Self-learning intelligent order sending method based on cloud big data

Abstract

The invention relates to the technical field of big data analysis, in particular to a self-learning intelligent order dispatching method based on cloud big data, and the method comprises the steps: obtaining historical clue data and the clue number of the current month; determining an order sending mode for the clues based on a comparison result of a ratio of the number of clues in the current month to the historical monthly average number of clues in the same period and a preset ratio, wherein the order sending mode comprises random order sending, basic completion duration determined based on a target closeness degree characterization value, order sending based on the determined basic completion duration, and scheduling resource order sending; when it is judged that scheduling resource order dispatching is carried out, a constraint mode is determined based on the historical order completion condition of scheduling resources, and constraint types comprise process node constraints and completion timeliness constraints; according to the method and the system, intelligent order dispatching is realized by improving the accuracy of analyzing the number of the tasks and the closeness degree of the target and the order dispatching platform, so that the task processing efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to a self-learning intelligent order dispatching method based on cloud-based big data. Background Technology

[0002] Against the backdrop of rapid development in the digital economy, various service industries (such as logistics and delivery, housekeeping services, repair and installation, and internet-based transportation) have experienced explosive growth in business volume. Users are also increasingly demanding faster service response times, more accurate resource matching, and higher quality task completion. Traditional order dispatching models are gradually revealing problems such as low efficiency, insufficient resource utilization, and difficulty in handling complex business scenarios, making them unable to meet the industry's needs for large-scale and refined operations.

[0003] Chinese Patent Application No. CN115545374A discloses an intelligent order dispatching method and system based on big data. The method includes: acquiring work order data; determining the location information of the ordering user based on the work order data; determining the ordering channel, service type, and product category based on the work order data; acquiring service provider information; acquiring service information of service personnel; performing big data processing on the ordering user's location information, ordering channel, service type, product category, service provider information, and service personnel's service information, and dispatching the service personnel who processed the work order data. This invention processes the ordering user's location information, ordering channel, service type, product category, service provider information, and service personnel's service information using big data algorithms, and can directly dispatch the corresponding service personnel to the ordering user, effectively improving order dispatching efficiency and reducing labor costs.

[0004] However, existing technologies still have the following problems: The system lacks the ability to dynamically adapt to fluctuations in business volume, has insufficient user segmentation and service priority division, lacks a refined resource constraint mechanism, and is not timely or accurate enough in terms of task risk warnings. Summary of the Invention

[0005] To address these issues, this invention provides a self-learning intelligent order dispatching method based on cloud-based big data, which overcomes the problems in existing technologies such as lack of dynamic adaptability to business volume fluctuations, insufficient user segmentation and service priority division, insufficiently refined resource constraint mechanisms, and inadequate timeliness and accuracy of task risk warnings.

[0006] To achieve the above objectives, this invention provides a self-learning intelligent order dispatching method based on cloud-based big data. It includes: Step S1: Obtain historical clue data and the number of clues for the current month; Step S2, based on the comparison between the ratio of the number of leads in the current month to the historical average number of leads in the same period and a preset ratio, determines the order dispatch method for the leads, including: Random order assignment The baseline completion time is determined based on the target's degree of proximity, and orders are dispatched based on this baseline completion time. And dispatching resources and assigning orders; Step S3: When determining to dispatch resources, the constraint method is determined based on the historical order completion status of the dispatch resources. The constraint types include: process node constraints and completion time constraints.

[0007] Furthermore, based on the comparison between the ratio of the number of leads in the current month to the historical average number of leads in the same period and a preset ratio, the order dispatch method for the leads is determined, including: Calculate the ratio of the number of leads in the current month to the historical average number of leads in the same month. If the ratio is less than or equal to the first preset ratio, then random order assignment will be determined. If the ratio is greater than the first preset ratio and less than or equal to the second preset ratio, then the basic completion time is determined based on the target's closeness characterization value, and an order is dispatched based on the determined basic completion time. If the ratio is greater than the second preset ratio, then it is determined that a scheduling resource dispatch order will be made.

[0008] Furthermore, the density characterization value of the target is calculated as follows: Determine the order frequency and number of orders for the target. Calculate the ratio of the order placement frequency to the preset order placement frequency to obtain the first density sub-parameter. The ratio of the number of orders placed to the preset number of orders is used to obtain the second degree of closeness sub-parameter. The first and second density sub-parameters are weighted and summed to obtain the density characterization value.

[0009] Furthermore, the determination of the basic completion time based on the target-related density representation value includes: The density rating is divided into multiple intervals, each corresponding to a preset base completion time. Among them, the basic completion time and the density characterization value are negatively correlated.

[0010] Furthermore, in step S3, determining the constraint method based on the historical order completion status of the scheduling resources includes: Calculate the ratio of the number of tasks completed on time to the total number of tasks. If the ratio is greater than or equal to the preset ratio, it is determined that only the completion time constraint will be applied; If the ratio is less than the preset ratio, then process node constraints and completion time constraints are determined.

[0011] Furthermore, the completion time constraint is as follows: a final completion deadline for the task is set, and this final completion deadline is treated as a hard constraint. If the time required for the remaining workload, predicted based on historical data, exceeds the actual remaining available time during task execution, an alert signal will be issued.

[0012] Furthermore, when issuing a reminder signal, the time difference between the predicted time required for the remaining workload based on historical data and the actual remaining available time is calculated, and an auxiliary scheduling mechanism is triggered based on the time difference analysis, including: If the time difference is greater than or equal to the preset time difference, the auxiliary scheduling mechanism is triggered. If the time difference is less than the preset time difference, it is determined that the auxiliary scheduling mechanism will not be triggered.

[0013] Furthermore, the process node constraints are as follows: The task is broken down into multiple logically related sub-steps, process nodes are identified, and completion criteria are set for each process node. If a node does not receive completion feedback within the specified time limit or the feedback result does not meet the standard, a node abnormality warning will be triggered.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: By screening idle resources and randomly allocating them, this invention ensures that all available resources can undertake tasks, avoiding some resources from being idle. When resources are scarce, scheduling resources to dispatch orders avoids overload and crashes. Timely activation of scheduled resources can prevent existing resources from being overloaded, thus avoiding the problems of task backlog and response timeouts. This solution ensures a basic user experience standard in both resource-sufficient and resource-scarce situations by using ratio-based hierarchical and close correlation with response time. When resources are abundant, all customers can receive a fast response, preventing ordinary customers from waiting too long due to idle resources. When resources are scarce, constraints ensure that all customers' tasks do not time out, preventing high-value customers from being ignored due to resource overload.

[0015] Furthermore, this invention enables rapid response for high-frequency, high-volume customers. Rapid response directly reduces their waiting anxiety, avoids dissatisfaction caused by service delays, and even reduces the risk of losing high-value customers, further consolidating their trust and dependence on the platform. Even for regular customers with lower engagement, the tiered response across four levels allows high-priority resources to focus solely on the urgent needs of high-frequency customers, while regular resources serve the regular needs of low-frequency customers, avoiding waste caused by resource misallocation. Clearly defined response time tiers help the customer service team establish priority ranking rules, eliminating the need to handle all customer requests simultaneously. Instead, orders from high-frequency customers are processed first, followed by regular customer requests. This orderly work rhythm reduces efficiency losses caused by chaotic multitasking for customer service, while also reducing workload and error rates caused by the backlog of urgent orders.

[0016] Furthermore, when the on-time completion rate of scheduled resources is greater than or equal to a preset ratio, this invention applies only completion time constraints. These resources typically possess stable execution capabilities, and reducing process node constraints can avoid over-management. When the on-time completion rate is less than the preset ratio, both process node constraints and completion time constraints are applied simultaneously. These resources may have insufficient progress control capabilities, and node constraints can help decompose tasks and promptly identify deviations, preventing delays in intermediate links from causing final timeouts. Process node constraints consume scheduling resources; for resources with high on-time completion rates, exempting node constraints can significantly reduce communication and time costs for both parties. Node constraints for resources with low on-time completion rates are not full-process monitoring, but rather targeted at key links affecting the final timeliness, effectively controlling progress while avoiding efficiency losses caused by over-monitoring.

[0017] Furthermore, the preset time difference of this invention is determined by the K percentile of historical warning events, rather than being subjectively set. This objectively reflects the urgency level that truly requires intervention, avoiding ineffective intervention for minor time differences and reducing additional system and manpower consumption. For extremely urgent situations where the time difference is greater than or equal to the preset value, auxiliary scheduling is immediately initiated. For minor warnings where the time difference is less than the preset value, only the current resources are reminded to adjust autonomously. This ensures that minor risks do not occupy auxiliary resources, nor do serious risks lead to ultimate failure due to response delays. Node warnings focus on whether intermediate links deviate from the plan, while time difference analysis focuses on whether there will be a timeout. The linkage between the two can cover the risks throughout the entire task execution cycle. When a single resource encounters an execution problem, the system can quickly transfer the task to other resources to avoid the task from timed out. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the workflow of the self-learning intelligent order dispatching method based on cloud-based big data of the present invention. Figure 2This is a flowchart illustrating the process of determining the dispatch method for a lead in the cloud-based big data-driven self-learning intelligent dispatch method of the present invention. Figure 3 This is a flowchart illustrating the process of determining the constraint method based on the historical order completion status of scheduling resources in the cloud-based big data self-learning intelligent order dispatching method described in this invention. Detailed Implementation

[0019] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] It should be noted that the data in this embodiment are all derived from a comprehensive analysis and evaluation of historical data from the six months prior to this determination and the corresponding historical determination results by the system described in this invention. Those skilled in the art will understand that the system described in this invention can determine the above-mentioned parameters for a single item by selecting the value with the highest proportion based on the data distribution as the preset standard parameter, using weighted summation to obtain the value as the preset standard parameter, substituting each historical data point into a specific formula and using the value obtained by that formula as the preset standard parameter, or other selection methods, as long as the system described in this invention can clearly define different specific situations in the single-item determination process through the obtained values.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0023] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0024] Please see Figures 1-3 As shown, Figure 1This is a flowchart illustrating the workflow of the self-learning intelligent order dispatching method based on cloud-based big data of the present invention. Figure 2 This is a flowchart illustrating the process of determining the dispatch method for a lead in the cloud-based big data-driven self-learning intelligent dispatch method of the present invention. Figure 3 This is a flowchart illustrating the process of determining the constraint method based on the historical order completion status of scheduling resources in the cloud-based big data self-learning intelligent order dispatching method described in this invention.

[0025] The self-learning intelligent order dispatching method based on cloud-based big data provided in this embodiment includes: Step S1: Obtain historical clue data and the number of clues for the current month; Step S2, based on the comparison between the ratio of the number of leads in the current month to the historical average number of leads in the same period and a preset ratio, determines the order dispatch method for the leads, including: Random order assignment The baseline completion time is determined based on the target's degree of proximity, and orders are dispatched based on this baseline completion time. And dispatching resources and assigning tasks.

[0026] Step S3: When determining to dispatch resources, the constraint method is determined based on the historical order completion status of the dispatch resources. The constraint types include: process node constraints and completion time constraints.

[0027] In this embodiment of the invention, the scheduling resource refers to the smallest business unit capable of independently completing the task of following up on leads. Specifically, in this embodiment, random task assignment first filters out the scheduling resources that are currently idle, that is, resources that are not executing tasks and can immediately accept new tasks, forming a resource pool to be allocated; then, through a preset random allocation algorithm, such as random sorting based on resource ID, random number extraction, etc., the single or multiple clues newly added in the current month are indiscriminately allocated to any available resources in the resource pool.

[0028] Specifically, in step S2, the order dispatch method for leads is determined based on the comparison between the ratio of the number of leads in the current month to the average number of leads in the same period in history and a preset ratio, including: Calculate the ratio of the number of leads in the current month to the historical average number of leads in the same month. If the ratio is less than or equal to the first preset ratio, then random order assignment will be determined. If the ratio is greater than the first preset ratio and less than or equal to the second preset ratio, then the basic completion time is determined based on the target's closeness characterization value, and an order is dispatched based on the determined basic completion time. If the ratio is greater than the second preset ratio, then it is determined that a scheduling resource dispatch order will be made.

[0029] Specifically, in this embodiment, the number of leads in the current month is the total number of leads generated within the current month (or period); the historical average number of leads in the same month is the average number of new leads in the same month (or similar business period) over the past several years (e.g., 3-5 years). The first preset ratio and the second preset ratio can be determined in the following way: collect the lead number data for each month over the past several years (e.g., 3-5 years); for each month in history, calculate the ratio of its monthly lead volume to the historical average number of leads in the same month to obtain a historical ratio sequence; perform distribution analysis on all calculated historical ratios; the first preset ratio can usually be set at a lower quantile of the historical ratio distribution, for example, the 30th quantile of the historical ratio, which means that when the business volume is lower than this proportion of the normal level of the same period in history, the system resources are very abundant; the second preset ratio can usually be set at a higher quantile of the historical ratio distribution, for example, the 75th quantile of the historical ratio, which means that when the business volume exceeds this proportion of the normal level of the same period in history, the system resources are very tight and the highest priority scheduling mode needs to be activated.

[0030] This invention employs a random allocation mechanism to select idle resources, ensuring that all available resources can handle tasks and preventing some resources from becoming idle. When resources are scarce, resource scheduling and task assignment prevent overload and crashes. Timely resource scheduling avoids task backlog and response timeouts caused by overloading existing resources. This solution uses ratio-based stratification and close correlation with response time to guarantee a basic user experience regardless of resource availability. When resources are abundant, all customers receive rapid responses, preventing ordinary customers from waiting excessively due to idle resources. When resources are scarce, constraints ensure that all customer tasks do not time out, preventing high-value customers from being overlooked due to resource overload.

[0031] Specifically, the density representation value of the target is calculated as follows: Determine the order frequency and number of orders for the target. Calculate the ratio of the order placement frequency to the preset order placement frequency to obtain the first density sub-parameter. The ratio of the number of orders placed to the preset number of orders is used to obtain the second degree of closeness sub-parameter. The first and second density sub-parameters are weighted and summed to obtain the density characterization value.

[0032] Specifically, the determination of the basic completion time based on the target's density representation value includes: The density rating is divided into multiple intervals, each corresponding to a preset base completion time. Among them, the basic completion time and the density characterization value are negatively correlated.

[0033] The preset order frequency and preset order count described in this embodiment of the invention can be determined by the following methods: Select complete business cycle data from the past 1-3 years; for example, if the business is annually cyclical, at least 3 complete years should be covered; stratify and statistically analyze by customer type, as the ordering behavior of different groups varies greatly, requiring stratified preset values ​​rather than a uniform benchmark; exclude extreme customer data, typically using the 3σ principle or quantile truncation, retaining the behavioral data of normal active customers; calculate the statistical characteristic values ​​of order frequency and order count for the selected valid samples; if the business objective is to prioritize serving highly engaged customers, the preset value can be set to the 70th percentile (i.e., 70% of customers are below this value, and customers above this value are highly engaged customers); if the objective is to cover the majority of customers, the 50th percentile can be used; however, the above values ​​are not limited to these, and those skilled in the art can adjust them according to the actual situation.

[0034] In this embodiment of the invention, the weights of the first and second closeness sub-parameters can be determined by the following method: Define a target variable and find an indicator that can measure the subsequent conversion effect, such as the repurchase amount in the next 3 months; collect a batch of historical customer data, each customer sample including: Feature 1 (X1): order frequency (e.g., average number of orders per month); Feature 2 (X2): total number of historical orders; Target variable (Y): repurchase amount in the next 3 months; Use models such as linear regression, with Y = β0 + β1 * X1 + Fit β2*X2; analyze the obtained standardized regression coefficients, assuming the standardized coefficients are: β1 (frequency) = 0.6, β2 (number of occurrences) = 0.3; calculate the weights: Wf (weight coefficient of the first tightness sub-parameter) = |β1| / (|β1|+|β2|) = 0.6 / (0.6+0.3) = 0.67; Wc (weight coefficient of the second tightness sub-parameter) = |β2| / (|β1|+|β2|) = 0.3 / (0.6+0.3) = 0.33, but the above values ​​are not limited to these, and those skilled in the art can adjust them according to the actual situation.

[0035] In this embodiment of the invention, the distribution of S (integrity value) values ​​based on historical customer data is used to divide the customer base, ensuring that the number of customers in each interval meets business expectations. For example, customers are divided into four categories based on their intimacy: core customers, important customers, ordinary customers, and marginal customers, with proportions of 10%, 20%, 50%, and 20%, respectively. The 90th, 70th, and 20th quantiles of historical customer S values ​​are calculated: Interval 1 (marginal customers): S < 20th quantile; Interval 2 (ordinary customers): 20th quantile <= S < 70th quantile; Interval 3 (important customers): 70th quantile <= S < 90th quantile; Interval 4 (core customers): S >= 90th quantile. h-quantiles; assign a preset base completion time T_base to each divided interval; T_base should be negatively correlated with the density representation value S, that is, the higher S (the closer the customers), the shorter T_base, which can be set directly based on historical experience; for example: S∈[1.5,+∞) corresponds to >T_base=4 hours; S∈[1.0,1.5) corresponds to >T_base=8 hours; S∈[0.5,1.0) corresponds to >T_base=24 hours; S∈[0,0.5) corresponds to >T_base=48 hours; analyze historical data to find the actual average completion time of customer orders in different S intervals, and use this as T_base.

[0036] This invention enables rapid response for high-frequency, high-volume customers. Rapid response directly reduces their waiting anxiety, avoids dissatisfaction caused by service delays, and even reduces the risk of losing high-value customers, further solidifying their trust and reliance on the platform. Even for less frequent, regular customers, a tiered response system with four levels allows high-priority resources to focus solely on the urgent needs of high-frequency customers, while regular resources serve the routine needs of less frequent, low-volume customers, avoiding waste caused by resource misallocation. Clearly defined response time tiers help the customer service team establish priority ranking rules, eliminating the need to handle all customer requests simultaneously. Instead, orders from high-frequency customers are processed first, followed by regular customer requests. This orderly work rhythm reduces efficiency losses caused by chaotic multitasking and lowers workload and error rates due to a backlog of urgent orders.

[0037] Specifically, in step S3, determining the constraint method based on the historical order completion status of the scheduling resources includes: Calculate the ratio of the number of tasks completed on time to the total number of tasks. If the ratio is greater than or equal to the preset ratio, it is determined that only the completion time constraint will be applied; If the ratio is less than the preset ratio, then process node constraints and completion time constraints are determined.

[0038] Specifically, the completion time constraint is as follows: set a final completion deadline for the task and use the final completion deadline as a hard constraint condition, and clearly inform the scheduling resources when dispatching the task that the task must be completed before the deadline; If the time required for the remaining workload, predicted based on historical data, exceeds the actual remaining available time during task execution, an alert signal will be issued.

[0039] The preset ratio described in this embodiment of the invention can be determined by the following method: performing distribution analysis on historical on-time completion rate data (by resource / team dimension) and calculating quantiles; setting the preset ratio to the 70%-80% quantile of historical data (indicating that when the on-time completion rate of a resource is higher than this value, its stability is sufficient and no additional node constraints are required); if the business has extremely high timeliness requirements (such as urgent orders), it can be increased to the 85%-90% quantile (only a very small number of highly stable resources are allowed to be exempt from node constraints), but the above values ​​are not limited to these, and those skilled in the art can adjust them according to the actual situation.

[0040] This invention applies only completion time constraints when the on-time completion rate of scheduled resources is greater than or equal to a preset ratio. These resources typically have stable execution capabilities, and reducing process node constraints can avoid over-management. When the on-time completion rate is less than the preset ratio, both process node constraints and completion time constraints are applied simultaneously. These resources may have insufficient progress control capabilities, and node constraints can help decompose tasks and detect deviations in a timely manner, avoiding delays caused by intermediate links that lead to final timeouts. Process node constraints consume scheduling resources. For resources with high on-time completion rates, exempting node constraints can significantly reduce communication and time costs for both parties. Node constraints for resources with low on-time completion rates are not full-process monitoring, but rather targeted at key links that affect the final timeliness, which can effectively control progress while avoiding efficiency losses caused by over-monitoring.

[0041] Specifically, when issuing a reminder signal, the time difference between the predicted time required for the remaining workload based on historical data and the actual remaining available time is calculated, and an auxiliary scheduling mechanism is triggered based on the time difference analysis, including: If the time difference is greater than or equal to the preset time difference, the auxiliary scheduling mechanism is triggered. If the time difference is less than the preset time difference, it is determined that the auxiliary scheduling mechanism will not be triggered.

[0042] Specifically, the process node constraints are as follows: The task is broken down into multiple logically related sub-steps, process nodes are identified, and completion criteria are set for each process node. If a node does not receive completion feedback within the specified time limit or the feedback result does not meet the standard, a node abnormality warning will be triggered.

[0043] The preset time difference in this embodiment of the invention can be determined by the following method: Collect all historical event data that trigger task reminder signals, calculate the difference (ΔT_historical) between the predicted remaining workload and the actual remaining available time for each event, forming a historical time difference dataset; sort the values ​​in this dataset in ascending order, and take the Kth percentile (K can be set to 5, 10, or 15 according to the business's risk tolerance) as the preset time difference; for example, when K=10, it means that historically only 10% of warning events have an urgency level (negative time difference) higher than this threshold. Once the current time difference reaches or exceeds this threshold, it is judged as extremely urgent. In urgent situations, an auxiliary scheduling mechanism must be triggered immediately. This mechanism includes placing tasks that are about to expire into an internal task-grabbing pool, adding an "urgent" tag and additional incentives to encourage other available or highly skilled scheduling resources to actively claim the task. The system automatically reclaims the task from the current resource and, based on the real-time load and skill matching of other resources, forcibly assigns it to a new resource most likely to complete it on time. Reserved backup resources (such as trainees, management support personnel, and cross-team support personnel) are activated to temporarily take over such urgent tasks. The task's priority is automatically increased, and the current resource's direct supervisor is notified. The supervisor can personally intervene or reassign the task.

[0044] The preset time difference in this invention is determined by the K percentile of historical warning events, rather than being subjectively set. This objectively reflects the urgency level that truly requires intervention, avoiding ineffective intervention for minor time differences and reducing additional system and manpower consumption. For extremely urgent situations where the time difference is greater than or equal to the preset value, auxiliary scheduling is immediately initiated. For minor warnings where the time difference is less than the preset value, only the current resources are prompted to adjust autonomously. This ensures that minor risks do not occupy auxiliary resources, nor do serious risks lead to ultimate failure due to response delays. Node warnings focus on whether intermediate links deviate from the plan, while time difference analysis focuses on whether there will be a timeout. The linkage between the two can cover the risks throughout the entire task execution cycle. When a single resource encounters an execution problem, the system can quickly transfer the task to other resources to avoid task timeout.

[0045] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0046] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A self-learning intelligent order dispatching method based on cloud-based big data, characterized in that, include: Step S1: Obtain historical clue data and the number of clues for the current month; Step S2, based on the comparison between the ratio of the number of leads in the current month to the historical average number of leads in the same period and a preset ratio, determines the order dispatch method for the leads, including: Random order assignment The baseline completion time is determined based on the target's degree of proximity, and orders are dispatched based on this baseline completion time. And dispatching resources and assigning orders; Step S3: When determining to dispatch resources, the constraint method is determined based on the historical order completion status of the dispatch resources. The constraint types include: process node constraints and completion time constraints.

2. The self-learning intelligent order dispatching method based on cloud-based big data as described in claim 1, characterized in that, In step S2, the order dispatch method for each lead is determined based on a comparison between the ratio of the number of leads in the current month to the historical average number of leads in the same period and a preset ratio, including: Calculate the ratio of the number of leads in the current month to the historical average number of leads in the same month. If the ratio is less than or equal to the first preset ratio, then random order assignment will be determined. If the ratio is greater than the first preset ratio and less than or equal to the second preset ratio, then the basic completion time is determined based on the target's closeness characterization value, and an order is dispatched based on the determined basic completion time. If the ratio is greater than the second preset ratio, then it is determined that a scheduling resource dispatch order will be made.

3. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 2, characterized in that, The method for calculating the density representation value of the target is as follows: Determine the order frequency and number of orders for the target. Calculate the ratio of the order placement frequency to the preset order placement frequency to obtain the first density sub-parameter. The ratio of the number of orders placed to the preset number of orders is used to obtain the second degree of closeness sub-parameter. The first and second density sub-parameters are weighted and summed to obtain the density characterization value.

4. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 3, characterized in that, The determination of the basic completion time based on the target's closeness representation value includes: The density rating is divided into multiple intervals, each corresponding to a preset base completion time. Among them, the basic completion time and the density characterization value are negatively correlated.

5. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 2, characterized in that, In step S3, the constraint method is determined based on the historical order completion status of the scheduling resources, including: Calculate the ratio of the number of tasks completed on time to the total number of tasks. If the ratio is greater than or equal to the preset ratio, it is determined that only the completion time constraint will be applied; If the ratio is less than the preset ratio, then process node constraints and completion time constraints are determined.

6. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 5, characterized in that, The completion time constraint is as follows: a final completion deadline for the task is set, and this final completion deadline is treated as a hard constraint. If the time required for the remaining workload, predicted based on historical data, exceeds the actual remaining available time during task execution, an alert signal will be issued.

7. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 6, characterized in that, When issuing a reminder signal, the time difference between the predicted time required for the remaining workload based on historical data and the actual remaining available time is calculated. Based on this time difference, an auxiliary scheduling mechanism is analyzed to determine whether to trigger, including: If the time difference is greater than or equal to the preset time difference, the auxiliary scheduling mechanism is triggered. If the time difference is less than the preset time difference, it is determined that the auxiliary scheduling mechanism will not be triggered.

8. The self-learning intelligent order dispatching method based on cloud-based big data according to claim 5, characterized in that, The process node constraints are as follows: The task is broken down into multiple logically related sub-steps, process nodes are identified, and completion criteria are set for each process node. If a node does not receive completion feedback within the specified time limit or the feedback result does not meet the standard, a node abnormality warning will be triggered.

Images