Method and system for diagnosing mechanical fault of pole-mounted circuit breaker

Abstract

The invention relates to the technical field of power equipment fault diagnosis, and particularly discloses a pole-mounted circuit breaker mechanical fault diagnosis method and system, and the method comprises the steps: firstly, synchronously collecting a dynamic force-displacement signal, a high-frequency acoustic emission signal and a broadband vibration signal in the switching-on and switching-off process of a circuit breaker operating mechanism; then, the acoustic emission and vibration signals are decoupled into an impact source component and a friction source component which are statistically independent through a blind source separation algorithm; performing envelope spectrum analysis on the impact source component to extract impact characteristics, and performing energy calculation on the friction source component to obtain friction noise energy characteristics; finally, on the basis of dynamic force-displacement curve fitting analysis, in combination with impact characteristics and friction noise energy characteristics, distinguishing and delimiting of mechanical wear and lubrication degradation faults are achieved. According to the invention, the problem that the existing diagnosis technology cannot effectively identify and distinguish the coupling fault of mechanical wear and lubrication deterioration is solved, and early warning and accurate diagnosis of the mechanical fault of the pole-mounted circuit breaker are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment fault diagnosis, and particularly relates to a mechanical fault diagnosis method and system for pole-mounted circuit breaker. BACKGROUND

[0002] As a key equipment of power distribution network, the reliability of the mechanical operating mechanism of the pole-mounted circuit breaker directly affects the safety of the power grid. At present, the state assessment is mainly performed by periodic maintenance and simple mechanical characteristic test (such as opening and closing time and speed) in engineering. However, the mechanical wear and lubrication state deterioration inside the mechanism often occur simultaneously and are coupled with each other, which are manifested as common phenomena such as increased operating resistance and abnormal action time. The traditional method can only find abnormal macroscopic parameters, but cannot distinguish the fault source, that is, it is difficult to determine whether the impact is intensified due to cam wear or the friction is increased due to lubricating grease aging.

[0003] The prior art has the following disadvantages: the existing mechanical fault diagnosis method for the pole-mounted circuit breaker cannot effectively distinguish and early identify the hidden mechanism failure problem caused by the mutual coupling of mechanical wear and lubrication deterioration, specifically including: the traditional vibration analysis or force-displacement curve monitoring can find the abnormal action of the mechanism, but cannot effectively distinguish whether the abnormal signal is caused by the mechanical impact generated by the micro wear on the cam surface or the increased friction resistance caused by the deterioration of the lubricating grease. SUMMARY

[0004] The purpose of the present application is to provide a mechanical fault diagnosis method and system for pole-mounted circuit breaker to solve the problems in the above background.

[0005] The purpose of the present application can be achieved by the following technical solutions: A mechanical fault diagnosis method for pole-mounted circuit breaker, comprising the following steps: S1: synchronously collecting dynamic force-displacement signals, high-frequency acoustic emission signals and wideband vibration signals of the pole-mounted circuit breaker when the operating mechanism of the pole-mounted circuit breaker performs opening and closing operation; S2: taking the high-frequency acoustic emission signals and the wideband vibration signals as observation signals, processing them through a blind source separation algorithm to obtain statistically independent impact source components and friction source components; S3: performing envelope spectrum analysis on the impact source components to extract impact features related to the rotating frequency of the operating mechanism; and simultaneously performing energy calculation on the friction source components to obtain friction noise energy features thereof; S4: constructing a standard dynamic force-displacement curve based on the dynamic force-displacement signal, fitting the actually collected dynamic force-displacement signal with the standard dynamic force-displacement curve, identifying a fitting deviation, and combining impact characteristics and friction noise energy characteristics to delimit a fault source of the fitting deviation and output a fault diagnosis result, wherein the fault source includes a mechanical wear fault dominated by the impact characteristics and a lubrication deterioration fault dominated by the friction noise energy characteristics.

[0006] As a further scheme of the present application, the S2 specifically comprises: The high-frequency acoustic emission signal and the wideband vibration signal are respectively preprocessed, and the preprocessing includes removing a direct current component and band-pass filtering to obtain two preprocessed observation signals. The two preprocessed observation signals are combined to form an observation signal vector, and a blind source separation algorithm based on a negative entropy maximization criterion is used to iteratively demix the observation signal vector until convergence, thereby obtaining a demixing matrix and corresponding two statistically independent source signal components. According to the time-domain waveform characteristics and frequency energy distribution of the two source signal components, a source signal component that presents pulse sparse characteristics in the time domain and whose frequency energy is mainly concentrated in the high-frequency band is determined as an impact source component, and a source signal component that presents continuous noise characteristics in the time domain and whose frequency energy is mainly concentrated in the medium and low frequency bands is determined as a friction source component.

[0007] As a further scheme of the present application, the S3 specifically comprises: The impact source component is subjected to envelope demodulation processing, and the envelope signal of the impact source component is obtained through Hilbert transform. The envelope signal is subjected to spectrum analysis, and in the obtained envelope spectrum, spectral peaks that accurately correspond to the fundamental frequency and harmonic frequencies of the operating mechanism rotation frequency are identified, and the amplitude weighted sum of these spectral peaks is taken as the impact characteristics. The sequence of the Tag energy operator of the friction source component in the entire opening and closing operation time range is calculated, and the statistical average value of the sequence of the Tag energy operator is calculated, and the statistical average value is taken as the friction noise energy characteristics.

[0008] As a further scheme of the present application, the envelope demodulation processing of the impact source component to obtain the envelope signal of the impact source component through Hilbert transform specifically comprises: The impact source component is subjected to complex wavelet transform, a complex wavelet basis function with analyticality is selected, and the wavelet coefficient modulus maximum value curve of the impact source component is obtained. Based on the wavelet coefficient modulus maximum value curve, the instantaneous amplitude sequence of the impact source component is extracted. The instantaneous amplitude sequence is subjected to smoothing processing, and the processed instantaneous amplitude sequence is taken as the envelope signal.

[0009] As a further aspect of the present invention: the spectral analysis of the envelope signal, and the identification of spectral peaks in the obtained envelope spectrum that precisely correspond to the fundamental frequency and harmonic frequencies of the operating mechanism's rotation frequency, specifically includes: A spectrum refinement method based on full-phase spectrum analysis is used to process the envelope signal to obtain a high-resolution envelope spectrum; A rotational frequency reference template is established based on the design parameters of the operating mechanism. The rotational frequency reference template includes the theoretical frequency values ​​of the fundamental frequency and the first six harmonics and their allowable fluctuation range. In the high-resolution envelope spectrum, find the spectral peaks that precisely correspond to each frequency component in the reference template.

[0010] As a further aspect of the present invention: S4 specifically includes: The dynamic time warping algorithm is used to nonlinearly align the actual acquired dynamic force-displacement signal with the standard dynamic force-displacement curve, and calculate the minimum path bending cost between the two curves. Based on the minimum path curvature cost, the local deformation features of the actual acquired signal relative to the standard curve are extracted. The local deformation features include phase lead region, phase lag region and amplitude anomaly region. Establishing a mapping relationship between local deformation characteristics and fault types, specifically including: By correlating the phase lead region with impact characteristics, mechanical wear faults can be identified; By correlating the phase lag region with the energy characteristics of friction noise, lubrication deterioration faults can be identified. Based on the quantified values ​​of impact characteristics and friction noise energy characteristics, combined with the distribution density of local deformation characteristics, the severity index of various faults is calculated, and diagnostic results containing fault type and severity are generated.

[0011] As a further aspect of the present invention: the calculation of the minimum path bending cost between two curves specifically includes: Construct a dynamic programming path search matrix, using the Euclidean distance between each sampling point of the actual acquired signal and each sampling point of the standard curve as the basic cost, and fill the dynamic programming path search matrix. Set asymmetric local path constraints to limit the path search direction to unidirectional expansion only along the standard curve direction, while limiting the continuous bending angle to no more than a set threshold. Using a multi-scale recursive calculation method, starting from the starting point of the path search matrix, the minimum cumulative cost to reach each position is calculated point by point until the minimum path curvature cost of the entire matrix is ​​obtained. The optimal nonlinear correspondence between two curves is determined by finding the transmission path with the minimum cumulative cost in the reverse tracing path search matrix.

[0012] As a further aspect of the present invention: the establishment of the mapping relationship between local deformation features and fault types specifically includes: The spatial distribution of the phase lead region is quantified, and its regional overlap with the high gradient change segment in the standard dynamic force-displacement curve is calculated to form a phase lead quantification index. A persistence analysis of the phase lag region is performed to calculate the proportion of its persistence length in the flat section of the standard dynamic force-displacement curve, thus forming a quantitative index of frictional lag. When the product of the phase lead quantification index and the impact characteristic exceeds the first threshold, it is determined that mechanical wear failure is dominant; when the product of the friction retardation quantification index and the friction noise energy characteristic exceeds the second threshold, it is determined that lubrication deterioration failure is dominant. For cases where two thresholds are exceeded simultaneously, the ratio of the two product values ​​is calculated, and the dominant fault type and coupling degree are determined based on the magnitude of the ratio.

[0013] As a further aspect of the present invention: the calculation process of the severity index is as follows: The impact characteristics are normalized, and the ratio of impact energy to background noise energy is used as the impact intensity factor. At the same time, the frequency of impact events during the monitoring period is calculated as the impact frequency factor. The frequency band characteristics of friction noise energy are decomposed to extract the relative change rate of noise energy in a specific frequency band compared with the historical baseline value. The friction degradation factor is then calculated by combining the absolute level of noise energy. The local deformation features are statistically analyzed by gridding and partitioning. The density of deformation feature points in each partition is calculated, and the average density of the three partitions with the highest density is calculated as the deformation concentration factor. The impact intensity factor, impact frequency factor, friction deterioration factor, and deformation concentration factor are weighted and fused together, with the weight of impact factors set to 0.4, the weight of friction factors set to 0.3, and the weight of deformation factors set to 0.3. The comprehensive failure severity index is obtained by weighted summation.

[0014] A mechanical fault diagnosis system for pole-mounted circuit breakers includes: The signal acquisition module is used to simultaneously acquire dynamic force-displacement signals, high-frequency acoustic emission signals, and broadband vibration signals of the pole-mounted circuit breaker when the operating mechanism of the pole-mounted circuit breaker performs opening and closing operations. The signal decoupling module is used to treat high-frequency acoustic emission signals and broadband vibration signals as observation signals, and process them through a blind source separation algorithm to decompose them into statistically independent impact source components and friction source components. The feature extraction module is used to perform envelope spectrum analysis on the impact source component and extract the impact features related to the rotation frequency of the operating mechanism; at the same time, it performs energy calculation on the friction source component to obtain its friction noise energy characteristics. The fault diagnosis module constructs a standard dynamic force-displacement curve based on the dynamic force-displacement signal, performs fitting analysis between the actual collected dynamic force-displacement signal and the standard dynamic force-displacement curve, and identifies the fitting deviation. Combining impact characteristics and friction noise energy characteristics, it delineates the root causes of the fitting deviation and outputs the fault diagnosis results. The root causes of the fault include mechanical wear faults dominated by impact characteristics and lubrication deterioration faults dominated by friction noise energy characteristics.

[0015] The beneficial effects of this invention are: (1) By combining multi-sensor signal synchronous acquisition with blind source separation technology, the impact component and friction noise component mixed in acoustic emission and vibration signals are effectively decoupled, overcoming the limitation of traditional single-parameter diagnostic methods in identifying complex faults. By using complex wavelet envelope demodulation and Tiger energy operator to extract impact features and friction energy features respectively, it is possible to clearly distinguish the two types of fault roots, mechanical wear and lubrication deterioration, based on the fitting analysis of dynamic force-displacement curves, which significantly improves the accuracy and reliability of early and latent fault diagnosis.

[0016] (2) Based on the dynamic time warping algorithm to identify fitting deviations, and combined with the distribution density of impact characteristics, friction noise energy characteristics and local deformation characteristics, a multi-factor weighted fault severity index was constructed. This index can not only determine the fault type, but also quantify and classify the severity of the fault, thereby providing an accurate basis for equipment maintenance decisions. It effectively avoids the shortcomings of traditional methods that can only determine "whether there is a fault" but cannot assess "fault severity", thus improving the pertinence and economy of maintenance work. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0019] The technical solutions of 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.

[0020] Please see Figure 1 As shown, this invention provides a method for diagnosing mechanical faults in pole-mounted circuit breakers, comprising the following steps: S1: When the operating mechanism of the pole-mounted circuit breaker performs opening and closing operations, the dynamic force-displacement signal, high-frequency acoustic emission signal and broadband vibration signal of the pole-mounted circuit breaker are collected simultaneously. S2: High-frequency acoustic emission signals and broadband vibration signals are used as observation signals and processed by blind source separation algorithm to decompose them into statistically independent impact source components and friction source components. S3: Perform envelope spectrum analysis on the impact source component to extract the impact characteristics related to the rotation frequency of the operating mechanism; at the same time, perform energy calculation on the friction source component to obtain its friction noise energy characteristics. S4: Construct a standard dynamic force-displacement curve based on the dynamic force-displacement signal, and perform fitting analysis between the actual collected dynamic force-displacement signal and the standard dynamic force-displacement curve to identify the fitting deviation; combine the impact characteristics and friction noise energy characteristics to delineate the root causes of the fitting deviation and output the fault diagnosis results, where the root causes of the fault include mechanical wear faults dominated by impact characteristics and lubrication deterioration faults dominated by friction noise energy characteristics.

[0021] In S1, when the operating mechanism of the pole-mounted circuit breaker performs opening and closing operations, the dynamic force-displacement signal, high-frequency acoustic emission signal, and broadband vibration signal of the pole-mounted circuit breaker are simultaneously acquired, specifically including: The acquisition of dynamic force-displacement signals is achieved through a combination of a six-dimensional force sensor and a laser displacement sensor mounted on the transmission chain of the operating mechanism. The six-dimensional force sensor is fixed to the key force point of the moving contact linkage, directly measuring the force and torque components in three directions during the opening and closing process. The transmitter and receiver of the laser displacement sensor are respectively mounted on the mechanism frame and the moving parts, and the linear displacement of the moving contact is accurately calculated by detecting the change in the reflected phase of the laser beam. The force and displacement signals are impedance matched and amplified by a signal conditioner before being synchronously recorded by a data acquisition card.

[0022] The high-frequency acoustic emission signal is acquired using a resonant acoustic emission sensor. This sensor is fixed to the position of the operating mechanism housing with the shortest vibration transmission path via a magnetic chuck. The sensor directly captures stress wave signals generated within the material due to defects such as microscopic wear and crack propagation.

[0023] The broadband vibration signal acquisition uses a triaxial integrated circuit piezoelectric accelerometer. This sensor is bonded to the exposed rigid housing surface of the operating mechanism with epoxy resin. The sensor synchronously detects the vibration acceleration in three mutually perpendicular directions during the movement of the mechanism. The output signal is converted into a voltage signal by a charge amplifier and then sent to the data acquisition system.

[0024] In S2, high-frequency acoustic emission signals and broadband vibration signals are used as observation signals and processed by a blind source separation algorithm to decompose them into statistically independent impact source components and friction source components, specifically including: First, the acquired high-frequency acoustic emission signals and broadband vibration signals were preprocessed separately. The preprocessing consisted of two consecutive operations: first, removing the DC component by subtracting the arithmetic mean of the signals; second, performing bandpass filtering using an eighth-order Butterworth filter. The bandpass range for the high-frequency acoustic emission signals was 80 kHz to 300 kHz, and for the broadband vibration signals, it was 2 kHz to 8 kHz. After preprocessing, two observation signals were obtained that eliminated baseline drift and preserved effective frequency band components.

[0025] Two preprocessed observation signals are combined into a two-dimensional observation signal vector, which is then processed using a blind source separation algorithm based on the negative entropy maximization criterion. The algorithm's implementation includes the following computational steps: First, the observation signal vector is centered by subtracting the mean vector of each channel. Second, the centered signal is whitened by calculating the eigenvalue decomposition of the covariance matrix to obtain the whitening matrix and performing a linear transformation on the signal, ensuring that the transformed signal components are uncorrelated and have unit variance. Third, a fixed-point iterative algorithm is used to solve for the separation matrix. By selecting the hyperbolic tangent function as the nonlinear function, an approximate value of the negative entropy is calculated, and the weight vector of the separation matrix is ​​continuously updated with the goal of maximizing negative entropy. The algorithm is considered convergent when the Euclidean distance between the weight vectors of two adjacent iterations is less than 0.0001. Finally, the unmixed matrix and the corresponding two statistically independent source signal components are obtained.

[0026] Component identification is performed based on the time-domain waveform characteristics and frequency-domain energy distribution of the two source signal components. Time-domain waveform characteristic analysis includes calculating the kurtosis coefficient and impulse factor of the signal. Source signal components with a kurtosis coefficient greater than 5 and an impulse factor greater than 10 are identified as having impulse sparsity characteristics. Frequency-domain energy distribution analysis involves calculating the power spectral density of the signal and observing the frequency bands where the signal energy is concentrated. Source signal components with energy mainly concentrated in the high-frequency band above 5 kHz are identified as impact source components, corresponding to mechanical impact events. Source signal components with energy mainly concentrated in the mid-to-low frequency band below 5 kHz and exhibiting continuous noise characteristics in the time domain are identified as friction source components, corresponding to friction processes. Through the combined judgment of the above time-domain and frequency-domain characteristics, the accurate identification and classification of the two source signal components are achieved.

[0027] In the component identification process, the specific criteria are set as follows: First, the kurtosis coefficient of each source signal component is calculated, which is the ratio of the fourth central moment to the square of the second central moment. Then, the impulse factor of each source signal component is calculated, which is the ratio of the peak value to the root mean square value. Simultaneously, the energy proportion of each source signal component in the frequency band above 5 kHz is calculated, i.e., the ratio of the power spectral density integral value in that frequency band to the power spectral density integral value across the entire frequency band. When a source signal component simultaneously satisfies a kurtosis coefficient greater than 5, an impulse factor greater than 10, and a high-frequency energy proportion exceeding 60%, it is determined to be an impulse source component. When a source signal component simultaneously satisfies a kurtosis coefficient less than 3, an impulse factor less than 5, and a low-frequency energy proportion exceeding 70%, it is determined to be a friction source component. Through this series of quantitative characteristic parameter calculations and threshold comparisons, the accuracy and reliability of the component identification results are ensured.

[0028] In S3, envelope spectrum analysis is performed on the impact source component to extract impact characteristics related to the rotation frequency of the operating mechanism; simultaneously, energy calculation is performed on the friction source component to obtain its friction noise energy characteristics, specifically including: The impact source component undergoes envelope demodulation processing, and the envelope signal of the impact source component is obtained through complex wavelet transform. Specifically, the implementation includes: firstly, selecting the Morlet complex wavelet as the basis function, whose mathematical expression is: ; in, For wavelet basis functions, For time variables, The center frequency parameter is set to 6.0. The imaginary unit, The base of the natural number is represented. A continuous wavelet transform is performed on the impulse source component, and its wavelet coefficients at various scales are calculated. The modulus values ​​of the wavelet coefficients are extracted to form a time-frequency distribution matrix. Then, by finding local maxima points of the modulus in the time direction, these maxima points are connected to form a wavelet coefficient modulus maxima curve. Based on this modulus maxima curve, a wavelet reconstruction algorithm is used to recover the instantaneous amplitude sequence of the signal. Finally, the instantaneous amplitude sequence is smoothed. An 11-point median filter is used to remove abnormal pulses, and a 21-point moving average filter is used to smooth random fluctuations to obtain the final envelope signal.

[0029] Spectral analysis of the envelope signal was performed, and a high-resolution envelope spectrum was obtained using the full-phase spectral analysis method. The specific steps of the full-phase spectral analysis included: first, the envelope signal was processed using a dual-window method, with a Hanning window for the front window and a rectangular window for the rear window, with a window length of 1024 points. The signal was then segmented and sampled with overlap, the overlap length being the window length minus 1. Windowing and Fast Fourier Transform were performed on each data segment. Finally, the spectral results of all segments were phase-aligned and averaged to obtain the envelope spectrum. A rotational frequency reference template was established based on the design parameters of the operating mechanism. This template included the theoretical frequency values ​​of the fundamental frequency and the first six harmonics, calculated based on the design rotational speed of the operating mechanism, with an allowable fluctuation range of ±2% of the theoretical value. In the obtained high-resolution envelope spectrum, local maxima were searched within a frequency range of ±2%, centered on the frequency values ​​in the reference template. When the amplitude of a spectral peak exceeded three times the average amplitude of adjacent frequency points, the corresponding characteristic spectral peak was considered to have been found. The amplitudes of these characteristic spectral peaks are weighted and summed according to their harmonic orders, with the weighting coefficients set to the reciprocal of the harmonic order. The resulting weighted sum is used as the impact characteristic.

[0030] Energy calculations are performed on the friction source components to obtain the energy characteristics of friction noise. Specifically, the Tiger energy operator is used for calculation, and its mathematical expression is as follows: ; in, This indicates that the Tiger energy operator at time [time] The output value, For the friction source component at time... The sampled values, and These are the sampled values ​​at adjacent time points. During calculation, the operator is first applied point-by-point to the friction source component throughout the entire opening and closing operation time range, resulting in the Tiger energy operator sequence. Then, outlier processing is performed on this sequence, removing outliers exceeding five standard deviations of the sequence mean. Finally, a statistical average is calculated from the processed sequence, which is the sum of the energy values ​​of all valid points divided by the number of valid points; this average is used as the friction noise energy characteristic.

[0031] In envelope spectral analysis, an adaptive thresholding method is used to identify characteristic spectral peaks. Specifically, a frequency window of ±2% is taken, centered on each theoretical frequency in the baseline template. The average and standard deviation of the amplitudes of all spectral lines within this window are calculated. If the amplitude of a spectral line exceeds the average plus three times the standard deviation, and this line is within ±0.5% of the theoretical frequency, it is considered a valid characteristic peak. For fundamental frequency characteristics, the peak amplitude must be at least five times the average background noise; for harmonic characteristics, the peak amplitude must be at least twice the average background noise. All identified valid characteristic peaks are used in the final impulse characteristic calculation.

[0032] The calculation of friction noise energy characteristics requires normalization. The normalization factor is the statistical average of the Tiger energy operator sequence of the normal friction source components of the operating mechanism under rated operating conditions. The calculated friction noise energy characteristic value is divided by this normalization factor to obtain the relative energy value. When the relative energy value is between 0.8 and 1.2, the lubrication condition is considered normal; when the relative energy value is between 1.2 and 1.5, slight lubrication degradation is considered; when the relative energy value is between 1.5 and 2.0, moderate lubrication degradation is considered; and when the relative energy value is greater than 2.0, severe lubrication degradation is considered. This classification provides a quantitative basis for subsequent fault diagnosis.

[0033] The calculation of the impact characteristic also considers the signal-to-noise ratio (SNR) of each harmonic component. For each identified characteristic spectral peak, its SNR is calculated, which is the ratio of the peak amplitude to the average background noise value within its frequency window. Only when the SNR is greater than 2.0 is the harmonic component included in the impact characteristic calculation. Simultaneously, different weighting coefficients are assigned to spectral peaks of different harmonic orders: 1.0 for the fundamental frequency, 0.8 for the second harmonic, 0.6 for the third harmonic, 0.4 for the fourth harmonic, 0.2 for the fifth harmonic, and 0.1 for the sixth harmonic. This weighting distribution reflects the higher importance of lower harmonics in the impact characteristic compared to higher harmonics. The final impact characteristic value is the sum of the spectral peak amplitudes of all effective harmonic components multiplied by their corresponding weighting coefficients.

[0034] In S4, a standard dynamic force-displacement curve is constructed based on the dynamic force-displacement signal. The actual acquired dynamic force-displacement signal is then fitted and analyzed with the standard dynamic force-displacement curve to identify the fitting deviation. Combining impact characteristics and friction noise energy characteristics, the root causes of the fitting deviation are delineated, and the fault diagnosis results are output. The root causes include mechanical wear faults dominated by impact characteristics and lubrication deterioration faults dominated by friction noise energy characteristics, specifically including: A dynamic time warping algorithm is used to nonlinearly align the actual acquired dynamic force-displacement signals with the standard dynamic force-displacement curve. First, a dynamic programming path search matrix is ​​constructed, where rows correspond to the sampling point indices of the standard curve, and columns correspond to the sampling point indices of the actual acquired signals. The fundamental cost of each element in the matrix is ​​obtained by calculating the Euclidean distance between the corresponding two points, using the following formula: ; in, Indicates basic value. The standard curve is in the th... Point of force, The standard curve is in the th... Displacement value of the point, These represent the actual signal at the th... Point of force, Indicates the actual signal at the th The displacement value of the point. Set asymmetric local path constraints, limiting the path search direction to unidirectional expansion only along the standard curve direction. Specifically, the constraint is... arrive ,from arrive ,from arrive Movement in three directions is possible, while limiting continuous bending angles to no more than 45 degrees. A multi-scale recursive calculation method is employed, starting from the initial point of the path search matrix. Begin by calculating the distance to each position point by point according to the following recursive formula. Minimum cumulative cost: ; in, Indicates from the starting point to the point The minimum cumulative cost. After calculation, the element in the lower right corner of the matrix. The value of is the minimum path bending cost between the two curves, where and These represent the total number of sampling points for the standard curve and the actual signal, respectively. By tracing the propagation path with the minimum cumulative cost in reverse, the optimal nonlinear correspondence between the two curves is determined.

[0035] Based on the minimum path curvature cost, local deformation features of the actual acquired signal relative to the standard curve are extracted. First, based on the correspondence between the two curves obtained through dynamic time warping, the time difference and amplitude difference between each corresponding point pair are calculated. Regions with continuously positive time differences for more than 5 sampling points are marked as phase-leading regions, and regions with continuously negative time differences for more than 5 sampling points are marked as phase-lag regions. Regions with amplitude differences whose absolute value exceeds 10% of the amplitude of the corresponding point on the standard curve are marked as amplitude-abnormal regions. Spatial distribution quantization is performed on the phase-leading regions, calculating their overlap with high-gradient change segments in the standard dynamic force-displacement curve. Specifically, the proportion of overlapping points to the total number of points is calculated to form a phase-leading quantization index. A persistence analysis is performed on the phase lag region, calculating its proportion of the smooth section of the standard dynamic force-displacement curve. Specifically, the ratio of the length of the phase lag region in the smooth section to the total length of the smooth section is calculated to form a quantitative index of frictional lag. .

[0036] Establish a mapping relationship between local deformation characteristics and fault types. When the phase leads the quantization index... Impact characteristics When the product exceeds the first threshold of 0.15, it is determined that mechanical wear failure is the dominant factor; when the frictional resistance quantification index Energy characteristics of frictional noise When the product of the two values ​​exceeds the second threshold of 0.12, it is determined that the lubrication deterioration fault is dominant. For cases where both thresholds are exceeded simultaneously, the ratio R of the two product values ​​is calculated. When R > 1.2, it is determined that the mechanical wear fault is dominant; when R < 0.8, it is determined that the lubrication deterioration fault is dominant; and when 0.8 ≤ R ≤ 1.2, it is determined that the fault is mixed.

[0037] Based on the quantified values ​​of impact characteristics and friction noise energy characteristics, combined with the distribution density of local deformation characteristics, the severity index of various faults is calculated. First, the impact characteristics are normalized, and the ratio of impact energy to background noise energy is calculated as the impact intensity factor. Simultaneously, the frequency of impact events occurring within the monitoring period is calculated as the impact frequency factor. The friction noise energy characteristics are decomposed into frequency bands, and the relative rate of change of noise energy in the 2kHz to 5kHz frequency band compared with the historical baseline value is extracted. The friction degradation factor is then calculated by combining the absolute level of noise energy. The calculation formula is: ; in, For the current noise energy, This is the historical baseline value. For the maximum allowable noise energy, and The weighting coefficients are 0.6 and 0.4, respectively. Local deformation features are statistically analyzed by gridding, dividing the force-displacement curve into a 10×10 grid. The density of deformation feature points within each grid is calculated, and the average density of the three grids with the highest density is selected as the deformation concentration factor. Finally, the factors are weighted and fused to calculate the comprehensive failure severity index S: ; According to the severity index The magnitude of the fault is used to classify the severity of the fault into four levels: S < 0.3 for minor, 0.3 ≤ S < 0.6 for moderate, 0.6 ≤ S < 0.9 for severe, and S ≥ 0.9 for dangerous. The final diagnostic results, including the fault type and severity, provide a basis for equipment maintenance decisions.

[0038] Please see Figure 2 As shown, a mechanical fault diagnosis system for pole-mounted circuit breakers includes: The signal acquisition module is used to simultaneously acquire dynamic force-displacement signals, high-frequency acoustic emission signals, and broadband vibration signals of the pole-mounted circuit breaker when the operating mechanism of the pole-mounted circuit breaker performs opening and closing operations. The signal decoupling module is used to treat high-frequency acoustic emission signals and broadband vibration signals as observation signals, and process them through a blind source separation algorithm to decompose them into statistically independent impact source components and friction source components. The feature extraction module is used to perform envelope spectrum analysis on the impact source component and extract the impact features related to the rotation frequency of the operating mechanism; at the same time, it performs energy calculation on the friction source component to obtain its friction noise energy characteristics. The fault diagnosis module constructs a standard dynamic force-displacement curve based on the dynamic force-displacement signal, performs fitting analysis between the actual collected dynamic force-displacement signal and the standard dynamic force-displacement curve, and identifies the fitting deviation. Combining impact characteristics and friction noise energy characteristics, it delineates the root causes of the fitting deviation and outputs the fault diagnosis results. The root causes of the fault include mechanical wear faults dominated by impact characteristics and lubrication deterioration faults dominated by friction noise energy characteristics.

[0039] The working principle of this invention is as follows: Dynamic force-displacement signals, high-frequency acoustic emission signals, and broadband vibration signals are simultaneously acquired during the opening and closing process of the operating mechanism. The acoustic emission and vibration signals are then used as observation signals, and a blind source separation algorithm is employed to decouple statistically independent impact and friction source components. Next, complex wavelet envelope demodulation and full-phase envelope spectrum analysis are performed on the impact source components to extract impact features related to the mechanism's rotation frequency. Simultaneously, the friction source components are calculated using the Tiger energy operator to determine their friction noise energy characteristics. Finally, a standard curve is constructed based on the dynamic force-displacement signals, and nonlinear fitting is performed using a dynamic time warping algorithm. Fitting deviations are identified, and the root causes of these deviations are delineated by combining the impact and friction noise energy characteristics. This results in a diagnostic assessment of the type and severity of mechanical wear or lubrication deterioration faults.

[0040] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A pole-mounted circuit breaker mechanical fault diagnostic method characterized by, The method comprises the following steps: S1: synchronously collect the dynamic force-displacement signal, high-frequency acoustic emission signal and wideband vibration signal of the column-mounted circuit breaker when the operating mechanism of the column-mounted circuit breaker performs opening and closing operations; S2: process the high-frequency acoustic emission signal and the wideband vibration signal as observation signals through a blind source separation algorithm to obtain statistically independent impact source components and friction source components through decomposition; S3: perform envelope spectrum analysis on the impact source components to extract impact features related to the rotating frequency of the operating mechanism; meanwhile, perform energy calculation on the friction source components to obtain friction noise energy features; S4: construct a standard dynamic force-displacement curve based on the dynamic force-displacement signal, perform fitting analysis on the actually collected dynamic force-displacement signal and the standard dynamic force-displacement curve, identify fitting deviations, and combine the impact features and the friction noise energy features to delimit the fault sources of the fitting deviations and output fault diagnosis results, wherein the fault sources include mechanical wear faults dominated by the impact features and lubrication deterioration faults dominated by the friction noise energy features.

2. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 1, wherein, The S2 specifically comprises: performing preprocessing on the high-frequency acoustic emission signal and the wideband vibration signal respectively, wherein the preprocessing comprises removing direct current components and band-pass filtering to obtain two preprocessed observation signals; grouping the two preprocessed observation signals into an observation signal vector, iteratively demixing the observation signal vector through a blind source separation algorithm based on a negative entropy maximization criterion until convergence is achieved, thereby obtaining a demixing matrix and corresponding two statistically independent source signal components; performing component identification according to the time-domain waveform features and frequency energy distribution of the two source signal components: determining a source signal component that presents pulse sparsity characteristics in the time domain and whose frequency energy is mainly concentrated in the high-frequency band as an impact source component; and determining a source signal component that presents continuous noise characteristics in the time domain and whose frequency energy is mainly concentrated in the medium and low frequency bands as a friction source component.

3. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 1, wherein, The S3 specifically comprises: performing envelope demodulation processing on the impact source component to obtain the envelope signal of the impact source component through Hilbert transform; performing spectrum analysis on the envelope signal, identifying spectrum peaks that accurately correspond to the base frequency and harmonic frequencies of the rotating frequency of the operating mechanism in the obtained envelope spectrum, and taking the weighted sum of the amplitudes of these spectrum peaks as the impact features; calculating the sequence of Teager energy operators of the friction source component within the entire opening and closing operation time range, and taking the statistical average value of the sequence of Teager energy operators as the friction noise energy features.

4. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 3, wherein, The envelope demodulation processing on the impact source component to obtain the envelope signal of the impact source component through Hilbert transform specifically comprises: performing complex wavelet transform on the impact source component, selecting a complex wavelet basis function with analyticality, and obtaining the wavelet coefficient modulus maximum value curve of the impact source component; extracting the instantaneous amplitude sequence of the impact source component based on the wavelet coefficient modulus maximum value curve; performing smoothing processing on the instantaneous amplitude sequence, and taking the processed instantaneous amplitude sequence as the envelope signal.

5. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 3, wherein, The spectrum analysis on the envelope signal to identify spectrum peaks that accurately correspond to the base frequency and harmonic frequencies of the rotating frequency of the operating mechanism in the obtained envelope spectrum specifically comprises: The envelope signal is processed by a spectrum refinement method based on full-phase spectrum analysis to obtain a high-resolution envelope spectrum; A rotating frequency reference template is established based on the design parameters of the operating mechanism, and the rotating frequency reference template contains the theoretical frequency values of the fundamental frequency and the first six harmonics and their allowable fluctuation ranges; In the high-resolution envelope spectrum, find the spectral peaks that accurately correspond to each frequency component in the reference template.

6. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 1, wherein, The S4 specifically comprises: The actual collected dynamic force-displacement signal and the standard dynamic force-displacement curve are nonlinearly aligned by using a dynamic time warping algorithm, and the minimum path bending cost between the two curves is calculated; Based on the minimum path bending cost, the local deformation characteristics of the actual collected signal relative to the standard curve are extracted, including the phase advance region, the phase lag region, and the amplitude abnormal region; The mapping relationship between the local deformation characteristics and the fault types is established, specifically including: The phase advance region is associated with the impact feature to identify mechanical wear faults; The phase lag region is associated with the friction noise energy feature to identify lubrication degradation faults; According to the quantitative values of the impact feature and the friction noise energy feature, and in combination with the distribution density of the local deformation characteristics, the severity index of each type of fault is calculated to generate a diagnosis result containing the fault type and the severity.

7. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 6, wherein, The calculation of the minimum path bending cost between the two curves specifically comprises: A dynamic programming path search matrix is constructed, taking the Euclidean distance between each sampling point of the actual collected signal and each sampling point of the standard curve as the basic value, and the dynamic programming path search matrix is filled; An asymmetric local path constraint condition is set to limit the path search direction to only allow one-way expansion along the standard curve direction, while limiting the continuous bending angle to not exceed a set threshold; The minimum cumulative cost value of reaching each position is calculated from the starting point of the path search matrix to the entire matrix by using a multi-scale recursive calculation method, and the minimum path bending cost of the entire matrix is obtained; The optimal nonlinear correspondence between the two curves is determined by tracing the transmission path of the minimum cumulative cost in the path search matrix in reverse.

8. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 6, wherein, The mapping relationship between the local deformation characteristics and the fault types is established, specifically including: The phase advance region is quantized in space, and the area overlap of the phase advance region with the high gradient change section in the standard dynamic force-displacement curve is calculated to form a phase advance quantization index; The phase lag region is analyzed for persistence, and the length ratio of the phase lag region in the flat section of the standard dynamic force-displacement curve is calculated to form a friction drag quantization index; When the product of the phase advance quantization index and the impact feature exceeds a first threshold, it is determined that the mechanical wear fault is dominant; when the product of the friction drag quantization index and the friction noise energy feature exceeds a second threshold, it is determined that the lubrication degradation fault is dominant; For the case where both thresholds are exceeded, the ratio of the two product values is calculated to determine the dominant fault type and the coupling degree according to the ratio size.

9. The pole-mounted circuit breaker mechanical fault diagnostic method of claim 6, wherein, The calculation process of the severity index is as follows: The impact feature is normalized, and the ratio of the impact energy to the background noise energy is taken as the impact intensity factor, and the occurrence frequency of the impact event in the monitoring period is calculated as the impact frequency factor; The friction noise energy features are band-decomposed, the relative change rate of the noise energy of a specific frequency band and a historical baseline value is extracted, and a friction degradation factor is calculated based on the absolute level of the noise energy; The local deformation features are grid-partitioned and counted, the deformation feature point density in each partition is calculated, and the average density of the three partitions with the largest density is selected as the deformation concentration factor; The impact intensity factor, impact frequency factor, friction degradation factor, and deformation concentration factor are fused by weighting, the impact factor weight is set to 0.4, the friction factor weight is set to 0.3, and the deformation factor weight is set to 0.3, and the comprehensive fault severity index is obtained by weighted summation.

10. A pole-mounted circuit breaker mechanical fault diagnostic system, characterized by, A mechanical fault diagnosis method for a pole-mounted circuit breaker is used to execute any one of claims 1-9, comprising: A signal acquisition module is configured to synchronously acquire dynamic force-displacement signals, high-frequency acoustic emission signals, and broadband vibration signals of the pole-mounted circuit breaker when the operating mechanism of the pole-mounted circuit breaker performs opening and closing operations. A signal decoupling module is configured to process the high-frequency acoustic emission signals and broadband vibration signals as observation signals through a blind source separation algorithm to obtain statistically independent impact source components and friction source components. A feature extraction module is configured to perform envelope spectrum analysis on the impact source components to extract impact features related to the rotating frequency of the operating mechanism, and to perform energy calculation on the friction source components to obtain friction noise energy features. A fault determination module is configured to construct a standard dynamic force-displacement curve based on the dynamic force-displacement signals, to perform fitting analysis on the actual acquired dynamic force-displacement signals and the standard dynamic force-displacement curve to identify fitting deviations, and to combine the impact features and the friction noise energy features to determine the fault source of the fitting deviations and output the fault diagnosis results, wherein the fault source includes mechanical wear faults dominated by the impact features and lubrication deterioration faults dominated by the friction noise energy features.

Images