Wide-range high-precision current transformer and compensation method

Abstract

The invention provides a wide-range high-precision current transformer and a compensation method. The input end of a voltage regulator of the current transformer applies alternating current voltage, and the output end of the voltage regulator is connected with a self-rising current standard current transformer; the self-current-rising standard current transformer comprises a current-rising converter and a standard current transformer CT1, and a secondary side of the self-current-rising standard current transformer is connected with the current transformer CT0 to be detected; the primary side of the current rising converter is connected with the voltage regulator, and the secondary side of the current rising converter is connected with the current transformer CT0 to be detected and the standard current transformer CT1; the primary side of the standard current transformer is connected with the sampling resistor R1, and the secondary side of the standard current transformer is connected with the current rising converter and the current transformer CT0 to be detected; the primary side of the to-be-tested current transformer CT0 is connected with the self-rising current standard current transformer, and the secondary side of the to-be-tested current transformer CT0 is connected with the rated load R0 and the auxiliary standard current transformer CT2; and the secondary side of the auxiliary standard current transformer is connected with the sampling resistor R2. Therefore, the measurement precision is effectively improved, and the problems of low precision of a wide-range current transformer and narrow range of a high-precision current transformer are solved.

Description

Technical Field

[0001] The present invention relates to the field of current measurement and protection, and in particular to a wide-range and high-precision current transformer and a compensation method. Background Art

[0002] With the acceleration of global energy transformation and grid modernization, the power system has increasingly stringent requirements for operation control, intelligent management, and equipment safety. Current transformers (CTs) are an indispensable part of the power system, which can safely convert large currents in the primary circuit into small currents for accurate measurement and effective control. However, the current level of intelligence is relatively low, and the simultaneous use of low-precision wide-range current transformers and high-precision narrow-range current transformers brings unnecessary redundancy and complexity. In the face of these challenges, it is imperative to develop high-precision, wide-range electromagnetic current transformers.

[0003] The causes of errors in electromagnetic current transformers usually include temperature changes, manufacturing defects, external interference, core saturation, etc.; the most important reason is the energy loss during the excitation process. At present, there are a lot of research on reducing current transformer errors. One of the methods is to explore new designs and technologies for current transformers. For example, optical current transformers based on the Faraday effect, coreless current transformers based on the Hall effect, magnetic sensors based on the tunnel magnetoresistance (TMR) effect, current sensors based on Rogowski coils, and other new current transformers can effectively improve the accuracy of current transformers. However, these current transformers usually have relatively complex internal structures, high costs, and are difficult to use widely.

[0004] Of course, the traditional current transformer itself can also be improved, such as using an adjustable resistor with an auxiliary transformer for demagnetization to avoid current transformer saturation. It is also possible to expand the precise measurement range by selecting a current transformer with a larger initial permeability and saturation permeability. Current technology has also studied the effects of different core and secondary winding arrangements on magnetic field distribution, current error, phase error, and measurement accuracy. However, these methods make the design more complicated, increase manufacturing difficulty, and reduce subsequent maintainability. Summary of the invention

[0005] The present invention provides a wide-range and high-precision current transformer and compensation method, including a voltage regulator, a self-boosting standard current transformer, a current transformer to be measured, an auxiliary standard current transformer, a sampling resistor and a rated load, characterized in that it has two stages of offline measurement and online compensation to effectively improve the measurement accuracy. It simultaneously solves the low-precision problem of wide-range current transformers and the narrow-range problem of high-precision current transformers, and avoids the redundancy and complexity of using low-precision wide-range current transformers and high-precision narrow-range current transformers at the same time.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: The invention discloses a current transformer with wide measuring range and high precision, comprising a voltage regulator, a self-boosting standard current transformer, a current transformer to be measured, an auxiliary standard current transformer, a sampling resistor and a rated load, wherein an AC voltage is applied to the input end of the voltage regulator, and the output end is connected to the self-boosting standard current transformer; the self-boosting standard current transformer comprises a current boost converter and a standard current transformer CT1, and a secondary side thereof is connected to a current transformer to be measured CT0; the primary side of the current boost converter is connected to the voltage regulator, and the secondary side is connected to the current transformer to be measured CT0 and the standard current transformer CT1; the primary side of the standard current transformer is connected to the sampling resistor R1, and the secondary side is connected to the current boost converter and the current transformer to be measured CT0; the primary side of the current transformer to be measured CT0 is connected to the self-boosting standard current transformer, and the secondary side is connected to the rated load R0 and the auxiliary standard current transformer CT2; the secondary side of the auxiliary standard current transformer CT2 is connected to the sampling resistor R2.

[0007] Furthermore, it includes two stages: offline measurement and online compensation, which specifically include the following processes: (1) In the offline measurement stage, the primary current and secondary current at different times need to be measured offline through the cooperation of the microcontroller, the sampling chip, the standard current transformer CT1, the auxiliary standard current transformer CT2 and the sampling resistors R1 and R2; wherein the standard current transformer CT1 is responsible for measuring the primary current, the auxiliary standard current transformer CT2 is responsible for measuring the secondary current, and the microcontroller and the sampling chip are responsible for collecting the values ​​of the primary current and the secondary current at different times; after each measurement, the closed-loop demagnetization method is used to remove the residual magnetism in the current transformers of CT1, CT2 and CT0; (2) In the online compensation stage, the auxiliary standard current transformer CT2 and the sampling resistor R2 are used to measure the secondary current at different times, and the collected data is processed by fast Fourier transform FFT to calculate the effective value of the time-varying current. At the same time, the main magnetic flux is calculated by the secondary current; the corresponding functional relationship between the main magnetic flux and the excitation current is obtained by the neural network trained by the multi-layer perceptron MLP, and the excitation current is calculated online, and the excitation current after compensation is calculated. i 2 * .

[0008] Specifically, the microcontroller is an STM32F103C8T6 microcontroller unit, and the sampling chip is an ADS131A04 24-bit synchronous AD sampling chip.

[0009] Specifically, the collected data is processed using the Fast Fourier Transform (FFT). The data is expressed as follows: ; in, It is the FFT k outputs, is a discrete frequency domain signal The period of , j is the imaginary unit, is a complex exponential function.

[0010] Specifically, the effective value of the time-varying current is calculated as follows: .

[0011] Furthermore, the main magnetic flux obtained by the neural network trained by the multi-layer perceptron MLP With excitation current Corresponding function relationship, for excitation current Perform online calculations and calculate the compensated i 2 * , specifically: by As the reference value, the most suitable function model is selected to calculate the corresponding excitation current , which is expressed as follows: ; in, Before compensation The effective value of is the effective value of the secondary current under the typical operating condition closest to the current operating condition; the excitation current obtained according to the above formula is For secondary current To compensate, the expression is as follows: .

[0012] Furthermore, the main magnetic flux obtained by the neural network trained by the multi-layer perceptron MLP With excitation current The corresponding functional relationship is as follows: The primary current at different times is obtained by offline measurement The secondary current After calculation, a large amount of main magnetic flux is obtained With excitation current The data is used to train the neural network, adjust the network parameters through the back propagation algorithm, and fit the main magnetic flux With excitation current Corresponding functional relationship; the neural network is a multi-layer perceptron MLP neural network, which is a feedforward neural network that can fit complex nonlinear functions by learning the mapping relationship between input and output.

[0013] Furthermore, the training process of the neural network is as follows: (1) Data preparation: Collect or generate the data set of the required fitting function, including the values ​​of the main magnetic flux and excitation current; (2) Network structure design: Design the structure of the MLP network, including an input layer, a hidden layer, and an output layer; the number of nodes in the input layer corresponds to the main magnetic flux, and the number of nodes in the output layer corresponds to the excitation current; (3) Training the network: Use the collected data to train the MLP and adjust the network parameters through the back-propagation algorithm to minimize the prediction error; (4) Verification and testing: Verify the performance of the MLP model on an independent test set to ensure that the model has the required generalization ability; (5) Result analysis: Analyze the results of MLP fitting and compare them with the actual main flux and excitation current data to evaluate the accuracy of the fitting.

[0014] The beneficial effects of the present invention are as follows: a wide-range high-precision current transformer and its compensation algorithm are proposed, including a voltage regulator, a self-current standard current transformer, a current transformer to be measured, an auxiliary standard current transformer, a sampling resistor and a rated load. The current transformer has two stages: offline measurement and online compensation, so as to effectively improve the measurement accuracy; at the same time, the low-precision problem of the wide-range current transformer and the narrow-range problem of the high-precision current transformer are solved, and the redundancy and complexity of using low-precision wide-range current transformers and high-precision narrow-range current transformers at the same time are avoided. The traditional wide-range protection current transformer has a large error when measuring small currents, and the traditional high-precision measurement current transformer has the shortcomings of easy saturation of fault current and narrow range, and combines the advantages of the two types of electromagnetic current transformers to obtain a high-precision wide-range current transformer that is not saturated in measuring fault current and has a certain accuracy in measuring different percentages of rated current and fault current. Through the analysis of the working principle of multi-layer perceptron (MLP), the training process of neural network fitting the corresponding functional relationship between main magnetic flux and excitation current is given, and the intelligent compensation of current transformer error is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is the equivalent circuit diagram of current transformer; Figure 2 This is a schematic diagram of the functional relationship between the main magnetic flux and the excitation current; Figure 3 Simulate the circuit diagram for the current transformer; Figure 4 It is a comparison diagram before and after compensation at rated current; Figure 5 It is the electrical wiring diagram of the current transformer; Figure 6 This is a comparison chart before and after compensation of closed-type CT with 20% rated current; Figure 7 This is a comparison chart before and after compensation of closed-type CT with 120% rated current; Figure 8 This is a comparison chart before and after compensation of open-type CT with 120% rated current; Fig. 9 This is a comparison chart before and after compensation of 5% rated current open-type CT. DETAILED DESCRIPTION

[0016] The present invention is further described below in conjunction with the accompanying drawings and specific implementation examples.

[0017] Figure 1 In the figure, the equivalent circuit diagram of the current transformer is shown, where is the primary current, is the secondary current, is the excitation current, It's the load.

[0018] According to Kirchhoff's current law and Faraday's law of electromagnetic induction, it satisfies: ; ;in and are the number of winding turns on the primary and secondary sides respectively. is the main magnetic flux. Due to the excitation current, and The transformation ratio is not met and there is a phase error. In addition, since the magnitude of the excitation current is related to factors such as the core material, the size and shape of the core, the magnetic field strength, the operating frequency and the ambient temperature, it is impossible to correct the secondary current by directly measuring the excitation current. Therefore, in order to improve the accuracy of the transformer, it is necessary to determine the magnitude of the excitation current in other ways.

[0019] By integrating the above formula, the following formula can be derived: ;in Indicates time The magnetic flux value at Represents the initial magnetic flux. Residual magnetism of current transformer Set to zero by demagnetization.

[0020] Discretizing the above formula yields the following formula: ;in is the sampling interval. , we can get the magnetic flux at the current moment .

[0021] The hysteresis curve describes the relationship between magnetic flux density B and magnetic field strength H. According to the electromagnetic field principle and the complete current law, the following formula can be obtained: ; From this we can infer that the magnetic flux and excitation current It also conforms to the relationship of the hysteresis curve, such as Figure 2 The present invention uses a multi-layer perceptron (MLP) to detect magnetic flux and excitation current The data is used to train the neural network to fit the magnetic flux and excitation current The corresponding function relationship.

[0022] In the online compensation stage, the current working conditions must be determined and a suitable neural network model must be selected to obtain a suitable corresponding functional relationship between magnetic flux and excitation current. The data is subjected to fast Fourier transform (FFT) to obtain the time-varying current Frequency domain representation of , the formula is as follows: ; in, Before compensation The effective value of It is the FFT outputs, is a discrete frequency domain signal The period of , j is the imaginary unit, is a complex exponential function, and then the following formula is used to calculate The effective value of : ; and As the reference value, the most suitable function model is selected to calculate the corresponding excitation current , the formula is as follows: ;in, is the effective value of the secondary current under the typical operating condition closest to the current operating condition. The excitation current obtained according to the above formula is For secondary current Compensation is performed to improve accuracy, and the formula is as follows: .

[0023] The following are simulation and experimental application examples of the present invention: The present invention uses the EMTP-RV electromagnetic transient analysis platform for simulation. The simulation circuit diagram constructed in the software is as follows: Figure 3 The hysteresis curve data comes from an actual wide-range current transformer, the rated load of the simulation circuit is 2.5VA, and the transformation ratio is 200:5.

[0024] The present invention conducts simulation verification under the condition that the primary side current is 100% of the rated current. The comparison effect before and after compensation is as follows: Figure 4 As shown. Figure 4 It can be clearly seen that the compensation effect is very significant. i 2 * Almost the same as the actual primary current i 1 completely overlaps, indicating that the algorithm proposed in this invention effectively improves the measurement accuracy of the current transformer under this condition. By using FFT to analyze the waveform, we obtain i 1. i 2 and i 2 * The peak value, phase angle and other information of the current are obtained. The corresponding current error is calculated using the information of the relevant current and phase angle. Phase Error , the two formulas are as follows: ;in, is the rated current ratio, and are the effective values ​​of the primary and secondary currents, respectively. and are the phase angles of the primary and secondary currents respectively; the above formula is used to calculate the data before and after compensation. When the secondary current is 100% of the rated current, the current error after compensation decreases from -0.33% to 0.03%, and the phase error decreases from 127.02 to 1.36. Its accuracy has been improved to level 0.1, which verifies the feasibility of high-precision and wide-range current transformers.

[0025] In addition, the present invention has also been verified by physical experiments, and a hardware experimental platform for current transformers has been constructed. Figure 5 The 220V AC power is sized by a transformer and then passes through a self-boosting standard current transformer to achieve a 5% to 1000% condition. i 1. The primary side sampling resistor of the standard current transformer can be used R 1. Test the secondary side of the current transformer with the rated load R 0 series connection, current i 2. Through the sampling resistor on the secondary side of the auxiliary standard current transformer R 2. Using STM32F103C8T6 microcontroller unit and ADS131A04 24-bit synchronous AD sampling chip R 1 and R After each measurement, the closed-loop demagnetization method is used for CT1, CT2, and CT0 to ensure that there is no residual magnetism in the current transformer.

[0026] The physical experiment described above is firstly carried out to verify the closed-type current transformer. Figure 6 The figure shows the comparison of compensation effect when the primary current is 20% of the rated current. Due to the existence of magnetizing current, there is a certain error in the secondary current measured before compensation. After algorithm compensation, the secondary current overlaps with the primary current. The current error is reduced from -0.86% to 0.03%, the phase error is reduced from 36.40 points to -1.57 points, and the accuracy is improved from 1C level to 0.1C level. Similarly, Figure 7 As shown in the figure, the close-distance current transformer under 120% conditions also has similar results. The current error is reduced from -0.69% to -0.02%, the phase error is reduced from 25.26 points to 1.14 points, and the accuracy is improved from 1C level to 0.1C level.

[0027] Since the open-type current transformer has the advantage of not needing to disconnect the circuit during installation and removal, it has been widely used in practical applications. However, since there is an air gap between its iron cores, the accuracy of measuring small currents is significantly lower. The present invention also selects the open-type current transformer for experiments.

[0028] Figure 8 The figure shows the comparison of the open-type current transformer before and after compensation under the condition of 20% rated current on the primary side. In this case, the open-type current transformer shows huge current and phase errors compared with the closed-type current transformer. However, the secondary current is almost consistent with the primary current after compensation. The current error drops sharply from -6.29% to 0.15%, and the phase error drops sharply from 478.26 points to 7.94 points. Fig. 9 It shows that even at 5% of the rated current, the open-type current transformer can accurately measure the actual current with the help of the compensation algorithm. The current error is reduced from -6.97% to 0.30%, and the phase error is reduced from 500.95 points to 13.79 points. The accuracy of the open-type current transformer, which was originally not suitable for measuring small currents, is increased to 0.2C level, greatly expanding its application range.

[0029] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A current transformer with wide range and high precision, characterized in that: It comprises a voltage regulator, a self-boosting standard current transformer, a current transformer to be tested, an auxiliary standard current transformer, a sampling resistor and a rated load. An AC voltage is applied to the input end of the voltage regulator, and the output end is connected to the self-boosting standard current transformer; the self-boosting standard current transformer comprises a current boost converter and a standard current transformer CT1, and its secondary side is connected to the current transformer to be tested CT0; the primary side of the current boost converter is connected to the voltage regulator, and the secondary side is connected to the current transformer to be tested CT0 and the standard current transformer CT1; the primary side of the standard current transformer is connected to the sampling resistor R1, and the secondary side is connected to the current boost converter and the current transformer to be tested CT0; the primary side of the current transformer to be tested CT0 is connected to the self-boosting standard current transformer, and the secondary side is connected to the rated load R0 and the auxiliary standard current transformer CT2; the secondary side of the auxiliary standard current transformer CT2 is connected to the sampling resistor R2.

2. The current transformer compensation method according to claim 1, characterized in that: It includes two stages: offline measurement and online compensation, and specifically includes the following processes: (1) In the offline measurement stage, the primary current and secondary current at different times need to be measured offline through the cooperation of the microcontroller, the sampling chip, the standard current transformer CT1, the auxiliary standard current transformer CT2 and the sampling resistors R1 and R2; wherein the standard current transformer CT1 is responsible for measuring the primary current, the auxiliary standard current transformer CT2 is responsible for measuring the secondary current, and the microcontroller and the sampling chip are responsible for collecting the values ​​of the primary current and the secondary current at different times; after each measurement, the closed-loop demagnetization method is used to remove the residual magnetism in the current transformers of CT1, CT2 and CT0; (2) In the online compensation stage, the auxiliary standard current transformer CT2 and the sampling resistor R2 are used to measure the secondary current at different times, and the collected data is processed by fast Fourier transform FFT to calculate the effective value of the time-varying current. At the same time, the main magnetic flux is calculated by the secondary current; the corresponding functional relationship between the main magnetic flux and the excitation current is obtained by the neural network trained by the multi-layer perceptron MLP, and the excitation current is calculated online, and the excitation current after compensation is calculated. i 2 * .

3. The compensation method according to claim 2, characterized in that: The microcontroller is an STM32F103C8T6 microcontroller unit, and the sampling chip is an ADS131A04 24-bit synchronous AD sampling chip.

4. The compensation method according to claim 2, characterized in that: The fast Fourier transform (FFT) is used to process the collected The data is expressed as follows: ； in, It is the FFT k outputs, is a discrete frequency domain signal The period of , j is the imaginary unit, is a complex exponential function.

5. The compensation method according to claim 2, characterized in that: The effective value of the time-varying current is calculated as follows: 。 6. The compensation method according to claim 2, characterized in that: The main magnetic flux obtained by the neural network trained by the multi-layer perceptron MLP With excitation current Corresponding function relationship, for excitation current Perform online calculations and calculate the compensated i 2 * , specifically: by As the reference value, the most suitable function model is selected to calculate the corresponding excitation current , which is expressed as follows: ； in, Before compensation The effective value of is the effective value of the secondary current under the typical operating condition closest to the current operating condition; the excitation current obtained according to the above formula is For secondary current To compensate, the expression is as follows: 。 7. The compensation method according to claim 2, characterized in that: The main magnetic flux obtained by the neural network trained by the multi-layer perceptron MLP With excitation current The corresponding functional relationship is as follows: The primary current at different times is obtained by offline measurement The secondary current After calculation, a large amount of main magnetic flux is obtained With excitation current The data is used to train the neural network, adjust the network parameters through the back propagation algorithm, and fit the main magnetic flux With excitation current Corresponding functional relationship; the neural network is a multi-layer perceptron MLP neural network, which is a feedforward neural network that can fit complex nonlinear functions by learning the mapping relationship between input and output.

8. The compensation method according to claim 7, characterized in that: The training process of the neural network is as follows: (1) Data preparation: Collect or generate the data set of the required fitting function, including the values ​​of the main magnetic flux and excitation current; (2) Network structure design: Design the structure of the MLP network, including an input layer, a hidden layer, and an output layer; the number of nodes in the input layer corresponds to the main magnetic flux, and the number of nodes in the output layer corresponds to the excitation current; (3) Training the network: Use the collected data to train the MLP and adjust the network parameters through the back-propagation algorithm to minimize the prediction error; (4) Verification and testing: Verify the performance of the MLP model on an independent test set to ensure that the model has the required generalization ability; (5) Result analysis: Analyze the results of MLP fitting and compare them with the actual main flux and excitation current data to evaluate the accuracy of the fitting.

Images