CN113326972B - Bus lane short-time speed prediction method based on real-time bus speed statistical data - Google Patents

Bus lane short-time speed prediction method based on real-time bus speed statistical data Download PDF

Info

Publication number
CN113326972B
CN113326972B CN202110491821.XA CN202110491821A CN113326972B CN 113326972 B CN113326972 B CN 113326972B CN 202110491821 A CN202110491821 A CN 202110491821A CN 113326972 B CN113326972 B CN 113326972B
Authority
CN
China
Prior art keywords
time
road section
bus
speed
data
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
Application number
CN202110491821.XA
Other languages
Chinese (zh)
Other versions
CN113326972A (en
Inventor
翟华伟
崔立成
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dalian Maritime University
Original Assignee
Dalian Maritime University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Dalian Maritime University filed Critical Dalian Maritime University
Priority to CN202110491821.XA priority Critical patent/CN113326972B/en
Publication of CN113326972A publication Critical patent/CN113326972A/en
Application granted granted Critical
Publication of CN113326972B publication Critical patent/CN113326972B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125—Traffic data processing
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
    • 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

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Health & Medical Sciences (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Educational Administration (AREA)
  • Software Systems (AREA)
  • Game Theory and Decision Science (AREA)
  • Biomedical Technology (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Primary Health Care (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention provides a bus lane short-time speed prediction method based on real-time bus speed statistical data. According to real-time speed statistics data of a bus section and in combination with historical speed statistics data, a space-time composite prediction model based on a gray relation analysis method, a one-dimensional convolution neural network, a self-adaptive extreme learning machine and a double-layer threshold recursion unit circulation neural network is adopted, DHSTN is short for short, and future short-time bus speed change conditions of the bus section are predicted. According to the method, the influence of the adjacent or similar bus sections on the bus speed of the target section is fully considered, the time and space dependence characteristics of real-time and historical bus speed statistical data are analyzed, and finally, the long-term and short-term dependence characteristics of the target section are analyzed by introducing the self-adaptive extreme learning machine neural network, so that the prediction precision of the short-term bus speed of the bus section can be effectively improved.

Description

Bus lane short-time speed prediction method based on real-time bus speed statistical data
Technical Field
The invention relates to the technical field of intelligent transportation, in particular to a bus lane short-time speed prediction method based on real-time bus speed statistical data.
Background
The bus speed prediction is gradually an important decision basis for optimizing bus operation scheduling, establishing an elastic departure schedule and improving the bus service level, and provides important support for reasonably arranging travel routes before traveling of passengers. Unlike traditional road traffic speed predictions, urban bus speed predictions rarely use linear methods alone for prediction, and nonlinear methods and combination methods become the main stream of bus speed predictions. The nonlinear method mainly analyzes potential nonlinear characteristics in the bus speed, can adaptively control the dynamic change of the bus speed, such as an artificial neural network model, a support vector machine model, a Bayesian network model and the like, generally requires more complex and time-consuming algorithm training, and is generally based on time-dependent characteristics of statistical data, and the influence of adjacent/similar road sections is ignored. In order to fully exert the advantages of each nonlinear model, some students combine a plurality of linear and nonlinear methods to construct a combined model, and predict short-time bus speed, such as a genetic algorithm-support vector machine combined prediction model, a K-neighbor algorithm-cyclic recurrent neural network combined prediction model and the like. The combination method fully develops the advantages of each single model in the interior, mutually compensates the defects, fully excavates the linear and nonlinear characteristics in the statistical data, has certain advantages in the aspects of prediction precision, self-adaption and the like, but has a complex construction process, and does not effectively analyze and excavate the space and time characteristics and long-term and short-term dependence characteristics of the statistical data of the bus speed.
Disclosure of Invention
According to the existing prediction method, the technical problems of influence and complex structure of adjacent/similar road sections and the like are not considered, and the short-time speed prediction method for the bus lane based on real-time bus speed statistical data is provided. According to the invention, a one-dimensional convolutional neural network, an extreme learning machine, a gray relation analysis method and a cyclic recurrent neural network are organically combined together to construct a space-time composite prediction algorithm, and the space and time characteristics and long-term and short-term dependence characteristics of the statistical data of the bus speed of the target bus road section are comprehensively analyzed and mined, so that the prediction precision of the algorithm is improved.
The invention adopts the following technical means:
a bus lane short-time speed prediction method based on real-time bus speed statistical data comprises the following steps:
s1, acquiring bus speed time series data of a target road section and adjacent road sections thereof, analyzing a correlation between the adjacent road sections and the target road section by adopting a gray scale relation analysis method based on entropy, and finally selecting m' adjacent road sections which meet a correlation constraint condition;
s2, selecting a target road sectionTime series data of speed of road traffic vehicle of d days before collected by m' adjacent road sectionsK is more than or equal to 1 and less than or equal to d, and a convolutional neural network is adopted to analyze the space characteristics between a target road section and an adjacent road section;
s3, based on time series data of speed of the front d-day public transport vehicleK is more than or equal to 1 and less than or equal to d, and a vehicle speed sampling value +.>Attention relation mechanism between the two pairs is based on the Attention relation mechanism pair ++>Correcting;
s4, respectively analyzing daily bus speed time series dataConstructing a double-layer threshold circulating unit circulating neural network, and mining long-term time dependence characteristic h of historical time sequence data t,ls ；
S5, analyzing time series data centering on the target road sectionBased on a convolution neural network and a single-layer threshold cyclic unit cyclic neural network, a combined short-time prediction model is established, and spatial characteristics and short-time characteristics of time sequence data of a target road section are excavated>
S6, analyzing long-term time dependence characteristic h of target road section t,ls And short time dependence characteristicsEffectively combining the two to obtain the time dependent characteristic of the target road section>Wherein->
S7, based on the time dependent characteristics of the target road sectionConstructing a neural network based on an adaptive extreme learning machine, fitting +.>The change of the speed of the bus at the next moment of the target road section is predicted,for the predicted value of the bus speed at the next moment, f is a function to be fitted, w f To connect weights, b f To bias, w f For the first learning parameter, b f Is the second learning parameter.
Further, step S1 includes:
s101, calculating a simple gray association degree GRG (m) of the target road section and the adjacent road section according to the following formula according to the time sequence vehicle speed data of the target road section and the adjacent road section:
wherein, gamma is a gray relation coefficient, x t (s) the average speed of the target road section at the t sampling time, x t (m) is the average speed of the adjacent road section at the T sampling time, T is the time series data length, E (m) is the mEntropy of gray relation coefficients of adjacent road sections;
s102, analyzing the magnitude of the simple gray association degree of each adjacent road section and the target road section, and sequentially sorting the adjacent road sections and the target road section according to descending order;
s103, for a given threshold epsilon, if GRG (m) is not less than epsilon, selecting the adjacent road section as the associated road section of the target road section.
Further, the convolutional neural network in step S2 is a one-dimensional convolutional neural network without a pooling layer.
Further, step S3 includes:
s301, inputting data by using extreme learning machine neural networkAnalysis is performed, setting the training error to +.>Wherein, k is more than or equal to 1 and less than or equal to d, and ∈>For the predicted value of the extreme learning machine neural network, < >>For the actual measurement value, f is an error calculation function, and the output weight of the neural network hidden layer of the extreme learning machine is obtained when the error is smaller than a given threshold value>
S302, outputting weight based on neural network hidden layer of extreme learning machine Calculating data according to the following formula>Importance measure of the elements relative to the measured value +.>
S303, based on input dataAnd its corresponding importance measure +.> Correction of elements in the input data, i.e. +.>
Further, step S4 includes:
s401, analyzing data of d days before, and analyzing data of the q th dayThe first layer threshold cycle unit analyzes the data change at time t of each day, i.e. +.>1≤q<d，/>The input of the threshold circulation unit;
s402, analyzing the data change of the previous d days output by the first layer threshold circulating unit at the time t1≤q<d, using a second layer threshold cycling unitAnalyzing the correlation between the data at the same time t of day, i.e. +.>The output of the q-th threshold cycle unit;
s403, based on the analysis results of S401 and S402, acquiring long-term time dependence characteristic h of the previous d days data t,ls 。
Further, step S5 includes:
s501, selecting bus speed time series data acquired by a target road section and m' selected adjacent road sections, analyzing the space characteristics between the target road section and the adjacent road sections by adopting a convolutional neural network, and further acquiring the target road section time series data'0' represents the current time of the target link;
s502, analyzing the time sequence data of the target road section by adopting an Attention relation mechanismRelation with data at next time, and for +.>Correcting the data;
s503, analyzing the time sequence data characteristics of the target road section by adopting a single-layer threshold circulating unit to obtain the short-time dependency characteristics of the current time of the target road section
According to the real-time speed statistical data of the bus section and by combining with the historical speed statistical data, the invention adopts a space-time composite prediction model based on a gray relation analysis method, a one-dimensional convolution neural network, a self-adaptive extreme learning machine and a double-layer threshold recursion unit circulation neural network to predict the future short-time speed change condition of the bus in the bus section. And fully considering the influence of adjacent or similar bus sections on the bus speed of a target section, combining the real-time statistical data and the historical statistical data of the bus speed, analyzing and selecting an adjacent or similar section with the highest association degree with the target bus section to form an analysis object based on an entropy gray relation analysis method, constructing real-time and historical bus speed statistical data of the same time scale, introducing a one-dimensional convolutional neural network, an Attention relation analysis mechanism and a threshold recursion cyclic neural network to construct a multi-layer composite space-time model, respectively analyzing the time and space dependence characteristics of the real-time and historical bus speed statistical data, and finally, introducing an adaptive extreme learning machine neural network to analyze the long-term and short-term dependence characteristics of the target section to predict the future bus speed change condition of the target bus section, wherein the specific flow chart is shown in figure 1. According to the method, the influence of the speeds of adjacent/similar bus sections on the target section is considered, the time and space characteristics and the long-term and short-term dependence characteristics of the bus speed of the bus section are comprehensively analyzed, and the prediction accuracy of the short-term bus speed of the bus section can be effectively improved.
Compared with the prior art, the invention has the following advantages:
1. according to the bus lane short-time speed prediction method based on the real-time bus speed statistical data, the correlation between adjacent or similar road sections is analyzed, the spatial correlation characteristics between the road sections are analyzed, and the speed prediction precision of the target road section is effectively improved by means of the speed statistical data of the adjacent road sections.
2. According to the bus lane short-time speed prediction method based on the real-time bus speed statistical data, the long-term and short-time dependence characteristics of the bus speed statistical data are comprehensively analyzed, and the bus speed prediction precision of the bus-specific road section is effectively improved.
In summary, the invention predicts the future short-time bus speed change condition of the bus road section by adopting a space-time composite prediction model based on a gray relation analysis method, a one-dimensional convolution neural network, a self-adaptive extreme learning machine and a double-layer threshold recursion unit circulation neural network according to the real-time speed statistical data of the bus road section and combining with the historical speed statistical data. And fully considering the influence of adjacent or similar bus sections on the bus speed of a target section, combining the real-time statistical data and the historical statistical data of the bus speed, analyzing and selecting an adjacent or similar section with the highest association degree with the target bus section to form an analysis object based on an entropy gray relation analysis method, constructing real-time and historical bus speed statistical data of the same time scale, introducing a one-dimensional convolutional neural network, an Attention relation analysis mechanism and a threshold recursion cyclic neural network to construct a multi-layer composite space-time model, respectively analyzing the time and space dependence characteristics of the real-time and historical bus speed statistical data, and finally, introducing an adaptive extreme learning machine neural network to analyze the long-term and short-term dependence characteristics of the target section to predict the future bus speed change condition of the target bus section, wherein the specific flow chart is shown in figure 1.
Based on the reasons, the intelligent bus management system can be widely popularized in the fields of intelligent traffic, intelligent bus management and the like.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to the drawings without inventive effort to a person skilled in the art.
FIG. 1 is a flowchart of the method for predicting short-term speed of a bus lane based on real-time bus speed statistics.
Fig. 2 is a basic flow chart for predicting short-term speed of a bus lane by applying the method of the invention.
FIG. 3 is a graph showing the predicted speed of a bus at a bus section from 7:30 to 9:30 on a weekday at 12 th week of 2018.
FIG. 4 is a graph showing the predicted speed of a bus at a bus segment from 16:30 to 18:30 on a non-working day at week 12 of 2018.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
As shown in fig. 1, the invention provides a bus lane short-time speed prediction method based on real-time bus speed statistical data, which comprises the following steps:
s1, acquiring bus speed time series data of a target road section and adjacent road sections thereof, analyzing a correlation between the adjacent road sections and the target road section by adopting an Entropy-based gray scale relation analysis method (Entropy-based Grey Relation Analysis, EGRA), and finally selecting m' adjacent road sections which meet a correlation constraint condition. The method specifically comprises the following steps:
s101, calculating a simple gray association degree (Grey Relevancy Grade) GRG (m) of the target road section and the adjacent road section according to the following formula according to the time sequence vehicle speed data of the target road section and the adjacent road section:
wherein, gamma is a gray relation coefficient, x t (s) the average speed of the target road section at the t sampling time, x t (m) is the average speed of the T sampling time of the adjacent road section, T is the length of time series data, E (m) is the entropy of gray relation coefficient of the m adjacent road section;
s102, analyzing the magnitude of the simple gray association degree of each adjacent road section and the target road section, and sequentially sorting the adjacent road sections and the target road section according to descending order;
s103, for a given threshold epsilon, if GRG (m) is not less than epsilon, selecting the adjacent road section as the associated road section of the target road section. The purpose of step S102 is to quickly complete the selection of the associated road segment in step S103 without spending too much time.
S2, selecting time series data of the speed of the front d-day bus collected by the target road section and the selected m' adjacent road sectionsAnd k is more than or equal to 1 and less than or equal to d, and a convolutional neural network (Convolutional Neural Network, CNN) is adopted to analyze the spatial characteristics between the target road section and the adjacent road sections. The convolutional neural network is a one-dimensional convolutional neural network without a pooling layer.
S3, based on time series data of speed of the front d-day public transport vehicleK is more than or equal to 1 and less than or equal to d, and vehicle speed time series data are establishedThe vehicle speed sampling value of the target road section at time t +.>Attention relation mechanism between the two pairs is based on the Attention relation mechanism pair ++>And (5) performing correction. The method specifically comprises the following steps:
s301, inputting data by using extreme learning machine neural network (Extreme Learning Machine, ELM)Analysis is performed, setting the training error to +.>Wherein (1)>For the predicted value of the extreme learning machine neural network, < >>For the actual measurement value, f is an error calculation function, and the output weight of the neural network hidden layer of the extreme learning machine is obtained when the error is smaller than a given threshold value>
S302, outputting weight based on neural network hidden layer of extreme learning machine Calculating data according to the following formula>Importance measure of the elements relative to the measured value +.>
S303, based on input dataAnd its corresponding importance measure +.>Correction of elements in the input data, i.e. +.>
S4, respectively analyzing daily bus speed time series dataConstructing a double-layer threshold cyclic unit (Gated Recurrent Unit, GRU) cyclic neural network, and mining long-term time dependence characteristic h of historical time series data t,ls . The method specifically comprises the following steps:
s401, analyzing data of d days before, and analyzing data of the q th dayThe first layer threshold cycle unit analyzes the data change at time t of each day, i.e. +.>1≤q<，/>The input of the threshold circulation unit;
s402, analyzing the data change of the previous d days output by the first layer threshold circulating unit at the time t1≤q<d, analyzing the correlation between the data at the same time t every day by using a second layer threshold cycle unit, namely +.>The output of the q-th threshold cycle unit;
s403, based on the analysis results of S401 and S402, acquiring long-term time dependence characteristic h of the previous d days data t,ls 。
S5, analyzing the time sequence number centering on the target road sectionAccording toBased on a Convolutional Neural Network (CNN) and a single-layer threshold cyclic unit cyclic neural network (GRU), a combined short-time prediction model is established, and spatial characteristics and short-time characteristics of time sequence data of a target road section are mined +.>The method specifically comprises the following steps:
s501, selecting bus speed time series data acquired by a target road section and m' selected adjacent road sections, analyzing the space characteristics between the target road section and the adjacent road sections by adopting a convolutional neural network, and further acquiring the target road section time series data'0' represents the current time of the target link;
s502, analyzing the time sequence data of the target road section by adopting an Attention relation mechanismRelation with data at next time, and for +.>Correcting the data;
s503, analyzing the time sequence data characteristics of the target road section by adopting a single-layer threshold circulating unit to obtain the short-time dependency characteristics of the current time of the target road section
S6, analyzing long-term time dependence characteristic h of target road section tls And short time dependence characteristicsEffectively combining the two to obtain the time dependent characteristic of the target road section>Wherein->
S7, based on the time dependent characteristics of the target road sectionConstruction of an adaptive extreme learning machine based neural network (Extreme Learning Machine, ELM), fitting ∈>Changing, predicting the changing condition of the speed of the bus at the next moment of the target road section, < ->For the predicted value of the bus speed at the next moment, f is a function to be fitted, w f To connect weights, b f To bias, w f For the first learning parameter, b f Is the second learning parameter.
The following describes the solution and effects of the present invention by a specific application example.
Fig. 2 is a basic flow chart for predicting short-term speed of a bus lane by applying the method of the invention. In this embodiment, selecting a bus speed in a plurality of bus route sections within a period from 5:30-22:40 of 11:5 to 16:5:12:40 in Dalian city is an example of the present invention. The bus speed data are aggregated at intervals of 5 minutes, 206 sample data can be obtained per day for each bus route section, and 8652 data samples are obtained in the whole time period. And (3) predicting the bus speed of the target section of the urban bus line by utilizing the steps 1-7. Firstly, constructing time sequence data for each bus line section by taking 5 minutes as a scale, analyzing the spatial relationship among the bus line sections by adopting EGRA, selecting the first 5 adjacent bus line sections with the highest GRGs, and reconstructing an input data matrix of the DHSTN model; based on bus route and data characteristics, calculating time and prediction precision for a balance algorithm, determining that the number of one-dimensional CNN convolution layers is 3, each layer has 35 feature graphs, and the depth of each layer of filter is 2, 3 and 4 in sequence; determining the hidden layer neuron number of the neural network of the extreme learning machine as 30 based on principal component analysis and experience accumulation, and selecting a threshold epsilon=10 (MSE error) of the neural network of the extreme learning machine; based on trial and error and experience accumulation, a double-layer threshold recursion unit is selected, the hidden layer dimension of the cyclic neural network is 14, the input layer dimension is 14, and a single output layer is formed. And finally, accurately predicting the speed of the bus at the target road section by utilizing the neural network of the extreme learning machine.
Fig. 3 and 4 show the predicted results of bus speeds for a target bus segment at 7:30-9:30 on a weekday early peak and 16:30-18:30 on a non-weekday late peak for the second week of 12 months 2018, compared to the predicted results of ARIMA (2, 1, 2), MLP, RBFNN, TM-CNN, FDL, and DNN-BTF methods.
To better illustrate the advantages of the present method over prediction accuracy, the prediction results were evaluated using mean absolute error (Mean Absolute Error, MAE), mean absolute percent error (Mean Absolute Percentage Error, MAPE) and root mean square error (Root Mean Square Error, RMSE) and compared to the prediction results of ARIMA (2, 1, 2), MLP, RBFNN, TM-CNN, FDL and DNN-BTF methods (see tables 1, 2).
TABLE 1 MAE, MAPE and RMSE error comparison (workday)
TABLE 2 MAE, MAPE and RMSE error comparison (non-workday)
The comparison result shows that the method provided by the invention is superior to the traditional prediction method in prediction precision.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.

Claims (6)

1. A bus lane short-time speed prediction method based on real-time bus speed statistical data is characterized by comprising the following steps:
s1, acquiring bus speed time series data of a target road section and adjacent road sections thereof, analyzing a correlation between the adjacent road sections and the target road section by adopting a gray scale relation analysis method based on entropy, and finally selecting m' adjacent road sections which meet a correlation constraint condition;
s2, selecting time series data of the speed of the front d-day bus collected by the target road section and the selected m' adjacent road sectionsAnalyzing the spatial characteristics between the target road section and the adjacent road section by adopting a convolutional neural network;
s3, based on time series data of speed of the front d-day public transport vehicleBuild vehicle speed timeThe sequence data and the vehicle speed sampling value +.>Attention relation mechanism between the two pairs is based on the Attention relation mechanism pair ++>Correcting;
s4, respectively analyzing daily bus speed time series dataConstructing a double-layer threshold circulating unit circulating neural network, and mining long-term time dependence characteristic h of historical time sequence data t,ls ；
S5, analyzing time series data centering on the target road sectionBased on a convolution neural network and a single-layer threshold cyclic unit cyclic neural network, a combined short-time prediction model is established, and spatial characteristics and short-time characteristics of time sequence data of a target road section are excavated>
S6, analyzing long-term time dependence characteristic h of target road section t,ls And short time dependence characteristicsEffectively combining the two to obtain the time dependent characteristic of the target road section>Wherein->
S7, based on the time dependent characteristics of the target road sectionConstructing a neural network based on an adaptive extreme learning machine, fitting +.>Changing, predicting the changing condition of the speed of the bus at the next moment of the target road section, < ->For the predicted value of the bus speed at the next moment, f is a function to be fitted, w f To connect weights, b f To bias, w f For the first learning parameter, b f Is the second learning parameter.
2. The bus lane short time speed prediction method according to claim 1, wherein step S1 comprises:
s101, calculating a simple gray association degree GRG (m) of the target road section and the adjacent road section according to the following formula according to the time sequence vehicle speed data of the target road section and the adjacent road section:
wherein, gamma is a gray relation coefficient, x t (s) the average speed of the target road section at the t sampling time, x t (m) is the average speed of the T sampling time of the adjacent road section, T is the length of time series data, E (m) is the entropy of gray relation coefficient of the m adjacent road section;
s102, analyzing the magnitude of the simple gray association degree of each adjacent road section and the target road section, and sequentially sorting the adjacent road sections and the target road section according to descending order;
s103, for a given threshold epsilon, if GRG (m) is not less than epsilon, selecting the adjacent road section as the associated road section of the target road section.
3. The bus lane short-time speed prediction method according to claim 1, wherein the convolutional neural network in step S2 is a one-dimensional convolutional neural network without a pooling layer.
4. The bus lane short time speed prediction method according to claim 1, wherein step S3 comprises:
s301, inputting data by using extreme learning machine neural networkAnalysis is performed, setting the training error to +.>Wherein, k is more than or equal to 1 and less than or equal to d, and ∈>For the predicted value of the extreme learning machine neural network, < >>For the actual measurement value, f is an error calculation function, and the output weight of the neural network hidden layer of the extreme learning machine is obtained when the error is smaller than a given threshold value>
S302, outputting weight based on neural network hidden layer of extreme learning machine Calculating data according to the following formula>Importance measure of the elements relative to the measured value +.>
S303, based on input dataAnd its corresponding importance measure +.> Correction of elements in the input data, i.e. +.>
5. The bus lane short time speed prediction method according to claim 1, wherein step S4 comprises:
s401, analyzing data of d days before, and analyzing data of the q th day The first layer threshold cycle unit analyzes the data change at time t of each day, i.e. +.>1≤q<d，/>The input of the threshold circulation unit;
s402, analyzing the data change of the previous d days output by the first layer threshold circulating unit at the time t1≤q<d, analyzing the correlation between the data at the same time t every day by using a second layer threshold cycle unit, namely +.> The output of the q-th threshold cycle unit;
s403, based on the analysis results of S401 and S402, acquiring long-term time dependence characteristic h of the previous d days data t,ls 。
6. The bus lane short time speed prediction method according to claim 1, wherein step S5 comprises:
s501, selecting bus speed time series data acquired by a target road section and m' selected adjacent road sections, analyzing the space characteristics between the target road section and the adjacent road sections by adopting a convolutional neural network, and further acquiring the target road section time series data'0' represents the current time of the target link;
s502, analyzing the time sequence data of the target road section by adopting an Attention relation mechanismRelation with data at next time, and for +.>Correcting the data;
s503, analyzing the time sequence data characteristics of the target road section by adopting a single-layer threshold circulating unit to obtain the short-time dependency characteristics of the current time of the target road section
CN202110491821.XA 2021-05-06 2021-05-06 Bus lane short-time speed prediction method based on real-time bus speed statistical data Active CN113326972B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110491821.XA CN113326972B (en) 2021-05-06 2021-05-06 Bus lane short-time speed prediction method based on real-time bus speed statistical data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110491821.XA CN113326972B (en) 2021-05-06 2021-05-06 Bus lane short-time speed prediction method based on real-time bus speed statistical data

Publications (2)

Publication Number Publication Date
CN113326972A CN113326972A (en) 2021-08-31
CN113326972B true CN113326972B (en) 2024-01-05

Family

ID=77414235

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110491821.XA Active CN113326972B (en) 2021-05-06 2021-05-06 Bus lane short-time speed prediction method based on real-time bus speed statistical data

Country Status (1)

Country Link
CN (1) CN113326972B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114358416A (en) * 2021-12-31 2022-04-15 广东工业大学 Public transport road network partitioning method, system, equipment and medium based on multi-source traffic data

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111540199A (en) * 2020-04-21 2020-08-14 浙江省交通规划设计研究院有限公司 High-speed traffic flow prediction method based on multi-mode fusion and graph attention machine mechanism
CN111554118A (en) * 2020-04-24 2020-08-18 深圳职业技术学院 Dynamic prediction method and system for bus arrival time
CN111598325A (en) * 2020-05-11 2020-08-28 浙江工业大学 Traffic speed prediction method based on hierarchical clustering and hierarchical attention mechanism
CN111613054A (en) * 2020-05-07 2020-09-01 浙江大学 Multi-step traffic speed prediction method cooperatively considering space-time correlation and contribution difference

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111540199A (en) * 2020-04-21 2020-08-14 浙江省交通规划设计研究院有限公司 High-speed traffic flow prediction method based on multi-mode fusion and graph attention machine mechanism
CN111554118A (en) * 2020-04-24 2020-08-18 深圳职业技术学院 Dynamic prediction method and system for bus arrival time
CN111613054A (en) * 2020-05-07 2020-09-01 浙江大学 Multi-step traffic speed prediction method cooperatively considering space-time correlation and contribution difference
CN111598325A (en) * 2020-05-11 2020-08-28 浙江工业大学 Traffic speed prediction method based on hierarchical clustering and hierarchical attention mechanism

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于状态空间神经网络的短期公交调度模型;高瑾;邓卫;季彦婕;;交通运输工程与信息学报(03);全文 *
基于非参数回归-粒子滤波模型的公交到站时间预测;张孝梅;陈旭梅;张溪;;公路交通科技(04);全文 *

Also Published As

Publication number Publication date
CN113326972A (en) 2021-08-31

Similar Documents

Publication Publication Date Title
CN108197739B (en) Urban rail transit passenger flow prediction method
CN109285346B (en) Urban road network traffic state prediction method based on key road sections
CN110570651B (en) Road network traffic situation prediction method and system based on deep learning
CN110766942B (en) Traffic network congestion prediction method based on convolution long-term and short-term memory network
Stathopoulos et al. Fuzzy modeling approach for combined forecasting of urban traffic flow
CN111696355A (en) Dynamic graph convolution traffic speed prediction method
CN112289034A (en) Deep neural network robust traffic prediction method based on multi-mode space-time data
CN112053560B (en) Short-time traffic flow prediction method, system and storage medium based on neural network
CN102469103B (en) Trojan event prediction method based on BP (Back Propagation) neural network
CN111063194A (en) Traffic flow prediction method
CN114299723B (en) Traffic flow prediction method
Chen et al. A self-adaptive Armijo stepsize strategy with application to traffic assignment models and algorithms
CN113762595B (en) Traffic time prediction model training method, traffic time prediction method and equipment
Vargas et al. Automobile spare-parts forecasting: A comparative study of time series methods
CN112766597A (en) Bus passenger flow prediction method and system
CN113326972B (en) Bus lane short-time speed prediction method based on real-time bus speed statistical data
CN113674524A (en) LSTM-GASVR-based multi-scale short-time traffic flow prediction modeling and prediction method and system
CN113449905A (en) Traffic jam early warning method based on gated cyclic unit neural network
Yu et al. Hierarchical Bayesian nonparametric approach to modeling and learning the wisdom of crowds of urban traffic route planning agents
CN116311921A (en) Traffic speed prediction method based on multi-spatial scale space-time converter
CN113469425A (en) Deep traffic jam prediction method
CN116933946A (en) Rail transit OD passenger flow prediction method and system based on passenger flow destination structure
CN110807508A (en) Bus peak load prediction method considering complex meteorological influence
CN111985731B (en) Method and system for predicting number of people at urban public transport station
Xiangdong et al. Prediction of short-term available parking space using LSTM model

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