A Hotel Management Method and System Based on Multimodal Data Behavioral Analysis

Abstract

This invention discloses a hotel management method and system based on multimodal data behavior analysis, belonging to the field of hotel management technology. The method includes: establishing a knowledge graph of delivery hotspot areas; mining the correlation features between the concentration of delivery tasks and the frequency of delivery path intersections based on the knowledge graph; determining the path intersection risk level of hotspot areas based on the correlation features; constructing a capability profile of delivery robots based on historical delivery trajectory data; determining the number of tasks assigned to delivery robots in hotspot areas based on the adaptation characteristics between the path intersection risk level of hotspot areas and the capability profile of delivery robots; and updating the path intersection congestion features in the delivery hotspot area knowledge graph based on the number of tasks assigned to delivery robots to dynamically adjust the knowledge graph. This invention effectively reduces the risk of congestion on hotel robot delivery paths and significantly improves delivery efficiency and service quality.

Description

Technical Field

[0001] This invention relates to the field of hotel management technology, and specifically to a hotel management method and system based on multimodal data behavior analysis. Background Technology

[0002] With the rapid development of smart hotel management, more and more hotels are adopting delivery robots to complete delivery tasks in guest rooms or public areas to improve delivery efficiency and reduce operating costs. However, in the actual operation of delivery robots, delivery requests are often highly concentrated in specific time periods or areas, leading to frequent intersections and congestion of delivery routes. Furthermore, due to differences in technical parameters and behavioral strategies, different delivery robots exhibit distinct individual characteristics when dealing with route intersections and congestion risks, further exacerbating the complexity of delivery management.

[0003] Existing hotel delivery management methods primarily allocate delivery tasks through fixed route planning or manual scheduling. This approach typically ignores the concentrated spatial and temporal distribution of delivery tasks and fails to adequately consider the behavioral differences of delivery robots in situations such as path intersections, abnormal stops, or detours. This makes it difficult to dynamically adjust delivery strategies, resulting in delayed response times, reduced delivery efficiency, and a significant impact on hotel service quality.

[0004] Furthermore, the correlation between the concentration of delivery tasks and the risk of congestion along delivery routes has not yet been systematically explored and utilized. Managers lack effective tools to identify and address the congestion risks caused by concentrated delivery tasks in advance, making it difficult to achieve reasonable allocation of delivery tasks and risk prevention. Existing technologies also lack a unified platform for structured and dynamic knowledge management of congestion risks in delivery hotspots and delivery routes, resulting in a lack of effective adaptability in delivery strategies. Summary of the Invention

[0005] The purpose of this invention is to provide a hotel management system and method based on multimodal data behavior analysis to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a hotel management method based on multimodal data behavior analysis, comprising:

[0008] Based on the historical time period distribution characteristics of delivery request signals, a knowledge graph of delivery hotspot areas reflecting the characteristics of cross-congestion along delivery routes is established.

[0009] Based on the knowledge graph of delivery hotspot areas, we can explore the correlation features between the concentration of delivery tasks and the frequency of intersection of delivery routes;

[0010] Based on the aforementioned correlation characteristics, the path intersection risk level of hotspot areas is determined; and a capability profile of the delivery robot to adapt to path intersection risks is constructed based on the robot's historical delivery trajectory data.

[0011] Based on the adaptation characteristics of the path intersection risk level in hotspot areas and the capability profile of delivery robots, the number of tasks to be assigned to delivery robots in hotspot areas is determined.

[0012] The path intersection congestion features in the delivery hotspot area knowledge graph are updated based on the number of tasks assigned to delivery robots, so as to dynamically adjust the delivery hotspot area knowledge graph.

[0013] Secondly, the present invention provides a hotel management system based on multimodal data behavior analysis, implemented based on the aforementioned hotel management method based on multimodal data behavior analysis, comprising:

[0014] The hotspot graph construction module is used to build a knowledge graph of delivery hotspot areas that reflects the characteristics of cross-congestion along delivery routes, based on the historical time period distribution characteristics of delivery request signals.

[0015] The cross-feature mining module is used to mine the correlation features between the concentration of delivery tasks and the frequency of cross-delivery routes based on the knowledge graph of delivery hotspot areas;

[0016] The risk capability modeling module is used to determine the path intersection risk level of hotspot areas based on the associated features; and to construct a capability profile of the delivery robot to adapt to path intersection risks based on the robot's historical delivery trajectory data.

[0017] The risk-adaptive loading module is used to determine the number of tasks to be assigned to delivery robots in hotspot areas based on the adaptation characteristics of the path intersection risk level and the capability profile of the delivery robots.

[0018] The graph dynamic update module is used to update the path intersection congestion features in the knowledge graph of delivery hotspot areas based on the number of tasks assigned to delivery robots, so as to dynamically adjust the knowledge graph of delivery hotspot areas.

[0019] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0020] This invention constructs a knowledge graph of delivery hotspot areas to accurately identify the spatiotemporal concentrated distribution characteristics of hotel delivery tasks and the intersection congestion nodes of routes. It effectively solves the technical deficiency of existing technologies that cannot systematically mine the correlation characteristics between the concentration of delivery tasks and the risk of route congestion. It realizes the structured and dynamic management of delivery hotspot areas and route congestion risks, and provides key data support for the intelligent allocation of subsequent delivery tasks.

[0021] This invention explores the synergistic variation characteristics between the concentration of delivery tasks and the frequency of intersections in delivery routes. It further combines these characteristics with the behavioral features of individual delivery robots to create robot capability profiles. This accurately distinguishes the individual differences among different delivery robots when dealing with the risk of congestion at intersections, effectively overcoming the problem of decreased delivery efficiency caused by existing technologies that do not consider individual robot differences when allocating robot tasks. This significantly improves the accuracy and rationality of delivery task allocation.

[0022] This invention intelligently determines the number of tasks assigned to delivery robots in hotspot areas by adapting the risk level of path intersections in hotspot areas to the capability profiles of delivery robots, and dynamically updates the knowledge graph of delivery hotspot areas in real time. This enables the delivery strategy to adaptively optimize in real time according to changes in delivery demand, significantly reducing the frequency of delivery path intersection conflicts and abnormal waiting, effectively improving delivery efficiency and service response speed, and enhancing the overall service quality and customer experience of the hotel. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0024] Figure 1 This is a flowchart illustrating a hotel management method based on multimodal data behavior analysis according to the present invention.

[0025] Figure 2 This is a framework diagram of a hotel management system based on multimodal data behavior analysis according to the present invention. Detailed Implementation

[0026] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided to make the description of this application more complete and comprehensive, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The drawings are merely illustrative illustrations of this application and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0027] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of the exemplary embodiments disclosed in this application. However, those skilled in the art will recognize that the technical solutions disclosed in this application can be practiced with one or more specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of the disclosure of this application.

[0028] Example 1

[0029] like Figure 1 As shown in the figure, this embodiment discloses a hotel management method based on multimodal data behavior analysis, including:

[0030] S101: Based on the historical time period distribution characteristics of delivery request signals, establish a knowledge graph of delivery hotspot areas that reflects the characteristics of cross-congestion along delivery routes;

[0031] It should be noted that the delivery request signal is specifically delivery request data sent by a hotel room or public area through a smart terminal, including the time the request was sent, the specific floor, the room number, and the type of requested item;

[0032] For example, the data format of a delivery request signal is as follows:

[0033]

[0034] in, The i-th delivery request signal; Request the floor number (e.g., "10"); Number the room (e.g., "1001"); The time the request was sent (e.g., "07:05:12"); For the requested item type (e.g., "breakfast").

[0035] In specific implementation, the establishment of a knowledge graph of delivery hotspot areas reflecting the characteristics of cross-congestion along delivery routes includes:

[0036] Based on delivery request signals, identify candidate hotspot areas where historical delivery tasks are concentrated and the frequency of delivery robot stops or detours is abnormal; specific implementation includes:

[0037] The total number of delivery requests received by each floor or area of ​​the hotel during different peak hours (such as breakfast hours 07:00-09:00) over a consecutive number of days (e.g., the last 30 days) is recorded as follows: , representing the number of delivery requests for the i-th floor or area during peak time t;

[0038] Set the threshold for the number of delivery requests to (For example This will be the number of delivery requests during peak hours over a consecutive number of days (e.g., 7 days or more). The floors or areas were initially identified as candidate areas for concentrated delivery tasks;

[0039] Based on historical delivery trajectory data collected by the robot's indoor positioning system (using a UWB indoor positioning system), the position coordinate sequence of the delivery robot during task execution is extracted: ,in, Let be the coordinates of the k-th position on the robot's delivery trajectory. This is the timestamp of the robot's location. The floor number where the location is located (e.g., "10th floor");

[0040] Based on historical delivery trajectory data, calculate the frequency of the following abnormal events occurring by the robot in candidate hotspot areas or adjacent areas:

[0041] Abnormal Dwelling Event: Duration of continuous dwell time of the delivery robot at a single location coordinate point (For example, 30 seconds);

[0042] Abnormal detour event: The distance by which the delivery robot deviates from the planned path. (For example, 2 meters); Specific calculation method:

[0043] Delivery robot dwell time at trajectory points:

[0044]

[0045] Among them, the robot is in coordinates If the location remains continuously for n consecutive timestamps, If so, it is considered an abnormal stay event;

[0046] Difference between the actual path length and the planned path length of the robot:

[0047]

[0048] If satisfied If the following occurs, it is considered an abnormal detour event, where: The length of the robot's actual delivery trajectory; The path length for the robot was planned based on the hotel's standard delivery route.

[0049] Statistical analysis of the total frequency of abnormal robot events (abnormal stopping or detouring) within candidate hotspot areas. Further set an abnormal event frequency threshold. (For example (times / month), if the frequency of abnormal events in the region If so, the area is ultimately identified as a candidate hotspot area;

[0050] To clearly define the spatial boundaries of delivery hotspot areas, this embodiment defines each candidate hotspot area as a closed polygon region, and its geometric boundaries are represented by a set of vertex coordinates: When the robot trajectory enters or leaves the candidate hotspot area, the ray method is used to determine the position, and a 0.5m buffer is added to reduce misjudgment caused by positioning noise.

[0051] Extract delivery trajectory data of delivery robots between the candidate hotspot areas to determine the spatial distribution characteristics of the intersections of actual delivery paths;

[0052] The spatial distribution characteristics of determining the intersection locations of actual delivery routes include:

[0053] Based on the historical delivery task data of the delivery robots, identify the locations where the robots actually stopped or experienced abnormal waiting during peak delivery periods;

[0054] For example, during peak delivery hours (such as 07:00-09:00), the identification of actual conflicts or abnormal waiting locations on the robot delivery route specifically includes:

[0055] For each delivery robot, extract a set of delivery trajectory points during peak hours based on historical delivery task data. The selected location coordinates show a dwell time exceeding the abnormal waiting threshold. (For example, trajectory points over 10 seconds):

[0056] The method for identifying abnormal waiting events is as follows:

[0057] For any continuous trajectory points If it satisfies:

[0058] Spatial distance between adjacent points (For example, 0.2 meters), meaning the position remains basically unchanged;

[0059] Corresponding time interval (For example, 10 seconds);

[0060] Then the coordinates point This has been identified as an abnormal waiting location;

[0061] In addition, the specific implementation method for identifying path conflict events is as follows:

[0062] During peak delivery periods, analyze the trajectory point data of different delivery robots. If any two delivery robots are at the same time (or the interval is less than a threshold), For example, the radius of points located at the same position within 5 seconds. If the distance is within 0.5 meters (for example), then that location is identified as a path conflict location.

[0063] The set of coordinates of all stopping points in the delivery robot's delivery route where actual path conflicts or abnormal waiting occur, identified using the above method, is denoted as: .

[0064] Determine the actual path direction correlation between abnormal stop locations and hotspot areas in the robot delivery route;

[0065] Specifically, based on the historical delivery trajectories of the delivery robots, a set of delivery trajectory paths is extracted between the robot's departure from one candidate hotspot area and its arrival at another candidate hotspot area, denoted as: ,in, Let m be the starting point of the hotspot region. The endpoint of the hotspot region n;

[0066] Determine the set of abnormal stopping locations Each abnormal stop location in With delivery route set Whether the spatial distance between each path is less than the path association determination distance threshold (e.g., 0.5 meters) to determine the path direction correlation between abnormal dwelling locations and hotspot areas;

[0067] The specific formula for determining the association between abnormal stop locations and paths is as follows:

[0068] If a certain delivery route exists At least one point ,satisfy: Then the abnormal location is identified. There is an actual path direction association with the hotspot areas m and n.

[0069] Based on the actual path direction association, analyze whether there are fixed or repeated path intersections between delivery routes in different hotspot areas;

[0070] Specifically, the frequency of path correlations between abnormal stop locations and hotspot areas within the historical delivery data period is statistically analyzed using the following method:

[0071] If a certain abnormal stopping location hotspot region pair (m,n) The frequency of path associations occurring during peak delivery hours exceeds the fixed path determination frequency threshold. (For example, 20 times / month), then the abnormal stay location is identified as a hotspot area (m, n). Forming fixed path intersections;

[0072] Repeat the above steps for all hotspot pairs to obtain the correspondence between hotspot pairs that form fixed or repeating path intersections and their location coordinates, denoted as a set: .

[0073] Based on the analysis results, the spatial distribution characteristics of the intersections of actual delivery routes between hotspot areas are generated.

[0074] Specifically, based on the coordinates of the intersection positions of the determined fixed or repeating paths. The set of feature points at the intersection of delivery routes is labeled as follows: ;

[0075] Further analysis of the feature point set Clustering processing (such as the DBSCAN spatial clustering algorithm) is performed to merge feature points that are spatially close, so that they represent the actual spatial distribution of the intersection of delivery paths between hotspot areas;

[0076] The coordinates of the center positions of each cluster obtained after clustering are output as the spatial distribution features of the intersection positions of the actual delivery routes, denoted as: .

[0077] Based on the spatial distribution characteristics, specific intersection nodes where delivery robots frequently experience path conflicts between hotspot areas are identified;

[0078] Specifically, using the obtained spatial distribution feature set C as input, the statistics of each intersection position are calculated. The total number of path conflicts or abnormal waits that occurred within the historical period is denoted as . ;

[0079] Set a threshold for the frequency of conflict events (For example, 20 times / month), select from set C those that meet the condition of having a conflict count greater than or equal to The position of the node is denoted as a specific set of intersection nodes: ;

[0080] With the set of cross nodes The location coordinates in the data serve as specific intersection nodes where frequent path conflicts occur between hotspot areas of the delivery robot.

[0081] Using the intersection nodes as bridging points for the delivery route association between hotspot areas, a delivery hotspot area knowledge graph representing the risk of route intersection congestion is established.

[0082] Specifically, with a specific set of intersection nodes The coordinates of each node in the graph serve as bridging points, establishing path relationships between candidate hotspot regions. These relationships are specifically represented as: region pairs... There is at least one intersection node between them, indicating a path association;

[0083] A knowledge graph of delivery hotspot areas is constructed using knowledge graph triples. A triple is defined as follows: ,in, These represent delivery hotspots and nodes, respectively. To characterize the intersection relationship of delivery routes between nodes in hotspot areas, specific attributes include: coordinates of associated nodes (i.e., the location of intersection nodes), frequency of conflict events, and information on peak delivery times;

[0084] For example, the relationship between hotspot areas at levels 10 and 11 can be represented as a triple:

[0085] ("Area 10", "Path Intersection Relationship (Intersection Point: (12.5m, 8.2m), Conflict Frequency: 35 times / month, Peak Hour: 07:00-09:00)", "Area 11" ;

[0086] The above set of relational triples forms a complete knowledge graph of delivery hotspot areas, which clearly and intuitively expresses the characteristics of cross-congestion of delivery routes between hotspot areas.

[0087] S102: Based on the knowledge graph of delivery hotspot areas, explore the correlation features between the concentration of delivery tasks and the frequency of intersection of delivery routes;

[0088] In specific implementation, the correlation characteristics between the concentration of delivery tasks and the frequency of intersection of delivery routes include:

[0089] Based on the knowledge graph of delivery hotspot areas, identify the historical distribution of delivery congestion events at the intersections of delivery routes between hotspot areas;

[0090] Specifically, based on the knowledge graph of delivery hotspot areas established in step S101, delivery congestion event data corresponding to the path intersection relationships between nodes in each hotspot area are extracted to form a historical distribution set of delivery congestion events. ,in, This represents the spatial coordinates of the i-th intersection node; This indicates the cumulative number of delivery congestion events that occurred at the intersection within a historical period (e.g., the most recent 30 days). This indicates the peak time period (e.g., 07:00-09:00) during which congestion events occur at the i-th intersection node.

[0091] For example, the intersection of the delivery hotspot area nodes on the 10th and 11th floors is located at (12.5m, 8.2m). If the cumulative number of delivery congestion events within the historical period is 35, and the peak period is 07:00-09:00, then it is represented as: ;

[0092] Based on the historical distribution of delivery congestion events, determine the peak periods of delivery congestion in each hotspot area;

[0093] Specifically, statistically analyze the historical distribution set of delivery congestion events. The number of delivery congestion events at intersections corresponding to various hotspot areas within a 24-hour period is denoted as the percentage of such events to the total number of events throughout the day. If the following conditions are met: Then, the hotspot area is identified as experiencing peak delivery congestion during the corresponding time period, where: The number of congestion events in a hotspot area within a specific time period; This represents the total number of congestion events in the hotspot area throughout the day. The threshold for the proportion of concentrated periods is set, for example, 0.5 (i.e. 50%).

[0094] For example, if the total number of congestion events on the 10th floor of the hotspot area is 50 throughout the day, 30 of them occur between 07:00 and 09:00, accounting for 0.6 (exceeding the threshold of 0.5). Therefore, the peak period for delivery congestion on the 10th floor of the hotspot area is determined to be between 07:00 and 09:00.

[0095] This ultimately results in a cluster of peak delivery congestion periods: For example, in specific form: .

[0096] For regions where the growth rate of delivery tasks exceeds a preset threshold during peak delivery congestion periods, the collaborative change characteristics between the degree of delivery task concentration and the frequency of path intersections are analyzed.

[0097] The analysis of the coordinated change characteristics between the concentration of delivery tasks and the frequency of path intersections includes:

[0098] Extract the specific growth rate trend of delivery request signals in each hotspot area during peak delivery congestion periods;

[0099] Specifically, taking the peak delivery congestion periods (e.g., 07:00-09:00) within a historical period (e.g., the last 30 days) as the analysis object, the peak periods of each hotspot area are further divided into several fixed-unit sub-periods (e.g., every 10 minutes); the number of new delivery requests in each sub-period is calculated. Delivery task growth rate The calculation formula is: ,in, The number of new delivery requests within the sub-time period; The duration of a unit sub-time period (e.g., 10 minutes);

[0100] For example, if the number of new delivery requests in the hotspot area of ​​floor 10 between 07:00 and 07:10 is 9, then the growth rate of delivery tasks during this period is: For example: .

[0101] Identify specific hotspots that cause a significant increase in the number of path intersection congestion after the rapid increase in the growth rate of the delivery request signal;

[0102] Specifically, based on the obtained set of growth rate trends Identify the changes in the growth rate of delivery requests over two consecutive sub-periods; if the growth rate of hotspot areas in a certain sub-period... Exceeding the set growth rate threshold (e.g., 1.0 per minute), and the percentage increase in the number of congestion events at intersections in the corresponding hotspot areas within subsequent adjacent time periods. If the percentage exceeds a set threshold (e.g., 30%), the hotspot area is recorded as a hotspot area with coordinated change characteristics.

[0103] For example, if the growth rate of delivery requests in hotspot area 10 is 1.2 per minute (exceeding the threshold of 1.0 per minute) during the period from 07:00 to 07:10, and the number of delivery congestion events during the period from 07:10 to 07:20 increases by 40% compared to the previous period (exceeding the 30% threshold), then area 10 is identified as a hotspot area of ​​coordinated change.

[0104] Forming a set of hotspots for coordinated change: ,For example: .

[0105] Determine whether the growth rate of delivery requests in the specific hotspot area and the trend of changes in the number of intersections and congestion on delivery routes both exhibit a rapid increase simultaneously.

[0106] Specifically, for the time series data of the growth rate of delivery requests and the number of congestion events at intersection nodes in each region, a rate of change threshold detection method is used to determine the starting point of their rapid increase:

[0107] First, calculate the growth rate of delivery requests and the rate of change of the number of congestion events at cross-nodes for each time period, denoted as follows:

[0108] Rate of change of growth rate of delivery requests ;

[0109] Change rate of intersection congestion frequency ;

[0110] Then, when the aforementioned rate of change exceeds their respective preset thresholds within two consecutive time periods (for example, the threshold for the rate of change of the request growth rate is 50%, and the threshold for the rate of change of the number of congestion events is 40%; the specific thresholds can be determined based on experimental data), the moment when the rate of change first exceeds the threshold is determined as the starting moment of the rapid increase in delivery requests. and the starting point of rapid increase in path congestion ;

[0111] Finally, determine whether the difference between the two starting times is less than the set synchronization judgment window. If the following conditions are met:

[0112]

[0113] This indicates that the concentration of delivery tasks and the frequency of intersection of delivery routes in the hotspot area exhibit a synergistic characteristic of simultaneous and rapid increases.

[0114] When the growth rate of delivery requests and the number of path intersection congestion both show a rapid increase, determine the coordinated change characteristics between the concentration of delivery tasks and the frequency of path intersections.

[0115] Specifically, the hotspot regions that simultaneously and rapidly increase as described above are recorded as the final set of regions exhibiting coordinated change characteristics: ,For example: .

[0116] When the collaborative change characteristics meet the cross-congestion triggering conditions, determine the correlation characteristics between the concentration of delivery tasks and the frequency of cross-delivery routes;

[0117] Specifically, after identifying the characteristic of a "simultaneous rapid increase" in both the growth rate of delivery requests and the number of path intersection congestion incidents, the triggering conditions for intersection congestion are further defined:

[0118] If a hotspot area meets the following conditions:

[0119] 1. Delivery request growth rate Greater than the set threshold (e.g., 1.0 units / minute), and;

[0120] 2. The increase in the number of congestion events at intersections in the two subsequent consecutive sub-periods. All are greater than the set threshold (e.g., 30%), and;

[0121] 3. The time difference between the start of rapid increase in request growth and the increase in congestion events is less than the synchronization judgment window. ;

[0122] The hotspot area is then deemed to meet the "cross-traffic congestion triggering condition," and the associated feature tuple is output. ;

[0123] Example associated feature tuple: (Region 10 layers", ["07:00", "07:10"], 1.2 items / minute, 40%, 8 minutes, "Yes").

[0124] S103: Determine the path intersection risk level of the hotspot area based on the aforementioned correlation characteristics; and construct a capability profile of the delivery robot to adapt to path intersection risks based on the robot's historical delivery trajectory data;

[0125] Specifically, based on the correlation characteristics between the concentration of delivery tasks in hotspot areas and the frequency of delivery route intersections determined in step S102, the route intersection risk level of each hotspot area is calculated. The specific method is as follows:

[0126] Based on the growth rate of delivery requests in hotspot areas The percentage increase in the number of intersections with corresponding regional routes Calculate the risk score of path intersection The dynamic quantile normalization method is used to define: This refers to the quantile position of the current period's delivery request growth rate relative to this metric across all regions. For example: The value range is truncated to [0,1]. The percentage increase in the number of path intersection congestion events is also calculated using dynamic quantile normalization. and For example, weighting factors ;

[0127] Then, risk levels are determined based on quantile thresholds from historical statistical data:

[0128] when At that time, hotspot area i was identified as a high-risk area;

[0129] when At that time, hotspot area i was identified as a medium-risk area;

[0130] when At that time, hotspot area i was determined to be at a low-risk level.

[0131] The ability profile for constructing delivery robots to adapt to path intersection risks includes:

[0132] Based on historical delivery data, extract the frequency and duration of specific coping behaviors taken by the robot when there is cross - congestion in the delivery path in hot spots;

[0133] Among them, the three specific coping behaviors include:

[0134] Active avoidance: The robot actively bypasses the congestion node and changes the delivery route;

[0135] Specifically, the spatial threshold distance at which the robot's trajectory data deviates from the original planned path (for example, deviating more than 0.5 meters from the central path), and the re - planned path is significantly different from the original planned route (the overall change in the path exceeds 10%); The specific judgment formula is: If<00004​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​The frequency and duration of each robot's behavior type during path intersection congestion within a historical period (e.g., the last 30 days) are statistically analyzed to obtain robot behavior data: ,in: This represents the j-th delivery robot; These represent the frequency of three types of behaviors: proactively avoiding obstacles, adjusting the order of tasks, and waiting in place. These represent the cumulative duration of the corresponding actions;

[0143] Determine the specific behavioral preferences of each delivery robot when faced with cross-path congestion in hot areas, such as proactively avoiding obstacles, adjusting task order, or waiting in place.

[0144] The determination of the specific behavioral preferences of each delivery robot when facing path intersection congestion in hotspot areas includes:

[0145] Based on the delivery robot's historical task records, the frequency at which the robot proactively changes its original delivery route when facing the risk of congestion at intersections is determined; the specific formula is:

[0146]

[0147] Where: the denominator represents the total number of actions taken by robot j during the congestion event.

[0148] Determine the historical statistical distribution of delivery delays caused by changes in robot delivery routes;

[0149] Specifically, this involves statistically analyzing data on the robot's historical proactive avoidance behavior, extracting the delivery delay duration (in seconds) caused by each route change, forming a delay duration data sequence, and then using histograms or kernel density estimation to obtain the probability distribution curve of the delivery delay duration. ;

[0150] Statistics on delivery delays caused by delivery robots adjusting task order and waiting in place.

[0151] Specifically, based on the robot's historical delivery task trajectory data, the delivery delay time after each "adjust task order" and "wait in place" behavior is statistically analyzed to form a delay time data set. The average delay time corresponding to each behavior type is then calculated using the following formula:

[0152]

[0153] in, These represent the delay time after the k-th occurrence of adjusting the task order and waiting in place, respectively.

[0154] The above steps provide the average delay time for three behaviors (active avoidance, adjusting task order, and waiting in place), which will provide a basis for determining the specific behavioral preferences of the delivery robot.

[0155] Based on the historical statistical distribution of the delivery delay time, determine the actual impact of the robot's proactive route change behavior on delivery efficiency;

[0156] The formula for calculating the actual impact of the robot's proactive route-changing behavior on delivery efficiency is as follows:

[0157]

[0158] in, This is the average delivery delay impact value, used to represent the actual impact of the robot's proactive route change behavior on delivery efficiency. This represents the delivery delay time after the k-th proactive route change. This indicates the total number of times the route was changed proactively.

[0159] Based on the differences in actual impact, determine the specific behavioral preferences of each delivery robot when facing path intersection congestion in hotspot areas;

[0160] Specifically, the behavioral preferences of delivery robots are clearly categorized as follows:

[0161] Active avoidance type: Frequently changes routes proactively (e.g.) ), and the average delay impact value of active avoidance Significantly lower than adjusting task order and waiting in place When the delay effect value is;

[0162] Cautious waiting type: moderate frequency of proactive route changes (e.g.) ), and the average delay impact value of active avoidance Significantly lower than adjusting task order and waiting in place When the difference in the delay effect value is not significant;

[0163] Reactive response type: Less frequent proactive route changes (e.g.) ), and the average delay impact value of active avoidance Greater than adjusting task order and waiting in place When the delay effect value is...

[0164] Analyze the specific impact relationship between the aforementioned behavioral preferences and the overall response delay of delivery tasks;

[0165] Specifically, for robots with various behavioral preference types within a historical period, the overall response delay time of all delivery tasks in hotspot areas during peak task periods is statistically analyzed, the average overall response delay time is calculated, the correlation between behavioral preferences and response delay is analyzed, and the degree of influence of behavioral preferences on response delay is determined.

[0166] Based on the specific influencing relationships described above, a capability profile of the delivery robot to adapt to path intersection risks that reflects individual differences among robots is constructed;

[0167] The specific attributes of the capability profile include: robot number, frequency of proactive avoidance behavior (%), average delivery delay impact value (seconds), behavior preference type, and historical average duration of overall task response delay;

[0168] Example capability profiles (see Table 1 below):

[0169] Table 1: Capability Profile Data Table

[0170]

[0171] The above methods enable an accurate description of the delivery robot's ability to adapt to path intersection risks, thus supporting the optimization of subsequent task allocation strategies.

[0172] S104: Based on the adaptation characteristics of the path intersection risk level in hotspot areas and the capability profile of delivery robots, determine the number of tasks to be assigned to delivery robots in hotspot areas;

[0173] The determination of the number of tasks assigned to delivery robots in hotspot areas includes:

[0174] Based on the hotspot area knowledge graph, identify several pairs of related areas with the highest risk of path intersection within the hotspot area;

[0175] Specifically, based on the risk levels (high, medium, low) of the hotspot areas determined in step S103 above, hotspot areas belonging to the high-risk level are paired up in pairs. Then, based on the historical frequency of conflict events between inter-regional nodes recorded in the knowledge graph, the top few groups (e.g., the top 5 groups) with the highest historical conflict frequency are selected as the set of high-risk associated region pairs, represented as: ;

[0176] For example, as shown in Table 2 below:

[0177] Table 2: Related Region Data Table

[0178]

[0179] Analyze the differences in historical completion efficiency of delivery tasks among the various delivery robots in the associated regions;

[0180] This step analyzes the fundamental relationship between the degree of delivery task backlog and the rate of decrease in the robot's delivery task processing speed, and uses historical data to build a regression model to predict the robot's real-time task handling capacity at a given task backlog level. ; and historical task completion efficiency indicators This is a macro-level statistical evaluation of the robot's long-term delivery performance, specifically its actual task-carrying capacity over multiple historical periods. The average value is used for the next step (analyzing the robot's adaptability to the risk of congestion on different regional routes);

[0181] Furthermore, behavioral preference types from the robot's capability profile are introduced as a correction factor for efficiency evaluation. This allows for personalized adjustments to real-time load capacity predictions, thereby accurately reflecting the impact of individual robot behavior differences on actual delivery efficiency.

[0182] Specifically, the correction factor is defined based on the robot's individual behavioral preference type:

[0183]

[0184] Subsequently, through regression analysis of historical data, the formula for calculating the decline in delivery speed was determined as follows:

[0185]

[0186] in, For the number of backlogged tasks, the coefficient is... Obtained through regression analysis of historical data;

[0187] The unified formula for calculating the actual task carrying capacity of the delivery robot j is as follows: ,in, This represents the historical average delivery speed of robot j under normal delivery conditions.

[0188] The analysis of historical differences in delivery task completion efficiency includes:

[0189] Determine the actual changes in delivery task processing speed when robots experience a backlog of delivery tasks in hotspot areas;

[0190] The formula for the actual delivery speed of the robot when there is a backlog of tasks is as follows: ,in, The actual delivery speed of robot j when it has a backlog of tasks; The number of delivery tasks completed by robot j within the statistical period is clearly disclosed from historical records; The length of the backlog of statistical tasks, for example, 10 minutes.

[0191] Analyze the relationship between the backlog of delivery tasks in hotspot areas and the rate of decrease in the processing speed of robot delivery tasks;

[0192] The formula for calculating the decrease in delivery speed is as follows: ,in, The extent of the decrease in robot delivery speed, This represents the historical average delivery speed of robot j under normal conditions.

[0193] Based on the aforementioned relationship, determine the robot's actual task carrying capacity under increased delivery pressure in hotspot areas;

[0194] Specifically, the calculation is performed according to the aforementioned formula for calculating the actual task carrying capacity of the delivery robot.

[0195] Based on the differences in actual task carrying capacity, analyze the differences in historical completion efficiency of delivery tasks;

[0196] Specifically, the formula for the historical completion efficiency difference of delivery tasks is: ,in, For the historical completion efficiency indicators of robot j is the number of tasks backlogged in the kth statistical period; M is the number of historical statistical periods.

[0197] Based on the differences in historical completion efficiency of the delivery tasks, the adaptability of each delivery robot to the risk of path intersection congestion in different related areas is determined.

[0198] Specifically, based on the calculated historical completion efficiency index of robot delivery tasks Define robot j as a pair of associated regions (m, n) Path intersection congestion risk adaptability index ,in, For delivery robot j, pair of regions (m,n) The adaptability index for path intersection congestion risk, with a value range of [0,1]. The closer the value is to 1, the higher the adaptability. To find the region (m,n) in the associated region The highest value among all historical task completion efficiency metrics for all delivery robots within the system.

[0199] Based on the differences in the adaptability of robots to cross-path congestion risks, the number of tasks to be assigned to delivery robots in hotspot areas is determined;

[0200] Specifically, the method for determining the number of delivery robot tasks is as follows: based on the aforementioned adaptability indicators. And consider the associated region pairs (m,n) Total number of delivery requests during a specific peak task period Delivery robot j operates in hotspot areas (m, n) Number of tasks assigned ,in, For hotspot regions (m,n) The sum of the adaptability indicators of all K delivery robots in the system. For hotspot regions (m,n) The total number of delivery requests within a statistical period (such as a peak task period, for example, 07:00-09:00); This is the floor symbol.

[0201] S105: Update the path intersection congestion features in the delivery hotspot area knowledge graph according to the number of tasks assigned to the delivery robots, so as to dynamically adjust the delivery hotspot area knowledge graph;

[0202] Specifically, based on the number of delivery robot tasks assigned in step S104, the latest delivery trajectory data after the robot actually completes the task will be used to update the path intersection congestion features in the knowledge graph. The specific update process includes the following steps:

[0203] After the robot completes its task, it uses an indoor positioning system (UWB positioning system) Real-time collection of the latest delivery trajectory points: ,in, Let i be the coordinates of the i-th trajectory point. The timestamp corresponding to the trajectory point;

[0204] Based on the historical path intersection coordinates stored within the original hotspot area knowledge graph, calculate whether any new abnormal stops or path intersection events have occurred in the latest trajectory. Specifically:

[0205] Determine the location point The criteria for determining whether a location is a new abnormal stop are: ;

[0206] Determine the location point The condition for determining whether a path is a new intersection is: ;

[0207] in, For the duration of stay, This is the threshold for abnormal dwell time, such as 30 seconds. Spatial distance function, specifically Euclidean distance; This is a set of historical abnormal locations where people stayed. This is a set of historical path intersection feature locations; Determine the distance threshold for the new location; Statistics on the number of trajectory intersections at the new location; A frequency threshold is set for fixed paths;

[0208] Newly detected abnormal stops or path intersections will be updated in the delivery hotspot area knowledge graph, generating new knowledge graph triples. The specific format is explicitly disclosed as follows:

[0209] The following is an example of a newly added relation triple: (Region 12","Path Intersection Relationship (Intersection Point: (15.3m, 9.1m), Conflict Count: 23 times / month, Peak Hour: 07:00−09:00)","Region 13");

[0210] Delete or weaken location feature points in the knowledge graph that have not experienced path intersection events for a long period of time (e.g., 3 consecutive months) or whose conflict frequency has significantly decreased.

[0211] After completing the above updates, a new knowledge graph of delivery hotspot areas will be formed to achieve dynamic and adaptive updates of the knowledge graph.

[0212] Example 2

[0213] like Figure 2 As shown in the example, the parts not detailed in this embodiment are as shown in Example 1. This embodiment discloses a hotel management system based on multimodal data behavior analysis, including:

[0214] The hotspot graph construction module 201 is used to establish a knowledge graph of delivery hotspot areas that reflects the characteristics of cross-congestion of delivery routes based on the historical time period distribution characteristics of delivery request signals.

[0215] The cross-feature mining module 202 is used to mine the correlation features between the concentration of delivery tasks and the frequency of cross-delivery routes based on the knowledge graph of delivery hotspot areas.

[0216] The risk capability modeling module 203 is used to determine the path intersection risk level of hotspot areas based on the associated features; and to construct a capability profile of the delivery robot to adapt to path intersection risks based on the robot's historical delivery trajectory data.

[0217] The risk adaptation and loading module 204 is used to determine the number of tasks to be assigned to the delivery robot in the hot spot area based on the adaptation characteristics of the path intersection risk level and the capability profile of the delivery robot in the hot spot area.

[0218] The graph dynamic update module 205 is used to update the path intersection congestion features in the knowledge graph of delivery hotspot areas according to the number of tasks assigned to the delivery robots, so as to dynamically adjust the knowledge graph of delivery hotspot areas.

[0219] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters, weights, and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0220] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired or wireless network. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0221] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A hotel management method based on multimodal data behavior analysis, characterized in that, include: Based on the historical time period distribution characteristics of delivery request signals, a knowledge graph of delivery hotspot areas reflecting the characteristics of cross-congestion along delivery routes is established. Based on the knowledge graph of delivery hotspot areas, we can explore the correlation features between the concentration of delivery tasks and the frequency of intersection of delivery routes; The path intersection risk level of hotspot areas is determined based on the aforementioned correlation characteristics; And based on the robot's historical delivery trajectory data, a profile of the delivery robot's ability to adapt to path intersection risks is constructed; The ability profile for constructing delivery robots to adapt to path intersection risks includes: Based on historical delivery data, the frequency and duration of specific responses taken by robots when delivery routes in hotspot areas become congested are extracted. Determine the specific behavioral preferences of each delivery robot when faced with cross-path congestion in hot areas, such as proactively avoiding obstacles, adjusting task order, or waiting in place. The determination of the specific behavioral preferences of each delivery robot when facing path intersection congestion in hotspot areas includes: Based on the historical task records of the delivery robot, determine the frequency at which the robot will proactively change its original delivery route when faced with the risk of congestion at intersections. Determine the historical statistical distribution of delivery delays caused by changes in robot delivery routes; Statistics on delivery delays caused by delivery robots adjusting task order and waiting in place. Based on the historical statistical distribution of the delivery delay time, determine the actual impact of the robot's proactive route change behavior on delivery efficiency; Based on the differences in actual impact, determine the specific behavioral preferences of each delivery robot when facing path intersection congestion in hotspot areas; Analyze the specific impact relationship between the aforementioned behavioral preferences and the overall response delay of delivery tasks; Based on the specific influencing relationships described above, a capability profile of the delivery robot to adapt to path intersection risks that reflects individual differences among robots is constructed; Based on the adaptation characteristics of the path intersection risk level in hotspot areas and the capability profile of delivery robots, the number of tasks to be assigned to delivery robots in hotspot areas is determined. The path intersection congestion features in the delivery hotspot area knowledge graph are updated based on the number of tasks assigned to delivery robots, so as to dynamically adjust the delivery hotspot area knowledge graph.

2. The hotel management method based on multimodal data behavior analysis according to claim 1, characterized in that, The establishment of a knowledge graph of delivery hotspot areas reflecting the characteristics of congestion at intersections of delivery routes includes: Based on delivery request signals, identify candidate hotspot areas where historical delivery tasks are concentrated and the frequency of delivery robot stops or detours is abnormal; specific implementation includes: Extract delivery trajectory data of delivery robots between the candidate hotspot areas to determine the spatial distribution characteristics of the intersections of actual delivery paths; Based on the spatial distribution characteristics, specific intersection nodes where delivery robots frequently experience path conflicts between hotspot areas are identified; Using the intersection nodes as bridging points for the delivery path associations between hotspot areas, a knowledge graph of delivery hotspot areas representing the risk of congestion at path intersections is established.

3. The hotel management method based on multimodal data behavior analysis according to claim 2, characterized in that, The spatial distribution characteristics of determining the intersections of actual delivery routes include: Based on the historical delivery task data of the delivery robots, identify the locations where the robots actually stopped or experienced abnormal waiting during peak delivery periods; Determine the actual path direction correlation between abnormal stop locations and hotspot areas in the robot delivery route; Based on the actual path direction association, analyze whether there are fixed or repeated path intersections between delivery routes in different hotspot areas; Based on the analysis results, spatial distribution characteristics of the intersections of actual delivery routes between hotspot areas are generated.

4. The hotel management method based on multimodal data behavior analysis according to claim 3, characterized in that, The correlation features between the concentration of delivery tasks and the frequency of intersection of delivery routes include: Based on the knowledge graph of delivery hotspot areas, identify the historical distribution of delivery congestion events at the intersections of delivery routes between hotspot areas; Based on the historical distribution of delivery congestion events, determine the peak periods of delivery congestion in each hotspot area; For regions where the growth rate of delivery tasks exceeds a preset threshold during peak delivery congestion periods, the collaborative change characteristics between the degree of delivery task concentration and the frequency of path intersections are analyzed. When the collaborative change characteristics meet the cross-congestion triggering conditions, determine the correlation characteristics between the concentration of delivery tasks and the frequency of cross-delivery routes.

5. The hotel management method based on multimodal data behavior analysis according to claim 4, characterized in that, The analysis of the coordinated changes between the concentration of delivery tasks and the frequency of route intersections includes: Extract the specific growth rate trend of delivery request signals in each hotspot area during peak delivery congestion periods; Identify specific hotspots that cause a significant increase in the number of path intersection congestion after the rapid increase in the growth rate of the delivery request signal; Determine whether the growth rate of delivery requests in the specific hotspot area and the trend of changes in the number of intersections and congestion on delivery routes both exhibit a rapid increase simultaneously. When the growth rate of delivery requests and the number of path intersection congestion both show a rapid increase, determine the coordinated change characteristics between the concentration of delivery tasks and the frequency of path intersections.

6. The hotel management method based on multimodal data behavior analysis according to claim 5, characterized in that, The determination of the number of tasks assigned to delivery robots in hotspot areas includes: Based on the hotspot area knowledge graph, identify several pairs of related areas with the highest risk of path intersection within the hotspot area; Analyze the differences in historical completion efficiency of delivery tasks among the various delivery robots in the associated regions; Based on the differences in historical completion efficiency of the delivery tasks, the adaptability of each delivery robot to the risk of path intersection congestion in different related areas is determined. Based on the differences in the adaptability of robots to cross-path congestion risks, the number of tasks assigned to delivery robots in hotspot areas is determined.

7. The hotel management method based on multimodal data behavior analysis according to claim 6, characterized in that, The analysis of historical differences in delivery task completion efficiency includes: Determine the actual changes in delivery task processing speed when robots experience a backlog of delivery tasks in hotspot areas; Analyze the relationship between the backlog of delivery tasks in hotspot areas and the rate of decrease in the processing speed of robot delivery tasks; Based on the aforementioned relationship, determine the robot's actual task carrying capacity under increased delivery pressure in hotspot areas; Based on the differences in the actual task carrying capacity, the differences in historical completion efficiency of delivery tasks are analyzed.

8. A hotel management system based on multimodal data behavior analysis, implemented based on the hotel management method based on multimodal data behavior analysis according to any one of claims 1-7, characterized in that, include: The hotspot graph construction module is used to build a knowledge graph of delivery hotspot areas that reflects the characteristics of cross-congestion along delivery routes, based on the historical time period distribution characteristics of delivery request signals. The cross-feature mining module is used to mine the correlation features between the concentration of delivery tasks and the frequency of cross-delivery routes based on the knowledge graph of delivery hotspot areas; The risk capability modeling module is used to determine the path intersection risk level of hotspot areas based on the associated features. And based on the robot's historical delivery trajectory data, a profile of the delivery robot's ability to adapt to path intersection risks is constructed; The risk-adaptive loading module is used to determine the number of tasks to be assigned to delivery robots in hotspot areas based on the adaptation characteristics of the path intersection risk level and the capability profile of the delivery robots. The graph dynamic update module is used to update the path intersection congestion features in the knowledge graph of delivery hotspot areas based on the number of tasks assigned to delivery robots, so as to dynamically adjust the knowledge graph of delivery hotspot areas.

Images