A method and system for performance evaluation of a current transformer

Abstract

This invention discloses a performance evaluation method and system for current transformers, belonging to the field of current transformer performance evaluation technology. It aims to solve the technical problem that existing technologies, due to their reliance on single monitoring methods, cannot accurately reflect the true operating state of current transformers under multiple physical fields. The method includes an intelligent evaluation module and a predictive application module. The intelligent evaluation module comprises a high-fidelity physical model unit, a data-driven calibration unit, and a dynamic evaluation unit. This invention constructs an electromagnetic-thermal coupled physical model by integrating the mutual inductance electromotive force formula and Fourier's law of heat conduction. This model comprehensively captures the true operating state of the equipment under the coupling of multiple physical fields. It correlates and weights the monitoring information of different physical quantities to generate a comprehensive health index that fully reflects the true state of the current transformer under the coupling of electrical, thermal, and mechanical multiple physical fields, thus addressing the problem of the current transformer's inability to reflect its true operating state under multiple physical fields.

Description

Technical Field

[0001] This invention relates to the field of current transformer performance evaluation technology, and more specifically, to a method and system for evaluating the performance of current transformers. Background Technology

[0002] Current transformers, as core metering and protection devices in power systems, are widely used in various stages of power generation, transmission, and distribution. The stability of their operation and the accuracy of their evaluation results directly affect the safe and reliable operation of the power system. With the development of power systems towards intelligence, high voltage, and large capacity, the operating conditions of current transformers are becoming increasingly complex, facing the combined effects of electromagnetic interference, temperature fluctuations, mechanical vibration, insulation aging, and other factors. This places higher demands on the comprehensiveness, accuracy, and dynamism of their performance evaluation.

[0003] With the development of smart grids and the integration of new energy sources, the operating environment of current transformers is becoming increasingly complex, and the requirements for real-time performance, accuracy, and predictability are constantly increasing. Existing technologies mostly focus on the real-time acquisition of electrical quantities such as secondary current and voltage, which can only reflect the local state of the equipment's electromagnetic characteristics, neglecting the additional impacts brought by the integration of new energy sources. These include non-electrical parameter changes such as additional winding heating caused by harmonic currents, mechanical vibration of the iron core due to load impacts, and accelerated insulation aging caused by long-term high temperatures. These multi-physical field factors are coupled with each other and are precisely the core causes of equipment performance degradation and failure. Single electrical quantity monitoring cannot comprehensively capture the true operating state of the equipment. Therefore, we propose a performance evaluation method and system for current transformers. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art, adapt to practical needs, and provide a performance evaluation method and system for current transformers, so as to solve the technical problem that the current technology is difficult to reflect the real operating status of current transformers under multiple physical fields due to the single monitoring method.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a performance evaluation system for a current transformer, comprising an intelligent evaluation module and a predictive application module; the intelligent evaluation module includes a high-fidelity physical model unit, a data-driven calibration unit, and a dynamic evaluation unit; the high-fidelity physical model unit is used to construct an initial digital twin characterizing the electromagnetic, thermal, and mechanical properties of the current transformer being evaluated; the data-driven calibration unit is used to dynamically calibrate the model parameters of the initial digital twin; the dynamic evaluation unit is used to calculate the real-time error and / or comprehensive health index of the current transformer being evaluated based on the calibrated digital twin; the predictive application module includes a performance degradation prediction unit; the performance degradation prediction unit is used to predict the performance degradation trend and remaining lifespan based on historical and current dynamic performance evaluation results.

[0006] Preferably, it also includes a multimodal data acquisition module, which includes an electrical parameter monitoring unit and a physical state monitoring unit; The electrical parameter monitoring unit is configured to acquire a reference electrical signal through a reference-level current transformer and monitor the current. ,Voltage Power factor parameter; The physical state monitoring unit is configured to collect data including temperature monitoring. Vibration acceleration Insulation resistance At least one non-electrical parameter in it.

[0007] Preferably, the high-fidelity physical model unit is constructed based on the electromagnetic-thermal coupling principle, and includes an electromagnetic induction part and a heat conduction part: The electromagnetic induction part uses the formula for mutual inductance electromotive force: ,in Indicates mutual inductance electromotive force; Represents the mutual inductance coefficient; This represents the rate of change of the primary current; The heat conduction section adopts Fourier's law of heat conduction: ,in Indicates heat flux density; Indicates thermal conductivity; This represents the temperature gradient.

[0008] Preferably, the data-driven calibration unit uses the least squares method to optimize model parameters, including: Define the loss function: , in, This represents the loss function, used to measure the deviation between the model's predicted values ​​and the actual values; Indicates the first The actual data values ​​collected vary in units depending on the monitoring parameters. Indicates the first The data values ​​predicted by the model; Indicates the number of data samples; Minimize the loss function using the gradient descent algorithm: , in, The units for representing model parameters vary depending on the parameter type. This represents the learning rate of the gradient descent algorithm. It is dimensionless and usually a decimal between 0 and 1. This indicates the loss function in the parameters The gradient at the point is used to match the model with the actual device.

[0009] Preferably, the prediction application module further includes an early warning generation unit, which generates early warning signals through the following logic: , in, This indicates a warning signal; 1 means a warning is generated, and 0 means no warning is generated. This represents the predicted performance metrics, such as error rate and temperature. The threshold represents the performance metric.

[0010] Preferably, the intelligent evaluation module is communicatively connected to the multimodal data acquisition module, and is used to establish and dynamically calibrate a digital twin of the current transformer being evaluated based on the electrical and non-electrical parameters, so as to output dynamic performance evaluation results.

[0011] Preferably, the prediction application module is communicatively connected to the intelligent evaluation module, and is used to perform trend prediction based on the dynamic performance evaluation results and generate predictive maintenance information.

[0012] A method for evaluating the performance of a current transformer, the method comprising the following steps: S100. The electrical parameters of the current transformer being evaluated and at least one non-electrical parameter reflecting its physical state are simultaneously acquired through the multi-modal data acquisition module. S200: The initial digital twin is constructed through the high-fidelity physical model unit of the intelligent evaluation module. Then, the model parameters of the initial digital twin are dynamically calibrated based on the collected parameters by the data-driven calibration unit. Finally, the dynamic evaluation unit outputs the dynamic performance evaluation results based on the calibrated digital twin. S300: The performance degradation prediction unit of the application module predicts the performance degradation trend and remaining lifespan based on historical and current dynamic performance evaluation results. At the same time, the early warning generation unit generates early warning information when the performance index is lower than the preset threshold or when a failure is predicted.

[0013] Preferably, the dynamic calibration of model parameters in step S200 specifically includes: Real-time acquisition of multimodal data as input, and continuous adjustment of mutual inductance coefficient using gradient descent algorithm. Thermal conductivity parameter; Establish a loss function to measure the deviation between the model's predicted values ​​and the actual values, thereby achieving a match between the model and the actual equipment; An adaptive learning rate adjustment strategy is adopted to ensure the stability and efficiency of parameter convergence.

[0014] Preferably, the performance degradation prediction in step S300 is implemented using an LSTM model, including: input gate: , in, Indicates the LSTM input gate at The output at any given time is a value between 0 and 1. This represents the weight matrix of the input gate; express The hidden state at any given moment; express Input data at any given time, such as historical error rate, temperature, etc.; This represents the bias term of the input gate; express The activation function, its mathematical expression is: ; Forgotten Gate: , in This indicates that the LSTM forget gate is in The output at any given time is a value between 0 and 1. A value close to 0 indicates that the information at the corresponding location will be forgotten; a value close to 1 indicates that it will be retained. The weight matrix representing the forget gate; The bias term representing the forget gate; Output gate: , in, This indicates that the LSTM output gate is in The output at any given time is a value between 0 and 1. This represents the weight matrix of the output gate; This represents the bias term of the output gate.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs an electromagnetic-thermal coupled physical model by integrating the mutual inductance electromotive force formula and Fourier's law of heat conduction. This model simultaneously covers both the electrical and non-electrical characteristics of the current transformer, comprehensively capturing the actual operating state of the equipment under the coupling of multiple physical fields. It correlates and weights the monitoring information of different physical quantities to generate a comprehensive health index that fully reflects the true state of the current transformer under the coupling of electrical, thermal, and mechanical physical fields. This makes the evaluation results no longer limited to electromagnetic characteristics, effectively capturing performance degradation caused by non-electrical factors such as harmonic heating, load impact vibration, and insulation aging. It solves the problem of one-sided evaluation based on a single electrical quantity and addresses the difficulty of existing technologies in reflecting the true operating state of the current transformer under multiple physical fields due to the limited monitoring capabilities of single quantities.

[0016] 2. This invention also constructs a digital twin using a high-fidelity physical model, providing a theoretical foundation for evaluation. The data-driven calibration unit uses the least squares method to define the loss function and employs a gradient descent algorithm to dynamically and online calibrate the key parameters of the digital twin. This hybrid driving mode of physical model and real-time data greatly reduces the deviation between the model and the actual equipment, making the evaluation results closer to real operating conditions, significantly improving accuracy, and effectively reducing the high false alarm rate caused by simply relying on threshold comparisons.

[0017] 3. This invention also utilizes an LSTM model to perform deep learning on historical and real-time comprehensive health indices, error rates, and other performance indicators through a performance degradation prediction unit in the application module. This model can effectively capture the long-term dependencies and trends of performance parameters, thereby accurately predicting performance degradation trajectories and estimating remaining service life. Combined with a deviation analysis algorithm, the difference between the model's predicted values ​​and the actual monitored values ​​is compared. When an anomaly occurs, the system can analyze which parameter in the multiphysics field deviates first, thereby performing root cause analysis and assisting in determining whether the fault originates from the core, winding, or insulation. This greatly improves the efficiency and accuracy of operation and maintenance decisions, upgrading the operation and maintenance approach from passive response to proactive early warning. It provides a direct and quantitative decision-making basis for predictive maintenance and effectively avoids sudden failures. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] like Figures 1 to 2 As shown, the present invention relates to a performance evaluation method and system for a current transformer, comprising a multi-mode data acquisition module, an intelligent evaluation module, and a predictive application module; The multimodal data acquisition module is used to simultaneously acquire the electrical parameters of the current transformer being evaluated and at least one non-electrical parameter reflecting its physical state. Through the electrical parameter monitoring unit and the physical state monitoring unit, it collects multi-dimensional operating data of the current transformer, providing comprehensive input for subsequent evaluation. The multimodal data acquisition module includes an electrical parameter monitoring unit and a physical state monitoring unit; The electrical parameter monitoring unit is configured to acquire a reference electrical signal through a reference-level current transformer, wherein the current of the current transformer is monitored. This indicates the current value or voltage on the primary or secondary side of the current transformer. This indicates the voltage value and power factor. It is the ratio of active power to apparent power, and other electrical parameters, which are sampled in real time, for example, by current sensors and voltage sensors.

[0020] The physical state monitoring unit is configured to collect at least one non-electrical parameter, including temperature, vibration, sound waves, or insulation status, wherein temperature is monitored. Vibration acceleration Insulation resistance Data is collected using physical conditions such as temperature sensors, vibration sensors, and insulation monitoring devices. Multi-dimensional data collection ensures the richness of subsequent model inputs, laying a data foundation for accurate assessment and prediction.

[0021] The intelligent evaluation module is communicatively connected to the multimodal data acquisition module and is used to establish and dynamically calibrate a digital twin of the current transformer under evaluation based on the electrical and non-electrical parameters, so as to output dynamic performance evaluation results; the intelligent evaluation module includes a high-fidelity physical model unit, a data-driven calibration unit, and a dynamic evaluation unit; The high-fidelity physical model unit is used to construct an initial digital twin characterizing the electromagnetic, thermal, and mechanical properties of the current transformer being evaluated. Based on the principles of electromagnetic induction and heat conduction, an electromagnetic-thermal coupling physical model of a current transformer is constructed, as shown in the following formula: The electromagnetic induction part uses the formula for mutual inductance electromotive force: ,in Indicates mutual inductance electromotive force; Represents the mutual inductance coefficient; It represents the rate of change of the primary current; it describes the electromagnetic conversion characteristics of the current transformer.

[0022] The heat conduction part uses Fourier's law of heat conduction: ,in Indicates heat flux density; Indicates thermal conductivity; It represents the temperature gradient and describes the heat transfer and temperature distribution of the current transformer.

[0023] Multiphysics coupling: This involves coupling electromagnetic losses (such as copper losses) with electromagnetic fields. ,in Indicates copper loss; Indicates winding resistance and iron loss (representing iron loss) is used as the heat source for heat conduction to realize electromagnetic-thermal coupling simulation, simulating the characteristics of current transformers under electro-thermal multi-physics fields.

[0024] The data-driven calibration unit is used to dynamically calibrate the model parameters of the initial digital twin; To address the discrepancy between the physical model and the actual equipment, a data-driven model calibration method is employed. Taking the least squares method as an example, the model parameters are optimized: Define the loss function between the model output and the actual data: , in, This represents the loss function, used to measure the deviation between the model's predicted values ​​and the actual values; Indicates the first The actual data values ​​collected vary in units depending on the monitoring parameters. Indicates the first The data values ​​predicted by the model; Indicates the number of data samples.

[0025] Minimize the loss function using the gradient descent algorithm and adjust model parameters (such as the mutual inductance coefficient). Thermal conductivity (etc.), the formula is: , in, The units for representing model parameters vary depending on the parameter type. This represents the learning rate of the gradient descent algorithm. It is dimensionless and usually a decimal between 0 and 1. This indicates the loss function in the parameters The gradient at the point is used to match the model with the actual device.

[0026] The dynamic evaluation unit is used to calculate the real-time error and / or comprehensive health index of the current transformer being evaluated based on the calibrated digital twin. Based on the calibrated physical model, real-time acquired multimodal data is input to calculate key performance indicators, such as: Error rate: , in, It represents the error rate, which measures the degree of deviation between the model's calculated values ​​and the measured values; This represents the measured current value; This represents the current value calculated by the model; This indicates the rated current of the current transformer.

[0027] Thermal stability index: The temperature field distribution is calculated based on the heat conduction model to determine whether it exceeds the heat resistance threshold of the insulating material.

[0028] The prediction application module is communicatively connected to the intelligent evaluation module and is used to perform trend prediction based on the dynamic performance evaluation results and generate predictive maintenance information. The prediction application module includes an early warning generation unit and a performance degradation prediction unit; The early warning generation unit is used to generate early warning information when the performance index is lower than a preset threshold or when a failure is predicted to occur. Set thresholds for performance metrics, such as error rate thresholds. Temperature threshold When the predicted performance metric exceeds the threshold, an early warning signal is generated. , in, This indicates a warning signal; 1 means a warning is generated, and 0 means no warning is generated. This represents the predicted performance metrics, such as error rate and temperature. The threshold represents the performance metric.

[0029] The performance degradation prediction unit is used to predict performance degradation trends and remaining lifetime based on historical and current dynamic performance evaluation results.

[0030] An LSTM model is used to learn historical performance metrics such as error rate and temperature to predict future performance degradation trends. The core unit of the LSTM is calculated as follows: Input Gate: , in, Indicates the LSTM input gate at The output at any given time is a value between 0 and 1. This represents the weight matrix of the input gate; express The hidden state at any given moment; express Input data at any given time, such as historical error rate, temperature, etc.; This represents the bias term of the input gate; express The activation function, its mathematical expression is: The output range of this function is It is often used in the gating unit of neural network LSTM to achieve weight control of information retention or forgetting by mapping the input to a value between 0 and 1. Forgotten Gate: , in This indicates that the LSTM forget gate is in The output at any given time is a value between 0 and 1. A value close to 0 indicates that the information at the corresponding location will be forgotten; a value close to 1 indicates that it will be retained. The weight matrix representing the forget gate; The bias term representing the forget gate; Cell status update: , in, express Cellular state at any given moment; The weight matrix representing cell state updates; Bias terms representing cell state updates; This represents the hyperbolic tangent activation function, with an output value between -1 and 1; Output gate: , in, This indicates that the LSTM output gate is in The output at any given time is a value between 0 and 1. This represents the weight matrix of the output gate; This represents the bias term of the output gate; Hidden state: , in, This represents the hidden state at time t. By training an LSTM model, the predicted performance metrics for future times are output, thus yielding the performance degradation curve.

[0031] A method for evaluating the performance of a current transformer, the method comprising the following steps: S100. The electrical parameters of the current transformer being evaluated and at least one non-electrical parameter reflecting its physical state are simultaneously acquired through the multi-modal data acquisition module. S200: The initial digital twin is constructed through the high-fidelity physical model unit of the intelligent evaluation module. Then, the model parameters of the initial digital twin are dynamically calibrated based on the collected parameters by the data-driven calibration unit. Finally, the dynamic evaluation unit outputs the dynamic performance evaluation results based on the calibrated digital twin. S300: The performance degradation prediction unit of the application module predicts the performance degradation trend and remaining lifespan based on historical and current dynamic performance evaluation results. At the same time, the early warning generation unit generates early warning information when the performance index is lower than the preset threshold or when a failure is predicted.

[0032] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A performance evaluation system for a current transformer, characterized in that, The system includes an intelligent assessment module and a predictive application module; The intelligent evaluation module includes a high-fidelity physical model unit, a data-driven calibration unit, and a dynamic evaluation unit; The high-fidelity physical model unit is used to construct an initial digital twin characterizing the electromagnetic, thermal, and mechanical properties of the current transformer being evaluated. The data-driven calibration unit is used to dynamically calibrate the model parameters of the initial digital twin; The dynamic evaluation unit is used to calculate the real-time error and / or comprehensive health index of the current transformer being evaluated based on the calibrated digital twin. The prediction application module includes a performance degradation prediction unit; The performance degradation prediction unit is used to predict performance degradation trends and remaining lifetime based on historical and current dynamic performance evaluation results.

2. The performance evaluation system for a current transformer according to claim 1, characterized in that, It also includes a multimodal data acquisition module, which comprises an electrical parameter monitoring unit and a physical state monitoring unit; The electrical parameter monitoring unit is configured to acquire a reference electrical signal through a reference-level current transformer and monitor the current. ,Voltage Power factor parameter; The physical state monitoring unit is configured to collect data including temperature monitoring. Vibration acceleration Insulation resistance At least one non-electrical parameter in it.

3. The performance evaluation system for a current transformer according to claim 2, characterized in that, The high-fidelity physical model unit is constructed based on the electromagnetic-thermal coupling principle, including an electromagnetic induction part and a heat conduction part: The electromagnetic induction part uses the formula for mutual inductance electromotive force: ,in Indicates mutual inductance electromotive force; Represents the mutual inductance coefficient; This represents the rate of change of the primary current; The heat conduction section adopts Fourier's law of heat conduction: ,in Indicates heat flux density; Indicates thermal conductivity; This represents the temperature gradient.

4. The performance evaluation system for a current transformer according to claim 3, characterized in that, The data-driven calibration unit uses the least squares method to optimize model parameters, including: Define the loss function: , in, This represents the loss function, used to measure the deviation between the model's predicted values ​​and the actual values; Indicates the first The actual data values ​​collected vary in units depending on the monitoring parameters. Indicates the first The data values ​​predicted by the model; representing the number of data samples; Minimize the loss function using the gradient descent algorithm: , in, The units for representing model parameters vary depending on the parameter type. This represents the learning rate of the gradient descent algorithm. It is dimensionless and usually a decimal between 0 and 1. This indicates the loss function in the parameters The gradient at the point is used to match the model with the actual device.

5. The performance evaluation system for a current transformer according to claim 4, characterized in that, The prediction application module also includes an early warning generation unit, which generates early warning signals through the following logic: ， in, This indicates a warning signal; 1 means a warning is generated, and 0 means no warning is generated. This represents the predicted performance metric value; The threshold represents the performance metric.

6. The performance evaluation system for a current transformer according to claim 5, characterized in that, The intelligent evaluation module is communicatively connected to the multimodal data acquisition module and is used to establish and dynamically calibrate a digital twin of the current transformer being evaluated based on the electrical and non-electrical parameters, so as to output dynamic performance evaluation results.

7. The performance evaluation system for a current transformer according to claim 6, characterized in that, The prediction application module is communicatively connected to the intelligent evaluation module and is used to perform trend prediction based on the dynamic performance evaluation results and generate predictive maintenance information.

8. A method for evaluating the performance of a current transformer, applicable to a performance evaluation system for a current transformer as described in any one of claims 1-5, characterized in that, The method includes the following steps: S100. The electrical parameters of the current transformer being evaluated and at least one non-electrical parameter reflecting its physical state are simultaneously acquired through the multi-modal data acquisition module. S200: The initial digital twin is constructed through the high-fidelity physical model unit of the intelligent evaluation module. Then, the model parameters of the initial digital twin are dynamically calibrated based on the collected parameters by the data-driven calibration unit. Finally, the dynamic evaluation unit outputs the dynamic performance evaluation results based on the calibrated digital twin. S300: The performance degradation prediction unit of the application module predicts the performance degradation trend and remaining lifespan based on historical and current dynamic performance evaluation results. At the same time, the early warning generation unit generates early warning information when the performance index is lower than the preset threshold or when a failure is predicted.

9. The performance evaluation method for a current transformer according to claim 8, characterized in that, The dynamic calibration of model parameters in step S200 specifically includes: Real-time acquisition of multimodal data as input, and continuous adjustment of mutual inductance coefficient using gradient descent algorithm. Thermal conductivity parameter; Establish a loss function to measure the deviation between the model's predicted values ​​and the actual values, thereby achieving a match between the model and the actual equipment; An adaptive learning rate adjustment strategy is adopted to ensure the stability and efficiency of parameter convergence.

10. The performance evaluation method for a current transformer according to claim 8, characterized in that, The performance degradation prediction in step S300 is implemented using an LSTM model, including: input gate: ， in, Indicates the LSTM input gate at The output at any given time is a value between 0 and 1. This represents the weight matrix of the input gate; express The hidden state at any given moment; express Input data at any given time, such as historical error rate and temperature; This represents the bias term of the input gate; express The activation function, its mathematical expression is: ; Forgotten Gate: ， in This indicates that the LSTM forget gate is in The output at any given time is a value between 0 and 1. A value close to 0 indicates that the information at the corresponding location will be forgotten; a value close to 1 indicates that it will be retained. The weight matrix representing the forget gate; The bias term representing the forget gate; Output gate: ， in, This indicates that the LSTM output gate is in The output at any given time is a value between 0 and 1. This represents the weight matrix of the output gate; This represents the bias term of the output gate.

Images