A smart transformer current monitoring method and system for a smart grid

Abstract

The application discloses a kind of intelligent transformer current monitoring method and system for smart grid, belong to the field of power system automation.The method comprises: in transformer outlet end, the first original signal of Rogowski coil output and the second original signal of Hall sensor output are synchronously obtained;First current data and second current data are generated by integral restoration and low-pass filtering, differential amplification and temperature compensation respectively;According to real-time current amplitude, full-range dynamic is divided into low-range area, fusion area and high-range area;In fusion area, dynamic weight coefficient is calculated based on logistic function and two-way current is weighted and fused, second current data is used in low-range area, and first current data is used in high-range area;When steady state and current are in low-range area or fusion area, first current data is corrected in real time by integral drift based on second current data.The application is suitable for online monitoring and protection of smart grid transformer.

Description

Technical Field

[0001] This application belongs to the field of power system automation, and specifically relates to an intelligent transformer current monitoring method and system for smart grids. Background Technology

[0002] Transformer current monitoring is a key component of the sensing layer of a smart grid. In existing technologies, traditional iron-core current transformers are prone to magnetic saturation under high current conditions such as short circuits, resulting in distorted output waveforms and loss of fault characteristics. While single Hall effect sensors have good linearity in the low current range, their range is limited, making it difficult to simultaneously meet the dual requirements of precise metering under light loads and fault monitoring under heavy loads.

[0003] While using differential sensors such as Rogowski coils can avoid magnetic saturation, their integration stage suffers from zero-point drift during long-term operation, affecting the accuracy of low-current applications, and their signal-to-noise ratio is insufficient.

[0004] Furthermore, existing solutions often employ simple threshold switching to combine different sensors, resulting in signal step noise during mode switching and affecting waveform continuity. Therefore, achieving current monitoring with no magnetic saturation, full-range coverage, no integral drift, and smooth transition is a pressing technical problem to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent transformer current monitoring method and system for smart grids, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for monitoring current in an intelligent transformer for smart grids includes the following specific steps: Simultaneously acquire the first original signal reflecting the rate of change of current and the second original signal reflecting the magnetic field strength at the transformer output terminals; The first original signal is integrated and low-pass filtered to generate the first current data; the second original signal is differentially amplified, and gain correction and zero-point offset correction are performed in the digital domain based on the real-time temperature information obtained from the sensor package to generate the second current data. Based on the real-time sensed current amplitude, the full-range monitoring range of the transformer is dynamically divided into three continuous functional zones: low-range zone, fusion zone, and high-range zone. When the current is determined to be within the fusion region, a weighted fusion operation is initiated. Dynamic weighting coefficients are calculated in real-time using a logistic function with the current current amplitude as the independent variable. The first current data and the second current data are then weighted and summed based on these dynamic weighting coefficients to obtain the final fused current data. When the current is determined to be within the low-range region, the second current data is directly used as the final current data. When the current is determined to be within the high-range region, the first current data is directly used as the final current data. When the transformer is in steady-state operation and the current is in the low range region or the fusion region, the second current data is used as a reference value to perform real-time correction of the integral drift of the first current data.

[0007] Furthermore, the first original signal is induced and output by a coreless Rogowski coil, which is a hollow ring structure made of non-ferromagnetic material with a thermal expansion coefficient of less than 20 ppm / degree Celsius. The second original signal is output by a magnetically balanced closed-loop Hall sensor, which operates in a zero-flux mode and has a built-in ring magnetic core with an air gap, a Hall element placed in the air gap, and a feedback winding wound on the magnetic core. The two Hall sensors are arranged in a dual-redundant symmetrical manner on both sides of the transformer output busbar, and their output voltage signals are arithmetically averaged to serve as the second original signal.

[0008] Furthermore, the Hall sensor is encapsulated in a metal box made of a highly conductive metal, which is connected at a single point on the printed circuit board level to form a Faraday cage, with only a narrow gap in the direction of the magnetic core opening to shield stray electromagnetic interference.

[0009] Furthermore, the second original signal undergoes gain correction and zero-point offset correction in the digital domain, specifically including: The microcontroller acquires the voltage across the thermistor embedded inside the Hall sensor near the Hall element in real time through a voltage divider circuit, and calculates the real-time value of the current ambient temperature based on the voltage divider ratio and the thermistor's calibration table. The microcontroller has a two-dimensional temperature compensation coefficient matrix pre-stored inside. This matrix is ​​obtained by placing the Hall sensor in a high and low temperature test chamber for temperature cycling and recording the relationship curve between magnetic induction intensity and Hall output voltage at each temperature point, and then using least squares fitting to obtain the gain correction coefficient and zero-point offset coefficient at each temperature point. In actual operation, the microcontroller performs linear interpolation on the two-dimensional temperature compensation coefficient matrix based on the real-time temperature to obtain the corresponding gain correction coefficient and zero-point offset coefficient, and completes digital correction according to the formula.

[0010] Furthermore, the full-range monitoring range of the transformer is dynamically divided into a low-range zone, a convergence zone, and a high-range zone, specifically including: A first threshold and a second threshold are preset, wherein the first threshold is set to 20% of the transformer's rated current and the second threshold is set to 80% of the transformer's rated current; The estimated value of the current amplitude is compared with the first threshold and the second threshold respectively: if the estimated value is less than the first threshold, it is determined that the current range is low; if the estimated value is greater than or equal to the first threshold and less than or equal to the second threshold, it is determined that the current range is fusion; if the estimated value is greater than the second threshold, it is determined that the current range is high. The first threshold and the second threshold can be remotely and dynamically adjusted by receiving threshold update instructions from the edge computing gateway via an industrial fieldbus.

[0011] Furthermore, the weighted fusion calculation is performed according to the following formula: in, For the final current data after fusion, This is the first current data. This is the second current data. These are the dynamic weighting coefficients calculated in real time based on the aforementioned logistic function; The logistic function is: in, For dynamic weighting coefficients, To obtain the current amplitude estimate by taking the absolute value of the second current data, The center amplitude point of the fusion region is determined jointly by the first threshold and the second threshold. The slope constant is used to control the smoothness of the weight transition; The slope constant The value of makes the logistic function approach 0 or 1 at the boundary of the fusion region, so as to achieve a smooth transition from fully accepting the second current data to fully accepting the first current data.

[0012] Furthermore, real-time correction of integral drift is performed on the first current data, specifically including: The calibration process is initiated when the transformer enters steady-state operation and the current is located in the low range region or the fusion region. The condition for determining that the transformer has entered steady state is that the fluctuation range of the second current data is less than the preset steady-state determination threshold within N consecutive sampling periods. After the calibration process is started, the DC component of the instantaneous residual between the first current data and the second current data is extracted by a moving average filter to obtain the drift error estimate. A reverse correction is applied to the first current data based on the drift error estimate to eliminate integral drift; When the transformer current crosses the second threshold and enters the high range region, the current drift error estimate is frozen, and the frozen value is used to correct the first current data in subsequent processing.

[0013] Furthermore, it also includes a self-diagnostic step: When it is determined that the current is in the fusion region, the absolute value of the residual between the first current data and the second current data is calculated; When the absolute value of the residual is continuously greater than the preset sensor deviation threshold, the abnormal state is accumulated and timed, and after the accumulated time reaches the preset continuous judgment time, a sensor health warning signal is automatically issued.

[0014] Furthermore, harmonic analysis is performed in parallel on the acquired final current data: Extract data frames of a preset time length from continuous sampling data, and apply Hanning window weighting to the data frames; The discrete spectrum of the windowed data frame is calculated using the Fast Fourier Transform algorithm. The amplitude of the fundamental component and the amplitude of specific order harmonic components are extracted from it, and the total harmonic distortion rate is calculated. The calculated total harmonic distortion rate is compared with a pre-stored transformer loss-harmonic correlation table to assess the current operating loss level of the transformer online.

[0015] A smart transformer current monitoring system for smart grids includes: The signal acquisition unit is used to simultaneously acquire a first raw signal reflecting the rate of change of current and a second raw signal reflecting the magnetic field strength at the transformer output terminal. The signal conditioning and conversion unit is used to perform integration and low-pass filtering on the first original signal to generate the first current data, and to perform differential amplification and temperature compensation on the second original signal to generate the second current data. The dynamic range division unit is used to dynamically divide the full-range monitoring range into a low-range zone, a fusion zone, and a high-range zone based on the real-time sensed current amplitude. The weighted fusion unit is used to calculate the dynamic weighting coefficients in real time within the fusion zone based on the logistic function with the current current amplitude as the independent variable, and to perform a weighted summation of the first current data and the second current data to output the final fused current data. A real-time integral drift calibration unit is used to correct the integral drift of the first current data in real time, based on the second current data, when the transformer is operating in steady state and the current is in the low range or convergence region; and, The microcontroller integrates a hardware floating-point arithmetic unit to execute the signal processing, interval partitioning, weight fusion, and drift calibration algorithms of the aforementioned units, and uploads the final current data in real time via the industrial fieldbus protocol.

[0016] In summary, this application includes at least one of the following beneficial technical effects: 1. This invention completely solves the problem of core magnetic saturation that easily occurs in traditional electromagnetic current transformers under fault conditions by introducing a coreless Rogowski coil as a sensing means for large and transient currents. This ensures that the secondary waveform remains undistorted even under a preset multiple of the rated current. Simultaneously, combined with a high-precision magnetically balanced Hall effect sensor, this invention successfully constructs a full-range monitoring system covering everything from zero current to extreme short-circuit current. The linear measurement range is significantly improved compared to traditional solutions, achieving perfect compatibility between light-load metering and fault monitoring.

[0017] 2. This invention provides a signal-to-noise ratio and measurement accuracy far exceeding traditional Rogowski coil solutions in the low-current range by implementing digital temperature compensation and amplification processing on the Hall sensor. By utilizing the steady-state high-precision characteristics of the Hall sensor to correct the zero-point drift of the Rogowski coil integrator in real time, the impact of accumulated errors on long-term operational stability is eliminated. This complementary dual-mode architecture ensures that the system achieves the predetermined industrial-grade metrological accuracy across the entire measurement range.

[0018] 3. This invention innovatively employs a type-function-based weighted fusion algorithm to replace the traditional threshold hard-switching mode, effectively eliminating signal jumps and noise caused by differences in sensor physical principles. Within the fusion range of load fluctuations, the system can achieve seamless integration of different modal data through weight allocation, ensuring the continuity and integrity of the current waveform. Simultaneously, because the algorithm runs efficiently in a high-performance microcontroller, the system possesses excellent transient response characteristics, providing accurate criteria for relay protection devices at a predetermined time level, ensuring the safe operation of the smart grid.

[0019] 4. This invention integrates a self-diagnostic mechanism and harmonic analysis function, enabling not only real-time output of high-quality current signals but also a comprehensive assessment of the sensor's own operating status and the transformer's power quality. This highly integrated intelligent monitoring solution reduces the number of external secondary devices and simplifies the hardware layout at the transformer's output terminals. Through the combination of digital output and edge computing, this invention provides a solid data foundation for remote operation and maintenance, load forecasting, and lifespan management of smart grids, significantly reducing the total lifecycle operating costs of grid assets. Attached Figure Description

[0020] Figure 1 A schematic diagram of the overall technical solution for an intelligent transformer current monitoring method used in smart grids; Figure 2 This is a schematic diagram illustrating the core principle of weighted fusion based on dynamic interval partitioning and logistic functions; Figure 3 The flowchart shows the logic flow for the synchronous acquisition and preprocessing stage of dual-channel raw signals. Figure 4 A schematic diagram of multi-level interaction and data flow for real-time integral drift correction based on steady-state reference data; Figure 5 This is a flowchart of a smart transformer current monitoring method for smart grids. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with the appendix. Figure 1 To be continued Figure 5 The present invention will be further described in detail below with reference to specific embodiments.

[0022] Firstly, the intelligent transformer current monitoring method for smart grids disclosed in this application relies on the deep integration of high-precision sensing hardware and real-time signal processing algorithms for its execution logic. By performing dual-modal sensing of the full-range current under transformer operating conditions, accurate coverage from weak load current to extreme short-circuit fault current is ensured.

[0023] The first step, S1, involves simultaneously deploying a Rogowski coil and a magnetically balanced Hall sensor at the transformer's output terminals to acquire a first raw signal reflecting the rate of change of current and a second raw signal reflecting the magnetic field strength, respectively. The aim is to cover the entire current range from its physical source and provide dual-channel data with high time synchronization accuracy for subsequent processing. This process includes the following steps.

[0024] Step S101, Physical configuration and installation of the Rogowski coil: The coreless Rogowski coil is precisely installed on the low-voltage side busbar of the transformer. The coil adopts a hollow toroidal structure made of non-ferromagnetic material, and its frame is manufactured by high-precision injection molding process using engineering plastic with high insulation strength and low coefficient of thermal expansion, thus physically eliminating the possibility of magnetic saturation of the iron core.

[0025] To ensure a constant mutual inductance coefficient over the long term, the inner diameter of the coil is customized according to the actual cross-sectional dimensions of the busbar, maintaining a preset uniform air gap between the inner wall of the coil and the outer surface of the busbar, such as 5 mm to 10 mm, to balance insulation and signal strength. The windings are arranged with uniform pitch using a precision winding machine, and the coil turn density is set to at least one turn for every degree of central angle, to eliminate measurement errors caused by uneven winding distribution and ensure the spatial resolution of the output signal.

[0026] The selected skeleton material has a coefficient of thermal expansion of less than 20 ppm / degree Celsius to ensure that the geometry of the coil remains highly stable in complex electromagnetic environments and extreme temperature changes ranging from -40 degrees Celsius to 85 degrees Celsius.

[0027] Step S102, signal induction of the Rogowski coil: when a current is generated on the primary side, the changing current excites a changing magnetic field in space. The aforementioned coreless Rogowski coil cuts the magnetic lines of force, and an induced electromotive force is generated at both ends of its winding.

[0028] Especially when a short-circuit fault triggers an impact of a preset multiple, such as 20 times or more of the rated current, because there is no iron core, the coil magnetic circuit operates entirely in the unsaturated linear region, and its output primary signal can completely capture the high-frequency transient current component. This primary signal is an induced electromotive force, the magnitude of which is proportional to the rate of change of the primary current with respect to time. This relationship is defined by the following formula: in, The instantaneous electromotive force of the first original signal is expressed in volts. The mutual inductance between the coil and the busbar, determined through calibration, is expressed in henries. This is the instantaneous value of the primary current, measured in amperes. Time, in seconds; It is the derivative of the current with respect to time, and its unit is amperes per second.

[0029] Step S103: Arrangement and closed-loop structure of the magnetic balance Hall sensor. A magnetic balance closed-loop Hall sensor is installed at a predetermined position near the outgoing busbar. This position needs to be verified through simulation or actual measurement to ensure effective magnetization at the opening of the sensor core.

[0030] The sensor incorporates a toroidal magnetic core with an air gap, made of high-permeability soft magnetic material, a Hall element placed within the air gap, a signal amplification circuit, and a feedback winding wound around the magnetic core. The sensor operates in a zero-flux mode: the magnetic field generated by the primary current creates a magnetic flux within the core. The Hall element detects this flux and outputs a voltage signal proportional to it. This voltage signal is amplified by a circuit with a fixed gain of G, driving the feedback winding to generate a reverse current. This reverse current flows through the Nf-turn feedback winding, establishing a reverse magnetic field within the core that is opposite in direction and equal in magnitude to the magnetic field generated by the primary current. This forces the net magnetic flux within the core to continuously approach zero.

[0031] At this point, the instantaneous value of the feedback current is proportional to the instantaneous value of the primary current. By measuring the voltage drop across the precision sampling resistor Rs generated by this feedback current, a second original signal accurately reflecting the instantaneous value of the primary current can be obtained. This closed-loop structure significantly improves the linearity and measurement accuracy of the sensor.

[0032] Step S104: Redundancy configuration and anti-interference shielding of Hall sensors. In order to offset the measurement error caused by the busbar position deviation, the Hall sensors adopt a dual redundancy configuration, with two independent magnetic balance closed-loop Hall sensors symmetrically distributed on both sides of the busbar.

[0033] The feedback currents of both components flow through their respective precision sampling resistors, and the output voltage signals are arithmetically averaged. This average value is then used as the final second raw signal. The entire sensor system, including the magnetic core, Hall element, and preamplifier circuit, is encapsulated in a metal box made of highly conductive metal with electrostatic shielding.

[0034] The metal box is connected to the system ground at a single point on the printed circuit board to form an effective Faraday cage. It has only a narrow gap in the direction of the magnetic core focusing opening, which effectively shields stray electromagnetic interference from transformer windings and other external high-voltage equipment, and significantly improves the signal-to-noise ratio in the low current range.

[0035] Step S105: High-precision synchronous acquisition of dual-channel signals. The system is equipped with a microcontroller. This microcontroller has a built-in high-precision real-time clock and acts as a slave clock for the IEEE 1588 protocol or directly receives wireless synchronization trigger signals to calibrate its local clock and synchronize it with the master clock of other monitoring nodes in the system. The synchronization process ensures that the clock deviation of each node is within 1 microsecond.

[0036] Meanwhile, at least two analog-to-digital conversion channels connected to the microcontroller sample the first raw signal output by the Rogowski coil and the second raw signal output by the Hall sensor in parallel and at equal intervals at a sampling frequency of not less than 12.8kHz under the unified beat of the synchronous clock. The sampled values ​​are then packaged into digital signal frames with their respective channel source identifiers and timestamps accurate to 1 microsecond, and sent to the subsequent processing stage.

[0037] In summary, step S1 completes the entire process from physical sensor deployment to synchronous acquisition of dual-mode signals. The Rogowski coil and Hall sensor complement each other in measurement range and are strictly aligned in time, providing the entire monitoring system with complete and faithful raw data ranging from weak light load currents to extreme short-circuit fault currents. These synchronously acquired dual signals will directly enter the conditioning and preprocessing stage in step S2, where they will undergo integration, amplification, and temperature compensation to generate the first current data and the second current data, respectively.

[0038] The next step, S2, takes the first and second raw signals output from step S1 and conditions and transforms them respectively to generate first and second current data that can be directly used by subsequent algorithms. The entire preprocessing process, taking into account the distinctly different signal characteristics of the two sensors, designs independent processing chains to ensure data accuracy from the source. Specifically, it includes the following steps.

[0039] Step S201: The active integrator circuit for the first original signal is constructed by inputting the weak induced electromotive force e(t) output from the Rogowski coil into the active integrator circuit, which is composed of a high input impedance operational amplifier. This integrator circuit employs three hardware-level measures to suppress drift: an automatic zero-calibration operational amplifier with an input offset voltage drift of less than 0.05 microvolts per degree Celsius is selected; and a low-loss polypropylene film capacitor with a loss tangent of less than 0.1% is selected as the integrating capacitor.

[0040] The integrating resistor is a high-precision thin-film resistor with a temperature coefficient of less than 10 ppm per degree Celsius. The parameters of these three types of devices are matched to suppress temperature drift and long-term drift to the greatest extent possible at the physical level.

[0041] Step S202, Integration and Low-Pass Filtering: The aforementioned integrator circuit performs integration on the first original signal e(t), and its transfer function restores the electromotive force signal, which is proportional to the current derivative, into a current waveform signal. The mathematical relationship of the integration operation is defined as follows: in, This is the voltage signal output by the integrating circuit, measured in volts. The resistance value of the integration input resistor, in ohms; This is the capacitance value of the integral feedback capacitor, in farads. The instantaneous electromotive force value of the first original signal in step S102 is expressed in volts. Time, in seconds; integral symbol This represents the accumulation of the electromotive force signal over time.

[0042] A second-order active low-pass filter is connected in series after the integrator circuit. The -3dB cutoff frequency of this filter is set to 5kHz. This frequency value can cover all harmonics of the transformer up to the 50th order and transient components with rise times on the order of tens of microseconds, while effectively filtering out high-frequency spurious interference. The output signal after integration and filtering is the first current data, denoted as... .

[0043] Step S203: Differential amplification and range matching of the second raw signal. The second raw signal output from the Hall sensor is linearly amplified using a differential amplifier circuit composed of a precision instrumentation amplifier. The gain of this instrumentation amplifier is set by an external gain setting resistor with an accuracy of 0.1%. Precise settings ensure that the peak-to-peak value of the amplified signal exactly matches the full-scale input range of the subsequent analog-to-digital converter.

[0044] For example, if the full-scale output voltage of the Hall sensor is ±100mV and the input range of the analog-to-digital converter is ±5V, then the gain should be set to 50 times.

[0045] Step S204: Real-time acquisition and digital temperature compensation of temperature information. A negative temperature coefficient thermistor is pre-embedded inside the Hall sensor package near the Hall element. The microcontroller collects the voltage across the thermistor in real time through a voltage divider circuit and calculates the real-time value of the current ambient temperature based on the voltage divider ratio and the thermistor's calibration table.

[0046] The microcontroller pre-stores a two-dimensional temperature compensation coefficient matrix. The matrix is ​​constructed by placing the Hall sensor in a high and low temperature test chamber and cycling the temperature in 5-degree increments within the range of -40 degrees Celsius to 85 degrees Celsius. The relationship curve between the magnetic induction intensity and the Hall output voltage is recorded at each temperature point. The gain correction coefficient and zero offset coefficient at each temperature point are obtained by least squares fitting.

[0047] In actual operation, the microcontroller performs linear interpolation in the coefficient matrix based on the measured temperature to obtain the corresponding gain correction coefficient. and zero offset coefficient Then, complete the digital correction using the following formula: in, The corrected current value is in amperes. The current reading before correction is given, in amperes. This is the gain correction factor, which is dimensionless. This is the zero-point offset coefficient, in amperes. The second current data obtained after this processing is denoted as... It exhibits consistent gain characteristics and zero-point stability over an ambient temperature range of -40°C to 85°C.

[0048] In summary, through step S2, the preprocessing stage converts the first raw signal output by the Rogowski coil into first current data suitable for full-range current characterization, and converts the second raw signal output by the Hall sensor into high-precision, temperature-drift-compensated second current data. The two data streams maintain the microsecond-level synchronization established in step S105, providing accurate, synchronized input data with eliminated inherent errors in their respective sensing links for the subsequent dynamic interval division in step S3 and the weighted fusion algorithm in step S4.

[0049] The next step, S3, involves dynamically dividing the full-range monitoring range of the transformer into three functional zones based on the real-time sensed current amplitude. This allows sensor data with different characteristics to be rationally utilized within their respective advantageous regions, providing decision boundaries for the subsequent weighted fusion algorithm. The specific steps are as follows: Step S301, threshold setting and interval definition: The system presets two current thresholds, namely the first threshold... and the second threshold The first threshold is set at 20% of the transformer's rated current, and the second threshold is set at 80% of the transformer's rated current. These two thresholds divide the entire monitoring range into three consecutive current intervals.

[0050] The low-range region corresponds to operating conditions where the current amplitude is less than the first threshold, i.e., a light-load operating state where the current is less than 20% of the rated current. Within this range, the primary current value is relatively small, and the signal-to-noise ratio of the first raw signal output by the Rogowski coil decreases significantly after integration. However, the Hall sensor, with its magnetic balance closed-loop structure, can provide measurement accuracy and signal-to-noise ratio far superior to that of the Rogowski coil channel.

[0051] The fusion zone corresponds to operating conditions where the current amplitude is between the first and second thresholds, i.e., a normal load condition with the current between 20% and 80% of the rated current. Within this range, both the Rogowski coil and the Hall sensor are in their respective good linear operating regions, and both current data have reliable measurement accuracy, providing conditions for data fusion.

[0052] The high-range region corresponds to operating conditions where the current amplitude exceeds the second threshold, i.e., heavy load or fault short-circuit conditions where the current exceeds 80% of the rated current. Within this range, continued current increase may cause the Hall sensor's magnetic core to gradually approach the saturation region, and its linearity begins to deteriorate. However, due to its coreless structure, the Rogowski coil can always maintain a linear response and continue to provide high-precision current change characterization.

[0053] Step S302, Interval Determination Logic: In actual operation, the microcontroller acquires an estimated value of the current current amplitude in each sampling period. This estimated value is taken as the absolute value of the second current data output in step S2, i.e., |I_HS|, and is used as the basis for interval determination.

[0054] The microcontroller compares the estimated value with the first threshold and the second threshold: if the estimated value is less than the first threshold, it is determined that the current range is low; if the estimated value is greater than or equal to the first threshold and less than or equal to the second threshold, it is determined that the current range is fusion; if the estimated value is greater than the second threshold, it is determined that the current range is high.

[0055] Step S303: Remote dynamic adjustment of thresholds. The first and second thresholds are not fixed preset values, but can be dynamically adjusted remotely according to the actual operating mode of the power grid. The microcontroller receives a threshold update command from the edge computing gateway or host computer system via the industrial fieldbus. The command contains the specific values ​​of the adjusted first and second thresholds. After receiving the update command, the microcontroller writes it to its internal non-volatile memory and uses the new thresholds for interval determination starting from the next sampling period.

[0056] The practical significance of this mechanism lies in adapting to the needs of different scenarios. For example, in the independent operation mode of a microgrid, the connected power of distributed power sources may only be a few kilowatts, corresponding to a primary current far lower than the conventional light-load level in grid-connected mode. In this case, the system can remotely adjust the first threshold from the default 20% of the rated current to 10% or even lower, thereby enhancing the ability to sense the weak connected current of distributed power sources and ensuring that low-current conditions can still be accurately classified into the low-range zone and measured by Hall sensors.

[0057] In summary, through step S3, the dynamic range division stage divides the full-range monitoring range into three functional areas—a low-range area, a fusion area, and a high-range area—based on the real-time current amplitude, and allows for remote dynamic optimization of threshold parameters according to the power grid operation scenario. This division provides clear fusion range boundaries for the weighted fusion algorithm in step S4, ensuring that weight coefficient calculation is only activated within the fusion area, while single sensor data is directly used in the low-range and high-range areas, avoiding unnecessary fusion computation overhead.

[0058] For step S4, within the fusion region determined in step S3, dynamic weighting coefficients are calculated using a logistic function, and the first current data and the second current data are weighted and summed to achieve a smooth transition between the two sensor modes and eliminate signal step noise caused by traditional hard switching. Specifically, the steps are as follows.

[0059] Step S401, definition of the fusion formula: When step S302 determines that the current is in the fusion region, the microcontroller performs a weighted fusion operation once in each sampling period. The final fused current data is calculated by the following formula: in, The final output current data after fusion is in amperes; The first current data generated in step S2 is the current value of the Rogowski coil channel after integration and filtering, in amperes; The second current data generated in step S2 is the current value of the Hall sensor channel after amplification and temperature compensation, in amperes. The dynamic weighting coefficients assigned to the first current data are dimensionless and their values ​​vary continuously between 0 and 1.

[0060] when When the value is 1, the fusion formula degenerates into The system fully accepts the data from the Rogowski coil channel; when When the value is 0, the fusion formula degenerates into The system fully accepts the data from the Hall sensor channel; when When the value is between 0 and 1, the two data streams are mixed and output proportionally.

[0061] Step S402, Calculation method of weighting coefficients, dynamic weighting coefficients It is not a fixed constant, but rather calculated in real time based on the current current amplitude using a logistic function. The mathematical expression of this function is: in, The current amplitude is the weighting coefficient, which is dimensionless and ranges from 0 to 1. This is an estimate of the current amplitude, in amperes. The central amplitude point of the fusion zone, in amperes; The slope constant used to control the smoothness of the weight transition, with units of 1 / Ampere; This represents an exponential function with the natural constant e as its base.

[0062] Current amplitude estimate The method for obtaining the data is as follows: The microcontroller obtains the second current data output in step S2. The absolute value, that is This is used as an estimate of the current amplitude. This selection is consistent with the basis for interval determination in step S302, ensuring that the input quantities for interval division and weight calculation are consistent.

[0063] Central amplitude point of the integration zone It is determined by the first threshold and the second threshold, and its calculation formula is as follows: ,in The first threshold, This is the second threshold. According to the default value given in step S301, if the rated current is... ,but , ,at this time .

[0064] Step S403, the basis for setting the slope constant, slope constant The value of directly affects the steepness of the transition of the weight curve in the fusion region. The larger the value, the more rapid the transition of the weight from 0 to 1, and the closer the curve is to a step switch; The smaller the value, the smoother the transition, and the wider the mixing range of the two sensor data. The specific value needs to be determined in conjunction with the actual width of the fusion region, so that the logistic curve is sufficiently close to 0 or 1 at the boundary of the fusion region, and the main transition is completed near the center of the fusion region.

[0065] One optional setting method is to take the width of the fusion region. ,make By default threshold ,but Under this setting, when hour, This means the weights have essentially decayed to zero, and the system primarily relies on Hall sensor data; when hour, This means the weight has essentially increased to 1, and the system primarily relies on Rogowski coil data; in the center of the fusion zone... place, The two types of sensor data are mixed with equal weights.

[0066] Step S404: Real-time updating and fusion of weight coefficients are performed, with the update frequency of the weight coefficients synchronized with the system sampling frequency. The microcontroller performs the following operations in each sampling cycle: obtain the latest data from step S2. Value and calculate According to the determination result of step S302, if the current range is low, directly... Set to 0; if in the high range region, directly set to 0. Set to 1; if in the fusion zone, call the formula in step S402 to calculate in real time. The precise value. Then, this... Substituting the values ​​into the fusion formula in step S401, we obtain... Then output it to the next processing stage.

[0067] This mechanism ensures that during transient processes with rapid load fluctuations, even if the current amplitude spans different ranges within a single sampling period, the fused signal can continuously track the true trend of the primary current without signal jumps caused by mode switching.

[0068] In summary, through step S4, the weighted fusion stage, centered on the logistic function, achieves a gradual weighted mixing of the first and second current data within the fusion region, establishing a smooth and continuous transition channel between the two sensor modes. In conjunction with step S3, the low-range and high-range regions directly utilize single-sensor data, while the fusion region performs weighted fusion, forming a complete partitioned adaptive processing strategy. The final current data output by this strategy... It will proceed directly to step S5, which is used for integral drift calibration, harmonic analysis, and bus output.

[0069] Finally, step S5 completes two core tasks: first, outputting the final current data generated in step S4 for subsequent analysis; and second, utilizing the steady-state high-precision characteristics of the second current data to perform real-time drift correction on the integral element upon which the first current data depends, ensuring that the Rogowski coil channel maintains an accurate initial state throughout long-term operation. Specifically, this includes the following steps: Step S501: The mechanism of integral drift and calibration triggering conditions. During long-term operation, the active integrator circuit of the Rogowski coil will gradually accumulate DC bias error at the integrator output due to the combined effects of factors such as the input bias current of the operational amplifier, the leakage current of the integrating capacitor, and temperature changes. This manifests as zero-point drift. This drift reduces the accuracy of the first current data in the low current range and affects the reliability of the initial state when entering the high range region.

[0070] To address this issue, the system initiates the calibration process when the transformer is operating in steady state and the current is in the low-range or convergence region. The microcontroller determines that steady state has been reached based on the following condition: within N consecutive sampling periods, the second current data... The fluctuation range is less than the preset steady-state judgment threshold, which is set to 2% of the rated current. The value of N is determined based on the sampling frequency and the characteristics of the power grid inertia. For example, at a sampling rate of 12.8kHz, N is 1280, corresponding to a continuous observation window of 100 milliseconds. During calibration, the second current data serves as a reference value unaffected by integral drift.

[0071] Step S502, Identification and Quantization of Accumulated Error: When the calibration trigger condition is met, the microcontroller processes the first current data. With the second current data Real-time comparison is performed. Since the current is in the low-range or convergence zone at this time, the two data streams should ideally reflect the same instantaneous current value. The deviation caused by integral drift manifests as... and The persistent difference in DC components between them.

[0072] The microcontroller extracts the DC component of this difference using a moving average filter. Specifically, it calculates the instantaneous residual between the two data streams at each sampling period. Where t is the current sampling time; the drift error estimate at the current time is obtained by arithmetically averaging the residual values ​​of M consecutive sampling periods. The value of M is the same as that of N, i.e., M is 1280. This moving average calculation filters out AC components and random noise, retaining only the stable DC deviation. The calculation relationship is as follows: in, This is an estimate of the integral drift error, in amperes. This is the length of the moving average window, i.e., the number of sampling points participating in the averaging. For at any time The instantaneous residual value, in amperes; The sampling period is the reciprocal of the sampling frequency, expressed in seconds. The index variable is used for summation, ranging from 0 to... .

[0073] Step S503: Dynamic adjustment of the initial integral value. The microcontroller adjusts the drift error estimate obtained by quantization in step S502. The initial state of the integrator circuit is dynamically compensated. In the digital domain, this compensation is achieved by directly applying a reverse correction to the first current data. The correction formula is: in, The first current data is after drift correction, and the unit is amperes; The current data before correction is in amperes. The current drift error estimate is calculated in step S502, and the unit is amperes.

[0074] This correction operation is continuously performed within each sampling period, once... As the calibration value approaches zero, the correction value also returns to zero. When the transformer transitions from steady state to heavy load or fault range, i.e., when the current crosses the second threshold and enters the high-range region, the microcontroller freezes the current reading. The value is then used to correct the first current data. This is because during the transient process of the fault current, the Hall sensor may enter the nonlinear region and is no longer suitable as a reference, while the initial state of the Rogowski coil channel has been calibrated to an accurate value before entering the high-range region.

[0075] Step S504: Output and upload of the final current data to the bus. The microcontroller will process the full-range final current data after the fusion processing in step S4 and the drift correction in this step. The data is uploaded in real time to the edge computing gateway of the smart grid via the industrial fieldbus protocol. The output data maintains the same sampling frequency as in step S105, i.e., no less than 12.8kHz, and the data bit width of each sampling point is 24 bits. The bus protocol adopts the Modbus-RTU or IEC 61850 standard message format. The data frame carries the instantaneous current value, timestamp, and current range interval identifier to meet the dual requirements of precision metering and fast relay protection for data real-time performance and integrity.

[0076] In summary, step S5, the data output and calibration stage, completes two key functions: first, it establishes a real-time integral drift correction closed loop based on the second current data and using the moving average residual as the error metric, ensuring that the Rogowski coil channel maintains an accurate initial state before long-term operation and entering the heavy-load range; second, it uploads the fused full-range current signal to the edge computing gateway in a high sampling rate and high bit-width digital format. The high-quality current data output in this stage will simultaneously support subsequent harmonic analysis and sensor self-diagnosis functions, providing a reliable data foundation for a comprehensive assessment of the transformer's operating status.

[0077] In addition, after the final current data is output in step S5, harmonic analysis and self-diagnosis functions are performed in parallel. The two functions share the same full-range current signal and are completed by the microcontroller with integrated high-performance floating-point arithmetic unit within the same operation cycle.

[0078] The microcontroller extracts a data frame of length L from the continuous sampling data within each harmonic analysis cycle. The value of L is chosen such that the data frame corresponds to a 100-millisecond time window; at a sampling rate of 12.8kHz, L is 1280 points. To suppress spectral leakage, a Hanning window weighting is applied to the extracted data frame. The window function is defined as: in, is the coefficient of the window function at the nth sampling point, which is dimensionless; n is the sampling point number; L is the total number of sampling points in the data frame.

[0079] The windowed data sequence is processed using a radix-2 decimation-time fast Fourier transform algorithm to calculate its discrete spectrum. The microcontroller then extracts the fundamental component amplitude from the spectrum. And the amplitudes of specific order harmonic components such as the 3rd, 5th, and 7th harmonics. Where h takes odd orders such as 3, 5, 7, etc. The total harmonic distortion rate is calculated by the following formula: Wherein, THD is the total harmonic distortion rate, expressed as a percentage; This represents the amplitude of the fundamental component, in amperes. The summation value is the amplitude of the h-th harmonic component, in amperes; summation symbol. This represents the summation of the squared amplitudes of each harmonic order h that needs to be statistically analyzed.

[0080] The microcontroller compares the calculated THD value with a pre-stored transformer loss-harmonic correlation table. This table, established through offline testing, records the correspondence between THD values ​​and additional iron losses and winding eddy current losses under different load levels. Based on the comparison results, the microcontroller assesses the current operating loss level of the transformer online and uploads the THD value and loss assessment results along with the current data frame.

[0081] The system's self-diagnostic mechanism utilizes sensor data comparison within the fusion zone. When step S302 determines that the current is within the fusion zone, the microcontroller calculates the first current data in each sampling cycle. With the second current data The absolute value of the residuals between: Where D is the absolute value of the residual in the current sampling period, in amperes.

[0082] An internal counter is set up in the microcontroller. The counter increments when D continuously exceeds a preset sensor deviation threshold, and resets to zero once D falls below the threshold. The sensor deviation threshold is set to 1% based on a ratio principle, meaning that when D... An abnormality is detected when the counter's accumulated value reaches a preset number of consecutive detections (C). The system automatically issues a sensor health warning signal. The value of C ensures that the abnormal state must persist for at least 50 milliseconds before triggering the warning; at a sampling rate of 12.8kHz, C corresponds to 640.

[0083] The warning signal is uploaded to the edge computing gateway in the form of an independent message via the industrial fieldbus. The message content includes the current deviation amplitude, duration stamp, and the sensor channel identifier that is recommended to be checked, so that maintenance personnel can remotely diagnose whether the sensor has experienced performance degradation or loose installation.

[0084] The harmonic analysis results, self-diagnostic status, and final current data output from step S4 are all uploaded through the same industrial fieldbus interface, supporting Modbus-RTU or IEC 61850 standards. The sampling frequency is maintained above 12.8kHz, and the data width is 24 bits. All signal processing, interval division, weight fusion, drift calibration, harmonic analysis, and self-diagnostic algorithms are serially completed within a single computation cycle by a microcontroller with an integrated hardware floating-point unit. The total delay from sampled data input to diagnostic result output is less than 50 microseconds.

[0085] On the other hand, the intelligent transformer current monitoring system for smart grids disclosed in this application includes: The signal acquisition unit is used to simultaneously acquire a first raw signal reflecting the rate of change of current and a second raw signal reflecting the magnetic field strength at the transformer output terminal. The signal conditioning and conversion unit is used to perform integration and low-pass filtering on the first original signal to generate the first current data, and to perform differential amplification and temperature compensation on the second original signal to generate the second current data. The dynamic range division unit is used to dynamically divide the full-range monitoring range into a low-range zone, a fusion zone, and a high-range zone based on the real-time sensed current amplitude. The weighted fusion unit is used to calculate the dynamic weighting coefficients in real time within the fusion zone based on the logistic function with the current current amplitude as the independent variable, and to perform a weighted summation of the first current data and the second current data to output the final fused current data. A real-time integral drift calibration unit is used to correct the integral drift of the first current data in real time, based on the second current data, when the transformer is operating in steady state and the current is in the low range or convergence region; and, The microcontroller integrates a hardware floating-point arithmetic unit to execute the signal processing, interval partitioning, weight fusion, and drift calibration algorithms of the aforementioned units, and uploads the final current data in real time via the industrial fieldbus protocol.

[0086] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0087] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for intelligent transformer current monitoring in smart grids, characterized in that, Includes the following steps: A first raw signal reflecting the rate of change of current and a second raw signal reflecting the magnetic field strength are simultaneously acquired at the transformer output terminals. The first raw signal is induced and output by a coreless Rogowski coil. The Rogowski coil adopts a hollow ring structure made of non-ferromagnetic material, and the coefficient of thermal expansion of its skeleton material is less than 20 ppm / degree Celsius. The second raw signal is output by a magnetically balanced closed-loop Hall sensor. The Hall sensor adopts a zero-flux operating mode and has a built-in ring magnetic core with an air gap, a Hall element placed in the air gap, and a feedback winding wound on the magnetic core. The two Hall sensors are distributed on both sides of the transformer output busbar in a dual-redundant symmetrical arrangement, and the voltage signals output by them are arithmetically averaged to obtain the second raw signal. The first original signal is integrated and low-pass filtered to generate the first current data. The second original signal is differentially amplified, and gain correction and zero-point offset correction are performed in the digital domain based on the real-time temperature information obtained from inside the sensor package to generate the second current data. Based on the real-time sensed current amplitude, the full-range monitoring range of the transformer is dynamically divided into three continuous functional zones: low-range zone, fusion zone, and high-range zone. When the current is determined to be within the fusion region, a weighted fusion operation is initiated. Dynamic weighting coefficients are calculated in real-time using a logistic function with the current current amplitude as the independent variable. The first current data and the second current data are then weighted and summed based on these dynamic weighting coefficients to obtain the final fused current data. The weighted fusion operation is specifically performed according to the following formula: Wherein, the dynamic weight coefficient Real-time calculation based on logistic functions: in, This is an estimated value for the current amplitude. The central amplitude point of the integration zone, The slope constant is... The value of makes the logistic function approach 0 or 1 at the boundary of the fusion region; When it is determined that the current is in the low range region, the second current data is directly used as the final current data; when it is determined that the current is in the high range region, the first current data is directly used as the final current data. When the transformer is in steady-state operation and the current is in the low-range region or the fusion region, the second current data is used as a reference value to perform real-time correction of the integral drift of the first current data. This real-time correction of the integral drift of the first current data specifically includes: initiating a calibration process when the transformer enters steady-state operation and the current is in the low-range region or the fusion region, wherein the condition for entering steady-state is: within N consecutive sampling periods, the fluctuation range of the second current data is less than a preset steady-state determination threshold; after the calibration process is initiated, the DC component of the instantaneous residual between the first current data and the second current data is extracted using a moving average filter to obtain a drift error estimate; a reverse correction is applied to the first current data based on the drift error estimate to eliminate integral drift; when the transformer current enters the high-range region, the current drift error estimate is frozen, and the frozen current drift error estimate is used to continue correcting the first current data in subsequent processing.

2. The intelligent transformer current monitoring method for smart grids according to claim 1, characterized in that, The Hall sensor is encapsulated in a metal box made of highly conductive metal. The metal box is connected to the system ground at a single point on the printed circuit board level to form a Faraday cage, with only a narrow gap in the direction of the magnetic core opening to shield stray electromagnetic interference.

3. The intelligent transformer current monitoring method for smart grids according to claim 1, characterized in that, The second original signal undergoes gain correction and zero-point offset correction in the digital domain, specifically including: The microcontroller acquires the voltage across the thermistor embedded inside the Hall sensor near the Hall element in real time through a voltage divider circuit, and calculates the real-time value of the current ambient temperature based on the voltage divider ratio and the thermistor's calibration table. The microcontroller has a two-dimensional temperature compensation coefficient matrix pre-stored inside. This matrix is ​​obtained by placing the Hall sensor in a high and low temperature test chamber for temperature cycling and recording the relationship curve between magnetic induction intensity and Hall output voltage at each temperature point, and then using least squares fitting to obtain the gain correction coefficient and zero-point offset coefficient at each temperature point. In actual operation, the microcontroller performs linear interpolation on the two-dimensional temperature compensation coefficient matrix based on the real-time temperature to obtain the corresponding gain correction coefficient and zero-point offset coefficient, and completes digital correction according to the formula.

4. The intelligent transformer current monitoring method for smart grids according to claim 1, characterized in that, The full-range monitoring range of the transformer is dynamically divided into a low-range zone, a convergence zone, and a high-range zone, specifically including: A first threshold and a second threshold are preset, wherein the first threshold is set to 20% of the transformer's rated current and the second threshold is set to 80% of the transformer's rated current; The estimated value of the current amplitude is compared with the first threshold and the second threshold respectively: if the estimated value is less than the first threshold, it is determined that the current range is low; if the estimated value is greater than or equal to the first threshold and less than or equal to the second threshold, it is determined that the current range is fusion; if the estimated value is greater than the second threshold, it is determined that the current range is high. The first threshold and the second threshold can be remotely and dynamically adjusted by receiving threshold update instructions from the edge computing gateway via an industrial fieldbus.

5. The intelligent transformer current monitoring method for smart grids according to claim 1, characterized in that, It also includes a self-diagnostic step: When it is determined that the current is in the fusion region, the absolute value of the residual between the first current data and the second current data is calculated; When the absolute value of the residual is continuously greater than the preset sensor deviation threshold, the abnormal state is accumulated and timed, and after the accumulated time reaches the preset continuous judgment time, a sensor health warning signal is automatically issued.

6. The intelligent transformer current monitoring method for smart grids according to claim 1, characterized in that, The final current data acquired is also used to perform harmonic analysis in parallel. Extract data frames of a preset time length from continuous sampling data, and apply Hanning window weighting to the data frames; The discrete spectrum of the windowed data frame is calculated using the Fast Fourier Transform algorithm. The amplitude of the fundamental component and the amplitude of specific order harmonic components are extracted from it, and the total harmonic distortion rate is calculated. The calculated total harmonic distortion rate is compared with a pre-stored transformer loss-harmonic correlation table to assess the current operating loss level of the transformer online.

7. An intelligent transformer current monitoring system for smart grids, characterized in that, The system comprising the method of any one of claims 1 to 6, wherein the method is: The signal acquisition unit is used to simultaneously acquire a first raw signal reflecting the rate of change of current and a second raw signal reflecting the magnetic field strength at the transformer output terminal. The signal conditioning and conversion unit is used to perform integral restoration and low-pass filtering on the first original signal to generate first current data, and to perform differential amplification and temperature compensation on the second original signal to generate second current data. The dynamic range division unit is used to dynamically divide the full-range monitoring range into a low-range zone, a fusion zone, and a high-range zone based on the real-time sensed current amplitude. The weighted fusion unit is used to calculate dynamic weighting coefficients in real time within the fusion region based on a logistic function with the current current amplitude as the independent variable, and to perform a weighted summation of the first current data and the second current data to output the final fused current data. A real-time integral drift calibration unit is used to perform real-time correction of the integral drift of the first current data based on the second current data when the transformer is operating in steady state and the current is in the low range region or the convergence region; and, The microcontroller integrates a hardware floating-point arithmetic unit to execute the signal processing, interval partitioning, weight fusion, and drift calibration algorithms of the aforementioned units, and uploads the final current data in real time via the industrial fieldbus protocol.

Images