CN-121998446-A - Intelligent decision method, device, equipment and storage medium for scenes

CN121998446ACN 121998446 ACN121998446 ACN 121998446ACN-121998446-A

Abstract

The application discloses a scene-oriented intelligent decision method, device, equipment and storage medium. The method comprises the steps of loading corresponding plan graphs for instantiation based on input target business scene information, creating initial unified context corresponding to a plan instance, determining ready node queues in the plan instance by a scheduler based on the plan graphs and the initial unified context, executing a flow plan based on the ready node queues, writing execution results, verification results and monitoring results in the unified context during execution, carrying out local reevaluation on affected edges and nodes in the plan graphs based on the updated unified context, determining subsequent ready node queues and executing until the nodes in the plan instance are executed. The application provides an intelligent decision platform of a light application and a shared center, which supports a general intelligent decision system with dynamic arrangement, real-time response and continuous verification, so that a business process has the capabilities of sensing environment change and dynamic autonomous optimization.

Inventors

GUO SHENGMIN
Song Chongxian
LIU YANGYANG
AN LIMIN
ZHANG RUILONG
Han xingguang
LI YUNCAI
XIA SHUDONG

Assignees

北京掌行通信息技术有限公司

Dates

Publication Date: 20260508
Application Date: 20251209

Claims (12)

1. The intelligent decision method facing to the scene is characterized by comprising the following steps: based on the input target business scene information, loading a corresponding plan graph for instantiation, and creating an initial unified context corresponding to the plan instance; a scheduler determines a ready node queue in the plan instance based on the plan graph and the initial unified context; Based on the ready node queue execution flow plan, writing the execution result, the verification result and the monitoring result in the execution period back into the unified context; And carrying out local reevaluation on the affected edges and nodes in the plan graph based on the updated unified context, determining a subsequent ready node queue and executing until the nodes in the plan instance are executed.
2. The method of claim 1, wherein executing a flow plan based on the ready node queue comprises: Executing based on the ready node queue to obtain an execution result; Monitoring the execution result by adopting a monitor preset by a node; Triggering early warning when the execution result is within a preset abnormal interval range, carrying out local reevaluation on the affected side and node of the early warning, determining a subsequent ready node queue according to the evaluation result, and executing.
3. The method of claim 1, wherein loading the corresponding plan view for instantiation based on the entered target business scenario information comprises: Loading a corresponding plan graph from a plan library based on the target business scene information; Creating an operation example based on the plan graph, adding candidate workpieces to each pluggable node in the plan graph, and adding corresponding security daemon to each preset key node in the plan graph; Binding node roles in the plan graph to corresponding users, and writing global constraint and version snapshot information into the unified context.
4. The method of claim 1, wherein creating an initial unified context for a plan instance comprises: receiving request information and original context information of a target service scene input by a user; The related sensors or external data sources are scheduled according to the need to complement the original context information, and data normalization processing is carried out based on a preset unified data dictionary; and creating the initial unified context corresponding to the plan instance based on the unified context information after normalization processing, wherein the unified context is used for recording the execution full-link data of the plan instance.
5. The method of claim 1, wherein the scheduler determining a ready node queue in the plan instance based on the plan graph and the initial unified context comprises: After the current node is executed, the scheduler traverses each outgoing edge to acquire preset judging conditions, the judging conditions are calculated based on the unified context, and if the conditions are met, the node pointed by the edge is added into a ready node queue.
6. The method of claim 1, wherein writing back the execution results, the validation results, the monitoring results during execution to the unified context based on the ready node queue execution flow plan comprises: When the current node is a calculation node, calling a candidate workpiece corresponding to the node to execute calculation to obtain a calculation result and a confidence coefficient; When the current node is a form node, executing man-machine interaction based on the form node to obtain an interactive form; when the current node is a task node, executing a service action based on the task node to obtain a service execution result; When the current node is a rule node, executing a logic rule based on the rule node and outputting branch selection to obtain a rule judgment result; And writing the calculation result, the confidence coefficient, the form, the service execution result and the rule judgment result after the node execution back into the unified context.
7. The method of claim 1 or 6, wherein writing back the execution results, the validation results, the monitoring results during execution to the unified context based on the ready node queue execution flow plan, comprises: When the current node is a preset key node, aggregating the node confidence degrees of the current node leading links participating in the decision to obtain the total confidence degree of the key node; when the total grasping degree is larger than or equal to a confidence degree threshold value and the requisite constraint set is all met and the monitoring index meets the standard, determining that the key node passes through, and executing a subsequent path; And when the total confidence level is smaller than a confidence level threshold value, the requisite constraint set does not meet or the monitoring index triggers an early warning prompt, invoking a verification service to verify, obtaining a verification result, and writing the verification result back into the unified context.
8. The method of claim 1, wherein locally reevaluating affected edges, nodes in the plan view based on the updated unified context, determining a subsequent ready node queue and performing comprises: Based on a context monitoring mechanism, obtaining a unified context increment, and determining affected edges and nodes in a plan graph based on the unified context increment; for the affected edge, reevaluating the judging condition of the edge based on the unified context increment, and dynamically adjusting the node set in the subsequent ready node queue according to the change of the condition value; for re-entrant nodes, re-executing and updating output based on the unified context delta; and for the key nodes, judging whether the key nodes pass or not again based on the unified context increment.
9. The method according to claim 1 or 8, wherein the plan view is obtained by arranging a scene arrangement tool according to a preset rule, The local reevaluation is carried out on the affected edges and nodes in the plan graph based on the updated unified context, and a subsequent ready node queue is determined, and the method further comprises the following steps: And if the current plan view cannot cover the subsequent ready node sequence, adopting the scene editing tool to dynamically edit the plan view.
10. A scene-oriented intelligent decision making device, comprising: the plan instantiation module is used for loading a corresponding plan graph for instantiation based on the input target business scene information and creating an initial unified context corresponding to the plan instance; An execution module for determining a ready node queue in the plan instance by a scheduler based on the plan graph and the initial unified context; the result write-back module is used for writing back an execution result, a verification result and a monitoring result in the execution period into the unified context based on the execution flow plan of the ready node queue; and the dynamic correction module is used for carrying out local reevaluation on the affected edges and nodes in the plan graph based on the updated unified context, determining a subsequent ready node queue and executing until the nodes in the plan instance are executed.
11. An electronic device comprising a processor and a memory storing program instructions, the processor being configured, when executing the program instructions, to perform the scene-oriented intelligent decision method of any of claims 1 to 9.
12. A computer readable medium having stored thereon computer readable instructions for execution by a processor to implement a scene oriented intelligent decision method according to any of claims 1 to 9.

Description

Intelligent decision method, device, equipment and storage medium for scenes Technical Field The application relates to the technical field of information processing and business decision, in particular to a scene-oriented intelligent decision method, a device, equipment and a storage medium. Background In key business scenes such as accident emergency response, operation scheduling, traffic operation management and the like, a series of steps such as data acquisition, model prediction, rule judgment, instruction issuing, effect monitoring and the like are required to be sequentially executed. Currently, enterprises commonly adopt BPM (business process management) as a core, and combine an independently deployed rule engine, model service and form system to support the business. In the prior art, a single service application is used as a construction unit, and the self service flow, a data model, decision rules and an interactive interface are internally cured, and data exchange is carried out only through an application program interface between different application systems, so that the problems of tight coupling of architecture splitting and systems, scattered judgment caliber, insufficient scene coordination, verification and flow disconnection and the like are caused. With the improvement of the intelligent degree of a software system and the increase of the complexity of a service, the requirements of the service on agility cooperation, intelligent decision and continuous evolution under the background of digital intelligence are difficult to support by the architecture of the prior art, and an innovative architecture is needed to realize the centralized management of common performance and the unified intelligent arrangement and execution of a scenerization process. Disclosure of Invention The embodiment of the application provides a scene-oriented intelligent decision method, device, equipment and storage medium, which at least solve the technical problems that business collaboration, centralized process management and dynamic real-time response of scene-oriented intelligent decision are difficult to realize in the related technology. According to an aspect of the embodiment of the present application, there is provided a scene-oriented intelligent decision method, including: based on the input target business scene information, loading a corresponding plan graph for instantiation, and creating an initial unified context corresponding to the plan instance; the scheduler determines a ready node queue in the plan instance based on the plan view and the initial unified context; Based on the ready node queue execution flow plan, writing the execution result, verification result and monitoring result in the unified context; and carrying out local reevaluation on affected edges and nodes in the plan graph based on the updated unified context, determining a subsequent ready node queue and executing until the nodes in the plan instance are executed. According to another aspect of the embodiment of the present application, there is also provided an intelligent decision device facing to a scene, including: the plan instantiation module is used for loading a corresponding plan graph for instantiation based on the input target business scene information and creating an initial unified context corresponding to the plan instance; The execution module is used for determining a ready node queue in the plan instance by the scheduler based on the plan graph and the initial unified context; the result write-back module is used for writing back the execution result, the verification result and the monitoring result in the execution period into the unified context based on the execution flow plan of the ready node queue; And the dynamic correction module is used for carrying out local reevaluation on the affected edges and nodes in the plan graph based on the updated unified context, determining a subsequent ready node queue and executing until the nodes in the plan instance are executed. According to yet another aspect of the embodiments of the present application, there is also provided an electronic device, including a memory, and a processor, where the memory stores a computer program, and the processor is configured to execute the above-mentioned intelligent scene-oriented decision method by using the above-mentioned computer program. According to a further aspect of embodiments of the present application, there is also provided a computer readable storage medium having a computer program stored therein, wherein the computer program is arranged to execute the above-described intelligent scene-oriented decision method when run. The technical scheme provided by the embodiment of the application can have the following beneficial effects: The application provides a scene-oriented intelligent decision method. By dynamic instantiation of the plan graph and construction of unified context, quick response of the service sc