An online monitoring and analysis system for potential risks in power grid inverter boosting

Abstract

This invention relates to the field of power electronic converter operation and control technology, and discloses an online monitoring and analysis system for potential hazards in grid inverter boost converters. The system includes an inverter unit, a sensing unit, and a control unit. The control unit synchronously acquires discrete sequences of DC bus voltage and zero-sequence leakage current, reconstructs common-mode voltage excitation, and extracts complex common-mode admittance at multiple characteristic frequencies. By calculating the frequency multiple relationship deviation of the admittance imaginary part sequence, the carrier frequency is shifted to avoid the series resonance region. When the voltage phasor amplitude is lower than the judgment threshold, an asymmetric factor is used to adjust the zero-vector dwell time allocation ratio for perturbation, exciting high-frequency common-mode energy. Subsequently, a multi-frequency admittance overdetermined matrix model is established, and the low-frequency equivalent insulation conductance is separated by inversion. The insulation state is then evaluated through smoothing filtering and differential operations. This invention overcomes the interference of common-mode resonance and operating condition excitation attenuation, improving the accuracy of online monitoring of insulation conductance parameters and the reliability of hazard early warning.

Description

Technical Field

[0001] This invention relates to the field of power electronic converter operation and control technology, specifically to an online monitoring and analysis system for potential risks in power grid inverter boosting. Background Technology

[0002] In grid-connected converter systems, the topology of high-power inverters combined with long-distance AC cables is quite common. To ensure the safe operation of the equipment, it is necessary to monitor the system's insulation status to ground online. Most existing passive insulation monitoring schemes use the common-mode voltage generated by the switching action of the converter as an excitation and calculate the insulation parameters of the system by detecting the zero-sequence leakage current.

[0003] Such methods have limitations in practical applications. Long-distance AC cables contain parasitic capacitance and inductance to ground, and these distributed parameters can induce common-mode circuit series resonance in certain high-frequency bands. Under resonance, the impedance frequency characteristics of the system are distorted, and the admittance data extracted based on conventional fixed frequency points deviates from the actual insulation impedance, leading to parameter calculation errors. At the same time, when the inverter is running under full load or high modulation index conditions, the amplitude of the output high-frequency common-mode voltage will attenuate, and the weak excitation signal will cause the collected leakage current to be submerged by electromagnetic environmental noise, causing the insulation parameter analytical model to fail due to insufficient signal-to-noise ratio. In addition, the insulating medium has a polarization effect in the AC electric field, and its high-frequency extracted conductivity parameters are superimposed with frequency-varying loss components, which cannot be directly equivalent to the low-frequency leakage conductivity characterizing the degree of insulation damage. The above-mentioned operating condition interference and medium physical characteristics make it difficult for existing insulation monitoring technologies to continuously extract accurate insulation assessment parameters during operation, which may cause the early warning of insulation degradation risks to fail or false alarm. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an online monitoring and analysis system for potential hazards in power grid inverter boost converters. This system solves the problems of inaccurate extraction of insulation equivalent conductivity parameters and false alarms in potential hazard warnings caused by interference from common-mode resonance, operating condition excitation attenuation, and polarization effects in existing grid-connected converter systems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: This invention provides an online monitoring and analysis system for potential risks in power grid inverter boosting, comprising an inverter unit, a sensing unit, and a control unit. The inverter unit includes a first inverter subunit, a second inverter subunit, and a third inverter subunit connected in sequence, with the third inverter subunit used to connect to an external output unit. The sensing unit includes a first sensing subunit installed on the first inverter subunit and a second sensing subunit installed on the third inverter subunit. The control unit is connected to the first sensing subunit, the second sensing subunit, and the second inverter subunit. The control unit is used to extract complex common-mode admittance based on electrical signals collected by the sensing unit, and reconstruct common-mode excitation by combining the operating state of the inverter unit, thereby separating the equivalent insulation conductance to assess the insulation damage risk status of the system. The above system combines equipment operating status quantities and sensing data to realize online calculation of insulation parameters without disconnecting from the grid.

[0006] Furthermore, the control unit establishes a hardware synchronization triggering mechanism based on internal pulse width modulation events to synchronously acquire the discrete sequence of DC bus voltage output by the first sensing subunit as an electrical signal, and the discrete sequence of zero-sequence leakage current output by the second sensing subunit as an electrical signal. The control unit is also used to synchronously read the real-time switching state variables of the second inverter subunit, and substitute the discrete sequence of DC bus voltage and the real-time switching state variables into the reconstruction equation to calculate the discrete sequence of common-mode voltage excitation. Through timing alignment of the underlying hardware, the phase offset error in the synchronous acquisition process of multi-source signals is reduced.

[0007] Furthermore, the control unit is used to determine the switching frequency, second harmonic frequency, and third harmonic frequency as characteristic frequencies based on the current pulse width modulation switching frequency; the voltage phasor and current phasor corresponding to the characteristic frequencies are extracted point by point using the sliding discrete Fourier transform algorithm; the complex common-mode admittance at the characteristic frequencies is calculated by performing complex division operations on the current phasor and the corresponding voltage phasor; this step uses multiple determined harmonic frequency bands to extract admittance data, providing a basis for the subsequent calculation of the frequency domain polarization characteristics of the insulating material.

[0008] Furthermore, the control unit is used to extract the imaginary part sequence of the complex common-mode admittance, calculate the relative deviations of the imaginary part data corresponding to the second harmonic frequency and the imaginary part data corresponding to the third harmonic frequency relative to the reference multiple of the imaginary part data corresponding to the switching frequency, and take the maximum value of the two as the frequency multiple relationship deviation; when the frequency multiple relationship deviation is greater than the preset linear deviation tolerance threshold, it is determined that the extracted characteristic frequency point falls into the series resonance region caused by parasitic inductance; since high-frequency parasitic inductance will cause the imaginary part of capacitive admittance to deviate from the linear increasing law in a specific frequency band, by calculating the deviation of the imaginary part multiple relationship, it is identified whether the system is affected by resonance interference.

[0009] In a preferred embodiment of the present invention, when it is determined that the system falls into the series resonance region, the control unit calculates the target value of the smoothed carrier frequency based on a preset frequency offset. If the target value of the smoothed carrier frequency does not reach the upper or lower limit of the safety switch frequency, the control unit modulates the carrier period register value of the internal pulse width modulation generator to shift the underlying pulse width modulation carrier frequency, and re-executes the data acquisition process after a set stable delay period. By actively adjusting the modulation frequency of the system, the feature extraction frequency band avoids the parasitic resonance region, ensuring the effectiveness of leakage current signal acquisition.

[0010] Furthermore, when the frequency multiple relationship deviation is less than or equal to the preset linear deviation tolerance threshold, the control unit calculates the voltage phasor amplitude corresponding to the switching frequency and compares the voltage phasor amplitude with the preset signal-to-noise ratio judgment threshold.

[0011] In a preferred embodiment of the present invention, if the comparison result shows that the voltage phasor amplitude is lower than the preset signal-to-noise ratio judgment threshold, the control unit is used to trigger the adaptive zero-sequence perturbation control logic. Under the constraint of keeping the action time of each non-zero effective vector and the total action time of the zero vector constant, the residence time distribution ratio of the all-zero vector and the all-one vector is adjusted by using an asymmetric factor to reconstruct the zero-sequence excitation time distribution of the inverter unit to excite high-frequency common-mode energy. This control logic, without affecting the inverter fundamental power output and the power quality of the grid side, injects common-mode excitation components into the system by adjusting the asymmetry of the zero vector distribution, thereby solving the problem of high-frequency characteristic signal submersion caused by environmental interference.

[0012] Furthermore, if the comparison result shows that the voltage phasor amplitude is greater than or equal to the preset signal-to-noise ratio threshold, the control unit is used to extract the real part sequence corresponding to the complex common-mode admittance at the characteristic frequency point; the real part of the admittance at each characteristic frequency point is characterized as the algebraic sum of the low-frequency equivalent insulation conductance and the frequency-varying attenuation term; a system of real part linear equations containing the unknown low-frequency equivalent insulation conductance and the unknown polarization frequency-varying factor is established, and the coefficient matrix and observation vector corresponding to the overdetermined matrix model are constructed; the real part sequence of multiple frequency points is jointly calculated by constructing the overdetermined matrix model, and the polarization frequency-varying component under high-frequency excitation conditions is removed.

[0013] Furthermore, the coefficient matrix consists of constants and normalized per-unit values ​​of the corresponding characteristic frequencies relative to the reference switching frequency; the control unit is used to perform least squares inversion operations to calculate the generalized inverse of the coefficient matrix, solve the overdetermined matrix model to obtain a parameter vector containing analytical results, and extract the low-frequency equivalent insulation conductance values ​​from the parameter vector and store them in the internal diagnostic result register.

[0014] Furthermore, the control unit performs an exponentially weighted moving average filtering algorithm on the extracted low-frequency equivalent insulation conductance to generate a smooth insulation conductance reference value; establishes a sliding time window and calculates the first-order difference value of conductance within adjacent sliding time windows; when the smooth insulation conductance reference value exceeds the preset conductance safety threshold, or the number of consecutive set values ​​of the first-order difference value of conductance exceeds the preset degradation rate threshold, it confirms that there is a potential for insulation layer damage in the system and generates an early warning message containing the characteristic parameters of the potential for damage; and establishes judgment conditions by combining the static absolute value of insulation conductance with the dynamic evolution slope, thereby reducing the false alarm rate of single transient over-limit caused by fluctuations in the field environment.

[0015] This invention provides an online monitoring and analysis system for potential hazards in power grid inverter boost converters. It offers the following advantages: 1. This invention extracts the complex common-mode admittance of multi-order characteristic frequency points and establishes an overdetermined matrix model containing the unknown low-frequency equivalent insulation conductance. It then uses the least squares inversion method to separate the polarization loss components of the insulating medium. This scheme does not require the addition of external signal injection hardware to the converter system. It directly uses the high-frequency harmonics generated by the inverter topology as the test excitation, eliminating the random fluctuation error of single-frequency measurement under strong electromagnetic interference environment and improving the analytical accuracy of low-frequency equivalent insulation conductance.

[0016] 2. This invention calculates the deviation of the frequency multiple relationship of the admittance imaginary part sequence. When the frequency point falls into the series resonance region caused by parasitic inductance, it modifies the carrier period register value to shift the underlying pulse width modulation carrier frequency. This mechanism can identify and actively avoid high-frequency common-mode series resonance caused by the distributed parameters of long AC cables online, ensure the algebraic linearity of the insulation admittance model solution process, and prevent the monitoring system from miscalculating insulation parameters and issuing false alarms in the resonant frequency band.

[0017] 3. When the voltage phasor amplitude is detected to be lower than the preset signal-to-noise ratio threshold, this invention adjusts the dwell time distribution ratio of the all-zero vector and the all-one vector using an asymmetric factor while maintaining the fundamental differential mode output unchanged. This control logic enhances the high-frequency common-mode voltage amplitude by unidirectionally biasing the zero vector's action time, solving the problem of weak signal extraction failure caused by excessive attenuation of the excitation signal in specific operating conditions such as high modulation index of the inverter system, and ensuring the continuity of online monitoring of insulation hazards. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the hardware architecture of the online monitoring system for insulation status of the converter system of the present invention; Figure 2 This is a schematic diagram of the overall process of the online monitoring and early warning method for the insulation status of the converter system according to the present invention; Figure 3 This is a schematic diagram of the common-mode excitation reconstruction and complex admittance demodulation control logic of the present invention; Figure 4 This is a schematic diagram of the carrier frequency adaptive translation control logic of the present invention; Figure 5 This is a schematic diagram of the adaptive zero-sequence perturbation control logic of the present invention; Figure 6 This is a schematic diagram of the high-frequency conductivity analysis and polarization separation solution logic of the present invention; Figure 7 This is a schematic diagram of the insulation condition diagnosis and early warning logic of the present invention; Figure 8 This is a schematic diagram illustrating the insulation conductivity smoothing filter and degradation evolution trend of the present invention; Figure 9 This is a schematic diagram illustrating the complex admittance phasor deviation and frequency shift recovery characteristics of the present invention; Figure 10 This is a schematic diagram comparing the common-mode voltage spectrum energy distribution before and after the zero-sequence perturbation operation of the present invention.

[0019] Among them, 10 is the inverter unit; 11 is the first inverter sub-unit; 12 is the second inverter sub-unit; 13 is the third inverter sub-unit; 14 is the output unit; 20 is the control unit; 30 is the sensing unit; 31 is the first sensing sub-unit; and 32 is the second sensing sub-unit. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] See attached document Figure 1 The present invention provides an online monitoring and analysis system for potential risks of power grid inverter boosting. The monitoring and analysis system may include an inverter unit 10, a control unit 20, and a sensing unit 30.

[0022] The inverter unit 10 includes a first inverter subunit 11, a second inverter subunit 12, and a third inverter subunit 13; the DC input side of the second inverter subunit 12 is connected to the first inverter subunit 11; the AC output side of the second inverter subunit 12 is connected to the third inverter subunit 13; the end of the third inverter subunit 13 is connected to the output unit 14, through which electrical energy is transmitted to the medium-voltage power grid.

[0023] The sensing unit 30 includes a first sensing subunit 31 and a second sensing subunit 32. The first sensing subunit 31 is installed at the first inverter subunit 11 and is used to measure the DC bus voltage. The second sensing subunit 32 is installed on the third inverter subunit 13 between the second inverter subunit 12 and the output unit 14, and is connected to a three-phase AC cable to obtain the zero-sequence leakage current signal to ground. The effective measurement frequency band of the second sensing subunit 32 covers three times the switching frequency of the inverter unit 10.

[0024] The control unit 20 is electrically connected to the drive circuits of the first sensing subunit 31, the second sensing subunit 32, and the second inverter subunit 12, respectively. The control unit 20 is used to receive the measurement signal transmitted by the sensing unit 30 and output the pulse width modulation signal to the second inverter subunit 12.

[0025] To establish the connection between the above system hardware to realize the online monitoring function of insulation hazards, the control unit 20 is configured to execute the hazard monitoring and analysis method based on the above hardware structure.

[0026] See attached document Figure 2 This invention provides an online monitoring and analysis method for potential risks in power grid inverter boosting, comprising the following steps: S100 uses the internal pulse width modulation event as a trigger reference to synchronously acquire the discrete sequence of DC bus voltage output by the first sensing subunit 31 and the discrete sequence of zero-sequence leakage current output by the second sensing subunit 32, synchronously acquire the current real-time switching state variable of the second inverter subunit 12, and reconstruct the common-mode voltage excitation discrete sequence of the system based on the discrete sequence of DC bus voltage and the real-time switching state variable. S200, based on the current pulse width modulation switching frequency, performs orthogonal demodulation of the common-mode voltage excitation discrete sequence and the zero-sequence leakage current discrete sequence by sliding discrete Fourier transform, extracts the voltage phasor and current phasor corresponding to the switching frequency, second harmonic frequency and third harmonic frequency, and calculates the complex common-mode admittance at the corresponding characteristic frequency point. S300: Extract the imaginary part sequence of the complex common-mode admittance, calculate the frequency multiple relationship deviation of the imaginary part sequence and compare the deviation with the preset linear deviation tolerance threshold. If the deviation is greater than the linear deviation tolerance threshold, it is determined that the current extracted frequency point is located in the series resonance region caused by parasitic inductance. Output the frequency jump control command to shift the underlying pulse width modulation carrier frequency by the preset frequency offset and return to re-execute S100. S400, if the deviation is less than or equal to the linear deviation tolerance threshold, calculate the voltage phasor amplitude. If the voltage phasor amplitude is lower than the preset signal-to-noise ratio judgment threshold, while keeping the total zero vector action time in the space vector pulse width modulation logic unchanged, adjust the dwell time distribution ratio of the two zero vectors to directionally excite the high-frequency common-mode energy and return to re-execute S100. S500: When the voltage phasor amplitude is greater than or equal to the preset signal-to-noise ratio judgment threshold, extract the real part sequence of complex common-mode admittance, establish a set of real linear equations containing unknown low-frequency equivalent insulation conductance and unknown polarization frequency variation factor to construct an overdetermined matrix model, and perform least squares inversion operation to solve the overdetermined matrix model to separate the low-frequency equivalent insulation conductance. S600 performs a sliding time window exponential weighted moving average filter on the low-frequency equivalent insulation conductance to obtain a smooth insulation conductance reference value, calculates the first-order difference value of conductance in adjacent time windows, and confirms that there is a potential insulation layer damage risk in the system when the smooth insulation conductance reference value exceeds the preset conductance safety threshold or the first-order difference value of conductance continuously exceeds the preset degradation rate threshold. It then generates an early warning message containing the characteristic parameters of the potential risk and sends it through the communication bus.

[0027] See attached document Figure 3 In specific implementations, step S100 provided by the present invention may include the following steps: S101, the control unit 20 establishes a hardware synchronous triggering mechanism for analog-to-digital conversion based on internal pulse width modulation events to perform discrete sampling of DC bus voltage and zero-sequence leakage current.

[0028] During the normal operation of the converter system, the pulse width modulation generator inside the control unit 20 outputs continuous drive pulses according to the space vector pulse width modulation algorithm. The control unit 20 directly uses the hardware interrupt event generated by the pulse width modulation generator at the top or bottom of the counter in each carrier cycle as the phase synchronization reference, and configures the internal high-frequency independent timer to generate a high-frequency trigger event sequence with a trigger frequency that is a fixed integer multiple of the pulse width modulation switching frequency and that the integer multiple is strictly greater than six to satisfy the Nyquist sampling theorem and the integer window constraint of the sliding discrete Fourier transform. The internal analog-to-digital conversion module is configured to capture the trigger event sequence and start synchronous sampling of the analog output signals of the first sensing subunit 31 and the second sensing subunit 32.

[0029] The analog-to-digital conversion processing control unit 20 acquires the discrete sequence of DC bus voltage measured by the first sensing subunit 31 and the discrete sequence of zero-sequence leakage current measured by the second sensing subunit 32. In this embodiment, the sampling and holding time settings and quantization encoding parameters of the analog-to-digital conversion module can be configured with reference to a conventional digital signal processor datasheet, which is a well-known technology in the field.

[0030] S102, the control unit 20 keeps synchronized with the sampling time of step S101 to read the real-time switching state variables of the second subunit 12 of the inverter at the corresponding discrete time point.

[0031] The second sub-unit 12 of the inverter includes a 3-phase bridge arm circuit. The control unit 20 extracts the 3-phase drive signal status corresponding to the current sampling time from the internal register to generate 3-phase real-time switching state variables. The above variables are represented as binary discrete functions defining the on / off states of the upper and lower switching tubes of the 3-phase bridge arm. When the upper tube of the corresponding phase bridge arm is on and the lower tube is off, the switching state variable of that phase takes the value of 1. When the upper tube of the corresponding phase bridge arm is off and the lower tube is on, the switching state variable of that phase takes the value of 0.

[0032] S103, the control unit 20 reconstructs the common-mode voltage excitation discrete sequence of the system based on the discrete sequence of DC bus voltage and the real-time switch state variables.

[0033] The high-frequency switching action of the second subunit 12 of the inverter causes a high-frequency potential jump at the AC output terminal of the system relative to the DC bus neutral point. This potential jump to ground is the common-mode excitation source that drives the zero-sequence leakage current to flow to ground through the cable parasitic capacitance and insulation resistance. The control unit 20 substitutes the acquired discrete sequence of DC bus voltage and the real-time switching state variables of the three-phase bridge arm into the reconstruction equation derived based on the converter topology to calculate the discrete sequence of common-mode voltage excitation. The calculation expression is as follows: ; In the formula, For the first The common-mode voltage excitation amplitude at each discrete sampling time; For the first DC bus voltage amplitude at each discrete sampling time; For the first Real-time switching state variables of phase A bridge arm at discrete sampling times; For the first Real-time switching state variables of phase B bridge arm at discrete sampling times; For the first Real-time switching state variables of the C-phase bridge arm at discrete sampling times; To characterize the positive integer index of the discrete sampling time point, the control unit 20 continuously stores the results obtained from the calculation at each discrete time point in the internal memory to complete the construction of the common-mode voltage excitation discrete sequence.

[0034] In this embodiment, step S200 provided by the present invention may include the following steps in specific implementation: S201, the control unit 20 determines the three characteristic frequency points to be extracted based on the current pulse width modulation switching frequency, namely the switching frequency itself, the second harmonic frequency of the switching frequency, and the third harmonic frequency of the switching frequency. The high-frequency switching action of the space vector pulse width modulation inside the inverter unit will naturally generate a broadband common-mode harmonic cluster centered on the switching frequency and its integer multiples. The frequency components of the first three energy concentrations within the measurement frequency band can be directly used as the characteristic frequency reference points for the subsequent admittance model. This is based on the fact that specific high-frequency points can fully reflect the objective properties of the physical state of the insulating medium under the complex frequency domain of the RC coupling.

[0035] S202, the control unit 20 performs a sliding discrete Fourier transform algorithm on the common-mode voltage excitation discrete sequence and the zero-sequence leakage current discrete sequence to iteratively extract the frequency domain phasor data corresponding to the above three characteristic frequency points. This algorithm achieves sliding recursive update by moving the new sampling point data into the data window and removing the oldest sampling point data. Before the algorithm starts, the control unit 20 sets the historical cache array to zero and assigns an initial value to avoid introducing uninitialized random dirty data, thereby accurately extracting the corresponding voltage phasor and current phasor. The operation expression is as follows: ; ; In the formula, For the first The discrete sampling time corresponds to the first... Historical voltage values ​​at characteristic frequency points; For the first The discrete sampling time corresponds to the first... Historical voltage values ​​at characteristic frequency points; This represents the total number of discrete sampling points contained within a single switching cycle. For the first The common-mode voltage excitation amplitude at each discrete sampling time; For delay Historical values ​​of common-mode voltage excitation amplitude for each sampling period; For the first The discrete sampling time corresponds to the first... Current phasor at the first characteristic frequency point; For the first The discrete sampling time corresponds to the first... Historical values ​​of current phasors at the characteristic frequency point of the first order; For the first The zero-sequence leakage current amplitude at each discrete sampling time; For delay Historical value of zero-sequence leakage current amplitude for each sampling period; is the base of the natural logarithm; The imaginary unit; Pi is a constant. The order of the extracted harmonic frequency; It is a positive integer index representing a discrete sampling time point.

[0036] S203, the control unit 20 monitors the data window filling status of the sliding discrete Fourier transform algorithm in real time. When the total number of filled discrete sampling points reaches the total number of discrete sampling points contained in a single cycle, the control unit 20 uses the extracted current phasor and the voltage phasor at the corresponding frequency point to perform complex division operation to calculate the complex common mode admittance at the corresponding characteristic frequency point. The complex common mode admittance physically characterizes the reciprocal of the system's impedance to ground under a specific high-frequency excitation and comprehensively includes the insulation conductance parameters, distributed parasitic capacitance parameters, and distributed parasitic inductance parameters of the system at the corresponding frequency point. Before performing the division operation, the control unit 20 checks the corresponding voltage phasor amplitude to avoid hardware anomalies caused by a division of zero. When the amplitude is close to zero, it forces the complex common-mode admittance of the current cycle to be zero and propagates it downwards to ensure that subsequent steps can correctly identify weak excitation states and skip invalid high-frequency resonance judgments, avoiding misjudgments triggered by historical dirty data. During the initial startup phase of the algorithm, the control unit 20 initializes the admittance calculation result of the previous cycle to zero to avoid calling undefined memory. The operation expression is as follows: ; In the formula, For the first Complex common-mode admittance at the characteristic frequency point of order; For the first The discrete sampling time corresponds to the first... Current phasor at the first characteristic frequency point; For the first The discrete sampling time corresponds to the first... The control unit 20 performs the above operations in parallel to obtain the complex common-mode admittance corresponding to the switching frequency, the complex common-mode admittance corresponding to the second harmonic frequency, and the complex common-mode admittance corresponding to the third harmonic frequency. In this embodiment, the logic for separating and combining the real and imaginary parts of complex division can be configured with reference to the instruction set of the digital signal processor. Configuring the logic for separating and combining the real and imaginary parts of complex division is a well-known technology in this field.

[0037] See attached document Figure 4 In specific implementations, step S300 provided by the present invention may include the following steps: S301, Control Unit 20 extracts the imaginary part sequence of complex common-mode admittance and calculates the frequency multiple relationship deviation of the imaginary part sequence. The physical circuit formed by the inherent parasitic inductance of long AC cables and the parasitic capacitance of the system to ground tends to have local series resonance under high-frequency excitation. This physical phenomenon will change the impedance frequency characteristics when the system impedance to ground is purely capacitive, thereby destroying the basic law that the imaginary part of admittance increases linearly with frequency. The control unit 20 extracts the imaginary part data corresponding to the complex common-mode admittance at three characteristic frequency points. Before performing the division algebra operation, it verifies the amplitude of the imaginary part data corresponding to the switching frequency to avoid hardware anomalies caused by division by zero. When the amplitude of the imaginary part data meets the non-zero safety condition, a high-frequency resonance distortion judgment model is established based on the imaginary part data corresponding to the switching frequency. The relative deviations of the imaginary part data corresponding to the second harmonic frequency and the third harmonic frequency are calculated respectively, and the maximum value of the two is taken as the current frequency multiple relationship deviation. If the amplitude of the imaginary part data is close to zero, the current frequency multiple relationship deviation is directly determined to be zero to skip the resonance shift logic. The calculation expression is as follows: ; In the formula, This represents the deviation in the frequency multiple relationship. This is a function to find the maximum value. This represents the imaginary part of the complex common-mode admittance corresponding to the switching frequency; This represents the imaginary part of the complex common-mode admittance corresponding to the second harmonic frequency; It represents the imaginary part of the complex common-mode admittance corresponding to the third harmonic frequency.

[0038] S302, the control unit 20 compares the calculated frequency multiple relationship deviation with the preset linear deviation tolerance threshold. The preset linear deviation tolerance threshold reflects the limit of natural fluctuation of parasitic parameters allowed when the system is in the non-resonant flat region. Its specific value is set as a fixed constant between 0.1 and 0.15 based on the historical statistical experience of the length and distribution parameters of AC cables on site. When the frequency multiple relationship deviation is greater than the preset linear deviation tolerance threshold, the control unit 20 determines that the currently extracted characteristic frequency point falls into the local series resonance band caused by parasitic inductance. At this frequency band, the imaginary part of the admittance exhibits nonlinear distortion, and continuing to perform model analysis based on the current frequency point data will cause algebraic errors in the calculation of the low-frequency equivalent insulation conductance.

[0039] S303, when it is determined that the system falls into the series resonance region, the control unit 20 first checks whether the system is in the resonance over-limit alarm lockout state. If it is in the lockout state, it directly suspends the current analysis calculation of the insulation equivalent conductance and jumps back to wait for the next control cycle to re-execute the data synchronization acquisition process of step S100. If the system is not in the lockout state, the control unit 20 calculates the smooth carrier frequency target value based on the preset frequency offset, where the preset frequency offset is set according to the typical physical bandwidth of the resonance band and is 500 Hz or 1000 Hz. Then, the control unit 20 verifies in real time whether the smooth carrier frequency target value exceeds the upper and lower limits of the safety switching frequency pre-calibrated by the power semiconductor device. If the target value of the smooth carrier frequency does not reach the upper and lower limits of the safety switch frequency, the control unit 20 modifies the carrier period register value of the internal pulse width modulation generator to shift the underlying pulse width modulation carrier frequency while keeping the active and reactive output states of the system unchanged, and maintains the currently accumulated frequency shift amount. After the system has gone through the set stable delay period, it jumps back to re-execute the data synchronization acquisition process of step S100, and through cross-cycle closed-loop iteration until the frequency multiple relationship deviation is less than or equal to the preset linear deviation tolerance threshold. If the calculated target value of the smoothed carrier frequency reaches the upper or lower limit of the safety switch frequency, the system is determined to be in a full-band resonance abnormal state. Frequency translation is stopped, a resonance over-limit alarm signal is output, and the alarm state is locked. During the locking period, the current carrier frequency is kept from further translation. At the same time, the analysis calculation of the current insulation equivalent conductance is suspended, and the system jumps back to wait for the next control cycle to re-execute the data synchronization acquisition process of step S100. The lock is released after receiving an external manual reset command or the device is powered on again, so as to avoid bus message congestion caused by frequent touching of the frequency physical limit. In this embodiment, the value range of the set stable delay period is 5 to 10 power frequency cycles.

[0040] In this embodiment, the numerical reload operation of the carrier period register can be configured with reference to the underlying driver logic of the digital signal processor. Configuring the numerical reload operation of the carrier period register is a well-known technology in this field.

[0041] See attached document Figure 5 In specific implementations, step S400 provided by the present invention may include the following steps: S401, the control unit 20 calculates the voltage phasor amplitude corresponding to the switching frequency under the condition that the frequency multiple relationship deviation is less than or equal to the preset linear deviation tolerance threshold and compares the amplitude with the preset signal-to-noise ratio judgment threshold. The high-frequency common-mode voltage amplitude naturally generated by the inverter second sub-unit 12 in a specific space vector pulse width modulation region, such as high modulation index or specific power factor conditions, will be objectively attenuated. This will result in weak extracted current phasor signals and cause the admittance analytical model to be affected by the electromagnetic environment on site, resulting in algebraic deviation. The preset signal-to-noise ratio judgment threshold is preset and stored in the internal register based on the quantization resolution of the system analog-to-digital conversion module and the background noise statistics parameters of the on site. In this embodiment, the preset signal-to-noise ratio (SNR) determination threshold ranges from 0.5 to 2.0 volts. When the voltage phasor amplitude is greater than or equal to the preset SNR determination threshold, the control unit 20 determines that the current excitation signal meets the model solution conditions, clears the weak signal measurement warning lock state, and enters the subsequent admittance real part data extraction and analysis process.

[0042] S402, if the voltage phasor amplitude is lower than the preset signal-to-noise ratio judgment threshold and the system is not in the weak signal measurement warning lock state, the control unit 20 triggers the adaptive zero-sequence perturbation control logic in the underlying pulse width modulation algorithm, and the control unit 20 reads the total zero vector action time and the action time of each non-zero effective vector obtained from the basic operation of the space vector pulse width modulation logic in the current control cycle. The control unit 20 performs subsequent perturbation adjustment operations under the constraint that the duration of each non-zero effective vector and the total duration of the zero vector remain constant within the underlying driver. This control constraint directly ensures, at the algorithm level, that the three-phase fundamental differential mode voltage sequence output from the second inverter subunit 12 to the third inverter subunit 13 maintains its original electrical state and avoids cross-coupling between the active power conversion circuit and the reactive power conversion circuit within the converter system. If the voltage phasor amplitude is lower than the preset signal-to-noise ratio judgment threshold but the system is in a weak signal measurement warning lockout state, the current insulation parameter analysis calculation is directly abandoned and the process jumps back to re-execute step S100.

[0043] S403, the control unit 20 reconstructs the zero-sequence excitation time distribution of the inverter topology by introducing a pre-set asymmetry factor to adjust the dwell time distribution ratio of the two basic zero vectors. In this embodiment, the pre-set asymmetry factor is a real number between 0.6 and 0.85. Under normal symmetrical modulation conditions, the all-zero vectors corresponding to the full conduction of the three-phase lower transistors and the all-one vectors corresponding to the full conduction of the three-phase upper transistors equally divide the total zero vector action time in the time domain. The control unit 20 uses the asymmetry factor to unidirectionally bias the dwell time of a specific zero vector within a single micro-switching cycle. Since the all-zero vectors and all-one vectors correspond to completely opposite DC-side terminal potentials to ground in the inverter bridge arm topology, the duty cycle of the physical common-mode voltage based on the above time bias operation is distorted in the switching waveform of a single switching cycle, thereby deriving a high-frequency common-mode component related to the switching frequency in addition to the fundamental frequency component. The calculation expression is as follows: ; ; In the formula, The dwell time of the all-zero vector within a single control cycle; The asymmetric factor set to control the reconstruction; The total zero vector action time within a single control cycle; The dwell time of the all-one vector within a single control cycle; when the control unit 20 executes a perturbation, if it is currently in the initial symmetric state, it directly sets the target asymmetric factor to the lower limit of the preset value range; otherwise, it calculates the target asymmetric factor by accumulating the steps. If the target asymmetric factor exceeds the upper limit of the preset value range, it determines that the excitation energy is saturated under the current working condition, outputs a weak signal measurement warning flag and locks the warning flag, and abandons the current insulation parameter analysis calculation. After resetting the internal asymmetry factor to the initial symmetric state of evenly sharing the total zero vector action time, while maintaining the non-perturbation symmetry state, wait for the next control cycle to jump back and re-execute the data synchronization acquisition process of step S100 until the system operating conditions change naturally so that the voltage phasor amplitude in the non-perturbation state recovers to above the judgment threshold and the lock is automatically released. Otherwise, after completing the register reload action, wait for the system to go through the set steady-state transition delay, and then jump back to re-execute the data synchronization acquisition process of step S100 to enter the closed-loop iteration of the next cycle until the voltage phasor amplitude reaches the signal-to-noise ratio condition; in this embodiment, the steady-state transition delay is set to 1 to 3 power frequency cycles. In this embodiment, the fixed step size ranges from 0.01 to 0.05, and the microprocessor's duty cycle register update action can be configured with reference to the space vector modulation algorithm framework of a conventional converter control system. Configuring the microprocessor's duty cycle register update action is a well-known technology in this field.

[0044] See attached document Figure 6 In specific implementations, step S500 provided by the present invention may include the following steps: S501, when the control unit 20 determines that the voltage phasor amplitude is greater than or equal to the preset signal-to-noise ratio determination threshold, it extracts the real part sequence corresponding to the complex common-mode admittance at three characteristic frequency points. In the operating environment of the converter system, the dipole inside the insulating material has a dielectric polarization loss physical phenomenon that varies with frequency under the action of the AC electric field. This physical phenomenon causes the real part of the admittance directly extracted from the high frequency to be superimposed with the pure low-frequency equivalent insulation conductance and the frequency-varying conductance component caused by the high-frequency polarization effect. The control unit 20 uses the real part data of the admittance corresponding to the extracted switching frequency, second harmonic frequency and third harmonic frequency as the basic observation sample to enter the conductance analytical separation algebraic operation process.

[0045] S502, the control unit 20 establishes a system of real-part linear equations containing unknowns of low-frequency equivalent insulation conductance and polarization frequency-varying factors. The control unit 20 characterizes the real part of the admittance at each characteristic frequency point as the algebraic sum of the low-frequency equivalent insulation conductance and frequency-varying attenuation terms. Then, it simultaneously constructs an observation equation system by combining the real-part data corresponding to the three extracted frequency points and transforms it into an overdetermined matrix model. The control unit 20 allocates contiguous address space in its internal memory and loads the coefficient matrix and observation vector corresponding to this overdetermined matrix model. The operational expressions are as follows: ; ; ; ; In the formula, The observation column vector consists of the real parts of the admittances at three characteristic frequency points; It is a coefficient matrix consisting of constants and normalized per-unit values ​​of the corresponding characteristic frequencies relative to the reference switching frequency; It is a parameter vector containing the state variables to be determined; This is the normalized per-unit value of the switching frequency; This is the normalized per-unit value of the second harmonic frequency; This is the normalized per-unit value of the third harmonic frequency; This is the real part of the complex common-mode admittance corresponding to the switching frequency; This is the real part of the complex common-mode admittance corresponding to the second harmonic frequency; This is the real part of the complex common-mode admittance corresponding to the third harmonic frequency; The low-frequency equivalent insulation conductivity after the high-frequency polarization component has been stripped is an unknown quantity. The unknown polarization frequency variation factor is used to comprehensively reflect the dielectric loss characteristics.

[0046] S503, the control unit 20 performs least squares inverse operation to solve the overdetermined matrix model and separate the low-frequency equivalent insulation conductance. The system has 3 observed samples, which is greater than the number of unknowns (2), thus placing it in an overdetermined state in algebraic calculation. Due to the normalized per-unit constant in the coefficient matrix, the product of the matrix transpose and the original matrix is ​​a constant full-rank non-singular matrix. The control unit 20 directly uses the least squares estimation algorithm to calculate the generalized inverse of the matrix and solve for the parameter vector values. The above matrix algebraic operation process provides an analytical solution based on the principle of minimizing the sum of squared residuals; the operation expression is as follows: ; In the formula, This is a parameter vector that includes the analytical results of low-frequency equivalent insulation conductance and polarization frequency variation factor; Extracting the coefficient matrix of the real equation system for an overdetermined matrix model; Extract the observation column vectors of the real equation system for the overdetermined matrix model; Coefficient matrix The transpose of the matrix; For the algebraic operation of obtaining the generalized inverse of the matrix, the control unit 20 extracts the first element of the obtained parameter vector, namely the low-frequency equivalent insulation conductance value, and stores it in the internal diagnostic result register for subsequent steps. In this embodiment, the configuration of floating-point arithmetic instructions for matrix transposition, matrix multiplication, and matrix inversion using the underlying mathematical operation library of the digital signal processor is a well-known technique in the art.

[0047] See attached document Figure 7 In specific implementations, step S600 provided by the present invention may include the following steps: S601, the control unit 20 receives the single-cycle low-frequency equivalent insulation conductance separated in step S500 and executes an exponentially weighted moving average filtering algorithm to generate a smoothed insulation conductance reference value. During grid-connected operation of the converter system, the AC lines are subjected to grid harmonics and transient voltage impacts, causing random numerical fluctuations in the discrete conductance sequence calculated in a single iteration. During the algorithm startup phase, the control unit 20 assigns an initial zero value to the historical smoothing buffer and performs weighted smoothing processing on the currently acquired low-frequency equivalent insulation conductance according to a fixed calculation rhythm. This recursive filtering operation extracts the long-term evolution trend data of the insulation impedance at the algebraic level and suppresses high-frequency random disturbance components caused by occasional operating conditions. The calculation expression is as follows: ; In the formula, For the first The smoothed insulation conductivity reference value for each discrete operation cycle; In this embodiment, to set the smoothing coefficient for data smoothing weights, The value range is from 0.01 to 0.1; For the first The calculated value of low-frequency equivalent insulation conductivity obtained by separating each discrete operation cycle; For the first Historical value of smoothed insulation conductivity reference for each discrete operation cycle; It is a positive integer index that represents the time point of discrete operation.

[0048] S602, the control unit 20 establishes a sliding time window based on the continuously acquired smooth insulation conductivity reference value and calculates the first-order difference value of conductivity in adjacent sliding time windows. When the insulation material undergoes physical degradation processes such as thermal aging and water tree penetration, its equivalent conductivity exhibits physical characteristics of absolute value increase and nonlinear evolution of the rate of change with time. The control unit 20 opens a first-in-first-out queue of fixed depth in the internal memory and stores the smooth insulation conductivity reference value as a sliding time window in the form of data pairs with time stamp information and monitors the data filling status of the queue in real time. When the total amount of data recorded in the queue reaches the preset window depth, the control unit 20 extracts the smoothed result of the current calculation cycle, subtracts the historical smoothed result with a fixed delay window length, and divides it by the actual physical time difference between the data nodes at both ends of the queue to obtain the difference value. During the startup phase when the queue is not full, the control unit 20 sets the first-order difference value of conductance to zero and temporarily suspends the triggering of subsequent dynamic over-limit judgment logic. The calculation expression is as follows: ; In the formula, For the first The first-order difference value of conductance under the time window node corresponding to each discrete operation cycle; For the first The smoothed insulation conductivity reference value for each discrete operation cycle; For delay Historical value of smoothed insulation conductivity reference for each operation cycle; In this embodiment, the fixed depth parameter of the sliding time window is used. The value of is a positive integer between 100 and 500; The window depth is calculated as the difference in actual physical time between the current computation cycle node at the head of the queue and the historical computation cycle node at the tail of the queue. The average allocation value; It is a positive integer index that represents the time point of discrete operation.

[0049] S603, the control unit 20 sets a preset conductivity safety threshold based on the background conductivity parameters and end-of-life attenuation tolerance of the insulated cable at the time of manufacture, and sets a preset degradation rate threshold based on the tangent slope of the fitted curve of the accelerated aging test of the insulation. In this embodiment, the preset conductivity safety threshold ranges from 10 to 50 micro Siemens, and the preset degradation rate threshold ranges from 0.1 to 0.5 micro Siemens per hour. The control unit 20 compares the current smooth insulation conductivity reference value with the preset conductivity safety threshold and sets a Level 1 static over-limit flag when the value is greater than the threshold. Once the Level 1 static over-limit flag is set, it will remain locked until an external manual reset command is received to prevent the insulation impedance from repeatedly changing near the threshold critical point due to physical fluctuations in ambient temperature and humidity. The control unit 20 synchronously converts the calculated first-order conductance difference value to a standard hourly rate using a unified time dimension and compares it with a preset degradation rate threshold. If the converted difference value is greater than the preset degradation rate threshold, an abnormal accumulation count is performed. When the abnormal accumulation count reaches a set number of times, a level 2 dynamic degradation flag is set. If the first-order conductance difference value falls back to or below the preset degradation rate threshold, the abnormal accumulation count is cleared to prevent false alarms caused by single transient fluctuations due to random high-frequency noise on site. At the same time, the triggered level 2 dynamic degradation flag is kept in a locked state until an external manual reset command is received to avoid repeated jumps in the warning state near the threshold critical point. In this embodiment, the value range of the set number of times is a positive integer between 3 and 10 times. When either of the two flag bits changes from a cleared state to a set state, the system is determined to meet the insulation degradation early warning conditions. The current time stamp information, absolute value of conductance, and differential value are extracted to generate a standard early warning message and report it to the external monitoring equipment. This avoids repeated message transmission during continuous over-limit periods, which could lead to communication bus congestion. In this embodiment, the communication transceiver and message framing and transmission logic at the controller's underlying layer can be configured with reference to the industrial fieldbus protocol specification. Configuring the communication transceiver and message framing and transmission logic at the controller's underlying layer is a well-known technology in this field. After completing the above-mentioned early warning judgment and message communication logic, regardless of whether the current cycle system is in a healthy state and has not triggered an early warning message, or is in an abnormal state and has sent an early warning message, the control unit 20 will reset the internal asymmetric factor to the initial symmetric state of evenly dividing the total zero vector action time to remove the perturbation. Then, the current operation cycle will end and jump back to re-execute step S100, thereby constructing a complete closed-loop control state machine to realize continuous and uninterrupted online monitoring of insulation hazards.

[0050] Application Examples: See attached document Figure 8 - Appendix Figure 10To better understand the technical solution of this invention, the following explanation uses a high-power string inverter system in a large photovoltaic power plant as an example: The inverter system is in grid-connected operation. The initial pulse width modulation switching frequency is set to 3000 Hz. The control unit synchronously acquires the 1500V DC bus voltage and leakage current signal, and extracts the complex common-mode admittance at 3000 Hz, 6000 Hz and 9000 Hz.

[0051] When the parasitic parameters of the cable to ground drift, the system calculates that the relationship between the imaginary part of the second harmonic admittance and the imaginary part of the first harmonic admittance exhibits nonlinear characteristics. The calculated frequency multiple relationship deviation is 0.25, which is greater than the preset linear deviation tolerance threshold of 0.15. The control unit determines that the system is in the series resonance region caused by parasitic inductance. The control unit shifts the switching frequency to 3500 Hz, and the recalculated deviation becomes 0.05. The imaginary admittance data shows a linear distribution.

[0052] When the photovoltaic system is operating at full load and the inverter enters the high modulation index operating range, the amplitude of the high-frequency common-mode excitation voltage decreases. The calculated voltage phasor amplitude at the switching frequency is 0.3 volts, which is lower than the preset signal-to-noise ratio threshold of 1.0 volts. The control unit triggers the perturbation logic, adjusts the asymmetry factor from 0.5 to 0.75, changes the dwell time of the all-zero vector, and the high-frequency common-mode voltage amplitude becomes 1.8 volts. This amplitude is greater than the preset signal-to-noise ratio threshold, and the total zero vector dwell time remains unchanged, so the grid-connected power of the system remains the same.

[0053] When the outer sheath of an AC cable is damaged and the insulation ages, the control unit uses an overdetermined matrix model to strip the polarization component and obtain the low-frequency equivalent insulation conductance. After filtering by a sliding time window, the smoothed insulation conductance reference value changes from 20 microsiemens to 45 microsiemens. The calculated first-order difference value of the conductance is 0.8 microsiemens per hour. If this difference value is greater than the preset degradation rate threshold of 0.5 microsiemens per hour for five consecutive times, the control unit determines that the system has experienced accelerated degradation, sets it to a level 2 dynamic deterioration flag, and sends an early warning message containing the current insulation conductance and degradation rate parameters to the external monitoring equipment. The system stops and performs the damaged cable replacement operation based on the received message.

[0054] A hardware-in-the-loop simulation test platform was built, which includes an equivalent network of resistors, capacitors, and inductors and inverter hardware. A power frequency leakage current monitoring method was introduced as a control group for comparative test experiments.

[0055] Standard resistors with different resistance values ​​were connected to the test platform to simulate the insulation working state. The parasitic capacitance parameter was set to 2 microfarads. The absolute error values ​​of the conductance measurement of the two methods were recorded and compared.

[0056] Table 1: Comparison of Insulation Equivalent Conductivity Measurement Results under Different Deterioration Conditions In a working environment with parasitic capacitance, the measurement results of the control group method have a positive deviation and trigger an over-limit alarm signal in a healthy state. The method provided by this invention extracts the low-frequency equivalent insulation conductance by solving the multi-frequency overdetermined equation system, and the relative error values ​​corresponding to each test state are all less than or equal to 2%.

[0057] The test period was set to 72 hours. During the experiment, the cable resonant frequency band offset condition and the inverter full-load high modulation operation condition were superimposed. The monitoring function operation parameters of the control group method and the method provided by the present invention were recorded.

[0058] Table 2: System Monitoring Performance Evaluation Table under Complex Operating Conditions The method provided by this invention includes a high-frequency resonance avoidance control step, an adaptive zero-sequence perturbation control step, and an insulation degradation differential evaluation step. The number of false alarms and the data loss time are both zero under the corresponding test conditions, and a trend warning signal is output before insulation breakdown.

[0059] In a two-dimensional coordinate system with time on the horizontal axis and conductivity on the vertical axis, the original conductivity calculation values ​​without filtering are distributed as a scattering pattern. The smooth insulation conductivity reference value obtained after sliding time window exponential weighted moving average filtering is a monotonically increasing curve, which corresponds to the evolution process of insulation material parameters.

[0060] In a complex plane coordinate system with the real part as the horizontal axis and the imaginary part as the vertical axis, the data points of the first, second, and third harmonic frequencies when the system is in a non-resonant state are arranged on a straight line passing through the origin. When the system is in a series resonant state, the data points of the high-frequency points are shifted in the negative direction of the imaginary axis and deviate from the straight line. After the control unit performs a carrier frequency translation operation, the shifted data points are rearranged on the straight line.

[0061] In a two-dimensional bar chart with frequency on the horizontal axis and voltage amplitude on the vertical axis, before the perturbation operation is performed, the amplitude of the characteristic spectral line at the switching frequency and the amplitude of the ambient noise are in the same range. After the control unit triggers the adaptive zero-sequence perturbation control logic, the amplitude of the spectral line at the switching frequency increases to six times the value before the perturbation, while the amplitude of the spectrum in the power frequency fundamental and low harmonic bands remains unchanged.

[0062] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An online monitoring and analysis system for potential hazards in power grid inverter boost converters, characterized in that, include: The inverter unit includes a first inverter subunit, a second inverter subunit, and a third inverter subunit connected in sequence, wherein the third inverter subunit is used to connect to an external output unit. The sensing unit includes a first sensing subunit installed on the first inverter subunit and a second sensing subunit installed on the third inverter subunit. The control unit is connected to the first sensing subunit, the second sensing subunit, and the second inverter subunit, respectively. The control unit is used to extract complex common-mode admittance based on the electrical signals collected by the sensing unit, and reconstruct common-mode excitation in combination with the operating state of the inverter unit, thereby separating the equivalent insulation conductance to assess the potential insulation damage status of the system.

2. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 1, characterized in that, The control unit establishes a hardware synchronization triggering mechanism based on internal pulse width modulation events to synchronously acquire the discrete sequence of DC bus voltage output by the first sensing subunit as the electrical signal output, and the discrete sequence of zero-sequence leakage current output by the second sensing subunit as the electrical signal output. The control unit is also used to synchronously read the real-time switching state variables of the second inverter subunit, and substitute the discrete sequence of DC bus voltage and the real-time switching state variables into the reconstruction equation to calculate the discrete sequence of common-mode voltage excitation.

3. The online monitoring and analysis system for potential hazards in power grid inverter boosting as described in claim 2, characterized in that, The control unit is used to determine the switching frequency, second harmonic frequency, and third harmonic frequency as characteristic frequency points based on the current pulse width modulation switching frequency. The voltage phasor and current phasor corresponding to the characteristic frequency points are extracted point by point using the sliding discrete Fourier transform algorithm. The complex common-mode admittance at the characteristic frequency point is calculated by performing complex division operations between the current phasor and the corresponding voltage phasor.

4. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 3, characterized in that, The control unit is used to extract the imaginary part sequence of the complex common-mode admittance, calculate the relative deviations of the imaginary part data corresponding to the second harmonic frequency and the imaginary part data corresponding to the third harmonic frequency relative to the reference multiple of the imaginary part data corresponding to the switching frequency, and take the maximum value of the two as the frequency multiple relationship deviation; when the frequency multiple relationship deviation is greater than the preset linear deviation tolerance threshold, it is determined that the extracted characteristic frequency point falls into the series resonance region caused by parasitic inductance.

5. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 4, characterized in that, When it is determined that the series resonant region is entered, the control unit calculates the target value of the smoothed carrier frequency based on the preset frequency offset. If the target value of the smoothed carrier frequency does not reach the upper and lower limits of the safety switch frequency, the control unit modifies the carrier period register value of the internal pulse width modulation generator to shift the underlying pulse width modulation carrier frequency, and re-executes the data acquisition process after a set stable delay period.

6. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 4, characterized in that, When the frequency multiple relationship deviation is less than or equal to the preset linear deviation tolerance threshold, the control unit calculates the voltage phasor amplitude corresponding to the switching frequency and compares the voltage phasor amplitude with the preset signal-to-noise ratio judgment threshold.

7. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 6, characterized in that, If the comparison result shows that the voltage phasor amplitude is lower than the preset signal-to-noise ratio threshold, the control unit is used to trigger the adaptive zero-sequence perturbation control logic; under the constraint of keeping the action time of each non-zero effective vector and the total action time of the zero vector constant, the residence time distribution ratio of the all-zero vector and the all-one vector is adjusted by using an asymmetric factor to reconstruct the zero-sequence excitation time distribution of the inverter unit to excite high-frequency common-mode energy.

8. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 6, characterized in that, If the comparison result indicates that the voltage phasor amplitude is greater than or equal to the preset signal-to-noise ratio threshold, the control unit is used to extract the real part sequence corresponding to the complex common-mode admittance at the characteristic frequency point; the real part of the admittance at each characteristic frequency point is characterized as the algebraic sum of the low-frequency equivalent insulation conductance and the frequency-varying attenuation term; a system of linear equations containing the real part of the low-frequency equivalent insulation conductance unknown and the polarization frequency-varying factor unknown is established, and the coefficient matrix and observation vector corresponding to the overdetermined matrix model are constructed.

9. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 8, characterized in that, The coefficient matrix consists of constants and normalized per-unit values ​​of the corresponding characteristic frequencies relative to the reference switching frequency; the control unit is used to perform least squares inversion operation to calculate the generalized inverse of the coefficient matrix, solve the overdetermined matrix model to obtain a parameter vector containing analytical results, and extract the low-frequency equivalent insulation conductance value from the parameter vector and store it in the internal diagnostic result register.

10. The online monitoring and analysis system for potential hazards in power grid inverter boost converters according to claim 9, characterized in that, The control unit is used to perform an exponentially weighted moving average filtering algorithm on the extracted low-frequency equivalent insulation conductance to generate a smooth insulation conductance reference value; establish a sliding time window and calculate the first-order difference value of conductance within adjacent sliding time windows; when the smooth insulation conductance reference value exceeds a preset conductance safety threshold, or when the number of consecutive set values ​​of the first-order difference value of conductance exceeds a preset degradation rate threshold, the system is confirmed to have a potential insulation layer damage risk and an early warning message containing risk feature parameters is generated.

Images