A non-intrusive power distribution network impedance spectrum online detection system and method based on wideband current pulse

Abstract

A non-intrusive power distribution network impedance spectrum online detection system and method based on wideband current pulse, the system comprises a signal injection module, a signal acquisition module, a signal processing and spectrum reconstruction module, an impedance feature recognition and abnormal diagnosis module, a control and communication module. The method comprises power-on initialization, detection timing determination, wideband pulse injection and synchronous acquisition, impedance characteristic analysis and abnormal diagnosis, adaptive closed-loop control and system protection. The system injects low-energy, short-time wideband current pulses in selected time slots to obtain line transient response signals without interfering with power supply and communication, and reconstructs the wideband impedance spectrum through time-frequency analysis to realize high-resolution online detection of the power distribution network.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a non-disruptive online impedance spectrum detection system and method for distribution networks based on broadband current pulses. Background Technology

[0002] Currently, in the operation of low-voltage distribution networks and the construction of integrated communication, induction, computing and control systems, real-time monitoring of line health status (including impedance characteristics, grounding conditions, insulation aging, etc.) is crucial to ensuring system safety and reliability.

[0003] The health status and potential defects of low-voltage distribution networks (LVDNs) (such as conductor aging, high-resistance contacts, and deterioration of cable insulation) are typically manifested in the impedance characteristics of the system at different frequencies. The changes in impedance spectrum. Therefore, accurate measurement of the broadband impedance spectrum is fundamental to achieving real-time diagnostics. When a very short current pulse is injected into the power grid, it can naturally excite the broadband required for LVDN diagnostics without the need for time-consuming point-by-point frequency scanning. Impedance spectrum of the line. The transient voltage response can be obtained through the acquisition. and transient current Perform a Fast Fourier Transform (FFT) and then calculate the ratio in the frequency domain. To achieve accurate impedance calculation, the signal acquisition module must have a high sampling rate and strictly synchronize the recording of voltage and current signals to ensure that time-domain information does not lead to deviations in frequency-domain phase angle calculations.

[0004] Existing technical solutions can be divided into two main categories:

[0005] High-frequency impedance detection technology based on controlled continuous injection (high-frequency injection method) is currently the mainstream online impedance measurement method. Its steps are as follows: generating characteristic harmonic current commands according to a harmonic signal generation algorithm → inputting the commands into a DC voltage feedback control loop for current tracking control → injecting high-frequency or harmonic currents into the power grid continuously or periodically by controlling the switching of the main circuit's transistors → selecting voltage and current signals for signal decomposition processing and analysis to obtain the power grid impedance. The system structure consists of a high-frequency signal source / control inverter used to generate continuous harmonic current commands; the coupling circuit / filter typically requires a complex external coupler to inject the high-frequency signal while isolating the high-frequency equipment from the mains power grid. This technology requires continuous or periodic injection of high-energy signals, which can easily interfere with communication and power quality.

[0006] The impedance spectrum acquisition technology based on natural transient disturbances in the system (passive transient method) involves the following steps: The system continuously monitors natural or operational transient events commonly found in the power system → identifies and captures high-energy transient signals, such as lightning strikes, operational surges, single-phase grounding faults, or capacitor switching → monitors the transient response current on cable grounding wires or cross-connection lines. Due to the strong amplitude of the impulse signal, the transient response is easily captured by the monitoring equipment, resulting in high detection accuracy → uses the monitored impulse and current signals to obtain the cable insulation impedance spectrum or admittance spectrum in the frequency domain → judges the insulation aging state based on the changes in impedance spectrum amplitude and the resonant frequency shift. The system structure consists of high-bandwidth transient sensors, such as voltage transformers / current transformers, used to capture high-energy transient responses; and high-speed data loggers used to record and analyze transient waveforms. The core drawback of this technology is that its excitation source is passive and random, making it impossible to achieve "on-demand" real-time online detection. The timing and quality of the measurement depend entirely on randomly occurring system events.

[0007] Patent document "Method and Apparatus for Measuring Power Grid Impedance" (CN117310293B) discloses a method for injecting current into the power grid by generating characteristic harmonic current commands, controlling the switching transistor to turn on and off, and then performing signal decomposition to obtain the power grid impedance. It represents an online measurement technique aimed at reducing disturbances while still based on continuous injection.

[0008] The patent document "Online Monitoring Method for Medium-Voltage Cable Insulation Based on Impulse Impedance Spectrum" (CN112881806B) discloses an online monitoring method that utilizes lightning impulse and switching impulse signals commonly found in power systems as excitations to monitor the current response on the cable grounding wire and obtain the cable insulation impedance spectrum in the frequency domain for determining the insulation aging state. This represents a technology for passively utilizing high-energy system transients for broadband diagnostics.

[0009] The paper "A Novel Online Monitoring Method for Overall Insulation Aging Based on System Power Disturbance" (Journal of Power System Technology, 2020) discloses an online monitoring method. This method utilizes the transient response of the grounding wire current under system power disturbances to identify the overall aging state of the cable by calculating the insulation admittance spectrum. It also falls under the category of passive broadband diagnostics using system transients.

[0010] The patent document "High-Frequency Detection Pulse Injection Method for Brushless DC Motor" (CN102780430B) discloses a high-frequency detection pulse injection method, which demonstrates the hardware implementation of coupling the high-frequency detection pulse signal to the power line through a coupling circuit. It is a representative of coupling and isolation technology in high-frequency injection methods.

[0011] Existing impedance detection technologies for power distribution networks mainly include offline testing methods and high-frequency signal injection methods.

[0012] Offline testing methods require specialized instruments for measurement during power outages or line cutovers. While offering high accuracy, the disconnect between the testing process and grid operation results in long testing cycles, complex operations, and high labor costs, failing to meet the real-time and continuous requirements of distribution networks. Consequently, potential line hazards are often only discovered passively after a serious fault occurs, lacking online monitoring and early warning capabilities.

[0013] While online methods using high-frequency signal injection can perform detection under power supply conditions, the high injection energy can interfere with power quality and communication signals. Furthermore, the signal requires a coupler to reach the line, resulting in a complex system structure and costly filtering circuits. Due to the single injection frequency or narrow spectral coverage, these methods suffer from significant deficiencies in frequency domain resolution and measurement accuracy, failing to accurately reflect the impedance variation characteristics of the line over a wide frequency range. In addition, continuous high-frequency signal injection can lead to the accumulation of electromagnetic interference on the line, which does not meet the current development requirements of "disruptive detection" and "low-power sensing" in distribution networks.

[0014] In summary, existing technologies generally suffer from drawbacks such as reliance on external equipment for detection, strong intrusiveness during the measurement process, limited spectral range, high system cost, and insufficient real-time performance and sensitivity. These problems are particularly prominent in the context of intelligent power distribution network construction, limiting the widespread application of impedance spectroscopy in equipment health assessment, fault identification, and aging diagnosis.

[0015] Therefore, there is a need for a non-disruptive online impedance spectrum detection system and method for power distribution networks that can overcome the shortcomings of existing technologies. Summary of the Invention

[0016] The technical problem to be solved by this application is to provide an online impedance spectrum detection system and method that overcomes the technical defects of existing power distribution network impedance detection technology, such as reliance on external equipment, strong interference in the measurement process, limited spectrum range, high system cost, and insufficient real-time performance and sensitivity.

[0017] According to one aspect of this application, a non-disruptive online impedance spectrum detection system for distribution networks based on broadband current pulses is provided, including a signal injection module, a signal acquisition module, a signal processing and spectrum reconstruction module, an impedance feature identification and anomaly diagnosis module, and a control and communication module.

[0018] The signal injection module receives a synchronous trigger signal generated by the control and communication module, and generates a broadband current pulse to inject into the distribution network line according to the synchronous trigger signal. The signal acquisition module acquires the voltage response and current response of the distribution network line and digitizes the voltage response and current response. The signal processing and spectrum reconstruction module receives the digitized voltage response and digitized current response output by the signal acquisition module and calculates the impedance spectrum of the distribution network line. The impedance feature identification and anomaly diagnosis module receives the impedance spectrum of the distribution network line output by the signal processing and spectrum reconstruction module to diagnose the health status of the line. The control and communication module receives the diagnostic data and feature data generated by the impedance feature identification and anomaly diagnosis module and reports the results to the distribution automation master station.

[0019] The signal injection module includes a controllable electronic switch, a reference load, and a drive circuit, used to generate broadband current pulse signals. The signal acquisition module includes a voltage transformer, a current transformer, and a high-speed analog-to-digital converter, used to simultaneously acquire transient response data of voltage and current of the distribution network line at the instant broadband current pulse signal is injected into the distribution network line. The signal processing and spectrum reconstruction module calculates and reconstructs the impedance spectrum of the distribution network line using digital signal processing algorithms, and performs spectrum smoothing and interpolation optimization. The impedance feature identification and anomaly diagnosis module utilizes the amplitude-frequency and phase-frequency characteristics of the impedance spectrum of the distribution network line to achieve line health status diagnosis through template matching or machine learning algorithms. The control and communication module coordinates the system operation process, realizing detection timing control, result reporting, and parameter adaptive optimization.

[0020] According to another aspect of this application, a method for online impedance spectrum detection of a distribution network based on broadband current pulses is provided, comprising the following steps:

[0021] S10 performs power-on initialization on the non-disruptive online impedance spectrum detection system for distribution networks based on broadband current pulses, completing system self-test, parameter loading, and sampling link calibration to prepare for online detection.

[0022] S20, determine the detection timing based on the load status and communication occupancy;

[0023] S30, Broadband Pulse Injection and Synchronous Acquisition, injects broadband current pulses into the distribution network lines during safe time slots and acquires the response of the distribution network lines in real time;

[0024] S40, Impedance Characteristic Analysis and Anomaly Diagnosis: Calculates the impedance spectrum of distribution network lines and extracts features for intelligent diagnosis.

[0025] S50, adaptive closed-loop control and system protection, realizes adaptive parameter adjustment, measurement quality improvement and safety protection action linkage, forming a complete closed-loop detection, analysis and optimization mechanism;

[0026] Step S20 includes the following steps:

[0027] S201, Periodicity determination, the detection period is controlled by the timer Timer_Detect. If the system time satisfy

[0028] Then proceed to step S202, where, Indicates the timestamp of the last detection;

[0029] S202, Load Condition Assessment, real-time monitoring of line current during the detection period. and in the sliding window Internal calculation of average load current and volatility variance The average load current for The fluctuation variance for Calculate the perturbation factor , ,like If the load enters a stable period, proceed to step S203; otherwise, return to step S201. The threshold for the perturbation factor. The maximum load variance during system calibration;

[0030] S203, Evaluate the communication occupancy status; if the communication unit communication bus in the control and communication module of claim 1 has been idle for a certain period of time... Then proceed to step S204; otherwise, return to step S201. Minimum communication bus idle time;

[0031] S204, if the current time is within the voltage trough window, proceed to step S30; otherwise, return to step S201. The voltage trough window is defined as the trough time of the fundamental voltage wave. Centered on, the width of the voltage trough is The time window; the trough moment The calculation method is as follows: ,in, This represents the value of the variable when it reaches its minimum value. This represents the real-time voltage waveform of the line. Represents the time variable, voltage trough width The calculation method is as follows ,in This is the proportionality coefficient. To align with the power frequency cycle.

[0032] According to some embodiments, step S30 includes the following steps:

[0033] S301, parameter loading and module self-test: the signal injection module performs parameter loading and module self-test;

[0034] S302 sends a synchronization trigger signal. It adopts a single master synchronous clock architecture. The master control MCU generates a synchronization trigger signal SYNC_TRIG, which is distributed to the signal injection module and the signal acquisition module through the isolation drive circuit.

[0035] S303, Generate a broadband current pulse according to the SYNC_TRIG synchronous trigger signal and inject it into the distribution network line;

[0036] S304, Transient Response Synchronous Acquisition, the signal acquisition module acquires the response of the distribution network line;

[0037] S305, signal preprocessing, performs DC component removal, windowing, and spectrum smoothing on the acquired signal;

[0038] S306, Time-Frequency Transformation and Impedance Calculation: Performs Fast Fourier Transform on preprocessed voltage and current signals to calculate the impedance spectrum of distribution network lines. ,in For the real part, It is the imaginary part;

[0039] S307, Spectrum Optimization and Curve Reconstruction: Spectrum reconstruction is performed to eliminate sampling noise and Fast Fourier Transform discretization errors. The spectrum reconstruction includes smoothing filtering and interpolation reconstruction. The smoothing filtering calculation formula is the smoothed impedance spectrum. ,in Represents the original discrete frequency points. N is the number of sampling points. The interpolation reconstruction includes cubic spline interpolation of sparse frequency points to obtain a high-resolution curve. ,in, This represents the high-resolution impedance spectrum after interpolation reconstruction. This represents the cubic spline interpolation operator. This represents the original discrete impedance data sequence. Represents the original discrete frequency points. This represents the target frequency variable.

[0040] According to some embodiments, step S40 includes the following steps:

[0041] S401, Impedance spectrum data preprocessing, the impedance spectrum After sampling and reconstruction, frequency band normalization and noise correction are performed. The frequency band normalization and noise correction include frequency logarithmic transformation, outlier removal, and amplitude frequency and phase frequency separation processing.

[0042] S402, Feature parameter extraction: Extracting typical characteristic quantities from the impedance spectrum that reflect line characteristics, load conditions, and contact degradation; the typical characteristic quantities include: low-frequency real part average value. Mid-frequency imaginary peak frequency Impedance phase angle slope Equivalent time constant Impedance fitting residual The average value of the low-frequency real part The peak frequency of the imaginary part of the intermediate frequency is a characteristic quantity reflecting the DC resistance and contact performance of the line. The impedance phase angle slope is a characteristic quantity related to the distribution of inductive and capacitive loads. The equivalent time constant is a characteristic quantity that reflects the rate of change of frequency response. The impedance fitting residual is a characteristic quantity used to characterize the dynamic response of a node. These are characteristic quantities that indicate the complexity or anomalies of a line.

[0043] S403, Impedance characteristic modeling and template library establishment, using a dual-path modeling method for impedance characteristic modeling, wherein one modeling path is a physical mechanism modeling path and the other modeling path is a data-driven modeling path;

[0044] S404, Abnormal Diagnosis and Status Determination, includes the following steps:

[0045] Step S40401, based on feature deviation Make a judgment. ,in, Indicates relative to the first Feature deviation of class state template This represents the feature vector currently being detected. Indicates the first Feature vectors of class templates This represents the square of the Euclidean distance. This represents the square of the template vector's magnitude. Indicates the index of the template or category; if If the condition is determined to be an unknown anomaly, proceed to step S40402. The threshold for determining feature deviation;

[0046] Step S40402: Calculate cosine similarity. ,in, Represents cosine similarity. This represents the dot product of vectors, used to calculate the maximum similarity between categories. Based on the maximum similarity, the corresponding category Determine the state category;

[0047] S405, Anomaly Level Quantization, Calculates Anomaly Level ,in, Indicating the severity level index of the anomaly. express The reference base value, express The reference base value, Indicates the best matching category similarity. Indicates the diagnosed state category. This represents the similarity residual, reflecting the degree of deviation between the current state and the standard template. For experience weights.

[0048] According to some embodiments, the physical mechanism modeling path in step S403 includes: fitting the impedance spectrum using an improved Cole model. ,in, This represents the total impedance of the line. For the resistance of the cable itself, The equivalent time constant, The fractal index, The range of values ​​is The least squares method is used to estimate ( Construct an equivalent parameter vector. .

[0049] According to some embodiments, the data-driven modeling path in step S403 includes: targeting historical datasets. The K-means clustering algorithm was used to classify the state categories. ,in, The k-th type of center corresponds to normal, slightly aged, and severely degraded operating states. Indicates the first The feature vector of each sample Indicates category index, This represents the total number of historical samples. This represents the square of the Euclidean distance. This indicates a minimize operation.

[0050] According to some embodiments, step S50 includes:

[0051] S501, Detection result reporting and data storage: The detection and diagnosis result data is uniformly encapsulated and reported by the control and communication module, and the data is stored locally at the same time.

[0052] S502, Performance Evaluation and Adaptive Decision, assesses measurement quality and anomaly risk after each round of testing to determine whether adaptive adjustment is needed;

[0053] S503, parameter optimization and protection execution, includes two parallel paths: one is the measurement parameter adaptive optimization path, and the other is the protection action linkage path.

[0054] S504, Closed-loop operation and cycle management, executed in a closed loop according to the cycle management mechanism: If Timer_Detect expires, execute steps S10 to S40; if the adaptive decision logic meets the conditions, execute step S503; if the line status is abnormal, execute circuit breaking or alarm; if the measurement quality index is greater than the threshold, the feature data is included in the template library for updating.

[0055] According to some embodiments, step S502 includes:

[0056] S5021, Calculate the signal-to-noise ratio ,in, The amplitude is the impedance spectrum after smoothing. This refers to the impedance spectrum residual or high-frequency noise component.

[0057] S5022, Calculate the measurement quality index ,in, As an ideal signal-to-noise ratio benchmark, The impedance spectrum fitting residuals, This serves as a reference value for the impedance fitting residual. This represents the change in the abnormality level between two consecutive detections. These are weighting coefficients;

[0058] S5023, if Then proceed to step S503, where, This is used to measure the quality index threshold.

[0059] According to some embodiments, step S502 further includes:

[0060] S5024, if the anomaly level increases and there is a continuous low signal-to-noise ratio Then proceed to step S503.

[0061] According to some embodiments, the adaptive optimization path for measurement parameters in step S503 includes:

[0062] If signal-to-noise ratio If it is too low, adjust the pulse amplitude;

[0063] If the high-frequency resolution is insufficient, adjust the pulse width;

[0064] If the power frequency interference is offset, adjust the center frequency of the band-stop filter;

[0065] If the spectral leakage is too large, adjust the window function type;

[0066] If the noise variance is large, adjust the adaptive averaging window.

[0067] The beneficial effects of this application are as follows:

[0068] A millisecond-level broadband current pulse signal is used as the excitation source. A broadband transient signal covering the frequency range from DC to hundreds of kHz is generated through a controllable electronic switch and a reference load circuit, achieving effective excitation of the impedance characteristics of the distribution network under low energy conditions. This pulse signal is short in time and continuous in frequency spectrum, which can simultaneously meet the requirements of broadband response acquisition and interference-free power supply and communication.

[0069] By synchronously acquiring voltage With current In the digital signal processing module, a windowed Fast Fourier Transform (FFT) combined with frequency domain smoothing and spline interpolation reconstruction is used to calculate the impedance spectrum. An improved dual-domain joint reconstruction algorithm is proposed: spectral smoothing is used in the frequency domain and transient deconvolution is used in the time domain to obtain impedance amplitude-frequency and phase-frequency characteristics with higher resolution and lower noise.

[0070] The system simultaneously monitors three signals: communication status, load fluctuations, and voltage waveforms, and constructs a three-condition joint decision logic based on C1 (communication idle), C2 (load stable), and C3 (voltage trough window). Only when these conditions are met... The detection process is triggered only when the excitation signal is injected, ensuring that the excitation signal is injected at the most stable and quietest moment of the power grid, thus realizing the detection strategy of "dynamic self-selection of disturbance-free window".

[0071] Extract a set of typical characteristic parameters from the reconstructed impedance spectrum. These parameters include the mean of the real part of the low-frequency circuit, the peak frequency of the imaginary part of the mid-frequency circuit, the phase angle slope, the equivalent time constant, and the spectral fitting residual. These parameters can reflect the changes in the resistance, inductance, capacitance, and dynamic response of the line, and can be used to intelligently identify conditions such as contact degradation, insulation aging, and grounding anomalies through template matching or machine learning models (SVM / CNN).

[0072] After the inspection is completed, the signal-to-noise ratio (SNR), measurement quality index (MQI), and changes in anomaly level are calculated to achieve a quantitative assessment of the inspection quality. The system automatically initiates a parameter adaptive adjustment process, dynamically correcting the pulse width, amplitude, filter center frequency, and window function type to improve subsequent measurement accuracy. This mechanism constructs a closed-loop control system from "detection → analysis → optimization → re-detection".

[0073] When the intelligent diagnostic module detects an anomaly level of L3 or higher, the control module immediately executes protective actions: first, it sends an alarm frame (including ID, status, and confidence level) through the communication channel, and then triggers the time-delay circuit breaker to achieve millisecond-level early warning and isolation control. Simultaneously, the action results and the updated feature vector are fed back to the template library for self-learning updates.

[0074] Significant advantages are achieved in method efficiency, detection accuracy, and engineering deployability through collaborative performance design. Firstly, to address the problem of "inability to detect online and the need for power outages," a joint judgment logic based on communication idle time, load stability, and voltage troughs is proposed. , , And, after determining the window, a transient current pulse with a peak value less than 1% of the rated current is injected, in milliseconds, so that the excitation is completed at a time imperceptible to the user, thereby achieving online impedance spectrum acquisition without interrupting power supply or affecting existing PLC / RF services. Secondly, to overcome the problem of "high-frequency injection being prone to interference and having limited resolution", the spectral characteristics of short-time broadband pulses are used ( By combining windowing / FFT, wavelet transform, and time-domain-frequency joint reconstruction, a single excitation can cover a wide frequency band, and high-resolution impedance curves can be obtained through spectral smoothing and cubic spline interpolation, significantly improving detection speed and spectral resolution. Thirdly, through hardware and software collaboration, the physical equivalent model is fitted (e.g., an improved Cole model) and the feature vectors are... The combination of template matching and machine learning ensures both physical interpretability and robustness in anomaly detection, thereby reducing reliance on expensive dedicated couplers / external devices and improving the economics of engineering deployment. This application constructs a closed-loop adaptive system: using SNR and MQI as quantitative indicators to judge measurement quality, automatically adjusting pulse width / amplitude, filter, and window parameters to improve the effectiveness of subsequent measurements, and using the diagnostic results for protection linkage (alarm + time-delay circuit breaker action) and online template library updates. This closed-loop mechanism not only continuously optimizes measurement performance and reduces false alarm rates, but also achieves millisecond-level response when serious anomalies are detected, providing both early warning and protection functions.

[0075] In summary, this application possesses significant technical advantages and strong engineering feasibility in achieving truly "disruptive, online, high-resolution, low-cost, and adaptive" impedance spectrum detection for distribution networks. The system injects low-energy, short-duration broadband current pulses within selected time slots to acquire the transient response signal of the line without interfering with power supply and communication. The broadband impedance spectrum is then reconstructed through time-frequency analysis, achieving high-resolution online detection of the distribution network. This application effectively avoids the dependence of traditional methods on external couplers, continuous excitation signals, and filtering circuits, while balancing real-time performance, low power consumption, wideband measurement, and high sensitivity. It can provide high-precision physical layer detection support for intelligent distribution areas and integrated sensing and control platforms. Attached Figure Description

[0076] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0077] Figure 1 A block diagram of a non-disruptive online impedance spectrum detection system for a distribution network based on a broadband current pulse, according to an example embodiment, is shown.

[0078] Figure 2 A flowchart of an online impedance spectrum detection method for a broadband current pulse-based distribution network according to an example embodiment is shown. Detailed Implementation

[0079] The embodiments of this application will now be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0080] Those skilled in the art should understand that the following specific embodiments or implementation methods are a series of optimized configurations listed in this application to further explain the specific application content. These configuration methods can be combined or used in conjunction with each other, unless this application explicitly states that some or a specific embodiment or implementation method cannot be associated with or used in conjunction with other embodiments or implementation methods. Furthermore, the following specific embodiments or implementation methods are only considered as optimized configurations and are not intended to limit the scope of protection of this application.

[0081] Example 1

[0082] Figure 1A block diagram of a non-disruptive online impedance spectrum detection system for a distribution network based on a broadband current pulse, according to an example embodiment, is shown.

[0083] like Figure 1 As shown, a non-disruptive online impedance spectrum detection system for distribution networks based on broadband current pulses mainly consists of the following five core modules:

[0084] Signal injection module (1): It consists of a controllable electronic switch (6), a reference load (7) and a drive circuit (8), and is responsible for generating a broadband current pulse signal with set parameters (pulse width, amplitude and timing) to achieve discontinuous, low-energy injection.

[0085] Signal acquisition module (2): includes voltage transformer (10), current transformer (9) and high-speed analog-to-digital converter (11), used to synchronously acquire voltage at the moment of injection. and current Transient response data.

[0086] Signal Processing and Spectrum Reconstruction Module (3): Calculates and reconstructs the impedance spectrum using digital signal processing algorithms (FFT, Short Time Fourier Transform (STFT), Wavelet Transform). And perform spectral smoothing and interpolation optimization.

[0087] Impedance feature identification and anomaly diagnosis module (4): Using the amplitude and phase frequency characteristics of the impedance spectrum, the health status of the line is diagnosed through template matching or machine learning algorithms.

[0088] Control and communication module (5): coordinates the system operation process and realizes detection timing control, result reporting and parameter adaptive optimization.

[0089] The specific implementation process of the functions and roles of each unit in the above system can be found in the implementation process of the corresponding steps in the method, and will not be repeated here.

[0090] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0091] Example 2

[0092] Figure 2 A flowchart of an online impedance spectrum detection method for a broadband current pulse-based distribution network according to an example embodiment is shown.

[0093] like Figure 2 As shown, a non-disruptive online impedance spectrum detection method for distribution networks based on broadband current pulses includes the following steps:

[0094] S10, system power-on initialization, S10 mainly includes three sub-steps:

[0095] S101, Hardware Self-Test and Parameter Preloading:

[0096] After the system is powered on, the control and communication module automatically executes a self-test program, sequentially checking the hardware status of the signal injection module, signal acquisition module, and signal processing module.

[0097] 1) Detection content and criteria

[0098] On-resistance of switching devices Should meet

[0099]

[0100] mutual inductor ratio error ,

[0101] ADC noise RMS < 3 LSB (Least Significant Bit).

[0102] Synchronous trigger clock error ,

[0103] If any parameter does not meet the requirements, the fault flag DEF_Error=1 is set, and subsequent detection is stopped.

[0104] 2) Loading system parameters

[0105] The system reads the following main parameter set from local storage or a remote configuration file:

[0106]

[0107] in: For the testing cycle; The pulse width is typically taken as... ; The peak current injected should not exceed 1% of the rated load. The sampling rate satisfies the Nyquist sampling theorem. ; This is the signal-to-noise ratio threshold, which defaults to 20dB. This is the threshold for impedance change, with a relative change of 5%.

[0108] 3) Calculation of sampling resolution and frequency domain accuracy

[0109] The system calculates the frequency resolution based on the sampling rate and the number of FFT points, using the following formula:

[0110]

[0111] in, For frequency resolution, Sampling rate, This represents the number of points in the FFT transform.

[0112] S102, Initialize status flags:

[0113] The control module sets and initializes the system running flags, as shown in Table 1:

[0114] Table 1

[0115]

[0116] The Timer_Detect is responsible for triggering the next stage (S20) "timing judgment" when the detection period is reached.

[0117] S103, Sensor calibration and sampling link synchronization:

[0118] To ensure the accuracy of the measurement data, the system performs a sampling link calibration and synchronization process upon startup.

[0119] 1) Voltage and current sensor gain and phase calibration

[0120] The system at standard impedance Inject test signal under load Measure the response and Calculate the complex transfer function using the Fast Fourier Transform (FFT):

[0121]

[0122] in, The complex transfer function obtained by measurement. Time-domain voltage signal Voltage spectrum after FFT transformation Time-domain current signal Current spectrum after FFT transformation For frequency.

[0123] Comparison with standard impedance characteristics The systematic error can be obtained as follows:

[0124]

[0125]

[0126] in, Indicates frequency Phase error or phase offset at the location; This represents the phase angle of the measured transfer function; The phase angle representing the standard impedance; Indicates frequency Amplitude gain error at the location; This indicates the magnitude of the measured transfer function; Indicates the magnitude of the standard impedance; For frequency.

[0127] The system according to and Generate a compensation table for phase and amplitude correction in subsequent measurements:

[0128]

[0129] in, This represents the corrected voltage spectrum signal; This represents the measured voltage spectrum signal; Indicates phase error; Indicates amplitude gain error; is the base of the natural logarithm; The imaginary unit; For frequency.

[0130] 2) ADC dynamic range and quantization noise

[0131] ADC bit width is The theoretical signal-to-noise ratio of the bit is:

[0132]

[0133] when hour, The actual total signal-to-noise ratio of the system meets the requirements. To ensure the relative error of impedance measurement .

[0134] 3) Sampling clock synchronization and jitter control

[0135] Set sampling period Requires triggering jitter .exist hour, Therefore, control The corresponding phase error is:

[0136]

[0137] This precision ensures that the phase spectrum measurement error is within [a certain range]. Within.

[0138] 4) Calibration qualification standards

[0139] Transformer ratio error ;

[0140] Phase shift ;

[0141] ADC channel bias

[0142] Shaking

[0143] Once the above conditions are met, the system sets the calibration completion flag CAL_OK=1 and proceeds to the next step S20.

[0144] S20, Detection Timing Determination:

[0145] This step comprehensively considers load current fluctuation characteristics, communication channel occupancy status, and system clock cycle to dynamically select the optimal injection time. The entire timing determination process includes three sub-steps:

[0146] S201, Periodicity Judgment:

[0147] The system controls the detection period through a timer named Timer_Detect. When the system time satisfy

[0148]

[0149] This triggers the detection process to enter stage S202. Among other things, This indicates the timestamp of the last detection.

[0150] To ensure uniform detection distribution and avoid conflicts with the communication cycle, the system uses a pseudo-random offset in each cycle. (range of values) Corrected trigger time:

[0151]

[0152] This randomization mechanism prevents the detection period from overlapping with the periodic structure of the power line communication (PLC) frame, thereby reducing the probability of interference.

[0153] If the system has not yet reached the detection cycle or the communication module is in data transmission state (BUSY=1), it will remain in standby state and enter the next cycle of polling.

[0154] S202, Load Condition Assessment:

[0155] The system monitors the line current in real time during the detection period. and in the sliding window The following two characteristic indicators are calculated internally:

[0156] 1) Average load current

[0157]

[0158] 2) Volatility and Variance

[0159]

[0160] when A larger value indicates frequent load fluctuations or the starting of motor-type equipment; when A lower value indicates a stable load.

[0161] To quantify the "stability", a perturbation factor is defined. :

[0162]

[0163] in This represents the maximum load variance during system calibration. If... (Without a disturbance factor threshold, the default threshold is 0.8), the load is considered to have entered a stable period and can proceed to the next stage.

[0164] In addition, the system uses Fast Fourier Transform (FFT) to... If frequency domain analysis is performed and there are no obvious spikes below 2 kHz (harmonic amplitude is less than 5% of the main wave amplitude), it can be determined that there is no high-power equipment switching impact in the current period.

[0165] S203, Uninterrupted Timing Decision Making:

[0166] Under the premise of stable load, the system further evaluates the communication occupancy status and power trough periods to comprehensively determine the injection window.

[0167] 1) Communication occupancy detection

[0168] The control module and the PLC communication unit share the bus occupancy flag BUSY. When a communication idle period is detected (BUSY=0 and duration>20ms), the system considers the communication link available for pulse injection.

[0169] 2) Voltage trough identification

[0170] The system is based on real-time voltage waveforms Extracting voltage troughs within each power frequency cycle :

[0171]

[0172] in, Indicates the time of voltage trough; This represents the value of the variable when it reaches its minimum value; This represents the real-time voltage waveform of the line. Represents a time variable.

[0173] Injecting near the trough (the crossover region where the voltage is approximately zero) can effectively reduce the amplitude of the interference current, making the injected signal "concealed" on the power supply waveform.

[0174] 3) Logic of Joint Judgment

[0175] The three conditions are defined in Table 2:

[0176] Table 2

[0177]

[0178] If all three conditions are met simultaneously If the system detects a value of 1, it sends a detection trigger signal TRIG_DETECT=1 and enters the next stage S30; otherwise, the system returns to S201 to continue polling and judging.

[0179] 4) Time window calculation formula

[0180] Voltage trough width With power frequency cycle The relationship of s is:

[0181]

[0182] in The proportionality coefficient is (0.05–0.1), meaning approximately [a certain value] per cycle. Window. This window is sufficient to complete the pulse injection ( ) and synchronous sampling initialization.

[0183] 5) Timing Decision-Making Process Logic

[0184] Communication idle detection → Load fluctuation judgment → Voltage trough detection → Logic AND decision of the three → If the conditions are met, start the pulse injection process.

[0185] S30, broadband pulse injection and synchronous acquisition.

[0186] Step S30 generates a controllable broadband current pulse under normal power grid supply conditions and simultaneously acquires the transient response signals of line voltage and current to achieve impedance spectrum analysis. High-resolution computation. Step S30 includes:

[0187] S301, Parameter Loading and Module Self-Test:

[0188] After detecting the trigger signal TRIG_DETECT = 1, the system control module (module 5) performs preparatory work before pulse injection:

[0189] 1) Parameter loading

[0190] Retrieve the current round of detection parameter set from non-volatile memory. ,in, This is the pulse width, typically valued at... ; This represents the pulse amplitude, with a typical value of [value missing]. Rated current; To analyze the upper limit frequency, it is generally... ; The average number of iterations is used for multiple stacking noise reduction. This is the gain of the pre-sampling amplifier.

[0191] 2) Module self-test

[0192] The control module detects the gate drive state of the electronic switch (MOSFET or IGBT) to confirm its turn-off delay. Check the clock synchronization status of the sampling module (clock drift). If any item is abnormal, an alarm FLAG_HWERR=1 is issued, and the current round of detection is terminated.

[0193] S302, synchronous trigger signal transmission:

[0194] The system adopts a single-master synchronous clock architecture. The main control MCU generates a synchronous trigger signal SYNC_TRIG, which is distributed to the signal injection module and the signal acquisition module via an isolation drive circuit. The two modules use the same clock source to ensure triggering error. Its triggering logic is expressed as follows:

[0195]

[0196] in The system clock calibration offset is precisely compensated by the FPGA; Indicates the injection trigger time; This indicates the current system reference time. This ensures that injection and sampling begin at the same reference time, achieving high-precision "time-domain alignment".

[0197] S303, broadband current pulse generation:

[0198] This step corresponds to the "signal injection module". The signal injection circuit consists of a controllable electronic switch, a drive circuit, and a reference load, forming a low-impedance closed loop.

[0199] 1) Mathematical model of pulse current

[0200] Injected current Defined in time Start, continue Unipolar rectangular pulse:

[0201]

[0202] Its spectrum (through Fourier transform) is

[0203]

[0204] in The spectrum is in The area is approximately flat, forming a broadband excitation.

[0205] 2) Segmented injection strategy (if) )

[0206] If a single pulse cannot be completed within the trough window, segmented injection is used:

[0207]

[0208] in, It is a step function. Inject time for segment k.

[0209] Subsequently, the overall response is synthesized after phase calibration in the frequency domain:

[0210]

[0211] Amplitude continuity is maintained through time alignment compensation.

[0212] 3) Drive timing control

[0213] The electronic switch gate is driven by a PWM waveform controlled by an FPGA, ensuring that the switch rises along the correct time. The drive signal sequence is as follows:

[0214] t0: Rising edge of SYNC_TRIG

[0215] t0+Δt_drv: Switch is on.

[0216] t0+Δt_drv+W1: Switch off.

[0217] Δt_drv is the gate delay compensation amount, and its typical value is... .

[0218] S304, Transient Response Synchronous Acquisition:

[0219] The signal acquisition module includes a current transformer, a voltage transformer, and a high-speed analog-to-digital converter. Its sampling rate... The Nyquist criterion should be satisfied. Typical configuration is .

[0220] Sampling signal:

[0221]

[0222] Record timestamp ,

[0223] The sampling trigger and the injected synchronization signal SYNC_TRIG are from the same source, and the error is the same. This ensures that the time-domain signal can be used for direct spectrum calculation.

[0224] The sampling front end is equipped with a band-stop filter to suppress... Power frequency component:

[0225]

[0226] in .

[0227] S305, Signal Preprocessing:

[0228] To improve the accuracy of spectrum analysis, the following steps are performed on the acquired signal:

[0229] 1) Remove DC component

[0230]

[0231] Similarly, deal with:

[0232]

[0233] 2) Adding windows

[0234] Using the Hanning window to reduce spectral leakage

[0235]

[0236] Receive windowing signal ,in This represents the total number of sampling points.

[0237] 3) Spectral smoothing

[0238] The subsequent FFT results were smoothed using a moving average method to compensate for high-frequency noise caused by pulse spikes.

[0239] S306, Time-Frequency Transformation and Impedance Calculation:

[0240] Perform a Fast Fourier Transform (FFT) on the preprocessed voltage and current signals:

[0241]

[0242]

[0243] The formula for calculating impedance spectrum is:

[0244]

[0245] Obtaining complex impedance ,in This is the real part (resistance). This is the imaginary part (reactor).

[0246] Amplitude and phase frequency characteristics:

[0247]

[0248] Calculated The data is stored in the impedance spectrum buffer Z_spectrum[] for subsequent feature analysis.

[0249] S307, Spectrum Optimization and Curve Reconstruction:

[0250] To eliminate sampling noise and FFT discretization error, a spectral reconstruction algorithm is used:

[0251] 1) Smoothing Filter

[0252]

[0253] 2) Interpolation Reconstruction

[0254] Cubic spline interpolation is used for sparse frequency points to obtain high-resolution curves:

[0255]

[0256] in, This represents the high-resolution impedance spectrum after interpolation reconstruction. This represents the cubic spline interpolation operator; This represents the original discrete impedance data sequence; Represents the original discrete frequency points; This represents the target frequency variable.

[0257] 3) Phase correction (used for segmented injection synthesis)

[0258] If segmented injection is used, phase compensation needs to be performed on each segment of the spectrum:

[0259]

[0260] in This represents the time offset of the k-th segment; Represents the equivalent combined impedance spectrum; Indicates the total number of segments in the segmented injection; Indicates a segmented index; Indicates the first The time offset of segment injection; Indicates frequency; This represents the phase compensation factor.

[0261] The final output is a complete broadband impedance spectrum. Frequency range The resolution is better than .

[0262] S40, Impedance Characteristic Analysis and Anomaly Diagnosis.

[0263] Step S40 aims to analyze the complex impedance spectrum obtained in step S30. Feature extraction, pattern recognition, and anomaly detection are performed to output the health status and diagnostic results of distribution network nodes, including...

[0264] S401, Impedance Spectrum Data Preprocessing:

[0265] Impedance spectrum After sampling and reconstruction, frequency band normalization and noise correction are first performed:

[0266] 1) Logarithmic transformation of frequency

[0267] Converting linear frequencies to a logarithmic scale enhances low-frequency detail resolution.

[0268]

[0269] The corresponding complex impedance transformation is:

[0270]

[0271] 2) Outlier removal

[0272] Using the sliding window standard deviation method:

[0273]

[0274] in, This represents a local average value. Indicates local standard deviation; Indicates the width of the sliding window; Indicates the index of the current frequency point; Indicates the summation index within the window; This represents the impedance data within the window.

[0275] like If the value is an outlier, it is considered an outlier and replaced with the local mean.

[0276] 3) Amplitude and phase frequency separation processing

[0277] Output two sequences and These are used for amplitude feature extraction and phase feature extraction, respectively.

[0278] S402, Feature Parameter Extraction:

[0279] From the impedance spectrum, typical characteristic quantities that reflect line characteristics, load conditions, and contact degradation are extracted, and the following characteristic set is defined: See Table 3 for details:

[0280] Table 3

[0281]

[0282] in, This represents the number of low-frequency sampling points; This is the low-frequency cutoff point, typically 100Hz; The critical frequency at which the Bode plot amplitude-frequency curve drops by 3dB; This represents the real part of the impedance, i.e., the resistive component; Represents frequency variables; This represents the imaginary part of the impedance, i.e., the reactance component; This represents the differential operator; here it indicates the taking of the derivative. Indicates the impedance phase angle; The logarithm of frequency; This represents the total number of frequency points calculated during the fitting process; Indicates the frequency point index; Indicates frequency The measured impedance value at the location; Indicates frequency The fitted impedance value at that point.

[0283] S403, Impedance Characteristic Modeling and Template Library Establishment:

[0284] A "dual-path modeling" approach is adopted, with one path being the physical mechanism modeling path (equivalent circuit method) and the other being the data-driven modeling path (feature clustering method).

[0285] 1) Fitting the physical equivalent model

[0286] Fitting the impedance spectrum using an improved Cole model:

[0287]

[0288] in, This refers to the low-frequency resistance, i.e., the total impedance of the line. This refers to the high-frequency resistance, i.e., the resistance of the cable itself. It is the equivalent time constant; It is the fractal index, and its value range is... .

[0289] The least squares method is used to estimate ( Construct an equivalent parameter vector. .

[0290] 2) Feature clustering model

[0291] For historical datasets The K-means clustering algorithm is used to classify the state categories:

[0292]

[0293] in The k-th type of center corresponds to the operating states such as "normal", "slightly aged", and "severely degraded". Indicates the first Feature vectors of each sample; Indicates a category index; This represents the total number of historical samples; Represents the square of the Euclidean distance; This indicates a minimize operation.

[0294] After the database is built, the template library structure is as follows, as shown in Table 4:

[0295] Table 4

[0296]

[0297] S404, Abnormal Diagnosis and Status Judgment:

[0298] The diagnostic process consists of two logical layers:

[0299] 1) Threshold judgment layer

[0300] Based on feature deviation determination:

[0301]

[0302] in, Indicates relative to the first Feature deviation of class state template; This represents the feature vector currently being detected; Indicates the first Feature vectors of class templates; Represents the square of the Euclidean distance; This represents the square of the template vector's magnitude; Indicates a template or category index.

[0303] like If the condition is not met, it is determined as an "unknown anomaly"; otherwise, it proceeds to the template matching layer.

[0304] 2) Template Matching Layer

[0305] Calculate cosine similarity:

[0306]

[0307] in, Indicates cosine similarity; This represents the dot product of vectors.

[0308] Calculate the category corresponding to the maximum similarity Output the diagnostic results:

[0309]

[0310] S405, Anomaly Level Quantification and Visualization Output:

[0311] The abnormality level of the diagnostic results was calculated using weighted rules:

[0312]

[0313] in An index indicating the severity level of the anomaly; This represents the reference value for characteristic parameter 1; This represents the reference value for characteristic parameter 3; Indicates the best matching category similarity; Indicates the diagnosed state category; This represents the similarity residual, reflecting the degree of deviation between the current state and the standard template. The experience weights are typically 0.3, 0.4, and 0.3, respectively. The grading is shown in Table 5.

[0314] Table 5

[0315]

[0316] The output includes a graphical impedance spectrum (amplitude-frequency + phase-frequency curves), template matching results (category name + similarity), anomaly level bar chart, and a diagnostic report JSON data package.

[0317] S50, Adaptive Closed-Loop Control and System Protection Logic.

[0318] This step is responsible for feeding back the impedance spectroscopy detection and anomaly diagnosis results to the system control module, enabling adaptive parameter adjustment, measurement quality improvement, and coordinated safety protection actions, thus forming a complete closed-loop detection-analysis-optimization mechanism. It includes the following steps:

[0319] S501, Detection result reporting and data storage:

[0320] After the detection and diagnosis results are generated by S40, they are uniformly encapsulated and reported by the control and communication module. The reported content is shown in Table 6.

[0321] Table 6

[0322]

[0323] The reporting system uses a half-duplex low-speed communication link, with a typical rate of 9600 bps, and message formatting is based on the Modbus / RTU protocol. Simultaneously, data is stored locally in EEPROM for subsequent trend analysis and historical comparison.

[0324] S502, Performance Evaluation and Adaptive Decision:

[0325] After each round of testing, the system assesses the measurement quality and anomaly risk to determine whether adaptive adjustments are necessary. This process includes the following three core calculations:

[0326] 1) Signal-to-noise ratio calculation

[0327]

[0328] in, The amplitude is the impedance spectrum after smoothing. This refers to the impedance spectrum residual or high-frequency noise component.

[0329] 2) Measure the Quality Index (MQI)

[0330]

[0331] in, As an ideal signal-to-noise ratio benchmark; The impedance spectrum fitting residual; This serves as a reference value for the impedance fitting residual. This represents the change in the abnormality level between two consecutive detections; These are the weighting coefficients.

[0332] when At that time, the adaptive optimization process is triggered.

[0333] in, This is used to measure the quality index threshold.

[0334] 3) Adaptive decision logic

[0335] The adaptive decision logic conditions are defined in Table 7:

[0336] Table 7

[0337]

[0338] in, The threshold for abnormal level changes is a critical increment value defined by the system to determine whether the health status of the line has significantly deteriorated.

[0339] The joint logic is:

[0340] like Then proceed to the parameter optimization process.

[0341] S503, Parameter Optimization and Protection Execution:

[0342] This step is crucial for achieving the system's "self-evolution" characteristic and involves two parallel paths:

[0343] 1) Adaptive optimization path for measurement parameters

[0344] The system automatically adjusts the detection pulse and signal processing strategy to improve the quality of subsequent measurements. Key details are shown in Table 8.

[0345] Table 8

[0346]

[0347] Where A1 represents the pulse amplitude; ← represents the assignment update operation; k1 represents the amplitude growth coefficient; W1 represents the pulse width; Indicates the pulse width increment step size; f_c represents the center frequency of the band-stop filter; ± indicates the adjustment direction; This indicates the frequency fine-tuning step size; Hann represents the Hanning window, which is the default window function used by the system and has a good balance between frequency resolution and sidelobe suppression; Blackman represents the Blackman window, which is a window function with greater sidelobe attenuation and less spectral leakage.

[0348] After the parameters are updated, the system recalculates the expected signal-to-noise ratio improvement factor:

[0349]

[0350] in, This represents the expected signal-to-noise ratio improvement factor; This indicates the optimized signal-to-noise ratio; This represents the signal-to-noise ratio before optimization.

[0351] like If the optimization is successful, Flag_Adapt is set to 1; otherwise, the optimization iteration continues for a maximum of 3 rounds.

[0352] 2) Protection action linkage path

[0353] When the anomaly level reaches L3 or above, the system activates protection logic, which is divided into two levels: software alarm and hardware action.

[0354] ① Software alarm: The control module sends an alarm frame through the communication interface.

[0355] [ID_node][State][Severity][Timestamp][CRC],

[0356] Wherein, ID_node is the node identifier, representing the unique address or number of the detection device that issued the alarm signal in the power distribution IoT, used by the host computer to identify the specific location of the fault; State is the status type code, representing the specific line status category determined by S404 diagnosis; Severity is the anomaly severity level, representing the risk level after S405 quantitative assessment, used to indicate the priority of maintenance personnel; Timestamp is the timestamp, representing the system time when the anomaly was detected or the alarm was generated, used for historical fault tracing and time sequence analysis; CRC is the cyclic redundancy check code, used to verify the integrity and correctness of the entire data frame during transmission, preventing data errors caused by communication interference.

[0357] At the same time, an audible and visual alarm is issued on the host computer interface.

[0358] ② Hardware action: If the detection node is a low-voltage feeder terminal, the control logic initiates a delayed action of the electronic circuit breaker:

[0359] Delay determination: ,

[0360] Typical value , ,

[0361] If the abnormality persists after the delay ends, the corresponding branch will be disconnected to prevent the fault from spreading.

[0362] ③ Status feedback: After the action is completed, a status frame is reported to the monitoring master station, and the operation map of the station area is updated.

[0363] S504, Closed-Loop Operation and Cycle Management:

[0364] The closed-loop operation of the entire system follows the cycle management mechanism shown in Table 9:

[0365] Table 9

[0366]

[0367] Periodic governing equations:

[0368]

[0369] in, For the next testing cycle after the update; This refers to the current or previous testing cycle; This is an index indicating the severity of the anomaly. This represents the boundary of the normal state. This is the boundary of mild abnormality; This marks the boundary between moderate and severe abnormalities.

[0370] Example 3

[0371] In a laboratory environment, a low-voltage power distribution line simulation platform was built to verify the non-intrusive impedance spectrum detection system of this application.

[0372] (1) Implementation plan and structure

[0373] Simulated power distribution line: A 50-meter-long copper core cable with a cross-sectional area of ​​2.5 mm² is used, and the end is connected to a variable load box (including resistive, inductive and capacitive loads) to simulate the operating conditions of an actual low-voltage power distribution network.

[0374] System hardware modules:

[0375] Signal injection module: MOSFET is used as a controllable electronic switch, the reference load is a 0.1Ω non-inductive power resistor, and the drive circuit generates precise pulse timing based on FPGA.

[0376] Signal acquisition module: Voltage transformer (1:1 ratio), current transformer (100:1 ratio) and 16-bit high-speed ADC (500kHz sampling rate) achieve synchronous sampling.

[0377] Processing and control unit: Embedded DSP+ARM dual-core processor, responsible for signal processing, feature extraction, diagnostic algorithms and communication control.

[0378] Communication interface: The RS485 module is used to communicate with the host computer and supports the Modbus / RTU protocol.

[0379] Software platform: LabVIEW-based host computer software used for parameter configuration, data visualization, diagnostic result display, and historical data management.

[0380] System connection method: The signal injection module is connected in parallel at the beginning of the line, the voltage and current transformers are connected to the line respectively, and the acquisition module and the control unit achieve time domain alignment through synchronous trigger signals.

[0381] (2) Specific parameter configuration

[0382] In laboratory testing, the system parameter settings are shown in Table 10:

[0383] Table 10

[0384]

[0385] (3) Steps and functions

[0386] This embodiment executes the complete detection process, and the steps are as follows:

[0387] S10, System power-on initialization:

[0388] The system performs a self-test after power-on: switch on-resistance. The current transformer error is less than 1%, and the ADC noise is less than 3 LSB.

[0389] Load parameter set , , , Initialize the flags (DEF_Enable=1, Timer_Detect=current time+60s).

[0390] Perform sensor calibration: at standard impedance Inject a test pulse and calculate the amplitude / phase compensation table.

[0391] Function: To ensure that the hardware is in good condition and that the measurement link accuracy meets the standards.

[0392] S20, Detection Timing Determination:

[0393] Periodicity check: Check every 60 seconds whether the detection time has been reached.

[0394] Load assessment: Calculate the current variance within the sliding window ,like It is then determined to be stable.

[0395] Communication and voltage detection: Confirm that the PLC communication is idle (BUSY=0) and within the voltage trough window (±0.5ms).

[0396] Joint judgment: If If so, a detection will be triggered.

[0397] Function: To dynamically select a time when there is no disturbance, so as to avoid interference with the power grid and communications.

[0398] S30, Broadband Pulse Injection and Synchronous Acquisition:

[0399] Parameter loading and self-test: Invocation , Confirm that the switch driver is working properly.

[0400] Synchronous Trigger: The FPGA sends a SYNC_TRIG signal to initiate synchronous startup with the acquisition module.

[0401] Pulse generation: A rectangular current pulse is injected 2ms after the MOSFET is turned on. Its spectrum covers 0-100kHz.

[0402] Synchronous acquisition: ADC records at a sampling rate of 500kHz and It also performs DC removal and Hanning window preprocessing.

[0403] Impedance calculation: obtained by FFT and ,calculate And perform spectral smoothing and interpolation reconstruction.

[0404] Function: To obtain high-resolution broadband impedance spectra.

[0405] S40, Impedance Characteristic Analysis and Anomaly Diagnosis:

[0406] Feature extraction: Calculating feature sets This includes the average real part of the low-frequency component and the peak frequency of the imaginary part of the mid-frequency component.

[0407] Template matching: Calculate the cosine similarity with pre-stored templates (normal, contact oxidation, cable aging, grounding abnormality).

[0408] Status judgment: Output diagnostic results (such as "contact oxidation") and abnormality level (such as L2).

[0409] Function: To enable intelligent identification and quantitative assessment of the health status of the circuit.

[0410] S50, Adaptive Closed-Loop Control and System Protection:

[0411] Results reporting: Encapsulated data (node ​​ID, timestamp, impedance spectrum, diagnostic results, SNR) is sent to the host computer.

[0412] Performance evaluation: calculation , ,like This triggers adaptive optimization.

[0413] Parameter optimization: Adjustment or To improve the signal-to-noise ratio.

[0414] Protection linkage: If Send an alarm frame and activate the time-delay circuit breaker ( ).

[0415] Function: To achieve closed-loop control of detection-analysis-optimization, thereby improving system robustness and safety.

[0416] (4) Effects of the Example

[0417] Laboratory tests have shown the following results:

[0418] High-precision impedance spectrum acquisition: The system successfully reconstructed the impedance amplitude-frequency and phase-frequency curves in the range of 0-100kHz, with a frequency resolution of 488 Hz, amplitude error <5%, and phase error <3°.

[0419] High accuracy of intelligent diagnosis: When simulating "contact oxidation" fault, the system correctly identifies the state (similarity > 0.85) and judges the abnormal level as L2; ​​under normal state, the similarity is > 0.9.

[0420] Non-interference verification: During pulse injection, the line voltage fluctuation was <0.5%, and the PLC communication bit error rate remained unchanged, confirming its "non-interference" characteristic.

[0421] Effective adaptive optimization: When the signal-to-noise ratio drops to 18dB, the system automatically adjusts the pulse amplitude to 2.2A, the SNR increases to 26dB, and the measurement quality index (MQI) recovers from 0.68 to 0.76.

[0422] Real-time performance and reliability: The entire process takes about 3 seconds, which meets the requirements of online monitoring; during a continuous 72-hour test, the system operated stably without any false alarms or missed alarms.

[0423] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A non-disruptive online impedance spectrum detection system for distribution networks based on broadband current pulses, characterized in that, It includes a signal injection module, a signal acquisition module, a signal processing and spectrum reconstruction module, an impedance feature identification and anomaly diagnosis module, and a control and communication module; The signal injection module receives a synchronous trigger signal generated by the control and communication module, and generates a broadband current pulse to inject into the distribution network line according to the synchronous trigger signal. The signal acquisition module acquires the voltage response and current response of the distribution network line and digitizes the voltage response and current response. The signal processing and spectrum reconstruction module receives the digitized voltage response and digitized current response output by the signal acquisition module and calculates the impedance spectrum of the distribution network line. The impedance feature identification and anomaly diagnosis module receives the impedance spectrum of the distribution network line output by the signal processing and spectrum reconstruction module to diagnose the health status of the line. The control and communication module receives the diagnostic data and feature data generated by the impedance feature identification and anomaly diagnosis module and reports the results to the distribution automation master station. The signal injection module includes a controllable electronic switch, a reference load, and a drive circuit, used to generate broadband current pulse signals; The signal acquisition module includes a voltage transformer, a current transformer, and a high-speed analog-to-digital converter, used to simultaneously acquire transient response data of voltage and current of the distribution network lines at the instant broadband current pulse signal is injected into the distribution network lines; the signal processing and spectrum reconstruction module calculates and reconstructs the impedance spectrum of the distribution network lines through digital signal processing algorithms, and performs spectrum smoothing and interpolation optimization; the impedance feature identification and anomaly diagnosis module utilizes the amplitude-frequency and phase-frequency characteristics of the impedance spectrum of the distribution network lines to achieve line health status diagnosis through template matching or machine learning algorithms; the control and communication module coordinates the system operation process, realizing detection timing control, result reporting, and parameter adaptive optimization.

2. A non-disruptive online impedance spectrum detection method for distribution networks based on broadband current pulses, characterized in that, Includes the following steps: S10, Power-on initialization is performed on the non-disruptive online impedance spectrum detection system for distribution networks based on broadband current pulses as described in claim 1, and system self-test, parameter loading, and sampling link calibration are completed to prepare for online detection; S20, determine the detection timing based on the load status and communication occupancy; S30, Broadband Pulse Injection and Synchronous Acquisition, injects broadband current pulses into the distribution network lines during safe time slots and acquires the response of the distribution network lines in real time; S40, Impedance Characteristic Analysis and Anomaly Diagnosis: Calculates the impedance spectrum of distribution network lines and extracts features for intelligent diagnosis. S50, adaptive closed-loop control and system protection, realizes adaptive parameter adjustment, measurement quality improvement and safety protection action linkage, forming a complete closed-loop detection, analysis and optimization mechanism; Step S20 includes the following steps: S201, Periodicity determination, the detection period is controlled by the timer Timer_Detect. If the system time satisfy Then proceed to step S202, where, Indicates the timestamp of the last detection; S202, Load Condition Assessment, real-time monitoring of line current during the detection period. and in the sliding window Internal calculation of average load current and volatility variance The average load current for The fluctuation variance for Calculate the perturbation factor , ,like If the load enters a stable period, proceed to step S203; otherwise, return to step S201. The threshold for the perturbation factor. The maximum load variance during system calibration; S203, Evaluate the communication occupancy status; if the communication unit communication bus in the control and communication module of claim 1 has been idle for a certain period of time... Then proceed to step S204; otherwise, return to step S201. Minimum communication bus idle time; S204, if the current time is within the voltage trough window, proceed to step S30; otherwise, return to step S201. The voltage trough window is defined as the trough time of the fundamental voltage wave. Centered on, the width of the voltage trough is The time window; the trough moment The calculation method is as follows: ,in, This represents the value of the variable when it reaches its minimum value. This indicates the real-time voltage waveform of the line. Represents the time variable, voltage trough width The calculation method is as follows ,in This is the proportionality coefficient. To align with the power frequency cycle.

3. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 2, characterized in that, Step S30 includes the following steps: S301, parameter loading and module self-test: the signal injection module of claim 1 performs parameter loading and module self-test; S302 sends a synchronization trigger signal, adopting a single master synchronous clock architecture. The master control MCU generates a synchronization trigger signal SYNC_TRIG, which is distributed to the signal injection module and the signal acquisition module of claim 1 via an isolation drive circuit. S303, Generate a broadband current pulse according to the SYNC_TRIG synchronous trigger signal and inject it into the distribution network line; S304, Transient response synchronous acquisition, the signal acquisition module of claim 1 acquires the response of the power distribution network line; S305, signal preprocessing, performs DC component removal, windowing, and spectrum smoothing on the acquired signal; S306, Time-Frequency Transformation and Impedance Calculation: Performs Fast Fourier Transform on preprocessed voltage and current signals to calculate the impedance spectrum of distribution network lines. ,in For the real part, It is the imaginary part; S307, Spectrum Optimization and Curve Reconstruction: Spectrum reconstruction is performed to eliminate sampling noise and Fast Fourier Transform discretization errors. The spectrum reconstruction includes smoothing filtering and interpolation reconstruction. The smoothing filtering calculation formula is the smoothed impedance spectrum. ,in Represents the original discrete frequency points. N is the number of sampling points. The interpolation reconstruction includes cubic spline interpolation of sparse frequency points to obtain a high-resolution curve. ,in, This represents the high-resolution impedance spectrum after interpolation reconstruction. This represents the cubic spline interpolation operator. This represents the original discrete impedance data sequence. Represents the original discrete frequency points. This represents the target frequency variable.

4. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 3, characterized in that, Step S40 includes the following steps: S401, Impedance spectrum data preprocessing, the impedance spectrum After sampling and reconstruction, frequency band normalization and noise correction are performed. The frequency band normalization and noise correction include frequency logarithmic transformation, outlier removal, and amplitude frequency and phase frequency separation processing. S402, Feature parameter extraction: Extracting typical characteristic quantities from the impedance spectrum that reflect line characteristics, load conditions, and contact degradation; the typical characteristic quantities include: low-frequency real part average value. Mid-frequency imaginary peak frequency Impedance phase angle slope Equivalent time constant Impedance fitting residual The average value of the low-frequency real part The peak frequency of the imaginary part of the intermediate frequency is a characteristic quantity reflecting the DC resistance and contact performance of the line. The impedance phase angle slope is a characteristic quantity related to the distribution of inductive and capacitive loads. The equivalent time constant is a characteristic quantity that reflects the rate of change of frequency response. The impedance fitting residual is a characteristic quantity used to characterize the dynamic response of a node. These are characteristic quantities that indicate the complexity or anomalies of a line. S403, Impedance characteristic modeling and template library establishment, using a dual-path modeling method for impedance characteristic modeling, wherein one modeling path is a physical mechanism modeling path and the other modeling path is a data-driven modeling path; S404, Abnormal Diagnosis and Status Determination, includes the following steps: Step S40401, based on feature deviation Make a judgment. ,in, Indicates relative to the first Feature deviation of class state template This represents the feature vector currently being detected. Indicates the first Feature vectors of class templates This represents the square of the Euclidean distance. This represents the square of the template vector's magnitude. Indicates the index of the template or category; if If the condition is determined to be an unknown anomaly, proceed to step S40402. The threshold for determining feature deviation; Step S40402: Calculate cosine similarity. ,in, Represents cosine similarity. This represents the dot product of vectors, used to calculate the maximum similarity between categories. Based on the maximum similarity, the corresponding category Determine the state category; S405, Anomaly Level Quantization, Calculates Anomaly Level ,in, Indicating the severity level index of the anomaly. express The reference base value, express The reference base value, Indicates the best matching category similarity. Indicates the diagnosed state category. This represents the similarity residual, reflecting the degree of deviation between the current state and the standard template. For experience weights.

5. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 4, characterized in that, The physical mechanism modeling path described in step S403 includes: fitting the impedance spectrum using an improved Cole model. ,in, This represents the total impedance of the line. The resistance of the cable itself. The equivalent time constant, The fractal index, The range of values ​​is The least squares method is used to estimate ( Construct an equivalent parameter vector. .

6. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 4, characterized in that, The data-driven modeling path described in step S403 includes: targeting historical datasets The K-means clustering algorithm was used to classify the state categories. ,in, The k-th type of center corresponds to normal, slightly aged, and severely degraded operating states. Indicates the first The feature vector of each sample Indicates category index, This represents the total number of historical samples. This represents the square of the Euclidean distance. This indicates a minimize operation.

7. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 4, characterized in that, Step S50 includes: S501, Detection result reporting and data storage: The detection and diagnosis result data is uniformly encapsulated and reported by the control and communication module described in claim 1, and the data is stored locally at the same time. S502, Performance Evaluation and Adaptive Decision, assesses measurement quality and anomaly risk after each round of testing to determine whether adaptive adjustment is needed; S503, parameter optimization and protection execution, includes two parallel paths: one is the measurement parameter adaptive optimization path, and the other is the protection action linkage path. S504, Closed-loop operation and cycle management, executed in a closed loop according to the cycle management mechanism: If Timer_Detect expires, execute steps S10 to S40; if the adaptive decision logic meets the conditions, execute step S503; if the line status is abnormal, execute circuit breaking or alarm; if the measurement quality index is greater than the threshold, the feature data is included in the template library for updating.

8. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 7, characterized in that, Step S502 includes: S5021, Calculate the signal-to-noise ratio ,in, The amplitude is the impedance spectrum after smoothing. This refers to the impedance spectrum residual or high-frequency noise component. S5022, Calculate the measurement quality index ,in, As an ideal signal-to-noise ratio benchmark, The impedance spectrum fitting residuals, This serves as a reference value for the impedance fitting residual. This represents the change in the abnormality level between two consecutive detections. These are weighting coefficients; S5023, if Then proceed to step S503, where, This is used to measure the quality index threshold.

9. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 8, characterized in that, Step S502 further includes: S5024, if the anomaly level increases and there is a continuous low signal-to-noise ratio Then proceed to step S503.

10. The method for online impedance spectrum detection of a distribution network based on broadband current pulses according to claim 7, characterized in that, The adaptive optimization path for measurement parameters in step S503 includes: If signal-to-noise ratio If it is too low, adjust the pulse amplitude; If the high-frequency resolution is insufficient, adjust the pulse width; If the power frequency interference is offset, adjust the center frequency of the band-stop filter; If the spectral leakage is too large, adjust the window function type; If the noise variance is large, adjust the adaptive averaging window.

Images