A method for analyzing passenger behavior on digital platforms and pushing personalized services

Abstract

The present invention relates to a method for analyzing passenger behavior on a digital platform and pushing personalized services, the method comprising: L1. When a passenger enters a digital platform, data information on the passenger's historical travel trajectory is obtained, image data information on the passenger is obtained in real time based on a built-in camera on the platform, data information on the passenger's position coordinate sequence is obtained in real time based on a built-in infrared sensor on the platform, data information on the pressure change time series of the platform is obtained in real time based on a built-in pressure sensor on the platform, a skeletal key point detection model based on a dynamic learning factor is constructed, and a limb motion feature matrix of the passenger is characterized to obtain data information on the limb motion feature matrix of the passenger. The present invention can not only accurately collect and analyze information such as the behavior, trajectory, and needs of passengers and pedestrians at the platform, but also inform the passenger of vehicle congestion in advance and push personalized ride recommendations and other services, thereby improving the operational efficiency, service quality, and travel experience of unmanned public transportation.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned public transportation, and in particular to a method for analyzing passenger behavior at a digital platform and pushing personalized services. Background Art

[0002] With the rapid development of technology, smart buses, as a key component of intelligent transportation, remain a major component of urban transportation. However, the current bus system still has many shortcomings in terms of passenger service. Traditional bus stops have a single function, mainly providing passengers with simple route information. They are unable to effectively analyze passenger behavior and make it difficult to provide personalized services based on individual passenger needs.

[0003] Therefore, under the current bus operation model, buses are unable to accurately grasp passengers' travel habits and waiting preferences. For example, passenger behavior characteristics such as waiting time at the platform, whether they carry luggage, and whether they have companions are not fully utilized. This leads to a lack of targeted bus operation and scheduling, making it impossible to rationally allocate vehicle resources according to passengers' actual needs, thus affecting the passenger travel experience and reducing the overall operational efficiency of the bus system.

[0004] At the same time, due to the lack of a personalized service push mechanism, passengers on public transportation are unable to obtain information relevant to their needs, such as surrounding scene recommendations, real-time traffic alerts, and bus transfer suggestions. This has led to a significant gap between the public transportation system's service diversity and user stickiness and the expectations of modern passengers.

[0005] Prior art, a Chinese patent (Application Number: 202411972313.3, Publication Number: CN 119848349A) discloses a smart bus shelter data processing method and system based on digital analysis. This method involves collecting seat occupancy information and air conditioning usage status within a target vehicle, calculating temperature distribution and congestion information for each seat, and transmitting this information to all bus shelters for data sharing. The system then receives a target passenger's request, identifies the target passenger's historical travel history and current boarding demand, and predicts their preferred behavior and the expected distance from the desired seat to the exit. A seat recommendation model is constructed based on the temperature distribution and congestion information, along with their preferred behavior and expected distance, to generate a target seat for recommendation. Once the target passenger confirms their use of the target seat, the data is updated and synchronized to all bus shelters, allowing seat recommendations to be made for the next passenger. This solution relies on device accuracy: the collection of data such as seat occupancy and air conditioning usage depends on the accuracy of the on-board equipment. Equipment failures or insufficient accuracy can lead to inaccurate calculations of temperature distribution and congestion information, impacting the accuracy of seat recommendations.

[0006] Limitations of Passenger Behavior Prediction: While passenger preferences are predicted based on historical ride records and current demand, passenger travel needs can change temporarily due to special circumstances. Predictions based solely on partial historical and current information cannot fully capture the complex and ever-changing realities of actual travel, leading to discrepancies between recommended seats and actual passenger needs. System Complexity and Maintenance Cost: This system involves complex operations such as data exchange between vehicles and multiple bus shelters, multiple information analyses, and model building. The system requires high costs to build and maintain, including equipment maintenance, data management, and technical personnel input.

[0007] In the prior art, a Chinese patent (application number: 201810938571.8, publication number: CN 108830264A) discloses a platform passenger detection system and method for an unmanned bus. The system includes an on-board computing terminal, an external front camera, an external rear camera, an external left camera, an external right camera, a display, a GPS global positioning module, a 4G network communication module, and a PC remote monitoring center. The on-board computing terminal includes an ARM controller, a DSP image stitching module, a DSP image detection module, a GPU module, and a CAN communication module. The DSP image stitching module is connected to the ARM controller and is used to process the video from four external cameras in real time and stitch them into a panoramic video. The DSP image detection module is connected to the ARM controller and is used to process the video from the internal front camera in real time and identify and detect passengers at the unmanned bus platform to automatically park the vehicle. The 4G network communication module is connected to the ARM controller. The four external camera modules are used to collect real-time video of the surrounding area outside the vehicle. This system relies heavily on external cameras to collect data. In severe weather (such as heavy rain, heavy snow, dense fog, etc.), the camera's field of view may be affected, resulting in unclear image acquisition, which in turn affects the accuracy of passenger detection and the normal operation of the system. The 4G network communication module is used to transmit information to the PC remote monitoring center. When the network signal is unstable or the network is congested, data transmission delays may occur, affecting the real-time remote monitoring and timely control of the vehicle. Summary of the Invention

[0008] In view of the above problems, the present invention provides a digital platform passenger behavior analysis and personalized service push method, which can not only accurately collect and analyze the behavior, trajectory, needs and other information of passengers and pedestrians at the platform, but also inform the vehicle congestion in advance and push personalized riding suggestions and other services, thereby improving the operational efficiency, service quality and passenger travel experience of unmanned buses.

[0009] In order to achieve the above-mentioned and other related purposes, the present invention provides the following technical solutions:

[0010] A method for analyzing passenger behavior on a digital platform and delivering personalized services, the method comprising:

[0011] L1. When a passenger enters the digital platform, the platform acquires historical travel trajectory data. The platform's built-in camera captures the passenger's image data in real time. The platform's built-in infrared sensor captures the passenger's location coordinate sequence data in real time. The platform's built-in pressure sensor captures the platform's pressure change time series data in real time.

[0012] L2. Based on the passenger image data information, construct a skeleton key point detection model based on a dynamic learning factor, characterize the passenger's body movement feature matrix, and obtain data information of the passenger's body movement feature matrix;

[0013] L3. Based on the data information of the passenger's body movement feature matrix, the data information of the passenger's position coordinate sequence, and the data information of the platform pressure change time series, a BP neural network regression prediction algorithm based on the Pearson correlation coefficient is used to predict the passenger's travel time at the platform, thereby obtaining the predicted travel time data information of the passenger's platform;

[0014] L4. Based on the predicted travel time data of the passenger at the station and the historical travel trajectory data of the passenger, a push service model for passenger travel is constructed, and a corresponding personalized push strategy is formulated. The push is sent to the mobile phone APP or WeChat applet of the platform passenger via the wireless network.

[0015] Furthermore, in step L2, the construction of a skeleton key point detection model based on a dynamic learning factor to characterize the passenger's body movement feature matrix includes:

[0016] L21. The passenger image data information is input into the convolutional neural network model to extract the passenger image features to obtain the passenger image feature matrix data information;

[0017] L22 based on the passenger image feature matrix data information, establish the passenger's skeleton key point extraction function Q,

[0018] ,

[0019] Where x is the data information of the passenger's image feature matrix, α, β and δ are weight coefficients, and the passenger's skeletal key points are extracted to obtain the data information of the passenger's skeletal key point coordinates;

[0020] L23. Based on the data information of the passenger's skeletal key point coordinates, construct a passenger's body movement characterization function W,

[0021] ,

[0022] Among them, y is the data information of the coordinates of the passenger's skeleton key points, γ1, γ2 and γ3 are dynamic learning factors, which characterize the passenger's limb movement feature matrix to obtain the data information of the passenger's limb movement feature matrix.

[0023] Furthermore, the dynamic learning factors γ1, γ2 and γ3 are,

[0024] ,

[0025] ,

[0026] ,

[0027] Among them, y is the data information of the coordinates of the passenger's skeleton key points.

[0028] Furthermore, in step L3, the use of the BP neural network regression prediction algorithm based on the Pearson correlation coefficient to predict the travel time of the passenger's platform includes:

[0029] L31. Based on the data information of the passenger's position coordinate sequence and the platform's pressure change time series data information, establish the Pearson correlation coefficient function R of the two variables,

[0030] ,

[0031] Among them, X is the data information of the passenger's position coordinate sequence, and Y is the data information of the platform's pressure change time series. The relationship between the two variables is characterized to obtain the data information of the relationship matrix between the passenger's position coordinates and the platform's pressure change;

[0032] L32. The data information of the relationship matrix between the passenger's position coordinates and the platform's pressure changes and the data information of the passenger's body movement feature matrix are input into the BP neural network regression prediction model for training and learning to obtain a trained BP neural network regression prediction model;

[0033] L33. Based on the trained BP neural network regression prediction model, the data information of the relationship matrix between the passenger's position coordinates and the platform pressure change and the data information of the passenger's body movement feature matrix are input to predict the passenger's travel time at the platform, and the predicted data information of the passenger's travel time at the platform is obtained.

[0034] Furthermore, the BP neural network regression prediction model includes an input layer, a hidden layer and an output layer, and the neuron function P of the hidden layer is,

[0035] ,

[0036] Among them, r1 is the data information of the relationship matrix between the passenger's position coordinates and the pressure change of the platform, r2 is the data information of the passenger's body movement feature matrix, ω1, ω2 and ω3 are the weight factors of the BP neural network.

[0037] Furthermore, the constraints of the weight factors ω1, ω2 and ω3 of the BP neural network are:

[0038] .

[0039] Furthermore, in step L4, the construction of a push service model for passenger travel and the formulation of a corresponding personalized push strategy include:

[0040] L41. Based on the predicted data information of the passenger's platform travel time and the passenger's historical travel trajectory data information, normalization processing is performed to obtain the normalized passenger's platform travel time and the passenger's historical travel trajectory data information;

[0041] L42. Based on the normalized passenger's travel time at the platform and the passenger's historical travel trajectory data, establish a target optimization function G for passenger travel trajectory.

[0042] ,

[0043] Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, h2 is the normalized data information of the passenger's historical travel trajectory, ρ1, ρ2 and ρ3 are penalty factors;

[0044] L43. Based on the target optimization function G of the passenger travel trajectory, the passenger travel trajectory is optimized to obtain data information of the optimized passenger travel trajectory, thereby formulating a corresponding personalized push strategy.

[0045] Furthermore, the corresponding personalized push strategies include a push strategy based on station actions, a push strategy based on travel habits, a push strategy based on real-time scenarios, and a push strategy based on interest preferences. The penalty factors ρ1, ρ2, and ρ3 are:

[0046] ,

[0047] ,

[0048] ,

[0049] Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, and h2 is the normalized data information of the passenger's historical travel trajectory.

[0050] In order to achieve the above-mentioned objectives and other related objectives, the present invention also provides a digital platform passenger behavior analysis and personalized service push system, including a computer device that is programmed or configured to execute the steps of any one of the digital platform passenger behavior analysis and personalized service push methods.

[0051] In order to achieve the above-mentioned purpose and other related purposes, the present invention also provides a computer-readable storage medium, which stores a computer program programmed or configured to execute any one of the digital platform passenger behavior analysis and personalized service push methods.

[0052] The present invention has the following positive effects:

[0053] 1. This invention builds a complete digital platform passenger behavior analysis and personalized service push system for unmanned buses, covering the digital platform perception layer, data transmission layer, data processing center and personalized service push module. The layers work together to realize the full process function from passenger behavior data collection to personalized service push.

[0054] 2. This invention adopts multiple sensor fusion technologies to comprehensively collect passenger behavior data at the platform through cameras, infrared sensors, pressure sensors and other equipment, providing a rich data source for in-depth analysis of passenger behavior.

[0055] 3. The present invention predicts the travel time of the passenger at the platform by adopting a BP neural network regression prediction algorithm based on the Pearson correlation coefficient, and combines it with the construction of a skeletal key point detection model based on dynamic learning factors to characterize the passenger's body movement feature matrix. It not only formulates a personalized service push strategy based on platform movements, travel habits, real-time scenarios and interest preferences, but also can provide diversified and personalized service content according to the different needs and actual conditions of passengers. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a schematic diagram of a push scenario of the present invention;

[0057] Figure 2 Schematic diagram of the method flow of the present invention;

[0058] Figure 3 A schematic diagram of a skeleton key point detection model based on dynamic learning factors according to the present invention;

[0059] Figure 4 Schematic diagram of the process of the BP neural network regression prediction algorithm based on the Pearson correlation coefficient of the present invention;

[0060] Figure 5This is a flow chart of the present invention for constructing a push service model for passenger travel. DETAILED DESCRIPTION

[0061] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0062] Example 1: Figure 1 or Figure 2 As shown, a method for analyzing passenger behavior on a digital platform and delivering personalized services comprises:

[0063] L1. When a passenger enters the digital platform, the platform acquires historical travel trajectory data. The platform's built-in camera captures the passenger's image data in real time. The platform's built-in infrared sensor captures the passenger's location coordinate sequence data in real time. The platform's built-in pressure sensor captures the platform's pressure change time series data in real time.

[0064] L2. Based on the passenger image data information, construct a skeleton key point detection model based on a dynamic learning factor, characterize the passenger's body movement feature matrix, and obtain data information of the passenger's body movement feature matrix;

[0065] L3. Based on the data information of the passenger's body movement feature matrix, the data information of the passenger's position coordinate sequence, and the data information of the platform pressure change time series, a BP neural network regression prediction algorithm based on the Pearson correlation coefficient is used to predict the passenger's travel time at the platform, thereby obtaining the predicted travel time data information of the passenger's platform;

[0066] L4. Based on the predicted travel time data of the passenger at the station and the historical travel trajectory data of the passenger, a push service model for passenger travel is constructed, and a corresponding personalized push strategy is formulated. The push is sent to the mobile phone APP or WeChat applet of the platform passenger via the wireless network.

[0067] In this embodiment, if Figure 3 As shown, in step L2, the construction of a skeleton key point detection model based on a dynamic learning factor and the characterization of the passenger's limb motion feature matrix include:

[0068] L21. The passenger image data information is input into the convolutional neural network model to extract the passenger image features to obtain the passenger image feature matrix data information;

[0069] L22 based on the passenger image feature matrix data information, establish the passenger's skeleton key point extraction function Q,

[0070] ,

[0071] Where x is the data information of the passenger's image feature matrix, α, β and δ are weight coefficients, and the passenger's skeletal key points are extracted to obtain the data information of the passenger's skeletal key point coordinates;

[0072] L23. Based on the data information of the passenger's skeletal key point coordinates, construct a passenger's body movement characterization function W,

[0073] ,

[0074] Among them, y is the data information of the coordinates of the passenger's skeleton key points, γ1, γ2 and γ3 are dynamic learning factors, which characterize the passenger's limb movement feature matrix to obtain the data information of the passenger's limb movement feature matrix.

[0075] In this embodiment, the dynamic learning factors γ1, γ2 and γ3 are,

[0076] ,

[0077] ,

[0078] ,

[0079] Among them, y is the data information of the coordinates of the passenger's skeleton key points.

[0080] In this embodiment, if Figure 4 As shown, in step L3, the use of the BP neural network regression prediction algorithm based on the Pearson correlation coefficient to predict the travel time of the passenger's platform includes:

[0081] L31. Based on the data information of the passenger's position coordinate sequence and the platform's pressure change time series data information, establish the Pearson correlation coefficient function R of the two variables,

[0082] ,

[0083] Among them, X is the data information of the passenger's position coordinate sequence, and Y is the data information of the platform's pressure change time series. The relationship between the two variables is characterized to obtain the data information of the relationship matrix between the passenger's position coordinates and the platform's pressure change;

[0084] L32. The data information of the relationship matrix between the passenger's position coordinates and the platform's pressure changes and the data information of the passenger's body movement feature matrix are input into the BP neural network regression prediction model for training and learning to obtain a trained BP neural network regression prediction model;

[0085] L33. Based on the trained BP neural network regression prediction model, the data information of the relationship matrix between the passenger's position coordinates and the platform pressure change and the data information of the passenger's body movement feature matrix are input to predict the passenger's travel time at the platform, and the predicted data information of the passenger's travel time at the platform is obtained.

[0086] In this embodiment, the BP neural network regression prediction model includes an input layer, a hidden layer and an output layer. The neuron function P of the hidden layer is:

[0087] ,

[0088] Among them, r1 is the data information of the relationship matrix between the passenger's position coordinates and the pressure change of the platform, r2 is the data information of the passenger's body movement feature matrix, ω1, ω2 and ω3 are the weight factors of the BP neural network.

[0089] In this embodiment, the constraints of the weight factors ω1, ω2 and ω3 of the BP neural network are:

[0090] .

[0091] Example 2: Based on the digital platform passenger behavior analysis and personalized service push method of Example 1, the present invention is further illustrated and described below.

[0092] like Figure 1 or Figure 2 As shown, a method for analyzing passenger behavior on a digital platform and delivering personalized services comprises:

[0093] L1. When a passenger enters the digital platform, the platform acquires historical travel trajectory data. The platform's built-in camera captures the passenger's image data in real time. The platform's built-in infrared sensor captures the passenger's location coordinate sequence data in real time. The platform's built-in pressure sensor captures the platform's pressure change time series data in real time.

[0094] L2. Based on the passenger image data information, construct a skeleton key point detection model based on a dynamic learning factor, characterize the passenger's body movement feature matrix, and obtain data information of the passenger's body movement feature matrix;

[0095] L3. Based on the data information of the passenger's body movement feature matrix, the data information of the passenger's position coordinate sequence, and the data information of the platform pressure change time series, a BP neural network regression prediction algorithm based on the Pearson correlation coefficient is used to predict the passenger's travel time at the platform, thereby obtaining the predicted travel time data information of the passenger's platform;

[0096] L4. Based on the predicted travel time data of the passenger at the station and the historical travel trajectory data of the passenger, a push service model for passenger travel is constructed, and a corresponding personalized push strategy is formulated. The push is sent to the mobile phone APP or WeChat applet of the platform passenger via the wireless network.

[0097] In this embodiment, if Figure 5 As shown, in step L4, the construction of a push service model for passenger travel and the formulation of a corresponding personalized push strategy include:

[0098] L41. Based on the predicted data information of the passenger's platform travel time and the passenger's historical travel trajectory data information, normalization processing is performed to obtain the normalized passenger's platform travel time and the passenger's historical travel trajectory data information;

[0099] L42. Based on the normalized passenger's travel time at the platform and the passenger's historical travel trajectory data, establish a target optimization function G for passenger travel trajectory.

[0100] ,

[0101] Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, h2 is the normalized data information of the passenger's historical travel trajectory, ρ1, ρ2 and ρ3 are penalty factors;

[0102] L43. Based on the target optimization function G of the passenger travel trajectory, the passenger travel trajectory is optimized to obtain data information of the optimized passenger travel trajectory, thereby formulating a corresponding personalized push strategy.

[0103] In this embodiment, the corresponding personalized push strategies include a push strategy based on station actions, a push strategy based on travel habits, a push strategy based on real-time scenarios, and a push strategy based on interest preferences. The penalty factors ρ1, ρ2, and ρ3 are:

[0104] ,

[0105] ,

[0106] ,

[0107] Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, and h2 is the normalized data information of the passenger's historical travel trajectory.

[0108] In this embodiment, the push strategy based on the site action is shown in the table:

[0109]

[0110] In this embodiment, the push strategy based on travel habits is shown in the following table:

[0111]

[0112] In this embodiment, the push strategy based on real-time scenarios is shown in the following table:

[0113]

[0114] In this embodiment, the push strategy based on interest preference is shown in the following table:

[0115]

[0116] In this embodiment, the present invention provides a digital platform passenger behavior analysis and personalized service push system, including a computer device that is programmed or configured to execute any one of the steps of the digital platform passenger behavior analysis and personalized service push method.

[0117] In this embodiment, the present invention provides a computer-readable storage medium storing a computer program programmed or configured to execute any one of the methods for analyzing passenger behavior on a digital platform and delivering personalized services.

[0118] Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0119] In summary, the present invention can not only accurately collect and analyze information such as the behavior, trajectory, and needs of passengers and pedestrians at the platform, but also inform passengers of vehicle congestion in advance and push personalized riding suggestions, thereby improving the operational efficiency, service quality and passenger travel experience of unmanned buses.

[0120] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A digital platform passenger behavior analysis and personalized service push method, characterized by: The method comprises: L1. When a passenger enters the digital platform, the platform acquires historical travel trajectory data. The platform's built-in camera captures the passenger's image data in real time. The platform's built-in infrared sensor captures the passenger's location coordinate sequence data in real time. The platform's built-in pressure sensor captures the platform's pressure change time series data in real time. L2. Based on the passenger image data information, construct a skeleton key point detection model based on a dynamic learning factor, characterize the passenger's body movement feature matrix, and obtain data information of the passenger's body movement feature matrix; L3. Based on the data information of the passenger's body movement feature matrix, the data information of the passenger's position coordinate sequence, and the data information of the platform pressure change time series, a BP neural network regression prediction algorithm based on the Pearson correlation coefficient is used to predict the passenger's travel time at the platform, thereby obtaining the predicted travel time data information of the passenger's platform; L4. Based on the predicted travel time data of the passenger at the station and the historical travel trajectory data of the passenger, a push service model for passenger travel is constructed, and a corresponding personalized push strategy is formulated. The push is sent to the mobile phone APP or WeChat applet of the platform passenger via the wireless network.

2. The digital platform passenger behavior analysis and personalized service push method according to claim 1 is characterized in that: In step L2, the construction of a skeleton key point detection model based on dynamic learning factors to characterize the passenger's body movement feature matrix includes: L21. The passenger image data information is input into the convolutional neural network model to extract the passenger image features to obtain the passenger image feature matrix data information; L22 based on the passenger image feature matrix data information, establish the passenger's skeleton key point extraction function Q, ， Where x is the data information of the passenger's image feature matrix, α, β and δ are weight coefficients, and the passenger's skeletal key points are extracted to obtain the data information of the passenger's skeletal key point coordinates; L23. Based on the data information of the passenger's skeletal key point coordinates, construct a passenger's body movement characterization function W, ， Among them, y is the data information of the coordinates of the passenger's skeleton key points, γ1, γ2 and γ3 are dynamic learning factors, which characterize the passenger's limb movement feature matrix to obtain the data information of the passenger's limb movement feature matrix.

3. The digital platform passenger behavior analysis and personalized service delivery method according to claim 2 is characterized by: The dynamic learning factors γ1, γ2 and γ3 are, ， ， ， Among them, y is the data information of the coordinates of the passenger's skeleton key points.

4. The method for analyzing passenger behavior on a digital platform and delivering personalized services according to claim 1, characterized in that: In step L3, the use of the BP neural network regression prediction algorithm based on the Pearson correlation coefficient to predict the travel time of the passenger's platform includes: L31. Based on the data information of the passenger's position coordinate sequence and the platform's pressure change time series data information, establish the Pearson correlation coefficient function R of the two variables, , Among them, X is the data information of the passenger's position coordinate sequence, and Y is the data information of the platform's pressure change time series. The relationship between the two variables is characterized to obtain the data information of the relationship matrix between the passenger's position coordinates and the platform's pressure change; L32. The data information of the relationship matrix between the passenger's position coordinates and the platform's pressure changes and the data information of the passenger's body movement feature matrix are input into the BP neural network regression prediction model for training and learning to obtain a trained BP neural network regression prediction model; L33. Based on the trained BP neural network regression prediction model, the data information of the relationship matrix between the passenger's position coordinates and the platform pressure change and the data information of the passenger's body movement feature matrix are input to predict the passenger's travel time at the platform, and the predicted data information of the passenger's travel time at the platform is obtained.

5. The digital platform passenger behavior analysis and personalized service delivery method according to claim 4 is characterized by: The BP neural network regression prediction model includes an input layer, a hidden layer and an output layer. The neuron function P of the hidden layer is: ， Among them, r1 is the data information of the relationship matrix between the passenger's position coordinates and the pressure change of the platform, r2 is the data information of the passenger's body movement feature matrix, ω1, ω2 and ω3 are the weight factors of the BP neural network.

6. The digital platform passenger behavior analysis and personalized service delivery method according to claim 5 is characterized by: The constraints of the weight factors ω1, ω2 and ω3 of the BP neural network are: 。 7. The method for analyzing passenger behavior on a digital platform and delivering personalized services according to claim 1, characterized in that: In step L4, the construction of a push service model for passenger travel and the formulation of a corresponding personalized push strategy include: L41. Based on the predicted data information of the passenger's platform travel time and the passenger's historical travel trajectory data information, normalization processing is performed to obtain the normalized passenger's platform travel time and the passenger's historical travel trajectory data information; L42. Based on the normalized passenger's travel time at the platform and the passenger's historical travel trajectory data, establish a target optimization function G for passenger travel trajectory. ， Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, h2 is the normalized data information of the passenger's historical travel trajectory, ρ1, ρ2 and ρ3 are penalty factors; L43. Based on the target optimization function G of the passenger travel trajectory, the passenger travel trajectory is optimized to obtain data information of the optimized passenger travel trajectory, thereby formulating a corresponding personalized push strategy.

8. The digital platform passenger behavior analysis and personalized service delivery method according to claim 7 is characterized by: The corresponding personalized push strategies include push strategies based on platform actions, push strategies based on travel habits, push strategies based on real-time scenarios, and push strategies based on interest preferences. The penalty factors ρ1, ρ2, and ρ3 are: ， ， ， Among them, h1 is the normalized data information of the passenger's travel time at the station where he is located, and h2 is the normalized data information of the passenger's historical travel trajectory.

9. A digital platform passenger behavior analysis and personalized service push system, including computer equipment, characterized in that: The computer device is programmed or configured to execute the steps of the method for analyzing digital platform passenger behavior and pushing personalized services as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program programmed or configured to execute the method for analyzing digital platform passenger behavior and pushing personalized services as described in any one of claims 1 to 8.

Images