CN-121984222-A - Early warning and operation and maintenance system and method based on digital twin and artificial intelligence

CN121984222ACN 121984222 ACN121984222 ACN 121984222ACN-121984222-A

Abstract

The invention discloses a digital twin and artificial intelligence based early warning and operation and maintenance system and method, wherein the system comprises a data acquisition and tracing layer for endowing unique tracing identification for whole system data; the hierarchical digital twin modeling layer is used for constructing a hierarchical digital twin model mapped with the physical system in real time, the multi-mode big data analysis layer is used for driving the AI model to realize fault prediction and health assessment by fusing power grid, equipment and environment data, the hierarchical early warning and decision layer is used for generating different levels of early warning and operation and maintenance decisions according to analysis results, and the dispatching operation and maintenance closed-loop control layer is used for realizing automatic generation, remote control and cooperative linkage with power grid dispatching. The invention solves the problems of difficult fault positioning, low early warning precision, delayed operation and maintenance response and insufficient cooperation with the power grid of the high-voltage cascade energy storage system, and remarkably improves the operation reliability, operation and maintenance efficiency and the safety of interaction with the power grid.

Inventors

Niu dongyang
PING XIAOFAN
YANG CHAORAN
Duan Zhaorong
LIU WEI
WEI YU
WANG NING
SONG JISHUO
CHEN GUOZHANG
SUN ZHIWEI
ZHAO HU
CHENG QIAN
LIU MINGYI
CAO XI
CAO CHUANZHAO
LEI HAODONG

Assignees

华能国际电力股份有限公司河北清洁能源分公司
中国华能集团清洁能源技术研究院有限公司
华能平山清洁能源有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260114

Claims (10)

1. An early warning and operation and maintenance system based on digital twinning and artificial intelligence, which is characterized by comprising: The data acquisition and tracing layer is used for acquiring power grid side data, equipment side data and environment data of the high-voltage cascade energy storage system, and attaching a tracing label containing an equipment unique identity to the acquired data to obtain a structured data set; The hierarchical digital twin modeling layer is used for constructing a hierarchical digital twin model which is mapped with the physical system in real time and updating the data of the data acquisition and tracing layer, and the hierarchical digital twin model comprises a device-level model, a system-level model and a power grid interaction digital twin model; The multi-mode big data analysis layer is used for carrying out fusion analysis on the structured data set and calling a built-in AI model to output fault diagnosis, health assessment and maintenance window prediction results; The grading early warning and decision layer is used for generating early warning signals of different grades and corresponding operation and maintenance decision suggestions according to the output result of the multi-mode big data analysis layer; And the dispatching operation and maintenance closed-loop control layer is used for executing the operation and maintenance decision suggestion, generating a maintenance plan and completing closed-loop operation and maintenance control in cooperation with the power grid dispatching system.
2. The digital twinning and artificial intelligence based early warning and operation and maintenance system according to claim 1, wherein the data acquisition and tracing layer comprises: The sensor groups are distributed in the energy storage system and are used for collecting power grid side data, equipment side data and environment data of the high-voltage cascade energy storage system; The edge computing gateway is used for receiving and processing the data acquired by the sensor group; The unique identity of the equipment is generated by adopting a hierarchical coding rule, and the traceability tag further comprises acquisition time, data type and interaction object information; the data acquisition and tracing layer is also connected to a power grid dispatching system data interface to acquire power grid load curve and dispatching window period data.
3. The digital twinning and artificial intelligence based early warning and operation and maintenance system according to claim 2, wherein the hierarchical digital twinning modeling layer comprises: the device-level model is constructed based on three-dimensional laser scanning and physical rendering technology and is used for carrying out three-dimensional modeling and state mapping on the battery cell, the BMS subunit and the PCS power module; the system level model is integrated with the equipment level model and is used for simulating a cooperative control process and a data interaction link between the BMS and the PCS; The power grid interaction level digital twin model is used for accessing a power grid digital twin interface, mapping the electrical state of a high-voltage direct hanging point and simulating the influence of faults on a power grid; the hierarchical digital twin modeling layer is configured with a real-time data stream driven update mechanism and a physical location calibration mechanism based on timed three-dimensional scanning.
4. A digital twinning and artificial intelligence based early warning and operation and maintenance system according to claim 3, wherein the multi-modal big data analysis layer comprises: a data processing frame of a distributed computing frame is adopted for processing and storing real-time stream data and historical data; The multi-mode data fusion algorithm is used for converting structured data, unstructured data and time sequence data into unified feature vectors; The AI model base comprises a fault tracing and diagnosis model, a device health assessment model and a maintenance window prediction model.
5. The system for early warning and operation and maintenance based on digital twinning and artificial intelligence according to claim 4, wherein, The fault tracing and diagnosing model is based on a random forest algorithm, abnormal data with tracing labels are input, and fault types, responsibility units and fault diffusion risks are output; The equipment health assessment model is based on an LSTM neural network, inputs historical attenuation data of equipment, and outputs the residual life and health grade of the equipment; The maintenance window prediction model is based on a genetic algorithm, and combines a power grid load curve and scheduling window period data to output an optimal maintenance window and maintenance duration.
6. The digital twinning and artificial intelligence based early warning and operation and maintenance system according to claim 5, wherein the hierarchical early warning and decision layer comprises: the early warning classification mechanism is used for classifying early warning into equipment-level early warning, system-level early warning and power grid influence-level early warning; The decision output form displays early warning information, operation and maintenance decision suggestions and fault tracing details through a visual interface and the hierarchical digital twin model; wherein the equipment-level early-warning pushing equipment cooperates with the responsibility division report and the emergency strategy of the fault, and the grid influence level early warning pushes grid influence assessment and emergency scheduling cooperative information.
7. The digital twinning and artificial intelligence based early warning and operation and maintenance system according to claim 6, wherein the dispatch operation and maintenance closed loop control layer comprises: The remote control interface is used for issuing a control instruction to the physical equipment through the hierarchical digital twin model and receiving execution feedback; The maintenance plan automatic generation module is used for automatically generating an operation and maintenance plan table containing maintenance content, time, application materials and personnel configuration based on the optimal maintenance window and the equipment health grade, and synchronizing the operation and maintenance plan table to the power grid dispatching system; the closed-loop feedback optimization module is used for collecting fault processing effect data after operation and maintenance are completed, and feeding the fault processing effect data back to the multi-mode big data analysis layer to update AI model parameters; And the emergency linkage module is used for automatically pushing fault information and emergency shutdown suggestions to the power grid dispatching system when the power grid impact level early warning is triggered.
8. A method for early warning and operation and maintenance based on digital twinning and artificial intelligence, applying the system of any one of claims 1-7, comprising the steps of: S1, collecting multi-source data of a high-voltage cascade energy storage system, and attaching a traceability tag containing a unique identity of equipment to the multi-source data to form a structured data set, wherein the structured data set comprises real-time stream data and historical data; s2, constructing a hierarchical digital twin model, and updating by adopting real-time streaming data, wherein the hierarchical digital twin model comprises an equipment-level model, a system-level model and a power grid interaction digital twin model; S3, carrying out fusion analysis on the structured data set, calling an AI model to carry out fault diagnosis, health evaluation and maintenance window prediction, and generating a hierarchical early warning signal and an operation and maintenance decision suggestion; S4, pushing early warning information according to the operation and maintenance decision and the hierarchical early warning signals, and performing visual display in the hierarchical digital twin model; S5, based on the operation and maintenance decision and the grading early warning signals, cooperating with power grid dispatching and executing operation and maintenance operation; And S6, collecting feedback data after operation and maintenance execution, and feeding back the feedback data to the AI model for parameter optimization to realize closed loop feedback.
9. The method for early warning and operation and maintenance based on digital twinning and artificial intelligence according to claim 8, wherein the step S2 specifically comprises: acquiring point cloud data of physical equipment by adopting three-dimensional laser scanning, and constructing a layered digital twin model comprising an equipment-level model, a system-level model and a power grid interactive digital twin model by combining a physical rendering technology; Driving the hierarchical digital twin model to perform real-time state mapping by using the real-time stream data, and updating state parameters of the hierarchical digital twin model; And calibrating physical position parameters of the hierarchical digital twin model through three-dimensional scanning periodically.
10. The method for early warning and operation and maintenance based on digital twinning and artificial intelligence according to claim 8, wherein the step S5 specifically comprises: If the early warning level is at the equipment level or the system level, generating an operation and maintenance schedule based on the predicted optimal maintenance window, submitting a scheduling application, and executing maintenance operation based on the equipment state and the identifier mapped by the hierarchical digital twin model after batch acquisition; if the early warning level is the power grid influence level, automatically triggering emergency linkage, pushing an emergency shutdown suggestion to a power grid dispatching system, and disconnecting the fault system from the power grid.

Description

Early warning and operation and maintenance system and method based on digital twin and artificial intelligence Technical Field The invention belongs to the technical field of operation and maintenance of energy storage systems, and particularly relates to an early warning and operation and maintenance system and method based on digital twin and artificial intelligence. Background The current energy storage system technical development route gradually evolves to high-voltage cascade direct-hanging rack structure, compared with traditional energy storage, the high-voltage cascade direct-hanging system has the advantages of high efficiency, excellent electric energy quality, large single machine capacity, high response speed and the like, but once the high-voltage cascade energy storage system is in failure, serious influence can be caused on the power grid and the energy storage system due to direct connection with a high-voltage power grid, and the battery, the BMS and the PCS of the high-voltage cascade energy storage system are integrally designed, so that when the failure occurs, the failure cannot be accurately positioned, namely equipment hardware defect, control logic problem or collaborative interaction failure is not achieved, failure responsibility is divided cleanly, tracing is difficult, and operation and maintenance efficiency is low. Because the high-voltage cascade energy storage system is directly coupled with the high-voltage power grid, maintenance operation needs to be applied to a power grid dispatching part, the traditional operation and maintenance rely on manual fault pre-judgment and then application is submitted, response delay exists, the best opportunity of fault intervention is easily missed, and even the power grid fluctuation risk is caused. In the prior art, threshold early warning is carried out based on single equipment parameters (such as voltage and current), high-voltage power grid side data (such as power grid voltage fluctuation and frequency deviation), equipment cooperative data (such as BMS-PCS control instruction stream) and environmental data (such as power module heat dissipation temperature) are not fused, so that early warning false alarm rate and false alarm rate are high, and hidden faults cannot be identified in advance. The existing energy storage digital twin model is mainly aimed at a low-voltage distributed system, only realizes the visual appearance of equipment, does not construct refined mapping aiming at the layered cooperative characteristics (battery cell-module-system-power grid) of a high-voltage cascade system, and cannot simulate the BMS-PCS cooperative control process and the power grid interaction influence in real time, so that operation and maintenance personnel are difficult to intuitively sense the internal state of the system. Although the prior art provides application ideas of big data analysis and digital twinning, solutions are not designed aiming at the core characteristics of ' integration ', high-voltage network coupling ' and the like of a high-voltage cascade direct-hanging system, and especially the full-flow closed-loop technical scheme of ' fault tracing-maintenance scheduling-early warning decision ' is lacking, so that the operation and maintenance requirements of the high-voltage cascade direct-hanging energy storage system cannot be met. Disclosure of Invention The invention aims to provide a digital twinning and artificial intelligence based early warning and operation system and method, which overcome the defects of the prior art. In order to achieve the above purpose, the present invention provides the following technical solutions: In a first aspect, the invention provides a digital twin and artificial intelligence based early warning and operation and maintenance system, which comprises a data acquisition and tracing layer, wherein the data acquisition and tracing layer is used for acquiring power grid side data, equipment side data and environment data of a high-voltage cascade energy storage system, and attaching a tracing label containing a unique identity of equipment to the acquired data to obtain a structured data set; The hierarchical digital twin modeling layer is used for constructing a hierarchical digital twin model which is mapped with the physical system in real time and updating the data of the data acquisition and tracing layer, and the hierarchical digital twin model comprises a device-level model, a system-level model and a power grid interaction digital twin model; The multi-mode big data analysis layer is used for carrying out fusion analysis on the structured data set and calling a built-in AI model to output fault diagnosis, health assessment and maintenance window prediction results; The grading early warning and decision layer is used for generating early warning signals of different grades and corresponding operation and maintenance decision suggestions according to the output result of the mult