A smart power distribution system monitoring system

By using an intelligent power distribution system monitoring system, and by combining edge computing units and multiple modules, the problems of fixed measurement benchmarks and inaccurate extraction of weak fault features in existing technologies have been solved, thus achieving accuracy and real-time performance of protection devices in inverter-type distributed power supply environments.

CN123121435APending Publication Date: 2026-11-17TIANJIN JIAFENG ELECTRIC POWER ENGINEERING CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611250200.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The measurement benchmarks and protection thresholds of existing power distribution system monitoring and protection devices are fixed and cannot track the dynamic changes in line current characteristics during the switching of inverter-type distributed power supply operation status. Furthermore, existing weak fault feature extraction methods are difficult to extract weak single-phase grounding fault features with high signal-to-noise ratio and no phase shift in the background of strong inverter switching noise and harmonics, resulting in maloperation, failure to operate, and inaccurate fault direction identification of protection devices.

Method used

An intelligent power distribution system monitoring system is adopted, including voltage and current transformer modules and edge computing units. It utilizes short-time feature extraction and status recognition modules, dual-time-scale reference correction modules, fractional-order generalized integral filtering modules, variational mode decomposition and reconstruction modules, and directional fault criterion modules to achieve adaptive correction of measurement references and high signal-to-noise ratio extraction of weak fault features. Real-time processing is performed through heterogeneous edge computing hardware.

Benefits of technology

It significantly reduces the probability of false tripping and failure to trip of protection devices in high-penetration scenarios, achieves accuracy and reliability of fault direction judgment, and meets the real-time requirements of distribution network protection actions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121435A_ABST
    Figure CN123121435A_ABST
Patent Text Reader

Abstract

This invention discloses an intelligent power distribution system monitoring system, belonging to the field of power distribution network monitoring and relay protection technology. The system includes a voltage and current transformer module and an edge computing unit. The edge computing unit identifies the operating status of inverter-type distributed power sources through a short time scale window, and performs rolling weighted correction on the fundamental current measurement reference value and protection threshold through a long time scale window. At the same time, it filters the zero-sequence transient signal without phase shift through a fractional generalized integrator, and reconstructs high signal-to-noise ratio fault feature components by combining variational mode decomposition, and calculates the zero-sequence power direction criterion. This invention realizes dynamic tracking and correction of the measurement reference and high signal-to-noise ratio extraction of weak fault features on the edge side, improves the accuracy of ground fault direction discrimination, reduces the probability of protection maloperation and failure to operate, and is suitable for power distribution network monitoring and protection with high penetration of distributed power sources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power distribution network monitoring and relay protection technology, specifically to an intelligent power distribution system monitoring system. Background Technology

[0002] Monitoring and protection devices for distribution networks typically consist of voltage transformers, current transformers, signal conditioning circuits, analog-to-digital converters, and protection criterion processing units. They are used to collect line voltage and current signals in real time and determine whether a line fault has occurred and whether tripping is necessary based on pre-set values. Three-stage current protection is a widely used protection method in distribution networks. Its basic principle is to pre-set fixed instantaneous overcurrent, time-limited instantaneous overcurrent, and overcurrent protection settings based on the line's rated load current and short-circuit calculated current. By comparing the measured effective current value with the fixed threshold, it is determined whether a line fault has occurred. For single-phase grounding faults, distribution networks generally adopt line and phase selection methods based on zero-sequence current, zero-sequence voltage amplitude, and phase comparison. This method is a conventional technical means well-known to those skilled in the art.

[0003] With the increasing proportion of distributed power sources, especially inverter-type distributed power sources, in power distribution networks, the network topology and fault characteristics of power distribution systems have undergone fundamental changes. The output current of inverter-type distributed power sources is constrained by the inverter control strategy, exhibiting highly nonlinear and response delay characteristics. Their short-circuit fault current is usually limited, making it easy for three-stage current protection relying on fixed thresholds to fail, fail to operate, or maloperate. Existing monitoring systems most similar to this invention generally adopt a fixed sampling architecture and fixed measurement benchmark with a single time scale, that is, using fixed rated voltage and rated current values ​​as measurement benchmarks, and using a pre-set setting table to realize protection criteria. For the extraction of fault characteristic signals, existing systems mostly use conventional Fourier transform or wavelet transform fundamental wave extraction methods. After detecting zero-sequence current, it is directly compared with a fixed threshold value to determine whether a ground fault has occurred.

[0004] Because the measurement benchmarks and protection thresholds of existing monitoring and protection devices remain fixed, while the output current of inverter-type distributed power sources is constrained by the inverter control strategy and exhibits nonlinear and limited characteristics, when the inverter-type distributed power source switches between operating states such as normal grid connection, low voltage ride-through, and anti-islanding disconnection, the steady-state and transient current characteristics of the line will change significantly. Fixed thresholds cannot track this dynamic change process, which in turn leads to the protection device malfunctioning or refusing to operate during the switching of operating states.

[0005] Since neither conventional Fourier transform nor ordinary wavelet transform has the ability to selectively suppress high-frequency switching noise of inverters, nor the ability to extract narrowband fault feature components without phase shift, and the zero-sequence current feature quantity corresponding to a single-phase ground fault has a weak amplitude, under the background of a large number of high-frequency harmonics and switching noise generated by inverter-type distributed power sources, existing methods will inevitably suffer from problems such as fault feature signals being submerged by noise, distorted extraction results, and even phase delay, which in turn leads to errors in the calculation of fault direction criteria and inaccurate line and phase selection results.

[0006] In addition, existing monitoring systems mostly adopt a simple functional module splicing method or rely on cloud platforms for big data delay calculations. It is difficult to simultaneously meet the three requirements of high-frequency noise suppression, weak signal extraction without phase shift, and dynamic correction of measurement reference at the edge of protection and monitoring devices, and cannot meet the strict requirements of distribution network protection devices for real-time operation.

[0007] In summary, there is an urgent need for an intelligent power distribution system monitoring system that can perform real-time adaptive correction of the measurement benchmark at the edge side and extract weak fault features with high signal-to-noise ratio and no phase shift in a strong noise background, in order to solve the problems of false tripping, failure to tripping, and inaccurate fault direction identification in the existing fixed threshold protection method. Summary of the Invention

[0008] The technical problem to be solved by this invention is that the measurement benchmark and protection threshold of existing power distribution system monitoring and protection devices are fixed and cannot track the dynamic changes of line current characteristics during the switching of inverter-type distributed power supply operation status. Moreover, the existing weak fault feature extraction methods are difficult to extract weak single-phase grounding fault features with high signal-to-noise ratio and no phase shift in the background of strong inverter switching noise and harmonics, thus causing the protection device to malfunction, fail to operate, and make inaccurate fault direction identification.

[0009] To address the aforementioned technical problems, this invention provides an intelligent power distribution system monitoring system. This system includes a voltage and current transformer module and an edge computing unit connected to it. The voltage and current transformer module is used to collect voltage and current signals from various monitoring points on the power distribution line. The edge computing unit includes a short-time feature extraction and state recognition module, a dual-time-scale reference correction module, a fractional-order generalized integral filtering module, a variational mode decomposition and reconstruction module, a directional fault criterion module, and a protection decision module. The short-time feature extraction and state recognition module performs short-time feature calculations on the collected voltage and current signals within a microsecond-level short-time-scale window to obtain a short-time feature vector. Based on the short-time feature vector, it identifies the current operating state of the inverter-type distributed power source and outputs a state identifier. The dual-time-scale reference correction module performs rolling weighting on the system fundamental current measurement reference value within a millisecond-level long-time-scale window based on the state identifier. The system corrects and dynamically updates the protection action threshold based on the corrected fundamental current measurement reference value. The fractional-order generalized integral filtering module performs fractional-order generalized integral filtering on the acquired zero-sequence transient current signal and transient voltage difference signal to obtain a filtered signal that suppresses harmonics in the non-target frequency band and inverter switching noise. The variational mode decomposition and reconstruction module performs variational mode decomposition on the filtered signal to obtain multiple intrinsic mode components and their corresponding center frequencies. Based on the fault characteristic frequency band, the target mode components are selected and superimposed and reconstructed to obtain reconstructed transient zero-sequence current components and reconstructed transient zero-sequence voltage components with high signal-to-noise ratio. The directional fault criterion module calculates the zero-sequence power direction criterion based on the reconstructed transient zero-sequence current components and reconstructed transient zero-sequence voltage components to determine the fault direction. The protection decision module is connected to the dual-time-scale reference correction module and the directional fault criterion module respectively to output the protection action signal based on the protection action threshold and the fault direction judgment result.

[0010] Furthermore, the adaptive weighting coefficient used for rolling weighted correction of the fundamental current measurement reference value and the margin coefficient used for updating the protection action threshold in the dual time scale reference correction module are set with different values ​​according to different operating states, so that the reference value has different tracking speeds under different operating states and the protection device has reasonable action margins under different operating states.

[0011] Furthermore, the fractional-order generalized integral filter module processes the signal using a fractional-order generalized integral transfer function. By adjusting the fractional order, it selectively extracts the fault characteristic frequency band without introducing phase delay. The variational mode decomposition and reconstruction module obtains the narrowband intrinsic mode components by constructing a constrained variational optimization model and iteratively solving it using the alternating direction multiplier method. The directional fault criterion module extracts the phasors corresponding to the reconstructed transient zero-sequence voltage component and the reconstructed transient zero-sequence current component, calculates the real part of the product of the transient zero-sequence voltage phasor and the transient zero-sequence current conjugate phasor, and uses it as the zero-sequence power direction criterion value.

[0012] Furthermore, the present invention also provides a hardware architecture for implementing the above system, which consists of a field-programmable gate array (FPGA) and a digital signal processor (DSP) to form heterogeneous edge computing hardware. The FPGA is responsible for multi-channel high-speed synchronous analog-to-digital acquisition and variational mode decomposition (VMD) low-level parallel filtering decomposition operations, while the DSP is responsible for operation status identification, dual time-scale reference correction operations, and direction criterion and protection logic operations. The two communicate with each other at high speed through a bidirectional data bus.

[0013] The technical effects that this invention can achieve include:

[0014] 1. Through a dual-time-scale adaptive measurement benchmark correction architecture, the measurement benchmark and protection action threshold can be dynamically tracked and corrected according to the operating status of the inverter-type distributed power source, which significantly reduces the probability of protection devices malfunctioning or refusing to operate in high-penetration scenarios.

[0015] 2. By combining the fractional generalized integral filtering module and the variational mode decomposition and reconstruction module for edge-side signal processing, it is possible to extract weak zero-sequence fault characteristics in real time at the edge without relying on cloud computing, thus meeting the real-time requirements of distribution network protection action windows.

[0016] 3. By calculating the zero-sequence power direction criterion based on the reconstructed transient zero-sequence voltage and current phasors, the accuracy and reliability of the single-phase grounding fault direction criterion are improved. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall system architecture of the present invention.

[0018] Figure 2 This is a flowchart of the dual-time-scale adaptive measurement reference correction process of the present invention;

[0019] Figure 3 This is a flowchart of the weak fault feature extraction and direction determination process of the present invention. Detailed Implementation

[0020] The following is in conjunction with the appendix Figure 1-3The specific embodiments of the present invention will be further described below. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0021] In terms of the overall system architecture, the intelligent power distribution system monitoring system provided in this embodiment includes a voltage and current transformer module, a signal conditioning circuit, an edge computing unit, and a protection decision and communication module. The voltage and current transformer module is installed at each monitoring point of the power distribution line and consists of a Rogowski coil and a capacitive voltage divider. The Rogowski coil is used to induce the high-frequency and transient components of the current, and the capacitive voltage divider is used to induce the high-frequency and transient components of the voltage. The two work together to output analog voltage and current signals. The output of the voltage and current transformer module is connected to the signal conditioning circuit, which performs filtering and anti-aliasing processing and level adaptation processing, and outputs the processed analog signal to the edge computing unit. The edge computing unit consists of a field-programmable gate array and a digital signal processor, and internally implements short-term protection decision and communication in the form of functional modules. The system comprises a short-time feature extraction and state recognition module, a dual-time-scale reference correction module, a fractional-order generalized integral filtering module, a variational mode decomposition and reconstruction module, and a directional fault criterion module. The transformer signal is processed sequentially by the short-time feature extraction and state recognition module and the dual-time-scale reference correction module to obtain the corrected measurement reference value and protection action threshold. Simultaneously, it is processed sequentially by the fractional-order generalized integral filtering module and the variational mode decomposition and reconstruction module, and then input into the directional fault criterion module to obtain the fault direction discrimination result. The outputs of the dual-time-scale reference correction module and the directional fault criterion module are jointly sent to the protection decision module. The protection decision module transmits the protection action signal, alarm information, and monitoring data to the distribution automation master station or local protection device via the communication interface to execute the corresponding tripping or alarm actions.

[0022] Regarding the dual-timescale adaptive measurement reference correction, this embodiment provides the following specific implementation: Instantaneous voltage and current signals from various monitoring points in the distribution network are acquired using a voltage and current transformer module. The acquired signals are then synchronously sampled at a sampling frequency of at least 1 MHz using a multi-channel high-speed analog-to-digital converter interface of a field-programmable gate array (FPGA), resulting in discretized voltage and current sequences. Within a microsecond-level short timescale window, the length of this window is denoted as... The typical value ranges from 20 microseconds to 100 microseconds. A field-programmable gate array (FPGA) performs high-frequency feature calculations on the voltage and current sequences within this window, extracting three short-time features: voltage dip depth, current rise rate, and frequency offset. These are combined to form a short-time feature vector. Based on this short-time feature vector, the digital signal processor (DSP) identifies the current operating state of the inverter-type distributed power source according to preset state discrimination rules and outputs a state identifier. This identifier represents one of three operating states: normal grid connection, low voltage ride-through, and anti-islanding. This indicates the sequence number of the current long-timescale window.

[0023] After obtaining the state identifier, the baseline value is corrected within a millisecond-scale long-term window, the length of which is denoted as . Typically, one power frequency cycle is used, i.e., 10 to 20 milliseconds, and the system fundamental current measurement reference value is adjusted by rolling weighting according to the following formula:

[0024] ;

[0025] In the formula, For the first The fundamental current measurement reference value is corrected for a long time scale window. This is the baseline value obtained after correction from the previous long-term time scale window. For the first The effective value of the fundamental current measured within each window. For state identifier The corresponding adaptive weighting coefficient has a value greater than 0 and less than 1. Different operating states correspond to different preset weight values, so that the reference value can track the change of actual load current at different speeds when the operating state is switched. In a specific embodiment, the weighting coefficient corresponding to the normal grid connection state is greater than the weighting coefficient corresponding to the low voltage ride-through state and the anti-islanding grid disconnection state, so that the reference value remains relatively stable during normal operation and can track the change of actual current more quickly in abnormal operating states.

[0026] Based on the revised fundamental current measurement reference value, the protection action threshold is dynamically updated according to the following formula:

[0027] ;

[0028] In the formula, For the first The protection action threshold corresponding to each window For state identifier Corresponding margin coefficients, with different preset margin coefficient values ​​for different operating states, are used to ensure that the protection device has a reasonable operating margin under different inverter-type distributed power supply operating states, and to correct the measurement reference value. With protection action threshold The output is sent to the protection criterion module for fault identification in the current and subsequent cycles, and simultaneously... The feedback is sent to the short-time feature extraction module as a reference baseline for the next long-time scale window, forming a closed-loop adaptive process in which the baseline value is dynamically corrected according to the operating state of the inverter-type distributed power source.

[0029] Regarding the joint extraction of weak fault features using fractional-order generalized integral filtering and variational mode decomposition, this embodiment provides the following specific implementation method: extracting the original zero-sequence current signal collected by the transformer... and transient voltage difference signal Input the fractional-order generalized integral filter module, which performs fractional-order generalized integral processing on the signal based on the following transfer function:

[0030] ;

[0031] In the formula, Let be the transfer function of the fractional-order generalized integral filter module. For the Laplace operator, This is the gain coefficient, used to adjust the filter bandwidth. The fundamental angular frequency of the system. The order is a fraction, with values ​​greater than 0 and not greater than 1, and is adjusted... This module enables selective extraction of fault characteristic frequency bands without introducing phase delay. After processing, the original signal is converted into a filtered signal that suppresses harmonics in non-target frequency bands and inverter switching noise. and .

[0032] The filtered signal is divided into lengths of... Windowing is used to define the expected number of intrinsic mode components for variational mode decomposition. and secondary penalty factor , framing the signal As the signal to be decomposed Input the variational mode decomposition module to construct the following constrained variational optimization model:

[0033] ;

[0034] In the formula, For the first Each intrinsic mode component For the first The central angular frequency corresponding to each modal component The preset total number of modal components, For the Dirac function, The imaginary unit, symbol This represents the convolution operation. Represents the L2 norm, The input signal to be decomposed is denoted as .

[0035] The variational model above is solved iteratively using the alternating direction multiplier method, with each modal component and center frequency updated alternately according to the following formula:

[0036] ;

[0037] ;

[0038] In the formula, For the first The iteration yields the... Frequency domain representation of each modal component Signal to be decomposed Frequency domain representation, For the frequency domain representation of Lagrange multipliers, This is a quadratic penalty factor used to control the strength of the modal bandwidth constraint. For the first After the [number]th iteration update The center frequency estimates of each modal component are obtained by iterating the above two equations repeatedly until convergence, yielding the final result. Each intrinsic mode component and its center frequency .

[0039] Based on prior knowledge of the characteristic frequency band of zero-sequence current in single-phase ground faults in distribution networks, from the obtained The center frequency is selected from the intrinsic modal components. The modal components falling within the fault characteristic frequency band are selected and reconstructed by superimposing the selected modal components to obtain the reconstructed transient zero-sequence current component with a high signal-to-noise ratio. For transient voltage difference signals Perform the same fractional-order generalized integral filtering, framing, variational mode decomposition, and target mode component screening and reconstruction processes as described above to obtain the corresponding high signal-to-noise ratio reconstructed transient zero-sequence voltage component. .

[0040] Based on the reconstruction and Extract their phasors separately and The zero-sequence power direction criterion is calculated according to the following formula:

[0041] ;

[0042] In the formula, This is the zero-sequence power direction criterion value. For the reconstructed transient zero-sequence voltage phasor, The phasor of the reconstructed transient zero-sequence current is denoted by [symbol]. This indicates the conjugate operation. This indicates the operation of taking the real part, based on... The sign and value of the fault are compared with the preset threshold value to determine whether a ground fault has occurred in the forward direction, the reverse direction, or no ground fault has occurred, and the judgment result is output to the protection decision module.

[0043] In terms of hardware implementation architecture, this embodiment adopts heterogeneous edge computing hardware composed of field-programmable gate arrays (FPGAs) and digital signal processors (DSPs). The sensing layer consists of Rogowski coil current sensing units and capacitive voltage divider sensing units. After being processed by the signal conditioning circuit, they are respectively connected to the multi-channel analog-to-digital conversion interface of the computing layer. In the computing layer, the FPGA undertakes the multi-channel high-speed synchronous analog-to-digital acquisition function and the variational mode decomposition low-level parallel filtering decomposition operation function. The DSP undertakes the fractional-order generalized integral parameter adaptive update function, inverter-type distributed power source status identification and short-time feature calculation function, dual-time-scale reference correction operation function, and direction criterion and protection logic operation function. The FPGA and the DSP exchange data at high speed through a bidirectional data bus. The output of the DSP is connected to the communication interface module. The communication interface module transmits protection action signals, alarm information, and monitoring data to the distribution automation master station or local protection execution unit, thereby forming a complete hardware data link from signal sensing, edge computing to protection action output.

[0044] Regarding the overall system workflow, this system continuously executes the aforementioned processing steps in parallel during actual operation: on the one hand, it performs dynamic adaptive correction of the fundamental current measurement reference value and protection action threshold according to the operating state of the inverter-type distributed power source; on the other hand, it performs high signal-to-noise ratio extraction of zero-sequence transient current and transient voltage difference and calculation of fault direction criteria. The outputs of the two processing links are jointly sent to the protection decision module for comprehensive judgment, that is, when the measured fundamental current exceeds the dynamically corrected protection action threshold... Or direction criteria When the preset ground fault criteria are met, the system determines that a corresponding type of fault has occurred and triggers protection actions or alarms, thereby enabling intelligent and accurate monitoring of the power distribution system's operating status in high-penetration inverter-type distributed power supply access scenarios.

[0045] As an alternative implementation, in the weak fault feature extraction stage, variational mode decomposition can also be replaced by wavelet transform or empirical mode decomposition. The signal after fractional generalized integral filtering is decomposed into multiple scales and combined with the same fractional generalized integral pre-filtering and target frequency band mode component reconstruction method to achieve the same high signal-to-noise ratio weak signal extraction effect as the aforementioned embodiment.

[0046] As an alternative implementation, in the inverter-type distributed power supply operation status identification stage, the status discrimination rules in the short-time feature extraction and status identification module can also be implemented by a pre-trained classification model. This classification model uses short-time feature vectors as input to classify and identify the operation status of the inverter-type distributed power supply. The remaining dual-time-scale benchmark correction, fractional-order generalized integral filtering and variational mode decomposition signal extraction, and direction criterion parts maintain the same technical solutions as the aforementioned embodiments, thereby further improving the status identification accuracy while maintaining the real-time performance of the edge side.

[0047] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, any modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope defined by the claims of the present invention.

Claims

1. An intelligent power distribution system monitoring system, characterized by, include: Voltage and current transformer modules are used to collect voltage and current signals at various monitoring points on power distribution lines; The edge computing unit is electrically connected to the voltage and current transformer module through a signal conditioning circuit, and includes a short-time feature extraction and state recognition module, a dual time-scale reference correction module, a fractional generalized integral filtering module, a variational mode decomposition and reconstruction module, and a directional fault criterion module. The short-time feature extraction and state recognition module is used to calculate a short-time feature vector based on the voltage signal and the current signal within a short time scale window, and to identify the current operating state of the inverter-type distributed power source based on the short-time feature vector, and output a state identifier. The dual-timescale reference correction module is used to perform rolling weighted correction on the fundamental current measurement reference value within a long timescale window based on the status identifier, and update the protection action threshold based on the corrected fundamental current measurement reference value. The duration of the long timescale window is greater than the duration of the short timescale window. The fractional-order generalized integral filtering module is used to perform fractional-order generalized integral filtering on the acquired zero-sequence transient current signal and transient voltage difference signal to obtain the filtered signal; The variational mode decomposition and reconstruction module is used to perform variational mode decomposition on the filtered signal to obtain multiple intrinsic mode components and their corresponding center frequencies. Based on the preset fault characteristic frequency band, the target mode components whose center frequencies fall into the fault characteristic frequency band are selected. The target mode components are superimposed and reconstructed to obtain the reconstructed transient zero-sequence current component and the reconstructed transient zero-sequence voltage component. The directional fault criterion module is used to calculate the zero-sequence power direction criterion based on the reconstructed transient zero-sequence current component and the reconstructed transient zero-sequence voltage component, and to determine the fault direction. The protection decision module is connected to the dual time-scale reference correction module and the directional fault criterion module respectively, and is used to output protection action signals based on the judgment results of the protection action threshold and the fault direction.

2. The intelligent power distribution system monitoring system of claim 1, wherein In the dual-time-scale reference correction module, the adaptive weighting coefficient used for rolling weight correction of the fundamental current measurement reference value and the margin coefficient used for updating the protection action threshold are set with different values ​​according to different operating states. The value of the adaptive weighting coefficient is greater than 0 and less than 1.

3. The intelligent power distribution system monitoring system of claim 1, wherein The status identifier is used to represent one of the three operating states: normal grid connection state, low voltage ride-through state, and anti-islanding grid disconnection state.

4. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The fractional-order generalized integral filter module uses a fractional-order generalized integral transfer function that includes a fractional order. The value of the fractional order is greater than 0 and not greater than 1. By adjusting the fractional order, selective extraction of fault characteristic frequency bands can be achieved without introducing phase delay.

5. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The variational mode decomposition and reconstruction module is used to frame the filtered signal, set the expected number of intrinsic mode components and the quadratic penalty factor, construct a constrained variational optimization model, and use the alternating direction multiplier method to iteratively solve the variational optimization model to obtain the multiple intrinsic mode components and their corresponding center frequencies.

6. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The directional fault criterion module is used to extract the phasors corresponding to the reconstructed transient zero-sequence current component and the reconstructed transient zero-sequence voltage component, and calculate the real part of the product of the reconstructed transient zero-sequence voltage phasor and the reconstructed transient zero-sequence current conjugate phasor as the zero-sequence power direction criterion.

7. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The edge computing unit consists of a field-programmable gate array (FPGA) and a digital signal processor (DSP). The FPGA is used to perform multi-channel high-speed synchronous analog-to-digital acquisition operations and low-level parallel filtering decomposition operations of the variational mode decomposition and reconstruction module. The DSP is used to perform state recognition operations of the short-time feature extraction and state recognition module, reference correction operations of the dual-time-scale reference correction module, and direction criterion operations of the directional fault criterion module. The FPGA and the DSP are connected via a bidirectional data bus.

8. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The voltage and current transformer module includes a Rogowski coil for sensing transient current components and a capacitive voltage divider for sensing transient voltage components. The output of the voltage and current transformer module is connected to a signal conditioning circuit, and the output of the signal conditioning circuit is connected to the edge computing unit.

9. The intelligent power distribution system monitoring system according to claim 1, characterized in that, The state discrimination rule in the short-time feature extraction and state recognition module is implemented by a pre-trained classification model. The classification model uses the short-time feature vector as input to identify the current operating state of the inverter-type distributed power source.