Current transformer zero flux control method and system based on magnetic potential balance constraint

Abstract

The invention relates to the technical field of power electronic measurement, and discloses a current transformer zero flux control method and system based on magnetic potential balance constraint, and the method comprises the steps: converting a current signal of a current transformer into a pulse sequence, and generating an event drive data format; processing the pulse sequence by using an event-driven wavelet decomposition algorithm, and outputting a multiband pulse stream; analyzing the multiband pulse stream, and calculating a coupling relationship between frequency components; calculating a magnetic potential balance compensation parameter by using a pulse frequency coding algorithm based on the coupling relationship between the frequency components; the magnetic potential balance compensation parameters are converted into an asynchronous control pulse sequence, and compensation current is generated and output to a magnetic potential balance execution mechanism; by introducing a neuromorphic calculation and event-driven processing mechanism, the problem of inherent delay of a traditional synchronous clock sampling method is solved, and microsecond-level response control over the current transformer in the high-frequency switching power supply system is achieved.

Description

Technical Field

[0001] This invention relates to the field of power electronic measurement technology, and more specifically, to a method and system for zero flux control of current transformers based on magnetomotive force balance constraints. Background Technology

[0002] With the rapid development of power electronics technology, high-frequency switching power supply systems are widely used in various scenarios. In these systems, current transformers, as key measurement components, need to respond to rapidly changing current signals within microseconds to ensure stable system operation and measurement accuracy.

[0003] Currently, traditional current transformer control methods are mainly based on synchronous clock sampling technology and employ algorithms such as wavelet transform and graph neural networks for signal processing. These methods have significant technical drawbacks in practical applications: First, their signal processing delay is typically in the millisecond range, failing to meet the stringent requirements of modern high-frequency power electronic systems for microsecond-level transient response; second, due to the large processing delay, the system cannot effectively capture and compensate for high-frequency components, leading to a decrease in the measurement accuracy of the current transformer and affecting the stability of the entire system; furthermore, during high-frequency switching, rapid current changes can easily cause core saturation in the current transformer, generating measurement errors and severely impacting the normal operation and safety of the system.

[0004] There is no effective solution to the above problems in the existing technology, especially in terms of microsecond-level response time and prevention of magnetic core saturation, where there is a significant technological gap. There is an urgent need to develop new current transformer control technologies. Summary of the Invention

[0005] This invention provides a zero-flux control method and system for current transformers based on magnetomotive force balance constraints, solving the technical problem in related technologies that current transformers cannot respond to rapidly changing current signals within microseconds.

[0006] This invention provides a zero-flux control method for current transformers based on magnetomotive force balance constraints, comprising: The current signal from the current transformer is converted into a pulse sequence to generate an event-driven data format. The pulse sequence is processed using an event-driven wavelet decomposition algorithm to output a multi-band pulse stream. The analysis of multi-band pulse streams and the calculation of coupling relationships between frequency components include: extracting the pulse occurrence time series for each pair of frequency components; standardizing the time series to eliminate differences in time scales across different frequency bands; calculating the standardized time difference and applying a time kernel function for weighting; normalizing the coupling weights of each frequency component to ensure the sum of the weights is 1; multiplying by the normalized coupling weights to obtain the coupling strength; and constructing a pulse timing parameter relationship graph model based on the coupling strength. Based on the coupling relationship between frequency components, the magnetic potential balance compensation parameters are calculated using a pulse frequency coding algorithm, including: normalizing the magnetic flux data to standardize the magnetic flux values ​​to the range of [-1, 1]; normalizing the pulse frequency code to make the numerical range of each frequency component consistent; and combining the normalized magnetic flux data with the pulse frequency code to calculate the magnetic potential balance compensation parameters through a weighted nonlinear mapping function. The magnetomotive force balance compensation parameters are converted into an asynchronous control pulse sequence, which generates a compensation current and outputs it to the magnetomotive force balance actuator.

[0007] Furthermore, the step of converting the current signal of the current transformer into a pulse sequence includes: The original current signal is normalized to normalize the current amplitude to the range of [0, 1]. The pulse density coding algorithm is applied to convert the normalized current signal into a pulse sequence; An adaptive threshold adjustment algorithm is executed. When the rate of change of the current signal exceeds a preset threshold, the time window is dynamically reduced and the pulse generation frequency is increased.

[0008] Furthermore, the event-driven wavelet decomposition algorithm utilizes a spiking neural network, which consists of the following components: Input layer: Receives pulse sequences and distributes them to different processing channels; Feature extraction layer: Composed of multiple spiking neurons, each neuron has different time-frequency filter characteristics; Frequency separation layer: Groups and clusters pulses with different frequency characteristics; Output layer: Generates multi-band pulse streams while preserving the time-frequency characteristics of the original signal.

[0009] Furthermore, the encoded pulse sequence is input into a neuromorphic processor for subsequent processing, including: The pulse sequence is timestamped to ensure that events are processed in the order of occurrence. A pulse sparsity algorithm is applied to remove redundant pulses and improve processing efficiency. The processed pulse sequence is distributed to different processing units of the neuromorphic processor; Achieving microsecond-level response speed through asynchronous parallel processing; Generate a standardized event stream.

[0010] Furthermore, the pulse sequence is processed using an event-driven wavelet decomposition algorithm, outputting a multi-band pulse stream including: The pulse sequence is input into an event-driven wavelet decomposition algorithm, which performs calculations based on pulse triggering and only executes when the signal changes. Perform impulse response transformation by convolving the input pulse with a preset wavelet-like basis function to extract the characteristics of different frequency components; Frequency-selective filtering is applied to allocate pulses to different processing channels based on their time-frequency characteristics; The aggregation process generates a multi-band pulse stream output, with each band corresponding to a frequency component of the current signal.

[0011] Furthermore, the step of converting the magnetomotive force balance compensation parameters into an asynchronous control pulse sequence includes: The compensation parameters are normalized to eliminate differences in dimensions and numerical ranges between different parameters; The standardized compensation parameters are mapped to the pulse amplitude and frequency space; The corresponding pulse density is generated by using a nonlinear mapping function based on the magnitude and rate of change of the parameters.

[0012] Furthermore, the step of generating the compensation current includes: The input control pulses are digitally filtered to remove high-frequency noise. Pulse sequences are standardized to unify their time-domain and frequency-domain characteristics; The standardized pulse sequence is converted into a compensation current amplitude by applying a direct mapping function; The generated current amplitude is calibrated to ensure it meets the input specifications of the actuator; Adjust the timing characteristics of the compensation current according to the pulse timing characteristics.

[0013] Furthermore, the step of generating a compensation current and outputting it to the magnetomotive force balance actuator includes: Monitor real-time changes in magnetic flux; Calculate the deviation of the magnetic flux from zero; The compensation current is adjusted in real time to make the magnetic flux approach zero.

[0014] Furthermore, the generation process of the asynchronous control pulse sequence utilizes an asynchronous control event-driven model, which includes the following key components: Pulse generator: dynamically generates control pulses based on compensation parameters; Event Queue Manager: Manages the ordering and scheduling of asynchronous pulse events; Direct digital-to-analog mapping unit: converts pulses into analog compensation currents; Magnet potential detection feedback unit: provides closed-loop control feedback.

[0015] This invention provides a zero-flux control system for a current transformer based on magnetomotive force balance constraints, used to execute the aforementioned zero-flux control method for a current transformer based on magnetomotive force balance constraints, comprising: Pulse conversion module: used to convert current signals into pulse sequences and generate event-driven data; Frequency band analysis module: used to implement event-driven wavelet decomposition and multi-band pulse stream output; Frequency coupling calculation module: used to process multi-band pulse streams and construct coupling relationships between frequency components; Magnetic potential balance calculation module: used to generate magnetic potential balance compensation parameters based on frequency coupling relationship; Control execution module: used to convert compensation parameters into control pulses and drive the magnetomotive force balance actuator.

[0016] The beneficial effects of this invention are as follows: traditional signal processing methods based on synchronous clocks require waiting for a complete sampling period before processing can begin, while the event-driven processing mechanism adopted in this invention enables the system to trigger calculations only when the signal changes, fundamentally eliminating this delay; By converting current signals into pulse sequences and using spiking neural networks for parallel processing, the system can respond to current changes on a microsecond timescale, which improves response speed compared to traditional millisecond-level processing methods. This invention can capture and respond to microsecond-level signal changes in real time, thus providing better compensation for high-frequency components in high-frequency switching power supplies. This enhanced compensation capability directly leads to the effective prevention of core saturation. By constructing the pulse timing parameter relationship diagram and applying pulse frequency encoding, the system can capture subtle changes in the current signal and perform precise magnetomotive force compensation based on these changes, thereby achieving more precise magnetomotive force balance and zero flux control. Event-driven processing mechanisms activate computation only when signals change, which greatly reduces unnecessary computational operations and lowers system power consumption. Through an adaptive threshold mechanism and parameter adjustment algorithm, this invention can adapt to current transformers under different operating conditions, thereby improving the applicability and stability of the system. Attached Figure Description

[0017] Figure 1 This is a flowchart of the zero flux control method for current transformers based on magnetomotive force balance constraints in this invention; Figure 2 It is a line graph comparing the response characteristics during the pulse code analysis of current jump events; Figure 3 It is a line graph comparing the relationship between the magnetomotive force balance compensation parameters and the magnetic flux; Figure 4 It is a bar chart comparing the changes in magnetic flux before and after compensation under different current conditions; Figure 5 It is a bar chart comparing the response speeds of different control methods; Figure 6 This is a bar chart comparing the performance of traditional synchronous control methods and the method of this invention in terms of flux density reduction rate and iron loss reduction rate at different operating frequencies. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses a zero-flux control method for current transformers based on magnetomotive force balance constraints, such as... Figure 1 As shown, it includes: Step 1: Convert the current signal of the current transformer into a pulse sequence to generate an event-driven data format; This step includes the following operations: Step 1.1, sampling the current signal; A high-speed analog-to-digital converter (ADC) is used to sample the secondary current signal of the current transformer to obtain an analog current signal.

[0020] Step 1.2, Pulse density encoding; The sampled analog current signal is converted into a pulse sequence using a pulse density coding algorithm. Specifically, the following operations are performed: The original current signal is normalized to normalize the current amplitude to the range of [0, 1] to facilitate subsequent processing. Analyze the normalized signal amplitude and establish the mapping relationship between amplitude and pulse density; Generate pulse events of corresponding density on the timeline; The event sequence is marked as a timestamp, address pair format.

[0021] Step 1.3, adjust the encoding parameters; An adaptive threshold adjustment algorithm is executed, and the rate of change of the current signal is monitored in real time. The encoding parameters are adjusted through the following steps: Calculate the rate of change of the signal within a short time window; Compare the calculation results with a preset threshold; When the rate of change exceeds the threshold, the time window is dynamically reduced and the pulse generation frequency is increased; When the rate of change decreases, the time window is gradually restored to its normal size.

[0022] like Figure 2 The figure illustrates the pulse coding analysis process for a current jump event. It demonstrates how the system adaptively adjusts the time window and increases the number of generated pulses as the current value increases rapidly (from 50A to 250A), visually showcasing the rapid response characteristics of the event-driven processing mechanism. The figure clearly shows that when the current changes drastically, the time window dynamically shrinks from 10μs to 2μs, while the number of generated pulses increases accordingly, ensuring the system can respond to current changes within microseconds.

[0023] Step 1.4, follow-up processing; The encoded pulse sequence is then input into a neuromorphic processor for further processing. Specifically, this includes: The pulse sequence is timestamped to ensure that events are processed in the order of occurrence. A pulse sparsity algorithm is applied to remove redundant pulses and improve processing efficiency. The processed pulse sequence is distributed to different processing units of the neuromorphic processor; Achieving microsecond-level response speed through asynchronous parallel processing; Generate a standardized event stream to provide input for the wavelet decomposition algorithm in the next step.

[0024] Furthermore, in the pulse density coding algorithm described above, a threshold prediction mechanism is used for particularly rapidly changing current signals. Pulses are generated in advance when the signal reaches a certain percentage of the predicted value, thereby further reducing processing delay.

[0025] Furthermore, in pulse frequency encoding, since the dynamic adjustment of the time window size affects the data distribution, the system introduces an adaptive normalization mechanism. This mechanism monitors the changes in data distribution caused by window adjustments in real time and dynamically adjusts the normalization parameters to ensure that pulse sequences generated under different time window sizes have consistent statistical characteristics. This adaptive normalization mechanism improves the stability and accuracy of the system under rapidly changing signal conditions.

[0026] Furthermore, to handle high-frequency components, the system performs power normalization on signals from different frequency channels to avoid bias when directly comparing high-frequency and low-frequency components. The power normalization process automatically adjusts the gain factor based on frequency response characteristics to ensure the balance of energy distribution across different frequency components, while preserving the relative phase information of each frequency component, thus providing a more reliable data foundation for subsequent frequency coupling analysis.

[0027] Step 2: Process the pulse sequence using an event-driven wavelet decomposition algorithm to output a multi-band pulse stream; This step includes the following operations: Step 2.1, Pulse sequence input; The pulse sequence is input into an event-driven wavelet decomposition algorithm, which performs calculations based on pulse triggering and only executes when the signal changes. Step 2.2, Impulse Response Transformation; Perform impulse response transformation by convolving the input pulse with a preset wavelet-like basis function to extract the characteristics of different frequency components; Furthermore, before performing the impulse response transformation, the system performs amplitude normalization on the input pulses, mapping the amplitudes of pulses from different sources to a uniform range. This normalization eliminates the amplitude differences between pulses generated by different signal sources, ensuring that subsequent convolution operations can accurately extract time-frequency characteristics without being affected by the original amplitudes. The normalization process employs a dynamic threshold mechanism, automatically adjusting the normalization parameters based on the pulse amplitude distribution within a short time window.

[0028] Step 2.3, frequency-selective filtering; Frequency-selective filtering is applied to allocate pulses to different processing channels based on their time-frequency characteristics; Step 2.4, aggregation results; The aggregation process generates a multi-band pulse stream output, with each band corresponding to a frequency component of the current signal.

[0029] Furthermore, the aforementioned event-driven wavelet decomposition algorithm utilizes a spiking neural network model in its hardware implementation. The spiking neural network model consists of the following components: Input layer: Receives pulse sequences and distributes them to different processing channels; Feature extraction layer: Composed of multiple spiking neurons, each neuron has different time-frequency filter characteristics, simulating wavelet basis functions; Frequency separation layer: Groups and clusters pulses with different frequency characteristics; Output layer: Generates multi-band pulse streams while preserving the time-frequency characteristics of the original signal.

[0030] Step 3: Analyze the multi-band pulse current and calculate the coupling relationship between frequency components; This step includes the following operations: Step 3.1: Receive multi-band pulse stream; Extract the time characteristic parameters of each frequency band, including pulse generation time, interval, and density; The system performs frequency-related weighting adjustments on multi-band pulse currents, taking into account the different physical meanings of different frequency components. The adjustment process includes: analyzing the energy contribution of each frequency component in the current signal, assessing the impact of different frequencies on core saturation, and establishing a frequency weighting mapping relationship. This weighting adjustment ensures that the actual importance of each frequency component is reasonably reflected in subsequent coupling relationship analysis, thus improving the practical value of the analysis results.

[0031] Step 3.2: Execute the activity propagation algorithm; Analyze the interaction between frequency components using the following steps: For each pair of frequency components, extract its pulse generation time sequence. and ; The time series is standardized to eliminate the differences in time scales across different frequency bands and normalize the time difference into relative time units. Calculate the standardized time difference and apply the time kernel function. Weighting; Coupling weights for each frequency component Normalization is performed to ensure that the sum of the weights is 1, thus guaranteeing the comparability of weights for different frequency channels; Multiply by the normalized coupling weights The coupling strength is obtained. ; Repeat the above calculation process to obtain the complete coupling matrix.

[0032] ; in, Represents a node and nodes The coupling strength between them and Representing nodes respectively and The pulse firing time, It is a time kernel function. These are coupling weights; The summation symbol is used.

[0033] Furthermore, the time kernel function Defined as an exponential decay function: ; in and Representing nodes respectively and The pulse firing time, It is a time kernel function; It is the base of the natural logarithm. This is the attenuation coefficient, used to control the effect of time differences on coupling strength. The attenuation rate is the rate at which time differences occur; the greater the time difference, the smaller the coupling effect. The time kernel function ensures that time-close pulse events have stronger interactions, while events with larger time intervals have less interaction.

[0034] Before performing critical coupling analysis, the system standardizes the calculated coupling strength matrix to ensure consistent comparison benchmarks across different frequency ranges. The standardization process includes: eliminating frequency-dependent biases, balancing the weights of high- and low-frequency components in the coupling analysis, and adjusting the dynamic range of the coupling strength to facilitate subsequent threshold selection. This standardization process improves the reliability and accuracy of cross-frequency range coupling relationship analysis.

[0035] Step 3.3, Key Coupling Analysis; Based on the calculated coupling matrix, execute the key coupling analysis algorithm: Set a coupling strength threshold to filter out weak coupling relationships; Clustering algorithms are used to group frequency components; Calculate the centrality index within each group and identify key coupling nodes.

[0036] Step 3.4: Output the results of key frequency coupling; The results of key frequency coupling are output, including frequency components that have a significant impact on core saturation and their interrelationships.

[0037] Furthermore, when executing the activity propagation algorithm, an adaptive time window technique is used, which can automatically adjust the time range of correlation analysis according to the characteristics of different frequency components, thereby improving the accuracy of coupling relationship analysis.

[0038] Furthermore, based on the above calculation results, the system implemented a pulse timing parameter relationship graph model, which consists of the following components: Graph structure layer: It consists of nodes and edges. Nodes represent the pulse characteristics of different frequency components, and edges represent the coupling relationship between frequency components. Time-series characteristic storage unit: records the time-series information of each node; Coupling matrix storage unit: stores the calculated coupling strength between frequency components; Path analysis module: used to find key nodes and coupling paths.

[0039] Step 4: Based on the coupling relationship between frequency components, calculate the magnetomotive force balance compensation parameters using a pulse frequency coding algorithm; This step includes the following operations: Step 4.1, pulse frequency encoding; The pulse frequency encoding algorithm is executed, and the following steps are performed: The number of pulses for each frequency component is counted within a fixed time window; Divide the number of pulses by the window length to obtain the pulse frequency code: ; in , , These represent the pulse frequencies of the 1st, 2nd, and nth frequency components, respectively. The total number of frequency components processed in the system; By combining the frequency code with the key frequency coupling relationship, an enhanced feature representation is generated; Step 4.2: Collect magnetic flux data; Collect magnetic flux data And perform the following preprocessing: The magnetic flux data is normalized to standardize the magnetic flux values ​​to the range of [-1, 1]. pulse frequency code Normalization is performed to ensure that the numerical range of each frequency component is consistent. The preprocessed magnetic flux data is combined with the pulse frequency code and input into the magnetic potential balance calculation algorithm.

[0040] Before calculating the magnetic potential balance difference, the system performs dimensional unification processing on all input parameters, especially the scale matching between physical quantities and computational quantities. Specifically, this processing includes converting magnetic flux data into a dimensionless form, applying frequency response characteristic correction to the pulse frequency code, and establishing a unified computational benchmark. This dimensional unification ensures that parameters with different physical meanings can be reasonably combined in the computational model, avoiding computational biases caused by dimensional inconsistencies.

[0041] Step 4.3, Calculation of magnetic potential balance difference; Calculate the magnetic potential balance difference based on the following relationship: ; in, Indicates the difference in magnetic potential balance; Indicates magnetic flux; , , They represent the 1st, 2nd, and 3rd respectively. The pulse frequency of each frequency component The total number of frequency components processed in the system; It is a function for calculating the difference in magnetic potential balance.

[0042] like Figure 3 The figure illustrates the calculation process of the magnetic potential balance compensation parameters. The graph, presented as a line graph, shows the relationship between the magnetic flux value and the calculated compensation parameters, demonstrating how the system calculates the corresponding compensation parameters to achieve magnetic potential balance control as the magnetic flux increases. It can be observed from the graph that the compensation parameters are always greater than the magnetic flux value, and the two exhibit an approximately linear relationship. This reflects the stability of the system's magnetic potential balance calculation algorithm, ensuring effective compensation control throughout the process of magnetic flux changes.

[0043] Furthermore, Defined as a weighted nonlinear mapping function: ; in It is a function for calculating the difference in magnetic potential balance; Indicates magnetic flux; Indicates the summation symbol; It is the first Weighting coefficients for each frequency component; , , They represent the 1st, 2nd, and 3rd respectively. The pulse frequency of each frequency component The total number of frequency components processed in the system; It is the first The nonlinear transformation function of each frequency component is defined as: ; in It is the first The pulse frequency of each frequency component; It is a non-linear adjustment coefficient.

[0044] The weighted nonlinear mapping function structure ensures that the linear relationship is dominant when the magnetic flux is low, while the influence of the nonlinear term is enhanced in the high magnetic flux region, which can accurately capture the complex relationship between the magnetic potential balance and the frequency component.

[0045] Step 4.4, adaptive calculation of compensation parameters; The adaptive calculation algorithm for compensation parameters is executed, and the parameters are optimized through the following steps: Calculate the changing trend of the magnetic potential balance difference; Adjusting compensation parameters based on gradient descent method; The learning rate is adaptively adjusted based on the dynamic characteristics of the magnetic source. Output the optimal compensation parameters.

[0046] Furthermore, in the above calculation of magnetic potential balance difference, piecewise linear logic was used for rapid calculation, and different calculation models were dynamically switched in the nonlinear region to quickly adapt to the range of current changes.

[0047] Furthermore, the magnetic potential balance model used in this step consists of the following components: Input layer: Receives magnetic flux signals and pulse frequency codes; Mapping layer: Transforms the input features into a feature space suitable for computation; Nonlinear computation unit: realizes the nonlinear mapping between magnetic potential balance differences and input characteristics; Parameter adaptive unit: Adjusts model parameters according to the dynamic characteristics of the real-time magnetic source; Output layer: Generates compensation parameters.

[0048] Step 5: Convert the magnetomotive force balance compensation parameters into an asynchronous control pulse sequence, generate a compensation current, and output it to the magnetomotive force balance actuator; This step includes the following operations: Step 5.1, parameter to pulse conversion; The parameter-to-pulse conversion algorithm is executed by following these steps: The compensation parameters are normalized and mapped to the standard range to eliminate the differences in dimensions and numerical ranges between different parameters. The standardized compensation parameters are mapped to the pulse amplitude and frequency space; The corresponding pulse density is generated by using a nonlinear mapping function based on the magnitude and rate of change of the parameters. The pulse sequence is arranged in chronological order to form an asynchronous control pulse sequence; Step 5.2, pulse-to-current mapping; Applying the pulse current mapping algorithm, perform the following calculation steps: The input control pulses are digitally filtered to remove high-frequency noise. Pulse sequences are standardized to unify their time-domain and frequency-domain characteristics; The standardized pulse sequence is converted into a compensation current amplitude by applying a direct mapping function; The generated current amplitude is calibrated to ensure it meets the input specifications of the actuator; Adjust the timing characteristics of the compensation current according to the pulse timing characteristics; Step 5.3, Zero Flux Control; The zero-flux control algorithm is executed to adjust the magnitude and direction of the compensation current in real time. Monitor real-time changes in magnetic flux; Calculate the deviation of the magnetic flux from zero; Adjust the compensation current so that the magnetic flux gets closer and closer to zero; Step 5.4, output compensation current; The output compensation current is sent to the magnetomotive force balance actuator to achieve zero flux control of the current transformer.

[0049] like Figure 4 As shown in the figure, the verification results of the zero flux control effect are presented. The changes in magnetic flux before and after compensation under different current conditions are compared, intuitively demonstrating the effectiveness of the zero flux control method. It is clearly seen from the figure that the magnetic flux after compensation is much lower than the initial magnetic flux, verifying that the system can maintain a magnetic flux suppression rate of over 94% across the entire current range, fully demonstrating the effectiveness of this method in preventing core saturation.

[0050] Furthermore, the pulse-to-current mapping algorithm employs adaptive frequency compensation technology, which can differentiate control pulses of different frequencies to improve the accuracy of compensation current.

[0051] Furthermore, the asynchronous control event-driven model in this step includes the following key components: Pulse generator: dynamically generates control pulses based on compensation parameters; Event Queue Manager: Manages the ordering and scheduling of asynchronous pulse events; Direct digital-to-analog mapping unit: eliminates the need for a traditional digital-to-analog converter, directly converting pulses into analog compensation current; Magnetic potential detection feedback unit: Real-time detection of magnetic flux, providing closed-loop control feedback.

[0052] like Figure 5 As shown in the figure, the response time and settling time of the traditional synchronous control method and the method of the present invention are compared under different current step amplitudes, intuitively demonstrating the superiority of the method of the present invention in terms of response speed. As can be seen from the figure, the response time and settling time of the method of the present invention are both about 90% lower than those of the traditional method, achieving microsecond-level response, which provides a reliable guarantee for real-time control of high-frequency power electronic systems.

[0053] like Figure 6 As shown, the performance of the traditional synchronization control method and the method of this invention in terms of flux density reduction rate and iron loss reduction rate are compared at different operating frequencies, demonstrating the superior and stable performance of the method of this invention under high-frequency conditions. The figure clearly shows that the traditional method achieves a flux density reduction rate of only 57% at high frequencies (100kHz), while the method of this invention maintains a rate of around 93%, and consistently maintains a high level of over 98% in terms of iron loss reduction, remaining almost unaffected even at extremely high frequencies. This is of great significance for the stable operation of high-frequency power electronic systems.

[0054] A zero-flux control system for a current transformer based on magnetomotive force balance constraints, used to execute the aforementioned zero-flux control method for a current transformer based on magnetomotive force balance constraints, includes: Pulse conversion module: used to convert current signals into pulse sequences and generate event-driven data; Frequency band analysis module: used to implement event-driven wavelet decomposition and multi-band pulse stream output; Frequency coupling calculation module: used to process multi-band pulse streams and construct coupling relationships between frequency components; Magnetic potential balance calculation module: used to generate magnetic potential balance compensation parameters based on frequency coupling relationship; Control execution module: used to convert compensation parameters into control pulses and drive the magnetomotive force balance actuator.

[0055] Here, the present invention provides an implementation example: This embodiment applies to a high-frequency fast charging system for electric vehicles. The system employs a high-frequency DC-DC converter implemented with SiC devices, featuring a switching frequency up to 100kHz, an operating voltage of 750V, and a maximum charging current of 400A. In this application scenario, the current transformer needs to accurately measure and control / protect the high-frequency pulsating current in real time. Traditional current transformers are prone to core saturation at such high frequencies, reducing measurement accuracy and generating additional losses. The method proposed in this invention effectively solves this problem.

[0056] In this example, a high-speed ADC with a sampling rate of 20MHz was used to sample the charging current signal. The analog current signal was converted into a pulse sequence using a pulse density coding algorithm. Taking a current jump at a certain moment as an example, when the current suddenly increases from 50A to 250A, a traditional current transformer requires tens to hundreds of microseconds to stabilize its response. However, after adopting this method, the pulse coding example data for the current jump event is shown in Table 1: Table 1: Example data of pulse codes for current jump events

[0057] When the system detects that the rate of change of current exceeds the preset threshold (>100A / μs), it automatically reduces the time window from 10μs to 2μs and increases the pulse generation frequency to ensure a rapid response to current jumps within 2μs.

[0058] In high-frequency charging systems, the current signal contains multiple frequency components, primarily the fundamental frequency (0Hz DC component), the switching frequency (100kHz), and its harmonic components. By processing the pulse sequence using an event-driven wavelet decomposition algorithm, the system identifies five main frequency components: DC component, 20kHz, 50kHz, 100kHz, and 150kHz. The wavelet decomposition algorithm is pulse-triggered and executes only when the signal changes, achieving highly efficient multi-band pulse stream output.

[0059] Based on the multi-band pulse stream output by the wavelet decomposition algorithm, the system constructed a pulse timing parameter relationship diagram, and the coupling strength matrix (normalized value) between frequency components is shown in Table 2: Table 2: Coupling strength matrix between frequency components (normalized values)

[0060] Analysis of the coupling strength matrix in Table 2 shows that the 100kHz switching frequency has a strong coupling relationship with both the DC and 50kHz components, indicating that the switching frequency has the most significant impact on core saturation. This information is crucial for subsequent calculations of magnetomotive force compensation parameters.

[0061] Based on the frequency coupling relationship obtained in the preceding steps, the system calculates the weight of each frequency component and generates magnetopotential balance compensation parameters. As described in step 4 of implementation method 1, the system collects magnetic flux data and performs normalization processing, standardizing the magnetic flux values ​​to the range [-1, 1] (Table 3 shows the range [0, 1] because the measured magnetic flux values ​​in this example are all positive). Then, the normalized magnetic flux data is combined with the pulse frequency code, and the compensation parameters are calculated using a weighted nonlinear mapping function. An example of the magnetopotential balance compensation parameter calculation is shown in Table 3. Table 3: Examples of Calculation for Magnet potential balance compensation parameters

[0062] Frequency component weight vector in the table The contribution values ​​corresponding to [DC component, 20kHz, 50kHz, 100kHz, 150kHz] are based on the aforementioned coupling strength matrix analysis. Among them, the switching frequency (100kHz) has the largest weight (0.45), which is consistent with the strong coupling relationship between the 100kHz frequency component and other components in the coupling strength matrix.

[0063] The system generates an asynchronous control pulse sequence based on the calculated compensation parameters, and converts the pulses into compensation current through direct digital-to-analog mapping. This process first normalizes the compensation parameters, then maps them to pulse amplitude and frequency spaces, and finally generates the corresponding pulse density through a nonlinear mapping function. The compensation current in Table 4 is the result after processing by the direct digital-to-analog mapping unit, which directly converts the pulses into analog compensation current without the need for a traditional digital-to-analog converter. To verify the zero-flux control effect, the system monitors and records the flux changes in real time. The verification data for the zero-flux control effect are shown in Table 4. Table 4: Verification data of zero flux control effect

[0064] As shown in Table 4, this method achieves effective control of magnetic flux, maintaining a magnetic flux suppression rate of over 94% across the entire current range, effectively preventing core saturation.

[0065] The response speeds of different control methods are compared in Table 5: Table 5: Comparison of response speeds of different control methods

[0066] The data in Table 5 show that, compared with the traditional synchronous control method, the zero flux control method implemented in this invention improves the response speed by about 10 times, shortens the settling time by about 9 times, and reduces the overshoot by about 2 / 3, which fully demonstrates the realization of microsecond-level response capability.

[0067] Table 6 shows a comparison of the core saturation prevention effects of different control methods: Table 6: Comparison of the core saturation prevention effects of different control methods

[0068] Table 6 clearly demonstrates the superiority of the method of the present invention under high-frequency conditions. As the operating frequency increases, the flux control effect of traditional methods decreases sharply, while the method of the present invention can maintain a flux density reduction rate of over 93% even at 100kHz, effectively preventing core saturation and significantly reducing iron losses. This result fully demonstrates the applicability and effectiveness of the present invention in high-frequency power electronic systems.

[0069] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A zero-flux control method for current transformers based on magnetomotive force balance constraints, characterized in that, include: The current signal from the current transformer is converted into a pulse sequence to generate an event-driven data format. The pulse sequence is processed using an event-driven wavelet decomposition algorithm to output a multi-band pulse stream. The analysis of multi-band pulse streams and the calculation of coupling relationships between frequency components include: extracting the pulse occurrence time series for each pair of frequency components; standardizing the time series to eliminate differences in time scales across different frequency bands; calculating the standardized time difference and applying a time kernel function for weighting; normalizing the coupling weights of each frequency component to ensure the sum of the weights is 1; multiplying by the normalized coupling weights to obtain the coupling strength; and constructing a pulse timing parameter relationship graph model based on the coupling strength. Based on the coupling relationship between frequency components, the magnetic potential balance compensation parameters are calculated using a pulse frequency coding algorithm, including: normalizing the magnetic flux data to standardize the magnetic flux values ​​to the range of [-1, 1]; normalizing the pulse frequency code to make the numerical range of each frequency component consistent; and combining the normalized magnetic flux data with the pulse frequency code to calculate the magnetic potential balance compensation parameters through a weighted nonlinear mapping function. The magnetomotive force balance compensation parameters are converted into an asynchronous control pulse sequence, which generates a compensation current and outputs it to the magnetomotive force balance actuator.

2. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The step of converting the current signal of the current transformer into a pulse sequence includes: The original current signal is normalized to normalize the current amplitude to the range of [0, 1]. The pulse density coding algorithm is applied to convert the normalized current signal into a pulse sequence; An adaptive threshold adjustment algorithm is executed. When the rate of change of the current signal exceeds a preset threshold, the time window is dynamically reduced and the pulse generation frequency is increased.

3. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The event-driven wavelet decomposition algorithm utilizes a spiking neural network, which consists of the following components: Input layer: Receives pulse sequences and distributes them to different processing channels; Feature extraction layer: Composed of multiple spiking neurons, each neuron has different time-frequency filter characteristics; Frequency separation layer: Groups and clusters pulses with different frequency characteristics; Output layer: Generates multi-band pulse streams while preserving the time-frequency characteristics of the original signal.

4. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The encoded pulse sequence is input into a neuromorphic processor for further processing, including: The pulse sequence is timestamped to ensure that events are processed in the order of occurrence. A pulse sparsity algorithm is applied to remove redundant pulses and improve processing efficiency. The processed pulse sequence is distributed to different processing units of the neuromorphic processor; Achieving microsecond-level response speed through asynchronous parallel processing; Generate a standardized event stream.

5. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The pulse sequence is processed using an event-driven wavelet decomposition algorithm, and the output multi-band pulse stream includes: The pulse sequence is input into an event-driven wavelet decomposition algorithm, which performs calculations based on pulse triggering and only executes when the signal changes. Perform impulse response transformation by convolving the input pulse with a preset wavelet-like basis function to extract the characteristics of different frequency components; Frequency-selective filtering is applied to allocate pulses to different processing channels based on their time-frequency characteristics; The aggregation process generates a multi-band pulse stream output, with each band corresponding to a frequency component of the current signal.

6. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The step of converting the magnetic potential balance compensation parameters into an asynchronous control pulse sequence includes: The compensation parameters are normalized to eliminate differences in dimensions and numerical ranges between different parameters; The standardized compensation parameters are mapped to the pulse amplitude and frequency space; The corresponding pulse density is generated by using a nonlinear mapping function based on the magnitude and rate of change of the parameters.

7. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The step of generating the compensation current includes: The input control pulses are digitally filtered to remove high-frequency noise. Pulse sequences are standardized to unify their time-domain and frequency-domain characteristics; The standardized pulse sequence is converted into a compensation current amplitude by applying a direct mapping function; The generated current amplitude is calibrated to ensure it meets the input specifications of the actuator; Adjust the timing characteristics of the compensation current according to the pulse timing characteristics.

8. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The step of generating a compensation current and outputting it to the magnetomotive force balance actuator includes: Monitor real-time changes in magnetic flux; Calculate the deviation of the magnetic flux from zero; The compensation current is adjusted in real time to make the magnetic flux approach zero.

9. The zero-flux control method for current transformers based on magnetomotive force balance constraints according to claim 1, characterized in that, The generation process of the asynchronous control pulse sequence utilizes an asynchronous control event-driven model, which includes the following key components: Pulse generator: dynamically generates control pulses based on compensation parameters; Event Queue Manager: Manages the ordering and scheduling of asynchronous pulse events; Direct digital-to-analog mapping unit: converts pulses into analog compensation currents; Magnet potential detection feedback unit: provides closed-loop control feedback.

10. A zero-flux control system for a current transformer based on magnetomotive force balance constraints, characterized in that, The method for executing the zero flux control method for a current transformer based on magnetomotive force balance constraints as described in any one of claims 1-9 includes: Pulse conversion module: used to convert current signals into pulse sequences and generate event-driven data; Frequency band analysis module: used to implement event-driven wavelet decomposition and multi-band pulse stream output; Frequency coupling calculation module: used to process multi-band pulse streams and construct coupling relationships between frequency components; Magnetic potential balance calculation module: used to generate magnetic potential balance compensation parameters based on frequency coupling relationship; Control execution module: used to convert compensation parameters into control pulses and drive the magnetomotive force balance actuator.

Images