A circuit abnormality identification method and system of a three-phase electric meter

Abstract

The application relates to the technical field of data recognition, in particular to a circuit abnormality recognition method and system of a three-phase electric meter, which comprises the following steps: Fourier transform is used to acquire a frequency spectrum sequence from a collected historical characteristic sequence; time domain differences and frequency domain differences are calculated according to corresponding historical normal sequences, abnormal characteristic sequences and frequency spectrum sequences of three phases under the same load state; and the weights of the time domain and the frequency domain are calculated through the time domain differences and the frequency domain differences; the single-phase abnormality degree of a target characteristic is calculated by using the time domain and the frequency domain weights of the target characteristic under each load state; the single-phase abnormality degree of each target characteristic on each phase is input into a control system, and circuit abnormality recognition of the three-phase electric meter is carried out according to a preset abnormality degree threshold. Therefore, the application solves the shortcomings of high false alarm rate of abnormality recognition caused by single time-frequency domain threshold and single time data ignoring global changes through the data recognition technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data recognition, and in particular to a circuit abnormality recognition method and system of a three-phase electric meter. BACKGROUND

[0002] Three-phase electric meters are widely used in industrial, commercial and residential places for real-time monitoring and recording of power consumption. Traditional methods mainly use voltage, current, power factor and other conventional parameters for circuit monitoring, relying on fixed threshold judgment, waveform analysis and simple rule engines, setting certain thresholds or tolerance ranges, and exceeding the set value is considered to be abnormal.

[0003] However, the rule engine based on a single time-frequency domain cannot adapt to dynamic load fluctuations, such as motor start-stop and photovoltaic power transients, resulting in a high false alarm rate. For example, the instantaneous current surge during motor start-up is often misjudged as overload, while gradual overload caused by device aging is ignored because the threshold is not triggered. Therefore, the existing technology for processing data recognition, i.e., abnormality recognition through single time-frequency domain feature threshold, has the disadvantage of high false alarm rate. SUMMARY

[0004] In order to solve the disadvantage of high false alarm rate of single time-frequency domain threshold and single time data ignoring global changes for abnormality recognition, the present application provides a circuit abnormality recognition method and system of a three-phase electric meter.

[0005] In the first aspect, the present application provides a circuit abnormality recognition method of a three-phase electric meter, which adopts the following technical solution:

[0006] A circuit abnormality recognition method of a three-phase electric meter, comprising the following steps:

[0007] Collecting historical normal signals, abnormal feature signals and to-be-recognized feature signals corresponding to each phase of the three-phase electric meter under different load conditions;

[0008] Using Fourier transform on the collected historical feature sequence to obtain a frequency spectrum sequence, calculating time domain differences and frequency domain differences based on the historical normal sequence, abnormal feature sequence and frequency spectrum sequence corresponding to the three phases under the same load condition, and calculating the weights of time domain and frequency domain through the time domain differences and the frequency domain differences;

[0009] Dividing the to-be-recognized feature sequence using a sliding window, calculating the time domain and frequency domain abnormality degrees based on the time-frequency domain features of the target window and the adjacent window, and calculating the single-phase abnormality degree of the target feature under each load condition using the time domain and frequency domain weights of the target feature;

[0010] Inputting the single-phase abnormality degree of each target feature on each phase into a control system, and performing circuit abnormality recognition of the three-phase electric meter according to a preset abnormality degree threshold.

[0011] The calculation of the single-phase abnormality degree comprises: using Fourier transform to obtain a frequency spectrum sequence corresponding to each window for each window divided by a sliding window, calculating a frequency domain abnormality degree based on the frequency spectrum sequences of the window at a previous time and adjacent windows, and calculating the single-phase abnormality degree of the target feature based on the time domain abnormality degree and the frequency domain abnormality degree of the target feature, to obtain the single-phase abnormality degree of each target feature on each phase.

[0012] Optionally, the voltage and current transformers are used to collect historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase electric meter under different load states.

[0013] Optionally, wavelet denoising is used to denoise the collected historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase electric meter under different load states.

[0014] Optionally, normalization is performed on the collected historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase electric meter under different load states to eliminate dimensional differences.

[0015] Optionally, the calculation of the weights of the time domain and the frequency domain comprises:

[0016] Fourier transform is used to obtain a frequency spectrum sequence corresponding to each feature sequence for the three-phase historical normal and abnormal feature sequences under different load states.

[0017] The time domain difference and the frequency domain difference of the target feature under each load state are calculated according to the three-phase historical normal and abnormal feature sequences of the target feature under different load states.

[0018] The weights of the time domain and the frequency domain are calculated based on the time domain difference and the frequency domain difference of the target feature under the same load state.

[0019] In a second aspect, the present application provides a circuit abnormality identification system of a three-phase electric meter, which adopts the following technical scheme:

[0020] A circuit abnormality identification system of a three-phase electric meter, which adopts the three-phase electric meter circuit abnormality identification method, comprises:

[0021] A signal collection module is configured to collect historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase electric meter under different load states.

[0022] The weight calculation module is configured to use Fourier transform on the collected historical characteristic sequence to obtain a frequency spectrum sequence, calculate time domain difference and frequency domain difference according to the historical normal sequence, the abnormal characteristic sequence and the frequency spectrum sequence corresponding to three phases under the same load state, and calculate the weight of the time domain and the frequency domain through the time domain difference and the frequency domain difference.

[0023] The single-phase abnormality degree calculation module is configured to divide the to-be-identified characteristic sequence using a sliding window, calculate time domain abnormality degree and frequency domain abnormality degree based on time-frequency domain features of a target window and adjacent windows, and calculate the single-phase abnormality degree of the target characteristic using the time domain weight and the frequency domain weight of the target characteristic under each load state.

[0024] The abnormality recognition module is configured to input the single-phase abnormality degree of each target characteristic on each phase into a control system, and perform circuit abnormality recognition of the three-phase ammeter according to a preset abnormality degree threshold.

[0025] The calculation of the single-phase abnormality degree includes: using Fourier transform on each window divided by the sliding window to obtain a frequency spectrum sequence corresponding to each window, calculating frequency domain abnormality degree based on the frequency spectrum sequence of the window at the previous moment and adjacent windows, and calculating the single-phase abnormality degree of the target characteristic based on the time domain abnormality degree and the frequency domain abnormality degree of the target characteristic, to obtain the single-phase abnormality degree of each target characteristic on each phase.

[0026] The application has the following technical effects: Fourier transform is used on the collected historical characteristic sequence to obtain a frequency spectrum sequence, time domain difference and frequency domain difference are calculated according to the historical normal sequence, the abnormal characteristic sequence and the frequency spectrum sequence corresponding to three phases under the same load state, and the weight of the time domain and the frequency domain is calculated through the time domain difference and the frequency domain difference. The to-be-identified characteristic sequence is divided using a sliding window, time domain abnormality degree and frequency domain abnormality degree are calculated based on time-frequency domain features of a target window and adjacent windows, and the single-phase abnormality degree of the characteristic is calculated using the time domain weight and the frequency domain weight of the characteristic under each load state. Time-frequency domain features are fused through adaptive weight to avoid high false alarm rate of a single time-frequency domain, and time series analysis is performed to avoid false alarm caused by instantaneous change of a single moment feature. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 is a flowchart of steps S1-S4 in the circuit abnormality recognition method of the three-phase ammeter. DETAILED DESCRIPTION

[0028] The embodiment of the application discloses a circuit abnormality recognition method of a three-phase ammeter, which refers to Figure 1 and includes the following steps:

[0029] S1: Collect historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals corresponding to each phase of a three-phase ammeter under different load states.

[0030] It should be noted that the specific scenario to which the present application is directed can be: real-time identification of circuit abnormalities of a three-phase electric meter through time-frequency domain fusion of the time-frequency domain differences and time-frequency domain weights of the target window and adjacent windows at the current moment.

[0031] In an embodiment of the present application, a voltage transformer is used to collect historical normal signals and abnormal voltage signals on each phase of the three-phase electric meter under different load states, and a current transformer is used to collect historical normal signals and abnormal current signals on each phase of the three-phase electric meter under different load states, and the voltage and current signals to be identified are obtained in the same way. Among them, one voltage transformer and one current transformer can be arranged on each phase. For example, the placement position is on the phase line of each phase. In addition, the collected signals can be denoised using wavelet denoising, and the collected signals can be normalized to eliminate dimension differences.

[0032] At this point, the three-phase historical normal and abnormal feature (current, voltage) sequences under different load states and the feature sequence (current, voltage) to be identified are obtained. Each phase of the feature sequence to be identified contains a group of feature sequences, and in the present scheme, there are three groups, each including two feature sequences of current and voltage.

[0033] S2: Fourier transform is used to obtain the frequency spectrum sequence of the collected historical feature sequence, the time domain difference and the frequency domain difference are calculated according to the corresponding historical normal sequence, abnormal feature sequence and frequency spectrum sequence of the three phases under the same load state, and the weights of the time domain and the frequency domain are calculated through the time domain difference and the frequency domain difference.

[0034] In an embodiment of the present application, since the three-phase circuit in daily life contains various load states, the fluctuation of current or voltage caused by load fluctuation under different load states is different in time-frequency domain, and when the circuit is abnormal, the abnormality of current and voltage is not obvious in a single time domain or frequency domain. The traditional circuit identification of three-phase electric meter mainly based on single time-frequency domain feature or equal weight fusion ignores the abnormality of time-frequency domain, which leads to the phenomenon of not timely identification or misidentification when the three-phase electric meter circuit detects abnormality.

[0035] Therefore, Fourier transform is used to obtain the frequency spectrum sequence of the collected historical feature sequence, the time domain difference and the frequency domain difference are calculated according to the corresponding historical normal sequence, abnormal feature sequence and frequency spectrum sequence of the three phases under the same load state, and the weights of the time domain and the frequency domain are calculated through the time domain difference and the frequency domain difference. The process of obtaining the abnormal identification weight is as follows:

[0036] First, Fourier transform is used to obtain the spectrum sequence corresponding to each feature sequence for the three-phase historical normal and abnormal feature (current, voltage) sequences under different load conditions. The feature sequence contains the value of the feature in the time series, and the spectrum sequence contains the amplitude corresponding to each frequency. Each feature sequence corresponds to a spectrum sequence.

[0037] Then, for target feature 'a' among multiple features, the target feature under each load condition is calculated based on the historical normal and abnormal feature sequences of the three-phase system under different load conditions. The time-domain difference can be calculated similarly based on the spectral sequence to determine the frequency-domain difference. As an example, using target features... Taking the calculation of time-domain differences as an example:

[0038]

[0039] in, This indicates the characteristics under this load condition. The time domain difference, This indicates the target features collected under this load condition. The number of normal (abnormal) feature sequences, This indicates the length of each feature sequence. Indicates the first The first normal feature sequence One element, Indicates the first The first abnormal feature sequence One element, Its function is to normalize and eliminate the dimensional differences between the time and frequency domains, allowing calculations to be performed between these differences. This is because there is a phase difference between different phases. Both normal and abnormal feature sequences are signals acquired on the same phase. Similarly, features are calculated based on the spectral sequence. Frequency domain differences The frequency domain difference is calculated based on the spectral sequence, and the calculation method is the same as that for the time domain difference.

[0040] Finally, regarding target features The time-domain and frequency-domain weights are calculated based on the time-domain and frequency-domain differences of this feature under the same load condition. Taking the calculation of the time-domain weight as an example:

[0041]

[0042] in, Representation of features Temporal weights in anomaly detection A larger value indicates the normal or abnormal characteristics of the circuit in the time domain. The greater the difference, represent the time domain difference of the feature under the load state, represent the frequency domain difference of the feature under the load state. Similarly, the frequency domain weight of the feature in abnormality identification . The division logic of time domain and frequency domain weights is that the time (frequency) domain difference is calculated by the difference between the normal signal and the abnormal signal in the time-frequency domain, and the greater the time (frequency) domain difference, the greater the difference between the circuit abnormality and the normal in the time (frequency) domain, and the more significant it is, which is more helpful to abnormality identification.

[0043] So far, the time domain weight and the frequency domain weight of each feature under each load state are obtained.

[0044] S3: using a sliding window to divide the to-be-identified feature sequence, calculating the time domain and frequency domain abnormality degrees based on the time-frequency domain features of the target window and the adjacent window, and calculating the single-phase abnormality degree of the target feature by using the time domain and frequency domain weights of the target feature under each load state.

[0045] In an embodiment of the present application, because there is load transient fluctuation in the use of the three-phase circuit, the transient fluctuation of the load causes a sudden value of the feature in the time-frequency domain, for example, the transient of the harmonic content when the motor starts and stops. The traditional abnormality identification usually uses a static fixed feature (current, voltage) threshold, which cannot distinguish between normal transient impact and real failure, harmonic transient caused by load fluctuation, etc., resulting in misidentification or delayed identification when performing circuit abnormality identification of the three-phase electric meter.

[0046] Therefore, the present application uses a sliding window to divide the to-be-identified feature sequence, calculates the time domain and frequency domain abnormality degrees based on the time-frequency domain features of the target window and the adjacent window, and calculates the single-phase abnormality degree of the feature by using the time domain and frequency domain weights of the feature under each load state. The calculation of the single-phase abnormality degree specifically includes the following steps:

[0047] (1) for a target feature and a phase , the to-be-identified feature sequence of the feature is divided by a sliding window, an exemplary window length is 5, and the time domain abnormality degree is calculated based on the sliding divided window:

[0048]

[0049] wherein, represents the time domain abnormality degree of the feature on the phase , represents the number of adjacent windows, represents the sliding window length, represents the i-th element in the target window, represents the i-th element in the target window, represents the i-th element in the i-th adjacent window, represents the i-th element in the target window, represents the i-th element in the i-th adjacent window, represents the i-th element in the target window, represents the i-th element in the target window, represents the i-th element in the target window, represents the i-th element in the i-th adjacent window, represents the i-th element in the i-th adjacent window, is a preset characteristic value and serves to eliminate dimension. The calculation logic of the abnormality degree is as follows: when the circuit is abnormal, the characteristic value generated at the current time instant is obviously different from that at the adjacent time instant, so the greater the difference between the target window and the adjacent window, the greater the probability of the current time instant being abnormal.

[0050] (2) For a target characteristic and a phase , the Fourier transform is used to obtain the frequency spectrum sequence corresponding to each window for each window divided in process (1), and the frequency domain abnormality degree is calculated based on the frequency spectrum sequences of the window at the previous time instant and the adjacent windows:

[0051]

[0052] wherein, represents the frequency domain abnormality degree of the phase characteristic, represents the number of adjacent windows, represents the length of the frequency spectrum sequence, represents the i-th element of the target window frequency spectrum sequence, represents the i-th element of the i-th adjacent window frequency spectrum sequence. The calculation logic of the abnormality degree is as follows: when the circuit is abnormal, the characteristic will produce unique harmonics in the frequency domain, and the difference between the target window and the adjacent window in the frequency domain can capture such changes, and the greater the difference, the greater the probability of producing harmonics, and the greater the probability of the amplitude of different frequencies changing, the greater the probability of the circuit being abnormal. (3) For a target characteristic and a phase , the single-phase abnormality degree of the phase characteristic is calculated based on the time domain abnormality degree and the frequency domain abnormality degree of the phase characteristic:

[0053]

[0054]

[0055] ​​​​​​​wherein, represents a single-phase abnormality degree of the feature on the phase represents a time-domain weight of the feature in the abnormality identification, Similarly, represents a time-domain abnormality degree of the feature on the phase Similarly.

[0056] (4) Similarly, the single-phase abnormality degree of each feature on each phase is obtained.

[0057] S4: The single-phase abnormality degree of each target feature on each phase is input into a control system, and circuit abnormality identification of the three-phase ammeter is performed according to a preset abnormality degree threshold.

[0058] In an embodiment of the present application, since three-phase circuit abnormality is usually first manifested on a single phase, an alarm signal is sent to the system when the single-phase abnormality degree of the feature on any phase exceeds a preset abnormality degree threshold, wherein the preset abnormality degree threshold is a standard of a corresponding numerical value of the judgment abnormality degree preset in advance. For example, the preset abnormality degree threshold is 0.7.

[0059] The present application also discloses a circuit abnormality identification system of a three-phase ammeter, which adopts the circuit abnormality identification method of the three-phase ammeter, and specifically comprises:

[0060] a signal acquisition module, configured to acquire historical normal signals, abnormal feature signals and to-be-identified feature signals of each phase of the three-phase ammeter under different load states;

[0061] a weight calculation module, configured to use Fourier transform to obtain a frequency spectrum sequence of the acquired historical feature sequence, calculate time-domain differences and frequency-domain differences according to the historical normal sequence, the abnormal feature sequence and the frequency spectrum sequence of the three phases under the same load state, and calculate the weights in the time domain and the frequency domain through the time-domain differences and the frequency-domain differences;

[0062] a single-phase abnormality degree calculation module, configured to divide the to-be-identified feature sequence using a sliding window, calculate time-domain and frequency-domain abnormality degrees based on time-frequency domain features of a target window and adjacent windows, and calculate the single-phase abnormality degree of the target feature by using the time-domain and frequency-domain weights of the target feature under each load state;

[0063] an abnormality identification module, configured to input the single-phase abnormality degree of each target feature on each phase into a control system, and perform circuit abnormality identification of the three-phase ammeter according to a preset abnormality degree threshold;

[0064] ​​​​​The single-phase abnormality degree calculation comprises: using Fourier transform to obtain a frequency spectrum sequence corresponding to each window by dividing each window of the sliding window, calculating a frequency domain abnormality degree based on the frequency spectrum sequences of the window and adjacent windows of a previous time, and calculating the single-phase abnormality degree of the target feature based on the time domain abnormality degree and the frequency domain abnormality degree of the target feature, so as to obtain the single-phase abnormality degree of each target feature in each phase.

[0065] In summary, the Fourier transform is used to obtain a frequency spectrum sequence of the collected historical feature sequence, the time domain difference and the frequency domain difference are calculated according to the three-phase historical normal and abnormal feature sequences and the frequency spectrum sequences under the same load state, and the weights of the time domain and the frequency domain are calculated through the time domain difference and the frequency domain difference. The target feature sequence is divided by using the sliding window, the time domain and frequency domain abnormality degrees are calculated based on the time-frequency domain features of the target window and adjacent windows, and the single-phase abnormality degree of the feature is calculated by using the time domain and frequency domain weights of the feature under each load state. The time-frequency domain features are fused by using the adaptive weight to avoid high false alarm rate of a single time-frequency domain, and the time series analysis is performed to avoid false alarm caused by instantaneous change of a single time feature.

[0066] The above are preferred embodiments of the present application, and do not limit the protection scope of the present application, so that: equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method of identifying a circuit abnormality of a three-phase wattmeter, characterized by, The method comprises the following steps: Collecting corresponding historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals of each phase of the three-phase electric meter under different load states; Using Fourier transform on the collected historical characteristic sequences to obtain frequency spectrum sequences, calculating time domain differences and frequency domain differences according to the corresponding historical normal sequences, abnormal characteristic sequences and frequency spectrum sequences of the three phases under the same load state, and calculating the weights of the time domain and the frequency domain through the time domain differences and the frequency domain differences; Dividing the to-be-identified characteristic sequences using a sliding window, calculating the time domain and frequency domain abnormal degrees based on the time-frequency domain characteristics of the target window and the adjacent window, and calculating the single-phase abnormal degree of the target characteristic using the time domain and frequency domain weights of the target characteristic under each load state; Inputting the single-phase abnormal degree of each target characteristic on each phase into a control system, and performing circuit abnormal identification of the three-phase electric meter according to a preset abnormal degree threshold; The calculation of the single-phase abnormal degree comprises: using Fourier transform on each window divided by the sliding window to obtain the frequency spectrum sequence corresponding to each window, calculating the frequency domain abnormal degree based on the frequency spectrum sequences of the window at the previous moment and the adjacent window, and calculating the single-phase abnormal degree of the target characteristic based on the time domain abnormal degree and the frequency domain abnormal degree of the target characteristic, so as to obtain the single-phase abnormal degree of each target characteristic on each phase; The calculation of the single-phase abnormal degree specifically comprises the following steps: (1) For a target feature and a phase , a sliding window is divided for a feature sequence to be identified of the target feature, and a time domain abnormality degree is calculated based on the sliding divided window: in, Indicates phase upper features The degree of temporal anomaly, Indicates the number of adjacent windows. Indicates the length of the sliding window. Indicates the first in the target window One element, Indicates the first The first neighboring window One element, Indicates the first [number] in the target window The slope of each element, Indicates the first Within the nearest window The slope of each element, It is a preset feature value whose function is to eliminate dimensions; the calculation logic of the degree of abnormality is as follows: when the circuit is abnormal, the feature value that will produce the abnormality at the current time is significantly different from that at the adjacent time. Therefore, the greater the difference between the target window and the adjacent windows, the greater the probability of the abnormality occurring at the current time. (2) For a target feature and a phase , the Fourier transform is used to obtain the corresponding frequency spectrum sequence of each window for process (1), and the frequency domain anomaly degree is calculated based on the frequency spectrum sequences of the window and adjacent windows at the previous time. in, Indicates phase upper features The degree of frequency domain anomaly, Indicates the number of adjacent windows. Indicates the length of the spectral sequence. The first spectral sequence of the target window represents the... One element, Indicates the first The spectral sequence of the nearest neighbor window The calculation logic for the degree of abnormality of each element is as follows: When the circuit is abnormal, the feature will generate unique harmonics in the frequency domain. This change can be captured by the difference in amplitude changes at each frequency between the target window and the adjacent windows in the frequency domain. The larger the difference, the greater the probability of generating harmonics. The greater the probability of amplitude changes at different frequencies, the greater the probability of circuit abnormality. (3) For a target feature and a phase , the single-phase abnormality degree of the phase feature is calculated based on the time-domain abnormality degree and the frequency-domain abnormality degree of the phase feature . wherein, denotes the degree of abnormality of the feature in phase , denotes the degree of abnormality of the feature in time , denotes the weight of the feature in time domain in abnormality recognition, likewise, denotes the degree of abnormality of the feature in phase , denotes the degree of abnormality of the feature in time likewise; (4) Similarly, the single-phase abnormal degree of each characteristic on each phase is obtained. The calculation of the weights of the time domain and the frequency domain comprises: Using Fourier transform on the historical normal and abnormal characteristic sequences of the three phases under different load states to obtain the frequency spectrum sequence corresponding to each characteristic sequence; Calculating the time domain differences and the frequency domain differences of the target characteristic under each load state according to the historical normal and abnormal characteristic sequences of the three phases of the target characteristic under different load states; Calculating the weights of the time domain and the frequency domain based on the time domain differences and the frequency domain differences of the target characteristic under the same load state.

2. The method of claim 1, wherein, Using voltage and current transformers to collect corresponding historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals of each phase of the three-phase electric meter under different load states.

3. The method of claim 2, wherein, Using wavelet denoising to perform denoising processing on the collected corresponding historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals of each phase of the three-phase electric meter under different load states.

4. The method of claim 2, wherein, Performing normalization to eliminate the dimensional differences of the collected corresponding historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals of each phase of the three-phase electric meter under different load states.

5. A system for identifying abnormality in a circuit of a three-phase wattmeter, which employs the method for identifying abnormality in a circuit of a three-phase wattmeter according to any one of claims 1 to 4, characterized by The method comprises the following steps: A signal collection module is configured to collect corresponding historical normal signals, abnormal characteristic signals and to-be-identified characteristic signals of each phase of the three-phase electric meter under different load states; A weight calculation module is configured to use Fourier transform on the collected historical characteristic sequences to obtain frequency spectrum sequences, calculate time domain differences and frequency domain differences according to the corresponding historical normal sequences, abnormal characteristic sequences and frequency spectrum sequences of the three phases under the same load state, and calculate the weights of the time domain and the frequency domain through the time domain differences and the frequency domain differences. The single-phase abnormality degree calculation module is configured to divide the to-be-identified feature sequence using a sliding window, calculate time-domain and frequency-domain abnormality degrees based on time-frequency domain features of a target window and adjacent windows, and calculate a single-phase abnormality degree of the target feature using time-domain and frequency-domain weights of the target feature under each load state. The abnormality identification module is configured to input the single-phase abnormality degree of each target feature on each phase into a control system, and perform circuit abnormality identification of the three-phase ammeter according to a preset abnormality degree threshold. The calculation of the single-phase abnormality degree includes: using Fourier transform to obtain a frequency spectrum sequence corresponding to each window for each window divided by the sliding window, calculating a frequency-domain abnormality degree based on frequency spectrum sequences of a window at a previous time and adjacent windows, and calculating a single-phase abnormality degree of the target feature based on the time-domain abnormality degree and the frequency-domain abnormality degree of the target feature, to obtain the single-phase abnormality degree of each target feature on each phase. The calculation of the single-phase abnormality degree specifically includes the following steps: (1) For a target feature and a phase , a sliding window is divided for a feature sequence to be identified of the target feature, and a time domain abnormality degree is calculated based on the sliding divided window: wherein, represents the phase of the above feature , the time-domain abnormality degree, represents the number of adjacent windows, represents the length of the sliding window, represents the first element in the target window, represents the first element in the first adjacent window, represents the first element in the first adjacent window, represents the slope of the first element in the target window, represents the slope of the first element in the first adjacent window, is a preset characteristic value and serves to eliminate the dimension; the calculation logic of the abnormality degree is that when the circuit is abnormal, the characteristic value of the current time will be obviously different from that of the adjacent time, so the greater the difference between the target window and the adjacent window, the greater the probability of the current time being abnormal. (2) For a target feature and a phase , the Fourier transform is used to obtain the corresponding frequency spectrum sequence of each window for process (1), and the frequency domain anomaly degree is calculated based on the frequency spectrum sequences of the window and adjacent windows at the previous time. wherein, represents the phase of the upper feature , the degree of abnormality in the frequency domain, represents the number of adjacent windows, represents the length of the spectrum sequence, represents the element of the target window spectrum sequence, the first element of the first adjacent window spectrum sequence, the first element of the first adjacent window spectrum sequence, the first element of the first adjacent window spectrum sequence, the first element of the first adjacent window spectrum sequence, the calculation logic of the degree of abnormality is that when the circuit is abnormal, the feature will produce a unique harmonic in the frequency domain. By the amplitude difference of each frequency in the frequency domain of the target window and the adjacent window, this change can be captured. The greater the difference, the greater the probability of generating harmonics. The greater the probability of the amplitude change of different frequencies, the greater the probability of circuit abnormality. (3) For a target feature and a phase , the single-phase abnormality degree of the phase feature is calculated based on the time-domain abnormality degree and the frequency-domain abnormality degree of the phase feature . wherein, represents the single-phase abnormality degree of the feature on the phase , represents the single-phase abnormality degree of the feature on the phase , represents the time-domain weight of the feature in the abnormality recognition, Similarly, represents the time-domain abnormality degree of the feature on the phase , represents the time-domain abnormality degree of the feature on the phase Similarly; (4) Similarly, the single-phase abnormality degree of each feature on each phase is obtained. The weight calculation module further includes: The transform module is configured to use Fourier transform to obtain a frequency spectrum sequence corresponding to each feature sequence for three-phase historical normal and abnormal feature sequences under different load states. The difference calculation module is configured to calculate time-domain and frequency-domain differences of the target feature under each load state according to the three-phase historical normal and abnormal feature sequences of the target feature under different load states. The target weight calculation module is configured to calculate time-domain and frequency-domain weights based on the time-domain and frequency-domain differences of the target feature under the same load state.

6. The system of claim 5, wherein, The signal acquisition module includes: The target signal acquisition module is configured to acquire historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase ammeter under different load states using voltage and current transformers. The denoising module is configured to perform denoising processing on the acquired historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase ammeter under different load states using wavelet denoising. The normalization module is configured to perform normalization to eliminate dimensional differences of the acquired historical normal signals, abnormal feature signals and to-be-identified feature signals corresponding to each phase of the three-phase ammeter under different load states.

Images