.png){kind=icon}
Return
AI-Based Customized Simulation Setup, Process Control, and Result Processing Technology for Distribution Networks
.png){kind=icon}
OA
AI
DOI：10.3390/pr14142333.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To address the two fundamental contradictions in distribution network digital simulation systems: “diversified simulation requirements versus standardized configuration” and “massive simulation outputs versus sparse decision-making information”, this paper builds upon existing work on Common Information Model (CIM)-based automatic simulation model generation and dynamic voltage regulation simulation and further constructs a user-oriented intelligent simulation service layer. This layer is collaboratively composed of an Orchestration_Agent (simulation orchestration agent) and an Analysis_Agent (result analysis agent), tasked with three responsibilities: based on multi-level simulation granularity (L0–L3) and a simulation template library, leveraging a large language model (LLM) to achieve natural language requirements parsing and automatic workflow orchestration; based on a simulation knowledge graph, implementing parameter recommendation, verification, and anomaly-adaptive recovery for process control; and based on a hybrid architecture of rule templates, statistical analysis, and causal graph models, achieving automatic result analysis, root cause reasoning, and structured report generation, with case feedback driving knowledge base iteration. Validation was conducted on data from a real 10 kV feeder with 91 distribution transformer areas over 30 consecutive days (2880 time cross-sections): comprehensive requirements-parsing accuracy of 96.3%, automatic parameter configuration coverage rate of 94.7%, anomaly identification recall/precision of 94.0%/96.9%, root cause reasoning accuracy of 92.1%, and the median end-to-end time per simulation shortened from approximately 36 min under the manual mode to 4.7 min. The results demonstrate that the proposed service layer provides a viable engineering technology pathway for the evolution of distribution network simulation from tool-oriented to service-oriented.
Keywords:
customized distribution network simulation
large language model agents (LLM agents)
natural language requirements parsing
automatic workflow orchestration
simulation process control
causal reasoning
simulation knowledge graph
root cause analysis
Journal
IF:
2.8
Papers:
7.3K
Citations:
3.7W
Organization
No organization information available
Cited Papers
Graph-Based Approaches and Functionalities in Retrieval-Augmented Generation: A Comprehensive Survey
PREAI
ShareSaveA Data–Physics-Driven Modeling Approach of Key Equipment for Large-Scale Distribution Network Simulation.png)
Electronics
ElectronicsIF0
PREAI
ShareSaveShareSaveShareSaveA Retrieval-Augmented Generation Approach for Data-Driven Energy Infrastructure Digital TwinsSMART CITIESOAAI
SMART CITIESIF5.5
OAAIShareSave
View more