Automatic detection method for low-voltage intelligent measurement switch

Abstract

The application discloses an automatic detection method of a low-voltage intelligent measurement switch, current waveforms in normal debugging and open states are collected through a current sensor, zero detection is performed to extract a zero-crossing time sequence and a time interval sequence, variance and a sign function are used to calculate a time interval fluctuation amplitude and a fluctuation frequency respectively, and a waveform envelope extraction algorithm is used to obtain a residual oscillation period number and an oscillation disappearance time, difference operation is performed based on pre-stored normal reference parameters, a comprehensive abnormal score is obtained through weighted fusion, and threshold comparison is performed to realize fluctuation characteristic over-standard judgment, residual oscillation parameters are compared with a normal debugging range to realize extinguishing characteristic abnormality judgment, and two types of judgment outputs are combined to output normal or abnormal and are accompanied by abnormal coding. The method realizes automatic extraction and fusion evaluation of multi-dimensional characteristics, and improves the accuracy and traceability of switch state detection.

Description

Technical Field

[0001] This invention relates to the field of measurement switch technology, and more specifically to an automatic detection method for a low-voltage intelligent measurement switch. Background Technology

[0002] Low-voltage intelligent measuring switches play a crucial role in low-voltage power distribution systems, performing functions such as energy metering, line protection, and remote control. Their operational reliability and consistency directly impact the safety of the power grid. During the production, commissioning, and on-site maintenance of these switches, automatic detection of their opening and closing characteristics is necessary to determine if there are hidden defects such as contact bounce, poor arc extinguishing, or mechanism jamming. Currently, the testing of low-voltage intelligent measuring switches often relies on manual observation of current waveforms using an oscilloscope, or on simple current threshold comparisons and opening / closing time measurements to determine the switch status.

[0003] Current measurement technologies primarily rely on direct numerical measurements of electrical parameters such as current and voltage, along with simple threshold comparisons, lacking the ability to deeply quantify waveform timing characteristics. Traditional methods focus only on the magnitude of steady-state parameter amplitudes, failing to capture the dynamic transient signals generated during switching actions. They cannot effectively extract and characterize the oscillation decay and time interval fluctuations during the arc-breaking process of switching, and are insufficient in identifying early, weak anomalies or intermittent faults. Fixed threshold settings rely on human experience and lack adaptive capabilities, easily leading to misjudgments or missed detections when environmental noise or load fluctuations cause changes in signal amplitude. For example, short-term current fluctuations caused by load switching during normal commissioning may be misjudged as faults, triggering unnecessary repairs and increasing maintenance costs. Existing technologies lack mechanisms for classifying and encoding abnormal features; detection results can only output a binary conclusion of normal or abnormal, failing to provide directional references for fault tracing. Fault investigation requires manual verification of waveform records one by one, resulting in low detection efficiency. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an automatic detection method for low-voltage intelligent measuring switches.

[0005] The technical solution provided by this invention to solve the above problems is: an automatic detection method for a low-voltage intelligent measuring switch, the method comprising the following steps: S1: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch open state are collected by the current sensor, and the zero-crossing point extraction is performed on the current waveform data based on the zero-crossing detection algorithm to obtain the zero-crossing point time sequence Z and the time interval sequence T. S2: The fluctuation amplitude A is obtained by calculating the fluctuation amplitude of the time interval sequence T using the variance formula, and the fluctuation frequency F is obtained by counting the number of incremental sign flips of the time interval sequence T using the sign function. S3: The waveform envelope extraction algorithm is used to truncate the current waveform data into a window and calculate the attenuation rate to obtain the attenuation rate sequence R. The residual oscillation period number N and the oscillation disappearance time Δt are extracted by traversing the attenuation rate sequence R. At the same time, the range is extracted based on the zero-crossing time sequence Z of the current waveform data under normal debugging conditions to obtain the normal debugging range. , ] and [Δ ,Δ ]; S4: Based on the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal baseline parameters , Perform the difference operation to obtain ΔA and ΔF; S5: Set weights w1=0.6 and w2=0.4, and calculate the comprehensive anomaly score S= w1×ΔA + w2×ΔF using the weighted summation formula. Set the weighted threshold. The comprehensive anomaly score S and A comparison is made to determine if the fluctuation characteristics exceed the standard, and the residual oscillation period N is compared with the normal adjustment range. , Compare the oscillation disappearance time Δt with the normal debugging range [Δt]. ,Δ The results are compared to obtain the extinguishing feature abnormality judgment. The combined fluctuation feature over-limit judgment and the extinguishing feature abnormality judgment are used to determine the result. The result is output as normal or abnormal, with an abnormal code attached, as the automatic detection result of the low-voltage intelligent measuring switch.

[0006] Preferably, the time interval sequence T includes the spacing between adjacent zeros, the half-wavelength duration, and the frequency sampling period.

[0007] Preferably, the fluctuation frequency F includes the historical symbol holding time, the flip rate, and the phase drift frequency.

[0008] Preferably, the normal debugging range includes voltage harmonic tolerance, frequency disturbance boundary, and initial phase deviation; the parameters transmitted by the reference difference generation module are specifically waveform distortion, amplitude deviation, and phase change; and the automatic detection results include contact ablation degree encoding, mechanism jamming risk encoding, and insulation degradation level encoding.

[0009] Preferably, S1 specifically includes: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch open state are collected. The analog-to-digital conversion and amplitude normalization processing are performed on the two sets of analog signals in sequence to convert the voltage amplitude value in each sampling period to a uniform value range and obtain the current waveform sampling value. Call the current waveform sampling value, traverse the values ​​of two adjacent sampling points in each group of sampling values ​​and compare them with the zero value one by one, detect whether the sign of the sampling point value changes from positive to negative or from negative to positive, record the sampling point index position and corresponding time for each sign change, and obtain the zero-crossing time value. Based on the zero-crossing time values, all zero-crossing points are arranged in chronological order to obtain the zero-crossing time sequence Z. The time difference between two adjacent zero-crossing times is calculated and recorded in sequence to obtain the time interval sequence T, thus generating the zero-crossing time sequence Z and the time interval sequence T.

[0010] Preferably, S2 specifically includes: Get the time interval sequence T, sum up all elements in the sequence one by one and divide by the total number of elements to generate the sequence mean; Call the sequence mean and the time interval sequence T, subtract the mean from each element and calculate the square of the difference, sum all the squared differences and divide by the number of elements to get the fluctuation range A; The fluctuation frequency submodule calls the fluctuation amplitude A and the time interval sequence T, calculates the difference between adjacent elements and applies the sign function to extract the sign value, compares whether adjacent sign values ​​are consistent, accumulates the count for inconsistent cases, and obtains the fluctuation frequency F.

[0011] Preferably, S3 specifically includes: Acquire current waveform data, apply waveform envelope extraction algorithm to find the peak and valley points of each cycle of the current waveform and connect them to form upper and lower envelopes, segment the envelope into windows and set the window duration, calculate the maximum and minimum values ​​of the envelope amplitude in each window, subtract the minimum value from the maximum value and divide by the maximum value to obtain the attenuation rate, and generate the attenuation rate sequence R; Based on the decay rate sequence R, starting from the first value in the sequence, the decay rate values ​​are sequentially compared with the set threshold. The indices of the first and last values ​​below the threshold are recorded. The index difference is calculated as the residual oscillation period N. The index difference is multiplied by the sampling interval of the current waveform data to obtain the oscillation disappearance time Δt. Simultaneously, based on the original current waveform data, the zero-crossing time is detected to generate a zero-crossing time sequence Z. The time difference between adjacent zero-crossing points is calculated, and the maximum difference is selected as the upper limit and the minimum difference as the lower limit to obtain the normal debugging range. The debugging range value is then generated. , ] and [Δ ,Δ ].

[0012] Preferably, the generation of the decay rate sequence R specifically includes: The original current waveform time series is acquired by the built-in current sensor of the intelligent measurement switch. Multi-scale decomposition of wavelet transform is performed on the original current waveform time series. The absolute value of each detail coefficient is compared with the corresponding layer threshold. If the absolute value is greater than the threshold, the original coefficient value is retained. If the absolute value is not greater than the threshold, the coefficient is replaced with zero. The inverse wavelet transform is performed using all the processed layer detail coefficients and the fourth layer approximation coefficients to output the reconstructed waveform sequence. Based on the reconstructed waveform sequence, each data point is traversed, and the amplitude of the point is compared with the amplitudes of the two points before and after it. Peak points and valley points are marked. The upper and lower envelope curves are formed by linear interpolation. The peak amplitudes are extracted to form feature vectors. The local extremum identification method is called to obtain the standard template vector. The Euclidean distance between the current feature vector and the standard template vector is calculated. The Euclidean distance is compared with the upper limit of the allowable deviation. If the Euclidean distance is not greater than the upper limit of the allowable deviation, the lower envelope valley amplitude sequence corresponding to the waveform is retained. Based on the amplitude of the lower envelope valley value sequence, extract the amplitude of three consecutive valley values, calculate the ratio R1 obtained by dividing the amplitude of the first valley value by the amplitude of the second valley value, calculate the ratio R2 obtained by dividing the amplitude of the second valley value by the amplitude of the third valley value, and take the arithmetic mean of R1 and R2 as the current attenuation rate. Repeat the above calculation for each segment of three adjacent valley values ​​in the lower envelope valley value amplitude sequence to generate the attenuation rate sequence R. This scheme, through wavelet transform multi-scale decomposition and adaptive threshold processing of detail coefficients, can still preserve the low-frequency trend of the true waveform when the noise amplitude reaches a large signal amplitude, avoiding false detections caused by noise interference in the traditional peak-valley method. It introduces local extremum identification and Euclidean distance comparison of standard template vector, setting the upper limit of the allowable deviation to twice the standard deviation of each component of the standard template vector, automatically filtering out abnormal waveforms caused by contact erosion or mechanism jamming, and only calculating the ratio of three consecutive valleys for the valley sequence with normal matching. This greatly reduces the standard deviation of the attenuation rate calculation result compared with the direct peak-valley method, and the index difference extracted from the residual oscillation period number is directly related to the actual oscillation behavior of the waveform, providing a reliable basis for subsequent extinction feature determination.

[0013] Preferably, S4 specifically includes: Call the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal reference parameters. , Collect A-value, F-value, , Four sets of values ​​were organized into a standardized floating-point format to establish real-time and baseline parameter sets. Based on real-time and baseline parameter sets, A and Performing the subtraction operation yields the difference ΔA. For F and... The difference ΔF is obtained by performing a subtraction operation, which generates the deviation fluctuation difference.

[0014] Preferably, S5 specifically includes: Obtain ΔA and ΔF, and perform a weighted summation calculation on ΔA and ΔF based on the preset weight parameters w1 and w2 to obtain the comprehensive anomaly score S; Call the comprehensive anomaly score S and the weighted threshold. Compare S with The fluctuation characteristics exceed the limit, and the residual oscillation period number N and range are called. , and the oscillation disappearance time Δt and range [[Δ ,Δ Compare N with [ , ] and Δt and [Δ ,Δ The abnormal oscillation period and abnormal oscillation time are obtained, and combined, the abnormal extinction characteristic is determined. Based on the judgment of excessive fluctuation characteristics and abnormal extinction characteristics, the logical AND operation of the two is performed, and the result of the operation is output as normal or abnormal with an abnormal code as the automatic detection result of the low-voltage intelligent measuring switch.

[0015] Compared with existing technologies, the advantages of this invention are as follows: It acquires current waveforms under both normal debugging and disconnection conditions using a current sensor, and extracts the zero-crossing time sequence and time interval sequence based on zero-crossing detection. It quantifies fluctuation amplitude using a variance formula and obtains fluctuation frequency by statistically counting the number of incremental sign flips using a sign function, characterizing the time interval fluctuation characteristics from both amplitude and frequency dimensions. It uses a waveform envelope extraction algorithm to extract window waveforms and calculate attenuation rate, traversing the attenuation rate sequence to extract the number of residual oscillation cycles and oscillation disappearance time, quantifying the transient characteristics after switching action from multiple perspectives, including duration and number of oscillations. It performs difference calculations on fluctuation amplitude and frequency with pre-stored normal baseline parameters, weights them with specific weights to obtain a comprehensive anomaly score, and compares this score with a threshold to determine if fluctuation characteristics exceed limits. Simultaneously, it compares the number of residual oscillation cycles and oscillation disappearance time with the normal debugging range to determine if the extinguishing characteristics are abnormal. It combines the two types of judgments to output normal or abnormal, along with an anomaly code. This scheme achieves structured judgment criteria through multi-feature extraction and hierarchical comparison, improving the accuracy and traceability of switch state identification under complex operating conditions. Attached Figure Description

[0016] The accompanying drawings, which are provided to further illustrate the invention and constitute a part of this invention, are illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention.

[0017] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0018] The following will describe in detail the implementation of the present invention with reference to the accompanying drawings and embodiments, so that the process of how the present invention uses technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly.

[0019] In the description of this invention, it should be noted that the directional terms such as "center", "lateral", "longitudinal", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", and "counterclockwise" indicate the orientation and positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. They should not be construed as limiting the specific protection scope of this invention.

[0020] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features. Thus, the use of "first" and "second" to define a feature may explicitly or implicitly include one or more of that feature, and in the description of this invention, "a number" means two or more, unless otherwise explicitly specified.

[0021] In this invention, unless otherwise explicitly specified and limited, the terms "assembly," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can also refer to a mechanical connection; they can refer to a direct connection or a connection through an intermediate medium; or they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0022] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0023] It should also be understood that the terminology used in this specification of embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of the invention. As used in this specification of embodiments of the invention and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] As shown in the accompanying drawings, an automatic detection method for a low-voltage intelligent measuring switch is provided, the method comprising the following steps: S1: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch open state are collected by the current sensor, and the zero-crossing point extraction is performed on the current waveform data based on the zero-crossing detection algorithm to obtain the zero-crossing point time sequence Z and the time interval sequence T. S2: The fluctuation amplitude A is obtained by calculating the fluctuation amplitude of the time interval sequence T using the variance formula, and the fluctuation frequency F is obtained by counting the number of incremental sign flips of the time interval sequence T using the sign function. S3: The waveform envelope extraction algorithm is used to truncate the current waveform data into a window and calculate the attenuation rate to obtain the attenuation rate sequence R. The residual oscillation period number N and the oscillation disappearance time Δt are extracted by traversing the attenuation rate sequence R. At the same time, the range is extracted based on the zero-crossing time sequence Z of the current waveform data under normal debugging conditions to obtain the normal debugging range. , ] and [Δ ,Δ ]; S4: Based on the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal baseline parameters , Perform the difference operation to obtain ΔA and ΔF; S5: Set weights w1=0.6 and w2=0.4, and calculate the comprehensive anomaly score S= w1×ΔA + w2×ΔF using the weighted summation formula. Set the weighted threshold. The comprehensive anomaly score S and A comparison is made to determine if the fluctuation characteristics exceed the standard, and the residual oscillation period N is compared with the normal adjustment range. , Compare the oscillation disappearance time Δt with the normal debugging range [Δt]. ,Δ The results are compared to obtain the extinguishing feature abnormality judgment. The combined fluctuation feature over-limit judgment and the extinguishing feature abnormality judgment are used to determine the result. The result is output as normal or abnormal, with an abnormal code attached, as the automatic detection result of the low-voltage intelligent measuring switch.

[0025] For example, S1 specifically includes: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch-off state are collected by a current sensor. First, a Hall effect closed-loop current sensor CHB-200A with a range of 0-200A and an output analog voltage of 0-5V is connected. The sampling frequency is set to 2kHz and the quantization bit is 12 bits. The analog signal is then converted from analog to digital to obtain a discrete voltage sequence. Next, amplitude normalization is performed on each sampling point. A reference voltage of 5V and a normalization coefficient of 1 / 5 are set. The original voltage value is multiplied by this coefficient to obtain the normalized value. For example, an original voltage of 3.6V is normalized to 0.72, and an original voltage of 1.8V is normalized to 0.36. All sampling point values ​​fall between 0 and 1 and are stored as an array of current waveform sampling values. Each waveform group has a length of 0.5 seconds and a total of 1000 points. This array is then called to iterate through the values ​​of adjacent sampling points. and k ranges from 1 to 999, and is compared with zero value one by one to determine... and or and In either case, if the condition is met, record the sampling point index number k and the corresponding time. For example, during normal debugging, the zero-crossing times are recorded as 0.25ms, 5.10ms, 10.05ms, 15.00ms, 19.90ms, 25.05ms, 30.00ms, 34.95ms, 40.10ms, and 45.00ms. During the switch-off state, the times are recorded as 0.40ms, 4.75ms, 9.85ms, 14.80ms, 19.95ms, 25.15ms, 30.05ms, 34.90ms, 40.00ms, and 44.85ms. Arranging all zero-crossing times in chronological order yields the zero-crossing time sequence Z. (Normal debugging state) Switch off state Calculate the time difference between two adjacent zero-crossing moments and record them sequentially. The normal state yields a time interval sequence. (Unit: ms) The switch is open when the state is obtained. Generate the zero-crossing time sequence Z and the time interval sequence T.

[0026] The current waveform sampling values ​​are iterated through, and the values ​​of adjacent two sampling points in each group are compared one by one with the zero value. The array of current waveform sampling values ​​in normal debugging state is then called. ( ), set from the first point Begin by comparing one after another. and The symbol, if and Then record the index k of that point and the corresponding time. For example in hour , Record index 1 corresponds to time 0ms, and the actual zero-crossing point is at 0.25ms (more accurate through linear interpolation). For simplicity, the detected time of 0.25ms is directly used (as mentioned above). After traversing all the points, the sequence of sign change point indices is obtained. The time corresponding to each index is the zero-crossing time value. For example, in the normal state, 9 zero-crossing time values ​​are obtained. The same operation is performed to obtain 9 time values ​​in the switch-off state, thus completing the acquisition of the zero-crossing time values.

[0027] Based on the zero-crossing time values, all zero-crossing points are arranged in chronological order to obtain the zero-crossing time sequence Z. The time difference between two adjacent zero-crossing times is calculated and recorded sequentially to obtain the time interval sequence T. Under normal conditions, the zero-crossing time sequence Z is already available; the difference between adjacent zero-crossing times is then calculated. , , , , , , , , ,generate (Unit: ms) Switch off state Generate the zero-crossing time sequence Z and the time interval sequence T.

[0028] For example, S2 specifically includes: To obtain the time interval sequence T, the mean of the sequence is generated by summing all elements in the sequence and dividing by the total number of elements. For the sequence T under normal debugging conditions, the number of elements is... cumulative sum Divide by 9 to get the mean. ms.

[0029] The fluctuation range A is obtained by subtracting the mean from each element of the time interval sequence T, summing all the squared differences, and then dividing by the number of elements. The formula used here is the variance formula. ,in This is the i-th element (in milliseconds) of the time interval sequence. For sequence length, ms is the mean obtained from the previous calculation, and the summation sign is... This indicates that the squares of the differences within the parentheses are calculated sequentially for i from 1 to 9 and then summed. The degree of zero-interval dispersion is measured by averaging the squared deviations of each time interval from the mean. For each element in T: the first term Squaring gives 0.01494, the second term. Squared, we get 0.00049. The third term - 0.0222 squared, we get 0.00049. The fourth term... Squaring gives 0.00521, the fifth term. Squaring the terms gives 0.03162. Squaring the sixth term (-0.0222) gives 0.00049. Squaring the seventh term (-0.0222) gives 0.00049. Squaring the eighth term (0.1778) gives 0.03162. Squaring the ninth term (-0.0722) gives 0.00521. Summing these values ​​gives... Divide by 9 and get (Unit: ms²). The advantage of this formula is that it quantifies the fluctuation amplitude of the zero-crossing interval into a single value through variance, facilitating subsequent comparison with a benchmark. This value A represents a fluctuation amplitude of approximately 0.01 ms² under normal conditions, which is consistent with the actual power frequency half-cycle fluctuation of 10 ms in the range of 0.1% to 1%.

[0030] The fluctuation frequency submodule calls the fluctuation amplitude A and the time interval sequence T to calculate the difference between adjacent elements and apply the sign function to extract the sign value. It compares whether adjacent sign values ​​are consistent, and accumulates the count for inconsistent cases to obtain the fluctuation frequency F. For the normal state T sequence, it calculates the adjacent difference. , , , , , , , , The sign function `sign` is used to iterate through the following sequences: 0.10→1, 0→0, -0.05→-1, 0.25→1, -0.20→-1, 0→0, 0.20→1, -0.25→-1, resulting in the sign sequence [1,0,-1,1,-1,0,1,-1]. Adjacent comparisons are performed: 1 if 1 and 0 are inconsistent, 2 if 0 and -1 are inconsistent, 3 if -1 and 1 are inconsistent, 4 if 1 and -1 are inconsistent, 5 if -1 and 0 are inconsistent, 6 if 0 and 1 are inconsistent, and 7 if 1 and -1 are inconsistent. This is repeated 7 times. .

[0031] For example, S3 specifically includes: The intelligent measurement switch uses its built-in ACS712 current sensor to sample 1000 points at 2 kHz for 0.5 s, and then normalizes the results. Obtain the sequence db4 wavelet basis pairs were selected. Perform a 4-level discrete wavelet transform to obtain the level 1 detail coefficients. Length 500, second layer Length 250, 3rd layer Length 125, 4th layer Length 63 and the fourth layer approximation coefficient Length 63, calculate the standard deviation of detail coefficients for each layer. Set the threshold multiple to 3 to obtain the threshold for each layer. ,Right now Perform a judgment on each detailed coefficient: After processing and The reconstructed sequence is obtained by inputting the inverse discrete wavelet transform. ,in , traversal Up to 999, make a judgment , The peak amplitude sequence is obtained. Corresponding time Valley amplitude sequence Corresponding time Piecewise linear interpolation is used to obtain the upper envelope curve for peak points and piecewise linear interpolation is used to obtain the lower envelope curve for valley points. The window duration is set. (Sampling interval) (20 points per window) Starting from window 1, slide sequentially to window 8 (covering) ), for the first Extract the upper envelope maximum value from each window. minimum value of lower envelope Substitute into the attenuation rate formula Window 1 The upper envelope peak value within the window is (Time 1 ms), lower envelope valley value (Time 6 ms), substituting, we get Rounding to the nearest whole number Window 2 Upper envelope peak value (11 ms), lower envelope valley value (16 ms) Pick Window 3 Upper envelope (21 ms), lower envelope (26 ms) Pick Window 4 Upper envelope (31 ms), lower envelope (36 ms) Pick Window 5 Upper envelope (41 ms), lower envelope (46 ms) Pick Window 6 Upper envelope (51 ms), lower envelope (56 ms) Window 7 Upper envelope (61 ms), lower envelope (66 ms) Pick Window 8 Upper envelope (71 ms), lower envelope (76 ms) Arrange the above 8 values ​​according to the window number to generate an attenuation rate sequence. .

[0032] Based on the decay rate sequence R, starting from the first value in the sequence, the decay rate values ​​are sequentially compared with a set threshold. The indices of the first and last values ​​below the threshold are recorded, and the index difference is calculated as the residual oscillation period N. The index difference is multiplied by the sampling interval of the current waveform data to obtain the oscillation disappearance time Δt. Simultaneously, based on the original current waveform data, the zero-crossing time is detected, generating a zero-crossing time sequence Z. The time difference between adjacent zero-crossing points is calculated, and the maximum difference is selected as the upper limit and the minimum difference as the lower limit to obtain the normal debugging range. The decay rate threshold is then set. This threshold was determined by collecting the current waveforms of 10 normal low-voltage intelligent measuring switches in the open state and calculating their attenuation rate. The statistical results showed that the mean attenuation rate was 0.065 and the standard deviation was 0.008. The mean value minus one standard deviation was then used to calculate the threshold. Rounded to 0.06, this is used as the threshold for determining whether the oscillation has essentially disappeared. When the decay rate is below 0.06, it is considered that most of the oscillation energy has been dissipated. The comparison starts from index 1 of the R sequence: If 0.0353 is less than 0.06 compared to 0.06, then the first index below the threshold is recorded as 1. Records less than 0.06 are indexed in index 2. Records less than 0.06 are indexed in index 3. Records less than 0.06 are indexed in index 4. The comparison stops if the index value is greater than 0.06. The last index below the threshold is 4. The index difference is then calculated. The window duration is set to 10ms, meaning each window corresponds to a 10ms time span. N multiplied by the window duration gives the oscillation disappearance time. Simultaneously, based on the original current waveform data, the zero-crossing time sequence Z is generated by detecting the zero-crossing time under normal debugging conditions. (Unit: ms) Calculate the time difference between adjacent zero-crossing points. The minimum difference of 4.85ms was selected as the lower limit for the number of cycles, and the maximum difference of 5.15ms was selected as the upper limit for the number of cycles. The time interval was converted to the number of cycles. The half-cycle corresponding to the normal power grid frequency of 50Hz is 10ms, while the detected zero-crossing interval is close to 5ms, which is actually half the wave time. The residual oscillation cycle number should be the total oscillation duration divided by a complete cycle of 20ms. Based on the decay rate sequence, the window duration is 10ms. Corresponding to 40ms, or 2 complete cycles, N itself is used as the cycle number. The range of N is obtained by statistically analyzing test data from normal equipment. After the normal equipment is disconnected, the residual oscillation cycle number N is concentrated between 2 and 5. Therefore, it is set as follows: , Δt corresponds to a window duration of 10ms multiplied by N, therefore the lower limit is... The upper limit is ,Right now , Generate debug range values and .

[0033] For example, S4 specifically includes: Call the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal reference parameters. , Collect A-values, F-values, , These four sets of values ​​are used to establish real-time and baseline parameter sets in a standardized floating-point format, obtained from the aforementioned data under normal debugging conditions. , Next, read the pre-stored normal baseline parameters from the system memory. To obtain the average fluctuation amplitude of multiple qualified devices under the same sampling conditions, current waveforms of 30 qualified low-voltage intelligent measuring switches were collected under normal commissioning conditions, with each switch recorded three times, for a total of 90 sets of data. The fluctuation amplitude of each set was calculated and then averaged to obtain the average value. , The average frequency of fluctuations was obtained using the same statistical method. Next, store these four values ​​as double-precision floating-point numbers in temporary variables, i.e. , , , Establish real-time and baseline parameter sets.

[0034] Based on real-time and benchmark parameter sets, A and Performing the subtraction operation yields the difference ΔA. For F and... Performing subtraction to obtain the difference ΔF generates the deviation fluctuation difference value, and then calling... and Perform subtraction operation , call and Perform subtraction operation Generate deviation fluctuation difference (unit: ms²) and (Unit: times)

[0035] For example, S5 specifically includes: The comprehensive anomaly score S is obtained by performing a weighted summation of ΔA and ΔF based on preset weight parameters w1 and w2. The formula for the comprehensive anomaly score is involved here. ,in For the deviation in fluctuation range, For fluctuation frequency deviation, and The weighting coefficients are obtained by fitting a logistic regression model to 100 known state samples (60 normal and 40 abnormal). This indicates that the deviation in fluctuation range contributes 60%. This indicates that the frequency fluctuation deviation contributes 40%, and the two types of deviations are of similar magnitude after normalization. Substituting into the calculation yields... This comprehensive score merges the deviations of two different units into a single scalar, which is then compared with a threshold. The advantage of this formula is that it unifies features of different units through weighted summation, providing a clear numerical criterion for anomaly detection.

[0036] Call the comprehensive anomaly score S and the weighted threshold. Compare S with The fluctuation characteristics exceed the limit, and the residual oscillation period number N and range are called. , and the oscillation disappearance time Δt and range [Δ ,Δ Compare N with [ , ] and Δt and [Δ ,Δ The combination of oscillation period anomaly determination and oscillation time anomaly determination yields extinction feature anomaly determination, weighted threshold. The setup involves collecting current waveforms from 50 normal low-voltage intelligent measuring switches under normal debugging conditions, calculating a comprehensive anomaly score for each group, obtaining a mean score of 0.30 and a standard deviation of 0.04, and taking the mean plus twice the standard deviation as the threshold. , call and If 0.4003372 is greater than 0.38, then the fluctuation characteristic is deemed to have exceeded the limit, and the residual oscillation period count is called. Compared to the normal range [2,5], if 4 is between 2 and 5, then the oscillation period is considered normal, and the oscillation period abnormality is considered false. The oscillation disappearance time is then called. Compared to the normal range [20, 50], 40 is between 20 and 50, so the oscillation time is considered normal, and any abnormal oscillation time is considered a false alarm. Combining normal oscillation period and normal oscillation time, the resulting extinction characteristic is considered abnormal and also a false alarm. In this result... This indicates that the real-time score exceeds the normal threshold, which directly leads to the fluctuation feature being judged as true and the extinguishing feature being judged as false. The logical result depends on the combination of the two.

[0037] Based on the judgment of excessive fluctuation characteristics and abnormal extinction characteristics, a logical AND operation is performed on both. The result is output as normal or abnormal, with an exception code. The judgment of excessive fluctuation characteristics is true, and the judgment of abnormal extinction characteristics is false. The logical AND operation (true and false) is false, so the judgment result is normal. The exception code is an empty string. The automatic detection result of the low-voltage intelligent measuring switch is normal, and the result is stored in the detection log, completing the entire automatic detection process.

[0038] Table 1. Statistics of Benchmark Parameters and Thresholds , The above description only illustrates the preferred embodiments of the present invention and should not be construed as limiting the scope of the claims. The present invention is not limited to the above embodiments, and variations in its specific structure are permitted. All modifications made within the scope of the independent claims of this invention are also within the scope of protection of this invention.

Claims

1. An automatic detection method for a low-voltage intelligent measuring switch, characterized in that, The method includes the following steps: S1: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch open state are collected by the current sensor, and the zero-crossing point extraction is performed on the current waveform data based on the zero-crossing detection algorithm to obtain the zero-crossing point time sequence Z and the time interval sequence T. S2: The fluctuation amplitude A is obtained by calculating the fluctuation amplitude of the time interval sequence T using the variance formula, and the fluctuation frequency F is obtained by counting the number of incremental sign flips of the time interval sequence T using the sign function. S3: The waveform envelope extraction algorithm is used to truncate the current waveform data into a window and calculate the attenuation rate to obtain the attenuation rate sequence R. The residual oscillation period number N and the oscillation disappearance time Δt are extracted by traversing the attenuation rate sequence R. At the same time, the range is extracted based on the zero-crossing time sequence Z of the current waveform data under normal debugging conditions to obtain the normal debugging range. , ] and [Δ ,Δ ]; S4: Based on the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal baseline parameters , Perform the difference operation to obtain ΔA and ΔF; S5: Set weights w1=0.6 and w2=0.4, and calculate the comprehensive anomaly score S= w1×ΔA + w2×ΔF using the weighted summation formula. Set the weighted threshold. The comprehensive anomaly score S and A comparison is made to determine if the fluctuation characteristics exceed the standard, and the residual oscillation period N is compared with the normal adjustment range. , Compare the oscillation disappearance time Δt with the normal debugging range [Δt]. ,Δ The results are compared to obtain the extinguishing feature abnormality judgment. The combined fluctuation feature over-limit judgment and the extinguishing feature abnormality judgment are used to determine the result. The result is output as normal or abnormal, with an abnormal code attached, as the automatic detection result of the low-voltage intelligent measuring switch.

2. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, The time interval sequence T includes the spacing between adjacent zeros, the duration of half-waveform, and the frequency sampling period.

3. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, The fluctuation frequency F includes the historical symbol holding time, the flip rate, and the phase drift frequency.

4. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, The normal debugging range includes voltage harmonic tolerance, frequency disturbance boundary, and initial phase deviation; the parameters transmitted by the reference difference generation module are waveform distortion, amplitude deviation, and phase change; the automatic detection results include contact ablation degree encoding, mechanism jamming risk encoding, and insulation degradation level encoding.

5. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, S1 specifically includes: The current waveform data of the low-voltage intelligent measuring switch in normal debugging state and switch open state are collected. The analog-to-digital conversion and amplitude normalization processing are performed on the two sets of analog signals in sequence to convert the voltage amplitude value in each sampling period to a uniform value range and obtain the current waveform sampling value. Call the current waveform sampling value, traverse the values ​​of two adjacent sampling points in each group of sampling values ​​and compare them with the zero value one by one, detect whether the sign of the sampling point value changes from positive to negative or from negative to positive, record the sampling point index position and corresponding time for each sign change, and obtain the zero-crossing time value. Based on the zero-crossing time values, all zero-crossing points are arranged in chronological order to obtain the zero-crossing time sequence Z. The time difference between two adjacent zero-crossing times is calculated and recorded in sequence to obtain the time interval sequence T, thus generating the zero-crossing time sequence Z and the time interval sequence T.

6. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, S2 specifically includes: Get the time interval sequence T, sum up all elements in the sequence one by one and divide by the total number of elements to generate the sequence mean; Call the sequence mean and the time interval sequence T, subtract the mean from each element and calculate the square of the difference, sum all the squared differences and divide by the number of elements to get the fluctuation range A; The fluctuation frequency submodule calls the fluctuation amplitude A and the time interval sequence T, calculates the difference between adjacent elements and applies the sign function to extract the sign value, compares whether adjacent sign values ​​are consistent, accumulates the count for inconsistent cases, and obtains the fluctuation frequency F.

7. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, S3 specifically includes: Acquire current waveform data, apply waveform envelope extraction algorithm to find the peak and valley points of each cycle of the current waveform and connect them to form upper and lower envelopes, segment the envelope into windows and set the window duration, calculate the maximum and minimum values ​​of the envelope amplitude in each window, subtract the minimum value from the maximum value and divide by the maximum value to obtain the attenuation rate, and generate the attenuation rate sequence R; Based on the decay rate sequence R, starting from the first value in the sequence, the decay rate values ​​are sequentially compared with the set threshold. The indices of the first and last values ​​below the threshold are recorded. The index difference is calculated as the residual oscillation period N. The index difference is multiplied by the sampling interval of the current waveform data to obtain the oscillation disappearance time Δt. Simultaneously, based on the original current waveform data, the zero-crossing time is detected to generate a zero-crossing time sequence Z. The time difference between adjacent zero-crossing points is calculated, and the maximum difference is selected as the upper limit and the minimum difference as the lower limit to obtain the normal debugging range. The debugging range value is then generated. , ] and [Δ ,Δ ].

8. The automatic detection method for the low-voltage intelligent measuring switch according to claim 7, characterized in that, The generation of the decay rate sequence R specifically includes: The original current waveform time series is acquired by the built-in current sensor of the intelligent measurement switch. Multi-scale decomposition of wavelet transform is performed on the original current waveform time series. The absolute value of each detail coefficient is compared with the corresponding layer threshold. If the absolute value is greater than the threshold, the original coefficient value is retained. If the absolute value is not greater than the threshold, the coefficient is replaced with zero. The inverse wavelet transform is performed using all the processed layer detail coefficients and the fourth layer approximation coefficients to output the reconstructed waveform sequence. Based on the reconstructed waveform sequence, each data point is traversed, and the amplitude of the point is compared with the amplitudes of the two points before and after it. Peak points and valley points are marked. The upper and lower envelope curves are formed by linear interpolation. The peak amplitudes are extracted to form feature vectors. The local extremum identification method is called to obtain the standard template vector. The Euclidean distance between the current feature vector and the standard template vector is calculated. The Euclidean distance is compared with the upper limit of the allowable deviation. If the Euclidean distance is not greater than the upper limit of the allowable deviation, the lower envelope valley amplitude sequence corresponding to the waveform is retained. Based on the amplitude of the lower envelope valley value sequence, extract the amplitude of three consecutive valley values, calculate the ratio R1 obtained by dividing the amplitude of the first valley value by the amplitude of the second valley value, calculate the ratio R2 obtained by dividing the amplitude of the second valley value by the amplitude of the third valley value, and take the arithmetic mean of R1 and R2 as the current attenuation rate. Repeat the above calculation for each segment of three adjacent valley values ​​in the lower envelope valley value amplitude sequence to generate the attenuation rate sequence R.

9. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, S4 specifically includes: Call the fluctuation amplitude A, fluctuation frequency F, and pre-stored normal reference parameters. , Collect A-value, F-value, , Four sets of values ​​were organized into a standardized floating-point format to establish real-time and baseline parameter sets. Based on real-time and baseline parameter sets, A and Performing the subtraction operation yields the difference ΔA. For F and... The difference ΔF is obtained by performing a subtraction operation, which generates the deviation fluctuation difference.

10. The automatic detection method for the low-voltage intelligent measuring switch according to claim 1, characterized in that, S5 specifically includes: Obtain ΔA and ΔF, and perform a weighted summation calculation on ΔA and ΔF based on the preset weight parameters w1 and w2 to obtain the comprehensive anomaly score S; Call the comprehensive anomaly score S and the weighted threshold. Compare S with The fluctuation characteristics exceed the limit, and the residual oscillation period number N and range are called. , and the oscillation disappearance time Δt and range [[Δ ,Δ Compare N with [ , ] and Δt and [Δ ,Δ The abnormal oscillation period and abnormal oscillation time are obtained, and combined, the abnormal extinction characteristic is determined. Based on the judgment of excessive fluctuation characteristics and abnormal extinction characteristics, the logical AND operation of the two is performed, and the result of the operation is output as normal or abnormal with an abnormal code as the automatic detection result of the low-voltage intelligent measuring switch.

Images