Order queue real-time interpolation method and device fused with user behavior prediction

Abstract

The invention provides an order queue real-time interpolation method and device fused with user behavior prediction, and relates to the technical field of order processing, standard making nodes are positioned according to order features, then dynamic offset analysis is performed on the standard making nodes according to historical behavior features, and updated making nodes are output; inserting the newly added coffee order into a real-time order queue according to the update making node; and according to the space-time-behavior coupling data, food taking time delay prediction is carried out, and a time delay dynamic correction value is output to trigger queue interpolation optimization in a real-time order queue. The technical problems that the existing automatic coffee machine adopts static scheduling to perform order fulfillment, so that the time deviation between making and meal taking is obvious, and resource waste is easily caused by the uncertainty influence of the meal taking behavior of the user are solved. The technical effects of guaranteeing dynamic and accurate matching of the beverage making completion time and the actual meal taking time of the user and avoiding order resource waste caused by abnormal performance of the user are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of order processing, and in particular to a method and device for real-time interpolation of order queues integrating user behavior prediction. Background Art

[0002] Existing automatic coffee machines generally use a static scheduling strategy based on fixed timing to process orders. Their production nodes are usually determined solely based on the preset average production time and the idealized food pickup path.

[0003] Such methods fail to account for the spatiotemporal fluctuations in users' actual mobility (such as sudden route changes and traffic delays), leading to significant discrepancies between drink preparation completion time and actual arrival time. Actual data shows that over 30% of users' meal pickup times deviate by more than five minutes from the scheduled window, leading to quality issues such as hot drinks cooling and iced drinks melting. Furthermore, static scheduling mechanisms are unable to effectively respond to the uncertainties of user behavior, such as order cancellations, modifications, or delayed pickups, increasing the waste rate of prepared drinks by over 15%.

[0004] In summary, existing automatic coffee machines use static scheduling to fulfill orders, which leads to significant deviations between the preparation and pickup times. At the same time, they are subject to the uncertainty of user pickup behavior and are prone to waste of resources. Summary of the Invention

[0005] The present invention provides a real-time interpolation method and device for order queues that integrates user behavior prediction, which is used to solve the technical problems that existing automatic coffee machines use static scheduling to fulfill orders, resulting in significant deviations between production and pickup time, and are prone to resource waste due to the uncertainty of user pickup behavior.

[0006] In view of the above problems, the present invention provides a real-time interpolation method and device for order queues that integrates user behavior prediction.

[0007] The first aspect of the present invention provides a real-time interpolation method for order queues that integrates user behavior prediction, the method comprising: after receiving a new coffee order sent by a new ordering user, locating a standard production node according to order characteristics; locally retrieving historical behavior characteristics according to a user identifier of the new coffee order; performing dynamic offset analysis on the standard production node according to the historical behavior characteristics, and outputting a production node offset, wherein the production node offset is identified by an offset vector; along the offset vector, time-series remapping of the standard production node using the production node offset is performed, and an updated production node is output; inserting the new coffee order into a real-time order queue according to the updated production node; within a preset behavior monitoring window, continuously collecting the spatiotemporal-behavioral coupling data of the new ordering user, and predicting the meal pickup delay based on the spatiotemporal-behavioral coupling data, and outputting a dynamic correction value for the delay; triggering queue interpolation optimization in the real-time order queue according to the dynamic correction value for the delay.

[0008] In one implementation, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the time delay, and the following processing is further performed:

[0009] When the dynamic correction value of the delay exceeds the preset correction threshold, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the delay until the dynamic correction value of the delay is less than the preset correction threshold; after the dynamic correction value of the delay is less than the preset correction threshold, the insertion position of the new coffee order in the real-time order queue is locked; when the real-time order queue progresses to the new coffee order, the order making instruction is started.

[0010] In one implementation, after receiving a new coffee order from a new ordering user, a node is created based on the order feature positioning criteria, and the following processing is also performed:

[0011] Extract the order beverage type, user order coordinates and user order time from the newly added coffee order; associate the production process parameters and the beverage timeliness label based on the order beverage type; after the interactive automatic coffee machine obtains the real-time equipment load, calculate and output the order production time based on the real-time equipment load and the production process parameters; calculate the meal pickup path time based on the user order coordinates; taking the user order time as the starting point, trace back along the meal pickup path time based on the order production time and the beverage timeliness label to locate the standard production node.

[0012] In one implementation, within a preset behavior monitoring window, the spatiotemporal-behavioral coupling data of the newly added ordering user is continuously collected, and the meal pickup delay is predicted based on the spatiotemporal-behavioral coupling data, and a dynamic correction value for the delay is output. The following processing is also performed:

[0013] Perform spatiotemporal stream data collection on the newly added ordering user to obtain the user's GPS trajectory and real-time environmental parameters; perform behavioral stream data collection on the newly added ordering user to obtain the APP interaction data stream, wherein the user's GPS trajectory, real-time environmental parameters and APP interaction data stream constitute the spatiotemporal-behavioral coupling data; perform fulfillment intention attenuation analysis on the spatiotemporal-behavioral coupling data, and output the dynamic correction value of the delay.

[0014] In one implementation, the spatiotemporal-behavior coupling data is subjected to a fulfillment intention attenuation analysis, the delay dynamic correction value is output, and the following processing is further performed:

[0015] The actual local path time is called from the user GPS trajectory; based on the proportional characteristics of the user GPS trajectory and the global trajectory of meal pickup, the standard local path time is decomposed from the meal pickup path time; the behavior intention attenuation feature is calculated and output based on the actual local path time and the standard local path time; the environmental intention attenuation feature of the real-time environmental parameters is matched; the order modification operation is called from the APP interactive data stream, and the fulfillment intention attenuation feature is output; the intention attenuation feature, the environmental intention attenuation feature and the fulfillment intention attenuation feature are used to traverse the delay correction information library, and the dynamic correction value of the delay is matched and output.

[0016] In one implementation, a dynamic offset analysis is performed on the standard production node based on the historical behavior characteristics, and a production node offset is output. The following processing is also performed:

[0017] Vectorize the historical behavior characteristics and output a meal pickup punctuality vector, a route stability vector and a cancellation risk vector; interactively obtain multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets; perform multivariate regression analysis on the multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets, and output an offset calculation function; complete the construction of an offset prediction model by configuring a calculation engine for the offset calculation function; input the meal pickup punctuality vector, route stability vector and cancellation risk vector into the offset prediction model to perform dynamic offset calculation, and output the production node offset; obtain the offset vector by normalizing the meal pickup punctuality vector, route stability vector and cancellation risk vector.

[0018] In one implementation, along the offset vector, the standard production node is time-sequentially remapped using the production node offset to output an updated production node, and the following processing is further performed:

[0019] Taking the standard production node as the starting point, the production node offset is reversed along the offset vector time sequence to locate and update the production interval; using the updated production interval as a constraint, multiple real-time load information of multiple coffee machine equipment units in the automatic coffee machine is called; order conflict judgment is performed based on the multiple real-time load information, and a conflict-free node is located in the target equipment unit as the updated production node.

[0020] According to a second aspect of the present invention, a real-time interpolation device for an order queue integrating user behavior prediction is provided, the device comprising: a production positioning unit for receiving a new coffee order sent by a new ordering user and locating a standard production node according to order characteristics; a behavior backtracking unit for locally retrieving historical behavior characteristics according to a user identifier of the new coffee order; an offset analysis unit for performing dynamic offset analysis on the standard production node according to the historical behavior characteristics and outputting a production node offset, wherein the production node offset is identified by an offset vector; a production update unit for performing time-series remapping on the standard production node along the offset vector using the production node offset and outputting an updated production node; an order interpolation unit for inserting the new coffee order into a real-time order queue according to the updated production node; a delay correction unit for continuously collecting spatiotemporal-behavioral coupling data of the new ordering user within a preset behavior monitoring window, and predicting the meal pickup delay based on the spatiotemporal-behavioral coupling data, and outputting a dynamic correction value for the delay; and an interpolation optimization unit for triggering queue interpolation optimization in the real-time order queue according to the dynamic correction value for the delay.

[0021] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0022] The method provided by the embodiment of the present invention locates the standard production node according to the order characteristics after receiving the new coffee order sent by the new ordering user; retrieves the historical behavior characteristics locally according to the user identifier of the new coffee order; performs dynamic offset analysis on the standard production node according to the historical behavior characteristics, and outputs the production node offset, wherein the production node offset is identified by an offset vector; performs time sequence remapping of the standard production node along the offset vector using the production node offset, and outputs an updated production node; inserts the new coffee order into the real-time order queue according to the updated production node; continuously collects the spatiotemporal-behavioral coupling data of the new ordering user within the preset behavior monitoring window, and predicts the meal pickup delay according to the spatiotemporal-behavioral coupling data, and outputs a dynamic correction value for the delay; triggers queue interpolation optimization in the real-time order queue according to the dynamic correction value for the delay. This achieves the technical effect of ensuring the dynamic and accurate matching of the beverage production completion time with the actual meal pickup time of the user, and avoiding the waste of order resources caused by abnormal user fulfillment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic flow chart of the method for real-time interpolation of order queues integrating user behavior prediction provided by the present invention is shown;

[0024] Figure 2 A structural diagram of the order queue real-time interpolation device integrated with user behavior prediction provided by the present invention is shown.

[0025] Explanation of the accompanying symbols: production positioning unit 1, behavior backtracking unit 2, offset analysis unit 3, production update unit 4, order interpolation unit 5, delay correction unit 6, interpolation optimization unit 7. DETAILED DESCRIPTION

[0026] The present invention provides a real-time interpolation method and device for order queues that integrates user behavior prediction, which is used to solve the technical problems that existing automatic coffee machines use static scheduling to fulfill orders, resulting in significant deviations between production and pickup time, and are prone to resource waste due to the uncertainty of user pickup behavior.

[0027] Below, the technical solutions of the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the accompanying drawings.

[0028] Example 1, a flowchart of a method for real-time interpolation of order queues integrating user behavior prediction provided by an embodiment of the present invention, see Figure 1 , the method comprising:

[0029] S100: After receiving a new coffee order from a new ordering user, a node is created based on the order feature positioning standard.

[0030] In one implementation, after receiving a new coffee order from a new ordering user, a node is created based on the order feature positioning standard. Step S100 of the method provided by the present invention includes:

[0031] S110: Extract the order beverage type, user order coordinates and user order time from the newly added coffee order.

[0032] S120: Associating the production process parameters and the beverage timeliness label according to the type of beverage ordered.

[0033] S130: After obtaining the real-time equipment load, the interactive automatic coffee machine calculates and outputs the order production time based on the real-time equipment load and production process parameters.

[0034] S140: Calculate the time required to pick up the meal based on the coordinates of the user's order.

[0035] S150: Starting from the time when the user places the order, based on the order production time and the drink timeliness label, the standard production node is located by tracing back along the meal pickup path.

[0036] Specifically, the system extracts three key pieces of data from newly ordered coffee orders: the type of drink ordered (e.g., Americano, cappuccino), the user's geographic coordinates at the time of order placement (typically obtained through app location tracking or manual input), and the timestamp of the order submission. This data forms the raw input for subsequent calculations and provides the foundational parameters for personalized scheduling.

[0037] According to the beverage type, the corresponding production process parameters (such as the extraction time of concentrated coffee liquid and the heating temperature of milk steam) and timeliness labels (such as the best drinking period for hot drinks is 8 minutes and the melting sensitive period for iced drinks is 15 minutes) are associated.

[0038] Interact with the automatic coffee machine's control system in real time to obtain the current equipment load status (such as the number of parallel production slots occupied and the grinder's operating frequency), and calculate the theoretical production time of the order based on the beverage's process parameters. For example, when the equipment load rate reaches 70%, espresso extraction may require an additional 10 seconds to wait for the queue to release resources.

[0039] Based on the geographic location coordinates of the newly added ordering user when placing the order, the path planning algorithm (integrated with map API or indoor positioning data) is called to calculate the estimated time it takes for the user to reach the pick-up point.

[0040] Using the user's order time as the initial anchor point, we work backwards to calculate the latest start time for drink preparation: adding the order preparation time and the food pickup time, and introducing a buffer time based on timeliness tags (e.g., hot drinks should be prepared 2 minutes in advance to allow for warming time), to determine the final standard preparation time. For example, if the user expects to arrive in 10 minutes and the preparation time is 5 minutes, the standard preparation time is set to the current time + 3 minutes.

[0041] This embodiment, after receiving a coffee order from a newly added ordering user, first determines the standardized beverage production start time node based on the key features of the order, thereby dynamically matching user demand with equipment production capacity, ensuring that the beverage production completion time is accurately aligned with the user's actual meal pickup time, and avoiding quality degradation due to premature production or waiting delays caused by late production.

[0042] S200: Locally retrieve historical behavior features based on the user identifier of the newly added coffee order.

[0043] After receiving a new coffee order, the user's historical behavior records are retrieved from the local database using a user identifier (such as a mobile phone number or membership ID). These behavioral characteristics include the user's past on-time pickup (such as average early or late arrival time), route preference and stability to the pickup point (such as whether the route is frequently changed), and the historical frequency of order cancellations or modifications (such as the probability of canceling an order on rainy days). For example, if a user's historical data shows an average of 5 minutes late pickup and a 30% cancellation rate on rainy days, these characteristics will be extracted for subsequent analysis.

[0044] S300: Performing dynamic offset analysis on the standard production node according to the historical behavior characteristics, and outputting a production node offset, wherein the production node offset is identified by an offset vector.

[0045] In one implementation, a dynamic offset analysis is performed on the standard production node based on the historical behavior characteristics to output the production node offset. Step S300 of the method provided by the present invention includes:

[0046] S310: Vectorize the historical behavior features and output a meal pickup punctuality vector, a route stability vector, and a cancellation risk vector.

[0047] S320: Interactively obtain multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets.

[0048] S330: Perform a multivariate regression analysis on the multiple sample punctuality vectors, the multiple sample stability vectors, the multiple sample risk vectors and the multiple sample node offsets, and output an offset calculation function.

[0049] S340: Complete the construction of the offset prediction model by configuring the calculation engine for the offset calculation function.

[0050] S350: Input the meal pickup punctuality vector, route stability vector and cancellation risk vector into the offset prediction model to dynamically calculate the offset and output the production node offset.

[0051] S360: The offset vector is obtained by normalizing the meal pickup punctuality vector, route stability vector and cancellation risk vector.

[0052] Specifically, this embodiment converts user historical behavior data into a computable vector form. The meal pickup punctuality vector contains the mean and fluctuation range of the user's arrival time (e.g., an average lateness of 3 minutes, with a standard deviation of 2 minutes). The route stability vector maps the discreteness of the user's path selection (e.g., 80% choose Route A, 20% choose Route B). The cancellation risk vector associates the user's cancellation probability in different scenarios (e.g., a 15% cancellation rate during the evening rush hour on weekdays, increasing to 40% on rainy days). For example, a user's vector might be represented as [3 minutes late, 80% of Route A arrivals, 40% cancellation rate on rainy days].

[0053] A large amount of user sample data is obtained from the historical order database, including multiple punctuality vectors, multiple route stability vectors, multiple risk vectors and their corresponding multiple node offset records for multiple other users. The node offset record here is the length of time to advance or delay the production of the node (such as a delay of 2 minutes).

[0054] Through multivariate regression analysis, a mathematical relationship between behavioral characteristics and node offsets was established. For example, the analysis found that for every 1-minute increase in the standard deviation of the punctuality vector, the time offset must increase by 0.5 minutes. For every 10% increase in the cancellation risk vector, the spatial offset must increase by 0.3 device units.

[0055] The resulting offset calculation function quantifies the specific impact of behavioral characteristics on node adjustments, and then transforms the function derived from regression analysis into a real-time computing engine. For example, a lightweight TensorFlow model deployed on an embedded device can process 50 offset prediction requests per second. The engine continuously receives real-time meal pickup punctuality vectors, route stability vectors, and cancellation risk vectors, and dynamically updates the offset calculation results.

[0056] The on-time pickup vector, route stability vector, and cancellation risk vector are input into the offset prediction model to dynamically calculate the offset and output the production node offset. For example, for a user with an average 4-minute lateness, high route volatility, and a 35% cancellation rate on rainy days, the model might output an adjustment strategy of "delaying production by 3 minutes."

[0057] After completing the vectorization and regression analysis of historical behavior characteristics, the three vectors of meal pickup punctuality, route stability, and cancellation risk are normalized and mapped to a unified numerical range (such as [-1, 1]) to generate a standardized offset vector.

[0058] The core goal of this step is to transform the impact of user behavior characteristics on production nodes into spatiotemporal adjustment parameters that can be directly manipulated by the equipment scheduling system. For example, if a user's historical behavior analysis shows that they are on average 5 minutes late for meal pickup (punctuality vector value 0.8), have high route selection volatility (route stability vector value -0.3), and have a 40% cancellation rate on rainy days (risk vector value 0.6), the resulting offset vector after normalization might be [0.8, -0.3, 0.6].

[0059] This embodiment dynamically adjusts the preliminarily determined standard production nodes based on the retrieved historical behavior characteristics of the user, thereby indirectly achieving the technical effect of ensuring that the user arrives just after the beverage is completed.

[0060] S400: along the offset vector, using the production node offset to perform time sequence remapping on the standard production node, and output an updated production node.

[0061] In one implementation, along the offset vector, the standard production node is time-sequentially remapped using the production node offset to output an updated production node. Step S400 of the method provided by the present invention includes:

[0062] S410: Taking the standard production node as the starting point, reversely calculate the production node offset along the offset vector time sequence to locate and update the production interval.

[0063] S420: Using the update production interval as a constraint, calling multiple real-time load information of multiple coffee machine equipment units in the automatic coffee machine.

[0064] S430: Determine order conflicts based on the multiple real-time load information, and locate a conflict-free node in the target equipment unit as the updated production node.

[0065] Specifically, the standard production node is used as the initial time reference, and the adjustable time window and spatial tolerance are reversely derived according to the components of each dimension of the normalized offset vector.

[0066] For example, a user's normalized offset vector is [0.8, -0.3, 0.6]. The punctuality component of 0.8 indicates a production delay (assuming a maximum allowable delay of 10 minutes, the actual delay is 8 minutes). The route stability component of -0.3 triggers the device switching logic (switching from the default device unit A to unit B, which has a lower load but is slightly farther away). The risk component of 0.6 is canceled, adding a 1.8-minute buffer (calculated as 30 seconds per 0.1). By combining these parameters, the original production time (for example, 10:00) is expanded to a flexible range (10:06.2 to 10:09.8), and device unit B is marked as a candidate.

[0067] This process quantifies user behavior characteristics into spatiotemporal parameters, dynamically expands the production window, and provides flexible adjustment space for subsequent conflict detection and load balancing. It not only adapts to users' habitual lateness or path fluctuations, but also avoids queue congestion caused by equipment overload.

[0068] After determining the flexible production time window (such as 10:06.2 to 10:09.8) and the candidate equipment unit (such as unit B), the working status data of all relevant coffee machine equipment units are retrieved in real time, including the number of orders in the current queue, the estimated completion time of each order, the upper limit of the equipment's parallel processing capacity, and other information.

[0069] Conflict detection is performed on candidate equipment units to determine whether inserting a new order within the flexible time window will cause time overlap or overload. For example, if the original order for unit B is due at 10:07, and the new order needs to be started at 10:06.2 and takes 5 minutes, the time overlap with the subsequent order (starting at 10:08) exceeds 75% (overlap time 3.8 minutes / 5 minutes), triggering a conflict flag. At this time, the new order is automatically switched to unit C (the target equipment unit), whose idle period completely covers 10:06.2 to 10:09.8 and has no parallel capacity limit. The order production task node without conflict located in the target equipment unit is used as the update production node.

[0070] This embodiment achieves the technical effect of updating the production node according to the user's behavior characteristics, ensuring that coffee production and user meal collection are compatible.

[0071] S500: Insert the newly added coffee order into the real-time order queue according to the updated production node.

[0072] Specifically, based on the updated timestamp and device unit of the production node, the newly added coffee order is inserted into the appropriate position in the real-time order queue. For example, if a user's order is determined to start at 10:06.2 on device unit C after adjustment, and there is an order starting at 10:04 (expected to be completed at 10:08) and a reserved window of 10:10 in the unit C queue, the new order is precisely inserted into the time period of 10:06.2-10:11.2, avoiding time overlap with the previous order (with a gap of 1.8 minutes) while preserving a full operation window for subsequent orders.

[0073] S600: Within the preset behavior monitoring window, continuously collect the spatiotemporal-behavioral coupling data of the newly added ordering user, and predict the meal pickup delay based on the spatiotemporal-behavioral coupling data, and output a dynamic correction value for the delay.

[0074] In one implementation, within a preset behavior monitoring window, the spatiotemporal-behavioral coupling data of the newly added ordering user is continuously collected, and the meal pickup delay is predicted based on the spatiotemporal-behavioral coupling data, and a dynamic correction value for the delay is output. The method step S600 provided by the present invention includes:

[0075] S610: Collect spatiotemporal flow data of the newly added ordering user to obtain the user's GPS trajectory and real-time environmental parameters.

[0076] S620: Collect behavior stream data of the newly added ordering user to obtain APP interaction data stream, wherein the user GPS trajectory, real-time environmental parameters and APP interaction data stream constitute the spatiotemporal-behavioral coupling data.

[0077] S630: Perform performance intention attenuation analysis on the spatiotemporal-behavior coupling data, and output the dynamic correction value of the delay.

[0078] In one implementation, the spatiotemporal-behavior coupling data is subjected to a fulfillment intention attenuation analysis to output the delay dynamic correction value. Step S630 of the method provided by the present invention includes:

[0079] S631: The time taken to call the actual local path from the user GPS track.

[0080] S632: Based on the ratio characteristics of the user's GPS trajectory and the global trajectory of meal pickup, the standard local path time is decomposed from the meal pickup path time.

[0081] S633: Calculate and output the behavior intention attenuation feature based on the actual local path time and the standard local path time.

[0082] S634: Match the environmental intention attenuation characteristics of the real-time environmental parameters.

[0083] S635: Call the order modification operation from the APP interactive data flow and output the fulfillment intention attenuation feature.

[0084] S636: Use the intention attenuation feature, environmental intention attenuation feature and performance intention attenuation feature to traverse the delay correction information library, match and output the delay dynamic correction value.

[0085] Specifically, through the integrated positioning service interface (such as GPS / Beidou), the user's movement trajectory is collected in real time, and the spatiotemporal flow data such as latitude and longitude coordinates, movement speed and direction changes are recorded. At the same time, third-party APIs such as meteorology and traffic are connected to obtain real-time environmental parameters (such as rainfall intensity and road congestion index).

[0086] For example, a user is moving along a street at walking speed and suddenly encounters a road section that is congested due to an accident. The real-time travel time of the road section is immediately revised from 5 minutes to 15 minutes, which serves as the key input for subsequent delay prediction.

[0087] The system monitors users' interactions within the coffee app throughout the entire process, including page dwell time, order detail viewing frequency, and modification or cancellation records. For example, if a user opens the order tracking page three times within 10 minutes of placing an order and attempts to change "dine-in" to "takeout," this behavior stream data will be captured in real time and analyzed in conjunction with spatiotemporal stream data (such as slowing movement speed) to determine whether their willingness to fulfill their order has decreased.

[0088] The specific data processing process for prediction of performance willingness decay is as follows:

[0089] The actual local route duration for the user's current location (e.g., the last 500 meters from a coffee shop) is extracted from the user's real-time GPS trajectory. For example, a user might spend 7 minutes traveling the last 500 meters due to red lights and heavy pedestrian traffic, while the historical average travel time for this section is only 3 minutes. This discrepancy directly reflects the degree of decline in the user's current mobility efficiency.

[0090] According to the proportion of GPS track points completed by the user in the global track of picking up the meal (such as 70% of the total path has been completed), the initial estimated time of picking up the meal is proportionally divided into the standard local path time required to complete the remaining local paths.

[0091] The user's actual local path time is compared with the standard local path time to calculate the behavioral intention decay feature. For example, if the standard time for the remaining 500 meters is 3 minutes, and the user actually took 5 minutes, the decay rate is (5-3) / 3≈66.7%. This value is encoded as a behavioral intention decay vector (e.g., [0.667]), which indicates the degree of decline in the user's willingness or ability to move.

[0092] Match the weights of the impact of real-time environmental parameters (such as rainfall, temperature, and traffic events) on environmental intent attenuation. For example, a red alert for heavy rain corresponds to an environmental attenuation coefficient of 0.6 (increasing the probability of delay by 60%). Combined with the 50% decrease in travel speed due to waterlogging on the user's current road section, an environmental intent attenuation feature vector (such as [0.6, 0.5]) is generated.

[0093] Analyze user order action records within the app to quantify fulfillment intention decay. For example, if a user clicks "Change Pickup Time" twice within 10 minutes of placing an order and cancels the original delivery method, this behavior is mapped to a fulfillment decay feature (e.g., [0.4]), indicating that their pickup certainty is reduced by 40%.

[0094] The intention decay characteristics, environmental intention decay characteristics, and fulfillment intention decay characteristics are input into the delay correction information database for pattern matching. For example, the delay correction information database stores historical cases: when the user behavior decay rate is greater than 50%, the environmental decay coefficient is greater than 0.5, and the fulfillment decay is greater than 0.3, the average delay correction value is +5 minutes.

[0095] After retrieving similar cases, the current actual parameters (66.7%, 0.6, and 0.4 in this example) are combined to perform weighted calculation to output the dynamic correction value of the delay (such as +6 minutes), and the value is fed back to the queue scheduling module in real time.

[0096] This embodiment continuously collects multiple data such as the user's geographic location movement trajectory, changes in the surrounding environment, and APP operation behavior, and comprehensively predicts the deviation between the user's actual meal pick-up time and the estimated time to correct the expected arrival time, achieving the technical effect of ensuring that queue scheduling always adapts to the user's real-time status.

[0097] S700: Trigger queue interpolation optimization in the real-time order queue according to the dynamic correction value of the delay.

[0098] In one implementation, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the time delay. Step S700 of the method provided by the present invention includes:

[0099] S710: When the dynamic correction value of the delay exceeds the preset correction threshold, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the delay until the dynamic correction value of the delay is less than the preset correction threshold.

[0100] S720: After the dynamic correction value of the delay is less than the preset correction threshold, the insertion position of the new coffee order in the real-time order queue is locked.

[0101] S730: When the real-time order queue progresses to the newly added coffee order, the order making instruction is started.

[0102] Specifically, in this embodiment, if the dynamic delay correction value exceeds a preset threshold (e.g., +3 minutes), queue interpolation optimization is immediately initiated. For example, if a user's delay is +5 minutes due to traffic congestion, the system first checks whether there is an adjacent order in the queue that can be replaced (e.g., a low-priority iced drink order), postpones it, and inserts it into the current user's adjusted time slot (e.g., from the original 10:12 to 10:17).

[0103] If after the adjustment, the dynamic correction value of the delay output after the time-space-behavior coupling data collection and fulfillment intention attenuation analysis of the user is still +4 minutes (still exceeding the threshold), then continue to look for the next replaceable order until the dynamic correction value of the delay output after the time-space-behavior coupling data collection and fulfillment intention attenuation analysis converges to within +2 minutes, indicating that the user will no longer breach the contract and fail to pick up the meal.

[0104] At this time, the final insertion position of the order in the queue is locked to prohibit subsequent order insertion interference, and when the real-time order queue progresses to the newly added coffee order, the order production instruction is started to produce the newly added coffee order on time.

[0105] This embodiment achieves the technical effect of ensuring that the error between the user's actual meal pick-up time and the optimal drink state window is controllable by continuously iterating the dynamic correction value of the delay until the dynamic correction value of the delay falls within the regression threshold (for example, the correction value is reduced to +2 minutes after multiple adjustments).

[0106] This embodiment as a whole achieves the technical effect of ensuring dynamic and accurate matching between the beverage production completion time and the user's actual meal pickup time, thereby avoiding waste of order resources caused by abnormal user fulfillment.

[0107] Example 2, based on the same inventive concept as the real-time interpolation method of order queues integrating user behavior prediction in the above embodiment, Figure 2 As shown, the present invention provides a real-time interpolation device for order queues integrating user behavior prediction, wherein the device includes:

[0108] The production positioning unit 1 is used to receive a new coffee order sent by a new ordering user and produce a node according to the order feature positioning standard.

[0109] The behavior backtracking unit 2 is used to locally retrieve historical behavior features according to the user identifier of the newly added coffee order.

[0110] The offset analysis unit 3 is configured to perform dynamic offset analysis on the standard production node according to the historical behavior characteristics, and output a production node offset, wherein the production node offset is identified by an offset vector.

[0111] The production updating unit 4 is configured to perform time sequence remapping on the standard production node along the offset vector using the production node offset, and output an updated production node.

[0112] The order interpolation unit 5 is used to insert the newly added coffee order into the real-time order queue according to the updated production node.

[0113] The delay correction unit 6 is used to continuously collect the spatiotemporal-behavioral coupling data of the newly added ordering user within a preset behavior monitoring window, predict the meal pickup delay based on the spatiotemporal-behavioral coupling data, and output a dynamic delay correction value.

[0114] The interpolation optimization unit 7 is used to trigger queue interpolation optimization in the real-time order queue according to the dynamic correction value of the delay.

[0115] In one implementation, the interpolation optimization unit 7 is further configured to:

[0116] When the dynamic correction value of the delay exceeds the preset correction threshold, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the delay until the dynamic correction value of the delay is less than the preset correction threshold; after the dynamic correction value of the delay is less than the preset correction threshold, the insertion position of the new coffee order in the real-time order queue is locked; when the real-time order queue progresses to the new coffee order, the order making instruction is started.

[0117] In one implementation, the manufacturing and positioning unit 1 is further configured to:

[0118] Extract the order beverage type, user order coordinates and user order time from the newly added coffee order; associate the production process parameters and the beverage timeliness label based on the order beverage type; after the interactive automatic coffee machine obtains the real-time equipment load, calculate and output the order production time based on the real-time equipment load and the production process parameters; calculate the meal pickup path time based on the user order coordinates; taking the user order time as the starting point, trace back along the meal pickup path time based on the order production time and the beverage timeliness label to locate the standard production node.

[0119] In one implementation, the delay correction unit 6 is further configured to:

[0120] Perform spatiotemporal stream data collection on the newly added ordering user to obtain the user's GPS trajectory and real-time environmental parameters; perform behavioral stream data collection on the newly added ordering user to obtain the APP interaction data stream, wherein the user's GPS trajectory, real-time environmental parameters and APP interaction data stream constitute the spatiotemporal-behavioral coupling data; perform fulfillment intention attenuation analysis on the spatiotemporal-behavioral coupling data, and output the dynamic correction value of the delay.

[0121] In one implementation, the delay correction unit 6 is further configured to:

[0122] The actual local path time is called from the user GPS trajectory; based on the proportional characteristics of the user GPS trajectory and the global trajectory of meal pickup, the standard local path time is decomposed from the meal pickup path time; the behavior intention attenuation feature is calculated and output based on the actual local path time and the standard local path time; the environmental intention attenuation feature of the real-time environmental parameters is matched; the order modification operation is called from the APP interactive data stream, and the fulfillment intention attenuation feature is output; the intention attenuation feature, the environmental intention attenuation feature and the fulfillment intention attenuation feature are used to traverse the delay correction information library, and the dynamic correction value of the delay is matched and output.

[0123] In one implementation, the offset analysis unit 3 is further configured to:

[0124] Vectorize the historical behavior characteristics and output a meal pickup punctuality vector, a route stability vector and a cancellation risk vector; interactively obtain multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets; perform multivariate regression analysis on the multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets, and output an offset calculation function; complete the construction of an offset prediction model by configuring a calculation engine for the offset calculation function; input the meal pickup punctuality vector, route stability vector and cancellation risk vector into the offset prediction model to perform dynamic offset calculation, and output the production node offset; obtain the offset vector by normalizing the meal pickup punctuality vector, route stability vector and cancellation risk vector.

[0125] In one implementation, the production and updating unit 4 is further configured to:

[0126] Taking the standard production node as the starting point, the production node offset is reversed along the offset vector time sequence to locate and update the production interval; using the updated production interval as a constraint, multiple real-time load information of multiple coffee machine equipment units in the automatic coffee machine is called; order conflict judgment is performed based on the multiple real-time load information, and a conflict-free node is located in the target equipment unit as the updated production node.

[0127] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A real-time interpolation method for order queues that integrates user behavior prediction is characterized by: include: After receiving a new coffee order from a new ordering user, create a node based on the order feature positioning criteria; Locally retrieving historical behavior features based on the user identifier of the newly added coffee order; Performing a dynamic offset analysis on the standard production node according to the historical behavior characteristics, and outputting a production node offset, wherein the production node offset is identified by an offset vector; along the offset vector, using the production node offset to perform time sequence remapping on the standard production node, and output an updated production node; Inserting the newly added coffee order into a real-time order queue according to the updated production node; Within a preset behavior monitoring window, continuously collect the spatiotemporal-behavioral coupling data of the newly added ordering user, predict the meal pickup delay based on the spatiotemporal-behavioral coupling data, and output a dynamic correction value for the delay; Queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the time delay.

2. The real-time interpolation method for order queues integrating user behavior prediction according to claim 1 is characterized in that: Triggering queue interpolation optimization in the real-time order queue according to the dynamic correction value of the time delay, including: When the dynamic correction value of the delay exceeds a preset correction threshold, queue interpolation optimization is triggered in the real-time order queue according to the dynamic correction value of the delay until the dynamic correction value of the delay is less than the preset correction threshold; After the dynamic correction value of the time delay is less than the preset correction threshold, locking the insertion position of the newly added coffee order in the real-time order queue; When the real-time order queue progresses to the newly added coffee order, the order making instruction is started.

3. The real-time interpolation method for order queues integrating user behavior prediction according to claim 1 is characterized in that: After receiving a new coffee order from a new ordering user, create a node based on the order feature positioning criteria, including: Extracting the order beverage type, user order coordinates, and user order time from the newly added coffee order; Associating production process parameters and beverage timeliness labels according to the type of beverage ordered; After obtaining the real-time equipment load, the interactive automatic coffee machine calculates and outputs the order production time based on the real-time equipment load and production process parameters; Calculate the time taken to pick up the meal based on the coordinates of the user's order; Taking the user's ordering time as the starting point, the standard production node is located by backtracking along the meal pickup path based on the order production time and the beverage timeliness label.

4. The real-time interpolation method for order queues integrating user behavior prediction according to claim 3 is characterized in that: Within a preset behavior monitoring window, continuously collect the spatiotemporal-behavioral coupling data of the newly added ordering user, perform meal pickup delay prediction based on the spatiotemporal-behavioral coupling data, and output a dynamic correction value for the delay, including: Collect spatiotemporal data of the newly added ordering user to obtain the user's GPS trajectory and real-time environmental parameters; Collecting behavior stream data of the newly added ordering user to obtain an APP interaction data stream, wherein the user GPS trajectory, real-time environmental parameters and APP interaction data stream constitute the spatiotemporal-behavioral coupling data; Performing performance intention attenuation analysis on the spatiotemporal-behavioral coupling data and outputting the dynamic correction value of the time delay.

5. The real-time interpolation method for order queues integrating user behavior prediction according to claim 4 is characterized in that: Performing performance intention attenuation analysis on the spatiotemporal-behavior coupling data and outputting the delay dynamic correction value includes: The time taken to call the actual local path from the user's GPS trajectory; Decomposing the standard local path time from the meal pickup path time based on the ratio characteristics of the user's GPS trajectory and the global meal pickup trajectory; Calculate and output the behavior intention attenuation feature according to the actual local path time and the standard local path time; An environmental intention attenuation characteristic matching the real-time environmental parameters; Invoke the order modification operation from the APP interaction data stream and output the fulfillment intention decay feature; The intention attenuation feature, the environmental intention attenuation feature and the performance intention attenuation feature are used to traverse the delay correction information library, and the delay dynamic correction value is matched and output.

6. The real-time interpolation method for order queues integrating user behavior prediction according to claim 1 is characterized in that: Performing dynamic offset analysis on the standard production node based on the historical behavior characteristics and outputting the production node offset includes: Vectorize the historical behavior features and output a meal pickup punctuality vector, a route stability vector, and a cancellation risk vector; Interactively obtain multiple sample punctuality vectors, multiple sample stability vectors, multiple sample risk vectors and multiple sample node offsets; Performing a multivariate regression analysis on the multiple sample punctuality vectors, the multiple sample stability vectors, the multiple sample risk vectors, and the multiple sample node offsets, and outputting an offset calculation function; By configuring a calculation engine for the offset calculation function, the construction of the offset prediction model is completed; Inputting the meal pickup punctuality vector, route stability vector, and cancellation risk vector into an offset prediction model to dynamically calculate the offset, and outputting the production node offset; The offset vector is obtained by normalizing the meal pickup punctuality vector, the route stability vector, and the cancellation risk vector.

7. The real-time interpolation method for order queues integrating user behavior prediction according to claim 3 is characterized in that: Performing time sequence remapping on the standard production node along the offset vector using the production node offset to output an updated production node, including: Taking the standard production node as the starting point, reversely inferring the production node offset along the offset vector time sequence to locate and update the production interval; Based on the update production interval as a constraint, calling multiple real-time load information of multiple coffee machine equipment units in the automatic coffee machine; Order conflict judgment is performed based on the multiple real-time load information, and a conflict-free node is located in the target equipment unit as the updated production node.

8. A real-time interpolation device for order queues integrating user behavior prediction, characterized in that: The steps for implementing the method according to any one of claims 1 to 7 include: A positioning unit is used to receive a new coffee order from a new ordering user and create a node based on the order feature positioning standard; a behavior tracing unit, configured to locally retrieve historical behavior features based on a user identifier of the newly added coffee order; an offset analysis unit, configured to perform dynamic offset analysis on the standard production node according to the historical behavior characteristics, and output a production node offset, wherein the production node offset is identified by an offset vector; a production updating unit, configured to perform time sequence remapping on the standard production node along the offset vector using the production node offset, and output an updated production node; An order interpolation unit, configured to insert the newly added coffee order into a real-time order queue according to the updated production node; A delay correction unit is configured to continuously collect spatiotemporal-behavioral coupling data of the newly added ordering user within a preset behavior monitoring window, predict the meal pickup delay based on the spatiotemporal-behavioral coupling data, and output a dynamic delay correction value; An interpolation optimization unit is used to trigger queue interpolation optimization in the real-time order queue according to the dynamic correction value of the delay.

Images