Method for optimizing urban commuting pattern by using artificial intelligence algorithm

Abstract

The invention relates to the technical field of urban governance, in particular to a method for optimizing an urban commuting pattern by using an artificial intelligence algorithm, which comprises the following steps of: identifying urban employment center data according to mobile phone signaling data, and summarizing and preprocessing the data; utilizing a Gaussian mixture model to identify space relative distribution characteristics and influence weights of the employment places and the employment centers; identifying a space relative change rule of a residence place and an employment place through an XGBoost model, and analyzing a job-residence relationship; solving the gravity model by using an XGBoost model to identify a commuting rule between the residence place and the employment place so as to obtain a commuting structure; simulating employment center adjustment through an intelligent optimization algorithm, and predicting changes of an employment place, a residence place and a commuting flow to generate a commuting pattern; the commuting pattern is evaluated by collecting actual commuting data and user feedback. The method can provide data support and scientific guidance for urban public facility pattern, urban quality improvement and urban fine governance.

Description

Technical Field

[0001] The present invention relates to the fields of geospatial artificial intelligence and urban governance technology, and in particular to a method for optimizing urban commuting patterns using an artificial intelligence algorithm. Background Art

[0002] Carbon emissions from urban commuting are an important part of urban carbon emissions. Optimizing the urban commuting pattern plays an important role in reducing urban commuting carbon emissions and improving urban sustainable development capabilities and environmental friendliness. The urban work-residence spatial structure refers to the spatial distribution characteristics of the residence and employment locations of employed people in the city and the population mobility between residence and employment locations. The work-residence spatial pattern is the main determinant of the commuting pattern. Optimizing the urban commuting pattern by adjusting the work-residence spatial pattern is a core task in urban governance and one of the core means to reduce urban commuting carbon emissions.

[0003] The rapid emergence of spatiotemporal big data and geographic artificial intelligence technologies in recent years has provided new ways to solve the above problems. New big data based on card swiping data, mobile phone signaling data, LBS social data, etc. have been widely used in the research of work-residence space.

[0004] In terms of the intrinsic relationship between urban spatial structure and urban commuting pattern, a considerable number of studies have measured the characteristics of urban spatial structure and urban commuting pattern respectively. The former explored the identification of diverse urban employment and residential spatial structures, while the latter focused on the evaluation of commuting distance, mode and demand, as well as commuting efficiency. In terms of the relationship between the two, by capturing the changing characteristics of urban residents' commuting, the relationship between urban commuting patterns and job-residence space can be obtained, and the changes can be effectively simulated through proxy models. In addition, there are studies that point out that residents who have changed their employment locations have the characteristics of "self-balancing" commuting, which can be achieved by optimizing local employment patterns or living units, thereby promoting local "relative balance between work and residence" and gradually promoting the optimization of the overall spatial pattern.

[0005] On the other hand, existing research is still relatively weak in establishing the exact relationship between urban spatial structure and commuting pattern, resulting in a lack of solid and reliable empirical evidence and quantitative support for alleviating commuting problems by optimizing spatial structure.

[0006] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0007] In response to the problems in the related technology, the present invention proposes a method for optimizing urban commuting patterns using an artificial intelligence algorithm to overcome the above-mentioned technical problems existing in the existing related technology.

[0008] To this end, the specific technical solution adopted by the present invention is as follows: A method for optimizing urban commuting patterns using an artificial intelligence algorithm, the optimization method comprising the following steps: S1. Identify urban employment center data based on mobile phone signaling data and pre-process the data aggregation; S2. Use Gaussian mixture model to identify the spatial relative distribution characteristics and influence weights of employment locations and employment centers; S3, using the XGBoost model to identify the spatial relative change patterns between residence and employment location and analyze the relationship between work and residence; S4, using XGBoost model to solve the gravity model to identify the commuting pattern between residence and employment to obtain the commuting structure; S5. Use intelligent optimization algorithms to simulate the adjustment of employment centers, predict changes in employment locations, residence locations, and commuting flows to generate commuting patterns; S6. Evaluate commuting patterns by collecting actual commuting data and user feedback.

[0009] Furthermore, the identifying of the city employment center data according to the mobile phone signaling data and pre-processing the data aggregation includes the following steps: S11. Identify high-precision grid data in mobile phone signaling data, including the grid center coordinates of the commuting departure and destination, the grid size, the commuting scale corresponding to the grid, and the permanent population size; S12, according to the commuting scale of the grid center coordinates, select the grid center with the largest commuting flow as the initial point, and enter the search radius required by the initial point center; S13, the point with the largest cumulative number of commuting flows outside the search radius as the center of the employment location; S14. Draw an area with a radius of the initial point with the employment center as the center, search for the next employment center outside the area, and repeat the process until all qualified employment centers are identified to form an employment center point set.

[0010] Furthermore, the use of the Gaussian mixture model to identify the spatial relative distribution characteristics of employment locations and employment centers and their impact weights includes the following steps: S21, generating a random point of commuting volume within the grid center coordinates of the destination according to the cumulative number of commuting flows in each group of grid centers; S22. Use Gaussian mixture models to identify the association between employment centers and job distribution and employment population distribution; S23. Use the head-tail split method to quantify the weight of each employment center.

[0011] Furthermore, the Gaussian mixture model used to identify the relationship between employment centers and job distribution and employment population distribution is:

[0012] In the formula, N represents the total number of working population in the city; represents the influence of employment centers in the mixed model; Indicates the employment location in the city x the number of commuters employed; k Represents the total number of elements in the employment center point set in the city; represents a two-dimensional Gaussian density function; Among them, the expression of the two-dimensional Gaussian density function is:

[0013] In the formula, Indicates k The geographical coordinates of each employment centerline point; Indicates the rate at which the influence of employment centers decays.

[0014] Furthermore, the use of the head-tail segmentation method to quantify the weight of each employment center includes the following steps: S231, calculating the average value of all the values ​​obtained in the above steps, considering the part greater than or equal to the average value as the head, and considering the part lower than the average value as the tail, wherein the head-tail index of the tail is 1, and the head-tail index of the head is 2; S232, repeat the above process for the data located at the head after segmentation, and assign head and tail indices of 2 and 3 to the tail data and head data obtained by re-identification respectively; S233. Repeat step S232 until the required number of grades is reached, and the weight of the employment center is represented according to the head-tail index. The larger the head-tail index, the higher the grade.

[0015] Furthermore, the analysis of the relationship between residence and employment by identifying the spatial relative change law of residence and employment through the XGBoost model includes the following steps: S31. Input the residential and employment distribution data of the commuting population in the two periods, and integrate the similarity weights of the residential and employment locations to construct a homogeneous relationship network; S32. Use homogeneous relationship network representation learning to automatically extract the characteristics of the change rules of residence and employment location; S33, using the extracted features as input points of the XGBoost model, and iteratively training to obtain the best analysis model; S34. By analyzing the feature importance of the XGBoost model, we can understand the relative spatial changes between residence and employment.

[0016] Furthermore, the method of using the XGBoost model to solve the gravity model to identify the commuting pattern between the place of residence and the place of employment to obtain the commuting structure includes the following steps: S41. Collect and input the network coordinates of the place of residence and the place of employment, the commuting scale and the population scale, where the commuting scale is the observed value of the commuting flow and the population scale is the observed value of the permanent population; S42. Using XGBoost to solve the gravity model principle, a relationship model between population size and the distance between residence and employment place is established; S43, using the collected data to estimate the parameters in the gravity model by the least square method, and using the cross-validation method to evaluate the gravity model; S44. Based on the established gravity model, the number of commuting population is predicted through the model, and the commuting patterns between residence and employment place are analyzed.

[0017] Furthermore, the method of simulating the adjustment of employment centers through intelligent optimization algorithms and predicting the changes in employment locations, residence locations, and commuting flows to generate commuting patterns includes the following steps: S51, establishing employment centers based on the spatial relative distribution characteristics, job-residence relationship and commuting structure obtained in the above steps; S52. Adjust the employment pattern according to the optimization objectives and constraints, and calculate the commuting cost function; S53. Use the adaptive differential evolution optimization algorithm to continue adjusting the employment structure for iterative optimization; S54. Repeat the above steps until the optimization target is achieved and the employment structure is generated.

[0018] Furthermore, the adjustment of the employment structure according to the optimization objectives and constraints and the calculation of the commuting cost function include the following steps: S521. Adjust the weight of employment centers according to adjustment targets and data analysis; S522. Recalculate the employment location distribution based on the spatial relative distribution characteristics and influence weights of employment locations and employment centers; S523. Recalculate the distribution of places of residence based on the relative spatial changes between places of residence and places of employment; S524. Recalculate the commuting flow distribution based on the commuting pattern between the place of residence and the place of employment; S525, recalculating the average commuting distance according to the calculation results obtained in the above steps; S526. Evaluate the degree of optimization of the adjusted commuting distance.

[0019] Furthermore, the use of the adaptive differential evolution optimization algorithm to continue adjusting the employment structure for iterative optimization includes the following steps: S531, selecting two individuals from the parent generation individuals to perform vector difference to generate a differential vector; S532, selecting another individual and summing the difference vector to generate an experimental individual; S533, performing a crossover operation on the parent individual and the corresponding experimental individual to generate a new offspring individual; S534. Perform a selection operation between the parent-generation individuals and the offspring-generation individuals, and save the individuals that meet the requirements into the next generation population.

[0020] The beneficial effects of the present invention are: 1. The present invention provides a method for optimizing the urban employment-residence spatial structure by considering the current characteristics of employment-residence centers and commuting patterns, and explores the derivation technology of the optimal urban employment spatial pattern based on an intelligent optimization algorithm, which can provide data support and scientific guidance for the urban public facilities pattern, urban quality improvement, and urban fine governance.

[0021] 2. The present invention uses the XGBoost model to identify the relative spatial change patterns between residence and employment. The XGBoost model can identify the relative spatial change patterns between residence and employment by analyzing a large amount of data, helping urban planning and transportation planning departments to better understand the patterns of population mobility. At the same time, by analyzing the spatial changes in the relationship between work and residence, it can help urban planning and transportation planning departments to better understand people’s travel needs and commuting patterns.

[0022] 3. The present invention artificially uses the XGBoost model to solve the gravity model to identify the commuting patterns between the place of residence and the place of employment. By analyzing the commuting patterns between the place of residence and the place of employment, it is possible to understand people’s commuting methods, distances, time and other related information, which helps to reveal the commuting patterns between different areas in the city. At the same time, the analysis results of commuting patterns can help transportation planning departments better understand people’s commuting needs and travel patterns, optimize the layout of the city’s transportation network, public transportation routes and transportation facilities, improve transportation efficiency and convenience, help reduce traffic congestion, alleviate commuting pressure, and improve urban traffic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1 is a flow chart of a method for optimizing urban commuting patterns using an artificial intelligence algorithm according to an embodiment of the present invention; Figure 2It is a general technical roadmap of a method for optimizing urban commuting patterns using an artificial intelligence algorithm according to an embodiment of the present invention; Figure 3 It is a schematic diagram of the working principle of an adaptive differential evolution algorithm in a method for optimizing urban commuting patterns using an artificial intelligence algorithm according to an embodiment of the present invention. Implementation

[0025] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0026] According to an embodiment of the present invention, a method for optimizing urban commuting patterns using an artificial intelligence algorithm is provided.

[0027] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figures 1 - 3 As shown, according to the method for optimizing the urban commuting pattern using an artificial intelligence algorithm according to an embodiment of the present invention, the optimization method includes the following steps: S1. Identify urban employment center data based on mobile phone signaling data and pre-process the data aggregation.

[0028] Specifically, the mobile phone signaling data is high-precision grid data, which includes the grid center coordinates of the commuting departure and destination. , the grid size is , the commuting scale corresponding to the grid is Flow The size of the permanent population is Pop。

[0029] In one embodiment, the identifying of city employment center data based on mobile phone signaling data and pre-processing of data aggregation includes the following steps: S11. Identify high-precision grid data in mobile phone signaling data, including the grid center coordinates of the commuting departure and destination, the grid size, the commuting scale corresponding to the grid, and the permanent population size; S12, according to the grid center coordinates The size of the commute Flow , select the grid center with the largest commuting flow as the initial point LMP 0 , enter the desired search radius from the initial point center R 0 ; S13, search for initial point LMP 0 Center, search radius R0 The point with the largest cumulative number of commuting flows is regarded as the employment center. LMP 1 ; S14, respectively based on the employment center LMP 1 Draw the initial point radius for the center R 0 Search for the next employment center outside this area LMP 2 , and iterates in sequence until all eligible employment centers are identified to form a set of employment center points.

[0030] Specifically, the employment location center point set is expressed as .

[0031] S2. Use Gaussian mixture model to identify the spatial relative distribution characteristics and influence weights of employment locations and employment centers.

[0032] Specifically, the employment center extraction result of step S1 is input, and the distribution characteristics of the number of employees in each employment location are identified through the Gaussian mixture model, and the influence weight of the employment center is quantified.

[0033] In one embodiment, the method of using a Gaussian mixture model to identify the spatial relative distribution characteristics of employment locations and employment centers and their influence weights includes the following steps: S21, according to each group of grid centers Cumulative number of commuting flows Flow , generate random points of commuting volume within the grid center coordinates of the destination; Specifically, a random point of commuting volume is generated within the grid center coordinates of the destination. arrive Generate within the range Random points.

[0034] S22. Use Gaussian mixture model to identify the association between employment centers and job distribution and employment population distribution.

[0035] The Gaussian mixture model used to identify the relationship between employment centers and job positions and employment population distribution is:

[0036] In the formula, N represents the total number of working population in the city; It represents the influence of employment center in the mixed model, which is the weight, also known as the mixing coefficient; Indicates the employment location in the city x the number of commuters employed; k Represents the set of employment centers in the city LMP The total number of elements in it; represents a two-dimensional Gaussian density function; Among them, the expression of the two-dimensional Gaussian density function is:

[0037] In the formula, Indicates k Geographical coordinates of each employment centerline point ; It represents the decay rate of the influence of employment centers, which is a two-dimensional covariance matrix.

[0038] Considering that the attenuation characteristics of the city center influence are universal and stable, all Gaussian mixture models are set to share a .

[0039] Its expression is:

[0040] and The solution is usually obtained by using expectation maximization ( EM )algorithm.

[0041] in, is the covariance, which indicates the direction of linear correlation of variables.

[0042] Its expression is:

[0043] S23. Use the head-tail split method to quantify the weight of each employment center.

[0044] In one embodiment, the quantifying the weight of each employment center using the head-tail segmentation method comprises the following steps: S231, calculate the average value of all the values ​​obtained in the above steps, regard the part greater than or equal to the average value as the head, and regard the part lower than the average value as the tail, where the head-tail index of the tail is HT i 1, the head and tail index of the head HT i is 2; S232, repeat the above process for the data at the head after segmentation, and assign head and tail indices to the tail data and head data obtained by re-identification respectively HT i For 2 and 3; S233, repeat step S232 until the number of classifications required is reached or the head data cannot be further subdivided, and then HTi Represents the weight of employment centers, head and tail index HT i The bigger it is, the higher the level.

[0045] S3. Use the XGBoost model to identify the spatial relative change patterns between residence and employment location and analyze the relationship between work and residence.

[0046] Specifically, XGBoost (eXtreme Gradient Boosting) is a machine learning model based on gradient boosting decision trees. It is a powerful model currently widely used in data science competitions and practical applications.

[0047] XGBoost uses a gradient boosting algorithm, which improves the model's predictive power and generalization performance by combining multiple weak learners (usually decision trees) into a strong learner. At the same time, the XGBoost model adds a regularization term to the objective function to control the complexity of the model and avoid overfitting. The complexity and generalization of the model are controlled by adjusting the regularization parameters. XGBoost can help understand and interpret data by analyzing the importance of each feature in the model. It can evaluate the importance of features by calculating indicators such as feature coverage, gain, and weight, thereby helping with feature selection and feature engineering.

[0048] The residential and employment distribution data of the commuting population in two periods were input, and the XGBosst model was used to fit the impact of job changes on residential patterns.

[0049] In one embodiment, the method of analyzing the relationship between residence and employment by identifying the spatial relative change law between residence and employment through the XGBoost model includes the following steps: S31. Input the residential and employment distribution data of the commuting population in the two periods, and integrate the similarity weights of the residential and employment locations to construct a homogeneous relationship network; S32. Use homogeneous relationship network representation learning to automatically extract the characteristics of the change rules of residence and employment location; S33, using the extracted features as input points of the XGBoost model, and iteratively training to obtain the best analysis model; S34. By analyzing the feature importance of the XGBoost model, we can understand the relative spatial changes between residence and employment.

[0050] The assumptions in the XGBoost model are T A decision tree, whose model equation is:

[0051] in Indicates jA regression tree function.

[0052] In the model t The objective function of a tree is:

[0053] Where Ω represents the complexity of a regression tree, which needs to be t The complexity of the regression trees is summed up: Among them, the hyperparameters and We can control the intensity of our punishments; T Indicates the number of leaf nodes of the current regression tree; Represents the sum of the squares of the values ​​of each leaf node, called L2 Regularization term.

[0054] When used to solve the correspondence between residence and employment, XGBoost is used to calculate the time when living in the grid i The working population on the graph is:

[0055] in, and Respectively t and t-1 Time Grid i The number of jobs in and Respectively Corresponding to and is the local coefficient of geographic coordinates; express t-1 Time Grid i The number of working people living on the represents the corresponding local coefficient; represents the intercept; Represents the residual term of the model.

[0056] The present invention takes into account that the distribution of residence and work place has spatial autocorrelation, and there is also spatial autocorrelation between the two, so the geographic coordinate variable is introduced Be reflected.

[0057] S4. Use the XGBoost model to solve the gravity model to identify the commuting patterns between residence and employment to obtain the commuting structure.

[0058] Input the network center coordinates, commuting scale and population size of the departure place and the employment place, and use the XGBoost described in step S3 to solve the gravity model, where the commuting scale is the observed value of the commuting flow, and the population size is the observed value of the permanent population.

[0059] In one embodiment, the method of using the XGBoost model to solve the gravity model to identify the commuting pattern between the place of residence and the place of employment to obtain the commuting structure includes the following steps: S41. Collect and input the network coordinates of the place of residence and the place of employment, the commuting scale and the population scale, where the commuting scale is the observed value of the commuting flow and the population scale is the observed value of the permanent population; S42. Using XGBoost to solve the gravity model principle, a relationship model between population size and the distance between residence and employment place is established; S43, using the collected data to estimate the parameters in the gravity model by the least square method, and using the cross-validation method to evaluate the gravity model; S44. Based on the established gravity model, the number of commuting population is predicted through the model, and the commuting patterns between residence and employment place are analyzed.

[0060] Specifically, the expression of the gravity model is:

[0061] in, represents the size of the commuting population from the origin grid to the destination grid; k represents a constant coefficient; and Respectively i and j the number of employed people living and working in the country; It represents the distance between the number of residences and the number of jobs of the employed population; and denote the elasticity of the population size of the origin and destination, respectively; It represents the measure of distance decay effect; represents the error term.

[0062] S5. Use intelligent optimization algorithms to simulate employment center adjustments, predict changes in employment locations, residences, and commuting flows, and generate commuting patterns.

[0063] Simulate the adjustment of employment centers and predict changes in employment locations, residence locations, and commuting flows. Use intelligent optimization algorithms to optimize commuting costs, substitute the results of steps S2, S3, and S4, and find the most appropriate adjustment plan for the spatial structure of employment and residence as the basis for optimizing the urban commuting pattern.

[0064] In one embodiment, the method of simulating the adjustment of employment centers through intelligent optimization algorithms, predicting the changes in employment locations, residence locations, and commuting flows to generate commuting patterns includes the following steps: S51, establishing employment centers based on the spatial relative distribution characteristics, job-residence relationship and commuting structure obtained in the above steps; S52. Adjust the employment structure according to the optimization objectives and constraints, and calculate the commuting cost function.

[0065] Specifically, the adjustment of the employment structure according to the optimization objectives and constraints and the calculation of the commuting cost function include the following steps: S521. Adjust the weight of employment centers according to adjustment targets and data analysis; S522. Recalculate the employment location distribution based on the spatial relative distribution characteristics and influence weights of employment locations and employment centers; S523. Recalculate the distribution of places of residence based on the relative spatial changes between places of residence and places of employment; S524. Recalculate the commuting flow distribution based on the commuting pattern between the place of residence and the place of employment; S525, recalculating the average commuting distance according to the calculation results obtained in the above steps; S526. Evaluate the degree of optimization of the adjusted commuting distance.

[0066] Specifically, the commuting cost function for evaluating the degree of optimization is:

[0067] in, represents the adjusted average commuting distance; C The simulated commuting distance representing the current situation; It represents the benefit of the optimization strategy, that is, the reduction in the average commuting distance. The smaller the value, the greater the reduction. It represents the regularization term, which represents the cost of the optimization strategy, that is, the range of change in the urban employment and residential pattern. The smaller the value, the closer the optimization result is to the status quo, and the smaller the economic and social cost required to achieve structural optimization.

[0068] The calculation formula is:

[0069] in, represents the regularization coefficient, and its size is adjusted to balance the relationship between the benefits of commuting distance compression and the costs of structural changes; and Represents the center before and after optimization j The weight of .

[0070] S53. Use the adaptive differential evolution optimization algorithm to continue adjusting the employment structure for iterative optimization.

[0071] Specifically, the use of the adaptive differential evolution optimization algorithm to continue adjusting the employment structure for iterative optimization includes the following steps: S531, selecting two individuals from the parent generation individuals to perform vector difference to generate a differential vector; S532, selecting another individual and summing the difference vector to generate an experimental individual; S533, performing a crossover operation on the parent individual and the corresponding experimental individual to generate a new offspring individual; S534. Perform a selection operation between the parent-generation individuals and the offspring-generation individuals, and save the individuals that meet the requirements into the next generation population.

[0072] S54. Repeat the above steps until the optimization target is achieved and the employment structure is generated.

[0073] S6. Evaluate commuting patterns by collecting actual commuting data and user feedback.

[0074] In one embodiment, the step of evaluating the commuting pattern by collecting actual commuting data and user feedback comprises the following steps: Collect relevant commuting data, including commuting time, commuting mode, commuting distance, traffic congestion, etc., and obtain data through questionnaires, traffic surveys, traffic monitoring systems, etc.

[0075] Analyze the collected commuting data, and use statistical methods, data visualization and other means to analyze indicators such as commuting time, commuting distance, and traffic congestion level to identify problems and room for improvement.

[0076] Collect user feedback on commuting patterns through questionnaires and interviews to understand users’ satisfaction and dissatisfaction with commuting time, traffic congestion, transportation selection, etc., as well as their suggestions and opinions.

[0077] Comprehensively analyze commuting data and user feedback, evaluate the commuting pattern, compare commuting conditions in different regions and time periods, identify problems and areas for improvement, and comprehensively consider user needs and transportation planning requirements.

[0078] Based on the evaluation results, suggestions are put forward to improve the commuting pattern, traffic improvement plans are proposed for traffic congestion points, the transportation network layout is optimized, and public transportation services are improved.

[0079] Based on the improvement suggestions put forward, an implementation plan is formulated and the improvement measures are gradually implemented.

[0080] After the improvement measures are implemented, the effects will be monitored and evaluated. By collecting commuting data and user feedback again, comparing the differences before and after the improvement, and evaluating the effectiveness and feasibility of the improvement measures.

[0081] In summary, with the help of the above technical solutions of the present invention, the present invention provides a method for optimizing the urban employment-residence spatial structure by considering the current characteristics of the employment-residence center and the commuting rules, and explores the derivation technology of the optimal urban employment spatial pattern based on the intelligent optimization algorithm, which can provide data support and scientific guidance for the urban public facilities pattern, urban quality improvement, and urban fine governance. The present invention identifies the spatial relative change law between the residence and the employment place through the XGBoost model. The XGBoost model can identify the spatial relative change law between the residence and the employment place by analyzing a large amount of data, helping urban planning and transportation planning departments to better understand the law of population mobility, and at the same time, by analyzing the spatial changes in the employment-residence relationship, it can help urban planning and transportation planning departments better understand people’s travel needs and commuting patterns. The present invention uses the XGBoost model to solve the gravity model to identify the commuting patterns between the place of residence and the place of employment. By analyzing the commuting patterns between the place of residence and the place of employment, people's commuting methods, distances, time and other related information can be understood, which helps to reveal the commuting patterns between different areas in the city. At the same time, the analysis results of commuting patterns can help transportation planning departments better understand people's commuting needs and travel patterns, optimize the layout of the city's transportation network, public transportation routes and transportation facilities, improve transportation efficiency and convenience, help reduce traffic congestion, alleviate commuting pressure, and improve urban traffic conditions.

[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing urban commuting patterns using an artificial intelligence algorithm, characterized in that: The optimization method includes the following steps: S1. Identify urban employment center data based on mobile phone signaling data and pre-process the data aggregation; S2. Use Gaussian mixture model to identify the spatial relative distribution characteristics and influence weights of employment locations and employment centers; S3, using the XGBoost model to identify the spatial relative change patterns between residence and employment location and analyze the relationship between work and residence; S4, using XGBoost model to solve the gravity model to identify the commuting pattern between residence and employment to obtain the commuting structure; S5. Use intelligent optimization algorithms to simulate the adjustment of employment centers, predict changes in employment locations, residence locations, and commuting flows to generate commuting patterns; S6. Evaluate commuting patterns by collecting actual commuting data and user feedback.

2. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 1, characterized in that: The method of identifying the urban employment center data according to the mobile phone signaling data and pre-processing the data aggregation includes the following steps: S11. Identify high-precision grid data in mobile phone signaling data, including the grid center coordinates of the commuting departure and destination, the grid size, the commuting scale corresponding to the grid, and the permanent population size; S12, according to the commuting scale of the grid center coordinates, select the grid center with the largest commuting flow as the initial point, and enter the search radius required by the initial point center; S13, the point with the largest cumulative number of commuting flows outside the search radius as the center of the employment location; S14. Draw an area with a radius of the initial point with the employment center as the center, search for the next employment center outside the area, and repeat the process until all qualified employment centers are identified to form an employment center point set.

3. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 1, characterized in that: The method of using the Gaussian mixture model to identify the spatial relative distribution characteristics of employment locations and employment centers and their impact weights includes the following steps: S21, generating a random point of commuting volume within the grid center coordinates of the destination according to the cumulative number of commuting flows in each group of grid centers; S22. Use Gaussian mixture models to identify the association between employment centers and job distribution and employment population distribution; S23. Use the head-tail split method to quantify the weight of each employment center.

4. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 3, characterized in that: The Gaussian mixture model used to identify the relationship between employment centers and job distribution and employment population distribution is: In the formula, N represents the total number of urban working population; represents the influence of employment centers in the mixed model; Indicates the employment location in the city x of commuters employed; k Represents the total number of elements in the employment center point set in the city; represents the two-dimensional Gaussian density function; the expression of the two-dimensional Gaussian density function is: In the formula, Indicates k The geographical coordinates of each employment centerline point; Indicates the rate at which the influence of employment centers decays.

5. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 4, characterized in that: The method of quantifying the weight of each employment center using the head-tail segmentation method comprises the following steps: S231, calculating the average value of all the values ​​obtained in the above steps, considering the part greater than or equal to the average value as the head, and considering the part lower than the average value as the tail, wherein the head-tail index of the tail is 1, and the head-tail index of the head is 2; S232, repeat the above process for the data located at the head after segmentation, and assign head and tail indices of 2 and 3 to the tail data and head data obtained by re-identification respectively; S233. Repeat step S232 until the required number of grades is reached, and the weight of the employment center is represented according to the head-tail index. The larger the head-tail index, the higher the grade.

6. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 1, characterized in that: The method of analyzing the relationship between residence and work by identifying the spatial relative change law of residence and work place through the XGBoost model includes the following steps: S31. Input the residential and employment distribution data of the commuting population in the two periods, and integrate the similarity weights of the residential and employment locations to construct a homogeneous relationship network; S32. Use homogeneous relationship network representation learning to automatically extract the characteristics of the change rules of residence and employment location; S33, using the extracted features as input points of the XGBoost model, and iteratively training to obtain the best analysis model; S34. By analyzing the feature importance of the XGBoost model, we can understand the relative spatial changes between residence and employment.

7. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 1, characterized in that: The method of using the XGBoost model to solve the gravity model to identify the commuting rules between the place of residence and the place of employment to obtain the commuting structure includes the following steps: S41. Collect and input the network coordinates of the place of residence and the place of employment, the commuting scale and the population scale, where the commuting scale is the observed value of the commuting flow and the population scale is the observed value of the permanent population; S42. Using XGBoost to solve the gravity model principle, a relationship model between population size and the distance between residence and employment place is established; S43, using the collected data to estimate the parameters in the gravity model by the least squares method, and using the cross-validation method to evaluate the gravity model; S44. Based on the established gravity model, the number of commuting population is predicted through the model, and the commuting patterns between residence and employment place are analyzed.

8. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 1, characterized in that: The method of simulating the adjustment of employment centers through intelligent optimization algorithms, predicting the changes in employment locations, residence locations, and commuting flows, and generating commuting patterns includes the following steps: S51, establishing employment centers based on the spatial relative distribution characteristics, job-residence relationship and commuting structure obtained in the above steps; S52. Adjust the employment pattern according to the optimization objectives and constraints, and calculate the commuting cost function; S53. Use the adaptive differential evolution optimization algorithm to continue adjusting the employment structure for iterative optimization; S54. Repeat the above steps until the optimization target is achieved and the employment structure is generated.

9. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 8, characterized in that: The adjustment of the employment pattern according to the optimization objectives and constraints and the calculation of the commuting cost function include the following steps: S521. Adjust the weight of employment centers according to adjustment targets and data analysis; S522. Recalculate the employment location distribution based on the spatial relative distribution characteristics and influence weights of employment locations and employment centers; S523. Recalculate the distribution of places of residence based on the relative spatial changes between places of residence and places of employment; S524. Recalculate the commuting flow distribution based on the commuting pattern between the place of residence and the place of employment; S525, recalculating the average commuting distance according to the calculation results obtained in the above steps; S526. Evaluate the degree of optimization of the adjusted commuting distance.

10. The method for optimizing urban commuting patterns using artificial intelligence algorithms according to claim 8, characterized in that: The method of using the adaptive differential evolution optimization algorithm to continuously adjust the employment structure for iterative optimization includes the following steps: S531, selecting two individuals from the parent generation individuals to perform vector difference to generate a differential vector; S532, selecting another individual and summing the difference vector to generate an experimental individual; S533, performing a crossover operation on the parent individual and the corresponding experimental individual to generate a new offspring individual; S534. Perform a selection operation between the parent-generation individuals and the offspring-generation individuals, and save the individuals that meet the requirements into the next generation population.

Images