CN-121981740-A - Digital twinning-based park management method and system

CN121981740ACN 121981740 ACN121981740 ACN 121981740ACN-121981740-A

Abstract

The invention relates to the technical field of park management and discloses a park management method and system based on digital twinning, wherein the park management method based on digital twinning comprises the steps of obtaining real-time operation data and environment parameter data of Internet of things equipment in a park, preprocessing, associating a park building information model with a standardized data set, extracting energy consumption time-space characteristics, identifying causal relations, calculating carbon emission of each region, constructing an energy consumption time-space association feature library, constructing a causal perception prediction model, determining a key regulation time window and region, constructing a candidate regulation strategy, evaluating, selecting an optimal regulation scheme to send a control instruction, collecting operation data after regulation and control execution in real time, evaluating regulation and control effects, establishing a regulation knowledge base, and driving decision optimization.

Inventors

NIU ZENGQIANG
MA ZHENFANG
GAO QIANG
WEN TIAN
LI YU
Yang Linqi

Assignees

创意银航(山东)技术有限公司

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. A digital twinning-based campus management method, comprising: Acquiring real-time operation data and environment parameter data of Internet of things equipment in a park, and performing space-time labeling to obtain an original data set; The method comprises the steps of preprocessing an original data set to obtain a standardized data set, constructing a static geometric model based on park BIM data and a building construction drawing, and associating the static geometric model with the standardized data set to obtain a digital twin simulation model; Extracting energy consumption time-consuming empty features based on a standardized data set, and identifying causal relations to obtain an energy consumption influence factor causal graph; Constructing a causal sensing prediction model based on an energy consumption carbon emission space-time correlation feature library and an energy consumption influence factor causal graph, and outputting an energy consumption and carbon emission prediction result; Determining a key regulation time window and a region based on the energy consumption and carbon emission prediction result, constructing a candidate regulation strategy, and evaluating; The control command is issued based on a predictive regulation command set, and the running data after regulation and control execution is collected in real time, and a digital twin simulation model and a causal sensing prediction model are updated; and establishing a regulation and control knowledge base, and driving decision optimization to obtain the intelligent energy carbon collaborative park management system.
2. The digital twinning-based campus management method of claim 1, wherein the step of obtaining the energy consumption influencing factor causal graph includes: Based on the historical energy consumption data of each monitoring point in the standardized data set, decomposing the energy time sequence into a trend item, a season item and a residual item by adopting a seasonal trend decomposition method; Acquiring key variables of a park comprehensive energy management system subsystem, carrying out the Granges causal examination two by two, and considering that the Granges causal relation exists when the F statistic is greater than a critical value; Identifying and eliminating false causal relation by adopting a conditional independence test method, introducing candidate confusion variables to perform partial correlation analysis, and eliminating false association caused by the confusion variables; And combining the domain knowledge and the physical rule of the park energy system, and adopting an expert rule verification method to carry out rationality verification on the identified causal relationship to construct an energy consumption influence factor causal graph.
3. The digital twinning-based campus management method of claim 1, wherein the step of constructing the causal awareness prediction model includes: A deep neural network technology is adopted to construct a prediction model integrating time dependence, space association and causal mechanism, wherein a time dependence branch adopts an LSTM to capture the time dependence of energy consumption, the space association adopts a graph neural network to capture the space propagation rule of the energy consumption, and the causal mechanism gives different attention weights to variable pairs according to a causal graph of energy consumption influencing factors; the pre-training causal perception prediction model parameters trained in the source domain park are migrated to the target domain park by adopting a migration learning method, and the data distribution characteristics of the new scene are adapted through fine tuning; introducing an online self-adaptive learning mechanism, and when data distribution deviation is detected, updating causal sensing prediction model parameters by adopting an incremental learning algorithm, so as to keep the adaptability to a dynamic environment; The structural information of the causal graph of the energy consumption influence factors is embedded into a neural network, and the causal perception prediction model is constrained to learn the feature representation conforming to the causal relation, so that the causal perception prediction model is constructed.
4. The digital twinning-based campus management method of claim 1, wherein the selecting the optimal regulation scheme includes: constructing a multi-objective optimization problem, taking energy consumption minimization and carbon emission minimization as objective functions, and taking equipment operation constraint and safety constraint as constraint conditions; Solving a multi-objective optimization problem by adopting an NSGA-II algorithm to obtain a pareto optimal solution set, wherein each solution corresponds to a regulation strategy; introducing a robust optimization method to process uncertainty factors, and generating a regulation strategy with robustness to parameter disturbance by constructing an uncertainty set and worst-case optimization; And performing simulation evaluation on the candidate regulation strategies in a digital twin simulation environment, calculating expected energy consumption and carbon emission effects of each strategy, and selecting the regulation strategy with optimal comprehensive performance.
5. The digital twinning-based campus management method according to claim 1, wherein the step of calculating the carbon emissions of each zone includes: Acquiring a real-time power grid carbon emission factor, and calculating carbon emission generated by power consumption by combining power consumption data of each region; calculating the carbon emission amount generated by the combustion of the fossil fuel based on the fossil fuel consumption data and the corresponding carbon emission coefficient of each region; taking the carbon emission reduction effect of new energy power generation in the park into consideration, and obtaining the actual carbon emission of each period of each region by adopting a net carbon emission calculation method; and introducing a carbon emission uncertainty quantification method, calculating a confidence interval of the carbon emission, and providing reliability assessment for regulation and control decisions.
6. The digital twinning-based campus management method of claim 1, wherein the step of updating the digital twinning simulation model includes: Establishing a data quality assessment mechanism, performing anomaly detection and data cleaning on newly acquired real-time data, and ensuring that the data quality meets the updating requirement of a causal sensing prediction model; A sliding window mechanism is adopted to manage historical data and real-time data, and when new data arrives, a training data set is updated and the time-out data is removed; introducing a concept drift detection algorithm, monitoring the change of data distribution, and triggering retraining of a causal sensing prediction model when the data distribution deviation is detected; and online updating is carried out on the causal perception prediction model by adopting an incremental learning algorithm, so that the adaptability of the causal perception prediction model to dynamic environment changes is maintained, and the real-time synchronization of the digital twin model and the physical park is ensured.
7. The digital twinning-based campus management method of claim 1, wherein the step of constructing and evaluating candidate regulation strategies includes: Establishing a regulation and control effect evaluation index system comprising an energy consumption reduction rate, a carbon emission reduction amount, economic benefits and equipment operation stability; comparing the actual operation data before and after regulation by adopting a comparison analysis method, and calculating the improvement degree of each evaluation index; Introducing a statistical significance test method, verifying the statistical significance of the regulation effect, and eliminating the influence of random factors; And constructing a visual display system of the regulation and control effect, generating a regulation and control effect report, and providing data support for the optimization of a follow-up regulation and control strategy.
8. The digital twinning-based campus management method according to claim 1, wherein the step of collecting the operation data after the regulation and control are performed in real time includes: establishing a distributed data acquisition network, and acquiring the running state, energy consumption data and environmental parameters of each device in real time through an Internet of things sensor and an intelligent instrument; Preprocessing the collected original data by adopting an edge computing technology, including data cleaning, format conversion and preliminary analysis, so as to reduce the data transmission load; establishing a data transmission protocol and a security mechanism, and ensuring the integrity and security of data in the transmission process; And comparing and analyzing the processed operation data with the historical reference data, identifying the regulation and control effect and the abnormal condition, and updating the state parameters of the digital twin model in real time.
9. The digital twinning-based campus management method of claim 1, wherein the step of building the regulatory knowledge base includes: collecting and sorting historical regulatory case data, including regulatory policies, execution conditions, regulatory effects, and environmental factors; identifying a correlation mode between a regulation strategy and a regulation effect by adopting a correlation rule mining algorithm, and extracting an effective regulation rule and a failure regulation mode; establishing a regulation knowledge representation framework based on an ontology, storing the regulation experience in a structural form, and supporting semantic query and reasoning of knowledge; And introducing a machine learning method to automatically update and optimize the regulation knowledge.
10. A digital twinning-based campus management system for performing the digital twinning-based campus management method of any one of claims 1-9, comprising: the data acquisition module is used for acquiring real-time operation data and environment parameter data of the Internet of things equipment in the park, and performing space-time annotation to obtain an original data set; The digital twin modeling module is used for preprocessing an original data set to obtain a standardized data set, constructing a static geometric model based on the campus BIM data and a building construction drawing and correlating the static geometric model with the standardized data set to obtain a digital twin simulation model; the space-time feature analysis module is used for extracting energy consumption time-space features based on the standardized data set, identifying causal relations and obtaining an energy consumption influence factor causal graph; the causal sensing prediction module is used for constructing a causal sensing prediction model based on the energy consumption carbon emission space-time correlation feature library and the energy consumption influence factor causal graph to obtain an energy consumption and carbon emission prediction result; The intelligent regulation decision module is used for determining a key regulation time window and a key regulation area based on the energy consumption and carbon emission prediction result, constructing a candidate regulation strategy and evaluating the candidate regulation strategy; The control execution and model updating module is used for issuing control instructions based on a predictive control instruction set, collecting operation data after control execution in real time, updating the digital twin simulation model, and obtaining an updated causal sensing prediction model and a digital twin simulation model; And the effect evaluation and knowledge management module is used for evaluating the regulation and control effect based on the updated causal perception prediction model and the digital twin simulation model, and establishing a regulation and control knowledge base and driving decision optimization to obtain the intelligent energy carbon collaborative park management system.

Description

Digital twinning-based park management method and system Technical Field The invention relates to the technical field of park management, in particular to a digital twinning-based park management method and system. Background With the development of the 'two carbon' target and the deep advancement of smart city construction, the park is an important carrier for urban economic development, and the energy management and carbon emission control level directly influence the regional sustainable development capability. The high and new technology industrial park has the characteristics of various building types, complex equipment systems, large energy consumption scale, variable operation modes and the like, and massive multi-source heterogeneous operation data are generated every day. The traditional park management mode mainly relies on manual inspection, independent system monitoring and experience driving decision, and has the problems of serious information island, response lag, low efficiency of resource allocation and the like. The digital twin technology is taken as an emerging technical paradigm of deep fusion of the physical world and the digital world, and provides a new technical path for park management by constructing high-fidelity digital mapping of physical entities. However, the prior art has significant shortcomings in addressing the need for carbon co-management. The method mainly solves the problems that the processing of the energy consumption and carbon emission data of a park is stopped at a simple statistics and visualization level, deep excavation of the coupling relation of the energy consumption and the carbon emission in different space regions in different time periods is lacking, the deep law of energy management cannot be identified, subsystems in the park are independently processed in the prior art, excavation of the hidden association relation among different subsystems is lacking, global collaborative optimization cannot be achieved, the monitoring and post analysis of the current state are mainly achieved in the prior art, the future trend prediction and advanced regulation and control capability based on space-time association characteristics are lacking, the energy management is in a passive response state, the fixed carbon emission factor is mostly adopted in the prior art when the carbon emission is calculated, the dynamic change of the carbon emission intensity of a power grid and the fluctuation influence of new energy power generation are not considered, the carbon emission accounting accuracy is not high, the existing digital twin model is mainly used for basic state mapping and visual display, and the advantages of the method in aspects such as multi-time period simulation, scene deduction and strategy verification are not fully exerted. Therefore, there is a need to provide an intelligent park management method capable of deeply mining energy consumption carbon emission space-time correlation characteristics, realizing cross-period predictive regulation and control and dynamic and accurate carbon accounting. Disclosure of Invention The invention provides a park management method and system based on digital twinning, which solve the technical problems of insufficient extraction of time-space correlation characteristics of energy consumption and carbon emission, weak predictive regulation and control capability and lack of cross-domain collaborative optimization in the related technology. The invention provides a park management method based on digital twinning, which comprises the following steps: Acquiring real-time operation data and environment parameter data of Internet of things equipment in a park, and performing space-time labeling to obtain an original data set; The method comprises the steps of preprocessing an original data set to obtain a standardized data set, constructing a static geometric model based on park BIM data and a building construction drawing, and associating the static geometric model with the standardized data set to obtain a digital twin simulation model; Extracting energy consumption time-consuming empty features based on a standardized data set, and identifying causal relations to obtain an energy consumption influence factor causal graph; constructing a causal sensing prediction model based on an energy consumption carbon emission space-time correlation feature library and an energy consumption influence factor causal graph to obtain an energy consumption and carbon emission prediction result; Determining a key regulation time window and a region based on the energy consumption and carbon emission prediction result, constructing a candidate regulation strategy, and evaluating; the control command is issued based on a predictive regulation command set, operation data after regulation and control execution are collected in real time, and a digital twin simulation model is updated to obtain an updated causal sensing prediction model and a d