Power system dynamic simulation and fault prediction method based on digital twinning

Abstract

The application belongs to the technical field of power systems, and discloses a power system dynamic simulation and fault prediction method based on digital twinning. The method comprises the following steps: S1: collecting multi-source heterogeneous data in a power system in real time based on a deployed sensor network, and preprocessing the multi-source heterogeneous data; wherein the multi-source heterogeneous data comprises power data and heat data; S2: calculating the preprocessed multi-source heterogeneous data based on a dynamic correlation model, and analyzing the thermal-electric relationship of power electronic equipment through the calculation result; S3: establishing a dynamic mapping function of the thermal-electric relationship in the digital twinning model. The application adopts the digital twinning model to fuse the thermal-electric relationship effect of the heat data characteristics of the power electronic equipment and the power data characteristics of the power grid, realizes dynamic simulation, improves simulation accuracy, and reduces the prediction deviation caused by model mismatch. At the same time, the virtual and real are linked, realizing the bidirectional feedback of the simulation result and fault prediction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power systems, and particularly relates to a power system dynamic simulation and fault prediction method based on digital twinning. BACKGROUND

[0002] A power system is composed of five links of power generation, power transmission, power transformation, power distribution and power consumption. Primary energy (such as coal, water energy, wind energy) is converted into electric energy in a power plant, transmitted through a high-voltage transmission line, and then distributed to end users after being stepped down by a substation. Its core function is to realize efficient conversion and reliable supply of electric energy from production to consumption, and to support industrial production, residential life and social operation. Power electronic devices play an important role in the power system, which is mainly used for power conversion and control of the power system, and can realize the core technology of efficient power conversion and intelligent control. The two are developed in coordination to promote the evolution of the power industry towards clean, intelligent and flexible. However, the temperature rise of power electronic devices will change their electrical characteristics (such as IGBT switching speed), cause transient process of power grid, and aggravate device heating, thereby causing faults of the power system.

[0003] With the continuous expansion of the scale and the increasing complexity of the power system, the traditional operation and maintenance management mode has been difficult to meet the real-time monitoring and fault prediction needs of the operation state of the power system. As a new technology, digital twinning technology realizes real-time monitoring, simulation and prediction of physical entities by constructing virtual models corresponding to the physical entities.

[0004] However, most of the current power system dynamic simulation and fault prediction methods do not consider the relationship effect of power electronic device heat data and power data, resulting in limitations of fault prediction. At the same time, the fault prediction is single, and the dynamic simulation and fault prediction are not combined through the digital twinning model, resulting in a decrease in the accuracy of prediction. SUMMARY

[0005] The application provides a power system dynamic simulation and fault prediction method based on digital twinning, aiming to solve the problem that the relationship effect of power electronic device heat data and power data is not considered in the prior art, resulting in limitations of fault prediction. At the same time, the fault prediction is single, and the dynamic simulation and fault prediction are not combined through the digital twinning model, resulting in a decrease in the accuracy of prediction.

[0006] The power system dynamic simulation and fault prediction method based on digital twinning, the method comprises:

[0007] S1: Real-time acquisition of multi-source heterogeneous data in the power system based on the deployed sensor network, and pre-processing of the multi-source heterogeneous data; wherein the multi-source heterogeneous data comprises power data and heat data;

[0008] S2: Based on the dynamic correlation model, the pre-processed multi-source heterogeneous data is calculated, and the thermal-electric relationship of the power electronic device is analyzed through the calculation result;

[0009] S3: The dynamic mapping function of the thermal-electric relationship is established in the digital twin model, and is mapped to the simulation grid node to obtain the temperature continuous value, and then the electromagnetic field and the thermal field are co-simulated dynamically;

[0010] S4: Based on the temperature continuous value and the dynamic simulation data, the preventive fault time of the device is judged and estimated, and then the historical data is mapped to the fault mode to determine and output the prediction result;

[0011] S5: According to the prediction result, the simulation parameters of the digital twin model are corrected to simulate the fault.

[0012] Further, the preprocessing includes signal denoising and filtering processing, standardization processing and data grading processing, which is used to convert and input the multi-source heterogeneous original data.

[0013] Further, the specific steps of S2 are as follows:

[0014] S2.1: Calculate the instantaneous power loss P of the electronic device through the dynamic correlation model and the multi-source heterogeneous data loss , which is used to represent the dynamic loss characteristics of the device in the switching process;

[0015] S2.2: The calculated instantaneous power loss P is input into the dynamic correlation model as a heat source loss , and the heat diffusion value is calculated to determine the conduction and diffusion efficiency of heat in the device;

[0016] S2.3: Calculate the on-resistance R based on the junction temperature in the multi-source heterogeneous data on ;

[0017] S2.4: The calculated on-resistance R is fed back to the dynamic correlation model in real time on , the on-resistance R is recalculated on , and the dynamic correlation model is updated.

[0018] Further, before the thermal-electric relationship mapping in S3, the thermal-electric relationship needs to be processed as follows:

[0019] 1) Data synchronization

[0020] The precision time protocol is deployed in the power system to provide a unified time reference for the power data and heat data collected by the sensor network;

[0021] 2) Spatial interpolation algorithm

[0022] The digital twin model is constructed based on a geometric model. CAD drawings or scanning data of the equipment are input into the geometric model to reflect the geometric characteristics of the equipment. The geometric model is divided into uniform virtual grids with a resolution of 1mm 3 The grid nodes need to cover the temperature measurement point positions and extend to the areas not directly measured.

[0023] Further, the spatial interpolation algorithm further comprises substituting the junction temperature of the temperature measurement point and its spatial coordinates (x, y, z) into a mapping function to update the parameters in real time; and then inputting the data into the interpolation algorithm to calculate the virtual temperature value of the virtual grid node through weighted summation.

[0024] Further, the virtual temperature value is used to fill in the blank areas between the discrete temperature measurement points to generate continuous temperature values and reflect the global heat state of the equipment.

[0025] Further, the dynamic mapping function is as follows:

[0026] R on (T j )=R0·[1+α(T j -T0)]

[0027] V th (T j )=V th0 ·[1+β(T j -T0)]

[0028] Wherein, R0 is a resistance reference value at room temperature, V th0 is a voltage parameter value at room temperature, T0 is a reference temperature, and α and β are temperature coefficients.

[0029] Further, the specific steps of S4 are as follows:

[0030] S4.1: Calculate the temperature continuous value at a certain time and the material safety limit value of the equipment to obtain a temperature rise rate value, and then determine whether the difference between the temperature rise rate value and the material safety limit value of the equipment is greater than 0;

[0031] S4.2: When it is determined that the equipment enters a fault risk state, calculate the preventive fault time t critical of the equipment to reach a critical state; the calculation formula is as follows:

[0032]

[0033] Wherein, T max -T j is the temperature difference between the current junction temperature and the material safety limit value;

[0034] S4.3: Match the current junction temperature value with the temperature range in the historical database to determine the specific fault mode and its probability of occurrence, and output the prediction results.

[0035] Furthermore, in step S4.1, if it is determined that the junction temperature continuously exceeds the material safety limit and the heating rate is greater than 0, then it is determined that the equipment is at risk of failure and proceeds to the next step; otherwise, monitoring continues, and the current temperature, heating rate, and timestamp are stored in the historical database.

[0036] Compared with the prior art, this application has at least the following beneficial effects:

[0037] Based on further analysis and research into existing technical problems, this application employs a digital twin model to integrate the thermal data characteristics of power electronic equipment with the thermal-electrical relationship effects of the power data characteristics of the power grid. Driven by real-time data from the digital twin model, dynamic simulation is achieved, improving simulation accuracy and reducing prediction bias caused by model mismatch. Simultaneously, virtual-real linkage enables bidirectional feedback between simulation results and fault prediction, forming a closed loop of "prediction-simulation-optimization" that makes dynamic adjustments more closely aligned with actual operating conditions. Attached Figure Description

[0038] Fig. 1 A flowchart illustrating a digital twin-based dynamic simulation and fault prediction method for power systems provided in one embodiment of this application;

[0039] Fig. 2 The flowchart of S2 in the method for dynamic simulation and fault prediction of power system based on digital twin provided in one embodiment of this application. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0041] like Figs. 1-2 As shown, the power system dynamic simulation and fault prediction method based on digital twins provided in this application includes the following steps:

[0042] S1: Real-time acquisition of multi-source heterogeneous data from the power system based on a deployed sensor network, followed by preprocessing of this data. The sensor network includes high-frequency voltage sensors, Hall current sensors, NTC thermistors, and temperature sensors. High-frequency voltage sensors and Hall current sensors monitor the power system's electrical data and capture high-frequency transient signals; temperature sensors and NTC thermistors monitor the thermal data of equipment in the power system, directly outputting electrical signals related to junction temperature, covering temperature changes in the equipment. The multi-source heterogeneous data includes both electrical and thermal data.

[0043] The preprocessing includes signal denoising and filtering, standardization and data classification, which is used to convert multi-source heterogeneous raw data into high-quality and standardized analysis input, achieving the balance between real-time and data storage, and facilitating the use of data in different periods. The specific content is as follows:

[0044] 1) Signal denoising and filtering

[0045] The raw data collected by the sensor may contain high-frequency noise (such as electromagnetic interference) or low-frequency drift (such as the slow offset of the temperature sensor).

[0046] Among them, high-frequency noise filtering: low-pass filter (such as cut-off frequency 10kHz) is used for power data signal to retain the effective frequency band;

[0047] Low-frequency drift correction: sliding window mean filter (window length 1-2 seconds) is used for temperature signal to eliminate slow temperature drift;

[0048] Outlier rejection: by setting a threshold range (such as voltage amplitude ±10%), abnormal data points caused by sudden interference are removed.

[0049] 2) Standardization

[0050] Power data conversion: the output voltage of the Hall current sensor is converted to current value (such as 1V=100A) through the calibration coefficient;

[0051] Temperature conversion: the resistance value of NTC thermistor is converted to temperature value through Steinhart-Hart equation;

[0052] Standardization: all data are normalized (such as power data is normalized to ±1p.u. and temperature is normalized to 0-1 range), which eliminates the influence of dimension difference on subsequent analysis.

[0053] 3) Data classification

[0054] Data classification is used to divide multi-source heterogeneous data into real-time level and historical level; among them, key features (such as junction temperature, current) are directly used for fault judgment; the rest of multi-source heterogeneous data is stored in historical database (such as time series database InfluxDB) for subsequent deep analysis and long-term trend mining. The historical database also stores the multi-dimensional thermal state data of the device under different working conditions, including junction temperature range, temperature rise rate, environmental temperature, load current and other parameters, as well as the corresponding fault mode (such as heat dissipation failure, overheat protection trigger, material aging) and its occurrence probability statistics. It provides a reference basis for matching subsequent prediction results.

[0055] S2: Based on the dynamic correlation model, the pre-processed multi-source heterogeneous data is calculated, and the thermal-electric relationship of power electronic equipment is analyzed through the calculation results. This step is the basic step of power system dynamic simulation and fault prediction. Through the bidirectional correlation of electricity and heat, the dynamic correlation model can accurately describe the influence of temperature on the electrical characteristics of electronic equipment, providing key mathematical description and input parameters for subsequent real-time monitoring, collaborative simulation and fault prediction, ensuring the high precision and reliability of the entire system. The specific steps are as follows:

[0056] S2.1: Calculate the instantaneous power loss P of electronic equipment through the dynamic correlation model and multi-source heterogeneous data loss , which is used to represent the dynamic loss characteristics of the device in the switching process, and provides key input parameters for subsequent thermal-electric relationship analysis. The calculation formula is as follows:

[0057] P loss =f(V ce ,I c , t sw )

[0058] Where V ce is the voltage, I c is the current, and t sw is the device switching time.

[0059] S2.2: Input the calculated instantaneous power loss P loss into the dynamic correlation model as a heat source, calculate the heat diffusion value, and determine the conduction and diffusion efficiency of heat in the device. The calculation formula is as follows:

[0060]

[0061] Where k is the thermal conductivity of the material, p is the density of the material, c p is the specific heat capacity, and T is the temperature.

[0062] S2.3: Calculate the on-resistance R on based on the junction temperature in the multi-source heterogeneous data to determine the influence of temperature on the electrical characteristics of the device. Where the increase of junction temperature T j will cause the carrier mobility of the semiconductor to decrease, and the on-resistance R on will increase exponentially. The calculation formula is as follows:

[0063]

[0064] Where E a is the material characteristic coefficient of the device, and R0 is the on-resistance at the reference temperature.

[0065] S2.4: Calculate the on-resistance R onReal-time feedback into the dynamic correlation model, recalculate the on-resistance R on , update the dynamic correlation model, ensure that the model always reflects the temperature impact under actual working conditions, so as to dynamically reflect the impact of temperature change on device behavior.

[0066] For components with large heat capacity (heat capacity C > 1000 J / ℃), introduce thermal time constant compensation mechanism, further subdivide time into multiple substeps, calculate thermal conduction process separately in each substep, improve the accuracy of thermal field calculation. Solve the problem of response lag caused by thermal inertia.

[0067] S3: Establish a dynamic mapping function of thermal-electric relationship in the digital twin model, and map it to the simulation grid nodes to get continuous temperature values, and then perform collaborative dynamic simulation of electromagnetic field and thermal field to realize high-precision simulation of thermal-electric coupling, and dynamically correct electrical parameters based on real-time junction temperature, accurately quantify the impact of temperature on power grid state, and provide reliable support for power system fault prediction. The specific content is as follows:

[0068] 1) Data synchronization

[0069] Precision time protocol (PTP) is deployed in the power system to provide a unified time reference for power data and heat data collected by high-frequency voltage sensors and Hall current sensors, as well as NTC thermistors and temperature sensors. Through synchronized clocks, ensure that the sampling timestamps of the two types of data are accurately aligned, with an error of microseconds.

[0070] Timestamp matching: Through PTP protocol, map the timestamps of power data and heat data of the two types of data to the same time axis, realize cross-scale synchronization.

[0071] For example, each frame of power data (50 μs) and each frame of heat data (1 ms) are aligned through interpolation or segmentation, ensuring timing consistency and achieving cross-scale time alignment.

[0072] 2) Spatial interpolation algorithm

[0073] The digital twin model is based on a geometric model, which inputs the CAD drawings or scanned data of the device into the geometric model to reflect the geometric characteristics of the device (such as position, heat dissipation channel, material thickness). The geometric model is divided into uniform virtual grids with a resolution of 1 mm3. The grid nodes need to cover the temperature measurement points and extend to the areas not directly measured (such as the inside of the chip and the surface of the heat sink).

[0074] Map the discrete temperature measurement points in the device to the corresponding nodes of the virtual grid through a dynamic mapping function to ensure consistent geometric positions. The dynamic mapping function is as follows:

[0075] R on (Tj ) = R0·[1 + a(T j -T0)]

[0076] V th (T j ) = V th0 ·[1 + b(T j -T0)]

[0077] Wherein, R0 is the resistance reference value at room temperature, V th0 is the voltage parameter value at room temperature, T0 is the reference temperature, and a and b are both temperature coefficients.

[0078] The junction temperature of the temperature measurement point and its spatial coordinates (x, y, z) are substituted into the mapping function, and the parameters are updated in real time. Then the data is input into the interpolation algorithm, and the virtual temperature value of the virtual grid node is calculated by weighted summation. The virtual temperature value is used to fill in the blank area between the discrete temperature measurement points, generate continuous temperature continuous value, and reflect the global heat state of the device. The temperature rise trajectory can be captured in real time, and reliable support is provided for fault prediction. The calculation formula is as follows:

[0079]

[0080] Wherein, w i is the weight assigned by the digital twin model, N is the number of adjacent temperature measurement points, and T i is the junction temperature of the temperature measurement point.

[0081] The temperature measurement points of the device are usually distributed discretely, and cannot directly reflect the full-field temperature distribution. The digital twin model converts the discrete temperature measurement data into continuous temperature continuous value through spatial interpolation algorithm.

[0082] S4: Based on the temperature continuous value and dynamic simulation data, the preventive fault time of the device is estimated, and the historical data is mapped to determine and output the prediction result. By comprehensively considering the junction temperature and its change rate, and combining the historical failure data, the probability of fault occurrence can be more accurately predicted, the accuracy of fault probability prediction is improved, and more reliable basis is provided for the maintenance and operation of the device. The specific steps are as follows:

[0083] S4.1: Calculate the temperature continuous value at a certain time with the material safety limit value of the device to obtain the temperature rise rate value, and then judge whether the difference between the temperature rise rate value and the material safety limit value of the device is greater than 0. If the junction temperature continuously exceeds the material safety limit value and the temperature rise rate value is greater than 0, it is determined that the device enters the risk of failure, and the next step is entered. Otherwise, continue to monitor, and store the current temperature, temperature rise rate value and time stamp in the historical database for long-term trend analysis, and pre-warning degradation to prompt potential risks. When the temperature falls below the safety limit value and the derivative value tends to 0, the warning is terminated and the normal monitoring mode is returned. The calculation formula is as follows:

[0084]

[0085] where Δt is the sampling period, T j (t) is the junction temperature value at the previous time t, T j (t+1) is the junction temperature value at the next time.

[0086] S4.2: When the device is determined to be at risk of failure, calculate the preventive failure time t critical This process combines the temperature difference between the current junction temperature and the material safety limit, as well as the temperature rise rate value, to quantify the time window for failure occurrence, providing a basis for subsequent emergency decision-making. By dynamically tracking the temperature rise rate, the system can calculate the preventive failure time in real time. The calculation formula is as follows:

[0087]

[0088] where T max -T j is the temperature difference between the current junction temperature and the material safety limit.

[0089] S4.3: Match the current junction temperature value with the temperature interval in the historical database to determine the specific failure mode and occurrence probability, and output the prediction result. Match the current detected thermal state (junction temperature range) with the failure data in the historical database, and determine the specific failure mode and occurrence probability based on the matching result.

[0090] For example, when the temperature continuous value is between 150-160℃ at a certain time, it is determined to be solder layer fatigue, with an occurrence probability of 80%; when the temperature continuous value is >160℃ at a certain time, it is determined to be chip burnout, with an occurrence probability >95%. This helps to take measures in advance to avoid further development of the failure.

[0091] S5: Modify the digital twin model simulation parameters to simulate the failure based on the prediction result. Upgrade from a fixed threshold strategy to a dynamic optimization strategy, which can flexibly adjust the control strategy according to the real-time state of the device and the failure prediction result, so that the device can maintain the best operating state under different working conditions. The specific content is as follows:

[0092] 1) Digital twin model modification

[0093] Input the prediction result into the digital twin model to update the model parameters. Modify the simulation parameters in the digital twin model according to the prediction result and run it. This facilitates the adjustment of model boundary conditions based on the prediction result.

[0094] For example: If the predicted junction temperature is close to the material limit, reduce the power load parameter; if the remaining life is insufficient, accelerate the simulation process of the material aging model.

[0095] Run the modified digital twin model to simulate the operation of the device under different fault conditions, obtain relevant performance data and indicators. Used to pre-empt the development process of the fault and verify potential risks.

[0096] 2) Indicator verification

[0097] The values of power grid safety, device safety, and power supply continuity are preset respectively. Test the performance of the candidate control strategy under extreme working conditions in the digital twin model, observe the changes in the indicators, and verify its feasibility through the above three indicators;

[0098] For example: Power grid safety: Ensure that the voltage fluctuation is < ± 7%, the frequency deviation is < 0.6 Hz, and the device operation will not adversely affect the power grid.

[0099] Device safety: Monitor the junction temperature to ensure that the junction temperature is < the material limit to prevent the device from being damaged due to overheating.

[0100] Power supply continuity: Control the load loss < 15% to ensure the continuity and stability of power supply.

[0101] 3) Dynamic optimization

[0102] Based on the verification results, the digital twin model can mark the data that fails the verification as "disabled" and store the data that successfully passes the verification in the database. Through reinforcement learning (RL), the decision logic is continuously optimized, the model is updated, and the power grid safety and device safety are better protected.

[0103] In the above digital twin-based power system dynamic simulation and fault prediction method, the digital twin model is used to integrate the thermal data characteristics of power electronic devices and the thermal-electric relationship effects of power grid power data characteristics, and based on the real-time data driving of the digital twin model, dynamic simulation is realized, the simulation accuracy is improved, and the prediction deviation caused by model mismatch is reduced. At the same time, virtual and real linkage is realized, bidirectional feedback of simulation results and fault prediction is realized, forming a "prediction-simulation-optimization" closed loop to make dynamic adjustment more in line with actual working conditions.

[0104] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present disclosure.

Claims

1. A power system dynamic simulation and fault prediction method based on digital twinning, characterized in that, The method comprises: S1: Real-time collection of multi-source heterogeneous data in a power system based on a deployed sensor network, and preprocessing of the multi-source heterogeneous data; wherein the multi-source heterogeneous data comprises power data and heat data; S2: Calculation of the preprocessed multi-source heterogeneous data based on a dynamic correlation model, analysis of the thermal-electric relationship of power electronic equipment through the calculation results; The specific steps of S2 are as follows: S2.1: Calculate the instantaneous power loss of electronic equipment by dynamic correlation model and multi-source heterogeneous data , for representing the dynamic loss characteristics of the device during the switching process; S2.2: Calculate the instantaneous power loss As a heat source input dynamic correlation model, the heat diffusion value is calculated to determine the conduction and diffusion efficiency of heat inside the device; S2.3: Calculate on-resistance based on junction temperature in multi-source heterogeneous data ; S2.4: Calculate the on-resistance Real-time feedback to the dynamic correlation model, recalculate the on-resistance , update the dynamic correlation model; S3: Establishing a dynamic mapping function of the thermal-electric relationship in a digital twin model, and mapping it to a simulation grid node to obtain a temperature continuous value, and then performing a collaborative dynamic simulation of the electromagnetic field and the thermal field; S4: Estimating the fault prevention time of the equipment based on the temperature continuous value and the dynamic simulation data, and then mapping the fault mode combined with historical data to determine and output a prediction result; S5: Correcting the simulation parameters of the digital twin model to simulate the fault according to the prediction result.

2. The digital-twin-based power system dynamic simulation and fault prediction method according to claim 1, characterized in that, The preprocessing comprises signal denoising and filtering processing, standardization processing, and data grading processing, which are used to convert and input the multi-source heterogeneous original data. 3.The digital-twin-based power system dynamic simulation and fault prediction method of claim 1, wherein, Before the thermal-electric relationship mapping in S3, the following processing of the thermal-electric relationship is required: 1) Data synchronization The precision time protocol is deployed in the power system to provide a unified time reference for the power data and heat data collected by the sensor network; 2) Spatial interpolation algorithm The digital twin model is constructed based on a geometric model, the CAD drawing or scanning data of the equipment is input into the geometric model to reflect the geometric characteristics of the equipment; the geometric model is divided into uniform virtual grids with a resolution of 1mm³, and the grid nodes need to cover the temperature measurement point positions and extend to the areas not directly measured.

4. The digital-twin-based power system dynamic simulation and fault prediction method according to claim 3, characterized in that, The space interpolation algorithm further comprises the junction temperature of the temperature measurement point and its spatial coordinates Substitute the mapping function, update the parameters in real time; then input the data into the interpolation algorithm, and calculate the virtual temperature value of the virtual grid node through weighted summation.

5. The digital-twin-based power system dynamic simulation and fault prediction method according to claim 4, characterized in that, The virtual temperature value is used to fill in the blank areas between the discrete temperature measurement points to generate a continuous temperature continuous value, reflecting the global heat condition of the equipment.

6. The digital-twin-based power system dynamic simulation and fault prediction method of claim 1, wherein, The dynamic mapping function is as follows: ； wherein, R0 is a resistance reference value at normal temperature, V0 is a voltage parameter value at normal temperature, T0 is a reference temperature, and are temperature coefficients.

7. The digital-twin-based power system dynamic simulation and fault prediction method of claim 1, wherein, The specific steps of S4 are as follows: S4.1: Calculate the temperature continuous value at a certain time with the material safety limit value of the equipment to obtain a temperature rise rate value, and then determine whether the difference between the temperature rise rate value and the material safety limit value of the equipment is greater than 0; S4.2: When the device is determined to be at risk of failure, the preventive failure time for the device to reach a critical state is calculated ; the calculation formula is as follows: ； wherein, is the difference between the current junction temperature and the material safety limit; S4.3: Match the current junction temperature value with the temperature interval in the historical database to determine the specific fault mode and occurrence probability, and output the prediction result.

8. The digital-twin-based power system dynamic simulation and fault prediction method according to claim 7, characterized in that, In S4.1, if the junction temperature continuously exceeds the material safety limit value and the temperature rise rate value is greater than 0, it is determined that the equipment is at risk of failure, and the next step is entered; Otherwise, continue to monitor and store the current temperature, temperature rise rate value, and time stamp in the historical database.

Images