Urban traffic demand prediction method and device, electronic equipment and storage medium

Abstract

The invention provides an urban traffic demand prediction method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining an urban grid map, mobile phone signaling data and urban feature statistical data of a target city; determining a target travel volume of each grid unit according to the mobile phone signaling data and the urban characteristic statistical data; constructing a grid feature vector of each grid unit according to the urban feature statistical data; for any starting grid unit and any target grid unit, respectively taking the corresponding grid feature vectors and the grid distance between the two grid units as target input parameters, inputting the target input parameters into a pre-trained depth gravity prediction model, and obtaining a travel selection probability between the starting grid unit and the target grid unit; and calculating traffic travel prediction data based on the travel selection probability and the target travel volume of the departure grid unit. According to the invention, multi-source data can be effectively fused by using the depth gravity model, and the precision of urban traffic demand prediction is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to, but are not limited to, the field of urban planning, and particularly relate to a method and device for predicting urban traffic demand, an electronic device, and a storage medium. Background Art

[0002] Urban traffic demand prediction is the core link of urban traffic planning and management, and is crucial for optimizing traffic resource allocation and alleviating congestion. Traffic demand refers to the total travel volume within a specific time and space range, which is comprehensively affected by various factors such as population distribution, land use, economic activities, and residents' travel habits, showing complex and dynamic characteristics.

[0003] In related technologies, traditional four-stage models are often used for prediction. However, this method relies on large-scale travel surveys, with slow data update, high cost, and a relatively fixed model structure, making it difficult to capture the rapid changes in demand. Some methods use time series or statistical regression models, which are effective in specific scenarios but usually difficult to comprehensively depict the complex interactive effects of multiple factors, resulting in insufficient prediction accuracy. Summary of the Invention

[0004] The present application aims to at least solve one of the technical problems existing in the prior art. For this purpose, the present application provides a method and device for predicting urban traffic demand, an electronic device, and a storage medium, which can effectively fuse multi-source data using a deep gravity prediction model to improve the accuracy of urban traffic demand prediction.

[0005] To achieve the above object, a first aspect of the embodiments of the present application proposes a method for predicting urban traffic demand, the method including: Obtain the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city; Determine the target travel volume of each grid unit in the urban grid map according to the urban grid map, the mobile phone signaling data, and the urban feature statistical data; Construct a grid feature vector for each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map; For any origin grid unit and any destination grid unit among multiple grid units, respectively input the corresponding grid feature vector and the grid distance between the two grid units as target input parameters into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability from the origin grid unit to the destination grid unit; Calculate traffic travel prediction data based on the travel selection probability and the target travel volume of the origin grid unit.

[0006] In some embodiments, determining the target travel volume of each grid cell in the urban grid map according to the urban grid map, the mobile phone signaling data, and the urban feature statistical data includes: Determining the historical baseline travel rate of each grid cell according to the urban grid map and the mobile phone signaling data; Extracting multiple grid attribute variables of each grid cell from the urban feature statistical data; wherein the multiple grid attribute variables include grid accessibility, land use mix, road traffic level, job-housing relationship coefficient, and grid housing price; Inputting the historical baseline travel rate and the multiple grid attribute variables into a pre-constructed travel rate regression model, so that the travel rate regression model outputs the target travel rate; Obtaining the population distribution data of each grid cell, and determining the target travel volume of each grid cell based on the target travel rate and the population distribution data.

[0007] In some embodiments, constructing a grid feature vector of each grid cell in the urban grid map according to the urban feature statistical data and the urban grid map includes: For each grid cell, parsing the urban feature statistical data to obtain the population feature, land use feature, traffic convenience feature, and industrial facility feature corresponding to the grid cell; Constructing a grid feature vector of each grid cell according to the population feature, the land use feature, the traffic convenience feature, and the industrial facility feature.

[0008] In some embodiments, for any origin grid cell and any destination grid cell among the multiple grid cells, respectively inputting the corresponding grid feature vector and the grid distance between the two grid cells as target input parameters into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability from the origin grid cell to the destination grid cell, includes: Respectively extracting the grid feature vectors of the origin grid cell and the destination grid cell as the first input parameter and the second input parameter; Calculating the grid distance between the origin grid cell and the destination grid cell, and taking the grid distance as the third input parameter; Input the first input parameter, the second input parameter, and the third input parameter as the target input parameter into the input layer of the depth gravity prediction model, so that the depth gravity prediction model performs feature learning and interaction processing on the first input parameter, the second input parameter, and the third input parameter through a neural network, and generates an internal representation of the interaction relationship between the departure grid cell and the destination grid cell; The depth gravity prediction model calculates and outputs the travel selection probability through the activation function of the output layer.

[0009] In some embodiments, the pre-training process of the depth gravity prediction model includes: Based on the mobile signaling data, construct a training data set, the training data set contains multiple groups of training samples, and each training sample includes: a departure grid cell feature vector and a destination grid cell feature vector corresponding to the first input parameter and the second input parameter respectively, a grid distance corresponding to the third input parameter, and a target label determined according to the mobile signaling data and representing the actual travel relationship between the departure grid cell and the destination grid cell; Input the departure grid cell feature vector, the destination grid cell feature vector, and the grid distance in the training sample into the depth gravity prediction model to obtain a model prediction result; Based on the model prediction result and the target label, calculate the prediction error using a predefined loss function; Based on a preset optimization algorithm, according to the prediction error, iteratively update the model parameters of the depth gravity prediction model until a preset convergence condition is met.

[0010] In some embodiments, calculating the traffic travel prediction data based on the travel selection probability and the target travel volume of the departure grid cell includes: For any departure grid cell and any destination grid cell in the urban grid map, multiply the target travel volume of the departure grid cell by the travel selection probability between the departure grid cell and the destination grid cell to obtain the predicted travel volume between the departure grid cell and the destination grid cell.

[0011] In some embodiments, the method further includes: Divide the mobile signaling data and the urban feature statistical data according to a preset population category and a preset time period to obtain classified mobile signaling data and classified urban feature statistical data corresponding to each population category in each time period; Determine the classified target travel volume of each population category in each grid cell for each time period according to the urban grid map, the classified mobile phone signaling data, and the classified urban feature statistical data; Construct a classified grid feature vector of each population category in each grid cell for each time period in the urban grid map according to the classified urban feature statistical data and the urban grid map; For any origin grid cell and any destination grid cell among multiple grid cells, respectively input the corresponding classified grid feature vector and the grid distance between the two grid cells as target input parameters into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the classified travel selection probability between the origin grid cell and the destination grid cell; Calculate the classified traffic travel prediction data of each population category in each grid cell for each time period based on the classified travel selection probability and the classified target travel volume of the origin grid cell.

[0012] In a second aspect, an embodiment of the present application provides an urban traffic demand prediction device, including: An acquisition module that acquires an urban grid map, mobile phone signaling data, and urban feature statistical data of a target city; A determination module that determines the target travel volume of each grid cell in the urban grid map according to the urban grid map, the mobile phone signaling data, and the urban feature statistical data; A construction module that constructs a grid feature vector of each grid cell in the urban grid map according to the urban feature statistical data and the urban grid map; A prediction module that, for any origin grid cell and any destination grid cell among multiple grid cells, respectively inputs the corresponding grid feature vector and the grid distance between the two grid cells as target input parameters into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability between the origin grid cell and the destination grid cell; A calculation module that calculates traffic travel prediction data based on the travel selection probability and the target travel volume of the origin grid cell.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, including: a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the urban traffic demand prediction method according to any one of the first aspect embodiments of the present application.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a program, which when executed by a processor, implements the urban traffic demand prediction method according to any one of the embodiments of the first aspect of the present application.

[0015] The urban traffic demand prediction method proposed in the embodiment of the present application includes: obtaining the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city; determining the target travel volume of each grid unit in the urban grid map according to the urban grid map, mobile phone signaling data, and urban feature statistical data; constructing a grid feature vector for each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map; for any origin grid unit and any destination grid unit among multiple grid units, respectively taking the corresponding grid feature vector and the grid distance between the two grid units as target input parameters and inputting them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability from the origin grid unit to the destination grid unit; calculating traffic travel prediction data based on the travel selection probability and the target travel volume of the origin grid unit.

[0016] The urban traffic demand prediction method proposed in the present application first obtains the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city, constructs a multi-dimensional data foundation including actual travel behaviors reflected by mobile phone signaling, spatial geographical information reflected by the grid map, and regional social and economic attributes reflected by urban feature statistical data, overcomes the one-sidedness or bias that may be brought by relying on a single data source, and lays a comprehensive data support for subsequent accurate prediction. Then, it independently determines the target travel volume of each grid unit, accurately simulates the total scale of travel generation and attraction in each region of the city, and provides a key benchmark for subsequent travel distribution prediction. Next, by constructing a grid feature vector for each grid unit that can reflect multi-dimensional information such as its function and facilities, and combining the physical distance between grid units, these structured spatial and attribute information are input into a pre-trained deep gravity prediction model; this deep learning model can learn and express complex, non-linear spatial interaction relationships, thereby calculating the accurate travel selection probability between any two grid units, and effectively capturing the internal law of residents' travel distribution choices. Finally, by combining the target travel volume of each origin grid unit representing "total travel volume" and the travel selection probability representing "distribution tendency" calculated independently in the first two steps, specific and quantified inter-grid traffic travel prediction data are calculated, realizing a complete and refined prediction from total travel volume to travel distribution. In summary, the present application can effectively integrate multi-source data using the deep gravity prediction model and improve the accuracy of urban traffic demand prediction.

[0017] Other features and advantages of the present application will be described in the following specification, and in part will be obvious from the specification, or can be understood by implementing the present application. The objectives and other advantages of the present application can be achieved and obtained through the structures specifically pointed out in the specification, claims and drawings. Description of the Drawings

[0018] Figure 1 is a schematic flowchart of a method for predicting urban traffic demand provided by an embodiment of the present application; Figure 2 is a schematic flowchart of a method for predicting urban traffic demand provided by another embodiment of the present application; Figure 3 is a schematic flowchart of a method for predicting urban traffic demand provided by another embodiment of the present application; Figure 4 is a schematic flowchart of a method for predicting urban traffic demand provided by another embodiment of the present application; Figure 5 is a schematic flowchart of a method for predicting urban traffic demand provided by another embodiment of the present application; Figure 6 is a schematic flowchart of a method for predicting urban traffic demand provided by another embodiment of the present application; Figure 7 is a schematic diagram of an apparatus for predicting urban traffic demand provided by an embodiment of the present application; Figure 8 is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present application. Detailed Embodiments

[0019] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] It should be noted that although functional module division is performed in the apparatus schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order from the module division in the apparatus or the order in the flowchart. Terms such as "first", "second", etc. in the specification, claims and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0022] With the continuous acceleration of the urbanization process, the urban population scale continues to expand, the urban spatial structure becomes increasingly complex, and the travel demands of residents show the characteristics of diversification and dynamism. As an important foundation for urban traffic planning, traffic management, and infrastructure construction, urban traffic demand forecasting is of great significance for optimizing traffic resource allocation, improving traffic operation efficiency, alleviating traffic congestion, and promoting the sustainable development of cities.

[0023] In related technologies, traditional urban traffic demand forecasting methods mainly include the four-stage method (i.e., trip generation and attraction, trip distribution, mode split, and route assignment), etc. These methods usually rely on large-scale household travel survey data. By statistical analysis and parameter calibration, a relationship model between travel demand and factors such as population, land use, and economic activities is established. However, traditional methods have the following deficiencies: Traditional methods highly rely on periodic large-scale travel surveys. The data collection cycle is long and the cost is high, making it difficult to timely reflect the dynamic changes of urban traffic demand, resulting in limitations in the timeliness and accuracy of prediction results. At the same time, the traditional model structure is relatively fixed and it is difficult to flexibly adapt to the rapid changes in urban spatial structure and residents' travel behavior. With the development of information technology, a large number of new data sources have emerged in the field of urban traffic, such as mobile phone signaling, GPS trajectories, etc. Traditional methods are difficult to effectively integrate multi-source heterogeneous data and cannot fully exploit and utilize the rich information contained in these data.

[0024] In recent years, with the rapid development of big data and artificial intelligence technologies, traffic demand forecasting methods based on machine learning have gradually emerged. Such methods can automatically learn the complex non-linear relationships in the data, improving the prediction accuracy and generalization ability. However, the existing deep learning methods in the field of traffic demand forecasting still face the following challenges: How to effectively integrate multi-source heterogeneous data to construct a data foundation that comprehensively reflects urban travel characteristics; how to achieve refined prediction of travel demand at high spatial resolution; how to combine the interpretability of traditional traffic models with the powerful expressive ability of deep learning models to improve the practicality of the model. Therefore, there is an urgent need for an urban traffic demand forecasting method that can integrate multi-source data, has high spatial resolution, and can effectively capture the complex relationships between urban space and travel behavior to meet the actual needs of modern urban traffic management and planning.

[0025] Based on this, the embodiments of the present application provide an urban traffic demand forecasting method, device, electronic device, and storage medium, which can effectively integrate multi-source data by using a deep gravity prediction model and improve the accuracy of urban traffic demand forecasting.

[0026] The urban traffic demand forecasting method, device, electronic device, and storage medium provided by the embodiments of the present application are specifically described through the following embodiments. First, the urban traffic demand forecasting method in the embodiments of the present application is described.

[0027] This application can be used in many general or specific computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This application can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0028] It should be noted that in each specific embodiment of this application, when it comes to relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first. Moreover, the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when this application embodiment needs to obtain sensitive personal information of the user, the user's separate permission or separate consent will be obtained through methods such as pop-up windows or redirecting to a confirmation page. After clearly obtaining the user's separate permission or separate consent, the necessary user-related data for the normal operation of this application embodiment will be obtained.

[0029] Figure 1 It is an optional flowchart of the urban traffic demand prediction method provided by the embodiment of this application. Figure 1 The method in [figure] may include but is not limited to steps 101 to 105.

[0030] Step 101, obtain the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city.

[0031] Step 102, determine the target travel volume of each grid unit in the urban grid map according to the urban grid map, mobile phone signaling data, and urban feature statistical data.

[0032] Step 103, construct a grid feature vector for each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map.

[0033] Step 104: For any origin grid cell and any destination grid cell among multiple grid cells, respectively take the corresponding grid feature vector and the grid distance between the two grid cells as target input parameters, and input them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel choice probability from the origin grid cell to the destination grid cell.

[0034] Step 105: Calculate traffic travel prediction data based on the travel choice probability and the target travel volume of the origin grid cell.

[0035] Steps 101 to 105 illustrated in the embodiments of the present application first obtain the urban grid map, mobile signaling data, and urban feature statistical data of the target city, and construct a multi-dimensional data foundation including actual travel behaviors reflected by mobile signaling, spatial geographical information reflected by the grid map, and regional social and economic attributes reflected by urban feature statistical data, overcoming the one-sidedness or deviation that may be brought by relying on a single data source, and laying a comprehensive data support for subsequent accurate prediction. Then, independently determine the target travel volume of each grid cell, accurately simulate the total scale of travel generation and attraction in each region of the city, and provide a key benchmark for subsequent travel distribution prediction. Next, by constructing a grid feature vector for each grid cell that can reflect its multi-dimensional information such as function and facilities, and combining the physical distance between grid cells, input these structured spatial and attribute information into a pre-trained deep gravity prediction model; this deep learning model can learn and express complex and non-linear spatial interaction relationships, so as to calculate the accurate travel choice probability between any two grid cells, effectively capturing the internal law of residents' travel distribution choices. Finally, combine the target travel volume of each origin grid cell, which represents the "total travel volume", and the travel choice probability, which represents the "distribution tendency", calculated independently in the previous two steps, and calculate specific and quantitative inter-grid traffic travel prediction data, realizing a complete and refined prediction from travel volume to travel distribution. In summary, the embodiments of the present application can effectively integrate multi-source data using a deep gravity prediction model and improve the accuracy of urban traffic demand prediction.

[0036] In step 101 of some embodiments, this step aims to collect multi-source data of the target city within a specific analysis period. One of them is the "urban grid map", which usually standardizes the urban geographical space according to basic geographical information. The one-kilometer resolution grid standard adopted in the implementation of the present invention means that the urban area is divided into grid cells with a side length of one kilometer, each having a unique identifier and coordinate range, and these grid cells constitute the basic units for subsequent spatial positioning, data aggregation, and analysis.

[0037] The second is the key dynamic behavior data source - "mobile phone signaling data". This refers to the spatio-temporal record data generated during the interaction between user terminals and communication base stations, which is provided by mobile communication network operators and has been anonymized and privacy desensitized. When mobile phone users are active in the mobile network, data records containing information such as timestamps, encrypted user identities, and the locations of connected base stations are generated. In the embodiments of the present invention, the core use of these massive mobile phone signaling data lies in accurately identifying and reconstructing the "travel chains" of urban residents, that is, identifying the actual travel trajectory sequences of users, including the starting positions of trips, possible intermediate stopping points, ending positions of trips, as well as the corresponding travel occurrence times and durations. Furthermore, the travel flows between different origin-destination pairs (OD pairs) can be statistically calculated. In addition, on the premise of complying with privacy protection regulations, non-identity-related information such as the types of user terminal devices and some application usage characteristics can also be combined to statistically infer group attributes such as the gender, age segments, possible residence or workplace of users, so as to support more detailed classification modeling.

[0038] The third is the "urban feature statistical data", which is a comprehensive data set aiming to comprehensively describe the internal attributes and external environments of each grid cell. This data set deeply integrates "basic geographic information data", such as detailed urban "road network data" (including road grades, lengths, topological connection relationships, etc.), "Point of Interest" (POI) data reflecting urban functional distributions (such as the locations and types of various commercial facilities, public service institutions, and transportation stations), and "Area of Interest" (AOI) data (such as large residential areas, park green spaces, and industrial areas). At the same time, it also includes traditional social and economic statistical indicators, such as the population density within each grid cell, the area ratios of different land use types (such as residential, commercial, office, industrial, green space, etc.), and the number of employment positions. These urban feature statistical data, especially the basic geographic information elements among them, not only provide a basis for understanding the urban spatial structure, but also, in the present invention, they can be used as explanatory variables in the benchmark travel generation model and are also the key input data sources for constructing the grid feature vectors required for the subsequent deep gravity prediction model.

[0039] In step 102 of some embodiments, it is necessary to quantify the "trip generation" link of urban traffic based on the basic data obtained in step 101. Step 102 aims to determine the "target trip volume" departing from each grid cell of the "urban grid map" within a specific time period (for example, the morning peak period of a working day). Here, the "target trip volume" generally refers to the trip generation volume in the field of traffic planning, that is, the total number of person-trips generated or attracted by a specific area per unit time. This trip volume is calculated or predicted through a certain analysis model or statistical method, comprehensively using the historical trip activity intensity inferred from mobile signaling data, factors related to trip generation in urban characteristic statistical data (such as population quantity, job quantity, land use nature, etc.), and the spatial units provided by the urban grid map. The target trip volume represents the potential total trip demand scale of each grid cell as a trip origin, and is the starting point for subsequent trip distribution calculation.

[0040] Please refer to Figure 2 , in some embodiments, step 102 may include but is not limited to steps 201 to 204.

[0041] Step 201, according to the urban grid map and mobile signaling data, determine the historical baseline trip rate of each grid cell.

[0042] Step 202, extract multiple grid attribute variables of each grid cell from the urban characteristic statistical data.

[0043] Step 203, input the historical baseline trip rate and multiple grid attribute variables into a pre-constructed trip rate regression model, so that the trip rate regression model outputs the target trip rate.

[0044] Step 204, obtain the population distribution data of each grid cell, and determine the target trip volume of each grid cell based on the target trip rate and the population distribution data.

[0045] In step 201 of some embodiments, to obtain the basic reference value of the trip volume in each area, it is necessary to calculate and determine the historical baseline trip rate of each grid cell in the urban grid map according to the spatial units defined by the urban grid map and in combination with the mobile signaling data of the historical time series. This process generally involves statistical analysis of the actual travel activities of residents reflected in the mobile signaling data. For example, identify the number of trips originating from each grid cell within a past representative period (such as several working days or weekends), and divide it by a certain baseline quantity of the grid cell (such as the estimated resident population or active population) to obtain a standardized ratio. This historical baseline trip rate reflects the inherent average trip generation level of each grid cell based on historical observations.

[0046] In step 202 of some embodiments, to gain a deeper understanding of the potential factors influencing travel, it is necessary to extract and quantify "multiple grid attribute variables" for "each grid cell" from "urban characteristic statistical data" covering multi-dimensional information such as urban planning, social economy, and transportation facilities. These variables are carefully selected based on economic geography theory and transportation planning practice to characterize the inherent characteristics of the grid cell, specifically including: Grid accessibility, which comprehensively measures the convenience of reaching other important functional areas in the city (such as employment centers, commercial areas, public service facilities, etc.) from this grid; Land use mix, which quantifies the level of coexistence of different functional land use types (such as residential, commercial, industrial, public service, green space, etc.) within the grid and reflects the potential of the local area to meet the diverse activity needs of residents; Road traffic level. This variable specifically also includes the road network density, intersection density within the grid, and the service coverage level of public transportation (such as bus stops, subway stations); Job-housing relationship coefficient, which is used to measure the balance between the number of job positions and the resident population size within the grid; Grid housing price, which is an important indirect indicator reflecting the regional social and economic status, attractiveness, and resident income level. Step 202 provides key explanatory variables for subsequent travel rate modeling and a multi-dimensional perspective for depicting the complex characteristics of the grid cell.

[0047] In step 203 of some embodiments, the "historical baseline travel rate" obtained in step 201 and the "multiple grid attribute variables" extracted in step 202 are jointly used as inputs and imported into a "pre-built travel rate regression model". This model is a mathematical relationship calibrated based on historical data, and it learns and quantifies the statistical correlation strength (reflected as regression coefficients) between the historical travel rate and various grid attribute variables. When applied, the predicted values of the grid attribute variables in the target year (prediction year) are substituted into the calibrated model, and combined with the historical baseline rate, the model can output a predictive "target travel rate". This target travel rate reflects the expected unit population travel intensity of this grid cell under the predicted urban development conditions (such as land use changes, transportation improvements, housing price fluctuations, etc.).

[0048] In step 204 of some embodiments, it is first necessary to obtain the "population distribution data of each grid cell", which usually comes from the prediction results of the population spatial distribution in the census and special surveys, and in some embodiments, the population data of different groups (such as classified by age and gender) and different time periods (weekdays, weekends, holidays) can be distinguished. Then, based on the "target travel rate" for a specific population and specific time period output in step 203, a multiplication calculation is performed with the "population distribution data" of the corresponding grid cell, corresponding population, and time period, so as to determine the "target travel volume of each grid cell". This "target travel volume" represents the total number of trips expected to be generated by a specific population within a specific grid cell during a specific time period under specific conditions in the target year. This step completes the conversion from travel rate to travel volume, providing the total travel volume control at the starting point for subsequent travel distribution prediction.

[0049] In some embodiments, the linear regression model formula for travel rate prediction is as follows:

[0050] is the travel rate of the a - type residents in grid r during the s - period in the t - th period (i.e., the target year), which is calibrated by the above - mentioned regression model and calculated after substituting the regression variables of the prediction year, and is the calculation target variable of this model; is the travel rate of the a - type residents in grid r during the s - period in the (t - 1) - th period (i.e., the base year), which is the ratio of the total traffic travel volume of the a - type population to the total number of people in this type, and is obtained by cleaning the relevant fields in the mobile signaling data; is the accessibility of grid r in the t - th period. Accessibility comprehensively measures the difficulty for people to reach other main activity area destinations from a certain area to obtain facility services, and has a direct effect on the generation of grid residents' travel; is the land - use mixing index of grid r in the t - th period, which is calculated from the current land - use status and planning data in the basic geographic information data, and can reflect the diversity characteristics of the internal spatial form of the block and the degree of mixed use of urban functional spaces, and is used to measure the degree to which the mixed - use of land types in the traffic zone meets the travel demands of residents within the zone; The road traffic level considers the urban road network and public transportation service facilities such as buses, including the road network density 、the road intersection density 、the number of bus stops within grid r ,which is obtained through spatial statistics of the traffic network and station data in the basic geographic information data; is the employment - residence relationship coefficient of grid r in the t - th period, which comes from the mobile signaling data and the regional statistical yearbook, and can be calculated by calling from the social - economic attribute database; Denote the housing price of grid r at time t, which is sourced from the open-source housing purchase platform data and can be retrieved from the socioeconomic attribute database. The housing price reflects the income characteristics of the residents and workers in the neighborhood and the possible characteristics of travel modes, and has a direct impact on the generation of trips by grid residents. - is the regression model coefficient, which is calibrated by substituting the base-year trip rate data and various regression variables into the trip rate model and then carried over to the prediction of the target-year trip rate. Then, through and multiplying with the population distribution data of each grid cell, the target trip volume can be obtained. , representing the target trip volume between grids i and j during the s time period for the a-th type of population at time t.

[0051] Through steps 201 to 204, the embodiments of the present application first captured the historical base trip rate of historical trips using mobile signaling data. Then, by introducing multiple grid attribute variables (such as grid accessibility, land use mix, etc.), the urban built environment and socioeconomic factors affecting trips were taken into account, enhancing the model's ability to perceive and explain regional differences. By using the trip rate regression model, which can, in a data-driven manner, integrate historical trends and multi-dimensional current characteristics, learn the linear relationship between them, and thus more accurately correct and predict the trip rate to obtain the target trip rate. Finally, by combining the population distribution data to complete the conversion from trip rate to trip volume, the synergistic effect of this series of steps makes the finally output target trip volume close to reality, significantly improving the accuracy and reliability of the entire urban traffic demand forecasting method.

[0052] In step 103 of some embodiments, it is necessary to prepare the input for the deep learning model by converting the complex urban characteristics of each grid cell into a format that the model can handle. This step constructs a "grid feature vector" for each grid cell based on the "urban feature statistical data" and the "urban grid map" obtained in step 101. The so-called "grid feature vector" refers to an ordered list (vector) containing multiple numerical elements, where each element represents the feature measurement of the grid cell in a specific dimension. For example, the vector may contain quantified feature values such as the population quantity of the grid, the quantity of different types of POIs, the land use mix index, the road network density, the public transportation accessibility index, etc. By aggregating and organizing the urban feature statistical data according to the spatial units of the urban grid map, feature vectors that can comprehensively and structurally describe the internal attributes of each grid cell are finally formed. These feature vectors will serve as the key input for the subsequent deep learning model to understand the functions and attractiveness of grid cells.

[0053] Please refer to Figure 3 , in some embodiments, step 103 may include, but is not limited to, steps 301 to 302.

[0054] Step 301: For each grid cell, analyze the urban feature statistical data to obtain the population characteristics, land use characteristics, transportation convenience characteristics, and industrial facility characteristics corresponding to the grid cell.

[0055] Step 302: Construct a grid feature vector for each grid cell based on the population characteristics, land use characteristics, transportation convenience characteristics, and industrial facility characteristics.

[0056] In step 301 of some embodiments, in order to comprehensively and meticulously understand the inherent attributes of the basic units that constitute the urban texture, it is necessary to analyze and extract information from the pre-collected and organized urban feature statistical data for each grid cell in the urban grid map. This process aims to screen and summarize the index data from the vast amount of urban information that can accurately reflect the status and characteristics of the specific grid cell in multiple key dimensions. Specifically, these data are organized into four major categories: population characteristics, which are used to describe the number of residents, age and gender structure, employment status, and the relationship between workplace and residence within the grid; land use characteristics, which aim to reveal the utilization nature, area proportion, and spatial mixing pattern of various types of land (such as residential, industrial, commercial, public service, transportation, green space, etc.) within the grid; transportation convenience characteristics, which are used to quantify the geographical location advantage of the grid cell, the development degree of the internal and external connected road network (covering highways, expressways, main roads, secondary roads, branch roads, etc.), and the coverage level of public transportation (bus stops, subway stations) services; and industrial facility characteristics, which reflect the economic vitality (such as real estate prices), business prosperity, and social service function supporting conditions within the grid, and are usually measured by the number of Points of Interest (POI) and the area of Areas of Interest (AOI) to measure the distribution density and scale of various facilities (such as catering, sanitation, education, commerce, entertainment, office, service, transportation auxiliary facilities, etc.).

[0057] In step 302 of some embodiments, the feature information from different dimensions and with diversity is integrated and formally expressed. The specific operation is to construct a grid feature vector for each grid cell according to the extracted feature indicators and their corresponding values. Here, the grid feature vector refers to an ordered list of numerical values or an array, and each element in the vector corresponds to a specific regional feature indicator value. For example, in the regional feature variable system including 4 major categories and a total of 30 specific indicators shown in Table 1 and Table 2, the constructed grid feature vector will sequentially include a series of quantitative values such as "total resident population", "residential land area", "location accessibility", "real estate price", "POI number and AOI area" of various facilities. The core purpose of constructing the grid feature vector is to generate a feature expression for computer processing for each grid cell, so that it can be used as a unified input unit and effectively applied to subsequent machine learning models.

[0058] Table 1 Population and Land Area Feature Indicators of Different Regional Feature Categories

[0059] Table 2 Traffic and Industry Feature Indicators of Different Regional Feature Categories

[0060] In step 104 of some embodiments, a pre-trained "deep gravity prediction model" can be used to simulate the choice behavior of travelers between different destinations. Here, the "deep gravity prediction model" is an advanced prediction model that draws on the idea of the traditional gravity model but adopts a deep learning network structure. The fact that the model has been pre-trained means that it has learned the complex non-linear relationships in the urban travel pattern using a large amount of historical travel data (usually from mobile phone signaling data). In this step, for any pair of grid cells (one as the "departure grid cell" and the other as the "destination grid cell"), their "grid feature vectors" constructed in step 103 and the "grid distance" between the two are jointly used as the input parameters of the deep gravity prediction model. The neural network inside the model will perform complex non-linear transformations, interactions and learning on the input feature vectors, and finally output a scalar value, which represents the relative possibility that a traveler departing from this departure grid cell chooses this destination grid cell, that is, the "travel choice probability". This probability reflects the choice tendency under the combined action of destination attractiveness, origin characteristics and travel impedance.

[0061] Please refer to Figure 4 , in some embodiments, step 104 may include, but is not limited to, steps 401 to 404.

[0062] Step 401: Extract the grid feature vectors of the departure grid cell and the destination grid cell respectively as the first input parameter and the second input parameter.

[0063] Step 402: Calculate the grid distance between the departure grid cell and the destination grid cell, and use the grid distance as the third input parameter.

[0064] Step 403: Input the first input parameter, the second input parameter, and the third input parameter as the target input parameters into the input layer of the deep gravity prediction model, so that the deep gravity prediction model performs feature learning and interaction processing on the first input parameter, the second input parameter, and the third input parameter through the neural network, and generates an internal representation of the interaction relationship between the departure grid cell and the destination grid cell.

[0065] Step 404: The deep gravity prediction model calculates and outputs the travel selection probability through the activation function of the output layer.

[0066] In step 401 of some embodiments, first extract the grid feature vectors representing the inherent attributes of the departure grid cell, and this vector is designated as the first input parameter; at the same time, extract the grid feature vectors representing the inherent attributes of the destination grid cell, and this vector is designated as the second input parameter.

[0067] In step 402 of some embodiments, calculate the spatial distance or time distance between the departure grid cell and the destination grid cell determined in step 401. This quantified distance value, that is, the grid distance, is determined as the third input parameter. The grid distance, as a key factor measuring travel impedance, directly reflects the physical or time cost required to move between two grid cells.

[0068] In step 403 of some embodiments, jointly use the first input parameter (departure grid feature vector) obtained in step 401, the second input parameter (destination grid feature vector), and the third input parameter (grid distance) calculated in step 402 as the target input parameters, and input them into the input layer of the pre-constructed deep gravity prediction model. Subsequently, the deep gravity prediction model uses its internal neural network structure to perform in-depth feature learning and interaction processing on these three input parameters. This process aims to discover and learn the complex, non-linear interaction relationship among the characteristics of the departure location, the characteristics of the destination, and the distance between them, so as to generate an internal representation that can characterize the travel interaction intensity between the specific departure grid cell and the destination grid cell.

[0069] In step 404 of some embodiments, the deep gravity prediction model passes the internal representation generated in step 403 to its output layer and applies a preset activation function (such as the Softmax function) to calculate and transform the internal representation. Through the calculation of the activation function, the model finally outputs a specific numerical value, which is the predicted travel selection probability from the origin grid cell to the destination grid cell. This probability value quantitatively expresses the likelihood of a travel occurring under the conditions of given origin, destination characteristics, and the distance between them.

[0070] Through steps 401 to 404, the embodiments of the present application can achieve refined, data-driven prediction of travel selection probability. First, by extracting the specific origin and destination grid feature vectors (step 401) and the grid distance between them (step 402), it is ensured that the model input includes the key microscopic factors and impedance factors affecting travel. Then, using the neural network of the deep gravity prediction model for feature learning and interaction processing (step 403), complex non-linear dependence relationships between various factors are captured, generating a deep understanding of the OD interaction relationship. Finally, the travel selection probability is calculated through the activation function of the output layer (step 404). This series of steps together constitute an effective mechanism for simulating and predicting complex travel selection behaviors using deep learning technology, improving the accuracy of prediction.

[0071] Please refer to Figure 5 , in some embodiments, the pre-training process of the deep gravity prediction model may include, but is not limited to, steps 501 to 504.

[0072] Step 501, based on mobile phone signaling data, construct a training data set. The training data set contains multiple groups of training samples, and each training sample includes: an origin grid cell feature vector and a destination grid cell feature vector corresponding to the first input parameter and the second input parameter respectively, a grid distance corresponding to the third input parameter, and a target label determined according to the mobile phone signaling data and characterizing the actual travel relationship between the origin grid cell and the destination grid cell.

[0073] Step 502, input the origin grid cell feature vector, the destination grid cell feature vector, and the grid distance in the training sample into the deep gravity prediction model to obtain the model prediction result.

[0074] Step 503, based on the model prediction result and the target label, calculate the prediction error using a predefined loss function.

[0075] Step 504, based on a preset optimization algorithm according to the prediction error, iteratively update the model parameters of the deep gravity prediction model until the preset convergence condition is met.

[0076] In step 501 of some embodiments, based on the collected mobile phone signaling data, it is converted into structured training samples. Each training sample specifically includes feature vectors representing the internal attributes in the departure grid cell and the destination grid cell respectively (these two respectively correspond to the first input parameter and the second input parameter in the subsequent model application stage), the grid distance quantifying the travel impedance between these two grid cells (corresponding to the third input parameter), and a target label. This target label is obtained through statistical analysis of the mobile phone signaling data and is used to truly represent the actual travel relationship from the departure grid cell to the destination grid cell (for example, whether there is travel, the frequency or probability of travel, etc.).

[0077] In step 502 of some embodiments, training samples are selected from the training dataset, and the feature vectors of the departure grid cell, the feature vectors of the destination grid cell, and the grid distance between them in the samples are input into the input layer of the deep gravity prediction model to be trained in a predetermined format. After receiving these inputs, the model will use its current model parameters to perform a series of calculations and transformations through the internal neural network layers, and finally generate a model prediction result in the output layer. The neural network layer includes 15 hidden layers, where the first 6 layers have a dimension of 256 and the others have a dimension of 128. In each hidden layer, the parameter matrix is applied to the input variables. The neurons in the hidden layer use the Leaky Rectified Linear Unit function (LeakyReLu) as the activation function. This prediction result represents the estimated value of the travel relationship between the departure and destination grid cells corresponding to the training sample based on the current parameter state of the model.

[0078] In step 503 of some embodiments, after obtaining the prediction result of the model, it is necessary to evaluate the accuracy of the prediction. For this purpose, the model prediction result obtained in step 502 is compared with the target label representing the actual situation determined in step 501 for this training sample. This comparison is quantified through a predefined loss function (such as the cross-entropy loss function). The value calculated by the loss function is the prediction error, which measures the degree of difference between the model prediction result and the true target label.

[0079] In step 504 of some embodiments, the calculated prediction error will be used to guide the parameter optimization process of the model. A preset optimization algorithm, such as the gradient descent method, is adopted to adjust the internal parameters of the deep gravity prediction model according to the prediction error calculated in step 503. The optimization objective is to minimize the value of the loss function, that is, to reduce the prediction error. This "calculate error - update parameter" process will be iterated repeatedly for the samples in the training dataset. In each iteration, the parameters of the model will be finely tuned in the direction of reducing the overall prediction error. This iterative process will continue until a preset convergence condition is met, such as the prediction error being lower than a certain threshold, the performance of the model on the validation set no longer improving significantly, or the preset maximum number of iterations being reached.

[0080] Through steps 501 to 504, the embodiments of the present application construct a complete pre-training process of a deep gravity prediction model based on supervised learning. First, by using real mobile signaling data to construct a training dataset containing rich features and actual travel labels (step 501), the data basis for the training process is ensured to be close to reality. Then, through the forward propagation of the model (step 502), the error calculation based on the loss function (step 503), and the iterative parameter update using the optimization algorithm (step 504), a closed-loop learning mechanism is formed. This mechanism enables the deep gravity prediction model to learn the complex and non-linear internal laws between the origin features, destination features, spatial distance, and actual travel choices from a large amount of actual travel data. The finally obtained trained model has its internal parameters fully optimized and can accurately capture and simulate real travel decision-making behaviors, so as to output high-precision travel choice probabilities in subsequent actual prediction tasks (such as step 104).

[0081] In step 105 of some embodiments, the synthesis calculation of the final traffic demand prediction result is performed. This step combines the total amount of trip generation and the probability of trip distribution based on the intermediate results obtained in the previous steps to calculate the specific OD (Origin-Destination) flow. Specifically, it is achieved by multiplying the "target trip volume" of each "departure grid cell" determined in step 102 by the "travel choice probability" from this departure grid cell to all possible "destination grid cells" in step 104. By performing this calculation for all possible origin-destination pairs (OD pairs), the complete "traffic travel prediction data" can be obtained, which usually appears as an OD matrix. Each element in the matrix represents the predicted traffic travel volume from grid cell i to grid cell j, the traffic connection intensity between regions within the city, and the specific flow distribution. The specific formula is as follows:

[0082]

[0083] In the formula, represents the travel distribution volume of the a-th type of population between grids i and j during the s-th period at time t, which is the target variable of this step; represents the travel selection probability of the a-th type of population between grids i and j during the s-th period at time t, which is obtained through the deep gravity prediction model; represents the target travel volume of the a-th type of population in grid i during the s-th period at time t; represents a series of traffic demand variables related to the travel grid, population, and period at time t, which are obtained by summarizing according to the demand.

[0084] Please refer to Figure 6 , in some embodiments, the method provided by the embodiments of the present application may further include, but is not limited to, steps 601 to 605.

[0085] Step 601: Divide the mobile signaling data and the urban feature statistical data according to the preset population categories and preset time periods to obtain the classified mobile signaling data and the classified urban feature statistical data corresponding to each population category in each time period.

[0086] Step 602: Determine the classified target travel volume of each population category in each grid unit in each time period according to the urban grid map, the classified mobile signaling data, and the classified urban feature statistical data.

[0087] Step 603: Construct the classified grid feature vectors of each population category in each grid unit in each time period in the urban grid map according to the classified urban feature statistical data and the urban grid map.

[0088] Step 604: For any origin grid unit and any destination grid unit among multiple grid units, respectively input the corresponding classified grid feature vectors and the grid distance between the two grid units as the target input parameters into the pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the classified travel selection probability from the origin grid unit to the destination grid unit.

[0089] Step 605: Calculate the classified traffic travel prediction data of each population category in each grid unit in each time period based on the classified travel selection probability and the classified target travel volume of the origin grid unit.

[0090] In step 601 of some embodiments, for mobile phone signaling data and urban feature statistical data, detailed division is performed according to preset population categories such as gender, age group, occupation type, etc. and preset time periods such as weekdays, weekends, holidays, morning and evening rush hours, etc., so as to obtain the classified mobile phone signaling data and classified urban feature statistical data corresponding to each time period and each population category.

[0091] In step 602 of some embodiments, by combining the urban grid map, the classified mobile phone signaling data, and the classified urban feature statistical data, the classified target travel volume of each grid unit is determined for each time period and each population category. By independently modeling the travel behaviors of different populations and different time periods, the travel demand differences of various populations in different time periods can be more accurately reflected, and the fineness and pertinence of the prediction results can be improved.

[0092] In step 603 of some embodiments, based on the classified urban feature statistical data and the urban grid map, for each grid unit, each time period, and each population category, a corresponding classified grid feature vector is constructed. This feature vector synthesizes multi-dimensional information such as population characteristics, land use characteristics, traffic convenience characteristics, and industrial facility characteristics, and is customized for different populations and time periods, thus providing a more discriminative and representative feature description for model input.

[0093] In step 604 of some embodiments, for any pair of origin grid units and destination grid units, their corresponding classified grid feature vectors are respectively extracted, and the grid distance between the two is calculated, and these are used as target input parameters and input into a pre-trained deep gravity prediction model. This model can output the classified travel selection probability from the origin grid unit to the destination grid unit according to the feature differences of different populations and time periods, and realize the accurate description of the travel distribution law for different populations and different time periods.

[0094] In step 605 of some embodiments, by combining the classified target travel volume of the origin with the classified travel selection probability to each destination, the fine-grained prediction of the OD travel volume among different populations, different time periods, and different spatial units is realized.

[0095] In summary, through the above steps 602 to 605, the embodiments of the present application can fully mine and utilize multi-source heterogeneous data to realize the refined prediction of urban traffic demand for different populations, different time periods, and different spatial units. This method not only improves the spatial and temporal resolution of traffic demand prediction, but also enhances the ability to depict the travel behavior differences of different populations, thus providing more targeted decision-making support for application scenarios such as urban traffic planning and travel service optimization.

[0096] Please refer to Figure 7, an embodiment of the present application further provides an urban traffic demand prediction device, which can implement the above-mentioned urban traffic demand prediction method, including: An acquisition module, which acquires the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city; A determination module, which determines the target travel volume of each grid unit in the urban grid map according to the urban grid map, mobile phone signaling data, and urban feature statistical data; A construction module, which constructs a grid feature vector of each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map; A prediction module, for any origin grid unit and any destination grid unit among multiple grid units, respectively takes the corresponding grid feature vector and the grid distance between the two grid units as target input parameters, and inputs them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability from the origin grid unit to the destination grid unit; A calculation module, which calculates traffic travel prediction data based on the travel selection probability and the target travel volume of the origin grid unit.

[0097] In a third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the urban traffic demand prediction method according to any one of the embodiments in the first aspect of the present application.

[0098] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, the storage medium stores a program, and when the program is executed by a processor, it implements the urban traffic demand prediction method according to any one of the embodiments in the first aspect of the present application.

[0099] Please refer to Figure 8 , Figure 8 schematically shows the hardware structure of an electronic device in another embodiment. The electronic device includes: A processor 801, which can be implemented by using a general-purpose CPU (Central Processing Unit, central processor), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present application; The memory 802 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 802 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 802 and are called by the processor 801 to execute the urban traffic demand prediction method of the embodiments of this application; The input / output interface 803 is used to implement information input and output; The communication interface 804 is used to implement communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or through wireless means (such as mobile network, WIFI, Bluetooth, etc.); The bus 805 transmits information between the various components of the device (such as the processor 801, the memory 802, the input / output interface 803, and the communication interface 804); Among them, the processor 801, the memory 802, the input / output interface 803, and the communication interface 804 are communicatively connected to each other inside the device through the bus 805.

[0100] The embodiments of this application also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned urban traffic demand prediction method is implemented.

[0101] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory can include high-speed random access memory, and can also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes a memory remotely set relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0102] The embodiments described in the embodiments of this application are to more clearly illustrate the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are equally applicable to similar technical problems.

[0103] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figures, or combine certain steps, or different steps.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0105] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations.

[0106] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0107] It should be understood that in the present application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" may mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or a similar expression means any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c may mean: a, b, c, "a and b", "a and c", "b and c", or "a, b, and c", where a, b, and c can be single or multiple.

[0108] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the above division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical or other forms.

[0109] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0110] In addition, in each embodiment of the present application, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0111] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. And the aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical discs that can store programs.

[0112] The preferred embodiments of the embodiments of the present application have been described above with reference to the drawings, but this does not limit the scope of the rights of the embodiments of the present application. Any modification, equivalent replacement, and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of the rights of the embodiments of the present application.

Claims

1. A method for predicting urban traffic demand, characterized in that, The method includes: Obtaining the urban grid map, mobile phone signaling data, and urban feature statistical data of the target city; Determining the target travel volume of each grid unit in the urban grid map according to the urban grid map, the mobile phone signaling data, and the urban feature statistical data; Constructing a grid feature vector of each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map; For any origin grid unit and any destination grid unit among multiple grid units, respectively taking the corresponding grid feature vector and the grid distance between the two grid units as target input parameters and inputting them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability between the origin grid unit and the destination grid unit; Calculating traffic travel prediction data based on the travel selection probability and the target travel volume of the origin grid unit.

2. The urban traffic demand forecasting method according to claim 1, wherein The determining the target travel volume of each grid unit in the urban grid map according to the urban grid map, the mobile phone signaling data, and the urban feature statistical data includes: Determining the historical baseline travel rate of each grid unit according to the urban grid map and the mobile phone signaling data; Extracting multiple grid attribute variables of each grid unit from the urban feature statistical data; wherein, the multiple grid attribute variables include grid accessibility, land use mix, road traffic level, job-housing relationship coefficient, and grid housing price; Inputting the historical baseline travel rate and the multiple grid attribute variables into a pre-constructed travel rate regression model, so that the travel rate regression model outputs a target travel rate; Obtaining the population distribution data of each grid unit, and determining the target travel volume of each grid unit based on the target travel rate and the population distribution data.

3. The urban traffic demand forecasting method according to claim 1, wherein The constructing a grid feature vector of each grid unit in the urban grid map according to the urban feature statistical data and the urban grid map includes: For each grid unit, parsing the urban feature statistical data to obtain the population feature, land use feature, traffic convenience feature, and industrial facility feature corresponding to the grid unit; Constructing a grid feature vector of each grid unit according to the population feature, the land use feature, the traffic convenience feature, and the industrial facility feature.

4. The urban traffic demand forecasting method according to claim 1, wherein, The for any origin grid unit and any destination grid unit among multiple grid units, respectively taking the corresponding grid feature vector and the grid distance between the two grid units as target input parameters and inputting them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel selection probability between the origin grid unit and the destination grid unit includes: Respectively extracting the grid feature vectors of the origin grid unit and the destination grid unit as the first input parameter and the second input parameter; Calculate the grid distance between the starting grid cell and the destination grid cell, and use the grid distance as the third input parameter; Take the first input parameter, the second input parameter, and the third input parameter as the target input parameters and input them into the input layer of the depth gravity prediction model, so that the depth gravity prediction model performs feature learning and interaction processing on the first input parameter, the second input parameter, and the third input parameter through a neural network, and generates an internal representation of the interaction relationship between the starting grid cell and the destination grid cell; The depth gravity prediction model calculates and outputs the travel selection probability through the activation function of the output layer.

5. The urban traffic demand prediction method according to claim 4, wherein The pre-training process of the depth gravity prediction model includes: Based on the mobile phone signaling data, construct a training data set. The training data set contains multiple groups of training samples. Each training sample includes: a starting grid cell feature vector and a destination grid cell feature vector corresponding to the first input parameter and the second input parameter respectively, a grid distance corresponding to the third input parameter, and a target label determined according to the mobile phone signaling data and representing the actual travel relationship between the starting grid cell and the destination grid cell; Input the starting grid cell feature vector, the destination grid cell feature vector, and the grid distance in the training sample into the depth gravity prediction model to obtain a model prediction result; Based on the model prediction result and the target label, calculate the prediction error using a predefined loss function; Based on a preset optimization algorithm, according to the prediction error, iteratively update the model parameters of the depth gravity prediction model until a preset convergence condition is met.

6. The urban traffic demand forecasting method according to claim 1, wherein The calculation of the traffic travel prediction data based on the travel selection probability and the target travel volume of the starting grid cell includes: For any starting grid cell and any destination grid cell in the urban grid map, multiply the target travel volume of the starting grid cell by the travel selection probability between the starting grid cell and the destination grid cell to obtain the predicted travel volume between the starting grid cell and the destination grid cell.

7. The urban traffic demand prediction method according to claim 1, characterized in that The method further includes: Divide the mobile phone signaling data and the urban feature statistical data according to a preset population category and a preset time period to obtain classified mobile phone signaling data and classified urban feature statistical data corresponding to each population category in each time period; According to the urban grid map, the classified mobile phone signaling data, and the classified urban feature statistical data, determine the classified target travel volume of each grid cell for each population category in each time period; According to the classified urban feature statistical data and the urban grid map, construct a classified grid feature vector for each grid cell for each population category in each time period in the urban grid map; For any departure grid cell and any destination grid cell among the multiple grid cells, the corresponding classified grid feature vector and the grid distance between the two grid cells are respectively used as target input parameters and input into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the classified travel choice probability from the departure grid cell to the destination grid cell; Based on the classified travel choice probability and the classified target travel volume of the departure grid cell, the classified traffic travel prediction data for each grid cell, each time period, and each population category is calculated.

8. An urban traffic demand prediction device, characterized in that, Including: An acquisition module that acquires the urban grid map, mobile signaling data, and urban feature statistical data of the target city; A determination module that determines the target travel volume of each grid cell in the urban grid map according to the urban grid map, the mobile signaling data, and the urban feature statistical data; A construction module that constructs the grid feature vector of each grid cell in the urban grid map according to the urban feature statistical data and the urban grid map; A prediction module that, for any departure grid cell and any destination grid cell among the multiple grid cells, respectively uses the corresponding grid feature vector and the grid distance between the two grid cells as target input parameters and inputs them into a pre-trained deep gravity prediction model, so that the deep gravity prediction model outputs the travel choice probability from the departure grid cell to the destination grid cell; A calculation module that calculates traffic travel prediction data based on the travel choice probability and the target travel volume of the departure grid cell.

9. An electronic device, characterized in that, Including: A memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements the urban traffic demand prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program, and when the program is executed by the processor, it implements the urban traffic demand prediction method according to any one of claims 1 to 7.

Images