Digital twinning-based transformer dynamic heat dissipation performance evaluation and prediction method

Abstract

The invention relates to the technical field of thermal management of power equipment, in particular to a method for evaluating and predicting dynamic heat dissipation performance of a transformer based on digital twinning, which comprises the following steps of: constructing a digital twinning body consistent with a physical entity of the transformer, and fusing real-time operation data and a multi-physics field simulation model; and dynamically evaluating the heat dissipation performance of the transformer under different operation conditions. And trend prediction is performed on key indexes such as hot-spot temperature rise and cooling efficiency by combining a deep learning algorithm, and early warning and regulation suggestions are output based on risk levels. According to the method, fine modeling and intelligent prediction of the thermal state of the transformer can be realized, the operation reliability and the operation and maintenance efficiency are improved, and the method is suitable for intelligent management scenes of various oil-immersed or dry-type transformers in an electric power system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment thermal management, and particularly relates to a transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning. BACKGROUND

[0002] The transformer is the core power transmission and transformation equipment in the power system, and a large amount of heat will be generated in the operation process. If the heat cannot be dissipated in time and effectively, the winding insulation will be aged, the service life will be shortened, and even a fault or shutdown will be caused. The traditional heat dissipation performance evaluation method mainly relies on offline simulation or empirical formula, and it is difficult to reflect the thermal characteristics in the actual working condition in real time, and it lacks the prediction ability of future trends.

[0003] In recent years, the development of digital twinning technology provides a new path for intelligent modeling and state evaluation of power equipment. Through the mapping relationship between the virtual model and the physical entity, real-time perception and prediction analysis of the equipment operating state can be realized. However, the existing technology is mainly used for electrical parameter modeling, and the dynamic simulation, data fusion and heat dissipation trend prediction research of the internal heat distribution of the transformer are still insufficient, and there is a lack of highly coupled multi-source data driven mechanism and intelligent analysis system.

[0004] Therefore, an evaluation method combining digital twinning modeling, thermal coupling simulation and intelligent prediction algorithm is needed to realize the dynamic perception, prediction and early warning and intelligent control of the heat dissipation performance of the transformer. SUMMARY

[0005] The present application relates to the technical field of power equipment thermal management, and particularly relates to a transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning.

[0006] To achieve the above-mentioned purpose, the present application provides a transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning. The method takes the transformer equipment as the object, constructs a one-to-one mapping digital twin model, and fuses multi-source operation data and thermal coupling simulation algorithm to realize accurate modeling, dynamic evaluation and intelligent prediction of the heat dissipation performance in the whole operation cycle. The specific steps include the following steps:

[0007] Transformer model information acquisition and simulation preparation stage: collect the basic information of the target transformer such as structure parameters, material properties, cooling mode, rated capacity, design drawings, etc., and establish a structure model for simulation; set boundary conditions according to the operating environment of the transformer, including ambient temperature, wind speed, installation method, load characteristics, etc., to provide initial conditions for subsequent simulation and digital twinning modeling.

[0008] The digital twin construction stage: a digital twin corresponding to the physical entity of the transformer is constructed, and the digital twin at least includes the following modules:

[0009] 1. A three-dimensional geometric structure modeling module for establishing a spatial topology structure and a thermal node distribution model of the transformer;

[0010] 2. A thermal physical parameter library for storing thermal conductivity, specific heat capacity, density, resistivity and oil parameters of different materials in the transformer;

[0011] 3. A multi-source data interface module for establishing real-time data communication with sensors installed on the transformer body and its surroundings;

[0012] 4. A multi-physical field coupling simulation module for dynamically calculating and updating the thermal distribution state based on the current operating data.

[0013] Real-time operation data acquisition and fusion stage: through the deployment of intelligent sensing devices in key parts such as windings, cores, oil tanks, and air coolers, continuous acquisition of operating data is carried out, and the operating data includes but is not limited to winding current, winding temperature, oil temperature, ambient temperature and humidity, oil flow speed, cooling air speed, core surface temperature, resistance change rate, etc. The collected data is preliminarily processed and abnormally filtered through an edge computing device, and is fused into the digital twin for real-time model updating.

[0014] Dynamic heat dissipation performance evaluation stage: based on the updated digital twin, the heat dissipation capacity of the transformer under the current load and environmental conditions is calculated by combining finite element thermal simulation and CFD fluid simulation, and multiple thermal performance indicators including hot spot temperature, oil flow trajectory, heat flux distribution, and cooler heat exchange efficiency are obtained to judge the effectiveness and safety margin of the heat dissipation system under the current operating state.

[0015] Intelligent prediction stage of heat dissipation trend: a heat dissipation performance prediction model is constructed using time series modeling technology and machine learning algorithms, and the prediction model can be selected from long short-term memory network (LSTM), convolutional neural network (CNN), XGBoost or a combination model thereof. By inputting the collected historical and current operating data, the prediction output of hot spot temperature rise, oil temperature change, cooling capacity degradation and other thermal indicators in a specific time window in the future is realized, and early warning of abnormal working conditions is supported.

[0016] Early warning mechanism and operation suggestion output stage: according to the prediction result and the set risk level threshold, a multi-level early warning signal is automatically generated, and the risk level at least includes five levels of normal, attention, moderate early warning, serious early warning and emergency shutdown; the system outputs corresponding operation suggestions according to different levels, including increasing the rotating speed of the cooling equipment, reducing the load of the transformer, planning maintenance or temporary shutdown, etc., and can realize linkage control with the substation monitoring system (such as SCADA or DCS).

[0017] Visual display and operation interaction stage: three-dimensional heat maps, trend line graphs, hotspot distribution cloud maps and other visualization means are used to present the current evaluation results and prediction trends on the operation interface, support users to query the historical heat evolution process, and can also be accessed remotely through mobile terminals to realize real-time supervision and rapid response to key heat indicators.

[0018] Preferably, the method adopts an edge-cloud collaborative mode on a system deployment architecture, wherein the edge computing node is responsible for on-site data acquisition, preliminary evaluation and model rapid response, the cloud platform is responsible for global data fusion, model training and multi-device collaborative analysis, and supports online updating and optimization of the model.

[0019] Through the above steps, the transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning can guarantee the modeling accuracy while realizing dynamic tracking and forward-looking judgment of heat performance changes, improving the intelligent level and safety of transformer heat dissipation system management.

[0020] The technical scheme provided by the embodiment of the application brings at least the following beneficial effects:

[0021] In the application, by constructing a digital twinning model of the transformer, real-time operation data and multi-physical field simulation are fused to realize dynamic evaluation and intelligent prediction of the heat dissipation performance of the transformer. Compared with the traditional method, the method has the advantages of strong real-time evaluation, high prediction accuracy, timely risk warning, high intelligent degree of operation and maintenance, etc. The method can effectively avoid failures caused by overheating, prolong the service life of the equipment, improve the operation and maintenance efficiency, and is suitable for heat management and intelligent monitoring scenes of various types of transformers, and has significant engineering application value and popularization prospect. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 It is a general architecture diagram of the system of the application;

[0023] Figure 2 It is a functional module structure diagram of the transformer digital twinning body of the application;

[0024] Figure 3 It is a flowchart of heat performance evaluation and prediction of the application. DETAILED DESCRIPTION

[0025] In order to better illustrate the technical solutions of the present application, the transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning is described in detail below in combination with the drawings. It should be understood that the embodiments are only used to explain the technical concept of the present application, and are not a limitation on the scope of protection.

[0026] The overall architecture of the system of the present application is shown in Figure 1 , which includes a data acquisition layer, an edge computing layer, a digital twinning modeling layer, an evaluation and prediction layer, and a display and control layer. The data acquisition layer monitors the operating state of the transformer through various sensors, and collects parameters including winding temperature, oil temperature, current, voltage, ambient temperature and humidity, wind speed, oil pump flow rate, etc. The data is uploaded to the edge computing device through industrial protocols (such as Modbus or CAN) for preliminary processing.

[0027] The edge computing layer is deployed near the field device and has data preprocessing, fault-tolerant computing and local simulation capabilities. In the case of network interruption or abnormality, the edge can complete rapid evaluation and alarm. The processed data will be uploaded to the cloud digital twinning modeling platform.

[0028] As shown in Figure 2 , the digital twin is composed of multiple functional modules, including a three-dimensional structure modeling module, a thermal-physical parameter library module, a multi-source data interface module, a multi-physical field simulation module, a prediction and early warning module, and a self-correction module. The three-dimensional structure modeling module establishes a geometric model according to the transformer structure drawing, covering key components such as windings, cores, oil tanks, and cooling devices. The thermal-physical parameter library records the thermal conductivity, specific heat capacity, density, and insulating medium properties of the materials, which can be dynamically adjusted to match the actual working conditions on site.

[0029] The multi-source data interface module interfaces with field collected data and historical operation records to provide input for simulation and prediction. The multi-physical field simulation module based on finite element analysis and CFD method, couples the simulation of electric-thermal-fluid three fields, simulates the heat generation, heat transfer, oil flow and heat exchange process of the cooler. The output results include hot spot temperature distribution, local temperature rise, cooling efficiency and other thermal performance indicators.

[0030] In the above thermal performance evaluation process, in order to realize the quantitative analysis of the hot spot temperature rise, the system introduces the following model formula based on the principle of heat balance:

[0031] ;

[0032] Where, represents the hot spot temperature rise of the transformer at time , is the total heat generation power, including copper loss and iron loss, which is dynamically calculated based on real-time collected current, resistance and other parameters; Equivalent thermal resistance reflects the impedance of heat conduction and dissipation processes and is affected by structure, oil flow, and material properties. This is a correction factor for the efficiency of the cooling device, which is used to adjust the actual heat exchange capacity of the cooling system under different operating conditions based on the dynamic changes in the speed of the air cooler or the flow rate of the oil pump.

[0033] This model formula provides a physically interpretable computational basis for the system assessment of hotspot temperature rise trends. Combined with actual measurement and simulation results, it can further enhance the accuracy of predictive model training.

[0034] Combination Figure 3 The process shown in the invention includes the following steps:

[0035] First, the initialization phase configures the digital twin model and system parameters, providing a structural foundation for simulation and prediction. Then, the system periodically collects on-site operational data and performs data preprocessing, including anomaly removal, interpolation completion, and unit normalization. The processed data is then input into the digital twin to drive the simulation calculation of the current thermal state.

[0036] During the evaluation phase, the simulation module outputs hotspot temperature rise values, radiator efficiency, and temperature difference distribution, forming a thermal performance evaluation vector. Subsequently, this evaluation data, along with historical data, is input into the trained prediction model. This model, based on an LSTM network or ensemble learning algorithm, can predict the hotspot temperature change trend and cooling system efficiency changes over several future time steps.

[0037] When the predicted results exceed the set threshold, the system triggers multi-level risk warnings. Risk levels are categorized based on the prediction results: normal, watch out, moderate warning, severe warning, and emergency shutdown. For different risk levels, the system can generate control recommendations, such as increasing the operating intensity of cooling equipment, reducing load, or scheduling backup equipment. Warning results are simultaneously transmitted to the SCADA system or DCS system, supporting automatic or manual intervention.

[0038] The system also visualizes simulation and prediction results, displaying 3D heatmaps, trend curves, and hotspot migration trajectories. Maintenance personnel can view the thermal status of equipment in real time through the platform interface, enabling remote monitoring and fault prevention. Simultaneously, all assessment and prediction records can be used for historical review, model optimization, and operational analysis.

[0039] Furthermore, this invention supports group modeling of multiple transformers, with each device corresponding to an independent digital twin. The cloud platform can collaboratively manage the thermal performance status of multiple devices and perform unified scheduling based on prediction results, achieving optimized thermal safety control at the station group level.

[0040] Through the above embodiments, the heat dissipation capability of the transformer under various operating environments can be accurately modeled, dynamically monitored and risk predicted, thereby improving the operation reliability and intelligent management level of the equipment.

[0041] The above merely provides a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning, characterized in that, The method comprises the following steps: Data acquisition and modeling preparation: Obtain the structural parameters, material properties, cooling system type and historical operation data of the target transformer, and establish an initial simulation model including a geometric model, an electro-thermal coupling model and boundary conditions; Digital twin construction: Construct a digital twin corresponding to the transformer entity, which includes a three-dimensional structure model, a thermal-physical parameter library, a real-time sensor data interface module and an adaptive simulation module for real-time mapping of the internal thermal state of the transformer; Real-time operation data fusion: Continuously collect operation data of key parts of the transformer through edge perception terminals, including but not limited to load current, winding temperature, oil temperature, ambient temperature and humidity, cooling air speed and conductor resistance changes, and fuse the data into the digital twin; Dynamic heat dissipation performance evaluation: Use multi-physics field coupling algorithm to simulate heat conduction, heat convection and heat radiation in the digital twin, and obtain dynamic heat dissipation performance indicators including hot spot temperature rise, oil flow velocity distribution, winding heat dissipation capacity, cooler efficiency, etc.; Prediction model construction and training: Based on the evaluation results, construct a heat dissipation feature time series, and use a deep learning model for fitting and prediction. The model input is the historical and current heat dissipation indicator sequence, and the output is the heat dissipation trend in the future specified time window; Early warning and decision support output: According to the prediction results, set up a multi-level early warning mechanism, when the prediction index approaches the critical temperature rise or heat dissipation failure threshold, generate a risk prompt, and provide operable operation optimization suggestions or maintenance strategies.

2. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The three-dimensional structure model in the digital twin is constructed based on physical modeling and finite element mesh partitioning of the transformer, and the thermal-physical parameter library includes the thermal conductivity, specific heat capacity, density and surface emissivity of different materials, and can be updated in real time according to the operating state.

3. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The multi-physics field coupling algorithm includes an electro-thermal coupling module, an oil and gas flow coupling module, and a dynamic adjustment module of the heat dissipation path, wherein the electro-thermal coupling module is used to process resistance heating and heat conduction effects, and the oil and gas flow module uses Navier-Stokes equation to solve the cooling medium flow path and velocity distribution.

4. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The prediction model is an integrated multi-model framework, including a long short-term memory network (LSTM) for capturing temperature change trends, a convolutional neural network (CNN) for extracting spatial heat distribution features, a random forest model for evaluating the impact of cooling system efficiency fluctuations on heat dissipation performance, and a prediction result fusion through ensemble learning.

5. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The early warning mechanism is based on self-defined risk level standards, divided into five levels: normal, mild warning, moderate warning, severe warning and emergency shutdown, and supports linkage with the SCADA system or DCS system of the transformer to realize automatic adjustment of air cooling / oil cooling intensity or trigger load limiting strategy.

6. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The method supports offline mode and online mode switching, offline mode for long-period evaluation and model training, online mode for real-time dynamic simulation and prediction, and the two modes can realize data synchronization and error compensation through a bidirectional correction mechanism.

7. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The hotspot temperature rise value in the evaluation process is obtained by fusing sensor data and simulation data, and a Kalman filtering algorithm is used for data fusion optimization to ensure that the accuracy of the temperature rise estimation is better than ±1 ℃.

8. The transformer dynamic heat dissipation performance evaluation and prediction method based on digital twinning of claim 1, wherein, The method is deployed in a collaborative computing architecture composed of an edge computing device and a cloud platform. The edge device completes data acquisition and local simulation, the cloud platform performs global modeling, long-period trend prediction, and multi-transformer group optimization scheduling tasks, and supports remote visual access and alarm push functions on the mobile terminal.

Images