A method and system for monitoring leakage current in smart grids

Abstract

This invention relates to the field of power grid monitoring and analysis technology, specifically to a method and system for monitoring leakage current in smart grids. The method includes: monitoring the leakage current of target equipment, acquiring the leakage current during a leakage event, decomposing it into current signals of different frequency bands using wavelet packet transform, analyzing the current signal data of different frequency bands to determine the strongly correlated frequency bands for each band, constructing a frequency band sequence set, analyzing and acquiring the duration of each frequency band sequence, multiple strongly correlated sequences, and their corresponding correlation durations, calculating the average duration of the frequency band sequences, constructing a risk warning model, performing calculations based on real-time monitoring data, and using this model as a basis for real-time log synchronization and warning information dissemination decisions. This invention solves the problems of insufficient frequency band correlation analysis of leakage signals, inability to accurately capture leakage characteristics, difficulty in predicting leakage development trends, and inaccurate warning in existing technologies, achieving more in-depth signal analysis and more accurate warnings.

Description

Technical Field

[0001] This invention relates to the field of power grid monitoring and analysis technology, specifically to a method and system for monitoring leakage current in a smart grid. Background Technology

[0002] Leakage current monitoring is used to detect abnormal current leakage in circuits in real time, preventing electric shock injuries and electrical fires, and ensuring personal and property safety. Its core function is to promptly identify leakage risks caused by equipment failures and automatically cut off power to prevent accidents from escalating.

[0003] Chinese patent document CN107015103A discloses a leakage current monitoring system for a smart grid, including: a detection module, a data processing module, and an alarm module. The detection module is used to collect leakage current parameter data and transmit it to the data processing module. The data processing module allocates the leakage current parameter data to different data processing groups according to the magnitude of the leakage current parameter and issues an instruction corresponding to the data processing group to which the leakage current parameter is allocated to the alarm module to perform an electrical fire alarm.

[0004] Existing technologies allocate processing groups and trigger alarms based on the magnitude of leakage current parameters. However, they cannot perform frequency band decomposition and correlation analysis on leakage current signals, thus failing to deeply mine signal characteristics and resulting in insufficient accuracy in early warning. Furthermore, they are unable to capture leakage development trends and cannot predict future leakage development trends, thus lacking foresight in early warning. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background technology by proposing a leakage current monitoring method and system for smart grids.

[0006] The technical solution of this invention: a leakage current monitoring method for smart grids, comprising the following steps:

[0007] S1. Define a leakage event, monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the leakage event occurs, and decompose it into N different frequency bands using wavelet packet transform. Analyze the current signal data of different frequency bands to determine the strongly correlated frequency bands of each frequency band.

[0008] S2. Construct a frequency band sequence set for the characteristic window period, process the frequency band sequence set, and analyze and obtain the maintenance duration, multiple strongly correlated sequences and correlation duration of each frequency band sequence based on the strongly correlated frequency bands in the frequency band sequence set.

[0009] S3. Obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the average duration of the frequency band sequence. Use this as an input factor to build a risk warning model. Perform calculations based on real-time monitoring data and use the data as a log for real-time synchronization and decision-making for issuing warning information.

[0010] Preferably, the methods for analyzing current signal data in different frequency bands include:

[0011] Based on several frequency bands existing at different time points within the feature window period, sort them according to the median value of the frequency bands to construct the frequency band sequence corresponding to the time point;

[0012] Based on multiple historical leakage events, multiple current signals in the same frequency band are obtained in each frequency band sequence and marked as the same type of current signal in that frequency band.

[0013] The mean, median, and variance of several similar current signals in each frequency band are obtained respectively. Based on the mean, median, and variance of the frequency, the feature vectors of each similar current signal are constructed and combined to obtain the feature vector set of the frequency band.

[0014] Preferably, the average similarity coefficient of the feature vector set is calculated using the following formula and used as the feature value Z of the frequency band:

[0015]

[0016] In the formula, A i and A j Let i and j be the feature vectors of two similar current signals with different numbers; i and j are feature vector numbers, and i and j are both positive integers; n is the total number of feature vectors in the feature vector set.

[0017] The standard deviation of the frequency band sequence is calculated based on the characteristic value Z of several frequency pairs in the frequency band sequence. The sequence standard deviation is then compared with a preset standard deviation threshold. If the sequence standard deviation is not greater than the standard deviation threshold, it is determined that there is a strong correlation between several frequency bands in the frequency band sequence; otherwise, it is determined that there is no strong correlation between several frequency bands in the frequency band sequence.

[0018] Preferably, strongly correlated frequency band sequences are marked as target sequences, and a local strongly correlated list of frequency bands is constructed based on the strongly correlated frequency bands of each frequency band in the target sequence;

[0019] Perform a union operation on several locally strongly correlated lists of the same frequency band for all target sequences to determine the globally strongly correlated list for each frequency band.

[0020] Preferably, a primary frequency band is added to the frequency bands with a global strong correlation list. The frequency bands with the primary frequency band mark are combined with the corresponding frequency bands in the global strong correlation list in turn to obtain several frequency pairs. The deviation of the frequency pairs is calculated, and the frequency pair corresponding to the minimum deviation value is obtained and marked as the target frequency pair. The unmarked frequency bands in the target frequency pair are taken as the strongly correlated frequency bands with the primary frequency band mark.

[0021] Preferably, S2 includes the following steps:

[0022] Step 1: Construct a frequency band sequence set based on the frequency band sequences of all time nodes within the feature window period;

[0023] Step 2: According to the sliding pointer operation, mark the frequency band sequences in the frequency band sequence set as selected sequences in turn. Determine whether there is a frequency band with a main frequency band label in the selected sequence. If not, set the strongly associated sequence and the corresponding associated duration of the selected sequence to empty.

[0024] If they exist, based on the strongly correlated frequency bands of each frequency band in the selected sequence, we infer multiple frequency band sequences that the selected sequence may trigger and treat them all as strongly correlated sequences. We then index the strongly correlated sequences in the frequency band sequence set and obtain the occurrence time of each strongly correlated sequence as a marker time point. Based on the interval between the start time point and the marker time point, we calculate the association duration of the corresponding strongly correlated sequence triggered by the selected sequence.

[0025] Step 3: Unmark the selected sequence and slide the pointer along the sort order of the frequency band sequence set and remark it; whenever a frequency band sequence that has been traversed is encountered, update its duration cumulatively; repeat the sequence operation until all frequency band sequences in the frequency band sequence set have been traversed.

[0026] Preferably, during real-time monitoring, the risk score RS of the acquired real-time frequency band sequence is output using a risk warning model. The expression of the risk warning model is as follows:

[0027]

[0028] In the formula, t keep,real This refers to the actual duration of maintenance; t keep,avg t0 is the average duration; k is the coefficient; t0 is the duration constant used to define the maximum and minimum risk scores.

[0029] Preferably, the decision to issue early warning information includes comparing the risk score with a preset score threshold. If the risk score is not greater than the score threshold, no early warning is issued; if the real-time risk value is greater than the risk threshold, an emergency early warning is issued.

[0030] Based on the strong correlation frequency bands and correlation duration of the real-time frequency band sequence, the future frequency band sequence is inferred to obtain the predicted frequency band. Based on the average duration of the predicted frequency band, the possible time point for emergency warning is inferred.

[0031] This invention also discloses a leakage current monitoring system for a smart grid, which applies the aforementioned leakage current monitoring method for a smart grid, specifically including:

[0032] The data acquisition and first processing and analysis module is used to define leakage events, monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the occurrence of the leakage event, decompose it into N different frequency bands using wavelet packet transform, analyze the current signal data of different frequency bands, and determine the strongly correlated frequency bands of each frequency band.

[0033] The second data analysis module is used to construct a frequency band sequence set for the characteristic window period, process the frequency band sequence set, and analyze and obtain the maintenance duration, multiple strongly correlated sequences and correlation duration of each frequency band sequence in the frequency band sequence set based on the strongly correlated frequency bands of each frequency band.

[0034] The risk warning modeling and decision-making module is used to obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the mean to determine the average duration of the frequency band sequence. This average duration is then used as an input factor to construct a risk warning model. The model is calculated based on real-time monitoring data and used as a log for real-time synchronization and decision-making for issuing warning information.

[0035] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects:

[0036] This invention decomposes leakage current signals into different frequency bands using wavelet packet transform, deeply analyzes the current signals in each frequency band to determine strongly correlated frequency bands, constructs a frequency band sequence set, and analyzes the duration of each frequency band sequence, strongly correlated sequences, and correlation duration. These factors are used as input factors to construct a risk warning model. By combining strongly correlated frequency bands and correlation duration, it predicts future frequency band sequences and emergency warning time points. This invention solves the problems of insufficient frequency band correlation analysis of leakage signals, inability to accurately capture leakage characteristics, difficulty in predicting leakage development trends, and inaccurate warning in existing technologies, achieving more in-depth signal analysis and more accurate warning. Attached Figure Description

[0037] Figure 1 This is a block diagram of a method according to Embodiment 1 of the present invention. Detailed Implementation

[0038] Example 1, as Figure 1 As shown, the present invention proposes a method for monitoring leakage current in a smart grid, comprising the following steps:

[0039] S1. Define a leakage current event. Monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the leakage current event occurs, and decompose it into N different frequency bands using wavelet packet transform. Analyze the current signal data of different frequency bands to determine the strongly correlated frequency bands of each band. Specifically, a leakage current event is defined by monitoring the leakage current of the target device, comparing the leakage current with the safety limit, obtaining the comparison result, and defining the leakage current event based on the comparison result. If the real-time leakage current is greater than the safety limit, the comparison result is defined as a leakage current event; otherwise, no operation is performed. The duration of the characteristic window period is determined based on big data testing of effective early warning accuracy. For example, the leakage current is decomposed into three layers using wavelet packets to obtain current signals of 8 frequency bands.

[0040] Methods for analyzing current signal data in different frequency bands include:

[0041] Based on several frequency bands existing at different time points within the feature window period, sort them according to the median value of the frequency bands to construct the frequency band sequence corresponding to the time point; for example, there are frequency bands P1, P2 and P3, whose corresponding median values ​​increase sequentially, then sorting frequency bands P1, P2 and P3 will result in a frequency band sequence including {P1, P2, P3}.

[0042] Based on multiple historical leakage events, multiple current signals in the same frequency band are obtained in each frequency band sequence and marked as the same type of current signal in that frequency band.

[0043] The mean, median, and variance of several similar current signals in each frequency band are obtained. Based on these mean, median, and variance, feature vectors for each similar current signal are constructed and combined to obtain the feature vector set for that frequency band. The average similarity coefficient of the feature vector set is calculated using the following formula and used as the feature value Z of the frequency band:

[0044]

[0045] In the formula, A i and A j Let i and j be the feature vectors of two similar current signals with different numbers; i and j are feature vector numbers, and i and j are both positive integers; n is the total number of feature vectors in the feature vector set.

[0046] The standard deviation of the frequency band sequence is calculated based on the characteristic value Z of several frequency pairs in the frequency band sequence. The sequence standard deviation is then compared with a preset standard deviation threshold. If the sequence standard deviation is not greater than the standard deviation threshold, it is determined that there is a strong correlation between several frequency bands in the frequency band sequence; otherwise, it is determined that there is no strong correlation between several frequency bands in the frequency band sequence. The standard deviation threshold is dynamically updated based on the first M historical leakage events and the first Q frequency band sequences corresponding to the occurrence time of each event. Specifically, M can be 10 and Q can be 1000. The sequence standard deviation of the most recent 10×1000 frequency band sequences is calculated, and the sequence standard deviations are sorted in ascending order. The sequence standard deviation at the 20th percentile position is taken as the standard deviation threshold.

[0047] Strongly correlated frequency band sequences are marked as target sequences. A local strongly correlated list of frequency bands is constructed based on the strongly correlated frequency bands of each frequency band in the target sequence. For example, under the target sequence {P1, P2, P3}, the local strongly correlated list of frequency band P1 is {P2, P3}, the local strongly correlated list of frequency band P2 is {P1, P3}, and the local strongly correlated list of frequency band P3 is {P1, P2}.

[0048] Perform a union operation on several local strongly correlated lists of the same frequency band under all target sequences to determine the global strongly correlated list of each frequency band; for example, under the target sequence {P3, P4, P5}, the local strongly correlated list of frequency band P3 is {P4, P5}; then for all target sequences {P1, P2, P3} and {P3, P4, P5}, the global strongly correlated list of frequency band P3 is {P1, P2, P4, P5}.

[0049] Add a primary frequency band label to the frequency bands in the globally strongly associated list. Combine the frequency bands with the primary frequency band label with the corresponding frequency bands in the globally strongly associated list sequentially to obtain several frequency pairs. Calculate the deviation of the frequency pairs, obtain the frequency pair with the minimum deviation and label it as the target frequency pair. The unlabeled frequency bands in the target frequency pair are considered strongly associated frequency bands with the primary frequency band label. It should be noted that for frequency bands without a primary frequency band label, their strongly associated frequency bands are empty. For example, for the globally strongly associated list {P1, P2, P4, P5} of the primary frequency band P3, the frequency pairs obtained are as follows:

[0050] If the deviations of the frequency pairs (P3, P1), (P3, P2), (P3, P4), and (P3, P5) increase sequentially, then the strongly correlated frequency band of frequency band P3 is P1. It should be noted that the strongly correlated frequency band is the frequency band that is most likely to appear after a specific time period for the currently appearing frequency band.

[0051] S2. Construct a frequency band sequence set for the characteristic window time period, process the frequency band sequence set, and based on the strongly correlated frequency bands of each frequency band, analyze and obtain multiple strongly correlated sequences and correlation durations for each frequency band sequence in the frequency band sequence set. The method includes the following steps:

[0052] Step 1: Construct a frequency band sequence set based on the frequency band sequences of all time points within the feature window period. It should be noted that constructing the frequency band sequence set based on all time points within the feature window period can be achieved by acquiring a frequency band sequence per second and combining the frequency band sequences from each second to obtain the frequency band sequence set. For example, if the unit time length is 1 second, and the frequency band sequences at time points t1, t2, and t3 are all selected sequences, then the duration of the selected sequence is 3 seconds.

[0053] Step 2: According to the sliding pointer operation, mark the frequency band sequences in the frequency band sequence set as selected sequences in turn. Determine whether there is a frequency band with a main frequency band label in the selected sequence. If not, set the strongly associated sequence and the corresponding associated duration of the selected sequence to empty.

[0054] If they exist, based on the strongly correlated frequency bands of each frequency band in the selected sequence, we infer multiple frequency band sequences that the selected sequence may trigger and treat them all as strongly correlated sequences. We then index the strongly correlated sequences in the frequency band sequence set and obtain the occurrence time of each strongly correlated sequence as a marker time point. Based on the interval between the start time point and the marker time point, we calculate the association duration of the corresponding strongly correlated sequence triggered by the selected sequence.

[0055] For example, if all frequency bands in the frequency band sequence {P1, P2} have a main frequency band label, and the strongly correlated frequency bands of frequency bands P1 and P2 are P3 and P4 respectively, then the strongly correlated frequency bands of the frequency band sequence {P1, P2} include {P1, P4}, {P2, P3}, {P2, P4}, {P1, P2, P3}, {P1, P2, P4}, and {P1, P2, P3, P4}. It should be noted that the strongly correlated sequence triggered by the frequency band with the main frequency band label in the selected frequency band may or may not include the strongly correlated frequency band of the main frequency band, resulting in multiple possibilities.

[0056] Step 3: Unmark the selected sequence and slide the pointer along the sort order of the frequency band sequence set and remark it; whenever a frequency band sequence that has been traversed is encountered, update its duration cumulatively; repeat the sequence operation until all frequency band sequences in the frequency band sequence set have been traversed.

[0057] S3. Obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the average duration to determine the average duration of the frequency band sequence. Use this as an input factor to construct a risk warning model. Perform calculations based on real-time monitoring data and use the results as a basis for real-time log synchronization and warning information dissemination decisions.

[0058] During real-time monitoring, the risk score RS of the acquired real-time frequency band sequence is output using a risk warning model. The expression of the risk warning model is as follows:

[0059]

[0060] In the formula, t keep,real This refers to the actual duration of maintenance; t keep,avg t0 is the average duration; k is a coefficient; t0 is a duration constant used to define the maximum and minimum risk scores.

[0061] The decision to issue early warning information includes comparing the risk score with a preset score threshold. If the risk score is not greater than the score threshold, no early warning is issued; if the real-time risk value is greater than the risk threshold, an emergency early warning is issued.

[0062] Based on the strong correlation frequency bands and correlation duration of the real-time frequency band sequence, the future frequency band sequence is inferred to obtain the predicted frequency band. Based on the average duration of the predicted frequency band, the possible time point for emergency warning is inferred.

[0063] By decomposing the leakage current signal into different frequency bands through wavelet packet transform, and deeply analyzing the current signal in each frequency band to determine the strongly correlated frequency bands, a frequency band sequence set is constructed, and the duration of each frequency band sequence, the strongly correlated sequence, and the correlation duration are analyzed. These factors are used as input factors to construct a risk warning model. Combining the strongly correlated frequency bands and the correlation duration, the model predicts future frequency band sequences and emergency warning time points. This solves the problems of insufficient frequency band correlation analysis of leakage signals, inability to accurately capture leakage characteristics, difficulty in predicting leakage development trends, and accurate warning in existing technologies, and achieves more in-depth signal analysis and more accurate warning.

[0064] Example 2: A leakage current monitoring system for a smart grid proposed in this invention is applied to a leakage current monitoring method for a smart grid proposed in Example 1, and specifically includes:

[0065] The data acquisition and first processing and analysis module is used to define leakage events, monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the occurrence of the leakage event, decompose it into N different frequency bands using wavelet packet transform, analyze the current signal data of different frequency bands, and determine the strongly correlated frequency bands of each frequency band.

[0066] The second data analysis module is used to construct a frequency band sequence set for the characteristic window period, process the frequency band sequence set, and analyze and obtain the maintenance duration, multiple strongly correlated sequences and correlation duration of each frequency band sequence in the frequency band sequence set based on the strongly correlated frequency bands of each frequency band.

[0067] The risk warning modeling and decision-making module is used to obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the mean to determine the average duration of the frequency band sequence. This average duration is then used as an input factor to construct a risk warning model. The model is calculated based on real-time monitoring data and used as a log for real-time synchronization and decision-making for issuing warning information.

[0068] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for monitoring leakage current in a smart grid, characterized in that, Includes the following steps: S1. Define a leakage event, monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the leakage event occurs, and decompose it into N different frequency bands using wavelet packet transform. Analyze the current signal data of different frequency bands to determine the strongly correlated frequency bands of each frequency band. Methods for analyzing current signal data in different frequency bands include: Based on several frequency bands existing at different time points within the feature window period, sort them according to the median value of the frequency bands to construct the frequency band sequence corresponding to the time point; Based on multiple historical leakage events, multiple current signals in the same frequency band are obtained in each frequency band sequence and marked as the same type of current signal in that frequency band. The mean, median, and variance of several similar current signals in each frequency band are obtained respectively. Based on the mean, median, and variance of the frequency, the feature vectors of each similar current signal are constructed and combined to obtain the feature vector set of the frequency band. The average similarity coefficient of the feature vector set is calculated using the following formula and used as the feature value Z of the frequency band: ； In the formula, and Let i and j be the feature vectors of two similar current signals with different numbers; i and j are feature vector numbers, and i and j are both positive integers; n is the total number of feature vectors in the feature vector set. The standard deviation of the frequency band sequence is calculated based on the characteristic value Z of several frequency pairs in the frequency band sequence. The sequence standard deviation is then compared with a preset standard deviation threshold. If the sequence standard deviation is not greater than the standard deviation threshold, it is determined that there is a strong correlation between several frequency bands in the frequency band sequence; otherwise, it is determined that there is no strong correlation between several frequency bands in the frequency band sequence. S2. Construct a frequency band sequence set for the characteristic window period, process the frequency band sequence set, and based on the strongly correlated frequency bands of each frequency band, analyze and obtain the maintenance duration of each frequency band sequence, multiple strongly correlated sequences and their corresponding correlation durations in the frequency band sequence set. S3. Obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the average duration of the frequency band sequence. Use this as an input factor to build a risk warning model. Perform calculations based on real-time monitoring data and use the data as a log for real-time synchronization and decision-making for issuing warning information.

2. The method for monitoring leakage current in a smart grid according to claim 1, characterized in that, Strongly correlated frequency band sequences are marked as target sequences, and a local strongly correlated list of frequency bands is constructed based on the strongly correlated frequency bands of each frequency band in the target sequence. Perform a union operation on several locally strongly correlated lists of the same frequency band for all target sequences to determine the globally strongly correlated list for each frequency band.

3. The method for monitoring leakage current in a smart grid according to claim 2, characterized in that, Add a primary frequency band label to the frequency bands in the global strongly correlated list. Combine the frequency bands with the primary frequency band label with the corresponding frequency bands in the global strongly correlated list in sequence to obtain several frequency pairs. Calculate the deviation of the frequency pairs, obtain the frequency pair corresponding to the minimum deviation value and mark it as the target frequency pair. Take the unlabeled frequency bands in the target frequency pair as the strongly correlated frequency bands with the primary frequency band label.

4. The leakage current monitoring method for a smart grid according to claim 3, characterized in that, S2 includes the following steps: Step 1: Construct a frequency band sequence set based on the frequency band sequences of all time nodes within the feature window period; Step 2: According to the sliding pointer operation, mark the frequency band sequences in the frequency band sequence set as selected sequences in turn. Determine whether there is a frequency band with a main frequency band label in the selected sequence. If not, set the strongly associated sequence and the corresponding associated duration of the selected sequence to empty. If they exist, based on the strongly correlated frequency bands of each frequency band in the selected sequence, we infer multiple frequency band sequences that the selected sequence may trigger and treat them all as strongly correlated sequences. We then index the strongly correlated sequences in the frequency band sequence set and obtain the occurrence time of each strongly correlated sequence as a marker time point. Based on the interval between the start time point and the marker time point, we calculate the association duration of the corresponding strongly correlated sequence triggered by the selected sequence. Step 3: Unmark the selected sequence and slide the pointer along the sort order of the frequency band sequence set and remark it; whenever a frequency band sequence that has been traversed is encountered, update its duration cumulatively; repeat the sequence operation until all frequency band sequences in the frequency band sequence set have been traversed.

5. The leakage current monitoring method for a smart grid according to claim 4, characterized in that, During real-time monitoring, the risk score RS of the acquired real-time frequency band sequence is output using a risk warning model. The expression of the risk warning model is as follows: ； In the formula, This refers to the actual duration of maintenance. The average duration of maintenance; k is a coefficient; This is a duration constant used to define the maximum and minimum risk scores.

6. The leakage current monitoring method for a smart grid according to claim 1, characterized in that, The decision to issue early warning information includes comparing the risk score with a preset score threshold. If the risk score is not greater than the score threshold, no early warning is issued. If the real-time risk value exceeds the risk threshold, an emergency warning will be issued; Based on the strong correlation frequency bands and correlation duration of the real-time frequency band sequence, the future frequency band sequence is inferred to obtain the predicted frequency band. Based on the average duration of the predicted frequency band, the possible time point for emergency warning is inferred.

7. A leakage current monitoring system for a smart grid, applied to the leakage current monitoring method for a smart grid as described in any one of claims 1 to 6, characterized in that, Specifically, it includes: The data acquisition and first processing and analysis module is used to define leakage events, monitor the leakage current of the target device, obtain the leakage current during a characteristic window period before the occurrence of the leakage event, decompose it into N different frequency bands using wavelet packet transform, analyze the current signal data of different frequency bands, and determine the strongly correlated frequency bands of each frequency band. The second data analysis module is used to construct a frequency band sequence set for the characteristic window period, process the frequency band sequence set, and analyze and obtain the maintenance duration, multiple strongly correlated sequences and correlation duration of each frequency band sequence in the frequency band sequence set based on the strongly correlated frequency bands of each frequency band. The risk warning modeling and decision-making module is used to obtain the duration of the same frequency band sequence in multiple historical leakage events and calculate the mean to determine the average duration of the frequency band sequence. This average duration is then used as an input factor to construct a risk warning model. The model is calculated based on real-time monitoring data and used as a log for real-time synchronization and decision-making for issuing warning information.

Images