Automobile rental intelligent management system and platform

Abstract

The invention relates to the technical field of data processing, and particularly discloses an intelligent management system and platform for car rental, and the method comprises an obtaining module which obtains rental car data, predetermined rental data, historical rental data and car rental behavior feature vectors; the calculation module is used for calculating a standardized behavior vector of the historical rental sub-data of which each car rental tag is completed in the historical rental data, and determining behavior clustering data of a plurality of clustering tags; the clustering module is used for calculating a clustering independent variable set of each clustering label and a significant weight, a clustering range or a dominant clustering set of each independent variable in the clustering independent variable set; the construction module is used for constructing a lease recommendation model; and the management module is used for realizing intelligent management of automobile leasing. The method can adapt to lease demands and performance capabilities of different user groups, realizes customized lease recommendation, optimizes guarantee deposit collection and vehicle deployment strategies, reduces resource idleness and performance disputes, and gives consideration to recommendation accuracy, risk controllability and operation efficiency.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent management system and platform for car rental. Background Technology

[0002] The traditional car rental industry has long relied on manual experience for management, initially using paper records to manage vehicle information and order data, which was inefficient and prone to errors. With the popularization of information technology, the industry gradually introduced simple management systems, but these systems only achieved electronic data storage and failed to effectively integrate and analyze rental vehicle data, reservation data, and historical rental data. This resulted in problems such as high vehicle idle rates and low matching of user needs. In recent years, big data and cluster analysis technologies have developed rapidly, and the car rental industry urgently needs an intelligent management system capable of deeply mining the value of data to address the pain points of traditional management models and drive the industry's transformation towards intelligentization.

[0003] Therefore, this invention proposes an intelligent management system and platform for car rental. Summary of the Invention

[0004] This invention provides an intelligent car rental management system and platform. By acquiring rental vehicle data, reservation rental data, historical rental data, and car rental behavior feature vectors, it calculates standardized behavior vectors for each rental tag in the historical rental data that represents a completed historical rental sub-data. It then determines the behavior clustering data for multiple cluster tags, calculates the set of clustering independent variables for each cluster tag, and determines the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables. This constructs a rental recommendation model, enabling intelligent management of car rentals. It can adapt to the rental needs and fulfillment capabilities of different user groups, achieving customized output of recommended vehicle tags and deposits. This avoids high-risk orders, optimizes deposit collection and vehicle allocation strategies, reduces resource idleness and fulfillment disputes, and balances recommendation accuracy, risk controllability, and operational efficiency, providing an integrated solution for intelligent management and control of the entire car rental process.

[0005] This invention provides an intelligent car rental management system, comprising: Acquisition Module: Acquires rental vehicle data, reservation rental data, and historical rental data from car rental service providers, and obtains rental behavior feature vectors for rental management; Calculation module: Based on the historical rental data of car rental service providers and the rental behavior feature vector of rental management, calculate the standardized behavior vector of each rental tag in the historical rental data as the completed historical rental sub-data, and determine the behavior cluster data of multiple cluster tags; Clustering module: Based on historical rental data and behavioral clustering data for each cluster label, calculate the set of clustering independent variables for each cluster label, as well as the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables; Building Module: Based on historical rental data, behavioral clustering data of all cluster labels, the set of clustering independent variables, and the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables, a rental recommendation model is built. Management module: Based on the rental vehicle data, reservation rental data and rental recommendation model of the car rental service provider, it realizes intelligent management of car rental.

[0006] Preferably, a car rental intelligent management system includes an acquisition module, comprising: Rental Vehicle Data Unit: Acquire rental vehicle data from car rental service providers. Rental vehicle data includes a set of rental vehicle models and rental data for each model in the rental vehicle set. Among them, the model rental data includes multiple rental vehicles and basic vehicle data, status tags, and vehicle status data for each rental vehicle. The basic vehicle data includes at least the vehicle number, vehicle age, and vehicle tags. Reservation Rental Data Unit: Acquire reservation rental data from car rental service providers. This data includes basic user information and usage tags for multiple reservation orders. Basic user information includes name, age, gender, driving experience, and address. Historical Rental Data Unit: Acquire historical rental data from car rental service providers. This historical rental data includes multiple historical rental data, which includes user basic information, usage tags, rental deposit, rental tags, vehicle number, user rental data, and historical driving data. Rental tags include cancellation and completion. Car rental behavior feature vector unit: Obtain the car rental behavior feature vector of rental management and the feature range of each behavior feature in the car rental behavior feature vector.

[0007] Preferably, a car rental intelligent management system includes a rental vehicle data unit and a historical rental data unit, comprising: First vehicle status data sub-unit: If the status tag is idle, the vehicle status data shall include at least the location, remaining battery power, remaining fuel, and accumulated mileage; The second vehicle status data sub-unit: If the status tag is "rented", the vehicle status data includes real-time driving data, driving behavior data and rental basic data; The third vehicle status data sub-unit: If the status label is "under maintenance", the vehicle status data shall at least include the location, maintenance progress, estimated operating time, and maintenance cost; User car rental data and historical driving data sub-units: When the car rental tag is canceled, the user car rental data and historical driving data are empty. When the car rental tag is completed, the user car rental data includes the actual pick-up location, the actual drop-off location, the driving distance, the driving route, the driving time interval, and the final payment settlement time. The driving time interval includes multiple driving time sub-intervals and the driving duration of each driving time sub-interval.

[0008] Preferably, a car rental intelligent management system includes a computing module comprising: Historical behavior value vector unit: Based on the car rental behavior feature vector, features are extracted from the historical driving data of each car rental sub-data where the car rental tag is not completed and the driving distance in the user car rental data to determine the historical behavior value vector of each car rental sub-data where the car rental tag is not completed. Driving difficulty score unit: Input the actual pick-up location, actual drop-off location, and driving route of the user car rental data in each historical rental sub-data with the car rental tag being completed into the route evaluation model to determine the driving difficulty score of each historical rental sub-data with the car rental tag being completed. Peak Unit: For each historical rental data segment with a rental tag indicating completion, perform peak judgment on each driving time sub-interval within the user rental data segment of the historical rental data segment with a rental tag indicating completion, and determine the peak interval and peak duration of each driving time sub-interval within the driving time interval of the user rental data segment of the historical rental data segment with a rental tag indicating completion. Nighttime Unit: For each rental period in the historical rental data where the rental tag is "completed", perform nighttime judgment on each driving time sub-interval in the driving time interval of the user rental data within the historical rental data where the rental tag is "completed" to determine the nighttime interval and nighttime duration of each driving time sub-interval in the driving time interval of the user rental data within the historical rental data where the rental tag is "completed". Peak-Night Unit: Based on the peak and night intervals of the driving time sub-intervals in the user car rental data of each historical rental sub-data where the car rental tag is completed, determine the peak-night duration of each driving time sub-interval in the driving time intervals of the user car rental data of each historical rental sub-data where the car rental tag is completed; Standardized Behavior Vector Unit: Based on the historical behavior value vector of each rental sub-data with a completed rental tag in the historical rental data, the driving difficulty score, the peak interval of all driving time sub-intervals in the driving time interval in the user's rental data, and the feature range of each behavior feature in the rental behavior feature vector, the standardized behavior vector of each rental sub-data with a completed rental tag in the historical rental data is calculated. Clustering Unit: Perform cluster analysis on the standardized behavioral vectors of all historical rental sub-data with the rental tag "completed" in the historical rental data to determine multiple behavioral clusters and cluster labels for each behavioral cluster. The behavioral clusters include behavioral cluster vectors and multiple historical rental sub-data with the rental tag "completed".

[0009] Preferably, a car rental intelligent management system includes a clustering module, comprising: Urban congestion label unit: Based on the address in the user's basic information for each completed historical rental sub-data in the historical rental data, the urban congestion label for each completed historical rental sub-data in the historical rental data is determined. The urban congestion label includes severe congestion, moderate to severe congestion, mild congestion, and smooth traffic. Behavioral Feature Value Vector Unit: Based on the standardized behavioral vector of each historical rental sub-data of the behavioral clustering data for each cluster label, the behavioral feature value vector of each behavioral feature for each cluster label is determined; The first determining unit: Based on the age, gender, driving experience, and city congestion labels of all historical rental sub-data of the behavioral clustering data for each cluster label, determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector for each cluster label; Significance probability unit: Determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector of each cluster label as independent variables, and determine the behavioral feature value vector of each behavioral feature of each cluster label as dependent variables. Perform one-way ANOVA to determine the significance probability of each independent variable and each dependent variable of each cluster label. Correction label unit: Based on the significance probabilities of all independent and dependent variables for each cluster label, determine the correction label for each independent and dependent variable for each cluster label; Clustering Independent Variable Set Unit: Based on the significance probabilities of all independent and dependent variables for each cluster label and the correction label, calculate the clustering independent variable set for each cluster label and the significance weight of each independent variable in the clustering independent variable set; Clustering range or dominant cluster set: Based on each independent variable in the clustering independent variable set of each cluster label, all values ​​of each independent variable in the historical rental sub-data of all car rental labels in the behavior clustering data of each cluster label, age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector, determine the clustering range or dominant cluster set of each independent variable in the clustering independent variable set of each cluster label.

[0010] Preferably, a car rental intelligent management system includes clustering ranges or dominant clustering units, comprising: Age or driving experience sub-unit: If the independent variable in the clustering independent variable set of the clustering label is age or driving experience, based on the age feature vector or driving experience feature vector of the clustering label, determine the clustering range in the clustering independent variable set of the clustering label where the independent variable is age or driving experience. Gender sub-unit: If the independent variable in the clustering independent variable set of the clustering label is gender, based on all the values ​​of gender in the historical rental sub-data of all car rental tags that are completed in the behavior clustering data of the clustering label and the gender feature vector, calculate the discrete proportion of each gender in the gender feature vector in the clustering independent variable set of the clustering label. If the discrete proportion of gender in the gender feature vector in the clustering independent variable set of the clustering label is greater than a preset threshold, determine that the gender in the gender feature vector in the clustering independent variable set of the clustering label is the dominant gender set with gender as the independent variable in the clustering independent variable set of the clustering label; otherwise, determine that the dominant cluster set with gender as the independent variable in the clustering independent variable set of the clustering label is the gender feature vector. Urban congestion label subunit: If the independent variable in the clustering independent variable set of the clustering label is the urban congestion label, based on all values ​​of the urban congestion label in the historical rental subdata of all car rental labels completed in the behavior clustering data of the clustering label and the congestion feature vector, calculate the discrete proportion of each urban congestion label in the congestion feature vector in the clustering independent variable set of the clustering label. Sort the discrete proportions of all urban congestion labels in the congestion feature vector in the clustering independent variable set of the clustering label from largest to smallest. Select the smallest position greater than a preset threshold from the sorted proportion sequence as the congestion position of the urban congestion label in the clustering independent variable set of the clustering label. Determine the urban congestion label corresponding to the first few discrete proportions in the proportion sequence, which is the dominant congestion set of the dominant clustering set in the clustering independent variable set of the clustering label as the congestion feature vector.

[0011] Preferably, a car rental intelligent management system includes the following building modules: Historical cancellation data unit: Based on the clustering range or dominant cluster set of all independent variables in the clustering independent variable set of each cluster label, and the user basic information of all historical rental sub-data with cancellation tags in the historical rental data, the historical rental sub-data with cancellation tags in the historical rental data is divided to determine the historical cancellation data for each cluster label; Cancellation rate unit: Based on historical cancellation data and behavioral clustering data for each cluster label, determine the cancellation rate for each cluster label; Clustering training data unit: Based on the rental deposit, purpose label, rental label in all completed historical rental sub-data of the behavioral clustering data of each cluster label, and the final payment end time in the user's rental data, the clustering training data for each cluster label is determined; Clustering Recommendation Model Unit: Based on the clustering training data for each cluster label, construct a clustering recommendation model for each cluster label; Rental recommendation model unit: Based on the clustering recommendation model of all clustering labels, the behavior clustering vector in the behavior clustering data, the set of clustering independent variables, the cancellation rate, the significance weight of all independent variables in the set of clustering independent variables, the clustering range or the dominant cluster set, a rental recommendation model is constructed.

[0012] Preferably, a car rental intelligent management system includes a management module comprising: Reservation Recommendation Unit: Input the user's basic information and purpose tags for each reservation order in the car rental service provider's reservation rental data into the rental recommendation model to determine the recommended vehicle tag and recommended deposit for each reservation order in the car rental service provider's reservation rental data; Management Unit: Based on the rental vehicle data of the car rental service provider and the recommended vehicle tags and recommended deposits of all reservation orders in the reservation rental data, intelligent management of car rental is achieved.

[0013] This invention provides an intelligent car rental management platform for executing any one of the intelligent car rental management systems in Examples 1 to 8.

[0014] The beneficial effects of this invention compared to existing technologies are as follows: By acquiring rental vehicle data, pre-booked rental data, historical rental data, and rental behavior feature vectors, this invention calculates standardized behavior vectors for each rental tag in the historical rental data that represents a completed historical rental sub-data. It then determines the behavior clustering data for multiple clustering tags, calculates the set of clustering independent variables for each clustering tag, and determines the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables. This constructs a rental recommendation model, enabling intelligent management of car rentals. It can adapt to the rental needs and fulfillment capabilities of different user groups, achieving customized output of recommended vehicle tags and deposits, avoiding high-risk orders, optimizing deposit collection and vehicle allocation strategies, reducing resource idleness and fulfillment disputes, and balancing recommendation accuracy, risk controllability, and operational efficiency. This provides an integrated solution for intelligent management and control of the entire car rental process.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of an intelligent car rental management system according to an embodiment of the present invention. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention. Example 1:

[0019] This invention provides an intelligent management system for car rental, with reference to... Figure 1 ,include: Acquisition Module: Acquires rental vehicle data, reservation rental data, and historical rental data from car rental service providers, and obtains rental behavior feature vectors for rental management; Calculation module: Based on the historical rental data of car rental service providers and the rental behavior feature vector of rental management, calculate the standardized behavior vector of each rental tag in the historical rental data as the completed historical rental sub-data, and determine the behavior cluster data of multiple cluster tags; Clustering module: Based on historical rental data and behavioral clustering data for each cluster label, calculate the set of clustering independent variables for each cluster label, as well as the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables; Building Module: Based on historical rental data, behavioral clustering data of all cluster labels, the set of clustering independent variables, and the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables, a rental recommendation model is built. Management module: Based on the rental vehicle data, reservation rental data and rental recommendation model of the car rental service provider, it realizes intelligent management of car rental.

[0020] In this embodiment, three types of key business data and one core analytical basis are acquired. The three types of key business data are rental vehicle data, reservation rental data, and historical rental data of car rental service providers. Rental vehicle data covers information related to all available rental vehicles under the service provider. Reservation rental data covers all rental reservation order information submitted by users. Historical rental data covers detailed records of all completed or canceled rental orders in the past. The core analytical basis is the car rental behavior feature vector of rental management, which contains various core indicators that can reflect the characteristics of users' car rental behavior.

[0021] In this embodiment, based on the historical rental data of car rental service providers and the car rental behavior feature vector of rental management, the system first processes all historical rental sub-data with incomplete car rental tags in the historical rental data, transforming these sub-data into standardized behavior vectors. Then, based on the standardized behavior vectors, cluster analysis is performed to group historical rental sub-data with similar car rental behavior characteristics into one category. Finally, multiple cluster labels and the behavior cluster data corresponding to each cluster label are determined. Each behavior cluster data represents a group of users with similar driving behaviors.

[0022] In this embodiment, by combining historical rental data and behavioral clustering data for each cluster label, the set of clustering independent variables corresponding to each cluster label is first determined through analysis and calculation of these two types of data. This set contains all factors that have a significant impact on the car rental behavior of users in that cluster. Then, for each independent variable in the set of clustering independent variables, its corresponding significant weight cluster range or dominant cluster set is calculated. The significant weight reflects the degree of influence of the independent variable on the clustering result, the cluster range clarifies the value range of the independent variable, and the dominant cluster set defines the core value category of the independent variable.

[0023] In this embodiment, multi-dimensional data and parameters are integrated to build a model capable of accurate rental recommendations. The operation of this module relies on multiple core data and parameters, including behavioral clustering data of all cluster labels in historical rental data, the set of clustering independent variables for each cluster label, and the significant weight clustering range or dominant cluster set of each independent variable in the set of clustering independent variables. Through the integrated analysis of these data and parameters, the inherent correlation between user car rental behavior characteristics and vehicle resource matching is explored, and finally a rental recommendation model is constructed. This model can output accurate vehicle recommendation results based on relevant user information, providing a decision-making basis for intelligent management of car rental.

[0024] In this embodiment, the rental recommendation model is combined with actual business data to achieve intelligent management of the entire car rental business process. The operation of this module relies on three types of core data: rental vehicle data of car rental service providers, reservation rental data, and the completed rental recommendation model. By inputting reservation rental data into the rental recommendation model, recommendation results are obtained for each reservation order. Combined with the actual situation of rental vehicle data, multiple management functions such as intelligent scheduling of vehicle resources, accurate matching of orders, and reasonable setting of deposits are realized.

[0025] The beneficial effects of the above technologies are as follows: By acquiring rental vehicle data, reservation rental data, historical rental data, and rental behavior feature vectors, the standardized behavior vectors for each rental tag in the historical rental data are calculated for completed historical rental sub-data. This determines the behavior clustering data for multiple cluster tags, calculates the set of clustering independent variables for each cluster tag, and the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables. A rental recommendation model is then constructed to achieve intelligent management of car rentals. This model can adapt to the rental needs and fulfillment capabilities of different user groups, enabling customized output of recommended vehicle tags and deposits. It avoids high-risk orders, optimizes deposit collection and vehicle allocation strategies, reduces resource idleness and fulfillment disputes, and balances recommendation accuracy, risk controllability, and operational efficiency, providing an integrated solution for intelligent management and control of the entire car rental process. Example 2:

[0026] Based on Embodiment 1, a car rental intelligent management system includes an acquisition module, comprising: Rental Vehicle Data Unit: Acquire rental vehicle data from car rental service providers. Rental vehicle data includes a set of rental vehicle models and rental data for each model in the rental vehicle set. Among them, the model rental data includes multiple rental vehicles and basic vehicle data, status tags, and vehicle status data for each rental vehicle. The basic vehicle data includes at least the vehicle number, vehicle age, and vehicle tags. Reservation Rental Data Unit: Acquire reservation rental data from car rental service providers. This data includes basic user information and usage tags for multiple reservation orders. Basic user information includes name, age, gender, driving experience, and address. Historical Rental Data Unit: Acquire historical rental data from car rental service providers. This historical rental data includes multiple historical rental data, which includes user basic information, usage tags, rental deposit, rental tags, vehicle number, user rental data, and historical driving data. Rental tags include cancellation and completion. Car rental behavior feature vector unit: Obtain the car rental behavior feature vector of rental management and the feature range of each behavior feature in the car rental behavior feature vector.

[0027] In this embodiment, rental vehicle data from car rental service providers is obtained. This rental vehicle data includes two core components: a set of rental vehicle models and rental data for each model within the rental vehicle set. The rental data for each model further covers multiple specific rental vehicles. For each rental vehicle, three types of key data need to be collected: basic vehicle data, status tags, and vehicle status data. The basic vehicle data must include at least three core pieces of information: vehicle number, vehicle age, and vehicle tag. The rental vehicle set includes sedans, SUVs, commercial vehicles, and new energy vehicles, etc. The vehicle number refers to a code used to uniquely identify each rental vehicle, which can be a license plate number. The vehicle age refers to the number of years the rental vehicle has been in use from its manufacturing date to the current date. The vehicle tag refers to the vehicle brand and model.

[0028] In this embodiment, the reservation rental data of the car rental service provider is obtained. The reservation rental data covers the relevant information of multiple reservation orders. The information of each reservation order is divided into two main categories: basic user information and usage tags. The basic user information is the key content to characterize the identity and qualifications of the reservation user, specifically including five items: name, age, gender, driving experience, and address. This information can help the rental service provider understand the user's situation in advance and provide a reference for subsequent order review and vehicle matching.

[0029] In this embodiment, historical rental data from car rental service providers is obtained. This historical rental data is compiled from historical rental sub-data generated from multiple historical car rental transactions. Each historical rental sub-data contains rich content, including user basic information, purpose tags, rental deposit, rental tags, vehicle number, user rental data, and historical driving data. Among them, the rental tags are key identifiers used to identify the final status of historical car rental orders, specifically including two categories: cancellation and completion. Purpose tags include business reception, commuting, self-driving tours, wedding cars, etc. The rental deposit refers to the deposit paid by the user to the service provider when renting a vehicle.

[0030] In this embodiment, the car rental behavior feature vector required in the rental management process is obtained, and the feature range corresponding to each behavior feature in the car rental behavior feature vector is also obtained. The car rental behavior feature vector includes average relative speed limit deviation, throttle depth, braking depth, average number of rapid accelerations, average number of rapid decelerations, average number of sharp turns, maximum deceleration, maximum longitudinal acceleration, etc.

[0031] The beneficial effects of the above technologies are: obtaining rental vehicle data, reservation rental data, and historical rental data from car rental service providers, and obtaining rental behavior feature vectors for rental management, can improve the availability and accuracy of rental data. Example 3:

[0032] Based on Example 2, a smart car rental management system includes a rental vehicle data unit and a historical rental data unit, comprising: First vehicle status data sub-unit: If the status tag is idle, the vehicle status data shall include at least the location, remaining battery power, remaining fuel, and accumulated mileage; The second vehicle status data sub-unit: If the status tag is "rented", the vehicle status data includes real-time driving data, driving behavior data and rental basic data; The third vehicle status data sub-unit: If the status label is "under maintenance", the vehicle status data shall at least include the location, maintenance progress, estimated operating time, and maintenance cost; User car rental data and historical driving data sub-units: When the car rental tag is canceled, the user car rental data and historical driving data are empty. When the car rental tag is completed, the user car rental data includes the actual pick-up location, the actual drop-off location, the driving distance, the driving route, the driving time interval, and the final payment settlement time. The driving time interval includes multiple driving time sub-intervals and the driving duration of each driving time sub-interval.

[0033] In this embodiment, for rental vehicles with the status label of "idle", the range of vehicle status data corresponding to them is defined. The vehicle status data in this state must include at least four core pieces of information: location, remaining battery power, remaining fuel, and accumulated mileage.

[0034] In this embodiment, for rental vehicles with the status tag "rented", the corresponding vehicle status data range is defined. The vehicle status data in this state includes three core categories: real-time driving data, driving behavior data, and rental basic data. This data can help rental service providers monitor the vehicle's operating status and the user's driving behavior in real time during the rental period, and promptly detect abnormal situations to ensure the safety of the vehicle and the user.

[0035] In this embodiment, for rental vehicles with the status label "under repair", the corresponding vehicle status data range is defined. The vehicle status data in this state must include at least four core pieces of information: location, repair progress, estimated operating time, and repair cost. This information can help rental service providers to keep track of the vehicle's repair progress and cost in real time, and rationally plan the time node for the vehicle to return to operation to optimize the overall scheduling efficiency of rental vehicles.

[0036] In this embodiment, for the historical rental sub-data of different car rental tags, the value rules for user car rental data and historical driving data are clearly defined. When the car rental tag is canceled, the corresponding user car rental data and historical driving data are empty because no actual car rental behavior has occurred for this type of order. When the car rental tag is completed, the corresponding user car rental data contains rich content, including the actual pick-up location, the actual return location, the driving distance, the driving route, the driving time interval, and the final payment settlement time. The driving time interval is further refined into multiple driving time sub-intervals and the driving duration corresponding to each driving time sub-interval.

[0037] The beneficial effects of the above technologies are: obtaining rental vehicle data and historical rental data from car rental service providers can further improve the availability and accuracy of rental data, providing standardized and high-value data support for subsequent intelligent analysis and recommendations. Example 4:

[0038] Based on Embodiment 2, a car rental intelligent management system includes a computing module, comprising: Historical behavior value vector unit: Based on the car rental behavior feature vector, features are extracted from the historical driving data of each car rental sub-data where the car rental tag is not completed and the driving distance in the user car rental data to determine the historical behavior value vector of each car rental sub-data where the car rental tag is not completed. Driving difficulty score unit: Input the actual pick-up location, actual drop-off location, and driving route of the user car rental data in each historical rental sub-data with the car rental tag being completed into the route evaluation model to determine the driving difficulty score of each historical rental sub-data with the car rental tag being completed. Peak Unit: For each historical rental data segment with a rental tag indicating completion, perform peak judgment on each driving time sub-interval within the user rental data segment of the historical rental data segment with a rental tag indicating completion, and determine the peak interval and peak duration of each driving time sub-interval within the driving time interval of the user rental data segment of the historical rental data segment with a rental tag indicating completion. Nighttime Unit: For each rental period in the historical rental data where the rental tag is "completed", perform nighttime judgment on each driving time sub-interval in the driving time interval of the user rental data within the historical rental data where the rental tag is "completed" to determine the nighttime interval and nighttime duration of each driving time sub-interval in the driving time interval of the user rental data within the historical rental data where the rental tag is "completed". Peak-Night Unit: Based on the peak and night intervals of the driving time sub-intervals in the user car rental data of each historical rental sub-data where the car rental tag is completed, determine the peak-night duration of each driving time sub-interval in the driving time intervals of the user car rental data of each historical rental sub-data where the car rental tag is completed; Standardized Behavior Vector Unit: Based on the historical behavior value vector of each rental sub-data with a completed rental tag in the historical rental data, the driving difficulty score, the peak interval of all driving time sub-intervals in the driving time interval in the user's rental data, and the feature range of each behavior feature in the rental behavior feature vector, the standardized behavior vector of each rental sub-data with a completed rental tag in the historical rental data is calculated. Clustering Unit: Perform cluster analysis on the standardized behavioral vectors of all historical rental sub-data with the rental tag "completed" in the historical rental data to determine multiple behavioral clusters and cluster labels for each behavioral cluster. The behavioral clusters include behavioral cluster vectors and multiple historical rental sub-data with the rental tag "completed".

[0039] In this embodiment, the feature vector of car rental behavior is used as the basis and standard for extraction. For all historical rental sub-data with the car rental tag as completed in the historical rental data, the feature extraction focuses on two key data types: historical driving data and driving distance in the user's car rental data. The historical behavior value vector corresponding to each historical rental sub-data with the car rental tag as completed is determined, which is a quantitative characterization of a user's single completed car rental behavior.

[0040] In this embodiment, three key pieces of information are selected from the user's rental data in each rental sub-data where the rental tag is not completed: the actual pick-up location, the actual drop-off location, and the driving route. These three pieces of information are input into a pre-built route evaluation model. Through the analysis and calculation of these route-related information by the route evaluation model, the driving difficulty score corresponding to each rental sub-data where the rental tag is not completed is finally determined. This score can intuitively reflect the complexity of the user's rental driving route.

[0041] In this embodiment, the route assessment model first structurally analyzes and extracts multi-dimensional features that affect driving difficulty, covering road and topological features such as the type of road covered by the route, the proportion of special road sections, and the degree of change of turns and slope. Combined with spatiotemporal correlation features such as historical congestion data for the corresponding driving time period, and based on the rental business scenario and a large amount of historical road conditions and user driving feedback data, differentiated weights are assigned to various features. After quantifying and assigning values ​​to each feature, the multi-dimensional features are transformed into standardized driving difficulty scores within a fixed range through weighted summation and standardization. At the same time, the feature weights and quantification rules are continuously dynamically adjusted based on new historical rental sub-data and user feedback to ensure that the score accurately matches the actual driving difficulty, and to output an objective and unified quantitative result of driving difficulty for each rental sub-data with a complete rental tag.

[0042] In this embodiment, for each historical rental data segment where the rental tag is not completed, the user rental data includes a driving time interval. This interval is further broken down into each driving time sub-interval within that driving time interval. Peak time judgment is performed for each driving time sub-interval. By judging each driving time sub-interval one by one, it is determined whether each driving time sub-interval belongs to the peak interval. At the same time, the duration of the peak interval within each driving time sub-interval is counted. Finally, the peak interval and peak duration corresponding to each driving time sub-interval within the driving time interval of the user rental data in each historical rental data segment where the rental tag is not completed are obtained.

[0043] In this embodiment, for each historical rental data segment where the rental tag is not completed, the user's rental data includes a driving time interval. The nighttime determination is carried out for each driving time sub-interval within this driving time interval. The determination is based on the time of darkness and light on the day of the trip. By judging each driving time sub-interval one by one, it is determined whether each driving time sub-interval belongs to the nighttime interval. At the same time, the duration of the nighttime interval within each driving time sub-interval is counted. Finally, the nighttime interval and nighttime duration corresponding to each driving time sub-interval within the driving time interval of the user's rental data in each historical rental data segment where the rental tag is not completed are obtained.

[0044] In this embodiment, based on the previously determined historical rental data, the peak and nighttime intervals corresponding to each driving time sub-interval within the driving time interval of the user's car rental data in each driving time sub-data where the car rental tag is completed are identified. At the same time, the duration of this period is calculated, and finally the peak-nighttime duration corresponding to each driving time sub-interval within the driving time interval of the user's car rental data in each historical rental data where the car rental tag is completed is determined.

[0045] In this embodiment, the standardized behavior vector unit calculates the standardized behavior vector for each completed historical rental sub-data item with a rental tag based on the historical behavior value vector of each rental tag in the historical rental data, the driving difficulty score, the peak interval of all driving time sub-intervals in the user's rental data, and the feature range of each behavior feature in the rental behavior feature vector. The calculation formula is expressed as follows: ; in, This represents the standardized behavior vector of the i-th car rental tag in the historical rental data, indicating that the rental was completed. N² represents the standardized behavior value of the j-th behavior feature value in the historical behavior value vector of the i-th car rental sub-data that is labeled as completed. N² represents the number of behavior feature values ​​in the historical behavior value vector. This represents the first and N2nd behavior feature values ​​in the historical behavior value vector of the i-th car rental sub-data that is a completed historical rental. These represent the j-th behavioral feature value in the historical behavioral value vector of the i-th car rental sub-data that is labeled as completed in the historical rental data. These represent the lower and upper bounds of the feature range of the j-th behavioral feature in the car rental behavior feature vector, respectively. This indicates the driving difficulty score of the i-th rental tag in the historical rental data, representing a completed historical rental sub-data set. This represents the driving duration of the k-th driving time sub-interval within the driving time interval of the user's car rental data in the i-th car rental sub-data that is marked as completed in the historical rental data. denoted as , where represents the peak duration of the k-th driving time sub-interval within the driving time interval of the user's car rental data for the i-th car rental sub-data segment with the 'completed' tag in the historical rental data; , where , where , represents the number of driving time sub-intervals within the driving time interval of the user's car rental data for the i-th car rental sub-data segment with the 'completed' tag in the historical rental data; and , where , where , represents the upper limit threshold for the driving difficulty score of the historical rental sub-data segment with the 'completed' tag in the historical rental data. This represents the nighttime duration of the k-th driving time sub-interval within the driving time interval of the user's car rental data, where the i-th car rental tag indicates a completed historical rental. This represents the peak-night duration of the k-th driving time sub-interval within the driving time interval of the user's car rental data in the i-th car rental sub-data that is marked as completed in the historical rental data. The second indicator function represents the k-th driving time sub-interval within the driving time interval of the user's car rental data, where the i-th rental tag in the historical rental data is "completed". Let α represent the j-th behavioral feature value in the historical behavioral value vector of the i-th rental sub-data where the rental tag is completed in the historical rental data. Let α represent the standardization adjustment coefficient.

[0046] In this embodiment, the standardized adjustment factor α ranges from 0.15 to 0.25, and can be 0.2.

[0047] In this embodiment, cluster analysis is performed on the standardized behavior vectors corresponding to all historical rental sub-data with incomplete rental tags in the historical rental data. Through cluster analysis, standardized behavior vectors with similar behavioral characteristics are grouped into one category, and finally multiple behavior cluster data and the cluster label corresponding to each behavior cluster data are determined. Each behavior cluster data contains two core components: a behavior cluster vector that can represent the core characteristics of the behavior category, and multiple historical rental sub-data with incomplete rental tags belonging to the category. Clustering achieves a refined classification of user car rental behavior.

[0048] The beneficial effects of the above technologies are as follows: Based on the historical rental data of car rental service providers and the car rental behavior feature vector of rental management, the standardized behavior vector of each rental tag in the historical rental data is calculated as a completed historical rental sub-data, and the behavior clustering data of multiple clustering tags is determined, so as to realize the refined stratification of user car rental behavior, provide high-value data support for rental recommendation and intelligent management, and improve the accuracy and scientific nature of behavior analysis. Example 5:

[0049] Based on Example 4, a car rental intelligent management system includes a clustering module, comprising: Urban congestion label unit: Based on the address in the user's basic information for each completed historical rental sub-data in the historical rental data, the urban congestion label for each completed historical rental sub-data in the historical rental data is determined. The urban congestion label includes severe congestion, moderate to severe congestion, mild congestion, and smooth traffic. Behavioral Feature Value Vector Unit: Based on the standardized behavioral vector of each historical rental sub-data of the behavioral clustering data for each cluster label, the behavioral feature value vector of each behavioral feature for each cluster label is determined; The first determining unit: Based on the age, gender, driving experience, and city congestion labels of all historical rental sub-data of the behavioral clustering data for each cluster label, determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector for each cluster label; Significance probability unit: Determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector of each cluster label as independent variables, and determine the behavioral feature value vector of each behavioral feature of each cluster label as dependent variables. Perform one-way ANOVA to determine the significance probability of each independent variable and each dependent variable of each cluster label. Correction label unit: Based on the significance probabilities of all independent and dependent variables for each cluster label, determine the correction label for each independent and dependent variable for each cluster label; Clustering Independent Variable Set Unit: Based on the significance probabilities of all independent and dependent variables for each cluster label and the correction label, calculate the clustering independent variable set for each cluster label and the significance weight of each independent variable in the clustering independent variable set; Clustering range or dominant cluster set: Based on each independent variable in the clustering independent variable set of each cluster label, all values ​​of each independent variable in the historical rental sub-data of all car rental labels in the behavior clustering data of each cluster label, age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector, determine the clustering range or dominant cluster set of each independent variable in the clustering independent variable set of each cluster label.

[0050] In this embodiment, a corresponding city congestion label is matched for each completed historical rental data segment with a car rental tag. The core basis for matching is the address included in the user's basic information in the historical rental data segment. This unit determines the corresponding city congestion label from four categories: severe congestion, moderate to severe congestion, mild congestion, and smooth traffic, based on the congestion situation in the area where the user's address is located, thereby achieving a precise association between the user's address and the city congestion scenario.

[0051] In this embodiment, for each cluster label, the standardized behavior vector of all historical rental sub-data in its behavior cluster data is extracted, and all values ​​of the same behavior feature are integrated and sorted out to finally form the behavior feature value vector corresponding to each behavior feature for each cluster label, thereby realizing the structured collection of behavior feature data.

[0052] In this embodiment, for each cluster label, the age, gender, driving experience information, and matched city congestion label corresponding to each sub-data are extracted from all historical rental sub-data with incomplete car rental labels contained in its behavior cluster data. The information of the same type is integrated and processed to generate the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector of the cluster label respectively.

[0053] In this embodiment, the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector of each cluster label are set as independent variables, and the behavioral feature value vector corresponding to each behavioral feature of each cluster label is set as dependent variables. One-way ANOVA is performed for each group of independent variables and dependent variables, and the significance probability between each group of variables is determined by the analysis results. The probability value intuitively reflects the degree of influence of the independent variables on the dependent variables.

[0054] In this embodiment, the calibration label unit determines the calibration label for each independent variable and each dependent variable of each cluster label based on the significance probabilities of all independent and dependent variables for each cluster label. The calculation formula is expressed as follows: ; in, This represents the saliency probability sequence of the p-th cluster label. Let Na represent the significance probabilities of the a-th independent variable and the b-th dependent variable for the p-th cluster label, N3 represent the number of independent variables, and N4 represent the number of dependent variables. Let c represent the significance probability of the c-th cluster label in the significance probability sequence. Let c represent the correction label for the c-th significance probability in the significance probability sequence of the p-th cluster label, and let b represent the correction label for the a-th independent variable and the b-th dependent variable.

[0055] In this embodiment, This indicates that the significance probabilities of all independent and dependent variables are sorted from smallest to largest.

[0056] In this embodiment, the clustering independent variable set unit: based on the significance probabilities of all independent and dependent variables for each cluster label and the correction label, calculates the clustering independent variable set for each cluster label and the significance weight of each independent variable in the clustering independent variable set. The calculation formula is expressed as follows: ; Let represent the third indicator function of the c-th significance probability correction label corresponding to the a-th independent variable and the b-th dependent variable in the significance probability sequence of the p-th cluster label. Let represent the significance weight of the a-th independent variable for the p-th cluster label. Let a represent the a-th independent variable. Let represent the set of clustering independent variables for the p-th cluster label.

[0057] The beneficial effects of the above technologies are as follows: Based on historical rental data and behavioral clustering data for each cluster label, the set of clustering independent variables for each cluster label, as well as the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables, can achieve more precise and refined clustering analysis, effectively eliminate interference from invalid variables, strengthen the intrinsic relationship between user attributes, congestion scenarios, and behavioral characteristics, provide high-precision data support for the construction of subsequent rental recommendation models, and improve the level of intelligent decision-making in rental management. Example 6:

[0058] Based on Example 5, a smart car rental management system, comprising a clustering range or dominant clustering unit, includes: Age or driving experience sub-unit: If the independent variable in the clustering independent variable set of the clustering label is age or driving experience, based on the age feature vector or driving experience feature vector of the clustering label, determine the clustering range in the clustering independent variable set of the clustering label where the independent variable is age or driving experience. Gender sub-unit: If the independent variable in the clustering independent variable set of the clustering label is gender, based on all the values ​​of gender in the historical rental sub-data of all car rental tags that are completed in the behavior clustering data of the clustering label and the gender feature vector, calculate the discrete proportion of each gender in the gender feature vector in the clustering independent variable set of the clustering label. If the discrete proportion of gender in the gender feature vector in the clustering independent variable set of the clustering label is greater than a preset threshold, determine that the gender in the gender feature vector in the clustering independent variable set of the clustering label is the dominant gender set with gender as the independent variable in the clustering independent variable set of the clustering label; otherwise, determine that the dominant cluster set with gender as the independent variable in the clustering independent variable set of the clustering label is the gender feature vector. Urban congestion label subunit: If the independent variable in the clustering independent variable set of the clustering label is the urban congestion label, based on all values ​​of the urban congestion label in the historical rental subdata of all car rental labels completed in the behavior clustering data of the clustering label and the congestion feature vector, calculate the discrete proportion of each urban congestion label in the congestion feature vector in the clustering independent variable set of the clustering label. Sort the discrete proportions of all urban congestion labels in the congestion feature vector in the clustering independent variable set of the clustering label from largest to smallest. Select the smallest position greater than a preset threshold from the sorted proportion sequence as the congestion position of the urban congestion label in the clustering independent variable set of the clustering label. Determine the urban congestion label corresponding to the first few discrete proportions in the proportion sequence, which is the dominant congestion set of the dominant clustering set in the clustering independent variable set of the clustering label as the congestion feature vector.

[0059] In this embodiment, when it is determined that the independent variable in the clustering independent variable set of the clustering label is age or driving experience, the analysis is carried out directly based on the age feature vector or driving experience feature vector corresponding to the clustering label. By sorting out the distribution of all age or driving experience values ​​in the feature vector, a reasonable range that can cover the age or driving experience of most users under the clustering label is defined. This range can be defined by calculating the mean and standard deviation, or by dividing by quantiles. This range is the clustering range in the clustering independent variable set of the clustering label where the independent variable is age or driving experience. This clustering range can accurately reflect the core characteristics of this type of user in terms of age or driving experience.

[0060] In this embodiment, when it is determined that the independent variable in the clustering independent variable set of the clustering label is gender, the gender values ​​contained in all historical rental sub-data of completed car rental tags in the behavior clustering data of the clustering label are first extracted. At the same time, the gender feature vector corresponding to the clustering label is retrieved, including male and female. Based on these two types of data, the discrete proportion of each gender in the gender feature vector in the clustering independent variable set of the clustering label is calculated. The discrete proportion directly reflects the distribution of the number of users of different genders under the clustering label. Then, the calculated discrete proportion of each gender is compared with a preset threshold. If the discrete proportion of a certain gender is greater than the preset threshold, the gender is determined as the dominant gender set in the clustering independent variable set of the clustering label where the independent variable is gender, i.e., male or female. If the discrete proportion of any gender is not greater than the preset threshold, the entire gender feature vector is determined as the dominant cluster set in the clustering independent variable set of the clustering label where the independent variable is gender, i.e., male and female. The preset threshold can be 75%.

[0061] In this embodiment, when it is determined that the independent variable in the clustering independent variable set of the clustering label is a city congestion label, the city congestion label values ​​contained in all completed historical rental sub-data of the behavior clustering data of the clustering label are first extracted. At the same time, the congestion feature vector corresponding to the clustering label is retrieved. Based on these two types of data, the discrete proportion of each city congestion label in the congestion feature vector of the clustering independent variable set of the clustering label is calculated. The discrete proportion reflects the distribution of different city congestion labels under the clustering label. Then, the discrete proportions of all city congestion labels are sorted from largest to smallest to obtain an ordered proportion sequence. Then, the smallest position with a discrete proportion greater than a preset threshold is selected from the sorted proportion sequence. This position is determined as the congestion position of the independent variable in the clustering independent variable set of the clustering label where the independent variable is a city congestion label. Finally, the city congestion labels corresponding to the first few discrete proportions in the proportion sequence are selected, and these city congestion labels are determined as the dominant congestion set of the independent variable in the clustering independent variable set of the clustering label where the independent variable is a city congestion label.

[0062] The beneficial effects of the above technology are as follows: Based on all values ​​of each independent variable in the set of clustering independent variables for each cluster label, all values ​​of each independent variable in the historical rental sub-data of all rental tags that are completed in the behavior clustering data of each cluster label, age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector, the clustering range or dominant cluster set of each independent variable in the set of clustering independent variables for each cluster label can be determined. This can achieve precise definition of the boundaries of clustering independent variables, improve the matching degree between clustering results and user behavior characteristics, provide refined and scenario-based parameter support for rental recommendation models, and enhance the scientificity and accuracy of intelligent management decisions. Example 7:

[0063] Based on Example 6, a smart car rental management system includes the following modules: Historical cancellation data unit: Based on the clustering range or dominant cluster set of all independent variables in the clustering independent variable set of each cluster label, and the user basic information of all historical rental sub-data with cancellation tags in the historical rental data, the historical rental sub-data with cancellation tags in the historical rental data is divided to determine the historical cancellation data for each cluster label; Cancellation rate unit: Based on historical cancellation data and behavioral clustering data for each cluster label, determine the cancellation rate for each cluster label; Clustering training data unit: Based on the rental deposit, purpose label, rental label in all completed historical rental sub-data of the behavioral clustering data of each cluster label, and the final payment end time in the user's rental data, the clustering training data for each cluster label is determined; Clustering Recommendation Model Unit: Based on the clustering training data for each cluster label, construct a clustering recommendation model for each cluster label; Rental recommendation model unit: Based on the clustering recommendation model of all clustering labels, the behavior clustering vector in the behavior clustering data, the set of clustering independent variables, the cancellation rate, the significance weight of all independent variables in the set of clustering independent variables, the clustering range or the dominant cluster set, a rental recommendation model is constructed.

[0064] In this embodiment, all historical rental data with the "cancelled" tag are categorized. The categorization is based on two core criteria: the clustering range or dominant cluster set of all independent variables in the clustering variable set for each clustering tag, and the user's basic information corresponding to the historical rental data with the "cancelled" tag. This unit compares the user's basic information for each cancellation sub-data entry with the clustering variable parameters of each clustering tag, assigning the cancellation sub-data that best matches the independent variable range or dominant set of a particular clustering tag to that clustering tag. This ultimately determines the historical cancellation data corresponding to each clustering tag, achieving a precise association between cancellation behavior and clustering tags.

[0065] In this embodiment, the order cancellation rate corresponding to each cluster label is calculated based on historical cancellation data and behavioral clustering data for each cluster label. Behavioral clustering data reflects all completed historical rental sub-data under that cluster label, while historical cancellation data reflects all cancelled historical rental sub-data under that cluster label. By quantitatively comparing these two types of data, the proportion of orders cancelled under that cluster label is obtained, which is the cancellation rate. The cancellation rate directly reflects the rental order fulfillment stability of the user group corresponding to that cluster label.

[0066] In this embodiment, clustering training data corresponding to each cluster label is extracted and organized. The extraction scope includes all historical rental sub-data where the rental label is not completed in the behavioral clustering data of each cluster label. The core content extracted includes rental deposit, usage label, rental label, and the final payment end time in the user's rental data. This data covers key information such as the core cost of the rental transaction, usage scenario, order status, and settlement node. After integrating this data, exclusive clustering training data for each cluster label is formed.

[0067] In this embodiment, a dedicated clustering recommendation model is constructed for each cluster label, based on the clustering training data corresponding to that label. The clustering recommendation model mines the inherent correlation between user rental behavior and core transaction data under that cluster label, establishing a precise mapping relationship between feature inputs and recommendation results. The model first performs structured analysis on the clustering training data, extracting core feature dimensions such as rental deposit, usage label, and final payment deadline, clarifying the correlation between each feature. For example, users with the business usage label have shorter final payment settlement times, and users with the tourism usage label have a much higher preference for large-space vehicles than commuting users. Next, based on these feature correlation patterns, the model binds the user's basic attributes, rental usage, and other potential input features with appropriate vehicle labels, deposit standards, and other output results. Simultaneously, it learns the user's performance characteristics under that cluster label, such as the impact of final payment settlement time patterns on deposit settings, and the correlation between vehicle usage time and deposit amount for different usage labels. Ultimately, this forms a recommendation rule system that is only suitable for the user group of that cluster label.

[0068] In this embodiment, multiple key parameters are integrated, including the clustering recommendation model for each clustering label, the behavior clustering vector in the behavior clustering data, the set of clustering independent variables for each clustering label, the cancellation rate of each clustering label, the significance weight of all independent variables in the set of clustering independent variables, and the clustering range or dominant cluster set of all independent variables in the set of clustering independent variables. The operational logic of this unit is as follows: First, the clustering recommendation model for each cluster label is incorporated into the global model framework as a basic module, retaining the personalized recommendation rules for different user groups in each module. Then, behavioral clustering vectors are introduced to calibrate the core features of each cluster label, ensuring that the feature benchmarks of each clustering module are unified and accurate, avoiding recommendation errors caused by cluster feature bias. At the same time, the set of clustering independent variables is integrated to clarify the core variable dimensions of model analysis, locking in variables that have a key impact on recommendation results, such as age, driving experience, gender, and city congestion labels. The cancellation rate of each cluster label is combined to quantify the fulfillment risk, adding a risk correction dimension to the recommendation results of clusters with high cancellation rates. The influence of core independent variables on recommendation results is highlighted by significant weights, such as driving experience having a higher weight than gender in vehicle type recommendations. The effective value boundaries of each variable are defined based on the cluster range or the dominant cluster set, ensuring that the variable range of model analysis conforms to the actual characteristics of users in that cluster. By integrating and coordinating multi-dimensional parameters, the originally independent cluster recommendation models are merged into a globally adaptable rental recommendation model. This model retains the personalized recommendation rules of each cluster label while balancing the recommendation needs and operational risks of different clusters from a global perspective, thus achieving accurate recommendations for all user groups.

[0069] In this embodiment, the rental recommendation model integrates multi-dimensional core parameters with dedicated recommendation models for each cluster to construct a global, multi-dimensional, and risk-controllable rental recommendation system, achieving accurate matching of rental needs for different user groups and effective management of performance risks. The model first receives input data such as the user's basic information and rental purpose. It then quickly locates the user's cluster label using parameters such as the set of cluster independent variables, significant weights, cluster range, or dominant cluster set, accurately matching the dedicated cluster recommendation model corresponding to that label. Next, it calibrates the recommendation benchmark of the cluster model using behavioral clustering vectors, correcting feature biases within the cluster to ensure the recommendation results align with the core behavioral characteristics of that cluster. Simultaneously, it introduces the cancellation rate parameter of the cluster label to perform risk correction on the recommendation results. For example, for cluster labels with high cancellation rates, it appropriately increases the recommended deposit standard to reduce performance risks, or prioritizes recommending vehicles with short idle periods to reduce resource idleness losses. Finally, it integrates the recommendation results of the dedicated cluster model and risk correction strategies, considering the impact of various parameters, and outputs recommended vehicle labels and recommended deposits tailored to the user.

[0070] The beneficial effects of the above technologies are as follows: Based on historical rental data, behavioral clustering data of all clustering labels, the set of clustering independent variables, and the significant weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables, a rental recommendation model can be constructed, which can achieve a dual improvement in recommendation accuracy and risk management capabilities, effectively avoid the waste of resources in high cancellation rate clustering, strengthen the model's adaptability to the rental needs and performance risks of different user groups, and provide comprehensive and accurate decision support for intelligent rental recommendations. Example 8:

[0071] Based on Example 7, a smart car rental management system includes a management module comprising: Reservation Recommendation Unit: Input the user's basic information and purpose tags for each reservation order in the car rental service provider's reservation rental data into the rental recommendation model to determine the recommended vehicle tag and recommended deposit for each reservation order in the car rental service provider's reservation rental data; Management Unit: Based on the rental vehicle data of the car rental service provider and the recommended vehicle tags and recommended deposits of all reservation orders in the reservation rental data, intelligent management of car rental is achieved.

[0072] In this embodiment, a unique recommendation result is generated for each reservation order in the car rental service provider's reservation data. This generation is based on the user's basic information and purpose tags for each reservation order. The implementation involves inputting these two types of information into a pre-built rental recommendation model, ultimately outputting a recommended vehicle tag and a recommended deposit for each reservation order. The user's basic information includes core attributes that characterize the user, such as name, age, gender, driving experience, and address. The purpose tags identify the specific scenario for the user's vehicle rental, such as tourism, commuting, or business. This information together provides the model with an accurate user needs and scenario profile, ensuring that the output recommended vehicle tags match user preferences and the recommended deposit matches the user's repayment ability.

[0073] In this embodiment, intelligent management of the entire car rental process is achieved by relying on two types of core data: rental vehicle data from car rental service providers, and recommended vehicle tags and recommended deposits corresponding to all pre-booked rental orders in the pre-booked rental data. The recommended results of all pre-booked orders are linked and integrated with vehicle resource data. Based on the recommended vehicle tags, vehicles of corresponding models are matched, prioritizing vehicles that are idle and meet user needs. Simultaneously, a deposit collection standard is set for each order based on the recommended deposit, and the scheduling plan is dynamically adjusted in conjunction with vehicle status data. For example, backup vehicles are reserved for orders clustered with high cancellation rates, and vehicle configuration is optimized for orders with different usage tags. Through this linked management, intelligent operation of core rental processes such as vehicle resource scheduling, deposit control, and order fulfillment is achieved.

[0074] The beneficial effects of the above technologies are as follows: Based on the rental vehicle data, reservation data, and rental recommendation models of car rental service providers, intelligent management of car rentals can be achieved. It can adapt to the rental needs and fulfillment capabilities of different user groups, enabling customized output of recommended vehicle tags and deposits, avoiding high-risk orders, optimizing deposit collection and vehicle allocation strategies, reducing resource idleness and fulfillment disputes, and balancing recommendation accuracy, risk controllability, and operational efficiency, providing an integrated solution for intelligent management and control of the entire car rental process. Example 9:

[0075] This invention provides an intelligent car rental management platform for executing any one of the intelligent car rental management systems in Examples 1 to 8.

[0076] The beneficial effects of the above technologies are as follows: By acquiring rental vehicle data, reservation rental data, historical rental data, and rental behavior feature vectors, the standardized behavior vectors for each rental tag in the historical rental data are calculated for completed historical rental sub-data. This determines the behavior clustering data for multiple cluster tags, calculates the set of clustering independent variables for each cluster tag, and the significance weight, clustering range, or dominant cluster set of each independent variable in the set of clustering independent variables. A rental recommendation model is then constructed to achieve intelligent management of car rentals. This model can adapt to the rental needs and fulfillment capabilities of different user groups, enabling customized output of recommended vehicle tags and deposits. It avoids high-risk orders, optimizes deposit collection and vehicle allocation strategies, reduces resource idleness and fulfillment disputes, and balances recommendation accuracy, risk controllability, and operational efficiency, providing an integrated solution for intelligent management and control of the entire car rental process.

[0077] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart management system for car rental, characterized in that, include: Acquisition module: Acquires rental vehicle data, reservation rental data, and historical rental data from car rental service providers, and obtains rental behavior feature vectors for rental management; Calculation module: Based on the historical rental data of car rental service providers and the rental behavior feature vector of rental management, calculate the standardized behavior vector of each rental tag in the historical rental data as the completed historical rental sub-data, and determine the behavior cluster data of multiple cluster tags; Clustering module: Based on historical rental data and behavioral clustering data for each cluster label, calculate the set of clustering independent variables for each cluster label, as well as the significance weight, clustering range, or dominant cluster set for each independent variable in the set of clustering independent variables; Building Module: Based on historical rental data, behavioral clustering data of all cluster labels, set of clustering independent variables, and the significance weight, clustering range or dominant cluster set of each independent variable in the set of clustering independent variables, a rental recommendation model is built; Management module: Based on the rental vehicle data, reservation rental data and rental recommendation model of the car rental service provider, it realizes intelligent management of car rental.

2. The intelligent car rental management system according to claim 1, characterized in that, The acquisition module includes: Rental Vehicle Data Unit: Acquire rental vehicle data from car rental service providers. Rental vehicle data includes a set of rental vehicle models and rental data for each model in the rental vehicle set. Among them, the model rental data includes multiple rental vehicles and basic vehicle data, status tags, and vehicle status data for each rental vehicle. The basic vehicle data includes at least the vehicle number, vehicle age, and vehicle tags. Reservation Rental Data Unit: Acquire reservation rental data from car rental service providers. This data includes basic user information and usage tags for multiple reservation orders. Basic user information includes name, age, gender, driving experience, and address. Historical Rental Data Unit: Acquire historical rental data from car rental service providers. This historical rental data includes multiple historical rental data, which includes user basic information, usage tags, rental deposit, rental tags, vehicle number, user rental data, and historical driving data. Rental tags include cancellation and completion. Car rental behavior feature vector unit: Obtain the car rental behavior feature vector of rental management and the feature range of each behavior feature in the car rental behavior feature vector.

3. The intelligent car rental management system according to claim 2, characterized in that, The rental vehicle data unit and the historical rental data unit include: First vehicle status data sub-unit: If the status tag is idle, the vehicle status data shall include at least the location, remaining battery power, remaining fuel, and accumulated mileage; The second vehicle status data sub-unit: If the status tag is "rented", the vehicle status data includes real-time driving data, driving behavior data and rental basic data; The third vehicle status data sub-unit: If the status label is "under maintenance", the vehicle status data shall at least include the location, maintenance progress, estimated operating time, and maintenance cost; User car rental data and historical driving data sub-units: When the car rental tag is canceled, the user car rental data and historical driving data are empty. When the car rental tag is completed, the user car rental data includes the actual pick-up location, the actual drop-off location, the driving distance, the driving route, the driving time interval, and the final payment settlement time. The driving time interval includes multiple driving time sub-intervals and the driving duration of each driving time sub-interval.

4. The intelligent car rental management system according to claim 2, characterized in that, The calculation module includes: Historical behavior value vector unit: Based on the car rental behavior feature vector, features are extracted from the historical driving data of each car rental sub-data where the car rental tag is not completed and the driving distance in the user car rental data to determine the historical behavior value vector of each car rental sub-data where the car rental tag is not completed. Driving difficulty score unit: Input the actual pick-up location, actual drop-off location, and driving route of the user car rental data in each historical rental sub-data with the car rental tag being completed into the route evaluation model to determine the driving difficulty score of each historical rental sub-data with the car rental tag being completed. Peak Unit: For each rental period in the historical rental data where the rental tag is "completed", peak judgment is performed on each driving time sub-interval within the driving time interval of the user rental data in the historical rental data where the rental tag is "completed" to determine the peak interval and peak duration of each driving time sub-interval within the driving time interval of the user rental data in the historical rental data where the rental tag is "completed". Nighttime Unit: Perform nighttime judgment on each driving time sub-interval in the driving time interval of each user car rental data in the historical rental data where the car rental tag is "completed" to determine the nighttime interval and nighttime duration of each driving time sub-interval in the driving time interval of each user car rental data in the historical rental data where the car rental tag is "completed". Peak-Night Unit: Based on the peak and night intervals of the driving time sub-intervals in the user car rental data of each historical rental sub-data where the car rental tag is completed, determine the peak-night duration of each driving time sub-interval in the driving time intervals of the user car rental data of each historical rental sub-data where the car rental tag is completed; Standardized Behavior Vector Unit: Based on the historical behavior value vector of each rental sub-data with a completed rental tag in the historical rental data, the driving difficulty score, the peak interval of all driving time sub-intervals in the driving time interval in the user's rental data, and the feature range of each behavior feature in the rental behavior feature vector, the standardized behavior vector of each rental sub-data with a completed rental tag in the historical rental data is calculated. Clustering Unit: Perform cluster analysis on the standardized behavioral vectors of all historical rental sub-data with the rental tag "completed" in the historical rental data to determine multiple behavioral clusters and cluster labels for each behavioral cluster. The behavioral clusters include behavioral cluster vectors and multiple historical rental sub-data with the rental tag "completed".

5. The intelligent car rental management system according to claim 4, characterized in that, The clustering module includes: Urban congestion label unit: Based on the address in the user's basic information for each completed historical rental sub-data in the historical rental data, the urban congestion label for each completed historical rental sub-data in the historical rental data is determined. The urban congestion label includes severe congestion, moderate to severe congestion, mild congestion, and smooth traffic. Behavioral Feature Value Vector Unit: Based on the standardized behavioral vector of each historical rental sub-data of the behavioral clustering data for each cluster label, the behavioral feature value vector of each behavioral feature for each cluster label is determined; The first determining unit: Based on the age, gender, driving experience, and city congestion labels of all historical rental sub-data of the behavioral clustering data for each cluster label, determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector for each cluster label; Significance probability unit: Determine the age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector of each cluster label as independent variables, and determine the behavioral feature value vector of each behavioral feature of each cluster label as dependent variables. Perform one-way ANOVA to determine the significance probability of each independent variable and each dependent variable of each cluster label. Correction label unit: Based on the significance probabilities of all independent and dependent variables for each cluster label, determine the correction label for each independent and dependent variable for each cluster label; Clustering Independent Variable Set Unit: Based on the significance probabilities of all independent and dependent variables for each cluster label and the correction label, calculate the clustering independent variable set for each cluster label and the significance weight of each independent variable in the clustering independent variable set; Clustering range or dominant cluster set: Based on each independent variable in the clustering independent variable set of each cluster label, all values ​​of each independent variable in the historical rental sub-data of all car rental labels in the behavior clustering data of each cluster label, age feature vector, gender feature vector, driving experience feature vector, and congestion feature vector, determine the clustering range or dominant cluster set of each independent variable in the clustering independent variable set of each cluster label.

6. The intelligent car rental management system according to claim 5, characterized in that, Clustering range or dominant cluster set units include: Age or driving experience sub-unit: If the independent variable in the clustering independent variable set of the clustering label is age or driving experience, based on the age feature vector or driving experience feature vector of the clustering label, determine the clustering range in the clustering independent variable set of the clustering label where the independent variable is age or driving experience. Gender sub-unit: If the independent variable in the clustering independent variable set of the clustering label is gender, based on all the values ​​of gender in the historical rental sub-data of all car rental tags that are completed in the behavior clustering data of the clustering label and the gender feature vector, calculate the discrete proportion of each gender in the gender feature vector in the clustering independent variable set of the clustering label. If the discrete proportion of gender in the gender feature vector in the clustering independent variable set of the clustering label is greater than a preset threshold, determine that the gender in the gender feature vector in the clustering independent variable set of the clustering label is the dominant gender set with gender as the independent variable in the clustering independent variable set of the clustering label; otherwise, determine that the dominant cluster set with gender as the independent variable in the clustering independent variable set of the clustering label is the gender feature vector. Urban congestion label subunit: If the independent variable in the clustering independent variable set of the clustering label is the urban congestion label, based on all values ​​of the urban congestion label in the historical rental subdata of all car rental labels completed in the behavior clustering data of the clustering label and the congestion feature vector, calculate the discrete proportion of each urban congestion label in the congestion feature vector in the clustering independent variable set of the clustering label. Sort the discrete proportions of all urban congestion labels in the congestion feature vector in the clustering independent variable set of the clustering label from largest to smallest. Select the smallest position greater than a preset threshold from the sorted proportion sequence as the congestion position of the urban congestion label in the clustering independent variable set of the clustering label. Determine the urban congestion label corresponding to the first few discrete proportions in the proportion sequence, which is the dominant congestion set of the dominant clustering set in the clustering independent variable set of the clustering label as the congestion feature vector.

7. The intelligent car rental management system according to claim 6, characterized in that, Build modules, including: Historical cancellation data unit: Based on the clustering range or dominant cluster set of all independent variables in the clustering independent variable set of each cluster label, and the user basic information of all historical rental sub-data with cancellation tags in the historical rental data, the historical rental sub-data with cancellation tags in the historical rental data is divided to determine the historical cancellation data for each cluster label; Cancellation rate unit: Based on historical cancellation data and behavioral clustering data for each cluster label, determine the cancellation rate for each cluster label; Clustering training data unit: Based on the rental deposit, purpose label, rental label in all completed historical rental sub-data of the behavior clustering data of each cluster label, and the final payment end time in the user's rental data, the clustering training data for each cluster label is determined; Clustering Recommendation Model Unit: Based on the clustering training data for each cluster label, construct a clustering recommendation model for each cluster label; Rental recommendation model unit: Based on the clustering recommendation model of all clustering labels, the behavior clustering vector in the behavior clustering data, the set of clustering independent variables, the cancellation rate, the significance weight of all independent variables in the set of clustering independent variables, the clustering range or the dominant cluster set, a rental recommendation model is constructed.

8. The intelligent car rental management system according to claim 7, characterized in that, The management module includes: Reservation Recommendation Unit: Input the user's basic information and purpose tags for each reservation order in the car rental service provider's reservation rental data into the rental recommendation model to determine the recommended vehicle tags and recommended deposit for each reservation order in the car rental service provider's reservation rental data; Management Unit: Based on the rental vehicle data of the car rental service provider and the recommended vehicle tags and recommended deposits of all reservation orders in the reservation rental data, intelligent management of car rental is achieved.

9. A smart management platform for car rental, characterized in that, Used to implement a smart car rental management system as described in any one of claims 1 to 8.

Images