Bus voltage multi-dimensional real-time monitoring analysis method, system, equipment and medium

Abstract

The invention discloses a bus voltage multi-dimensional real-time monitoring and analyzing method, system, equipment and medium, and belongs to the technical field of power system monitoring and fault prediction.The method comprises the steps that high-frequency synchronous sampling is conducted on bus voltage signals, discrete time sequence voltage signals are obtained and preprocessed, preprocessed voltage signals are obtained, and the discrete time sequence voltage signals are obtained; the method comprises the following steps: synchronously extracting multi-dimensional characteristic parameters of voltage, obtaining a characteristic parameter set, calculating a health degree index comprehensively representing a voltage state by utilizing a dynamic weighted fusion algorithm based on the characteristic parameter set, calculating an abnormal degree of a current value according to a historical data sequence of the health degree index, and performing fault grade judgment according to the abnormal degree. And performing trend extrapolation based on the health degree index and the historical data sequence of the key parameters in the characteristic parameter set, and generating predictive maintenance suggestions. According to the method, the crossing from passive alarm to predictive maintenance is realized, and the intelligent level and the operation and maintenance reliability of state monitoring of the power system are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring and fault prediction technology, specifically to a method, system, equipment, and medium for multidimensional real-time monitoring and analysis of bus voltage. Background Technology

[0002] Traditional monitoring methods mainly rely on threshold judgment of voltage amplitude and harmonic analysis techniques based on Fast Fourier Transform (FFT). In recent years, with the advancement of digital signal processing technology, phase-locked loop (PLL) technology has been widely used for tracking the fundamental phase and frequency, while adaptive filtering algorithms have also played a role in noise suppression. Existing technologies typically employ a serial processing architecture: filtering and denoising are performed first, followed by FFT analysis or PLL calculation, and finally, the output results are compared with a fixed threshold. Although some advanced systems have introduced multi-parameter monitoring, such as simultaneously acquiring the effective voltage value and total harmonic distortion (THD), each parameter is often judged independently, lacking a deep fusion mechanism. Furthermore, to improve analysis accuracy, improved algorithms such as windowed interpolation FFT and enhanced PLL based on second-order generalized integrators have been adopted to reduce spectral leakage and asynchronous sampling errors.

[0003] However, existing technologies still have significant limitations. First, in the signal preprocessing stage, conventional LMS filters are sensitive to impulse interference and are prone to weight mismatch due to sudden noise, thus affecting the accuracy of subsequent analysis. Although there are methods such as amplitude limiting filtering, they lack the ability to accurately suppress the fusion interference of impulse and Gaussian noise while preserving signal details. Second, the parallelization of feature extraction is insufficient, and the serial execution of FFT and PLL operations is difficult to meet high real-time requirements, especially when it is necessary to synchronously acquire multi-dimensional parameters such as amplitude, phase, frequency, and harmonics, where the delay problem is prominent. More importantly, existing methods lack an effective fusion mechanism for multi-dimensional parameters: most schemes only set a fixed threshold for a single parameter, ignoring the coupling relationship between parameters and their dynamic characteristics over time, resulting in insufficient sensitivity to complex faults and high false alarm and false negative rates. Finally, traditional threshold judgment mechanisms lack adaptive capabilities, cannot identify slowly occurring performance degradation trends, and fail to correlate real-time monitoring with component-level health status, thus failing to achieve true predictive maintenance. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by this invention are: poor signal preprocessing effect in complex noise environments, insufficient real-time performance and single dimension of multi-dimensional feature extraction, and independent judgment of multiple parameters, lack of intelligent fusion and early warning capabilities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for multi-dimensional real-time monitoring and analysis of bus voltage, comprising the following steps, The bus voltage signal is sampled at high frequency to obtain a discrete-time series voltage signal, which is then preprocessed to obtain a preprocessed voltage signal. Multidimensional feature parameters of the voltage are extracted synchronously to obtain a feature parameter set. Based on the feature parameter set, a health index that comprehensively represents the voltage state is calculated using a dynamic weighted fusion algorithm. According to the historical data sequence of the health index, a dynamic threshold algorithm is used to calculate the degree of abnormality of the current value, and the fault level is judged according to the degree of abnormality. Based on the historical data sequence of the health index and key parameters in the feature parameter set, trend extrapolation is performed to generate predictive maintenance suggestions.

[0007] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the preprocessing includes filtering the discrete time series voltage signal using a nonlinear adaptive filter. Filtering using a nonlinear adaptive filter includes initial filtering and error calculation, nonlinear impulse detection, nonlinear processing, and weight update and output.

[0008] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the initial filtering and error calculation include: initially using a transverse finite impulse response filter based on the LMS algorithm to perform preliminary filtering on the discrete time series voltage signal to obtain the preliminarily filtered signal; The formula for calculating the preliminary error signal is expressed as follows: in, For error signals, For the expected response, The signal is after preliminary filtering, and n is the time index; The nonlinear pulse detection includes calculating the instantaneous amplitude of the error signal and comparing it with a dynamic threshold. The dynamic threshold is dynamically calculated from the average value and standard deviation of the error signal amplitude. The average value and standard deviation are recursively estimated through a first-order low-pass filter. When the error signal is greater than the dynamic threshold, it is determined that there is strong pulse interference at the current sampling point, and the trigger flag is set to 1. The formula for calculating the dynamic threshold is as follows: in, For dynamic thresholds, The mean of the recursive estimate of the absolute value of the error signal is given by K, where K is the threshold coefficient. The standard deviation is a recursive estimate of the absolute value of the error signal. The nonlinear processing includes pausing the weight update of the LMS algorithm when the trigger flag is 1, and starting the spline interpolation replacement mechanism. Using M uncontaminated error data points before and after the contaminated data point error signal, a cubic spline interpolation function is constructed, and the interpolation estimate at the current time is calculated to replace the original error signal value contaminated by the pulse. The weight update and output include, for moments without pulse interference (i.e., when the flag is 0), the LMS filter updates the filter weight vector using a standard algorithm, expressed as: in, Let n be the weight vector of the filter at time n. Let n be the weight vector of the filter at time n+1. Step size factor The input vector.

[0009] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the set of characteristic parameters includes: effective voltage value, total harmonic distortion rate, amplitude and phase of each harmonic, instantaneous phase of the fundamental wave, real-time frequency and frequency deviation. The process of obtaining the feature parameter set includes using a spectrum analysis unit and a fundamental wave tracking unit to process the preprocessed discrete-time series voltage signal stream in parallel, and simultaneously extracting multi-dimensional feature parameters of the voltage to obtain the feature parameter set. The outputs of the spectrum analysis unit and the fundamental wave tracking unit together constitute the feature parameter set. The spectrum analysis unit includes applying a window function to the preprocessed discrete-time series voltage signal data with a data window length of N, performing FFT operation to obtain the spectrum, and using an interpolation algorithm to correct the frequency, amplitude, and phase of the fundamental and harmonic spectral lines in the spectrum, and outputting the harmonic amplitude sequence, harmonic phase sequence, and THD value. The fundamental wave tracking unit includes receiving a preprocessed discrete-time series voltage signal using an SOGI quadrature signal generator, generating two signals orthogonal to the input fundamental wave, calculating the instantaneous phase error based on the two signals using a phase detector, filtering out high-frequency components in the phase error using a loop filter, and outputting a frequency correction amount. The voltage-controlled oscillator integrates the frequency correction amount output by the loop filter to generate the instantaneous phase and real-time frequency of the fundamental wave.

[0010] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the FFT operation includes windowing using the Blackman-Harris window function and correcting the spectral lines of the windowed FFT using a dual-spectral-line interpolation algorithm. The Blackman-Harris window function is expressed as: in, Here, n is the window function value, n is the time index, and N is the window length. , , and For window function coefficients; Correcting the spectrum of a windowed FFT involves calculating the frequency deviation and the corrected amplitude for the highest spectral line and adjacent spectral lines within a specified range of the target harmonic frequency. The formula is as follows: in, For frequency deviation, Let x be the complex spectral amplitude of the k-th spectral line. and The complex spectral amplitudes of the left and right adjacent spectral lines of the k-th spectral line are given. The constant related to the main lobe width of the window function. As a regulating factor, The corrected true amplitude, Let be the real part of the window function's spectrum. This is the imaginary component of the window function's spectrum function; The dynamic weighted fusion algorithm includes sub-health mapping, calculating dynamic weights, and performing health fusion. The sub-health mapping involves normalizing each parameter in the feature parameter set to the [0,1] interval using an improved sigmoid mapping function to obtain the sub-health score, expressed by the formula: in, This is the quantified value of the health status of the i-th feature parameter. The current feature parameter value, Let be the average value of the i-th feature parameter within the historical data window. Let be the standard deviation of the i-th feature parameter within the historical data window. This is the tolerance coefficient for the fluctuation range. This is the S-curve steepness adjustment coefficient. It is an exponential function; The formula for calculating dynamic weights is expressed as follows: in, The dynamic weights are the weights assigned to the sub-health score of the i-th feature parameter. For information entropy, Let m be the probability that the historical data value of the i-th parameter falls within the m-th discrete interval, where m is the interval index. The formula for health score fusion is expressed as: in, This is a comprehensive voltage health index.

[0011] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the dynamic threshold algorithm includes: continuously storing the latest L health index values ​​to form a first-in-first-out sliding window; calculating the mean and standard deviation of the data within the sliding window in real time; calculating the degree of abnormality of the current value; and setting an abnormality threshold for comparison to determine the fault level. The formula for calculating the degree of anomaly is expressed as follows: Where Z represents the degree of abnormality of the current value. This represents the mean of the data within the sliding window. The standard deviation of the data within the sliding window. To avoid zero denominators for extremely small positive numbers; set a first threshold. Second threshold ,when When a serious system malfunction is detected, the highest-level alarm is immediately triggered, and preset protective actions are executed; when If the system is found to be in a state of degradation or early failure, an early warning signal is triggered to alert maintenance personnel, but no immediate hard-line tripping measures are taken.

[0012] As a preferred embodiment of the multidimensional real-time monitoring and analysis method for bus voltage described in this invention, the trend extrapolation includes fitting and predicting the historical data sequence of health index and total harmonic distortion rate using the triple exponential smoothing method. The exponential smoothing method includes initializing the model, and for each new time point, after receiving new health index and harmonic distortion rate data, recursively updating the model components, and making predictions based on the updated model components. The formula for recursively updating model components is expressed as follows: in, Let be the horizontal component at time t, i.e., the base value of the time series at time t. For horizontal smoothing parameters, The comprehensive health status observation value at time t. Let be the seasonal component at time tr, and r be the length of a period. The horizontal component at time t-1 The trend component at time t-1, Let be the trend component at time t. For trend smoothing parameters, Let be the seasonal component at time t. For seasonal smoothing parameters; The prediction formula is: in, Let T be the predicted value at time t+T, where T is the prediction step size. For the seasonal component at the future time t+T; When the health index predicted by the trend extrapolation results continues to decline and falls below the warning threshold in the next T hours, and the total harmonic distortion rate is predicted to continue to rise, predictive maintenance recommendations are generated.

[0013] This invention provides a multi-dimensional real-time monitoring and analysis system for bus voltage.

[0014] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-dimensional real-time monitoring and analysis system for bus voltage, comprising: a signal preprocessing and anti-interference module, a multi-dimensional feature parallel extraction module, a dynamic weighted fusion health assessment module, and a trend prediction and maintenance decision module.

[0015] The signal preprocessing and anti-interference module is used to suppress Gaussian white noise and periodic pulse interference in the signal by using a composite filter, identify pulse events in real time through dynamic thresholds, and trigger an interpolation substitution mechanism to eliminate the influence, thus providing a clean voltage signal source for subsequent analysis.

[0016] The multidimensional feature parallel extraction module is used to accurately calculate the effective voltage value, total harmonic distortion rate, and amplitude and phase of each harmonic by using a windowed interpolation FFT algorithm in a parallel processing architecture. At the same time, it tracks the instantaneous phase, frequency, and frequency deviation of the fundamental voltage in real time by using an enhanced phase-locked loop algorithm.

[0017] The dynamic weighted fusion health assessment module is used to introduce information entropy theory, adaptively map each feature parameter to a sub-health value according to its own historical statistical characteristics, calculate the entropy value of each parameter data sequence to measure uncertainty, and perform dynamic weight allocation to obtain the voltage health index through fusion.

[0018] The trend prediction and maintenance decision module is used to learn the historical trends of health indicators and key parameters by applying time series prediction algorithms to predict future trends.

[0019] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the aforementioned method for multidimensional real-time monitoring and analysis of bus voltage.

[0020] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the aforementioned method for multidimensional real-time monitoring and analysis of bus voltage.

[0021] The beneficial effects of this invention are as follows: By employing a preprocessing step combining high-frequency synchronous sampling and nonlinear adaptive filtering, it achieves precise suppression of strong pulse interference in the bus voltage signal and a significant improvement in signal fidelity, laying a high-quality data foundation for subsequent multidimensional feature extraction. Ultimately, it achieves the beneficial effect of accurately capturing minute voltage anomalies even in complex electromagnetic environments. By utilizing a parallel processing architecture of a spectrum analysis unit and a fundamental wave tracking unit to synchronously extract multidimensional feature parameters such as the effective voltage value, harmonic distortion rate, fundamental wave phase, and frequency deviation, it achieves a comprehensive characterization of the time-frequency domain characteristics of the voltage signal, breaking through the limitations of traditional single-dimensional feature analysis and improving the sensitivity for identifying early faults and potential degradation trends. Furthermore, by introducing a dynamic weighted fusion algorithm and implementing a feature extraction method based on information entropy theory, it achieves a comprehensive characterization of the voltage signal's time-frequency domain characteristics. The adaptive weight allocation of characteristic parameters enables intelligent fusion and comprehensive quantification of multi-dimensional characteristic parameters, constructing a health index that dynamically reflects voltage state changes. This avoids evaluation bias caused by fixed weights and improves the accuracy of state characterization. By adopting an improved Z-score dynamic threshold algorithm combined with a sliding window to calculate the degree of anomaly of the health index in real time, adaptive hierarchical judgment of voltage state anomalies is achieved. This distinguishes between severe faults and early degradation trends, reduces false alarms and missed alarms, and optimizes operation and maintenance decision response. By using the triple exponential smoothing method to extrapolate the trends of the health index and total harmonic distortion rate and correlate them with the degradation physical model of the bus filter capacitor, predictive analysis of the future evolution trend of the voltage state is achieved, enabling the generation of maintenance suggestions in advance. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 The above is a flowchart of a multi-dimensional real-time monitoring and analysis method for bus voltage provided in one embodiment of the present invention.

[0024] Figure 2This is a system block diagram of a multi-dimensional real-time monitoring and analysis system for bus voltage provided in one embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for multi-dimensional real-time monitoring and analysis of bus voltage, including: S1: Perform high-frequency synchronous sampling of the bus voltage signal to obtain a discrete time series voltage signal, and perform preprocessing to obtain a preprocessed voltage signal. Simultaneously extract multi-dimensional feature parameters of the voltage to obtain a set of feature parameters.

[0027] Furthermore, the preprocessing includes filtering the discrete-time series voltage signal using a nonlinear adaptive filter; Filtering using a nonlinear adaptive filter includes initial filtering and error calculation, nonlinear impulse detection, nonlinear processing, and weight update and output.

[0028] Furthermore, the initial filtering and error calculation includes using a transverse finite impulse response filter based on the LMS algorithm to perform preliminary filtering on the discrete time series voltage signal to obtain the preliminarily filtered signal. The formula for calculating the preliminary error signal is expressed as follows: in, For error signals, For the expected response, The signal is after preliminary filtering, and n is the time index; The nonlinear pulse detection includes calculating the instantaneous amplitude of the error signal and comparing it with a dynamic threshold. The dynamic threshold is dynamically calculated from the average value and standard deviation of the error signal amplitude. The average value and standard deviation are recursively estimated through a first-order low-pass filter. When the error signal is greater than the dynamic threshold, it is determined that there is strong pulse interference at the current sampling point, and the trigger flag is set to 1. The formula for calculating the dynamic threshold is as follows: in, For dynamic thresholds, The mean of the recursive estimate of the absolute value of the error signal is given by K, where K is the threshold coefficient. The standard deviation is a recursive estimate of the absolute value of the error signal. The nonlinear processing includes pausing the weight update of the LMS algorithm when the trigger flag is 1, and starting the spline interpolation replacement mechanism. Using M uncontaminated error data points before and after the contaminated data point error signal, a cubic spline interpolation function is constructed, and the interpolation estimate at the current time is calculated to replace the original error signal value contaminated by the pulse. The weight update and output include, for moments without pulse interference (i.e., when the flag is 0), the LMS filter updates the filter weight vector using a standard algorithm, expressed as: in, Let n be the weight vector of the filter at time n. Let n be the weight vector of the filter at time n+1. Step size factor The input vector.

[0029] It should be noted that the set of characteristic parameters includes the effective value of voltage, total harmonic distortion rate, amplitude and phase of each harmonic, instantaneous phase of the fundamental wave, real-time frequency and frequency deviation; The process of obtaining the feature parameter set includes using a spectrum analysis unit and a fundamental wave tracking unit to process the preprocessed discrete-time series voltage signal stream in parallel, and simultaneously extracting multi-dimensional feature parameters of the voltage to obtain the feature parameter set. The outputs of the spectrum analysis unit and the fundamental wave tracking unit together constitute the feature parameter set. The spectrum analysis unit includes applying a window function to the preprocessed discrete-time series voltage signal data with a data window length of N, performing FFT operation to obtain the spectrum, and using an interpolation algorithm to correct the frequency, amplitude, and phase of the fundamental and harmonic spectral lines in the spectrum, and outputting the harmonic amplitude sequence, harmonic phase sequence, and THD value. The fundamental wave tracking unit includes receiving a preprocessed discrete-time series voltage signal using an SOGI quadrature signal generator, generating two signals orthogonal to the input fundamental wave, calculating the instantaneous phase error based on the two signals using a phase detector, filtering out high-frequency components in the phase error using a loop filter, and outputting a frequency correction amount. The voltage-controlled oscillator integrates the frequency correction amount output by the loop filter to generate the instantaneous phase and real-time frequency of the fundamental wave.

[0030] S2: Based on the feature parameter set, a dynamic weighted fusion algorithm is used to calculate the health index that comprehensively represents the voltage state. Based on the historical data sequence of the health index, a dynamic threshold algorithm is used to calculate the degree of abnormality of the current value, and the fault level is judged based on the degree of abnormality.

[0031] Furthermore, the FFT operation includes performing a windowing operation using the Blackman-Harris window function and correcting the spectral lines of the windowed FFT using a bispectral interpolation algorithm; The Blackman-Harris window function is expressed as: in, Here, n is the window function value, n is the time index, and N is the window length. , , and For window function coefficients; Correcting the spectrum of a windowed FFT involves calculating the frequency deviation and the corrected amplitude for the highest spectral line and adjacent spectral lines within a specified range of the target harmonic frequency. The formula is as follows: in, For frequency deviation, Let x be the complex spectral amplitude of the k-th spectral line. and The complex spectral amplitudes of the left and right adjacent spectral lines of the k-th spectral line are given. The constant related to the main lobe width of the window function. As a regulating factor, The corrected true amplitude, Let be the real part of the window function's spectrum. This is the imaginary component of the window function's spectrum function; The dynamic weighted fusion algorithm includes sub-health mapping, calculating dynamic weights, and performing health fusion. The sub-health mapping involves normalizing each parameter in the feature parameter set to the [0,1] interval using an improved sigmoid mapping function to obtain the sub-health score, expressed by the formula: in, This is the quantified value of the health status of the i-th feature parameter. The current feature parameter value, Let be the average value of the i-th feature parameter within the historical data window. Let be the standard deviation of the i-th feature parameter within the historical data window. This is the tolerance coefficient for the fluctuation range. This is the S-curve steepness adjustment coefficient. It is an exponential function; The formula for calculating dynamic weights is expressed as follows: in, The dynamic weights are the weights assigned to the sub-health score of the i-th feature parameter. For information entropy, Let m be the probability that the historical data value of the i-th parameter falls within the m-th discrete interval, where m is the interval index. The formula for health score fusion is expressed as: in, This is a comprehensive voltage health index.

[0032] Furthermore, the dynamic threshold algorithm includes continuously storing the latest L health index values ​​to form a first-in-first-out sliding window, calculating the mean and standard deviation of the data within the sliding window in real time, calculating the degree of abnormality of the current value, setting an abnormality threshold for comparison, and determining the fault level. The formula for calculating the degree of anomaly is expressed as follows: Where Z represents the degree of abnormality of the current value. This represents the mean of the data within the sliding window. The standard deviation of the data within the sliding window. To avoid zero denominators for extremely small positive numbers; set a first threshold. Second threshold ,when When a serious system malfunction is detected, the highest-level alarm is immediately triggered, and preset protective actions are executed; when If the system is found to be in a state of degradation or early failure, an early warning signal is triggered to alert maintenance personnel, but no immediate hard-line tripping measures are taken.

[0033] S3: Based on the historical data sequence of key parameters in the health index and feature parameter set, perform trend extrapolation to generate predictive maintenance suggestions.

[0034] Furthermore, the trend extrapolation includes fitting and predicting the historical data sequence of health indicators and total harmonic distortion using a triple exponential smoothing method. The exponential smoothing method includes initializing the model, and for each new time point, after receiving new health index and harmonic distortion rate data, recursively updating the model components, and making predictions based on the updated model components. The formula for recursively updating model components is expressed as follows: in, Let be the horizontal component at time t, i.e., the base value of the time series at time t. For horizontal smoothing parameters, The comprehensive health status observation value at time t. Let be the seasonal component at time tr, and r be the length of a period. The horizontal component at time t-1 The trend component at time t-1, Let be the trend component at time t. For trend smoothing parameters, Let be the seasonal component at time t. For seasonal smoothing parameters; The prediction formula is: in, Let T be the predicted value at time t+T, where T is the prediction step size. For the seasonal component at the future time t+T; When the health index predicted by the trend extrapolation results continues to decline and falls below the warning threshold in the next T hours, and the total harmonic distortion rate is predicted to continue to rise, predictive maintenance recommendations are generated.

[0035] Furthermore, the maintenance recommendations are specifically for bus filter capacitors because increased THD and decreased overall health are often closely related to an increase in the equivalent series resistance or a decrease in capacitance. The algorithm can embed an empirical formula based on a physical model of capacitor degradation, which includes model coefficients and initial normal system values. When the equivalent series resistance exceeds the failure threshold, a specific maintenance instruction of "recommending to check and replace the bus filter capacitor" is triggered. This creatively combines abstract algorithm predictions with specific physical faulty components and degradation models, achieving a complete closed loop from monitoring, analysis, diagnosis to maintenance decision-making. The empirical formula is expressed as: in, This is an estimate of the equivalent series resistance of the capacitor. and These are the model coefficients. The current total harmonic distortion (THD) The initial total harmonic distortion (THD) For the current ripple frequency, This is the initial ripple frequency.

[0036] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0037] Example 2, refer to Figure 2 This is the second embodiment of the present invention. This embodiment provides a multi-dimensional real-time monitoring and analysis system for bus voltage, including: a signal preprocessing and anti-interference module, a multi-dimensional feature parallel extraction module, a dynamic weighted fusion health assessment module, and a trend prediction and maintenance decision module.

[0038] The signal preprocessing and anti-interference module is used to suppress Gaussian white noise and periodic pulse interference in the signal by using a composite filter, identify pulse events in real time through dynamic thresholds, and trigger an interpolation substitution mechanism to eliminate the influence, thus providing a clean voltage signal source for subsequent analysis.

[0039] The multidimensional feature parallel extraction module is used to accurately calculate the effective voltage value, total harmonic distortion rate, and amplitude and phase of each harmonic by using a windowed interpolation FFT algorithm in a parallel processing architecture. At the same time, it tracks the instantaneous phase, frequency, and frequency deviation of the fundamental voltage in real time by using an enhanced phase-locked loop algorithm.

[0040] The dynamic weighted fusion health assessment module is used to introduce information entropy theory, adaptively map each feature parameter to a sub-health value according to its own historical statistical characteristics, calculate the entropy value of each parameter data sequence to measure uncertainty, and perform dynamic weight allocation to obtain the voltage health index through fusion.

[0041] The trend prediction and maintenance decision module is used to learn the historical trends of health indicators and key parameters by applying time series prediction algorithms to predict future trends.

[0042] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0043] Example 3 is the third embodiment of the present invention, which differs from the previous two embodiments in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0044] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0045] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0046] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

Claims

1. A method for multi-dimensional real-time monitoring and analysis of bus voltage, characterized in that: The method comprises the steps of: The method comprises the steps of: The method comprises the steps of: The pre-processing comprises filtering the discrete-time sequence voltage signal by using a nonlinear adaptive filter.

2. The method of claim 1, wherein the method comprises: The filtering by using the nonlinear adaptive filter comprises initial filtering and error calculation, nonlinear pulse detection, nonlinear processing, and weight updating and output. The initial filtering and error calculation comprises preliminarily filtering the discrete-time sequence voltage signal by using a transversal finite impulse response filter based on an LMS algorithm to obtain a preliminarily filtered signal.

3. The method of claim 2, wherein the method comprises: The formula for calculating the preliminary error signal is: The nonlinear pulse detection comprises calculating the instantaneous amplitude of the error signal and comparing the instantaneous amplitude with a dynamic threshold value, wherein the dynamic threshold value is obtained by dynamically calculating the average value and the standard deviation of the error signal amplitude, and the average value and the standard deviation are recursively estimated by using a first-order low-pass filter; when the error signal is greater than the dynamic threshold value, it is determined that there is strong pulse interference at the current sampling point, and a flag bit is triggered to be 1. wherein, is an error signal, is a desired response, is a preliminary filtered signal, n is a time index; The formula for calculating the dynamic threshold value is: The nonlinear processing comprises, when the flag bit is 1, suspending the weight updating of the LMS algorithm, and starting a spline interpolation replacement mechanism; M uncontaminated error data points before and after the contaminated data point are used to construct a cubic spline interpolation function, and an interpolation estimation value at the current time is calculated to replace the original error signal value contaminated by the pulse; wherein is a dynamic threshold, is a recursive estimate of the mean of the absolute value of the error signal, K is a threshold coefficient, is a recursive estimate of the standard deviation of the absolute value of the error signal; The weight updating and output comprise, for the time not interfered by the pulse, i.e. when the flag bit is 0, the LMS filter uses a standard algorithm to update the filter weight vector, and the formula is: The characteristic parameter set comprises the voltage effective value, the total harmonic distortion rate, the harmonic amplitude and phase, the instantaneous fundamental phase, the real-time frequency, and the frequency deviation. wherein, is the weight vector of the filter at time n, is the weight vector of the filter at time n+1, is a step factor, is an input vector.

4. The method of claim 3, wherein the method comprises: The characteristic parameter set is obtained by synchronously extracting the multi-dimensional characteristic parameters of the voltage by using a spectrum analysis unit and a fundamental tracking unit to process the pre-processed discrete-time sequence voltage signal stream in parallel, wherein the spectrum analysis unit and the fundamental tracking unit output together to constitute the characteristic parameter set. The spectrum analysis unit comprises applying a window function to a data window length N pre-processed discrete-time sequence voltage signal data, performing FFT operation to obtain a frequency spectrum, and using an interpolation algorithm to correct the frequency, amplitude and phase of the fundamental and harmonic spectral lines in the frequency spectrum, and outputting a harmonic amplitude sequence, a harmonic phase sequence and a THD value. ​ The fundamental tracking unit comprises: a SOGI quadrature signal generator receiving the preprocessed discrete-time sequence voltage signal, generating two signals orthogonal to the input fundamental, and calculating the instantaneous phase error based on the two signals using a phase detector; a loop filter filtering out the high-frequency components in the phase error and outputting the frequency correction; and a voltage-controlled oscillator integrating the frequency correction output by the loop filter to generate the instantaneous phase and real-time frequency of the fundamental.

5. The method of claim 4, wherein the method comprises: The FFT operation comprises: performing a windowing operation using a Blackman-Harris window function, and correcting the spectrum lines of the windowed FFT using a dual-line interpolation algorithm; The Blackman-Harris window function is expressed as: wherein, is a window function value, n is a time index, N is a window length, , , and is a window function coefficient; The correction of the spectrum lines of the windowed FFT comprises: calculating the frequency deviation and the corrected amplitude for the highest spectrum line and the adjacent spectrum line within the specified range of the target harmonic frequency, and the formula is expressed as: wherein is the frequency deviation, is the complex spectral amplitude of the kth spectral line, and is the complex spectral amplitude of the left and right adjacent spectral lines of the kth spectral line, is a window function main lobe width related constant, is an adjustment factor, is the corrected real amplitude, is the real part of the window function spectral function, is the imaginary part of the window function spectral function; The dynamic weighting fusion algorithm comprises: sub-health mapping, calculation of dynamic weights, and health degree fusion; The sub-health mapping comprises: normalizing each parameter in the set of characteristic parameters to the interval [0, 1] using an improved S-shaped mapping function to obtain the sub-health degree, and the formula is expressed as: wherein, is a health state quantification value for the i-th characteristic parameter, is a current characteristic parameter value, is an average value of the i-th characteristic parameter over the historical data window, is a standard deviation of the i-th characteristic parameter over the historical data window, is a fluctuation range tolerance coefficient, is an S-curve steepness adjustment coefficient, is an exponential function; The formula for calculating the dynamic weights is expressed as: wherein, is a dynamic weight, i.e. a weight assigned to the sub-health degree of the i-th feature parameter, is an information entropy, is a probability that the historical data value of the i-th parameter falls in the m-th discrete interval, m being an interval index; The health degree fusion formula is expressed as: wherein, is the integrated voltage health index.

6. The method of claim 4, wherein the method comprises: The dynamic threshold algorithm comprises: continuously storing the values of the latest L health indicators to form a first-in-first-out sliding window, calculating the mean and standard deviation of the data in the sliding window in real time, calculating the abnormality degree of the current value, setting an abnormality degree threshold for comparison, and determining the fault level; The formula for calculating the abnormality degree is expressed as: wherein Z is the current value of the abnormality degree, is the mean of the data in the sliding window, is the standard deviation of the data in the sliding window, is a very small positive number to avoid zero denominator; set the first threshold and the second threshold , when , it is determined that the system has a serious failure, and the highest level of alarm is triggered immediately, and the preset protective action is performed; when , it is determined that the system has a degradation or early failure trend, and a warning signal is triggered to prompt the operation and maintenance personnel to pay attention, but the strong measure of tripping is not immediately executed.

7. The method of claim 4, wherein the method comprises: The trend extrapolation comprises: using a cubic exponential smoothing method to fit and predict the historical data sequence of the health indicators and the total harmonic distortion rate; ​ The exponential smoothing method comprises: model initialization, for each new time point, receiving new health indicators and harmonic distortion rate data, recursively updating the model components, and predicting based on the updated model components; The formula for recursively updating the model components is expressed as: wherein, is the level component at time t, i.e. the base value of the time series at time t, is the level smoothing parameter, is the observed value of the comprehensive health degree at time t, is the seasonal component at time t-r, r being the length of a cycle, is the level component at time t-1, is the trend component at time t-1, is the trend component at time t, is the trend smoothing parameter, is the seasonal component at time t, is the seasonal smoothing parameter; The prediction formula is: wherein, is a predicted value for a future time t+T, T being a prediction step, is a seasonal component for a future time t+T. When the predicted health indicators in the trend extrapolation result continue to decline and fall below the warning threshold within T hours in the future, and at the same time the predicted total harmonic distortion rate continues to rise, a predictive maintenance suggestion is generated.

8. A bus voltage multi-dimensional real-time monitoring and analyzing system, applying a bus voltage multi-dimensional real-time monitoring and analyzing method as claimed in any one of claims 1 to 7, characterized in that, It comprises: a signal preprocessing and anti-interference module, a multi-dimensional feature parallel extraction module, a dynamic weighting fusion health degree evaluation module, and a trend prediction and maintenance decision module; The signal preprocessing and anti-interference module is configured to use a composite filter to suppress Gaussian white noise and periodic pulse interference in the signal, identify pulse events in real time through a dynamic threshold, trigger an interpolation replacement mechanism to eliminate the influence, and provide a clean voltage signal source for subsequent analysis; The multi-dimensional feature parallel extraction module is configured to use a parallel processing architecture to accurately calculate the voltage effective value, the total harmonic distortion rate, and the amplitudes and phases of each harmonic through a windowed interpolation FFT algorithm, and simultaneously track the instantaneous phase, frequency, and frequency deviation of the fundamental voltage in real time through an enhanced phase-locked loop algorithm. The dynamic weighting fusion health degree evaluation module is used for introducing information entropy theory, adaptively mapping each characteristic parameter into a sub health degree according to its historical statistical characteristics, calculating the entropy value of each parameter data sequence to measure uncertainty, and performing dynamic weight distribution to obtain a voltage health degree index through fusion. The trend prediction and maintenance decision module is used for applying a time series prediction algorithm to learn the historical trend of the health degree index and the key parameters, and predict the future trend. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8. The processor executes the computer program to realize the steps of the bus voltage multi-dimensional real-time monitoring analysis method in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the bus voltage multi-dimensional real-time monitoring analysis method in any one of claims 1 to 7.

Images