Method, device and equipment for completely adjusting robot process automation RPA and medium

Abstract

The invention discloses a complete dispatching method, device and equipment for robot process automation RPA and a medium, and relates to the field of financial science and technology, and the method comprises the steps: responding to a complete dispatching request, extracting a business type, a customer type and reservation handling time of a target task, and forming task data; obtaining a target time parameter based on the task data; extracting all uncompleted historical tasks from the RPA cluster and obtaining historical time parameters; using a priority scheduling algorithm to detect whether the RPA cluster satisfies a scheduling condition according to the target time parameter and the historical time parameter; and if yes, associating the target task to a to-be-scheduled queue of the target server according to the scheduling optimization model, and executing a complete scheduling operation. According to the embodiment of the invention, the financial service logic and the RPA scheduling technology are deeply fused, so that the problems of low efficiency, non-uniform resource allocation, insufficient time response and the like in the traditional complete scheduling process are solved, and a standardized and intelligent automatic solution is provided for financial institutions.

Description

Technical Field

[0001] The present invention is applicable to the field of financial technology, and in particular relates to a due diligence method, device, equipment and medium for Robotic Process Automation (RPA). Background Art

[0002] Amidst heightened financial regulation, banks are now subject to strict due diligence procedures for corporate banking operations, such as account opening, credit approval, and anti-money laundering due diligence. These procedures involve cross-system data querying, risk indicator calculation, and evidence retention. However, traditional manual operations present pain points, including cumbersome processes leading to long customer wait times, high manual error rates, and uneven resource allocation during peak hours.

[0003] Although Robotic Process Automation (RPA) technology can achieve process automation, it does not fully consider the special time attributes and special technical requirements of financial business in complex scenarios with multiple financial outlets, multiple business types, and multiple time constraints. As a result, it is impossible to dynamically adapt to the rigid requirements of customer appointment times, and it is difficult to achieve efficient utilization of RPA cluster resources in multi-task concurrent scenarios. There is an urgent need to propose an intelligent scheduling method that integrates time evaluation and resource optimization. Summary of the Invention

[0004] Based on this, the present invention provides a due diligence method, device, equipment and medium for robotic process automation (RPA) to solve the problems of low resource utilization, long customer waiting time, inability to prioritize VIP customers and high server energy consumption in the due diligence process of existing banks' public account business.

[0005] In a first aspect, an embodiment of the present invention provides a due diligence method for Robotic Process Automation (RPA), the method comprising:

[0006] In response to the due diligence request for the target task from the business front-end system, the target business type, target customer type, and target appointment processing time of the target task are extracted to form task data;

[0007] Calculating the deadline time point and estimated task duration of the target task based on the task data to obtain target time parameters of the target task;

[0008] Extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task;

[0009] Use the priority scheduling algorithm to integrate and sort the target time parameters and historical time parameters. Based on the integrated sorting results, check whether the RPA cluster meets the scheduling conditions for the target task.

[0010] If so, the target task is associated with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and the due diligence operation is performed on the target task when the system time reaches the task start time of the target task; wherein, the task start time is calculated by the deadline and the estimated task duration.

[0011] In a second aspect, an embodiment of the present invention further provides a due diligence device for Robotic Process Automation (RPA), the device comprising:

[0012] Due diligence request response module, used to respond to the due diligence request for the target task from the business front-end system, extract the target business type, target customer type and target appointment processing time of the target task to form task data;

[0013] A target time parameter acquisition module is used to calculate the deadline time point of the target task and the estimated task duration based on the task data to obtain the target time parameters of the target task;

[0014] A historical time parameter acquisition module is used to extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task;

[0015] The scheduling condition determination module is used to integrate and sort the target time parameters and historical time parameters using a priority scheduling algorithm. Based on the integrated sorting results, it detects whether the RPA cluster meets the scheduling conditions for the target task.

[0016] The due diligence task execution module is used to, if so, associate the target task with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and perform due diligence on the target task when the system time reaches the task start time of the target task; wherein the task start time is calculated by the deadline and the estimated task duration.

[0017] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising:

[0018] at least one processor; and

[0019] a memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a due diligence method for robotic process automation (RPA) as described in any embodiment of the present invention.

[0021] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a due diligence method for robotic process automation (RPA) as described in any embodiment of the present invention when executed.

[0022] The embodiment of the present invention realizes efficient scheduling of financial due diligence tasks through intelligent processes. First, it converts unstructured business requirements into computable task data; converts the time sensitivity of financial business into clear time parameters to ensure that tasks are executed within the minimum time window that meets the timeliness requirements, avoiding idle resources and task timeouts; the multi-objective optimization model combines server resource utilization and task waiting time to achieve intelligent matching of tasks and servers, improve the overall processing efficiency and resource utilization of the cluster, and is particularly suitable for financial scenarios with multiple branches and multiple businesses. By deeply integrating financial business logic with RPA scheduling technology, it solves the problems of low efficiency, uneven resource allocation, and insufficient timeliness response in the traditional due diligence process, and provides financial institutions with standardized and intelligent automation solutions.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 This is a flowchart of a due diligence method for Robotic Process Automation (RPA) provided according to the first embodiment of the present invention;

[0026] Figure 2 This is a flowchart of another due diligence method for Robotic Process Automation (RPA) provided according to the second embodiment of the present invention;

[0027] Figure 3 This is a structural diagram of a due diligence device for Robotic Process Automation (RPA) according to a third embodiment of the present invention;

[0028] Figure 4 This is a structural diagram of an electronic device for implementing a due diligence method for Robotic Process Automation (RPA) according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Example 1

[0032] Figure 1 This is a flowchart of a due diligence method for Robotic Process Automation (RPA) provided in the first embodiment of the present invention. This embodiment is applicable to the situation where due diligence is conducted on the bank's public business based on RPA. The method can be executed by a due diligence device for Robotic Process Automation (RPA), which can be implemented in the form of hardware and / or software and can be configured in the bank's core business processing system. Figure 1 As shown, the method includes:

[0033] S110 , in response to the due diligence request for the target task from the business front-end system, extracting the target business type, target customer type, and target appointment processing time of the target task to constitute task data.

[0034] A due diligence request is the core instruction that triggers due diligence in financial business processes. Essentially, it serves as a digital vehicle for translating business needs into automated tasks. In banking scenarios, when a customer initiates a business transaction requiring risk review, such as opening a public account or applying for a loan, the front-end business system (e.g., teller terminals or corporate online banking) automatically or manually generates a due diligence request based on regulatory requirements and internal processes, driving the RPA system to perform subsequent investigation operations. A due diligence request contains multiple key elements: at the basic business level, it includes the business type (account opening, loan approval, etc.) and customer number, which are used to accurately identify the business scenario and customer identity; at the time constraint level, it includes the appointment processing time. The business type identifies the business scenario requiring due diligence. For example, in the banking business mapping, account opening = 001 and account change = 002. The target customer type is essentially a customer classification label that determines the maximum waiting time threshold. For example, if the customer type is VIP, the maximum waiting time threshold is 30 minutes; if the customer type is ordinary, the maximum waiting time threshold is 45 minutes; and if the customer type is high-risk, the maximum waiting time threshold is 60 minutes. The appointment time is the time when the customer arrives at the bank to handle business (such as 2025-05-15 10:30:00).

[0035] S120 , calculating the deadline time point and estimated task duration of the target task based on the task data to obtain target time parameters of the target task.

[0036] In financial due diligence scenarios, different business types (such as account opening and loan approval) have different time requirements. For example, regulations require loan approval to be completed within 3 working days, while VIP customers who make account opening appointments may require due diligence to be completed 1 hour in advance. Mathematical models can be used to convert the above business rules into specific time parameters.

[0037] The calculation of the deadline time point requires a comprehensive consideration of the appointment processing time, business type, and customer attributes in the task data. For example, if the task data contains "the target appointment processing time is 10:30 on May 15, 2025", the deadline is usually set 60 minutes in advance according to the business rules (that is, due diligence is completed before 9:30); if the task data is marked as "VIP customer type", it may be further shortened to 30 minutes in advance. The calculation of the estimated task duration will first extract the average execution time of similar businesses from the historical database (such as the historical average time of "opening a public account" is 90 minutes), and then correct it according to the customer type. The output of the target time parameter integrates the deadline time point and the estimated task duration into structured data, which is directly used for subsequent scheduling condition detection. For example, the deadline time point calculated for a task is "2025-05-15

[0038] 10:30", and the estimated task duration is "105 minutes". Based on this, we can determine whether the current time meets the scheduling condition of "current time + 105 minutes is earlier than 09:30". At the same time, this duration is also used as one of the basis for server load evaluation.

[0039] S130: Extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and the remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task.

[0040] The RPA cluster's task management database batch filters out all unfinished historical tasks, including those in progress, queued, and waiting. This ensures that the collected data truly reflects the cluster's current task backlog and resource usage. For example, in a due diligence scenario for corporate banking, there may be dozens of pending due diligence tasks across different branches and business types.

[0041] For each unfinished historical task, two core time parameters are extracted. On the one hand, the historical deadline time point set when the task was created is obtained. This time point reflects the original time constraint requirements of the task, which may be determined based on factors such as the customer's appointment time and the business processing deadline; on the other hand, the remaining task duration is calculated, that is, the estimated execution time required for the unfinished part of the task. For tasks that are being executed, the system calculates the remaining duration by real-time monitoring of the execution time and combining it with the total estimated duration of the task; for tasks waiting in line, the originally set estimated task duration is directly used. For example, the historical deadline time point of a loan due diligence task is 15:00 on the same day, and it has been executed for 30 minutes. The original estimated total duration is 90 minutes, and the remaining task duration is 60 minutes.

[0042] The historical deadlines and remaining durations of each unfinished historical task are integrated into historical time parameters to form a structured dataset. By summing the remaining durations of all unfinished historical tasks, it is possible to determine whether the cluster currently has sufficient idle resources and time windows to accommodate new target tasks. This provides critical data foundation for subsequent task consolidation and scheduling condition testing using priority scheduling algorithms, preventing task timeouts or resource imbalances caused by blind scheduling.

[0043] S140: Use a priority scheduling algorithm to integrate and sort the target time parameters and historical time parameters, and detect whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting result.

[0044] The target task's deadline, estimated duration, remaining duration, and deadline of unfinished historical tasks in the cluster are processed uniformly. An algorithm such as Earliest Deadline First (EDF) is used to calculate a priority score based on task urgency and resource consumption. After the integration and sorting is completed, the RPA cluster's scheduling conditions for the target task are tested from two dimensions: time and resources. In the time dimension, the RPA cluster verifies whether the target task can be completed before the deadline, and evaluates whether inserting the target task will cause historical tasks to time out. In the resource dimension, the RPA cluster checks whether there are servers with resource utilization rates below the threshold, and whether the total queue duration of the target server plus the duration of the target task exceeds its available duration. If any of the conditions are not met, subsequent manual intervention or task reallocation processes are triggered.

[0045] S150. If so, the target task is associated with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and when the system time reaches the task start time of the target task, the due diligence operation is performed on the target task; wherein, the task start time is calculated by the deadline and the estimated task duration.

[0046] Once the RPA cluster is determined to meet scheduling requirements, the system first uses a scheduling optimization model to comprehensively consider factors such as server load, processing capacity, and task type compatibility to select the optimal target server from the cluster. The target tasks are then inserted into the target server's scheduling queue in ascending order of deadline, and the trigger timing is calculated based on the formula "task start time = deadline - estimated task duration." For example, if a task has a deadline of 10:30 and an estimated duration of 60 minutes, the system automatically triggers the RPA robot to perform due diligence at 9:30. This reverse calculation mechanism ensures that tasks do not prematurely consume resources while accurately meeting timelines. During the execution phase, the server invokes the RPA robot in queue order, completing automated operations based on the data source instructions in the due diligence request, achieving closed-loop management from scheduling decision-making to business execution.

[0047] The embodiment of the present invention realizes efficient scheduling of financial due diligence tasks through intelligent processes. First, it converts unstructured business requirements into computable task data; converts the time sensitivity of financial business into clear time parameters to ensure that tasks are executed within the minimum time window that meets the timeliness requirements, avoiding idle resources and task timeouts; the multi-objective optimization model combines server resource utilization and task waiting time to achieve intelligent matching of tasks and servers, improve the overall processing efficiency and resource utilization of the cluster, and is particularly suitable for financial scenarios with multiple branches and multiple businesses. By deeply integrating financial business logic with RPA scheduling technology, it solves the problems of low efficiency, uneven resource allocation, and insufficient timeliness response in the traditional due diligence process, and provides financial institutions with standardized and intelligent automation solutions.

[0048] Example 2

[0049] Figure 2 This is a flowchart of another due diligence method for Robotic Process Automation (RPA) provided in Example 2 of the present invention. This example is refined based on Example 1. Specifically, Figure 2 As shown, the method includes:

[0050] S210 , in response to the due diligence request for the target task from the business front-end system, extract the target business type, target customer type, and target appointment processing time of the target task to constitute task data.

[0051] S220: Based on the task data, query the preset rule library for the longest waiting time corresponding to the target customer type, and calculate the deadline time point of the target task by the target appointment processing time and the longest waiting time.

[0052] According to the customer type in the target task data (such as VIP customers, ordinary enterprises, high-risk customers, etc.), the corresponding maximum waiting time is retrieved from the preset rule library. The rule library adopts a key-value pair structure, for example: VIP customers: 15 minutes; "ordinary enterprises": 30 minutes; "high-risk customers": 45 minutes. Taking the target appointment processing time as the benchmark, the deadline is calculated by the appointment time and the maximum waiting time. For example, if a customer makes an appointment to handle business at 10:30 on May 15, 2025, and the maximum waiting time corresponding to his customer type is 30 minutes, the deadline is 9:30, ensuring that RPA completes due diligence preparations before the customer arrives at the store. This reverse calculation mechanism avoids the time redundancy that may be caused by the traditional "start time + estimated duration" model, and compresses resource usage to the minimum window.

[0053] S230. Extract a historical processing duration sample set corresponding to the target business type from a historical task database based on the task data, and input the sample set into a pre-trained duration prediction model to calculate the estimated task duration of the target task, thereby obtaining the target time parameter of the target task.

[0054] Retrieve historical tasks from the historical task database that fully match the target business type and extract their actual processing times to form a sample set. For example, extract the average processing time of 200 "loan due diligence" tasks from the past three months, which is 75 minutes. Use a gradient boosting tree or deep learning model to train the model using the historical sample set. For example, consider a loan due diligence task involving three query systems, with a moderate cluster load, and a model-estimated task duration of 30 minutes.

[0055] The rule base in this embodiment ensures that service duration requirements for different customer types are rigidly enforced, avoiding timeliness deviations caused by manual configuration and improving the ability to deliver on financial service commitments. By leveraging historical data and predictive models to adapt to business process changes in real time, this reduces estimation errors compared to fixed duration configurations and provides a more reliable time benchmark for scheduling decisions.

[0056] S240: Extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and the remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task.

[0057] Optionally, extracting all unfinished historical tasks from the RPA cluster and obtaining the historical deadline time points and remaining task durations of each unfinished historical task may include:

[0058] Get all unfinished tasks from the RPA cluster management system;

[0059] Among all unfinished tasks, the first unfinished historical task that has been assigned to a server and is located in the queue to be scheduled is divided into the queue task set, and the second unfinished historical task that has not been assigned to a server is divided into the free task set;

[0060] For the first unfinished historical task in the queue task set, read the historical deadline time point and the remaining task duration of each first unfinished historical task from the local state of the assigned server;

[0061] For the second unfinished historical tasks in the free task set, obtain the historical deadline time point set when the task was created, and based on the task type of each second unfinished task, query the average duration of similar tasks from the historical task database as the remaining task duration of each second unfinished task.

[0062] First, all unfinished tasks (including those in execution and queued) are extracted from the RPA cluster management system's global task pool and divided into two categories based on their allocation status: the queued task set includes tasks assigned to a specific server's pending scheduling queue (the first category of unfinished tasks), whose execution depends on the target server's resource scheduling; the free task set includes pending tasks that have not yet been assigned a server (the second category of unfinished tasks), which are in the cluster's global task pool awaiting allocation. For tasks in the queued task set, real-time time parameters can be directly read from the target server's local state database. For unassigned tasks in the free task set, a hybrid "rule + historical data" estimation method can be used. For example, the historical deadline can be directly used as the deadline set when the task was created. The remaining task duration can be retrieved from the historical task database based on the average execution time of similar tasks.

[0063] This embodiment of the present invention utilizes hierarchical management to avoid cluttering task data, enabling the system to analyze local server load and global resource pressure in a targeted manner, providing clear data support for scheduling decisions. By combining real-time data with historical estimates, the time parameters of all unfinished tasks are accessible and calculable, avoiding scheduling decision biases caused by missing data and enhancing system robustness.

[0064] S250: Use the priority scheduling algorithm to integrate and sort the target time parameters and the historical time parameters, and detect whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting result.

[0065] Furthermore, the target time parameters and historical time parameters are integrated and sorted using a priority scheduling algorithm. Based on the integrated sorting results, the RPA cluster is checked to see if it meets the scheduling conditions for the target task. This may include:

[0066] Merging the target time parameter and each historical time parameter into a time parameter set, and arranging the time parameter set in ascending order according to the end time point;

[0067] According to the sorted time parameter set, the target task and each unfinished historical task form an ordered task queue, and locate the target task in the ordered task queue;

[0068] Starting from the location of the target task, obtain the current tasks in order from front to back;

[0069] Calculate the total processing time of all tasks between the first task in the ordered task queue and the current task, and verify whether the total processing time is less than the time difference between the deadline of the current task and the current system time;

[0070] If so, the process returns to executing the operation starting from the location of the target task and sequentially obtaining the current task in order from the front to the back. When the current task is determined to be the last task in the ordered task queue, the scheduling condition for the target task is determined to be met.

[0071] If not, it is determined that the scheduling conditions for the target task are not met; wherein the total processing time is the sum of the estimated task time of the target task and / or the remaining task time of the unfinished historical tasks.

[0072] The target task's deadline, estimated duration, and the remaining duration and deadline of historical tasks are combined into a unified set of time parameters and sorted in ascending order by deadline. For example, if a target task has a deadline of 15:00 and an estimated duration of 60 minutes, a historical task A has a deadline of 14:30 and a remaining duration of 30 minutes, and a historical task B has a deadline of 16:00 and a remaining duration of 90 minutes, then the order of the sorted set is: Task A → Target Task → Task B. This sorting follows the Earliest Deadline First (EDF) principle, ensuring that time-sensitive tasks are processed first, which meets the requirements of rigid appointment time guarantees in financial services (such as completing due diligence before a customer arrives at the store). The sorted time parameters are mapped to a specific task queue, where each task contains a two-tuple of "deadline time point - processing duration". By traversing the queue to locate the index position of the target task, the scheduling algorithm is ensured to only focus on the impact range after the target task is inserted, avoiding resource waste caused by global calculations.

[0073] Starting from the target task's position, the current tasks are retrieved in queue order. The total processing time from the first task in the queue to the current task is calculated and compared with the available time window of "current task deadline - system time". For example, the initial state is: system time 13:00, the target task is second in the queue, the first task is Task A (deadline 14:30, 30 minutes remaining), and the target task is estimated to take 60 minutes. First verification: Total processing time = 30 + 60 = 90 minutes, available time = 14:30 - 13:00 = 90 minutes → 90 ≤ 90, which meets the condition. Verification continues with the next task (Task B, deadline 16:00, 90 minutes remaining). Second verification: Total processing time = 30 + 60 + 90 = 180 minutes, available time = 16:00 - 13:00 = 180 minutes → 180 ≤ 180, which meets the condition until the end of the queue, confirming that the entire task is schedulable.

[0074] That is, when the total processing time of all tasks from the target task to the end of the queue does not exceed the difference between the corresponding deadline and the system time, it means that inserting the target task will not cause any task to time out, and the RPA cluster resources are sufficient to support it. If the total processing time of a task exceeds the available time window, a scheduling conflict is determined and the manual intervention mechanism is triggered. The embodiment of the present invention combines deadline sorting with dynamic window verification to ensure that the insertion of the target task will not cause any task in the queue to time out, and achieves accurate matching of cluster resources and task timeliness. Through progressive verification starting from the target task position, the full-link impact range after task insertion is covered, avoiding global scheduling failures caused by local resource conflicts.

[0075] Optionally, after checking whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting results, the following steps may also be performed:

[0076] When the RPA cluster is detected as not meeting the conditions for scheduling the target task, a manual intervention request is generated, including the target business type, target customer type, target appointment processing time, target deadline, and target estimated task duration.

[0077] Push manual intervention requests to the preset management terminal, or trigger the work order system to create manual scheduling tasks based on manual intervention requests.

[0078] If the RPA cluster determines it cannot meet the scheduling conditions for the target task, a manual intervention request is automatically generated containing five key pieces of information. These include business attribute information (target business type and customer type), which allows for quick manual identification of the business scenario and risk level. Time constraints (target appointment time, deadline, and estimated task duration) provide a visual representation of the task's time urgency. For example, if a task is scheduled for 2025-05-15 10:00, with a deadline of 9:30 and an estimated duration of 90 minutes, the system detects that the task would require 90 minutes to complete at 8:00 AM, but the RPA cluster resources require 120 minutes, triggering manual intervention.

[0079] Requests can be sent to a pre-set management terminal (such as the head office's risk control center console) via pop-up window, text message, or email, accompanied by animated reminders and priority indicators (red indicates high-risk tasks). For example, a teller might receive a "High-risk customer loan due diligence timeout warning" in the teller system and click to view detailed parameters. This can also automatically trigger the enterprise-level work order system to generate manual scheduling tasks, such as "[Urgent] High-risk customer due diligence timeout requires manual intervention," which can be assigned to a senior reviewer to manually adjust the task priority and reallocate RPA resources.

[0080] The embodiment of the present invention is a fault-tolerant supplementary link for intelligent scheduling of RPA clusters. Through structured information transmission and multi-channel triggering mechanisms, it ensures the continuity of business processes in the event of scheduling failures. When the system's automatic scheduling fails, tasks can be quickly transferred to manual processing processes through manual intervention requests, avoiding business interruptions due to insufficient RPA cluster resources and ensuring the smooth progress of time-sensitive businesses such as finance. Structured information transmission reduces manual re-entry and information confirmation time, and the dual-trigger mode adapts to different levels of urgency. For example, urgent tasks are quickly responded to through terminal push, and complex tasks are handled through standardized processes in the work order system, improving the overall scheduling fault tolerance capability. A closed-loop management of "intelligent scheduling-anomaly detection-manual backup" is formed to make up for the limitations of the automation system and enhance the stability and risk resistance of the RPA cluster in complex business scenarios.

[0081] S260. If so, the target task is associated with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and when the system time reaches the task start time of the target task, the due diligence operation is performed on the target task; wherein, the task start time is calculated by the deadline and the estimated task duration.

[0082] Furthermore, associating the target task with the to-be-scheduled queue of the adapted target server in the RPA cluster according to the scheduling optimization model may include:

[0083] Obtain the CPU utilization, memory usage, and total remaining task duration of each server's scheduled queue in the RPA cluster.

[0084] Taking minimizing the global task waiting time as the first optimization goal, the total remaining task duration of each server is added to the estimated task duration of the target task to obtain the estimated waiting time of the target task on each server;

[0085] Taking maximizing resource utilization as the second optimization goal, the comprehensive resource utilization of each server is calculated based on the CPU utilization and memory occupancy of each server;

[0086] Calculate the comprehensive scheduling score of each server based on the estimated waiting time and comprehensive resource utilization of each server;

[0087] The server with the highest comprehensive score is selected as the target server, and the target task is inserted into the to-be-scheduled queue of the target server based on the task start time of the target task.

[0088] First, the CPU utilization, memory occupancy, and total remaining task duration of the queue to be scheduled of each server in the cluster are collected in real time to construct a profile of the server resource status. Then, based on the dual-objective optimization model, on the one hand, with the goal of minimizing task waiting time, the total duration of each server queue is added to the estimated duration of the target task to obtain the estimated waiting time of the task on different servers; on the other hand, with the goal of maximizing resource utilization, the comprehensive resource utilization is calculated by combining the CPU and memory occupancy. By normalizing the time waiting index and the resource utilization index and then weighting them to calculate the comprehensive scheduling score, the server with the highest comprehensive score is selected as the target server. Finally, based on the start time of the target task (determined by the estimated duration reversed from the deadline), the tasks are inserted into the queue to be scheduled of the target server in ascending order of the deadline, ensuring that the tasks achieve efficient utilization of cluster resources while meeting the time requirements.

[0089] The embodiment of the present invention is the core link for RPA clusters to achieve intelligent task allocation. Through multi-dimensional resource evaluation and optimization model calculation, the target task is accurately matched to the optimal server. This mechanism can significantly improve the scheduling efficiency and resource utilization of RPA clusters in high-concurrency financial scenarios by dynamically balancing task waiting time and resource load.

[0090] Optionally, a comprehensive scheduling score for each server is calculated based on the estimated waiting time and comprehensive resource utilization of each server, which may include:

[0091] According to the formula: U 综合 =ω1×U cpu +ω2×U 内存 , calculate the comprehensive resource utilization of each server U 综合 Among them, U cpu is the CPU utilization, U 内存 is the memory occupancy rate, ω1 is the weight coefficient of CPU utilization, ω2 is the weight coefficient of memory occupancy rate, ω1+ω2=1;

[0092] According to the formula: F=(1-U 综合 ), calculate the idle resource ratio F of each server;

[0093] According to the formula: The comprehensive scheduling score S of each server is calculated; wherein α is the weight coefficient of the first optimization objective, β is the weight coefficient of the second optimization objective, and α+β=1.

[0094] First, build a weighted model for server resource usage: 综合 =ω1×U cpu +

[0095] ω2×U 内存 Calculate resource utilization. For example, if the CPU utilization of server S1 is 70% and the memory utilization is 60%, and if ω1 = 0.6, then U 综合 =66%. Among them, the weight ω1 can be increased for computing-intensive tasks; the weight ω2 can be increased for memory-intensive tasks to achieve business awareness of resource evaluation. Through the formula F = (1-U 综合 ) Convert resource utilization into idle resource ratio as a quantitative indicator of resource optimization goal. For example, U 综合 = 66%, then its idle resource ratio F = 34%. The higher F, the more idle the server is, and the more tasks it can handle. This conversion shifts resource utilization from pursuing high utilization to pursuing low occupancy, which is consistent with the direction of time optimization goals. The final comprehensive score formula It is used to achieve a balance between time and resources, where 1 / estimated waiting time is used to convert waiting time into a time efficiency score (e.g., waiting for 120 minutes corresponds to a score of approximately 0.0083). The shorter the waiting time, the higher the score. The larger the proportion of idle resources F, the more idle the resources, and the higher the score. The priority is dynamically adjusted through α and β (e.g., when α = 0.7, more emphasis is placed on time efficiency).

[0096] During peak periods of financial services, increasing the α value prioritizes timeliness; during off-peak periods, increasing the β value improves resource utilization, achieving flexible scheduling that balances timeliness and cost. Differentiated scheduling is achieved by adjusting the weights of servers with different configurations. For example, assigning β = 0.6 to a high-performance server gives it priority in taking on tasks when resources are sufficient. When a server experiences temporary performance fluctuations, U 综合 An increase in F decreases the overall score, automatically reducing task allocation to avoid failures. The weighted model in this embodiment of the present invention adjusts CPU and memory weights based on business scenarios, making resource evaluation more tailored to actual load characteristics. By converting time and resource dimensions into a unified scoring standard through mathematical formulas, subjective decision-making bias is avoided, improving the scientific nature and automation of cluster task allocation.

[0097] The embodiment of the present invention realizes efficient scheduling of RPA clusters through full-process intelligent design: ① Task hierarchical management, dividing unfinished tasks into queue tasks and free tasks. The former reads the real-time parameters of the server, and the latter estimates the duration based on historical data, accurately reflecting the cluster load status, and providing multi-dimensional data support for scheduling; ② The priority scheduling algorithm sorts tasks by deadline, and verifies the time window through iteration to ensure that there is no timeout risk after the target task is inserted, thereby ensuring business timeliness; ③ When scheduling fails, a manual intervention request with complete business information is generated, and manual intervention is triggered through the management terminal or work order system, forming a closed loop of automation and manual backup to ensure business continuity; ④ The multi-objective optimization model comprehensively calculates the scheduling score based on server resource utilization and task waiting time, dynamically matches the optimal server, improves cluster resource utilization and task execution efficiency, and adapts to the differentiated needs of financial business and regulatory compliance requirements.

[0098] Example 3

[0099] Figure 3 This is a schematic diagram of the structure of a due diligence device for Robotic Process Automation (RPA) provided in the third embodiment of the present invention. Figure 3 As shown, the device includes:

[0100] Due diligence request response module 310 is used to respond to the due diligence request for the target task from the business front-end system and extract the target business type, target customer type and target appointment processing time of the target task to form task data;

[0101] A target time parameter acquisition module 320 is configured to calculate the deadline time point and the estimated task duration of the target task based on the task data to obtain the target time parameter of the target task;

[0102] The historical time parameter acquisition module 330 is used to extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameter corresponding to each unfinished historical task;

[0103] The scheduling condition determination module 340 is used to integrate and sort the target time parameters and historical time parameters using a priority scheduling algorithm, and detect whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting results;

[0104] The due diligence task execution module 350 is used to, if so, associate the target task to the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and perform a due diligence operation on the target task when the system time reaches the task start time of the target task; wherein the task start time is calculated by the deadline and the estimated task duration.

[0105] The embodiment of the present invention realizes efficient scheduling of financial due diligence tasks through intelligent processes. First, it converts unstructured business requirements into computable task data; converts the time sensitivity of financial business into clear time parameters to ensure that tasks are executed within the minimum time window that meets the timeliness requirements, avoiding idle resources and task timeouts; the multi-objective optimization model combines server resource utilization and task waiting time to achieve intelligent matching of tasks and servers, improve the overall processing efficiency and resource utilization of the cluster, and is particularly suitable for financial scenarios with multiple branches and multiple businesses. By deeply integrating financial business logic with RPA scheduling technology, it solves the problems of low efficiency, uneven resource allocation, and insufficient timeliness response in the traditional due diligence process, and provides financial institutions with standardized and intelligent automation solutions.

[0106] Optionally, based on the above embodiments, the target time parameter acquisition module 320 may include:

[0107] A first deadline calculation unit is configured to query a preset rule base for a maximum waiting time corresponding to a target customer type based on the task data, and calculate a deadline for the target task using the target appointment processing time and the maximum waiting time;

[0108] The estimated task duration calculation unit is used to extract a historical processing duration sample set corresponding to the target business type from the historical task database based on the task data, and input the sample set into a pre-trained duration prediction model to calculate the estimated task duration of the target task.

[0109] Optionally, based on the above embodiments, the historical time parameter acquisition module 330 may include:

[0110] The unfinished task acquisition unit is used to obtain all unfinished tasks from the RPA cluster management system;

[0111] The task division unit is used to divide the first unfinished historical task that has been assigned to a server and is located in the to-be-scheduled queue into a queue task set, and divide the second unfinished historical task that has not been assigned to a server into a free task set.

[0112] A first historical time parameter acquisition unit is configured to read, for each first unfinished historical task in the queue task set, a historical deadline time point and a remaining task duration of each first unfinished historical task from the local state of the assigned server;

[0113] The second historical time parameter acquisition unit is used to obtain the historical deadline time point set when the task is created for the second unfinished historical task in the free task set, and query the average duration of similar tasks from the historical task database based on the task type of each second unfinished task as the remaining task duration of each second unfinished task.

[0114] Optionally, based on the above embodiments, the scheduling condition determination module 340 may include:

[0115] a time parameter merging unit, configured to merge the target time parameter and each historical time parameter into a time parameter set, and arrange the time parameter set in ascending order according to the end time point;

[0116] a target task locating unit, configured to form an ordered task queue with the target task and each unfinished historical task according to the sorted time parameter set, and locate the target task in the ordered task queue;

[0117] The task traversal unit is used to obtain the current tasks in order from the front to the back, starting from the location of the target task;

[0118] The time difference verification unit is used to calculate the total processing time of all tasks between the first task in the ordered task queue and the current task, and verify whether the total processing time is less than the time difference between the deadline of the current task and the current system time;

[0119] a first scheduling condition determination unit, configured to, if yes, return to executing an operation of sequentially acquiring the current task from the location of the target task in a front-to-back order, and determine that the scheduling condition for the target task is satisfied when determining that the current task is the last task in the ordered task queue;

[0120] The second scheduling condition determination unit is used to determine that the scheduling condition for the target task is not met if no; wherein the total processing time is the sum of the estimated task time of the target task and / or the remaining task time of the unfinished historical tasks.

[0121] Optionally, based on the above embodiments, the present invention may further include: a manual intervention unit configured to detect whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting result, and if the RPA cluster does not meet the scheduling conditions for the target task, generate a manual intervention request including the target business type, target customer type, target appointment processing time, target deadline, and target estimated task duration of the target task;

[0122] Push manual intervention requests to the preset management terminal, or trigger the work order system to create manual scheduling tasks based on manual intervention requests.

[0123] Optionally, based on the above embodiments, the due diligence task execution module 350 may include:

[0124] The RPA cluster parameter acquisition unit is used to obtain the CPU utilization and memory usage of each server in the RPA cluster and the total remaining task duration of each server's scheduling queue;

[0125] A first optimization module is configured to take minimizing the global task waiting time as a first optimization goal, add the total remaining task duration of each server to the estimated task duration of the target task, and obtain the estimated waiting time of the target task on each server;

[0126] The second optimization module is configured to calculate the comprehensive resource utilization of each server based on the CPU utilization and memory occupancy of each server, taking maximizing resource utilization as the second optimization goal;

[0127] A scheduling score calculation unit, used to calculate the comprehensive scheduling score of each server based on the estimated waiting time and comprehensive resource utilization of each server;

[0128] The target server determination unit is used to select the server with the highest comprehensive score as the target server, and insert the target task into the to-be-scheduled queue of the target server based on the task start time of the target task.

[0129] Optionally, based on the above embodiments, the scheduling score calculation unit can also be used to calculate the scheduling score according to the formula: 综合 =ω1×U cpu +ω2×U 内存 , calculate the comprehensive resource utilization of each server U 综合 Among them, U cpu is the CPU utilization, U内存 is the memory occupancy rate, ω1 is the weight coefficient of CPU utilization, ω2 is the weight coefficient of memory occupancy rate, ω1+ω2=1;

[0130] According to the formula: F=(1-U 综合 ), calculate the idle resource ratio F of each server;

[0131] According to the formula: The comprehensive scheduling score S of each server is calculated; wherein α is the weight coefficient of the first optimization objective, β is the weight coefficient of the second optimization objective, and α+β=1.

[0132] A due diligence device for robotic process automation (RPA) provided in an embodiment of the present invention can execute a due diligence method for robotic process automation (RPA) provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0133] Example 4

[0134] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0135] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0136] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0137] Processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors that run machine learning model algorithms, digital signal processors (DSPs), and any other suitable processor, controller, or microcontroller. Processor 11 executes the various methods and processes described above, such as a due diligence method for Robotic Process Automation (RPA).

[0138] That is, in response to the due diligence request for the target task from the business front-end system, the target business type, target customer type and target appointment processing time of the target task are extracted to form the task data;

[0139] Calculating the deadline time point and estimated task duration of the target task based on the task data to obtain target time parameters of the target task;

[0140] Extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task;

[0141] Use the priority scheduling algorithm to integrate and sort the target time parameters and historical time parameters. Based on the integrated sorting results, check whether the RPA cluster meets the scheduling conditions for the target task.

[0142] If so, the target task is associated with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and the due diligence operation is performed on the target task when the system time reaches the task start time of the target task; wherein, the task start time is calculated by the deadline and the estimated task duration.

[0143] In some embodiments, a due diligence method for a robotic process automation (RPA) may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the due diligence method for a robotic process automation (RPA) described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute a due diligence method for a robotic process automation (RPA) in any other appropriate manner (e.g., by means of firmware).

[0144] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0145] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0146] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0147] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0148] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0149] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0150] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0151] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A due diligence method for Robotic Process Automation (RPA), characterized by: The method comprises: In response to the due diligence request for the target task from the business front-end system, the target business type, target customer type, and target appointment processing time of the target task are extracted to form task data; Calculating the deadline time point and estimated task duration of the target task based on the task data to obtain target time parameters of the target task; Extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task; Use the priority scheduling algorithm to integrate and sort the target time parameters and historical time parameters. Based on the integrated sorting results, check whether the RPA cluster meets the scheduling conditions for the target task. If so, the target task is associated with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and the due diligence operation is performed on the target task when the system time reaches the task start time of the target task; wherein, the task start time is calculated by the deadline and the estimated task duration.

2. The method according to claim 1, characterized in that Calculate the target task deadline and estimated task duration based on the task data, including: Based on the task data, the preset rule library is searched for the longest waiting time corresponding to the target customer type, and the deadline of the target task is calculated by the target appointment processing time and the longest waiting time; A historical processing duration sample set corresponding to the target business type is extracted from a historical task database based on the task data, and the sample set is input into a pre-trained duration prediction model to calculate the estimated task duration of the target task.

3. The method according to claim 2, characterized in that Extract all unfinished historical tasks from the RPA cluster and obtain the historical deadlines and remaining task durations for each unfinished historical task, including: Get all unfinished tasks from the RPA cluster management system; Among all unfinished tasks, the first unfinished historical task that has been assigned to a server and is located in the queue to be scheduled is divided into the queue task set, and the second unfinished historical task that has not been assigned to a server is divided into the free task set; For the first unfinished historical task in the queue task set, read the historical deadline time point and the remaining task duration of each first unfinished historical task from the local state of the assigned server; For the second unfinished historical tasks in the free task set, obtain the historical deadline time point set when the task was created, and based on the task type of each second unfinished task, query the average duration of similar tasks from the historical task database as the remaining task duration of each second unfinished task.

4. The method according to claim 3, characterized in that The priority scheduling algorithm is used to integrate and sort the target time parameters and historical time parameters. Based on the integrated sorting results, the RPA cluster is checked to see if it meets the scheduling conditions for the target task, including: Merging the target time parameter and each historical time parameter into a time parameter set, and arranging the time parameter set in ascending order according to the end time point; According to the sorted time parameter set, the target task and each unfinished historical task form an ordered task queue, and locate the target task in the ordered task queue; Starting from the location of the target task, obtain the current tasks in order from front to back; Calculate the total processing time of all tasks between the first task in the ordered task queue and the current task, and verify whether the total processing time is less than the time difference between the deadline of the current task and the current system time; If so, the process returns to executing the operation starting from the location of the target task and sequentially obtaining the current task in order from the front to the back. When the current task is determined to be the last task in the ordered task queue, the scheduling condition for the target task is determined to be met. If not, it is determined that the scheduling conditions for the target task are not met; wherein the total processing time is the sum of the estimated task time of the target task and / or the remaining task time of the unfinished historical tasks.

5. The method according to claim 4, characterized in that After checking whether the RPA cluster meets the scheduling conditions for the target task based on the integrated sorting results, the following steps are also performed: When the RPA cluster is detected as not meeting the conditions for scheduling the target task, a manual intervention request is generated, including the target business type, target customer type, target appointment processing time, target deadline, and target estimated task duration. Push manual intervention requests to the preset management terminal, or trigger the work order system to create manual scheduling tasks based on manual intervention requests.

6. The method according to any one of claims 1 to 5, characterized in that The target task is associated with the scheduling queue of the adapted target server in the RPA cluster based on the scheduling optimization model, including: Obtain the CPU utilization, memory usage, and total remaining task duration of each server's scheduled queue in the RPA cluster. Taking minimizing the global task waiting time as the first optimization goal, the total remaining task duration of each server is added to the estimated task duration of the target task to obtain the estimated waiting time of the target task on each server; Taking maximizing resource utilization as the second optimization goal, the comprehensive resource utilization of each server is calculated based on the CPU utilization and memory occupancy of each server; Calculate the comprehensive scheduling score of each server based on the estimated waiting time and comprehensive resource utilization of each server; The server with the highest comprehensive score is selected as the target server, and the target task is inserted into the to-be-scheduled queue of the target server based on the task start time of the target task.

7. The method according to claim 6, characterized in that Based on the estimated waiting time and comprehensive resource utilization of each server, the comprehensive scheduling score of each server is calculated, including: According to the formula: U 综合 =ω1×U cpu +ω2×U 内存 , calculate the comprehensive resource utilization of each server U 综合 Among them, U cpu is the CPU utilization, U 内存 is the memory occupancy rate, ω1 is the weight coefficient of CPU utilization, ω2 is the weight coefficient of memory occupancy rate, ω1+ω2=1; According to the formula: F=(1-U 综合 ), calculate the idle resource ratio F of each server; According to the formula: The comprehensive scheduling score S of each server is calculated; wherein α is the weight coefficient of the first optimization objective, β is the weight coefficient of the second optimization objective, and α+β=1.

8. A due diligence device for Robotic Process Automation (RPA), characterized by: The device comprises: Due diligence request response module, used to respond to the due diligence request for the target task from the business front-end system, extract the target business type, target customer type and target appointment processing time of the target task to form task data; A target time parameter acquisition module is used to calculate the deadline time point of the target task and the estimated task duration based on the task data to obtain the target time parameters of the target task; A historical time parameter acquisition module is used to extract all unfinished historical tasks from the RPA cluster, obtain the historical deadline time point and remaining task duration of each unfinished historical task, and obtain the historical time parameters corresponding to each unfinished historical task; The scheduling condition determination module is used to integrate and sort the target time parameters and historical time parameters using a priority scheduling algorithm. Based on the integrated sorting results, it detects whether the RPA cluster meets the scheduling conditions for the target task. The due diligence task execution module is used to, if so, associate the target task with the to-be-scheduled queue of the target server that matches the target task in the RPA cluster according to the scheduling optimization model, and perform due diligence on the target task when the system time reaches the task start time of the target task; wherein the task start time is calculated by the deadline and the estimated task duration.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a due diligence method for robotic process automation (RPA) according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement a due diligence method for robotic process automation (RPA) according to any one of claims 1 to 7 when executed.

Images