Method for identifying misoperation of high-voltage switch cabinet

Abstract

The invention discloses a misoperation identification method for a high-voltage switch cabinet, relates to the technical field of misoperation identification of the high-voltage switch cabinet, and aims at sensing failure caused by dust coverage and misjudgment caused by electromagnetic interference. The high-robustness displacement track is reconstructed by using the vibration energy of the cabinet body, and the physical limitation of a traditional optical / magnetic induction scheme is broken through; aiming at an electromagnetic interference scene, designing a load adaptive ripple envelope extraction algorithm, combining anti-interference feature quantization and frequency domain mask training, and keeping current features stable under strong pulses; in addition, the problem of time domain signal drift is solved by aligning a motor current phase with a vibration spectrum; ridge regression weight optimization is combined with regularization self-adaption, so that the virtual displacement mapping precision is improved; and the environment sensing module dynamically switches decision weights to realize dust-electromagnetic dual-scene seamless protection.

Description

Technical Field

[0001] The present invention relates to the technical field of misoperation identification of high-voltage switch cabinets, and particularly to a misoperation identification method for high-voltage switch cabinets. Background Art

[0002] In high-dust operation scenarios such as mines and cement plants, high-voltage switch cabinets are long-term exposed to coal ash and metal dust environments; these micron-sized particulate matters continuously penetrate into the interior of the cabinet body, forming conductive scale; key components such as displacement sensors and photoelectric sensors relied on by existing anti-misoperation technologies are extremely easy to be covered by dust on the surface; for example, for the handcart position detection device, when the dust accumulates to a thickness of 0.1 mm at the induction window, the infrared beam scattering rate increases sharply, resulting in the system being unable to identify the true displacement of the handcart; more intractably, the dust in the humid mine tunnel will form a hard crust at the rotating shaft of the interlocking mechanism after being mixed with water vapor, causing mechanical jamming.

[0003] Non-contact monitoring solutions that have emerged in recent years, such as millimeter-wave radars and laser rangefinders, although they can reduce physical contact, the diffusive dust in the mine still causes serious signal attenuation; for example, the intelligent switch cabinet in the publication number CN216043169U adopts dual acoustic and optical alarms. In practical applications, the voice reminder will be drowned out in a strong noise environment, and the flashing indicator light has insufficient recognition in a dim roadway.

[0004] At present, when the sensor readings drift, the existing solutions either directly trigger false alarms or discard data, resulting in the failure of protection. Although some adaptive filtering algorithms can dynamically calibrate, they do not model the non-linear relationship between dust concentration and signal attenuation. The key lies in regarding dust as a pure interference factor and ignoring the fact that its physical deposition process itself carries device state information; therefore, there is an urgent need for a misoperation identification method for high-voltage switch cabinets to solve such problems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a misoperation identification method for high-voltage switch cabinets to solve the problems that dust in mines causes sensor failure, electromagnetic interference causes frequent false alarms, and existing solutions cannot take into account the robustness in harsh environments.

[0007] To solve the above technical problems, the present invention provides the following technical solutions:

[0008] An embodiment of the present invention provides a misoperation identification method for high-voltage switch cabinets, which includes,

[0009] Step S1, collecting multi-source sensor data, including cabinet door displacement signals, motor drive current waveforms, cabinet body vibration spectra, and environmental dust concentration data;

[0010] Step S2, dynamically compensate and process the multi-source sensor data to generate anti-interference operation timing data;

[0011] Step S3, extract the operation intention feature vectors, including the current ripple envelope feature parsed from the motor drive current waveform and the predetermined frequency band energy feature parsed from the cabinet vibration spectrum;

[0012] Step S4, input the operation intention feature vectors into the pre-trained operation behavior analysis model to output the misoperation risk level;

[0013] Step S5, perform hierarchical interlock actions according to the misoperation risk level.

[0014] As a preferred solution of the misoperation identification method for a high-voltage switch cabinet according to the present invention, wherein: the dynamic compensation process in Step S2 includes:

[0015] Construct a non-linear mapping relationship between the dust concentration and the displacement signal attenuation amplitude;

[0016] When a sudden change in the displacement signal attenuation amplitude is detected, activate the vibration reconstruction unit.

[0017] As a preferred solution of the misoperation identification method for a high-voltage switch cabinet according to the present invention, wherein: in the vibration reconstruction unit of Step S2, the physical conversion from vibration spectrum energy to displacement includes:

[0018] Perform energy aggregation on the continuously filtered frequency bands:

[0019] Wherein, Aggregate energy of the characteristic frequency band, unit dB, Sampling time, unit s, Lower limit frequency of the characteristic frequency band, unit Hz, Upper limit frequency of the characteristic frequency band, unit Hz, Gain compensation coefficient of the passband, Integral independent variable frequency, unit Hz, Vibration spectrum energy density, unit dB;

[0020] The aggregated energy is converted into a sequence linearly related to the mechanical displacement using an exponentially weighted model:

[0021] Wherein, Virtual displacement sequence, unit mm, Sampling time, unit s, Frequency component index, dimensionless, Total number of frequency components participating in the mapping, dimensionless, The Component energy displacement conversion weight, unit , The instantaneous energy density of the component, in dB, The center frequency of the component, in Hz, background noise power spectrum, in dB, nonlinear energy index, dimensionless, with an empirical range of 0.8 - 1.2;

[0022] The weight vector is solved by ridge regression in closed form, and an adaptive expression of the regularization factor is given:

[0023] ,

[0024] ,

[0025] where weight vector , in units of , sample energy matrix, composed of , in dB, indicating the corresponding moment of the sampling sequence number, indicating the sample index, transpose of ridge regression regularization factor, dimensionless, identity matrix, dimensionless, actual displacement observation vector, in mm, regularization ratio coefficient, with an empirical range of 0.01 - 0.05, dimensionless, is the maximum singular value of , in dB.

[0026] As a preferred solution of the method for identifying misoperations of high - voltage switchgear according to the present invention, wherein: the execution logic of the vibration reconstruction unit includes:

[0027] Extract the characteristic frequency band in the cabinet vibration spectrum that is strongly correlated with mechanical displacement;

[0028] Based on the phase information of the motor drive current waveform, align the energy of the characteristic frequency band in the time domain;

[0029] Generate a virtual displacement sequence that is linearly correlated with the actual displacement of the cabinet door;

[0030] The determination method of the characteristic frequency band strongly correlated with mechanical displacement is:

[0031] Collect the vibration spectrum samples of the switchgear trolley within the standard displacement range;

[0032] Calculate the Pearson correlation coefficient between the energy of each frequency component and the displacement distance;

[0033] Screen continuous frequency bands with a correlation coefficient exceeding a first threshold and a signal-to-noise ratio exceeding a second threshold.

[0034] As a preferred solution of the method for identifying misoperations of a high-voltage switchgear according to the present invention, wherein: the extraction of the current ripple envelope feature in step S3 includes:

[0035] Perform band-pass filtering on the motor drive current waveform to isolate the ripple component;

[0036] Calculate the rising edge slope of the ripple component and the number of zero crossings per unit time.

[0037] As a preferred solution of the method for identifying misoperations of a high-voltage switchgear according to the present invention, wherein: in step S3, the calculation of the anti-interference feature of the motor drive current ripple envelope includes:

[0038] Construct a band-pass impulse response with a center frequency adjusted according to the load:

[0039] ,

[0040] wherein, is the impulse response of the point filter, is the discrete-time index, dimensionless, is the ripple center frequency, in Hz, is the half bandwidth, in Hz, is the sampling frequency, in Hz, is the Kaiser window coefficient;

[0041] Perform the Hilbert transform on the filtered output to obtain the analytic signal:

[0042] ,

[0043] wherein, is the ripple envelope amplitude, in A, is the current waveform after band-pass, in A, is the Hilbert transform operator, is the imaginary unit, dimensionless, is the sampling time, in s;

[0044] Calculate two anti-interference features within the observation window , and the calculation formula is:

[0045] ,

[0046] ,

[0047] wherein, is the envelope energy normalization ratio, dimensionless, is the envelope rising edge variation coefficient, dimensionless, is the observation window length, unit s, is the th sampling moment, unit s, is the sample index, dimensionless, is the point envelope amplitude, unit A, is the median absolute deviation operator, is the median operator; the final anti-interference eigenvector is denoted as .

[0048] As a preferred solution of the method for identifying misoperation of a high-voltage switch cabinet according to the present invention, wherein: the training of the operation behavior analysis model in step S4 includes:

[0049] Injecting standard electromagnetic pulse interference samples into historical operation data;

[0050] Adopting frequency domain masking technology to enhance the robustness of vibration spectrum features;

[0051] Optimizing the decision boundary of the convolutional neural network through adversarial training.

[0052] As a preferred solution of the method for identifying misoperation of a high-voltage switch cabinet according to the present invention, it further includes:

[0053] Environmental anti-interference mode switching, when the environmental electromagnetic field strength exceeds the standard:

[0054] Suppressing the current ripple envelope feature in the operation intention eigenvector;

[0055] Boosting the decision weight of the cabinet vibration spectrum feature to a preset value;

[0056] The triggering conditions for the environmental anti-interference mode switching include:

[0057] Real-time monitoring of the spectral distribution of the environmental electromagnetic field strength;

[0058] When it is detected that the energy in the 15 kHz - 100 kHz frequency band continuously exceeds the standard, it is determined as an electromagnetic interference scenario;

[0059] If the duration of exceeding the standard exceeds the set value, start the frequency domain feature priority decision mechanism.

[0060] As a preferred solution of the method for identifying misoperation of a high-voltage switch cabinet according to the present invention, wherein: the hierarchical interlocking action in step S5 includes:

[0061] When the misoperation risk level is at the first level, trigger the cabinet door electromagnetic locking device;

[0062] When the risk level of misoperation is level two, simultaneously perform the opening of the circuit breaker and the closing of the earthing switch.

[0063] Generate a security event log containing the hash value of the feature vector.

[0064] As a preferred solution of the method for identifying misoperation of a high-voltage switchgear cabinet described in the present invention, wherein: the method for determining the characteristic frequency band includes:

[0065] Analyze the vibration frequency spectrum distribution during the movement of the switchgear cabinet handcart.

[0066] Select the frequency band with the correlation coefficient between the energy change and the displacement distance greater than 0.8.

[0067] Exclude the frequency band dominated by environmental background noise.

[0068] The beneficial effects of the present invention are as follows: The present invention aims at the problems of sensing failure caused by dust coverage and misjudgment caused by electromagnetic interference. Through the physical conversion model from vibration spectrum energy to displacement, when the displacement sensor is blocked by dust, the vibration energy of the cabinet body is used to reconstruct a highly robust displacement trajectory, breaking through the physical limitations of traditional optical / magnetic induction schemes; for the electromagnetic interference scenario, a load-adaptive ripple envelope extraction algorithm is designed, combined with anti-interference feature quantization and frequency domain mask training, to keep the current characteristics stable under strong pulses; in addition, the time-domain signal drift problem is solved by aligning the motor current phase with the vibration spectrum; the ridge regression weight optimization is combined with regularization adaptation to improve the virtual displacement mapping accuracy; the environmental perception module dynamically switches the decision weights to achieve seamless protection in the dust - electromagnetic double scenario; and through millisecond-level risk classification interlocking, electromagnetic locking / circuit breaker opening, the passive alarm is converted into active interception, significantly reducing the risk of serious accidents.

[0069] The present invention reuses existing industrial sensors and edge computing units without modifying the structure of the switchgear cabinet, and is particularly suitable for deployment in harsh environments such as mines and substations, providing intelligent protection for the operation safety of high-voltage equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0071] Figure 1 It is a schematic flow chart of the method for identifying misoperation of the high-voltage switchgear cabinet in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification.

[0073] In the following description, numerous specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may be practiced in other ways different from those described herein. Persons skilled in the art may make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0074] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0075] Embodiment 1, referring to Figure 1 , this embodiment provides a method for identifying misoperations of a high-voltage switchgear, including the following steps:

[0076] Step S1, collect multi-source sensor data, including cabinet door displacement signals, motor drive current waveforms, cabinet vibration spectra, and environmental dust concentration data;

[0077] Step S2, perform dynamic compensation processing on the multi-source sensor data to generate anti-interference operation timing data;

[0078] The dynamic compensation processing in Step S2 includes:

[0079] Construct a non-linear mapping relationship between dust concentration and displacement signal attenuation amplitude;

[0080] When a sudden change in the attenuation amplitude of the displacement signal is detected, activate the vibration reconstruction unit;

[0081] In the vibration reconstruction unit of Step S2, perform physical conversion from vibration spectrum energy to displacement, including:

[0082] Perform energy aggregation on the continuously filtered frequency bands:

[0083] ,

[0084] Among them, Characteristic frequency band aggregated energy, unit dB, Sampling time, unit s, Lower limit frequency of the characteristic frequency band, unit Hz, Upper limit frequency of the characteristic frequency band, unit Hz, Gain compensation coefficient of the passband, Integral independent variable frequency, unit Hz, Vibration spectrum energy density, unit dB; Integration compresses the high-dimensional spectrum into a single energy trajectory, and noise is dissipated within the integration window. Gain compensation corrects for differences in the sensing link; The resulting trajectory changes monotonically with the door body travel, and remains smooth even in the presence of pulse interference. The impact of dust occlusion on the energy curve is negligible, laying a stable foundation for subsequent displacement mapping;

[0085] The aggregated energy is transformed into a sequence linearly related to the mechanical displacement using an exponentially weighted model:

[0086] ,

[0087] where, Virtual displacement sequence, unit mm, Sampling time, unit s, Frequency component index, dimensionless, Total number of frequency components participating in the mapping, dimensionless, The Component energy displacement conversion weight, unit , The Instantaneous energy density of the component, unit dB, The Center frequency of the component, unit Hz, Background noise power spectrum, unit dB, Nonlinear energy index, dimensionless, empirical range 0.8 - 1.2; The exponentially weighted model maintains a consistent slope in both small strokes and acceleration sections, solving the problem of kinks in the energy displacement relationship under different operating conditions; The noise spectrum is subtracted component by component, and the virtual displacement trajectory has a smoothing property for small signal fluctuations. The rising and falling sections are symmetric, and can be directly fed into the interlock threshold determination without additional filtering;

[0088] The weight vector is solved using ridge regression in closed form, and an adaptive expression for the regularization factor is given:

[0089] ,

[0090] ,

[0091] where, Weight vector , unit , Sample energy matrix, composed of , unit dB, Indicates the time corresponding to the sampling sequence number, Indicates the sample index, The transpose of, Ridge regression regularization factor, dimensionless, Identity matrix, dimensionless, Actual displacement observation vector, unit: mm Regular proportionality coefficient, empirical range: 0.01 - 0.05, dimensionless is the maximum singular value of, unit: dB; The closed - form solution avoids oscillations in the iterative process and maintains stable convergence when the samples are sparse or the frequency components are highly correlated; The regularization factor is adaptively adjusted according to the matrix spectral norm, which not only restricts the excessive weight but also prevents under - fitting; After calibration, the correlation between the virtual displacement curve and the laser - measured curve is greatly improved, and the fluctuation in repeated experiments is controlled within an acceptable range, providing a reliable input for the multi - source fusion module;

[0092] The execution logic of the vibration reconstruction unit includes:

[0093] Extract the characteristic frequency band in the cabinet vibration spectrum that is strongly correlated with mechanical displacement;

[0094] Based on the phase information of the motor drive current waveform, align the energy of the characteristic frequency band in the time domain;

[0095] Generate a virtual displacement sequence that is linearly correlated with the actual displacement of the cabinet door;

[0096] The determination method of the characteristic frequency band that is strongly correlated with mechanical displacement is:

[0097] Collect the vibration spectrum samples of the switch cabinet trolley within the standard displacement range;

[0098] Calculate the Pearson correlation coefficient between the energy of each frequency component and the displacement distance;

[0099] Screen the continuous frequency bands whose correlation coefficient exceeds the first threshold and the signal - to - noise ratio exceeds the second threshold;

[0100] The determination methods of the characteristic frequency band include:

[0101] Analyze the vibration spectrum distribution during the movement of the switch cabinet trolley;

[0102] Select the frequency bands whose correlation coefficient between the energy change and the displacement distance is greater than 0.8;

[0103] Exclude the frequency bands dominated by environmental background noise;

[0104] Step S3, extract the operation intention feature vector, including the current ripple envelope feature parsed from the motor drive current waveform and the energy feature of the predetermined frequency band parsed from the cabinet vibration spectrum;

[0105] The extraction of the current ripple envelope feature in step S3 includes:

[0106] Perform band - pass filtering on the motor drive current waveform to separate the ripple component;

[0107] Calculate the rising edge slope of the ripple component and the number of zero crossings per unit time;

[0108] In step S3, calculate the anti-interference characteristics of the motor drive current ripple envelope, including:

[0109] Construct a band-pass impulse response with a center frequency adjusted according to the load:

[0110] ,

[0111] where, is the impulse response of the point filter, is the discrete-time index, dimensionless, is the ripple center frequency, in Hz, is the half bandwidth, in Hz, is the sampling frequency, in Hz, is the Kaiser window coefficient; the adaptive center frequency moves synchronously with the drift of the load current to avoid misfiltering the actual ripple component; the sidelobe attenuation of the Kaiser window is fast, suppressing adjacent harmonic leakage, and the half bandwidth dynamically converges according to the statistical results of the ripple energy distribution, and can maintain a stable passband gain during the electromagnetic interference rise period; the filtered waveform only retains the current commutation ripple, and the fundamental wave and spike pulses are attenuated by more than 30 dB, and no additional notch compensation is required for subsequent envelope solution;

[0112] Perform the Hilbert transform on the filtered output to obtain the analytic signal:

[0113] ,

[0114] where, is the ripple envelope amplitude, in A, is the current waveform after band-pass, in A, is the Hilbert transform operator, is the imaginary unit, dimensionless, is the sampling time, in s; the analytic signal method gives the instantaneous amplitude and phase at one time, avoiding the two-channel error caused by the splitting of the upper and lower envelopes; symmetric filtering can keep the Hilbert phase shift at 90°, and the boundary effect is suppressed by zero-phase extension; the envelope trajectory has a high coupling degree with the mechanical shock period, and occasional pulses are converted into isolated spikes, which is convenient for statistical filtering and improves the credibility of subsequent robust indicators;

[0115] Calculate two anti-interference characteristics within the observation window , and the calculation formula is:

[0116] ,

[0117] ,

[0118] Among them, is the envelope energy normalization ratio, dimensionless, is the envelope rising edge variation coefficient, dimensionless, is the observation window length, unit s, is the th sampling moment, unit s, is the sample index, dimensionless, is the th point envelope amplitude, unit A, is the median absolute deviation operator, is the median operator; the final anti-interference feature vector is denoted as ; here Comparing the mean square amplitude and the arithmetic mean amplitude can quickly identify the ripple distortion caused by abnormal armature commutation, Using MAD to characterize the rising edge jitter has natural robustness to pulse interference and occasional glitches; both exponents are dimensionless, facilitating cross-device normalization;

[0119] Step S4, input the operation intention feature vector into the pre-trained operation behavior analysis model, and output the misoperation risk level;

[0120] The training of the operation behavior analysis model in step S4 includes:

[0121] Inject standard electromagnetic pulse interference samples into historical operation data;

[0122] Adopt frequency domain masking technology to enhance the robustness of vibration spectrum features;

[0123] Optimize the decision boundary of the convolutional neural network through adversarial training;

[0124] Step S5, perform hierarchical interlock actions according to the misoperation risk level;

[0125] The hierarchical interlock actions in step S5 include:

[0126] When the misoperation risk level is at the first level, trigger the cabinet door electromagnetic locking device;

[0127] When the misoperation risk level is at the second level, synchronously perform the circuit breaker opening and the grounding switch closing;

[0128] Generate a security event log containing the hash value of the feature vector;

[0129] This method further includes:

[0130] Environmental anti-interference mode switching, when the environmental electromagnetic field intensity exceeds the standard:

[0131] Suppress the current ripple envelope feature in the operation intention feature vector;

[0132] Increase the decision weight of the cabinet vibration spectrum feature to a preset value;

[0133] The triggering conditions for environmental anti-interference mode switching include:

[0134] Real-time monitor the spectral distribution of the environmental electromagnetic field intensity;

[0135] When it is detected that the energy in the 15 kHz - 100 kHz frequency band continuously exceeds the standard, it is determined as an electromagnetic interference scenario;

[0136] If the duration of exceeding the standard exceeds the set value, start the frequency-domain feature priority decision-making mechanism.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying misoperation of a high-voltage switchgear, characterized in that, including Step S1, collecting multi-source sensor data, including cabinet door displacement signals, motor drive current waveforms, cabinet vibration spectra, and environmental dust concentration data; Step S2, dynamically compensating and processing the multi-source sensor data to generate anti-interference operation timing data; Step S3, extracting operation intention feature vectors, including current ripple envelope features parsed from the motor drive current waveform and predetermined frequency band energy features parsed from the cabinet vibration spectrum; Step S4, inputting the operation intention feature vectors into a pre-trained operation behavior analysis model to output the misoperation risk level; Step S5, performing hierarchical interlock actions according to the misoperation risk level.

2. The misoperation recognition method of a high-voltage switchgear according to claim 1, characterized in that The dynamic compensation processing in Step S2 includes: Constructing a non-linear mapping relationship between dust concentration and displacement signal attenuation amplitude; When a sudden change in the displacement signal attenuation amplitude is detected, activating the vibration reconstruction unit.

3. The method for identifying misoperation of a high-voltage switchgear according to claim 2, wherein, In the vibration reconstruction unit of Step S2, performing physical conversion from vibration spectrum energy to displacement, including: Performing energy aggregation on continuously filtered frequency bands: ， Among them, Characteristic frequency band aggregation energy, unit: dB, Sampling time, unit: s, Lower limit frequency of characteristic frequency band, unit: Hz, Upper limit frequency of characteristic frequency band, unit: Hz, Gain compensation coefficient of passband, Integral independent variable frequency, unit: Hz, Vibration spectrum energy density, unit: dB; Converting the aggregated energy into a sequence linearly related to mechanical displacement using an exponentially weighted model: ， Among them, Virtual displacement sequence, unit: mm, Sampling time, unit: s, Frequency component index, dimensionless, Total number of frequency components participating in mapping, dimensionless, The Component energy displacement conversion weight, unit , The Component instantaneous energy density, unit: dB, The Component center frequency, unit: Hz, Background noise power spectrum, unit: dB, Nonlinear energy index, dimensionless, empirical range 0.8 - 1.2; Solving the weight vector using ridge regression in closed form and giving an adaptive expression of the regularization factor: ， ， Among them, weight vector , unit , sample energy matrix, consisting of units of dB, indicating the corresponding time of the sampling sequence number, indicating the sample index, transpose of ridge regression regularization factor, dimensionless, identity matrix, dimensionless, actual displacement observation vector, unit mm, regularization ratio coefficient, empirical range 0.01 - 0.05, dimensionless, is the maximum singular value of, unit dB.

4. The misoperation recognition method of a high-voltage switchgear according to claim 2, characterized in that, The execution logic of the vibration reconstruction unit includes: Extracting the characteristic frequency band in the cabinet vibration spectrum that is strongly correlated with mechanical displacement; Performing time-domain alignment on the energy of the characteristic frequency band based on the phase information of the motor drive current waveform; Generating a virtual displacement sequence linearly related to the actual displacement of the cabinet door; The determination method of the characteristic frequency band strongly correlated with mechanical displacement is: Collecting vibration spectrum samples of the switch cabinet trolley within the standard displacement range; Calculating the Pearson correlation coefficient between the energy of each frequency component and the displacement distance; Selecting continuous frequency bands with a correlation coefficient exceeding the first threshold and a signal-to-noise ratio exceeding the second threshold.

5. The misoperation identification method of a high-voltage switchgear according to claim 1, characterized in that The extraction of the current ripple envelope features in Step S3 includes: Performing band-pass filtering on the motor drive current waveform to separate the ripple component; Calculating the rising edge slope and the number of zero crossings per unit time of the ripple component.

6. The method for identifying misoperation of a high-voltage switchgear according to claim 5, characterized in that, In Step S3, performing anti-interference feature calculation on the motor drive current ripple envelope, including: Constructing a band-pass impulse response with a center frequency adjusted according to the load: ， wherein, is the point filter impulse response, is the discrete-time index, dimensionless, is the ripple center frequency, in Hz, is the half bandwidth, in Hz, is the sampling frequency, in Hz, is the Kaiser window coefficient; Perform a Hilbert transform on the filtered output to obtain an analytic signal: ， Among them, is the ripple envelope amplitude, unit A, is the current waveform after band-pass, unit A, is the Hilbert transform operator, is the imaginary unit, dimensionless, is the sampling time, unit s; In the observation window Calculate two anti-interference features, and the calculation formula is: ， ， Among them, is the envelope energy normalization ratio, dimensionless, is the envelope rising edge variation coefficient, dimensionless, is the observation window length, unit s, is the th sampling moment, unit s, is the sample index, dimensionless, is the th point envelope amplitude, unit A, is the median absolute deviation operator, is the median operator; the final anti-interference eigenvector is denoted as .

7. The method for identifying misoperation of a high-voltage switchgear according to claim 1, characterized in that, The training of the operation behavior analysis model in Step S4 includes: Injecting standard electromagnetic pulse interference samples into historical operation data; Using frequency domain masking technology to enhance the robustness of vibration spectrum features; Optimizing the decision boundary of the convolutional neural network through adversarial training.

8. The misoperation recognition method of a high-voltage switchgear according to claim 1, characterized in that, It also includes: Environmental anti-interference mode switching. When the environmental electromagnetic field strength exceeds the standard: Suppressing the current ripple envelope features in the operation intention feature vectors; Increasing the decision weight of the cabinet vibration spectrum features to a preset value; The triggering conditions for the environmental anti-interference mode switching include: Real-time monitoring of the spectral distribution of the environmental electromagnetic field strength; When it is detected that the energy in the 15 kHz - 100 kHz frequency band continuously exceeds the standard, it is determined as an electromagnetic interference scenario; If the exceeding standard duration exceeds the set value, starting the frequency domain feature priority decision mechanism.

9. The misoperation recognition method of a high-voltage switchgear as described in claim 1, characterized in that, The hierarchical interlock actions in Step S5 include: When the misoperation risk level is at the first level, triggering the cabinet door electromagnetic locking device; When the misoperation risk level is level two, simultaneously execute the opening of the circuit breaker and the closing of the earthing switch; Generate a security event log containing the hash value of the feature vector.

10. A misoperation recognition method for a high-voltage switchgear according to claim 4, characterized in that, The determination method of the feature frequency band includes: Analyze the vibration frequency spectrum distribution during the movement of the switchgear trolley; Select the frequency band where the correlation coefficient between the energy change and the displacement distance is greater than 0.8; Exclude the frequency band dominated by environmental background noise.

Images