An intelligent management method for autonomous order matching of online car-hailing

By cleaning and normalizing the online car-hailing order data, the convolutional neural network is used to analyze the order coverage details, and the order matching management coefficient is generated to regulate the online car-hailing matching process, which solves the unsolved supply and demand contradiction in the existing technology. The problem has improved the matching efficiency and user experience of online car-hailing.

CN118429170BActive Publication Date: 2024-11-15BEIJING BAILONG MAYUN TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410420573.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-09
Publication Date
2024-11-15
Estimated Expiration
2044-04-09

AI Technical Summary

Technical Problem

The supply and demand contradictions in the existing online ride-hailing platforms have not been effectively resolved during the order matching process, resulting in high vehicle air driving rate and slow or inaccurate order matching process, which increases operating costs and aggravates road traffic congestion.

Method used

By obtaining and cleaning the data related to online car-hailing orders, normalizing the processing, and using convolutional neural network to analyze the order coverage details, generate order matching management coefficients to regulate the matching process of online car-hailing, and improve the Supply and demand adaptability.

Benefits of technology

It reduces the situation where vehicles run empty and users cannot get a taxi, improves the supply and demand adaptability between users and online car-hailing, reduces operating costs and optimizes traffic flow.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118429170B_ABST
    Figure CN118429170B_ABST
Patent Text Reader

Abstract

The present invention discloses an intelligent management method for autonomous order matching of online car-hailing, and relates to the field of online car-hailing order management. The method comprises the following steps: first, when matching online car-hailing with orders on a platform, the present invention obtains order-related data generated by the online car-hailing and performs data cleaning, and normalizes the cleaned order data as a sample for model training, and then analyzes the online car-hailing order matching situation by obtaining order coverage details information generated during order matching management, and timely regulates and manages the matching of online car-hailing orders according to the adaptability of order matching, thereby reducing the situation where vehicles run empty and users cannot get a car, and improving the supply and demand adaptation between users and online car-hailing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of online car-hailing order management, and in particular to an intelligent management method for autonomous order matching of online car-hailing. Background Art

[0002] Online ride-hailing is a taxi service that is booked and dispatched through an Internet platform. With the popularization of smartphones and Internet technology, people can more conveniently use mobile applications to book and pay for travel services, allowing online ride-hailing to develop rapidly in the context of the sharing economy. Through the online ride-hailing platform, individual car owners can share their vehicles with people who need to travel, thereby improving vehicle utilization and reducing travel costs.

[0003] The existing technology has the following shortcomings: Although the online car-hailing operation mode has alleviated the difficulty of hailing a taxi compared to traditional taxis, its characteristics still lead to the contradiction between supply and demand not being effectively resolved. Since passengers' travel needs are affected by various random factors and drivers are blindly searching for passengers, the online car-hailing platform cannot dispatch vehicles in advance and effectively to allocate travel orders, resulting in a high empty driving rate of online car-hailing vehicles and a slow or inaccurate order matching process, which not only increases the operating costs of the platform, but also aggravates the road traffic congestion in the city.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide an intelligent management method for autonomous order matching of online car-hailing vehicles. When matching online car-hailing vehicles with orders on the platform, order-related data generated by the online car-hailing vehicles is obtained and data cleaning is performed, and the cleaned order data is normalized as a sample for model training. Then, the order matching situation of the online car-hailing vehicles is analyzed by obtaining the order coverage detail information generated in the process of order matching management. According to the adaptability of the order matching, the matching of the online car-hailing vehicles is timely regulated and managed, thereby reducing the situation where vehicles run empty and users cannot get a car, and improving the supply and demand adaptation between users and online car-hailing vehicles, so as to solve the shortcomings of the background technology.

[0006] In order to achieve the above-mentioned purpose, the present invention provides the following technical solution: an intelligent management method for autonomous order matching of online car-hailing, comprising the following steps:

[0007] When the platform matches online ride-hailing vehicles with orders, it obtains order-related data generated by the online ride-hailing vehicles and performs data cleaning, and normalizes the cleaned order data;

[0008] The order data is processed through a convolutional neural network and the model is trained to obtain the order coverage details information generated during the order matching management process. The order coverage details information includes node information diffusion, verification error information, and regional stability information.

[0009] The node information diffusion, verification error information and regional stability information are used to generate the order matching management coefficient and analyze the order matching of online ride-hailing vehicles to determine the adaptability of order matching;

[0010] The matching of online ride-hailing orders is managed based on the adaptability of order matching.

[0011] Preferably, when the platform matches online ride-hailing vehicles with orders, the order-related data generated by the online ride-hailing vehicles is obtained and the data is cleaned, and the cleaned order data is normalized, including:

[0012] Use data processing tools or programming languages ​​to scan the data for duplicate data in order-related data, identify duplicate records, select to mark the duplicate data, and leave one copy;

[0013] Use statistical methods or visualization tools to identify missing values ​​in the data set and delete or fill in missing data in order-related data;

[0014] Extract features from the cleaned data and use min-max scaling to scale the features of the order data;

[0015] Calculate the minimum and maximum values ​​of the features and normalize each feature.

[0016] Preferably, the node information diffusion includes a diffusion convergence index, the verification error information includes a verification error deviation value, and the regional stability information includes a regional order matching value.

[0017] Preferably, the logic for obtaining the diffusion convergence index is as follows:

[0018] Obtain the number of node feature vectors N when the graph convolutional network performs information transmission, obtain the initial feature vector of the node in each information transmission process, and establish the initial vector set OFi = {OF1, OF2, ..., OFm}, obtain the corresponding final feature vector, and establish the final feature vector set FFi = {FF1, FF2, ..., FFm}, where m represents the dimension of the vector and the number of elements in the vector. Calculate the vector dot product value and vector norm based on the initial feature vector and the final feature vector. The calculation formula of the vector dot product value is: The calculation formula for the vector norm is: The convergence value of the node feature vector is calculated based on the vector dot product value and the vector norm: SIM = (1 + FS) / 2. The diffusion convergence index is calculated for each node feature vector. The calculation expression is:

[0019] Preferably, the logic for obtaining the verification error deviation value is as follows:

[0020] Get the number of samples n for model training, and get the predicted value xc and the true value yc of the order matching after training, and calculate the predicted mean by summarizing and the true mean Calculate the correlation value and error value. The correlation value calculation formula is: The error value calculation expression is: The verification error deviation value is calculated based on the correlation value and the error value, and the calculation expression is: YZW = |Cov(x, y)| / WC.

[0021] Preferably, the logic for obtaining the regional order matching value is as follows:

[0022] Get the allocation time of each order and the time when the driver accepts the order within a unit of time, and get the actual number of orders JD 接单 and the actual allocated order quantity JD 实际 , calculate the order value: JZ = JD 接单 / JD 实际 For successfully accepted orders, obtain the total matching order duration ZS of all successfully accepted orders, and calculate the matching order duration value: SC = ZS / JD 实际 , get the matching order duration value of historical time Calculate the regional order matching value, the calculation expression is:

[0023] Preferably, the node information diffusion, verification error information and regional stability information are used to generate an order matching management coefficient and the order matching situation of the online car-hailing service is analyzed to determine the adaptability of the order matching. The specific steps include:

[0024] The obtained diffusion convergence index, verification error deviation value, and regional order matching value are used to generate an order matching management coefficient;

[0025] Compare the generated order matching management coefficient with the matching management threshold;

[0026] If the order matching management coefficient is greater than the matching management threshold, a normal matching signal is generated;

[0027] If the order matching management coefficient is less than or equal to the matching management threshold, the training model is generated as a regulation management signal to notify personnel to perform management regulation.

[0028] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0029] The present invention obtains order-related data generated by the online ride-hailing vehicle and cleans the data when matching the online ride-hailing vehicle with orders on the platform, and normalizes the cleaned order data as a sample for model training. The order matching situation of the online ride-hailing vehicle is then analyzed by obtaining the order coverage detail information generated in the process of order matching management. The matching of the online ride-hailing vehicle orders is timely regulated and managed according to the adaptability of the order matching, thereby reducing the situation where vehicles run empty and users cannot get a vehicle, and improving the supply and demand adaptation between users and online ride-hailing vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present application 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 recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0031] Figure 1 The present invention is a flow chart of an intelligent management method for autonomous order matching of online ride-hailing vehicles. DETAILED DESCRIPTION

[0032] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0033] Embodiment 1: The present invention provides Figure 1 The method for intelligent management of autonomous order matching of online car-hailing vehicles shown in the figure comprises the following steps:

[0034] When the platform matches online ride-hailing vehicles with orders, it obtains order-related data generated by the online ride-hailing vehicles and performs data cleaning, and normalizes the cleaned order data;

[0035] The order data is processed through a convolutional neural network and the model is trained to obtain the order coverage details information generated during the order matching management process. The order coverage details information includes node information diffusion, verification error information, and regional stability information.

[0036] The node information diffusion, verification error information and regional stability information are used to generate the order matching management coefficient and analyze the order matching of online ride-hailing vehicles to determine the adaptability of order matching;

[0037] The matching of online ride-hailing orders is managed based on the adaptability of order matching.

[0038] Autonomous order matching of online ride-hailing refers to the process in which passengers actively initiate orders through mobile applications and the system autonomously matches vehicles. Passengers select the departure point and destination, specify the travel time, and confirm the order through the mobile application of the online ride-hailing platform. The user's order information will be transmitted to the server of the online ride-hailing platform. The system monitors the location and idle status of nearby drivers in real time. The matching algorithm of the online ride-hailing platform is calculated based on factors such as the passenger's order information, the location of nearby drivers, and the timeliness of the order to find the most suitable driver. After the driver accepts the order, the vehicle starts to go to the boarding point specified by the passenger. In some areas or time periods with high order frequency, the number of online ride-hailing vehicles in the area is predicted and controlled to meet the order matching needs.

[0039] Order matching demand prediction is to predict the road traffic status in the future based on the historical and current order demand data in the urban traffic road network, and to improve the accuracy of the order demand prediction model by using a large amount of order demand data from online car-hailing or taxi platforms;

[0040] Urban order demand refers to the total amount of online car-hailing or taxi orders generated in a specific time period in the city, representing the taxi volume of urban residents' travel needs. This demand data has complex spatiotemporal dynamics. In terms of space, order demand is constrained by the topological structure of the traffic network, and is also affected by the order demand conditions in other regions and the historical order demand conditions. In terms of time, order demand changes over time and is affected by the cyclical nature of people's daily activities. In addition, order demand will also be affected by external factors, such as weather conditions, holidays, and emergencies, which makes the overall demand show strong uncertainty.

[0041] In terms of space, the demand for urban orders is constrained by the topological structure of the transportation network. Some areas may have higher order demand because they have busy commercial centers, residential areas or transportation hubs. At the same time, order demand is also affected by the order status of adjacent areas, forming a spatial correlation. This spatial correlation means that when the order demand in some areas increases, the demand in surrounding areas may also increase, resulting in a certain degree of overall agglomeration;

[0042] Order demand data shows dynamic changes in time, and people's daily activities have certain periodicity, such as morning and evening peak hours. Therefore, the order demand will fluctuate significantly in different time periods of the day. In addition, specific dates, such as weekends or holidays, may also have a significant impact on order demand. Changes in the clock and periodic activities require the forecasting and scheduling system to fully consider the time factor to cope with demand fluctuations in different time periods;

[0043] Order demand is also affected by various external factors. Weather conditions are an important factor. Bad weather can cause dramatic fluctuations in order volume, such as an increase in rainy or snowy weather. In addition, unexpected factors such as social activities, large-scale events, or emergencies may also have a temporary impact on order demand. Accurate identification and timely response to these external factors are essential for optimizing scheduling and improving service levels. Therefore, in order demand analysis, these complex spatiotemporal dynamics and the interference of external factors should be considered to achieve more accurate vehicle scheduling and service optimization.

[0044] When an order is matched with an online ride-hailing vehicle through the platform, many factors need to be considered to achieve accurate order matching, check and proofread the order-related data, calibrate uncertain data, and eliminate redundant and irrelevant data to improve data quality;

[0045] If the key information of an order, such as passenger name, contact information, vehicle information, departure location, destination, etc., is null or not provided, these data can be regarded as missing data. If the time information such as the timestamp, estimated arrival time, and actual arrival time of the order is missing, it also falls into the category of missing data. Completely duplicated data in duplicate data refers to completely identical order records, including the case where all fields such as order ID, passenger information, and itinerary information are the same, which can be defined as completely duplicated data.

[0046] Some fields are repeated: In some cases, only some fields may be the same, but there are identical records in the key fields, which can also be considered as duplicate data. For example, the same order ID but different other information;

[0047] Duplicates within a time window: If there are identical order records within a certain time window, they can also be defined as duplicate data. For example, the same user places multiple identical orders within the same day;

[0048] For duplicate data in order-related data, use data processing tools or programming languages ​​to scan the data, identify duplicate records, and choose to mark the duplicate data and leave a copy, or directly delete all duplicate records. Depending on the specific business needs, perform more detailed checks on specific fields, such as unique identifiers or key information, to ensure the uniqueness of the data.

[0049] For missing data in order-related data, use statistical methods or visualization tools to identify missing values ​​in the data set, delete or fill in missing data. For records containing a large number of missing values, consider deleting the entire record, but ensure that the deletion does not cause serious loss of information, and use interpolation, mean, median, mode and other methods to fill in missing values ​​to maintain data integrity. The filling method depends on the nature of the data and the missing pattern. For some cases, machine learning models can be used to predict missing values, that is, when there is a correlation between the pattern of missing values ​​and other variables, that is, different processing processes are performed according to different situations.

[0050] The cleaned order-related data is taken as order data and normalized to fall within a smaller general range. Normalization is a data preprocessing technique that scales the data so that it falls within a smaller general range. In the processing of online car-hailing order data, normalization can help avoid the impact of scale differences between different features on the model and improve the stability and convergence speed of model training. Use minimum-maximum scaling to scale the order data features to a specified range, usually [0, 1]. The specific steps are as follows:

[0051] For each feature, calculate the minimum value (min) and maximum value (max) of the feature;

[0052] For each value in each feature, normalize it using the following formula: nor =(XX min ) / (X max -X min ), where X is the original eigenvalue, X min , X max are the minimum and maximum values ​​of the feature respectively;

[0053] For each feature, the minimum value of the feature is converted to 0, the maximum value is converted to 1, and each other value is converted to a decimal between 0 and 1. The calculation formula is:

[0054] The normalized data is put into the model for training, and the spatial correlation of the order data is processed through the convolutional neural network, as follows:

[0055] The order demand data is organized in the format of an image, with the time dimension as the depth (channel) of the image, the geographic coordinates as the height and width of the image, and the order demand quantity is mapped to the intensity of the image pixels, for example, using color depth to represent the demand intensity;

[0056] Construct a convolutional neural network, including input layer, convolution layer, and pooling layer;

[0057] Input layer: The size of the input layer should match the size of the image, and each channel corresponds to the order demand data of one time step;

[0058] Convolutional layer: Use convolutional layers to capture spatial correlation. The size of the convolution kernel can be selected based on the geographic scope of the data, and the receptive field of the sliding window is used to extract spatial dependencies.

[0059] Pooling layer: Use the pooling layer to reduce the dimensionality of the data, retain key features, and enhance the perception of spatial dependencies;

[0060] Model training: Choose an appropriate loss function, such as mean squared error for regression problems, to measure the difference between the model prediction and actual demand;

[0061] It should be noted that in practical applications, it is necessary to adjust the model structure and hyperparameters according to the specific data distribution and problem characteristics. The characteristics of time series can be considered, and it may be necessary to combine models such as LSTM (Long Short-Term Memory Network) to better capture time correlation.

[0062] Obtain order coverage details generated during order matching management to analyze the accuracy of online car-hailing prediction and matching orders. Order coverage details include node information diffusion, verification error information, and regional stability information.

[0063] The node information diffusion includes the diffusion convergence index, the verification error information includes the verification error deviation value, and the regional stability information includes the regional order matching value. After collection, the diffusion convergence index, verification error deviation value, and regional order matching value are calibrated as KSS, YZW, and QYD, respectively.

[0064] The diffusion convergence index generated during model training has an important impact on the accuracy of analyzing order matching intelligent management. The diffusion convergence index is used to measure the degree of retention of node information by the graph convolutional network and the similarity change of node features during information transmission, which is of key significance to the accuracy of order matching intelligent management. The diffusion convergence index has the following functions:

[0065] Accuracy and prediction performance: A higher diffusion convergence index usually means that the model better retains the original information of the node during information transmission, which helps to improve the accuracy and prediction performance of the model. A low diffusion convergence index may reduce the accuracy of the model because the node characteristics change greatly during the transmission process, and the model is difficult to effectively capture and use information, resulting in a decrease in the accuracy of the model's prediction of orders;

[0066] Ability to understand node information: A higher diffusion convergence index helps to deeply understand how the model processes and changes node information and improves the understanding of order data. A low diffusion convergence index may make the model's processing of node information more unstable and reduce the interpretability of order data.

[0067] Therefore, by obtaining the diffusion convergence index generated during model training, the model training compliance can be analyzed, and then the number of information transmission layers, neighbor node selection strategy, etc. can be set to optimize model performance and improve matching accuracy;

[0068] The logic for obtaining the diffusion convergence index is as follows:

[0069] Obtain the number of node feature vectors N when the graph convolutional network performs information transmission, obtain the initial feature vector of the node in each information transmission process, and establish the initial vector set OFi = {OF1, OF2, ..., OFm}, obtain the corresponding final feature vector, and establish the final feature vector set FFi = {FF1, FF2, ..., FFm}, where m represents the dimension of the vector and the number of elements in the vector. Calculate the vector dot product value and vector norm based on the initial feature vector and the final feature vector. The calculation formula of the vector dot product value is: The calculation formula for the vector norm is: The convergence value of the node feature vector is calculated based on the vector dot product value and the vector norm: SIM = (1 + FS) / 2. The diffusion convergence index is calculated for each node feature vector. The calculation expression is:

[0070] It should be noted that the dimension of the node feature vector usually refers to the number of features representing the order node. Each order node can be represented as a feature vector, which contains various information about the order. The dimension determines the length of this feature vector, that is, the number of features contained in the vector. In online car-hailing orders, the dimension of the node feature vector may include the following example features, spatiotemporal information, user information, vehicle information, environmental factors, etc. The choice of dimension depends on the requirements of the task and model, and the specific dimension is determined according to the actual situation.

[0071] The validation error deviation value during model training affects the accuracy of analyzing order matching intelligent management. The validation error deviation value indicates the difference between the order matching situation predicted by the model and the actual order matching situation. The validation error deviation value is calculated when evaluating the model on the validation set. It indicates the overall prediction error of the model for the samples in the validation set on the regression problem, and reflects the prediction accuracy of the model for the order matching situation. The validation error deviation value affects the following aspects:

[0072] Prediction accuracy of the model: The validation error deviation value is the difference between the model's prediction value and the true value, so it directly reflects the model's prediction accuracy for the target variable. A smaller validation error deviation value indicates that the model has more accurate predictions on the validation set.

[0073] Effect of feature selection and: If the validation error deviation value is large, it may be that the model does not make good use of the input features, or the feature selection effect is not good. It is necessary to improve the model performance by adjusting the feature set, so as to make targeted improvements and optimizations.

[0074] The logic for obtaining the verification error deviation value is as follows:

[0075] Get the number of samples n for model training, and get the predicted value xc and the true value yc of the order matching after training, and calculate the predicted mean by summarizing and the true mean Calculate the correlation value and error value. The correlation value calculation formula is: The error value calculation expression is: The verification error deviation value is calculated based on the correlation value and the error value, and the calculation expression is: YZW = |Cov(x, y)| / WC.

[0076] The regional order matching value has an impact on the analysis of order matching intelligent management. The regional order matching value indicates the relationship between the number of orders successfully matched by the online car-hailing platform and the total number of orders assigned to drivers in the region and the corresponding time. This indicator measures the order matching effect of the online car-hailing system in a specific area. The regional order matching value reflects the following benefits:

[0077] Matching effect evaluation: The regional order matching value reflects the proportion of successfully matched orders in a specific area. A high matching value means that the platform successfully matches more orders in the area, which can be used as an indicator of good matching effect;

[0078] Time dimension analysis: Considering the corresponding time situation, the regional order matching value can be associated with time, such as hourly, daily or weekly analysis. This helps to identify the changes in order matching effects in different time periods and helps the platform optimize the matching strategy to adapt to the periodic demand fluctuations;

[0079] Optimize order allocation strategy: Low regional order matching values ​​may indicate poor matching results. The platform can adjust the matching algorithm and improve the driver allocation strategy based on the analysis results to improve the order matching results in a specific area.

[0080] By comprehensively analyzing the regional order matching values ​​and the corresponding time based on the regional order matching values, the online ride-hailing platform can have a more comprehensive and in-depth understanding of the status of order matching intelligent management, thereby improving and optimizing system operations in a targeted manner.

[0081] The logic for obtaining the regional order matching value is as follows:

[0082] Get the allocation time of each order and the time when the driver accepts the order within a unit of time, and get the actual number of orders JD 接单 and the actual allocated order quantity JD 实际 , calculate the order value: JZ = JD 接单 / JD 实际 For successfully accepted orders, obtain the total matching order duration ZS of all successfully accepted orders, and calculate the matching order duration value: SC = ZS / JD 实际 , get the matching order duration value of historical time Calculate the regional order matching value, the calculation expression is:

[0083] It should be noted that the unit time is the daytime and nighttime time periods set according to actual needs, and the historical time is also set according to needs, such as the same day in the month, weekends, etc.

[0084] Analyze order demand forecasts based on node information diffusion, verification error information, and regional stability information, and obtain the accuracy of order intelligent matching based on the analysis results;

[0085] The obtained diffusion convergence index, verification error deviation value, and regional order matching value are used to generate the order matching management coefficient, and the order matching management coefficient is calibrated as G x , based on the formula: In the formula, α, β, and γ are the preset proportional coefficients of the diffusion convergence index, the verification error deviation value, and the regional order matching value, respectively, and α, β, and γ are all greater than 0;

[0086] It should be noted that the formula is obtained by technical personnel in this field by collecting multiple groups of sample data and setting corresponding preset proportional coefficients for each group of sample data. The specific calculation process is: substituting the obtained diffusion convergence index, verification error deviation value, and regional order matching value into the formula to form an equation group, and then screening and averaging the calculated preset proportional coefficients to obtain the preset proportional coefficients. Since each group of sample data is different, the specific values ​​of the preset proportional coefficients under different sample data are also different.

[0087] It can be seen from the formula that the larger the diffusion convergence index, the smaller the verification error deviation value, and the larger the regional order matching value, that is, the larger the performance value of the order matching management coefficient, the better the adaptability of the order matching, and the more efficient and accurate the order matching is, which improves the quality and operational efficiency of the online car-hailing service. The smaller the diffusion convergence index, the larger the verification error deviation value, and the smaller the regional order matching value, that is, the smaller the performance value of the order matching management coefficient, the worse the adaptability of the order matching is. The online car-hailing service has poor adaptability to the matched orders and is more likely to have empty loads, thereby affecting customer experience and reducing cost-effectiveness.

[0088] A larger diffusion convergence index indicates a faster approach to the optimal solution. When generating orders, the best order allocation plan can be found more quickly, which improves the efficiency of order matching and reduces the possibility of empty loads or empty trips.

[0089] A smaller validation error deviation value indicates that the model performs better on unseen data, which helps improve prediction accuracy and ensures more accurate prediction of order demand in actual operations, thereby improving the quality of order matching;

[0090] A larger regional order matching value indicates that order management can more effectively meet the needs of different regions. Improving the regional order matching value helps optimize vehicle scheduling, reduce empty driving rate, and increase vehicle utilization, thereby improving service quality and operational efficiency.

[0091] Compare the generated order matching management coefficient with the matching management threshold, and generate a control management signal and a normal matching signal based on the comparison result;

[0092] After obtaining the generated order matching management coefficient, the generated order matching management coefficient is compared with the matching management threshold. If the order matching management coefficient is greater than the matching management threshold, a normal matching signal is generated, indicating that the user's ordering situation is compatible with the online car-hailing order situation, and the supply and demand between the user and the online car-hailing are met;

[0093] If the order matching management coefficient is less than or equal to the matching management threshold, the training model is generated as a regulation management signal, indicating that there is a problem with the order management and that timely regulation management is required. Relevant personnel should be notified to manage the order and vehicles should be allocated through manual or other means to reduce the situation where users travel without cars and online ride-hailing vehicles run empty.

[0094] It should be noted that the relevant threshold information in this embodiment is pre-set by professionals and will not be explained in detail here.

[0095] The present invention obtains order-related data generated by the online ride-hailing vehicle and cleans the data when matching the online ride-hailing vehicle with orders on the platform, and normalizes the cleaned order data as a sample for model training. The order matching situation of the online ride-hailing vehicle is then analyzed by obtaining the order coverage detail information generated in the process of order matching management. The matching of the online ride-hailing vehicle orders is timely regulated and managed according to the adaptability of the order matching, thereby reducing the situation where vehicles run empty and users cannot get a vehicle, and improving the supply and demand adaptation between users and online ride-hailing vehicles.

[0096] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0097] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0098] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0099] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent management method for autonomous order matching of online ride-hailing vehicles, characterized in that: The steps include: When the platform matches online ride-hailing vehicles with orders, it obtains order-related data generated by the online ride-hailing vehicles and performs data cleaning, and normalizes the cleaned order data; The order data is processed through a convolutional neural network and the model is trained to obtain the order coverage details generated during the order matching management process. The order coverage details include node information diffusion, verification error information, and regional stability information. The node information diffusion, verification error information and regional stability information are used to generate the order matching management coefficient and analyze the order matching of online ride-hailing vehicles to determine the adaptability of order matching; Manage the matching of online ride-hailing orders based on the suitability of order matching; The node information diffusion includes the diffusion convergence index, the verification error information includes the verification error deviation value, and the regional stability information includes the regional order matching value; The logic for obtaining the diffusion convergence index is as follows: Obtain the number of node feature vectors N when the graph convolutional network performs information transmission, obtain the initial feature vector of the node in each information transmission process, and establish the initial vector set OFi = {OF1, OF2, ..., OFm}, obtain the corresponding final feature vector, and establish the final feature vector set FFi = {FF1, FF2, ..., FFm}, where m represents the dimension of the vector and the number of elements in the vector. Calculate the vector dot product value and vector norm based on the initial feature vector and the final feature vector. The calculation formula of the vector dot product value is: The calculation formula for the vector norm is: The convergence value of the node feature vector is calculated based on the vector dot product value and the vector norm: SIM = (1 + FS) / 2. The diffusion convergence index is calculated for each node feature vector. The calculation expression is: The logic for obtaining the verification error deviation value is as follows: Get the number of samples n for model training, and get the predicted value xc and the true value yc of the order matching after training, and calculate the predicted mean by summarizing and the true mean Calculate the correlation value and error value. The correlation value calculation formula is: The error value calculation expression is: The verification error deviation value is calculated based on the correlation value and the error value, and the calculation expression is: YZW = |Cov(x, y)| / WC; The logic for obtaining the regional order matching value is as follows: Get the allocation time of each order and the time when the driver accepts the order within a unit of time, and get the actual number of orders JD 接单 and the actual allocated order quantity JD 实际 , calculate the order value: JZ = JD 接单 / JD 实际 For successfully accepted orders, obtain the total matching order duration ZS of all successfully accepted orders, and calculate the matching order duration value: SC = ZS / JD 实际 , get the matching order duration value of historical time Calculate the regional order matching value, the calculation expression is:

2. According to claim 1, a method for intelligent management of autonomous order matching for online car-hailing, characterized in that: When the platform matches online ride-hailing vehicles with orders, it obtains order-related data generated by the online ride-hailing vehicles and performs data cleaning, and normalizes the cleaned order data, including: Use data processing tools or programming languages ​​to scan the data for duplicate data in order-related data, identify duplicate records, select to mark the duplicate data, and leave one copy; Use statistical methods or visualization tools to identify missing values ​​in the data set and delete or fill in missing data in order-related data; Extract features from the cleaned data and scale the features of the order data; Calculate the minimum and maximum values ​​of the features and normalize each feature.

3. According to claim 2, a method for intelligent management of autonomous order matching for online car-hailing, characterized in that: The node information diffusion, verification error information and regional stability information are used to generate the order matching management coefficient and analyze the order matching situation of online car-hailing to determine the adaptability of order matching. The specific steps include: The obtained diffusion convergence index, verification error deviation value, and regional order matching value are used to generate an order matching management coefficient; Compare the generated order matching management coefficient with the matching management threshold; If the order matching management coefficient is greater than the matching management threshold, a normal matching signal is generated; If the order matching management coefficient is less than or equal to the matching management threshold, the training model is generated as a regulation management signal to notify personnel to perform management regulation.