Substation digital twin modeling and intelligent diagnosis method based on multi-source data fusion

Abstract

The application discloses a substation digital twin modeling and intelligent diagnosis method based on multi-source data fusion, and belongs to the technical field of intelligent operation and maintenance of power equipment. Point cloud, thermal imaging and vibration spectrum multi-source data are collected through a three-dimensional laser scanner, an infrared thermal imager and a vibration sensor; time and space registration and fusion of the multi-source data are realized by adopting an improved ICP algorithm combined with an adaptive weight distribution strategy; an electromagnetic-thermal-force multi-physical field coupling model is constructed based on the fused data, and a partial differential equation is solved through a physical information neural network; a multi-dimensional evaluation index system covering insulation, mechanical and thermal states is established, and abnormality detection and intelligent diagnosis are realized by using a graph convolution network. The application effectively improves data fusion precision, model prediction accuracy and fault diagnosis reliability, significantly improves the intelligent level of substation operation and maintenance, and provides technical support for safe and stable operation of a power grid.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology of power equipment, specifically involving a digital twin modeling and intelligent diagnosis method for substations based on multi-source data fusion. Background Technology

[0002] With the advancement of smart grid construction, substation equipment condition monitoring and fault diagnosis face new challenges. Traditional single-data-source monitoring methods have the following limitations: 1) Point cloud data only provides geometric information and lacks physical characteristics; 2) Thermal imaging data reflects surface temperature and is difficult to infer internal conditions; 3) Vibration data is only partially effective and cannot comprehensively assess the health status of equipment. Existing digital twin technologies are mostly based on single-physics-field modeling, making it difficult to accurately simulate the real behavior of substation equipment under the coupling of multiple physics fields. Especially in critical facilities such as 500kV GIS substations, the complex coupling effects of electromagnetic fields, temperature fields, and stress fields may lead to model prediction biases, affecting the accuracy of condition assessment. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a digital twin modeling and intelligent diagnosis method for substations based on multi-source data fusion. This method solves the limitations of traditional single-source data source monitoring and the inaccuracy of multi-physics coupling simulation in existing digital twin technologies, thereby improving the accuracy of substation equipment status assessment and the reliability of fault prediction, and enhancing the level of intelligent operation and maintenance of substations.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion includes the following steps:

[0006] Step 1: Data Acquisition: Multi-source data from the substation is acquired using a 3D laser scanner, an infrared thermal imager, and vibration sensors. This multi-source data includes point cloud data, thermal imaging data, and vibration spectrum data. Specifically, the 3D laser scanner acquires the geometric morphology information of the substation equipment, generating point cloud data; the infrared thermal imager captures the temperature distribution on the equipment surface, obtaining thermal imaging data; and the vibration sensors collect vibration signals during equipment operation, which are then processed to form vibration spectrum data.

[0007] Step 2: Establish a multi-source data spatiotemporal registration model and use an improved ICP algorithm to achieve data fusion. First, a unified spatiotemporal coordinate system is constructed, and multi-source data with different sampling rates and resolutions are mapped to this coordinate system to solve the data registration problem. The registration error function is shown in formula (1):

[0008] (1)

[0009] Where R is the rotation matrix, t is the translation vector, and w i These are weighting coefficients based on data credibility.

[0010] Multi-source data fusion employs an adaptive weight allocation strategy, with weight coefficient w. i Based on signal-to-noise ratio and measurement accuracy, the calculation formula is as follows:

[0011] (3)

[0012] SNR i Let ε be the signal-to-noise ratio of the i-th data source. i This represents measurement error.

[0013] Step 3: Construct a multiphysics coupling model of substation equipment based on fused data, including electromagnetic-thermal-mechanical multi-field coupling equations:

[0014] (2)

[0015] Where A is the magnetic vector potential, T is the temperature field, and u is the displacement field;

[0016] The multiphysics coupling solution employs a physics-based neural network, which embeds physical laws into the model training process, improving the model's physical consistency and prediction accuracy. The loss function is shown in equation (4):

[0017] (4)

[0018] Among them, L data For data fitting loss, L physics The residuals of the physical equations are used to achieve an accurate solution to the multiphysics coupling equations by minimizing this loss function.

[0019] Step 4: Establish an equipment condition assessment index system, which includes insulation condition indexes, mechanical condition indexes, and thermal condition indexes.

[0020] Insulation condition indicators: dielectric loss factor tanδ, partial discharge quantity;

[0021] Mechanical condition indicators: vibration characteristic frequency, displacement amplitude;

[0022] Thermal state indicators: temperature gradient, hot spot temperature.

[0023] Based on this state assessment index system, a deep learning-based anomaly detection model is used to achieve intelligent diagnosis. Anomaly detection employs a graph convolutional network (GCNN), where node features incorporate multi-physics data, and edge weights are determined based on device connectivity. By leveraging the GCNN, the relationships between various device components can be fully explored, improving the accuracy of anomaly detection and the precision of fault location.

[0024] This technology proposes a digital twin modeling and intelligent diagnosis method for substations based on multi-source data fusion, which has the following advantages and beneficial effects:

[0025] 1. This invention integrates multi-source data such as three-dimensional laser scanning, infrared thermal imaging, and vibration monitoring, which overcomes the limitations of traditional single-source data monitoring. It can comprehensively acquire the status information of substation equipment from multiple dimensions such as geometric shape, temperature distribution, and vibration characteristics, providing rich data support for equipment status assessment.

[0026] 2. A unified spatiotemporal coordinate system was established, and an improved ICP algorithm and adaptive weight allocation strategy were adopted to achieve multi-source data fusion. This effectively solved the registration problem of data with different sampling rates and resolutions, improved the accuracy of data fusion, and ensured the reliability and consistency of the fused data.

[0027] 3. An electromagnetic-thermal-mechanical multiphysics coupling model was constructed and solved using a neural network based on physical information. This model can accurately simulate the real operating behavior of substation equipment under the coupling of multiple physics fields, reduce model prediction bias, and improve the accuracy of equipment condition assessment.

[0028] 4. A multi-dimensional condition assessment index system covering insulation, mechanical and thermal conditions has been established. Combined with graph convolutional networks, anomaly detection and fault diagnosis are realized. This system can comprehensively and accurately assess the health status of equipment, promptly identify potential fault hazards, improve the reliability of fault prediction and the accuracy of fault location, provide strong technical support for the intelligent operation and maintenance of substations, help reduce operation and maintenance costs, and ensure the safe and stable operation of the power grid. Detailed Implementation

[0029] The present invention will be further described below with reference to the embodiments. It should be noted that these are merely examples and descriptions of the inventive concept. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the inventive concept or exceed the scope defined in the claims, they should all be considered to fall within the protection scope of the present invention.

[0030] The present invention will be further described in detail below with reference to specific embodiments.

[0031] This embodiment takes a transformer as the research object and uses the substation digital twin modeling and intelligent diagnosis method based on multi-source data fusion described in this invention to realize transformer condition assessment and fault early warning. The specific steps are as follows:

[0032] Step 1, Data Collection:

[0033] A transformer in a 500kV substation was selected as the monitoring object, and multi-source data was collected using the following equipment:

[0034] 3D laser scanner: A certain model of high-precision 3D laser scanner is used to scan the external structure of the transformer, collect point cloud data, and the scanning range covers the entire transformer body and its auxiliary components;

[0035] Infrared thermal imager: A high-resolution infrared thermal imager is selected to collect temperature data of key parts such as transformer tank, bushings, and windings under normal transformer operation, and to obtain thermal imaging data.

[0036] Vibration sensors: Vibration sensors are installed at different locations on the transformer casing (including the tank wall, core, winding ends, etc.) to collect vibration signals during transformer operation. The collected vibration signals are then processed using Fourier transform and other methods to obtain vibration spectrum data.

[0037] Step 2: Multi-source data fusion:

[0038] Establish a unified spatiotemporal coordinate system. With the center point of the transformer as the origin, establish a three-dimensional rectangular coordinate system and map the acquisition time and spatial location of point cloud data, thermal imaging data and vibration spectrum data to this coordinate system.

[0039] An improved ICP algorithm is used to register multi-source data. The registration error is calculated according to formula (1). The registration error is minimized by iteratively optimizing the rotation matrix R and the translation vector t.

[0040] Calculate the signal-to-noise ratio (SNR) SNRi and measurement error εi for each data source: For point cloud data, the SNR is obtained by calculating the ratio of signal strength to noise intensity, and the measurement error is determined based on the scanner's technical parameters and the actual measurement environment; for thermal imaging data, the SNR and measurement error are calculated based on the thermal imager's noise level and temperature measurement accuracy; for vibration spectrum data, the SNR is obtained by analyzing the amplitude distribution of the vibration signal and the background noise level, and the measurement error is determined based on the sensor's calibration data.

[0041] Calculate the weight coefficient wi of each data source according to formula (3), substitute the weight coefficient into the registration error function, complete the fusion of multi-source data, and obtain the unified data after fusion.

[0042] Step 3: Construction of a multiphysics coupling model:

[0043] An electromagnetic-thermal-mechanical multiphysics coupling model of a transformer is constructed based on fused data. The multiphysics coupling equation is shown in Equation (2).

[0044] A neural network based on physical information is constructed. The network structure includes an input layer, a hidden layer, and an output layer. The input layer is the fused multi-source data, and the output layer is the magnetic vector potential A, the temperature field T, and the displacement field u.

[0045] The loss function is determined as shown in Equation (4), where Ldata is obtained by calculating the mean square error between the network prediction and the actual measurement, and Lphysics is obtained by calculating the residual of the multiphysics coupling equation.

[0046] The gradient descent algorithm is used to train the neural network. The network parameters are continuously adjusted to minimize the loss function until the network converges, resulting in a well-trained multiphysics coupling model.

[0047] Based on the specific characteristics of the transformer, a thermal field model and a vibration propagation model are constructed:

[0048] 1. Thermal field model

[0049]

[0050] 2. Vibration Propagation Model

[0051]

[0052] 3. Anomaly detection based on attention mechanism

[0053]

[0054] Step 4, Intelligent Diagnosis:

[0055] Based on the operating characteristics and fault types of transformers, and in accordance with the condition assessment index system described in this invention, data such as dielectric loss factor tanδ, partial discharge, vibration characteristic frequency, displacement amplitude, temperature gradient, and hot spot temperature are collected.

[0056] A graph convolutional network is constructed, with each component of the transformer as a network node. The node features are multi-physics data (including magnetic vector potential, temperature field, displacement field data, and various state evaluation index data). The edge weights are determined based on the mechanical and electrical connections between the components of the transformer.

[0057] An attention mechanism is introduced to optimize the anomaly detection performance of graph convolutional networks. The attention mechanism model is as follows:

[0058]

[0059] The graph convolutional network is trained using historical fault data and normal operation data to obtain a trained anomaly detection model;

[0060] The real-time collected and fused data is input into the trained multiphysics coupling model to obtain the real-time multiphysics distribution of the transformer. Combined with the state assessment index data, it is input into the anomaly detection model to realize intelligent assessment of the transformer state and fault early warning.

[0061] By applying this embodiment, the operating status of the transformer can be monitored in real time and accurately, and problems such as insulation aging, mechanical failure, and overheating can be detected in a timely manner. This provides a scientific basis for the operation and maintenance of the transformer, effectively reduces the failure rate, and improves the power supply reliability of the substation.

[0062] The above is an exemplary description of the invention. Obviously, the specific implementation of the invention is not limited to the above-described manner. Any non-substantial improvement made using the inventive concept and technical solution of the invention, or the direct application of the inventive concept and technical solution to other situations without modification, is within the protection scope of the invention.

Claims

1. A method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion, characterized in that, Includes the following steps: Step 1: Collect multi-source data of the substation, including point cloud data, thermal imaging data and vibration spectrum data, using a 3D laser scanner, infrared thermal imager and vibration sensor; Step 2: Establish a multi-source data spatiotemporal registration model, and use an improved ICP algorithm to achieve data fusion. The registration error function is: （1） Where R is the rotation matrix, t is the translation vector, and w i These are weighting coefficients based on data credibility. Step 3: Construct a multiphysics coupling model of substation equipment based on fused data, including electromagnetic-thermal-mechanical multi-field coupling equations: （2） Where A is the magnetic vector potential, T is the temperature field, and u is the displacement field; Step 4: Establish an equipment status assessment index system and realize intelligent diagnosis based on a deep learning-based anomaly detection model.

2. The method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion according to claim 1, characterized in that, In step 2, the multi-source data fusion adopts an adaptive weight allocation strategy, with weight coefficient w. i Based on signal-to-noise ratio and measurement accuracy, the calculation formula is as follows: （3） SNR i Let ε be the signal-to-noise ratio of the i-th data source. i This represents measurement error.

3. The method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion according to claim 1, characterized in that... In step 3, the multiphysics coupling solution uses a neural network based on physical information, and the loss function is: （4） Among them, L data For data fitting loss, L physics This represents the residual of the physical equation.

4. The method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion according to claim 1, characterized in that... The status assessment index system in step 4 includes: Insulation condition indicators: dielectric loss factor tanδ, partial discharge quantity; Mechanical condition indicators: vibration characteristic frequency, displacement amplitude; Thermal state indicators: temperature gradient, hot spot temperature.

5. The method for digital twin modeling and intelligent diagnosis of substations based on multi-source data fusion according to claim 4, characterized in that, Anomaly detection employs a graph convolutional network, whose node features contain multiphysics data, and edge weights are determined based on device connectivity.

Images