Intelligent power distribution box multi-loop electricity larceny prevention real-time monitoring method

Abstract

The application relates to the technical field of electric variable measurement, and particularly discloses a real-time monitoring method for multi-loop electricity larceny prevention of an intelligent distribution box, which collects a power frequency voltage zero-crossing signal at an incoming line end, divides a power frequency period into phase intervals according to instantaneous amplitudes, synchronously latches voltage, current and reflection coefficient amplitude-frequency responses in each interval, and adds phase labels; recursive impedance parameter tracking is carried out by taking the phase labels as indexes, and weights are dynamically adjusted according to fluctuation amplitudes to obtain a dynamic background impedance baseline; the dynamic background impedance baseline is converted to a high-order cumulant domain, phase coupling characteristics in a diagonal slice which are stable with voltage fluctuation are extracted as intrinsic labels; an expected reflection coefficient curve is generated by recursively generating the intrinsic label state according to an instantaneous voltage amplitude, and a deviation residual is obtained by point-by-point comparison between the expected reflection coefficient curve and a measured value; the residual is subjected to discrete degree accumulation statistics in a sliding window, and is compared with a tolerance upper limit which is dynamically updated according to historical residuals, so that hardware tampering is determined and an alarm is output; the application can accurately identify hardware tampering under the condition of retaining lightning protection devices at the incoming line end.

Description

Technical Field

[0001] This invention relates to the field of electrical variable measurement technology, specifically to a real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes. Background Technology

[0002] In low-voltage power distribution systems, intelligent distribution boxes serve as the core node for power distribution and metering management. The safe operation of their incoming lines and multiple outgoing circuits directly impacts the fairness of trade settlement and the reliability of power supply between the power supply and consumption sides. Electricity theft, especially through hardware tampering methods such as shunt short circuits, undervoltage, insertion of unauthorized impedance components, and phase shifting using power electronic devices at the incoming line, leads to distorted power metering data, causing economic losses to power supply companies and proving difficult to detect through routine inspections. With the diversification of electricity consumption scenarios and the increasing sophistication of electricity theft methods, traditional methods of controlling theft, relying solely on the balance comparison of main meters and branch circuits or periodic on-site inspections, are no longer sufficient to meet the demands for real-time and precise location tracking.

[0003] Under the power distribution safety regulations that require parallel connection of nonlinear surge protection devices at the incoming line, the dynamic parasitic impedance drift generated by the device with the millisecond-level periodic change of the instantaneous amplitude of the power frequency voltage is highly superimposed in the frequency domain with the reflection coefficient spectrum characteristics caused by illegal short circuits or the insertion of unauthorized impedance components. This makes it impossible for existing impedance spectrum monitoring schemes based on fixed references or static thresholds to distinguish between the two, which in turn leads to the engineering contradiction of missing weak tampering behavior when the alarm threshold is forced to be raised to cover the dynamic drift range, or triggering false alarms when the grid voltage fluctuates normally. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes, so as to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes includes the following steps:

[0007] S1, acquire the zero-crossing signal of the power frequency voltage at the incoming end, divide each power frequency cycle into continuous phase intervals according to the instantaneous voltage amplitude, synchronously latch the instantaneous voltage value, instantaneous current value and amplitude-frequency response of the reflection coefficient at the incoming end in each phase interval, and add a phase tag to each set of latched data;

[0008] S2, using the phase label as an index, performs recursive impedance parameter tracking on the amplitude-frequency response of the reflection coefficient in each interval. During the tracking process, the recursive weight is dynamically adjusted according to the fluctuation amplitude of the parameters in each phase interval to obtain the equivalent impedance sequence that changes synchronously with the voltage phase, which serves as the dynamic background impedance baseline.

[0009] S3 transforms the dynamic background impedance baseline to the higher-order cumulative domain, extracts the phase coupling characteristics in the diagonal slice of its cumulative, and uses the part of the phase coupling characteristics that remains stable with voltage fluctuations as the intrinsic marker of nonlinear parasitic coupling.

[0010] S4. Based on the current instantaneous voltage amplitude, the intrinsic marker is recursively analyzed to generate the expected reflection coefficient curve after the current dynamic compensation. The measured reflection coefficient at the real-time incoming line is compared with the expected reflection coefficient curve point by point to obtain the deviation residual.

[0011] S5 performs cumulative statistics on the dispersion of the deviation residual within a sliding window, compares the statistical results with the tolerance upper limit dynamically updated by the historical residuals, and determines that there is hardware tampering at the incoming line when the statistical results continue to exceed the tolerance upper limit, and outputs the corresponding circuit alarm.

[0012] As a further aspect of the present invention: the addition of a phase tag to each group of latched data specifically includes:

[0013] Capture the continuous zero-crossing edges of the incoming power frequency voltage, take the time between adjacent zero-crossing edges as one power frequency cycle, divide the power frequency cycle into each phase interval according to a preset number of equal divisions, and assign a unique corresponding phase label to each phase interval.

[0014] At the beginning of each phase interval, the instantaneous value sampling of the incoming line voltage, the instantaneous value sampling of the current, and the amplitude-frequency response latching of the reflection coefficient are triggered simultaneously, and the three are associated as a single synchronous record under the same phase tag;

[0015] The instantaneous voltage values, instantaneous current values, and amplitude-frequency response of the reflection coefficient under each phase label are arranged in chronological order to form a latch sequence indexed by phase interval. Each group of data in the latch sequence is uniquely identified by its corresponding phase label.

[0016] As a further aspect of the present invention: the calculation process of the dynamic background impedance baseline is as follows:

[0017] Using the latched sequence as input, the historical amplitude-frequency response of the reflection coefficient and the current real-time measured value are extracted according to the phase label. The rate of change of the amplitude of the difference between the two is calculated as an instantaneous measure of the fluctuation amplitude of the current parameter within the phase interval corresponding to the phase label.

[0018] The recursive weights under the phase interval are adjusted according to the magnitude of the instantaneous metric. When the instantaneous metric increases, the recursive weights are increased to accelerate the tracking of the current change. When the instantaneous metric decreases, the recursive weights are decreased to smooth the tracking of fluctuations. The adjusted recursive weights are applied to the weighted sum of the current measured value and the tracked value at the previous moment to obtain the reconstructed response value sequence under the phase label after dynamic weighting.

[0019] After classifying the reconstructed response value sequence by phase label, impedance parameter inversion calculation is performed to obtain the equivalent impedance value corresponding to each phase label. The equivalent impedance values ​​under all phase labels are arranged in order of phase label and output as an equivalent impedance sequence that changes synchronously with the voltage phase.

[0020] As a further aspect of the present invention: the calculation process for the equivalent impedance value is as follows:

[0021] Using phase labels as indices, the reconstructed response value sequence is split into amplitude-frequency response subsequences corresponding to each phase interval;

[0022] For each amplitude frequency response subsequence, the inflection point frequency with the most dramatic slope jump in its amplitude frequency curve is detected and recorded as the characteristic frequency under the phase label;

[0023] Each characteristic frequency of the current power frequency cycle is compared with the historical characteristic frequency of the same phase label of the previous power frequency cycle. The change in the real and imaginary parts of the equivalent impedance is determined according to the frequency offset direction and offset amount. The change is then superimposed on the historical equivalent impedance value to output the current equivalent impedance value.

[0024] As a further aspect of the present invention: the process of obtaining the intrinsic marker is as follows:

[0025] Construct continuous time series of impedance values ​​changing along the power frequency period in each phase interval according to the phase label, calculate the third-order cumulant of the continuous time series and take its diagonal slice;

[0026] Extract the position of each spectral peak and its corresponding amplitude ratio from the diagonal slice, and combine them into the coupling structure descriptor under each phase label;

[0027] The coupling structure descriptor of the current power frequency cycle is compared one by one with the historical coupling structure descriptor of the same phase label of the previous power frequency cycle. The descriptor with the difference between the two within the preset allowable range is retained, and all retained descriptors are used as intrinsic markers.

[0028] As a further aspect of the present invention: the calculation of the third-order cumulant of the continuous time series and the taking of its diagonal slices specifically includes:

[0029] The impedance values ​​of multiple power frequency cycles within each phase interval are arranged in chronological order according to the phase label, forming the impedance cycle sequence of each phase label.

[0030] For each impedance periodic sequence, the periodic sequence is multiplied point by point with its own first lag sequence and second lag sequence. All multiplication results are categorized and accumulated according to the first lag and the second lag respectively to obtain the third-order cumulative distribution of each phase label.

[0031] Extract all elements from each third-order cumulant distribution where the first lag and the second lag are equal, and combine them to output a diagonal slice of phase label.

[0032] As a further aspect of the present invention: the calculation process of the deviation residual is as follows:

[0033] Obtain the instantaneous voltage amplitude sequence of the current power frequency cycle, and weight the instantaneous voltage amplitude with the descriptive quantity of the corresponding phase label in the intrinsic label according to the phase label to obtain the voltage-weighted state reference quantity under each phase label;

[0034] Perform a differential operation between the current state reference value of the current phase tag and the historical state reference value of the same phase tag in the previous power frequency cycle. Based on the direction and amplitude of the differential result, correct the current state reference value and generate a corrected state prediction value.

[0035] The corrected state prediction values ​​of all phase labels are arranged in the order of phase labels and matched with the preset impedance-reflection coefficient mapping constraint. The expected reflection coefficient curve after dynamic compensation is output. The difference between the expected reflection coefficient curve and the real-time measured reflection coefficient at the incoming line is calculated point by point to obtain the deviation residual.

[0036] As a further aspect of the present invention: the generation of the corrected state prediction value specifically includes:

[0037] The differential result between the current state reference value of the phase tag and the historical state reference value of the same phase tag in the previous power frequency cycle is used as input to determine the positive or negative direction of the differential result, and at the same time read the absolute value of the differential result.

[0038] The correction polarity is determined based on the positive and negative directions. The correction is matched with a preset nonlinear correction coefficient table based on the absolute value. The correction step size corresponding to the current absolute value is retrieved from the nonlinear correction coefficient table. The correction polarity is applied to the correction step size to obtain the correction amount.

[0039] The correction value is added to the current phase tag's state reference value to obtain the corrected state prediction value, and the state prediction value is output to the correction step of the next phase tag as the historical state reference value.

[0040] As a further aspect of the present invention: the output corresponding to the loop alarm specifically includes:

[0041] The deviation residuals are filled into the sliding window in chronological order. The positive deviations and negative deviations within the window are summed separately according to their positive and negative polarities. The difference between the positive deviations and negative deviations is taken as the discrete cumulative statistic.

[0042] Read the discrete cumulative statistics from multiple historical windows, select the statistics with the smallest absolute value, and multiply their average by a preset amplification factor to obtain the current tolerance limit.

[0043] The absolute value of the discrete cumulative statistics of the current window is compared with the current tolerance limit. If the absolute value is greater than the current tolerance limit and the absolute value of the discrete cumulative statistics of the previous window is greater than the tolerance limit of the previous window, it is determined that there is hardware tampering at the incoming line and the corresponding loop alarm is output.

[0044] The beneficial effects of this invention are:

[0045] (1) This invention divides the power frequency cycle into continuous phase intervals according to the instantaneous voltage amplitude and adds phase labels. Within each phase interval, it independently tracks the dynamic background impedance baseline. Combined with the stability discrimination of phase coupling characteristics in the higher-order cumulative domain, it can effectively distinguish between the nonlinear parasitic effect of the surge protection device at the incoming end caused by normal fluctuations in the grid voltage and the impedance mutation caused by actual hardware tampering. This method does not require the removal or bypass of the mandatory lightning protection device. It can maintain the ability to continuously measure the impedance characteristics of the incoming end under the constraints of power distribution safety regulations, avoiding the problem of reference template failure caused by nonlinear drift of the protection device, thereby improving the reliability and adaptability of anti-theft monitoring under complex power grid conditions.

[0046] (2) This invention adopts a dual confirmation and judgment method that combines cumulative statistics of discreteness within a sliding window with dynamic updates of the tolerance upper limit of historical residuals. The difference between the positive and negative polarities of the deviation residuals after separation and accumulation is used as the discrete cumulative statistics, and the mean of the statistics with the smallest absolute value among the most recent sliding windows is multiplied by a preset amplification factor as the tolerance upper limit. This realizes the online adaptive adjustment of the alarm threshold to the field operating environment. This mechanism does not rely on a fixed threshold setting and can dynamically fluctuate according to the actual line background noise level. It can effectively reduce occasional false alarms of exceeding the threshold caused by load switching or power grid harmonic fluctuations. At the same time, when abnormal deviations persist, it quickly outputs the corresponding circuit alarm, providing maintenance personnel with accurate abnormal circuit location information. Attached Figure Description

[0047] The invention will now be further described with reference to the accompanying drawings.

[0048] Figure 1 This is a flowchart of the method of the present invention;

[0049] Figure 2 This is a flowchart of the calculation process for the dynamic background impedance baseline in this invention;

[0050] Figure 3 This is a flowchart of the process for obtaining intrinsic markers in this invention. Detailed Implementation

[0051] 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.

[0052] Please see Figure 1 As shown, this invention provides a real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes, comprising the following steps:

[0053] S1, acquire the zero-crossing signal of the power frequency voltage at the incoming end, divide each power frequency cycle into continuous phase intervals according to the instantaneous voltage amplitude, synchronously latch the instantaneous voltage value, instantaneous current value and amplitude-frequency response of the reflection coefficient at the incoming end in each phase interval, and add a phase tag to each set of latched data;

[0054] S2, using the phase label as an index, performs recursive impedance parameter tracking on the amplitude-frequency response of the reflection coefficient in each interval. During the tracking process, the recursive weight is dynamically adjusted according to the fluctuation amplitude of the parameters in each phase interval to obtain the equivalent impedance sequence that changes synchronously with the voltage phase, which serves as the dynamic background impedance baseline.

[0055] S3 transforms the dynamic background impedance baseline to the higher-order cumulative domain, extracts the phase coupling characteristics in the diagonal slice of its cumulative, and uses the part of the phase coupling characteristics that remains stable with voltage fluctuations as the intrinsic marker of nonlinear parasitic coupling.

[0056] S4. Based on the current instantaneous voltage amplitude, the intrinsic marker is recursively analyzed to generate the expected reflection coefficient curve after the current dynamic compensation. The measured reflection coefficient at the real-time incoming line is compared with the expected reflection coefficient curve point by point to obtain the deviation residual.

[0057] S5 performs cumulative statistics on the dispersion of the deviation residual within a sliding window, compares the statistical results with the tolerance upper limit dynamically updated by the historical residuals, and determines that there is hardware tampering at the incoming line when the statistical results continue to exceed the tolerance upper limit, and outputs the corresponding circuit alarm.

[0058] In S1, the zero-crossing signal of the incoming line voltage is acquired. Each power frequency cycle is divided into continuous phase intervals according to the instantaneous voltage amplitude. The instantaneous voltage value, instantaneous current value, and amplitude-frequency response of the reflection coefficient at the incoming line are synchronously latched within each phase interval. A phase tag is added to each set of latched data, specifically including:

[0059] Capture the continuous zero-crossing edges of the power frequency voltage at the input end. Set up voltage sampling points at the input end, and detect the zero-crossing points of the rising edge when the power frequency voltage waveform crosses from the negative half cycle to the positive half cycle and the zero-crossing points of the falling edge when it crosses from the positive half cycle to the negative half cycle by voltage comparison. The rising edge zero-crossing point and the falling edge zero-crossing point are collectively referred to as zero-crossing edges, and the time length between two adjacent zero-crossing edges is taken as a power frequency cycle.

[0060] The power frequency cycle is divided into phase intervals according to a preset number of equal divisions, and each phase interval is assigned a unique corresponding phase label.

[0061] The phase interval is divided as follows: within a power frequency cycle consisting of two adjacent zero-crossing edges, the first zero-crossing edge is taken as the starting boundary, and the time length between the starting boundary and the next zero-crossing edge is divided equally according to a preset number of equal parts to obtain continuous and non-overlapping time segments. Each equal-width time segment is a phase interval.

[0062] The preset number of equal divisions is set to a positive integer greater than or equal to 256, preferably 256, meaning that each power frequency cycle is evenly divided into 256 phase intervals. Each phase interval is assigned an incrementally increasing numerical number according to its chronological order, and this numerical number is the phase label corresponding to that phase interval.

[0063] At the beginning of each phase interval, the instantaneous value sampling of the incoming line voltage, the instantaneous value sampling of the current, and the amplitude-frequency response latching of the reflection coefficient are triggered simultaneously, and the three are associated as a single synchronous record under the same phase tag.

[0064] The method for determining the starting time of a phase interval is as follows: the starting boundary of the power frequency cycle is taken as the starting time of the first phase interval, and the starting time of each subsequent phase interval is the time reached after the starting time of the previous phase interval is superimposed with the time width of one phase interval.

[0065] At the beginning of each phase interval, a voltage sampling trigger pulse is sent to the voltage sampling channel, a current sampling trigger pulse is sent to the current sampling channel, and a latch trigger pulse is sent to the reflection coefficient measurement port. The three channels complete data acquisition at the same time and store the acquisition results in the same temporary storage address. This temporary storage address is associated with the current phase tag, forming a synchronous record under the same phase tag.

[0066] The sampling of instantaneous voltage values ​​is achieved through analog-to-digital conversion. The sampling frequency is matched with the phase interval width. Specifically, when the power frequency period is 20 milliseconds and the preset number of equal divisions is 256, the width of each phase interval is 0.078125 milliseconds, corresponding to a sampling frequency of 12.8 kHz. One instantaneous voltage value is collected at the beginning of each phase interval.

[0067] The sampling of instantaneous current values ​​is also achieved through analog-to-digital conversion, and is triggered simultaneously with the sampling of instantaneous voltage values, with the sampling frequency remaining the same as that for voltage sampling.

[0068] The latching operation of the reflection coefficient amplitude-frequency response is as follows: a wideband probe signal is injected at the input end via coupling. This probe signal is a sinusoidal sweep signal with a constant amplitude, a sweep range of 10 kHz to 1 MHz, and a sweep step of 1 kHz. At the beginning of each phase interval, the ratio of the reflected signal amplitude to the incident signal amplitude at the current injected frequency is latched simultaneously. This ratio is the reflection coefficient amplitude-frequency response value at the current frequency point. The reflection coefficient amplitude-frequency response value corresponding to this frequency point is recorded and included in the synchronous record of this phase interval.

[0069] The instantaneous voltage values, instantaneous current values, and amplitude-frequency response of the reflection coefficient under each phase label are arranged in chronological order to form a latch sequence indexed by phase interval.

[0070] The arrangement method is as follows: the numerical order of each phase tag within a power frequency cycle from smallest to largest is used as the basis for arrangement. The instantaneous voltage value, instantaneous current value and the full-band reflection coefficient amplitude-frequency response value latched in the phase interval under each phase tag are combined into a set of data. All sets of data are arranged in ascending order of the phase tag numerical order to form a latch sequence.

[0071] Each data set in the latch sequence is uniquely identified by its corresponding phase tag. Subsequent steps use the phase tag as an index to extract the synchronization record of the corresponding phase interval from the latch sequence.

[0072] Please see Figure 2 As shown, in S2, using the phase label as an index, the amplitude-frequency response of the reflection coefficient in each interval is recursively tracked for impedance parameters. During the tracking process, the recursive weights are dynamically adjusted according to the parameter fluctuation amplitude in each phase interval to obtain an equivalent impedance sequence that changes synchronously with the voltage phase, serving as a dynamic background impedance baseline. Specifically, this includes:

[0073] Using the latched sequence as input, the historical amplitude-frequency response of the reflection coefficient and the current real-time measured value are extracted according to the phase label. The rate of change of the amplitude of the difference between the two is calculated as the instantaneous measure of the fluctuation amplitude of the current parameter within the phase interval corresponding to the phase label.

[0074] The specific operation method is as follows: For each phase interval identified by a phase tag, the current measured value of the amplitude-frequency response of the reflection coefficient under that phase tag in the current power frequency cycle is read from the latch sequence, and the historical value of the amplitude-frequency response of the reflection coefficient under the same phase tag in the previous power frequency cycle is read from the historical storage area; the amplitude of the current measured value is subtracted from the amplitude of the historical value to obtain the amplitude difference; the absolute value of the amplitude difference is divided by the amplitude of the historical value to obtain the amplitude change rate, which is the instantaneous measure of the fluctuation amplitude of the current parameter.

[0075] The larger the instantaneous metric, the more significant the jump in the amplitude-frequency response of the reflection coefficient within the phase interval compared to the same moment in the previous power frequency cycle; the smaller the instantaneous metric, the more stable the reflection characteristics within the phase interval.

[0076] After obtaining the instantaneous metric corresponding to each phase label, it is stored in a temporary cache for use in the next recursive weight adjustment process.

[0077] The recursive weights under this phase interval are adjusted according to the magnitude of the instantaneous metric. When the instantaneous metric increases, the recursive weights are increased to accelerate the tracking of the current change. When the instantaneous metric decreases, the recursive weights are decreased to smooth the tracking of fluctuations. The adjusted recursive weights are applied to the weighted sum of the current measured value and the tracked value at the previous moment to obtain the dynamically weighted reconstructed response value sequence under this phase label.

[0078] The adjustment of the recursive weights is implemented in the following way:

[0079] Three instantaneous metric threshold intervals are defined, with the first interval being when the instantaneous metric is less than or equal to... The second interval is where the instantaneous metric is greater than and less than or equal to The third interval is where the instantaneous metric is greater than When the instantaneous measurement falls into the first interval, the recursive weight is set to... When the instantaneous measurement falls into the second interval, the recursive weight is set to... When the instantaneous measurement falls into the third interval, the recursive weight is set to... ;

[0080] The current measured value under the current phase label is multiplied by the recursive weight to obtain the weighted current value. The previous tracking value under the same phase label is multiplied by the difference between the recursive weight and the current value to obtain the weighted historical value. The weighted current value and the weighted historical value are added together to obtain the dynamically weighted reconstructed response value. The above calculation is performed one by one in each power frequency cycle according to the time sequence. The obtained reconstructed response values ​​are indexed by phase label and arranged in time sequence to obtain the reconstructed response value sequence.

[0081] The initial value setting method for the tracking value in the above dynamic weighting process is as follows: In the first power frequency cycle after the system is first powered on, since there is no tracking value at the previous moment, the measured amplitude-frequency response of the reflection coefficient of each phase label in the first power frequency cycle is directly used as the initial value of the tracking value at the previous moment under that phase label, and updated cycle by cycle according to the aforementioned weighted accumulation method from the second power frequency cycle.

[0082] After classifying the reconstructed response value sequence by phase label, impedance parameter inversion calculation is performed to obtain the equivalent impedance value corresponding to each phase label. The equivalent impedance values ​​under all phase labels are arranged in order of phase label and output as an equivalent impedance sequence that changes synchronously with the voltage phase.

[0083] The specific implementation of the inversion calculation is as follows: using the phase label as an index, extract all reconstructed response values ​​belonging to the same phase label from the reconstructed response value sequence, arrange them in chronological order according to the power frequency cycle, and form the amplitude-frequency response subsequence under that phase label; for each amplitude-frequency response subsequence, detect the inflection point frequency of the slope jump in its amplitude-frequency curve and record it as the characteristic frequency under the current power frequency cycle of that phase label; compare each characteristic frequency of the current power frequency cycle with the historical characteristic frequency of the same phase label in the previous power frequency cycle, that is, each phase label has only one characteristic frequency value, determine the change in the real part and imaginary part of the equivalent impedance according to the frequency offset direction and offset amount, and superimpose the change amount onto the historical equivalent impedance value to output the current equivalent impedance value.

[0084] It should be noted that for each phase tag, a characteristic frequency is extracted in each power frequency cycle. Therefore, each phase tag corresponds to a characteristic frequency value in each cycle, and the characteristic frequency sequence of the phase tag can be formed in different cycles according to the time sequence.

[0085] The detection process for the inflection point frequency where the slope of the amplitude-frequency curve changes most sharply is as follows:

[0086] For each amplitude-frequency response subsequence, read the amplitude values ​​of two adjacent frequency points in ascending order of frequency, calculate the amplitude value of the next frequency point minus the amplitude value of the previous frequency point divided by the frequency interval, and obtain the forward slope value at each frequency point.

[0087] Then, the slope change rate at each frequency point is obtained by performing a difference calculation on two adjacent forward slope values.

[0088] The frequency point at which the absolute value of the rate of change of the slope reaches its maximum value is taken as the inflection point frequency of the most dramatic slope jump in the amplitude-frequency response subsequence.

[0089] The method for determining the frequency offset direction and offset amount is as follows: each characteristic frequency of the current power frequency cycle is recorded as the current characteristic frequency, and the historical characteristic frequency of the same phase label of the previous power frequency cycle is recorded as the historical characteristic frequency; the current characteristic frequency is calculated by subtracting the historical characteristic frequency from the current characteristic frequency to obtain the frequency offset amount. If the frequency offset amount is positive, the offset direction is positive; if it is negative, the offset direction is negative; if it is zero, the offset direction remains unchanged.

[0090] Let the current characteristic frequency be Historical characteristic frequency is Frequency offset .

[0091] Change in real part Based on frequency offset ratio ,make ,in This represents the real part of the historical equivalent impedance value. Offset direction sign (positive) negative That is, when the characteristic frequency shifts positively, the real part increases; when it shifts negatively, the real part decreases.

[0092] Change in the imaginary part The polarity is determined based on the offset direction, and the amplitude is taken as follows: The absolute value of, where This represents the quality factor of the amplitude-frequency response subsequence under this phase label at its historical characteristic frequencies. This represents the absolute value of the imaginary part of the historical equivalent impedance. If the frequency shifts positively, the imaginary part changes in the direction of increasing inductive reactance (i.e., the positive value of the imaginary part increases); if the frequency shifts negatively, the capacitive reactance increases (the negative value of the imaginary part increases, i.e., the algebraic value of the imaginary part decreases). Specifically, let the change in the imaginary part... Then the current imaginary part .

[0093] The real and imaginary parts are superimposed and combined to form the current equivalent impedance value.

[0094] In the first power frequency cycle after the system is first powered on, there is no historical equivalent impedance value. At this time, the impedance value obtained by inverting the latched data is directly used as the initial historical equivalent impedance value of the phase tag, and no superposition operation is performed.

[0095] The quality factor is calculated by dividing the amplitude at the historical characteristic frequency in the amplitude-frequency response subsequence by the frequency bandwidth corresponding to a 3 dB decrease in amplitude. This frequency bandwidth is the amplitude at the historical characteristic frequency. The difference between the upper and lower frequency limits corresponding to a multiple of 1.

[0096] When the frequency offset is in the positive direction, the imaginary part of the equivalent impedance shows an increasing trend of inductiveity; when the frequency offset is in the negative direction, the imaginary part of the equivalent impedance shows an increasing trend of capacitiveness.

[0097] The method for applying the change in the real and imaginary parts of the equivalent impedance to the historical equivalent impedance value is as follows:

[0098] Read the historical equivalent impedance value under the same phase tag in the previous power frequency cycle. This historical equivalent impedance value is a complex number, which includes a real part and an imaginary part.

[0099] The direction of change of the imaginary part is determined based on the direction of frequency offset. If it is a positive offset, the absolute value of the change of the imaginary part is added to the imaginary part of the historical equivalent impedance value. If it is a negative offset, the absolute value of the change of the imaginary part is subtracted from the imaginary part of the historical equivalent impedance value.

[0100] The change in the real part of the equivalent impedance is determined as follows: the absolute value of the frequency offset is calculated and divided by the absolute value of the historical characteristic frequency to obtain the frequency offset ratio. The real part of the historical equivalent impedance value is multiplied by the frequency offset ratio to obtain the change in the real part. Then, the change in the real part is added to the real part of the historical equivalent impedance value.

[0101] The real and imaginary parts obtained from the above calculations are combined to output the current equivalent impedance value.

[0102] The equivalent impedance values ​​output to all phase tags are arranged in ascending order of phase tag size, forming an equivalent impedance sequence that changes synchronously with the voltage phase, and are output as the dynamic background impedance baseline to the next step.

[0103] Please see Figure 3 As shown, in S3, the dynamic background impedance baseline is transformed to a higher-order cumulant domain, and the phase coupling characteristics in its cumulant diagonal slice are extracted. The portion of the phase coupling characteristics that remains stable with voltage fluctuations is used as the intrinsic marker of nonlinear parasitic coupling, specifically including:

[0104] Construct continuous time series of impedance values ​​changing along the power frequency cycle within each phase interval according to the phase label.

[0105] The equivalent impedance sequence is used as input. This equivalent impedance sequence contains the equivalent impedance values ​​corresponding to each phase label. The equivalent impedance values ​​are arranged in ascending order of the phase labels and stored cycle by cycle according to the time sequence of the power frequency cycle.

[0106] For each phase tag, the equivalent impedance value of the phase tag under each power frequency cycle is extracted from the equivalent impedance sequence. The extracted equivalent impedance values ​​are arranged in the order of arrival of the power frequency cycle to form a continuous time series under the phase tag.

[0107] In this continuous time series, each element corresponds to the equivalent impedance value of that phase label within one power frequency cycle. The index order of the sequence corresponds to the chronological order of the power frequency cycles; that is, the first element is the equivalent impedance value of the first power frequency cycle, the second element is the equivalent impedance value of the second power frequency cycle, and so on. For all phase labels, the above extraction and sorting operations are performed to obtain the continuous time series corresponding to each phase label. The length of each continuous time series is equal to the total number of power frequency cycles collected.

[0108] Calculate the third-order cumulants of each continuous time series and take its diagonal slice.

[0109] The calculation method for third-order cumulants is as follows:

[0110] For each continuous time series under a phase label, the series is regarded as a one-dimensional numerical sequence arranged by time index, and the values ​​in the series are the equivalent impedance values.

[0111] According to the definition of third-order cumulants, the statistical average of the product of the values ​​at any three time points in the sequence is calculated. The interval between these three time points is called the first lag and the second lag, respectively.

[0112] The specific operation is as follows: take each value in the sequence as the starting point, multiply the value at the starting point by the value at the position corresponding to the starting point after shifting backward by the first lag, and then multiply by the value at the position corresponding to the starting point after shifting backward by the second lag to obtain the product result; sum up the product results of the same first lag and the same second lag under all starting points and divide by the total number of starting points to obtain the cumulative value under the combination of the first lag and the second lag.

[0113] The values ​​of the first lag and the second lag are determined as follows: the first lag is any integer value from zero to the maximum lag, and the second lag is any integer value from zero to the maximum lag, where the maximum lag is equal to the length of the continuous time series minus one.

[0114] The above process iterates through all combinations of first and second lags to obtain the third-order cumulant distribution of the continuous time series. This third-order cumulant distribution is an array of cumulant values ​​with the first and second lags as two independent variables.

[0115] The specific calculation method for the third-order cumulant is expressed by the following formula:

[0116] ;

[0117] in, Indicates the first lag is And the second lag is The third-order cumulative value at time, Indicates the length of a continuous time series. Indicates the index number representing the numerical position in the sequence. Indicates the th continuous time series The impedance value at each location, Indicates the th continuous time series The impedance value at each location, Indicates the th continuous time series The impedance value at each location, This represents the first lag and its value is an integer ranging from zero to the maximum lag. This represents the second lag, and its value is an integer ranging from zero to the maximum lag. Indicates taking and The larger value in the range.

[0118] Extract a diagonal slice from the third-order cumulant distribution. This diagonal slice is a subset of all elements in the third-order cumulant distribution whose first lag is equal to the second lag.

[0119] The specific operation is as follows: In the array of the above third-order cumulant distributions, find all positions where the value of the first lag is equal to the value of the second lag, i.e. All the corresponding cumulative values ​​are arranged in order from zero to the maximum lag, and combined to form a diagonal slice of the continuous time series.

[0120] The positions of each spectral peak and their corresponding amplitude ratios are extracted from the diagonal slices and combined to form the coupling structure descriptors under each phase label.

[0121] After the diagonal slice is output, it is a sequence of values ​​arranged in order of the first lag, and each value in the sequence is a cumulative value.

[0122] The spectral peak detection of the numerical sequence is performed as follows: starting from the first value, each value is compared with its adjacent preceding and following values. If the current value is greater than both the preceding and following values, the position of the current value is marked as a spectral peak position. The cumulative value at all spectral peak positions is recorded as the spectral peak amplitude.

[0123] For each spectral peak position, the amplitude ratio corresponding to that peak position is calculated as follows: the cumulative value at that peak position is divided by the arithmetic mean of all cumulative values ​​in the diagonal slice to obtain the amplitude ratio of that peak position.

[0124] All spectral peak positions are arranged in ascending order of their indices, and each spectral peak position is paired with its corresponding amplitude ratio and recorded to form the coupling structure descriptor under that phase label.

[0125] The coupling structure descriptor of the current power frequency cycle is compared one by one with the historical coupling structure descriptor of the same phase label of the previous power frequency cycle. The descriptor with the difference between the two within the preset allowable range is retained, and all retained descriptors are used as intrinsic markers.

[0126] The comparison method is as follows:

[0127] For each phase tag, take the position of each spectral peak and its amplitude ratio in the coupled structure descriptor calculated under the current power frequency cycle, and compare it one by one with the amplitude ratio at the corresponding spectral peak position in the historical coupled structure descriptor of the same phase tag in the previous power frequency cycle, and calculate the absolute difference between the two amplitude ratios.

[0128] If the absolute difference is less than or equal to the upper limit of the preset allowable range, then the position of the spectral peak and its amplitude ratio under the current power frequency cycle are retained.

[0129] If the absolute difference is greater than the upper limit of the preset allowable range, then the position of the spectral peak and its amplitude ratio are discarded.

[0130] The upper limit of the preset allowable range is determined as follows: take the arithmetic mean of all amplitude ratios in the historical coupling structure descriptors of the same phase label in the previous power frequency cycle, and take 0.2 times the arithmetic mean as the upper limit of the preset allowable range.

[0131] All the retained spectral peak positions and their corresponding amplitude ratios are sorted by spectral peak position index and combined to output as the intrinsic markers under that phase label.

[0132] This intrinsic marker represents the stable coupling structure exhibited by the impedance value within each phase interval as a function of voltage fluctuation rather than external alteration, and serves as the input for state recursion in the next step.

[0133] The nonlinear parasitic coupling refers to the equivalent impedance coupling effect formed by the nonlinear surge protection device connected in parallel at the input terminal, whose parasitic parameters such as junction capacitance and leakage resistance change nonlinearly with the bias voltage during the periodic change of the instantaneous amplitude of the power frequency voltage. The impedance change generated by this effect has a definite correspondence with the voltage phase and repeats in each phase interval within the power frequency cycle, which is different from the impedance abrupt change caused by external hardware tampering.

[0134] In S4, the intrinsic marker is recursively analyzed based on the current instantaneous voltage amplitude to generate the expected reflection coefficient curve after dynamic compensation. The measured reflection coefficient at the real-time incoming line is then compared point by point with the expected reflection coefficient curve to obtain the deviation residual, which specifically includes:

[0135] Obtain the instantaneous voltage amplitude sequence of the current power frequency cycle, and weight the instantaneous voltage amplitude with the descriptive quantity of the corresponding phase label in the intrinsic label according to the phase label to obtain the voltage-weighted state reference quantity under each phase label.

[0136] The instantaneous voltage amplitude sequence is obtained as follows: Voltage sampling points are set at the input end, using the same sampling time points as the phase interval divisions in step S1, i.e., the start time of each phase interval. The instantaneous power frequency voltage value at that time is collected. The instantaneous voltage values ​​collected at the start time of each phase interval are arranged in ascending order of phase labels to form the instantaneous voltage amplitude sequence for the current power frequency cycle. The descriptive quantity corresponding to the phase label in the intrinsic label is the position of all spectral peaks retained under that phase label and their amplitude ratios.

[0137] The specific operation method of weighted association is as follows:

[0138] For each phase tag, the instantaneous voltage amplitude under that phase tag is multiplied by the amplitude ratio at the first spectral peak position in the intrinsic mark under that phase tag to obtain the first weighted component under that phase tag;

[0139] Multiply the instantaneous voltage amplitude under the phase label by the amplitude ratio at the second spectral peak position in the intrinsic mark under the phase label to obtain the second weighted component under the phase label;

[0140] This process continues until all spectral peak positions under the intrinsic markers of that phase label have been traversed;

[0141] All weighted components are arranged in ascending order of peak position index and combined to form the voltage-weighted state reference quantity under that phase label. If the intrinsic marker under a certain phase label is empty or contains only one peak position, the corresponding state reference quantity only contains the number of weighted components that actually exist.

[0142] The current state reference value of the phase tag is differentially calculated with the historical state reference value of the same phase tag in the previous power frequency cycle. The current state reference value is corrected based on the direction and amplitude of the differential result to generate the corrected state prediction value.

[0143] The difference operation is performed as follows:

[0144] For each weighted component in the state reference quantity under the current phase label, according to its corresponding spectral peak position index, find the historical weighted component at the same spectral peak position index in the historical state reference quantity of the same phase label in the previous power frequency cycle, and subtract the historical weighted component value from the current weighted component value to obtain the difference value at that spectral peak position.

[0145] The difference values ​​at all spectral peak positions are sorted by peak position index to form the difference result. This difference result contains information on positive and negative directions and the magnitude of the absolute value, which is used in subsequent correction steps.

[0146] The differential result between the current phase tag's state reference value and the historical state reference value of the same phase tag in the previous power frequency cycle is used as input to determine the positive or negative direction of the differential result, and at the same time, the absolute value of the differential result is read.

[0147] The method for determining positive and negative directions is as follows: for each difference value in the difference result, if the difference value is greater than zero, it is marked as a positive direction; if the difference value is less than zero, it is marked as a negative direction; and if the difference value is equal to zero, it is marked as a zero direction.

[0148] The absolute value is read as follows: for each difference value, take its absolute value. That is, if the difference value is negative, take its opposite; if the difference value is positive, take itself. This gives the absolute value of the difference.

[0149] The correction polarity is determined based on the positive and negative directions. The correction is matched with a preset nonlinear correction coefficient table based on the absolute value. The correction step size corresponding to the current absolute value is retrieved from the nonlinear correction coefficient table. The correction polarity is applied to the correction step size to obtain the correction amount.

[0150] It should be noted that each amplitude in the instantaneous voltage amplitude sequence is normalized to a value between 0 and 1 based on the rated voltage. Simultaneously, the amplitude ratio in the intrinsic label is obtained by dividing the cumulative value by the average of the cumulative values ​​of the diagonal slices, and under normal circumstances, its value also falls within the range of 0 to 1. Therefore, the absolute value of the weighted component obtained by multiplying the two, and the difference value calculated from it, will not exceed 1. Thus, the absolute value range of the difference value can be set to... .

[0151] The pre-defined nonlinear correction coefficient table is constructed as follows: The absolute value of the difference is used as the lookup index. This absolute value ranges from 0 to 1. This range is divided into 10 consecutive intervals: the first interval is for absolute values ​​greater than or equal to 0 and less than 0.1; the second interval is for absolute values ​​greater than or equal to 0.1 and less than 0.2; the third interval is for absolute values ​​greater than or equal to 0.2 and less than 0.3; the fourth interval is for absolute values ​​greater than or equal to 0.3 and less than 0.4; the fifth interval is for absolute values ​​greater than or equal to 0.4 and less than 0.5; the sixth interval is for absolute values ​​greater than or equal to 0.5 and less than 0.6; and the seventh interval is for absolute values ​​greater than or equal to... The absolute values ​​of the intervals are 0.6 and less than 0.7, the eighth interval is greater than or equal to 0.7 and less than 0.8, the ninth interval is greater than or equal to 0.8 and less than 0.9, and the tenth interval is greater than or equal to 0.9 and less than or equal to 1. The correction step size for each interval increases in ascending order of the interval number. The correction step size for the first interval is 0.02, for the second interval it is 0.04, for the third interval it is 0.07, for the fourth interval it is 0.10, for the fifth interval it is 0.14, for the sixth interval it is 0.18, for the seventh interval it is 0.23, for the eighth interval it is 0.28, for the ninth interval it is 0.34, and for the tenth interval it is 0.40.

[0152] The correction step size corresponding to each interval in the above nonlinear correction coefficient table is an empirical parameter preset by least-squares fitting and piecewise linearization based on the measured response characteristics of typical nonlinear surge protection devices at the incoming line under different voltage biases. In practical applications, this coefficient table can be calibrated and updated according to the model of the protection device in the specific distribution box, but in this embodiment, the fixed table is sufficient to cover common operating conditions.

[0153] When looking up a table, the correction step size corresponding to the interval is read based on the interval into which the absolute value of the current difference falls.

[0154] The method for determining the corrected polarity is as follows:

[0155] If the direction of the difference value is positive, the correction polarity is positive; if the direction of the difference value is negative, the correction polarity is negative; if the direction is zero, the correction polarity is zero.

[0156] The specific method for applying the corrected polarity to the corrected step size is as follows:

[0157] If the correction polarity is positive, the correction step size is taken as a positive value as the correction amount; if the correction polarity is negative, the correction step size is taken as a negative value as the correction amount.

[0158] If the correction polarity is zero, then the correction amount is zero.

[0159] The correction value is added to the current phase tag's state reference value to obtain the corrected state prediction value, and this state prediction value is output to the correction step of the next phase tag as the historical state reference value.

[0160] The overlay method is as follows:

[0161] For each weighted component in the state reference quantity under the current phase label, take the weighted component value and add the correction amount at the corresponding spectral peak position to obtain the corrected predicted component value at the spectral peak position.

[0162] The corrected predicted component values ​​at all spectral peak positions are sorted by spectral peak position index and then combined to form the corrected state prediction value under that phase label.

[0163] When processing the same phase tag in the next power frequency cycle, this predicted state value will be used as a historical state reference value from the previous power frequency cycle for subsequent differential calculations.

[0164] Arrange the corrected state prediction values ​​of all phase labels in the order of phase labels and match them with the preset impedance-reflection coefficient mapping constraints. Output the expected reflection coefficient curve after dynamic compensation. Calculate the point-by-point difference between the expected reflection coefficient curve and the real-time measured reflection coefficient at the incoming line to obtain the deviation residual.

[0165] The preset method for the impedance-reflection coefficient mapping constraint is as follows:

[0166] For each phase label, the conversion relationship between the equivalent impedance value and the reflection coefficient under that phase label is set as follows: the reflection coefficient is equal to the equivalent impedance value minus the standard impedance value, divided by the equivalent impedance value plus the standard impedance value, where the standard impedance value is 50 ohms.

[0167] The matching process is as follows: take each weighted component in the corrected state prediction value under each phase label as the impedance value, substitute them one by one into the above conversion relationship to obtain the expected reflection coefficient value at each frequency point under each phase label, arrange the expected reflection coefficient values ​​at all frequency points under all phase labels in the order of phase label and frequency to obtain the expected reflection coefficient curve.

[0168] The method for obtaining the real-time measured reflection coefficient at the incoming line end is as follows: at the incoming line end, the ratio of the reflected signal amplitude to the incident signal amplitude at each current frequency point is continuously measured through coupling to obtain the measured reflection coefficient curve.

[0169] The point-by-point difference calculation method is as follows: For each frequency point on the expected reflection coefficient curve, find the measured value at the same frequency point on the actual reflection coefficient curve, subtract the expected value from the measured value to obtain the difference at that frequency point; arrange the differences at all frequency points in frequency order and output them as the deviation residual.

[0170] In S5, the dispersion of the deviation residual is accumulated and statistically analyzed within a sliding window. The statistical result is compared with the tolerance upper limit dynamically updated from historical residuals. When the statistical result continuously exceeds the tolerance upper limit, it is determined that there is hardware tampering at the incoming line, and the corresponding loop alarm is output, specifically including:

[0171] The deviation residuals are filled into the sliding window in chronological order. The positive and negative deviation residuals within the window are summed separately according to their positive and negative polarities. The difference between the sum of positive and negative deviations is taken as the discrete cumulative statistic.

[0172] The deviation residual comes from the output of step S4, which is a sequence of differences arranged in order of frequency points. Each difference in the sequence is a real number, which can be positive, negative or zero, corresponding to the case where the measured reflection coefficient at the input end is higher than, lower than or equal to the expected reflection coefficient.

[0173] The sliding window is a storage area that stores the deviation residuals in the order of the power frequency cycle. This storage area can hold a fixed number of deviation residual data within consecutive power frequency cycles. The window width is set to include the deviation residual data of the current power frequency cycle and the 9 most recent power frequency cycles, for a total of 10 power frequency cycles.

[0174] Each time the deviation residual data of a new power frequency cycle arrives, the data is filled into the position of the first stored deviation residual data in the sliding window to cover it, so that the sliding window always contains the deviation residual data of the most recent 10 power frequency cycles.

[0175] For each deviation residual value within the current sliding window, determine its positive or negative value:

[0176] If the residual value of the deviation is greater than zero, it is marked as a positive deviation, and its value is added to the sum of positive deviations.

[0177] If the residual value of the deviation is less than zero, it is marked as a negative deviation, and its absolute value is accumulated to the total negative deviation.

[0178] If the deviation residual value is equal to zero, it is not accumulated;

[0179] After traversing all deviation residuals within the sliding window, the sum of positive deviations and the sum of negative deviations are obtained. The difference between the sum of positive deviations and the sum of negative deviations is used as the discrete cumulative statistic of the current sliding window.

[0180] Read the discrete cumulative statistics from multiple historical windows, select the statistics with the smallest absolute value, and multiply their average by a preset amplification factor to obtain the current tolerance limit.

[0181] The discrete cumulative statistics of multiple historical windows are stored in a separate historical statistics storage area. This storage area saves the discrete cumulative statistics calculated for the 20 most recent sliding windows before the current sliding window, and each statistic is a real number.

[0182] Once the historical statistics storage area is full with 20 statistics, each time a new discrete cumulative statistic is calculated, the new statistic is stored in the storage area and the statistic stored earliest in the storage area is removed, so that the storage area always retains the statistics of the most recent 20 sliding windows.

[0183] Select the five statistics with the smallest absolute values ​​from the above 20 statistics, add their absolute values ​​together and divide by 5 to obtain the arithmetic mean of the absolute values ​​of the five statistics; multiply the arithmetic mean by the preset magnification factor of 1.5, and the product is used as the current tolerance limit.

[0184] The absolute value of the discrete cumulative statistics of the current window is compared with the current tolerance limit. If the absolute value is greater than the current tolerance limit and the absolute value of the discrete cumulative statistics of the previous window is greater than the tolerance limit of the previous window, it is determined that there is hardware tampering at the incoming line and the corresponding loop alarm is output.

[0185] The calculation method for the previous window tolerance limit is the same as that for the current tolerance limit. That is, when calculating the previous window tolerance limit, the previous window is used as the current window, and the arithmetic mean of the five statistics with the smallest absolute values ​​among the discrete cumulative statistics of the 20 most recent windows before the previous window is multiplied by 1.5.

[0186] The comparison operation is as follows:

[0187] Read the absolute value of the current window discrete cumulative statistic, read the current tolerance limit, and compare the absolute value of the current window discrete cumulative statistic with the current tolerance limit;

[0188] Simultaneously, the absolute value of the discrete cumulative statistic of the previous window and the upper tolerance limit of the previous window are read and compared.

[0189] If the absolute value of the current window's discrete cumulative statistic is greater than the current tolerance limit and the absolute value of the previous window's discrete cumulative statistic is greater than the previous window's tolerance limit, then the double confirmation condition is met, and it is determined that there is hardware tampering at the incoming line.

[0190] If the above conditions are not met, the current deviation residual is determined to be within the normal fluctuation range.

[0191] If hardware tampering is detected, an alarm signal is output. The alarm signal contains the circuit number in the distribution box corresponding to the incoming line terminal that was detected as having been tampered with.

[0192] The circuit number is determined as follows: based on the circuit number of the distribution box corresponding to the selected incoming terminal when obtaining the measured reflection coefficient in step S4, this number is output as the circuit identifier in the alarm signal.

[0193] The alarm signal is output to the local alarm indication port of the distribution box in the form of a switch level, and is simultaneously uploaded to the remote monitoring terminal in the form of a data frame through the communication interface.

[0194] The working principle of this invention is as follows: The zero-crossing signal of the incoming power frequency voltage is acquired, and the power frequency cycle is divided into continuous phase intervals according to its instantaneous amplitude. The instantaneous voltage value, instantaneous current value, and amplitude-frequency response of the reflection coefficient within each phase interval are synchronously latched and labeled with a phase tag. Using this phase tag as an index, a recursive impedance tracking method is used to track the reflection coefficient of each interval, and the recursive weight is dynamically adjusted according to the fluctuation amplitude to obtain an equivalent impedance sequence that changes synchronously with the voltage phase as a dynamic background impedance baseline. This dynamic background impedance baseline is converted to a higher-order cumulative domain, and the phase coupling characteristics in its diagonal slices are extracted, retaining the portion that remains stable with voltage fluctuations as an intrinsic marker of nonlinear parasitic coupling. Based on the current instantaneous voltage amplitude, the intrinsic marker is recursively analyzed to generate a dynamically compensated expected reflection coefficient curve. This curve is then compared point-by-point with the measured reflection coefficient to obtain the deviation residual. The dispersion of the deviation residual is accumulated and statistically analyzed within a sliding window and compared with a tolerance upper limit dynamically updated by historical residuals. When the statistical result continuously exceeds the tolerance upper limit, it is determined that there is hardware tampering at the incoming end, and a corresponding circuit alarm is output.

[0195] 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 real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes, characterized in that, Includes the following steps: S1, acquire the zero-crossing signal of the power frequency voltage at the incoming end, divide each power frequency cycle into continuous phase intervals according to the instantaneous voltage amplitude, synchronously latch the instantaneous voltage value, instantaneous current value and amplitude-frequency response of the reflection coefficient at the incoming end in each phase interval, and add a phase tag to each set of latched data; S2, using the phase label as an index, performs recursive impedance parameter tracking on the amplitude-frequency response of the reflection coefficient in each interval. During the tracking process, the recursive weight is dynamically adjusted according to the fluctuation amplitude of the parameters in each phase interval to obtain the equivalent impedance sequence that changes synchronously with the voltage phase, which serves as the dynamic background impedance baseline. S3 transforms the dynamic background impedance baseline to the higher-order cumulative domain, extracts the phase coupling characteristics in the diagonal slice of its cumulative, and uses the part of the phase coupling characteristics that remains stable with voltage fluctuations as the intrinsic marker of nonlinear parasitic coupling. S4. Based on the current instantaneous voltage amplitude, the intrinsic marker is recursively analyzed to generate the expected reflection coefficient curve after the current dynamic compensation. The measured reflection coefficient at the real-time incoming line is compared with the expected reflection coefficient curve point by point to obtain the deviation residual. S5 performs cumulative statistics on the dispersion of the deviation residual within a sliding window, compares the statistical results with the tolerance upper limit dynamically updated by the historical residuals, and determines that there is hardware tampering at the incoming line when the statistical results continue to exceed the tolerance upper limit, and outputs the corresponding circuit alarm.

2. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 1, characterized in that, The process of attaching a phase tag to each group of latched data specifically includes: Capture the continuous zero-crossing edges of the incoming power frequency voltage, take the time between adjacent zero-crossing edges as one power frequency cycle, divide the power frequency cycle into each phase interval according to a preset number of equal divisions, and assign a unique corresponding phase label to each phase interval. At the beginning of each phase interval, the instantaneous value sampling of the incoming line voltage, the instantaneous value sampling of the current, and the amplitude-frequency response latching of the reflection coefficient are triggered simultaneously, and the three are associated as a single synchronous record under the same phase tag; The instantaneous voltage values, instantaneous current values, and amplitude-frequency response of the reflection coefficient under each phase label are arranged in chronological order to form a latch sequence indexed by phase interval. Each group of data in the latch sequence is uniquely identified by its corresponding phase label.

3. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 1, characterized in that, The calculation process for the dynamic background impedance baseline is as follows: Using the latched sequence as input, the historical amplitude-frequency response of the reflection coefficient and the current real-time measured value are extracted according to the phase label. The rate of change of the amplitude of the difference between the two is calculated as an instantaneous measure of the fluctuation amplitude of the current parameter within the phase interval corresponding to the phase label. The recursive weights under the phase interval are adjusted according to the magnitude of the instantaneous metric. When the instantaneous metric increases, the recursive weights are increased to accelerate the tracking of the current change. When the instantaneous metric decreases, the recursive weights are decreased to smooth the tracking of fluctuations. The adjusted recursive weights are applied to the weighted sum of the current measured value and the tracked value at the previous moment to obtain the reconstructed response value sequence under the phase label after dynamic weighting. After classifying the reconstructed response value sequence by phase label, impedance parameter inversion calculation is performed to obtain the equivalent impedance value corresponding to each phase label. The equivalent impedance values ​​under all phase labels are arranged in order of phase label and output as an equivalent impedance sequence that changes synchronously with the voltage phase.

4. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 3, characterized in that, The calculation process for the equivalent impedance value is as follows: Using phase labels as indices, the reconstructed response value sequence is split into amplitude-frequency response subsequences corresponding to each phase interval; For each amplitude frequency response subsequence, the inflection point frequency with the most dramatic slope jump in its amplitude frequency curve is detected and recorded as the characteristic frequency under the phase label; Each characteristic frequency of the current power frequency cycle is compared with the historical characteristic frequency of the same phase label of the previous power frequency cycle. The change in the real and imaginary parts of the equivalent impedance is determined according to the frequency offset direction and offset amount. The change is then superimposed on the historical equivalent impedance value to output the current equivalent impedance value.

5. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 1, characterized in that, The process of obtaining the intrinsic marker is as follows: Construct continuous time series of impedance values ​​changing along the power frequency period in each phase interval according to the phase label, calculate the third-order cumulant of the continuous time series and take its diagonal slice; Extract the position of each spectral peak and its corresponding amplitude ratio from the diagonal slice, and combine them into the coupling structure descriptor under each phase label; The coupling structure descriptor of the current power frequency cycle is compared one by one with the historical coupling structure descriptor of the same phase label of the previous power frequency cycle. The descriptor with the difference between the two within the preset allowable range is retained, and all retained descriptors are used as intrinsic markers.

6. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 5, characterized in that, The calculation of the third-order cumulant of a continuous time series and the taking of its diagonal slices specifically includes: The impedance values ​​of multiple power frequency cycles within each phase interval are arranged in chronological order according to the phase label, forming the impedance cycle sequence of each phase label. For each impedance periodic sequence, the periodic sequence is multiplied point by point with its own first lag sequence and second lag sequence. All multiplication results are categorized and accumulated according to the first lag and the second lag respectively to obtain the third-order cumulative distribution of each phase label. Extract all elements from each third-order cumulant distribution where the first lag and the second lag are equal, and combine them to output a diagonal slice of phase label.

7. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 1, characterized in that, The calculation process for the deviation residual is as follows: Obtain the instantaneous voltage amplitude sequence of the current power frequency cycle, and weight the instantaneous voltage amplitude with the descriptive quantity of the corresponding phase label in the intrinsic label according to the phase label to obtain the voltage-weighted state reference quantity under each phase label; Perform a differential operation between the current state reference value of the current phase tag and the historical state reference value of the same phase tag in the previous power frequency cycle. Based on the direction and amplitude of the differential result, correct the current state reference value and generate a corrected state prediction value. The corrected state prediction values ​​of all phase labels are arranged in the order of phase labels and matched with the preset impedance-reflection coefficient mapping constraint. The expected reflection coefficient curve after dynamic compensation is output. The difference between the expected reflection coefficient curve and the real-time measured reflection coefficient at the incoming line is calculated point by point to obtain the deviation residual.

8. The method for real-time monitoring of multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 7, characterized in that, The generation of the corrected state prediction value specifically includes: The differential result between the current state reference value of the phase tag and the historical state reference value of the same phase tag in the previous power frequency cycle is used as input to determine the positive or negative direction of the differential result, and at the same time read the absolute value of the differential result. The correction polarity is determined based on the positive and negative directions. The correction is matched with a preset nonlinear correction coefficient table based on the absolute value. The correction step size corresponding to the current absolute value is retrieved from the nonlinear correction coefficient table. The correction polarity is applied to the correction step size to obtain the correction amount. The correction value is added to the current phase tag's state reference value to obtain the corrected state prediction value, and the state prediction value is output to the correction step of the next phase tag as the historical state reference value.

9. The real-time monitoring method for multi-circuit anti-theft electricity in intelligent distribution boxes according to claim 1, characterized in that, The output corresponding loop alarm specifically includes: The deviation residuals are filled into the sliding window in chronological order. The positive deviations and negative deviations within the window are summed separately according to their positive and negative polarities. The difference between the positive deviations and negative deviations is taken as the discrete cumulative statistic. Read the discrete cumulative statistics from multiple historical windows, select the statistics with the smallest absolute value, and multiply their average by a preset amplification factor to obtain the current tolerance limit. The absolute value of the discrete cumulative statistics of the current window is compared with the current tolerance limit. If the absolute value is greater than the current tolerance limit and the absolute value of the discrete cumulative statistics of the previous window is greater than the tolerance limit of the previous window, it is determined that there is hardware tampering at the incoming line and the corresponding loop alarm is output.

Images