the large-scale activity data and the influence characteristic information.
The average speed prediction result determining module is used for obtaining an average speed prediction result by utilizing a short-time large-scale activity period road section average speed prediction model according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm. The construction process of the road section average speed prediction model in the short-time large-scale activity period specifically comprises the following steps: adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model; constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers; and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
In practical application, the system for predicting the average vehicle speed of the road section during the short-time large-scale activity further comprises: and the removing and screening module is used for removing and screening the original data of the floating car and the original data of the detector.
The invention analyzes and researches the traffic running state under the influence of short-time large-scale activities from two angles of time and space, searches the potential law of the traffic running state, realizes the prediction of the average speed of surrounding road sections when a Beijing worker stadium holds the activities by establishing a random forest model, and provides the strategy and the suggestion of the running organization of the surrounding road network aiming at the defects of the surrounding road control under the influence of the short-time large-scale activities. The advantages of the invention are as follows: the Python data analysis and data mining technology is used as a support, preprocessing such as extraction, elimination and screening is carried out on data analysis indexes, data quality is improved, and data acquisition cost is reduced. The method is characterized in that a random forest-based prediction model of the average speed of the road section in the short-time large-scale activity period is constructed, the random forest can process high-dimensional data, and the method has the characteristics of strong generalization capability, high training speed and the like. Therefore, the method has the advantages of high prediction speed and high precision. The floating car data with the granularity of 5min time interval can well reflect the running state of the vehicles on the road, and has better stability; the detector number of the detector data is used for matching the road section and the direction to be analyzed, the time is used for determining the time interval, and the traffic flow data is used for analyzing the state of the road section; and the relatively comprehensive and complete large-scale activity data including the week of the large-scale activity, the holding date, the weather, the name of the activity, the type of the activity, the holding venue, the number of the participants of the activity and the starting and ending time is added, and a solid foundation is laid for the subsequent analysis by relying on the database analysis and data mining technology. Therefore, the method provided by the invention can be used for predicting the vehicle speed of the road around the influence of large-scale activities.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. A method for predicting average vehicle speed of a road section during a short-time large-scale activity is characterized by comprising the following steps:
acquiring original data of a floating car, original data of a detector and large-scale activity data;
analyzing the running speed and section traffic volume of the peripheral road section of the large-scale activity data to determine influence characteristic information; the influence characteristic information comprises an influence time period, an influence degree and a speed change;
determining influence factors according to the large-scale activity data and the influence characteristic information;
obtaining an average speed prediction result by utilizing a road section average speed prediction model in a short-time large-scale activity period according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm.
2. The method for predicting the average vehicle speed of the road section during the short-time large-scale activity according to claim 1, wherein after the acquiring of the raw floating car data, the raw detector data and the large-scale activity data, the method further comprises:
and removing and screening the floating car original data and the detector original data.
3. The method for predicting the average vehicle speed of the road section during the short-time large-scale activity according to claim 1, wherein the analyzing the operation speed and the section traffic volume of the peripheral road section on the large-scale activity data to determine the influence characteristic information specifically comprises:
analyzing the running speed of the peripheral road sections of the large-scale activity data, and determining the peripheral road section influence time period in the influence time period, the peripheral road section influence degree in the influence degree and the speed change;
and analyzing the cross section traffic volume of the large-scale activity data, and determining the cross section influence time interval in the influence time interval and the cross section influence degree in the influence degree.
4. The method for predicting the average vehicle speed of the road section during the short-time large-scale activity according to claim 1, wherein the construction process of the prediction model of the average vehicle speed of the road section during the short-time large-scale activity specifically comprises the following steps:
adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model;
constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers;
and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
5. A system for predicting average vehicle speed of a road segment during a short-term large-scale activity, comprising:
the acquisition module is used for acquiring original data of the floating car, original data of the detector and large-scale activity data;
the influence characteristic information determining module is used for analyzing the running speed and the section traffic volume of the peripheral road section of the large-scale activity data and determining influence characteristic information; the influence characteristic information comprises an influence time period, an influence degree and a speed change;
the influence factor determining module is used for determining influence factors according to the large-scale activity data and the influence characteristic information;
the average speed prediction result determining module is used for obtaining an average speed prediction result by utilizing a short-time large-scale activity period road section average speed prediction model according to the floating car original data, the detector original data and the influence factors; the prediction model of the average speed of the road section in the short-time large-scale activity period is constructed by using a random forest algorithm.
6. The short term large activity segment average vehicle speed prediction system according to claim 5, further comprising:
and the removing and screening module is used for removing and screening the original data of the floating car and the original data of the detector.
7. The system for predicting the average vehicle speed of the road section during the short-time large-scale activity according to claim 5, wherein the influence characteristic information determining module specifically comprises:
the peripheral road section operation speed analysis unit is used for carrying out peripheral road section operation speed analysis on the large-scale activity data and determining the peripheral road section influence time interval in the influence time interval, the peripheral road section influence degree in the influence degree and the speed change;
and the section traffic volume analysis unit is used for carrying out section traffic volume analysis on the large-scale activity data and determining the section influence time interval in the influence time interval and the section influence degree in the influence degree.
8. The system for predicting the average vehicle speed of the short-time large-scale active section according to claim 5, wherein the construction process of the model for predicting the average vehicle speed of the short-time large-scale active section specifically comprises:
adjusting parameters by using floating car training data, detector training data and an influence factor training set, and determining the optimal iteration times and the optimal characteristic number of the random forest model;
constructing a random forest model according to the floating car training data, the detector training data, the influence factor training set, the optimal iteration times and the optimal characteristic numbers;
and optimizing the parameters of the random forest model according to the average absolute error, the mean square error and the judgment coefficient to obtain a short-time large-scale active period road section average speed prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111246104.7A CN114186710A (en) | 2021-10-26 | 2021-10-26 | Method and system for predicting average speed of road section during short-time large-scale activity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111246104.7A CN114186710A (en) | 2021-10-26 | 2021-10-26 | Method and system for predicting average speed of road section during short-time large-scale activity |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114186710A true CN114186710A (en) | 2022-03-15 |
Family
ID=80540000
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111246104.7A Pending CN114186710A (en) | 2021-10-26 | 2021-10-26 | Method and system for predicting average speed of road section during short-time large-scale activity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114186710A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150241242A1 (en) * | 2014-02-21 | 2015-08-27 | Iteris, Inc. | Short-term travel-time prediction modeling augmented with radar-based precipitation predictions and scaling of same |
| CN107103753A (en) * | 2016-02-22 | 2017-08-29 | 财团法人资讯工业策进会 | Traffic time prediction system, traffic time prediction method, and traffic model establishment method |
| CN109086964A (en) * | 2018-07-03 | 2018-12-25 | 南京邮电大学 | MR coverage rate influence factor determination method based on random forest |
| CN111080029A (en) * | 2019-12-26 | 2020-04-28 | 山东大学 | Urban traffic road speed prediction method and system based on multi-path segment space-time correlation |
-
2021
- 2021-10-26 CN CN202111246104.7A patent/CN114186710A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150241242A1 (en) * | 2014-02-21 | 2015-08-27 | Iteris, Inc. | Short-term travel-time prediction modeling augmented with radar-based precipitation predictions and scaling of same |
| CN107103753A (en) * | 2016-02-22 | 2017-08-29 | 财团法人资讯工业策进会 | Traffic time prediction system, traffic time prediction method, and traffic model establishment method |
| CN109086964A (en) * | 2018-07-03 | 2018-12-25 | 南京邮电大学 | MR coverage rate influence factor determination method based on random forest |
| CN111080029A (en) * | 2019-12-26 | 2020-04-28 | 山东大学 | Urban traffic road speed prediction method and system based on multi-path segment space-time correlation |
Non-Patent Citations (3)
| Title |
|---|
| SHUGUAN YANG等: ""understanding and Predicting Travel Time with Spatio-Temporal Features of Network Traffic Flow, Weather and Incidents"", 《IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE》, vol. 11, no. 3, 20 June 2019 (2019-06-20), pages 12 - 28, XP011736250, DOI: 10.1109/MITS.2019.2919615 * |
| 李瑞敏等: ""基于BP神经网络与D-S证据理论的路段平均速度融合方法"", 《交通运输工程学报》, vol. 14, no. 5, 31 October 2014 (2014-10-31), pages 111 - 118 * |
| 杨游云等著: "《Python广告数据挖掘与分析实践》", vol. 1, 31 March 2021, 机械工业出版社, pages: 120 - 124 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110264709B (en) | Method for predicting traffic flow of road based on graph convolution network | |
| CN109544932B (en) | Urban road network flow estimation method based on fusion of taxi GPS data and gate data | |
| CN110836675B (en) | Decision tree-based automatic driving search decision method | |
| CN110377807B (en) | Urban group functional relation and space pattern analysis method and system | |
| CN110807919A (en) | Urban road network traffic operation situation evaluation method based on vehicle passing data | |
| CN113436433B (en) | Efficient urban traffic outlier detection method | |
| CN111696369A (en) | Whole-city road time-division vehicle type traffic flow prediction method based on multi-source geographic space big data | |
| CN109886724B (en) | Robust resident travel track identification method | |
| CN112463898B (en) | Noise map updating method combining speed and noise monitoring data | |
| CN113222385B (en) | Method for constructing and evaluating driving condition of electric automobile | |
| CN112101132B (en) | Traffic condition prediction method based on graph embedding model and metric learning | |
| CN113449780B (en) | Intra-road berth occupancy prediction method based on random forest and LSTM neural network | |
| CN115422747A (en) | Method and device for calculating discharge amount of pollutants in tail gas of motor vehicle | |
| Chen et al. | Adaptive ramp metering control for urban freeway using large-scale data | |
| CN114299742A (en) | Dynamic recognition and updating recommendation method for speed limit information of expressway | |
| Treboux et al. | A predictive data-driven model for traffic-jams forecasting in smart santader city-scale testbed | |
| Al Mahmud et al. | Impact of pedal powered vehicles on average traffic speed in dhaka city: A cross-sectional study based on road class and timestamp | |
| CN114186710A (en) | Method and system for predicting average speed of road section during short-time large-scale activity | |
| CN114139984B (en) | Urban traffic accident risk prediction method based on flow and accident cooperative sensing | |
| Husni et al. | Predicting traffic conditions using knowledge-growing Bayes classifier | |
| CN116989801A (en) | Map matching method and device for low-frequency long tracks of complex road network | |
| Shepelev et al. | Forecasting the amount of traffic-related pollutant emissions by neural networks | |
| CN115497306A (en) | Speed interval weight calculation method based on GIS data | |
| CN115830855A (en) | High speed based on two-state division road section passing time prediction method | |
| CN106529778A (en) | Bus ride comfort index construction method based on smart phone |
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 |