CN114819366A - OD passenger flow short-time prediction method, device, equipment and storage medium thereof - Google Patents

OD passenger flow short-time prediction method, device, equipment and storage medium thereof Download PDF

Info

Publication number
CN114819366A
CN114819366A CN202210483804.6A CN202210483804A CN114819366A CN 114819366 A CN114819366 A CN 114819366A CN 202210483804 A CN202210483804 A CN 202210483804A CN 114819366 A CN114819366 A CN 114819366A
Authority
CN
China
Prior art keywords
data
time
generate
passenger flow
historical
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
Application number
CN202210483804.6A
Other languages
Chinese (zh)
Inventor
王成
李心怡
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.)
Huaqiao University
Original Assignee
Huaqiao 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 Huaqiao University filed Critical Huaqiao University
Priority to CN202210483804.6A priority Critical patent/CN114819366A/en
Publication of CN114819366A publication Critical patent/CN114819366A/en
Pending legal-status Critical Current

Links

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
    • 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
    • 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
    • G06N3/08—Learning methods
    • 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
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
    • 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

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

Abstract

The invention provides a short-time OD passenger flow prediction method, a device, equipment and a storage medium thereof, wherein the method comprises the steps of obtaining a historical passenger IC data set and external influence factor characteristic data, and carrying out OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set; converting the historical OD data set to generate a time sequence, stacking the time sequence to generate an input data set, dividing the time sequence in a preset time period by taking delta t as a prediction time interval, and stacking the divided data; calling a trained neural network model to process an input data set to generate OD passenger flow prediction data; and adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data. In addition, the OD prediction method in the prior art neglects the close relationship between OD distribution, historical passenger flow and external factors while considering the time sequence characteristics on the basis of solving the problems of data sparsity and availability, so that the result accuracy is low.

Description

OD passenger flow short-time prediction method, device, equipment and storage medium thereof
Technical Field
The invention relates to the technical field of deep learning neural networks and passenger flow prediction, in particular to an OD passenger flow short-time prediction method, device and equipment and a storage medium thereof.
Background
The OD (Origin-Destination, OD) passenger flow distribution is the travel volume from a certain starting point to a terminal point in a period of time, and reflects the travel demand and rule of passengers; in order to effectively improve the attraction and the operation organization level of urban public transport, for operators, the OD prediction result is used for carrying out real-time dynamic scheduling on vehicles, and the method is favorable for improving the travel experience of passengers. At present, researches in the short-time OD prediction field are less, most of the public transport short-time OD prediction methods in the market are time series methods and k nearest neighbor methods, although the methods are simple in operation, the prediction accuracy is low, and the statistical models suitable for processing the time series have the common defects that the time-lag variables are assumed to have linear relations, the internal structure cannot be represented by a large and common function like a nonlinear model, and the dynamic and nonlinear characteristics in the traffic field are difficult to solve.
In order to solve the problems of short-time OD prediction methods in the current market, the application of a deep learning method to OD prediction is widely developed in recent years, but the short-time OD prediction method applying the deep learning method in the prior art is not comprehensive in consideration of the problems that passenger flow distribution influence factors and OD prediction have unique characteristics, ignores the close relation between OD distribution and historical passenger flow and external factors such as weather conditions and the like on the basis of solving the problems of sparsity and usability of OD data, and causes low accuracy of OD prediction results.
In view of this, the present application is presented.
Disclosure of Invention
The invention discloses an OD passenger flow short-time prediction method, device, equipment and a storage medium thereof, which can effectively solve the problem that the OD prediction result is not high in accuracy due to close relations between OD distribution, historical passenger flow and external factors such as weather conditions and the like on the basis of neglecting the problems of sparsity and usability of OD data and considering the time sequence characteristics of the passenger flow on the basis of applying a deep learning method to the OD prediction method in the prior art.
The invention discloses a short-time OD passenger flow prediction method, which comprises the following steps:
acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set;
converting the historical OD data set to generate a time sequence, stacking the time sequence to generate an input data set, wherein the time sequence in a preset time period is divided by using delta t as a prediction time interval, and the divided data are stacked;
calling a trained neural network model to process the input data set to generate OD passenger flow prediction data;
and adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
Preferably, the historical OD dataset is subjected to conversion processing to generate a time series, and the time series is subjected to stacking processing to generate an input dataset, specifically:
converting the historical OD data set to generate a plurality of groups of time sequences of preset time periods;
dividing the time sequence of each preset time period by using delta t as a prediction time interval, and stacking the divided time sequences of each preset time period respectively to generate an input data set, wherein the input data set comprises proximity data, trend data, periodic data, real-time inbound passenger flow data and external factor data.
Preferably according to a formula
Figure BDA0003628751140000031
Calculating the proximity data; according to the formula
Figure BDA0003628751140000032
Calculating the trend data; according to the formula
Figure BDA0003628751140000033
Calculating the periodic data; according to the formula
Figure BDA0003628751140000034
Calculating the real-time inbound passenger flow data; according to the formula
Figure BDA0003628751140000035
Calculating the external factor data; wherein l m Characteristic length of external factors,/ e For the length of the proximity sequence,/ tr To trend sequence length, l pe Is the periodic sequence length, p trending time span, q periodic time span, M t Is a feature vector of the external factors at the prediction time interval t.
Preferably, before invoking the trained neural network model to process the input data set, the method further includes:
acquiring a plurality of groups of historical input data sets;
processing the historical input data set to generate a plurality of groups of data result values;
dividing the data result value into a training set and a test set according to a preset proportion;
and calling the training set as input data of the neural network, training the neural network, calling the test set to test the trained neural network, and generating a neural network model.
Preferably, the historical input data set is subjected to data processing to generate a plurality of sets of data result values, specifically:
according to the formula
Figure BDA0003628751140000041
Carrying out Min-Max standardization processing on the proximity data, trend data, periodic data and real-time inbound passenger flow data in the historical input data set to generate and map to [0,1]Data result values between ranges, where X max Maximum value of single-column characteristic data, X min Is the minimum value of the single-column characteristic data;
carrying out One-Hot coding digitization processing on discrete external factor data in the historical input data set to generate a digitized data result value;
and carrying out Min-Max linear normalization processing on the continuous external factor data in the historical input data set to generate a data result value which is scaled to the range of [0,1 ].
Preferably, the training set is called as input data of the neural network, the neural network is trained, the test set is called to test the trained neural network, and a neural network model is generated, specifically:
establishing a neural network, and inputting the training set into the neural network for training as input data;
inputting the test set into a neural network, and calculating the Root Mean Square Error (RMSE) and the average absolute error (MAE) of the test set to serve as judgment standards;
and when the root mean square error RMSE and the average absolute error MAE reach preset values, storing the architecture and network parameters of the neural network, and generating a neural network model.
Preferably according to a formula
Figure BDA0003628751140000051
Calculating the Root Mean Square Error (RMSE) according to the formula
Figure BDA0003628751140000052
Calculating the mean absolute error MAE, wherein: m is ij And m ij1 The traffic flow is divided into an actual value and a predicted value from the station i to the station j.
The invention discloses an OD passenger flow short-time prediction device, which comprises:
the data acquisition unit is used for acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set;
an input parameter obtaining unit, configured to perform conversion processing on the historical OD data set to generate a time sequence, and perform stacking processing on the time sequence to generate an input data set, where Δ t is used as a prediction time interval to divide the time sequence within a preset time period, and stack divided data;
the prediction data generation unit is used for calling the trained neural network model to process the input data set and generate OD passenger flow prediction data;
and the adjusting unit is used for adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
The invention discloses an OD passenger flow short-time prediction device which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein the processor realizes the OD passenger flow short-time prediction method when executing the computer program.
The invention discloses a readable storage medium, which is characterized by storing a computer program, wherein the computer program can be executed by a processor of a device where the storage medium is located, so as to realize the OD passenger flow short-time prediction method.
In summary, according to the method, the apparatus, the device and the storage medium for short-term prediction of OD passenger flow provided by this embodiment, first, according to the characteristic that the passenger flow has a time sequence, and according to the characteristics of spatial and temporal distribution, the time attribute of the OD data set is divided into three components, namely, time proximity, periodicity and trend; the ConvLSTM network is used for respectively modeling the three attributes, the close relation between OD distribution and external factors such as weather conditions and the like is considered, the consideration factors are comprehensive, the generalization capability of the model is improved, a prediction result with better precision can be obtained, and the real-time dynamic scheduling of the vehicle is adjusted according to the final prediction result. Therefore, the problem that the OD prediction result is not high in accuracy due to the fact that the close relation between OD distribution and historical passenger flow and travel external factors such as weather conditions and the like is considered on the basis of solving the problems of sparsity and usability of OD data and neglecting the problem that OD data are sparse in distribution and usability is solved by the aid of the short-time OD prediction method applied to the bus rapid transit in the prior art is solved.
Drawings
Fig. 1 is a schematic flow chart of a short-term OD passenger flow prediction method according to a first aspect of the present invention.
Fig. 2 is a schematic flow chart of a short-term OD passenger flow prediction method according to a second aspect of the present invention.
Fig. 3 is a schematic overall structure diagram of a neural network model provided in an embodiment of the present invention.
Fig. 4 is a schematic diagram illustrating the influence of the passenger flow timing characteristics according to the embodiment of the present invention.
Fig. 5 is a block diagram of an OD traffic short-term prediction device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings of the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
The following detailed description of specific embodiments of the invention refers to the accompanying drawings.
Referring to fig. 1, a first embodiment of the invention provides a method for OD traffic short-term prediction, which includes:
s101, acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set.
And S102, converting the historical OD data set to generate a time sequence, stacking the time sequence to generate an input data set, wherein the time sequence in a preset time period is divided by using delta t as a prediction time interval, and the divided data are stacked.
Specifically, step S102 includes:
converting the historical OD data set to generate a plurality of groups of time sequences of preset time periods;
dividing the time sequence of each preset time period by using delta t as a prediction time interval, and stacking the divided time sequences of each preset time period respectively to generate an input data set, wherein the input data set comprises proximity data, trend data, periodic data, real-time inbound passenger flow data and external factor data.
According to the formula
Figure BDA0003628751140000091
Calculating the proximity data; according to the formula
Figure BDA0003628751140000092
Calculating the trend data; according to the formula
Figure BDA0003628751140000093
Calculating the periodic data; according to the formula
Figure BDA0003628751140000094
Calculating the real-time inbound passenger flow data; according to the formula
Figure BDA0003628751140000095
Calculating the external factor data; wherein l m Characteristic length of external factors,/ e Is the length of the proximity sequence, l tr To trend sequence length, l pe Is the periodic sequence length, p trending time span, q periodic time span, M t Is a feature vector of the external factors at the prediction time interval t.
In recent years, short-term network traffic prediction has been extensively studied, and various data-driven and model-based prediction methods have been used to predict speed/traffic, boarding/alighting needs, entering/exiting needs, OD needs of traffic. Among them, unlike other short-term traffic forecasts, OD demand short-term forecasts have three unique features, as shown in table 1:
table 1:
Figure BDA0003628751140000096
Figure BDA0003628751140000101
short-term prediction of OD demand requires input of a more careful design model and takes sparsity into account compared to other prediction tasks. However, in the prior art, the short-time OD prediction method of the bus rapid transit, which applies the deep learning method, has the problem that the OD prediction result is not high in accuracy due to the close relationship between OD distribution, historical passenger flow and external factors such as weather conditions and the like, and the close relationship between the OD distribution and the historical passenger flow and the external factors such as the weather conditions and the like are considered on the basis of solving the problems of sparsity and usability of OD data.
Specifically, in the present embodiment, first, the characteristic data of the external influence factor (such as weather conditions) and the historical passenger IC card swiping data are acquired and OD matching is performed to obtain the required OD data. Secondly, OD data are converted into a time sequence, the current time is assumed to be t, delta t is taken as a prediction time interval, and the OD required by the bus rapid transit OD at the next moment is predicted t+1 Acquiring data with delta t as time interval for each component, respectively setting corresponding time sequence length, splicing the time sequence length along a time dimension to construct OD c 、OD pe 、OD tr 、INFLOW、M t Five input parameters.
Specifically, in the present embodiment, the OD pairs of 44 stations of the express bus in xiamen city and the OD pairs of 289 stations of the subway in shanghai city are taken as examples: the prediction time interval Δ t is 15min, and the used historical OD data sets are OD data of 6:00-23:00 (working day) and weather data of 7/1/8/3/2019 in xiamen city and 4/1/4/29/2015 in shanghai city.
Referring to fig. 2, in a possible embodiment of the present invention, before invoking the trained neural network model to process the input data set, the method further includes:
s201, acquiring multiple groups of historical input data sets;
s202, performing data processing on the historical input data set to generate a plurality of groups of data result values;
specifically, step S202 includes:
according to the formula
Figure BDA0003628751140000111
Carrying out Min-Max standardization processing on the proximity data, trend data, periodic data and real-time inbound passenger flow data in the historical input data set to generate and map to [0,1]Data result values between ranges, where X max Maximum value of single-column characteristic data, X min Is the minimum value of the single-column characteristic data;
carrying out One-Hot coding digitization processing on discrete external factor data in the historical input data set to generate a digitized data result value;
and carrying out Min-Max linear normalization processing on the continuous external factor data in the historical input data set to generate a data result value which is scaled to the range of [0,1 ].
Specifically, in this embodiment, the history is input to the OD of the data set c 、OD pe 、OD tr Min-Max normalization of the four input parameters INFLOW to map the resulting values to [0, 1%]Wherein the formula used is
Figure BDA0003628751140000112
For input parameter M of discrete variable in historical input data set t Digitizing by using One-Hot coding; for the input parameter M of the continuous variable in the historical input data set t Scaling to [0,1] again by Min-Max Linear normalization]Ranges to generate sets of data result values.
S203, dividing the data result values into a training set and a test set according to a preset proportion;
specifically, in this embodiment, the OD pairs of 44 stops of the express bus in xiamen city and the OD pairs of 289 stops of the subway in shanghai city are taken as examples: the data were divided into training and test sets at a ratio of 4:1 for a total of five weeks; wherein the first four weeks of data are used to train the model and the last week of data is used for testing.
And S204, calling the training set as input data of the neural network, training the neural network, calling the test set to test the trained neural network, and generating a neural network model.
Specifically, step S204 includes:
establishing a neural network, and inputting the training set into the neural network for training as input data;
inputting the test set into a neural network, and calculating the Root Mean Square Error (RMSE) and the average absolute error (MAE) of the test set to serve as judgment standards;
and when the root mean square error RMSE and the average absolute error MAE reach preset values, storing the architecture and network parameters of the neural network, and generating a neural network model.
According to the formula
Figure BDA0003628751140000121
Calculating the Root Mean Square Error (RMSE) according to the formula
Figure BDA0003628751140000131
Calculating the mean absolute error MAE, wherein: m is ij And m ij1 The traffic flow is divided into an actual value and a predicted value from the station i to the station j.
Specifically, in the present embodiment, a neural network model is constructed according to the model overall structure as shown in fig. 3. Firstly, according to the characteristics of the space-time distribution of the passenger flow, the time attributes of the OD flow are divided into three categories, including time proximity, periodicity and trend components. The three components have the same network structure and are composed of a CAS-CNN module and a time evolution module. Secondly, a CAS-CNN module is constructed by combining a channel attention mechanism and the split CNN, so that the problem of data sparsity of an OD matrix is solved, input data are weighted, and importance degrees among the ODs at different times are captured. And thirdly, constructing an auxiliary information coding module, and coding auxiliary information, including real-time inbound passenger flow data and external influence factor data blended for making up the data availability problem of the OD. And secondly, constructing a time evolution module, and modeling the three historical space-time passenger flow components by using a convolution long-short term memory network (ConvLSTM) respectively to model the proximity, periodicity and trend of the space-time data. And finally, dynamically aggregating the three components by a parameter matrix-based method, distributing different weights for different components, and further fusing the weights with external factors to obtain a final prediction result.
In this embodiment, a training set is used as an input parameter to train a neural network, data of a test set is input into the neural network to be tested, a root mean square error RMSE and an average absolute error MAE are used as evaluation indexes of a prediction result, and when an effect achieved by testing the test set by a model is not ideal, parameters of a network model need to be adjusted until the test effect is ideal; and when the effect of the model on testing the test set is ideal, storing the trained network model architecture and the trained network parameters.
Specifically, in this embodiment, the OD pairs of 44 stops of the express bus in xiamen city and the OD pairs of 289 stops of the subway in shanghai city are taken as examples: the neural network model can be trained by adopting an Adam optimizer; wherein the learning rate is set to 0.0001, the weight attenuation parameter is set to 0.00002, the batch-size is set to 32, the loss function is MAE, the number of model iterations is 100, the prediction time interval Δ T is 15min, T e ＝T tr ＝T p 5, daily and weekly trend span q p 1.
S103, calling the trained neural network model to process the input data set, and generating OD passenger flow prediction data.
Specifically, in the present embodiment, by molding the present moldThe model is compared with five models, namely an HA model, an ARIMA model, an LSTM model and a CAS-CNN model, and the model is compared with 4 variant models constructed based on the model, so that the effectiveness of the model is evaluated according to the quality of evaluation indexes. Wherein the HA (historical average) model, uses
Figure BDA0003628751140000141
Calculating the result as the predicted OD; the ARIMA model is used for constructing an Auto ARIMA model for each OD pair; the CAS-CNN model and the LSTM model are both short-time orbit OD prediction methods, with parameter values consistent with the models herein. The results are verified by using an ablation experiment, firstly, periodic and trend passenger flow data are removed, and the passenger flow time sequence characteristics are verified, so that the prediction precision is improved; secondly, removing the CAS-CNN module, and verifying the influence of data sparsity on a prediction result; and finally, removing the real-time inbound passenger flow data, and verifying the data to relieve the data availability problem.
In this embodiment, the OD pairs of 44 stations of the express public transport in xiamen city and the OD pairs of 289 stations of the subway in shanghai city are taken as examples: and verifying and comparing the OD passenger flow short-time prediction method and other reference models on a express public transport data set and a track data set of Shanghai city based on the RMSE and MAE evaluation indexes. As shown in table 2:
table 2:
Figure BDA0003628751140000151
in order to verify the effectiveness of each module in the proposed model CAS-CNN-ConvLSTM, ablation experiments were performed, 3 variant models based on this model were constructed and experiments were performed on two data sets. As shown in table 3:
table 3:
Figure BDA0003628751140000152
further, in order to verify the influence of the passenger flow time sequence characteristics on the prediction result, experimental verification is carried out on the basis of the express public transportation data set in the city of mansion.
As shown in Table 2, from the prediction results of the models, HA and ARIMA are poor in prediction effect, and LSTM and CAS-CNN show better accuracy, which indicates that the spatio-temporal characteristics are helpful for improving the prediction accuracy. Compared with 5 models, the OD passenger flow short-time prediction method based on CAS-CNN-ConvLSTM shows better accuracy, and spatial characteristics and time regularity of OD flows are learned because the model considers the passenger flow space-time distribution characteristics more comprehensively. Based on the OD passenger flow short-time prediction method, 3 additional models were constructed according to conditions for experiments, as shown in table 3. As can be seen from table 3, the prediction accuracy is reduced after removing SENet, splitting CNN, and real-time inbound passenger flow, and thus, the CAS-CNN module effectively alleviates the problem of data sparsity and the importance of treating different time components differently, and at the same time, the problem of data availability can be effectively alleviated by integrating real-time inbound passenger flow data. As shown in fig. 4, the influence of the passenger flow time sequence characteristics on the prediction result is verified by taking a express bus data set in the city of mansion as an example. The prediction effect obtained by comprehensively considering the passenger flow time sequence characteristics and the weather factors is remarkable, and the information is helpful for improving the prediction precision.
And S104, adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
In summary, the neural network model is an OD short-time passenger flow prediction model based on CAS-CNN-ConvLSTM, which can not only alleviate the problems of data sparsity and availability, but also model different time component importance degrees through a channel attention mechanism. In addition, the model also considers the time sequence characteristics of the passenger flow, namely the approaching rule, the week rule, the trend rule and the external factors. The effectiveness of the model is evaluated on two real data sets of different cities, the performance of the model is obviously superior to that of 5 reference models, and the model is proved to be more suitable for OD passenger flow short-time prediction and can meet the requirements of engineering. The OD passenger flow short-time prediction method relieves the problems of data sparsity and usability, and the importance degree of OD data under different time is treated differently.
Referring to fig. 5, a second embodiment of the invention provides an OD traffic short-term prediction apparatus, comprising:
the data acquisition unit 101 is used for acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set;
an input parameter obtaining unit 102, configured to perform conversion processing on the historical OD data set to generate a time sequence, and perform stacking processing on the time sequence to generate an input data set, where Δ t is used as a prediction time interval to divide the time sequence within a preset time period, and stack the divided data;
the prediction data generation unit 103 is configured to invoke a trained neural network model to process the input data set, and generate OD passenger flow prediction data;
and the adjusting unit 104 is used for adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
A third embodiment of the present invention provides an OD passenger flow short-time prediction apparatus, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the OD passenger flow short-time prediction method as described in any one of the above items when executing the computer program.
A fourth embodiment of the present invention provides a readable storage medium, which is characterized by storing a computer program, wherein the computer program is executable by a processor of a device on which the storage medium is located, so as to implement the OD passenger flow short-time prediction method as described in any one of the above.
Illustratively, the computer programs described in the third and fourth embodiments of the present invention may be partitioned into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions that describe the execution of the computer program in the implementation of an OD traffic short-time prediction device. For example, the device described in the second embodiment of the present invention.
The Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, the processor is a control center of the OD traffic short-time prediction method, and various interfaces and lines are used to connect the whole implementation to various parts of the OD traffic short-time prediction method.
The memory may be used to store the computer programs and/or modules, and the processor may implement the various functions of an OD passenger flow short-time prediction method by running or executing the computer programs and/or modules stored in the memory and invoking the data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, a text conversion function, etc.), and the like; the storage data area may store data (such as audio data, text message data, etc.) created according to the use of the cellular phone, etc. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.
Wherein the implemented module, if implemented in the form of a software functional unit and sold or used as a stand-alone product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method embodiments may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.
It should be noted that the above-described device embodiments are merely illustrative, where the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. In addition, in the drawings of the embodiment of the apparatus provided by the present invention, the connection relationship between the modules indicates that there is a communication connection therebetween, and may be specifically implemented as one or more communication buses or signal lines. One of ordinary skill in the art can understand and implement it without inventive effort.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention.

Claims (10)

1. An OD passenger flow short-time prediction method is characterized by comprising the following steps:
acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set;
converting the historical OD data set to generate a time sequence, stacking the time sequence to generate an input data set, wherein the time sequence in a preset time period is divided by using delta t as a prediction time interval, and the divided data are stacked;
calling a trained neural network model to process the input data set to generate OD passenger flow prediction data;
and adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
2. An OD passenger flow short-term prediction method according to claim 1, wherein the historical OD dataset is converted to generate a time series, and the time series are stacked to generate an input dataset, specifically:
converting the historical OD data set to generate a plurality of groups of time sequences of preset time periods;
dividing the time sequence of each preset time period by using delta t as a prediction time interval, and stacking the divided time sequences of each preset time period respectively to generate an input data set, wherein the input data set comprises proximity data, trend data, periodic data, real-time inbound passenger flow data and external factor data.
3. An OD passenger flow short-term prediction method as claimed in claim 2, characterized in that it is based on a formula
Figure FDA0003628751130000011
Calculating the proximity data; according to the formula
Figure FDA0003628751130000012
Calculating the trend data; according to the formula
Figure FDA0003628751130000021
Calculating the periodic data; according to the formula
Figure FDA0003628751130000022
Calculating the real-time inbound passenger flow data; according to the formula
Figure FDA0003628751130000023
Calculating the external factor data; wherein l m Characteristic length of external factors,/ e For the length of the proximity sequence,/ tr To trend sequence length, l pe Is the periodic sequence length, p trending time span, q periodic time span, M t Is a feature vector of the external factors at the prediction time interval t.
4. The method of claim 1, wherein prior to invoking the trained neural network model to process the input data set, further comprising:
acquiring a plurality of groups of historical input data sets;
processing the historical input data set to generate a plurality of groups of data result values;
dividing the data result value into a training set and a test set according to a preset proportion;
and calling the training set as input data of the neural network, training the neural network, calling the test set to test the trained neural network, and generating a neural network model.
5. The OD passenger flow short-term prediction method according to claim 4, characterized in that the historical input data set is subjected to data processing to generate a plurality of sets of data result values, specifically:
according to the formula
Figure FDA0003628751130000024
Carrying out Min-Max standardization processing on the proximity data, trend data, periodic data and real-time arrival passenger flow data in the historical input data set to generate and map to [0,1]Data result values between ranges, where X max Maximum value of single-column characteristic data, X min Is the minimum value of the single-column characteristic data;
carrying out One-Hot coding digitization processing on discrete external factor data in the historical input data set to generate a digitized data result value;
and carrying out Min-Max linear normalization processing on the continuous external factor data in the historical input data set to generate a data result value which is scaled to the range of [0,1 ].
6. The OD passenger flow short-term prediction method of claim 4, wherein the training set is called as input data of a neural network to train the neural network, the test set is called to test the trained neural network to generate a neural network model, and specifically:
establishing a neural network, and inputting the training set into the neural network for training as input data;
inputting the test set into a neural network, and calculating the Root Mean Square Error (RMSE) and the average absolute error (MAE) of the test set to serve as judgment standards;
and when the root mean square error RMSE and the average absolute error MAE reach preset values, storing the architecture and network parameters of the neural network, and generating a neural network model.
7. An OD passenger flow short-term prediction method as claimed in claim 6, characterized by being based on a formula
Figure FDA0003628751130000031
Calculating the Root Mean Square Error (RMSE) according to the formula
Figure FDA0003628751130000032
Calculating the mean absolute error MAE, wherein: m is a unit of ij And m ij1 The traffic flow is divided into an actual value and a predicted value from the station i to the station j.
8. An OD traffic short-term prediction device, comprising:
the data acquisition unit is used for acquiring a historical passenger IC data set and external influence factor characteristic data, and performing OD matching processing on the historical passenger IC data set and the external influence factor characteristic data to generate a historical OD data set;
an input parameter obtaining unit, configured to perform conversion processing on the historical OD data set to generate a time sequence, and perform stacking processing on the time sequence to generate an input data set, where Δ t is used as a prediction time interval to divide the time sequence within a preset time period, and stack the divided data;
the prediction data generation unit is used for calling the trained neural network model to process the input data set and generate OD passenger flow prediction data;
and the adjusting unit is used for adjusting the real-time dynamic scheduling of the vehicle according to the OD passenger flow prediction data.
9. An OD passenger flow short-term prediction device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the OD passenger flow short-term prediction method of any one of claims 1 to 7 when executing the computer program.
10. A readable storage medium, in which a computer program is stored, the computer program being executable by a processor of a device in which the storage medium is located to implement the OD passenger flow short-time prediction method according to any one of claims 1 to 7.
CN202210483804.6A 2022-05-06 2022-05-06 OD passenger flow short-time prediction method, device, equipment and storage medium thereof Pending CN114819366A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210483804.6A CN114819366A (en) 2022-05-06 2022-05-06 OD passenger flow short-time prediction method, device, equipment and storage medium thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210483804.6A CN114819366A (en) 2022-05-06 2022-05-06 OD passenger flow short-time prediction method, device, equipment and storage medium thereof

Publications (1)

Publication Number Publication Date
CN114819366A true CN114819366A (en) 2022-07-29

Family

ID=82512044

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210483804.6A Pending CN114819366A (en) 2022-05-06 2022-05-06 OD passenger flow short-time prediction method, device, equipment and storage medium thereof

Country Status (1)

Country Link
CN (1) CN114819366A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110991775A (en) * 2020-03-02 2020-04-10 北京全路通信信号研究设计院集团有限公司 Deep learning-based rail transit passenger flow demand prediction method and device
CN111242292A (en) * 2020-04-26 2020-06-05 北京全路通信信号研究设计院集团有限公司 OD data prediction method and system based on deep space-time network
CN111242395A (en) * 2020-04-26 2020-06-05 北京全路通信信号研究设计院集团有限公司 Method and device for constructing prediction model for OD (origin-destination) data
CN112001548A (en) * 2020-08-25 2020-11-27 北京交通大学 OD passenger flow prediction method based on deep learning
CN112215408A (en) * 2020-09-24 2021-01-12 交控科技股份有限公司 Rail transit passenger flow volume prediction method and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110991775A (en) * 2020-03-02 2020-04-10 北京全路通信信号研究设计院集团有限公司 Deep learning-based rail transit passenger flow demand prediction method and device
CN111242292A (en) * 2020-04-26 2020-06-05 北京全路通信信号研究设计院集团有限公司 OD data prediction method and system based on deep space-time network
CN111242395A (en) * 2020-04-26 2020-06-05 北京全路通信信号研究设计院集团有限公司 Method and device for constructing prediction model for OD (origin-destination) data
CN112001548A (en) * 2020-08-25 2020-11-27 北京交通大学 OD passenger flow prediction method based on deep learning
CN112215408A (en) * 2020-09-24 2021-01-12 交控科技股份有限公司 Rail transit passenger flow volume prediction method and device

Similar Documents

Publication Publication Date Title
Xie et al. Short‐term traffic volume forecasting using Kalman filter with discrete wavelet decomposition
Xue et al. Short‐term bus passenger demand prediction based on time series model and interactive multiple model approach
Grubesic et al. Broadband telecommunications and regional development
CN101541016B (en) Method for predicting data and equipment
CN107071193A (en) The method and apparatus of interactive answering system accessing user
CN106227792A (en) Method and apparatus for pushed information
CN111881243B (en) Taxi track hot spot area analysis method and system
CN103095496A (en) Prediction method and device for network flow
CN111357051A (en) Speech emotion recognition method, intelligent device and computer readable storage medium
Liu et al. A MRT daily passenger flow prediction model with different combinations of influential factors
CN109102468B (en) Image enhancement method and device, terminal equipment and storage medium
CN114819366A (en) OD passenger flow short-time prediction method, device, equipment and storage medium thereof
CN116432665B (en) Dialogue model construction method, text generation method, device, system and equipment
CN111625629A (en) Task-based conversational robot response method, device, robot and storage medium
CN111104951A (en) Active learning method and device and terminal equipment
CN116258253A (en) Vehicle OD prediction method based on Bayesian neural network
CN114565080A (en) Neural network compression method and device, computer readable medium and electronic equipment
CN115249084A (en) Large-scale electric vehicle charging load hybrid prediction method considering periodic characteristics
CN113313386A (en) Intelligent voice investigation system and method for automobile financial risk
Jose et al. A stacked long short-term memory neural networks for parking occupancy rate prediction
CN112561186A (en) Short-term power load prediction method
Qu et al. An improved character recognition algorithm for license plate based on BP neural network
Bai et al. A NN-GM (1, 1) model-based analysis of network traffic forecasting
CN111178489B (en) Conversation robot engine flow distribution method and device
Junejo et al. Modified hybrid grey model (1, 1) to forecast cellular subscribers

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