A method for intelligent analysis and checking of a substation monitoring information table

Abstract

The application discloses a kind of substation monitoring information table intelligent analysis and checking method, belong to the technical field of power system automation.This method includes: obtaining the historical monitoring information table sample of checking substation monitoring information table, signal entry is carried out word segmentation processing and named entity recognition, information element is extracted and normalized into structured record;The integrity score and consistency score of historical sample are calculated, and the reference sample set is formed by screening reference sample;The transition probability matrix is statistically divided according to the segment position of reference sample name, and the signal set should be matched according to the type of equipment and voltage level clustering to determine;Switching remote signaling opening position period is extracted to correct the coefficient linkage review;Abnormal judgment result, coefficient correction record and linkage review conclusion are output to check the result, and the sample confirmed by artificial review is incorporated into reference sample set iterative update.

Description

Technical Field

[0001] This invention relates to the field of power system automation technology, and in particular to a method for intelligent analysis and verification of substation monitoring information tables. Background Technology

[0002] The substation monitoring information table is the fundamental document for data exchange between the dispatch automation system and the substation automation system. It records key parameters such as the name, point number, and telemetry coefficient of signal entries including telemetry, remote signaling, remote control, and remote adjustment. The accuracy of the monitoring information table directly affects the reliability of the dispatch master station's monitoring and control of the substation's operating status. Currently, the verification of the monitoring information table mainly relies on manual item-by-item comparison, which is inefficient, prone to omissions, and difficult to effectively identify hidden errors (such as telemetry coefficient deviations and non-standard signal naming).

[0003] Existing methods for verifying monitoring information tables typically use fixed rule templates to check name format, without establishing a naming standardization evaluation model based on statistical learning from historical samples. When there are unconventional combinations or misaligned segments in the signal entry names, the fixed rule templates cannot make anomaly judgments based on the transition probability distribution of terms in historical samples, resulting in an insufficient detection rate for naming anomalies.

[0004] The existing monitoring information table verification method does not take into account the differences in signal configuration under different equipment types and voltage levels. It uses a uniform signal configuration list for comparison, which leads to some equipment type-specific required signals being misjudged as missing, or some low-frequency configuration signals being misjudged as redundant, resulting in a high false alarm rate for configuration anomalies.

[0005] Existing methods for verifying telemetry coefficients in monitoring information tables typically rely on theoretical calculations or manual comparisons based on experience. They fail to utilize the electrical constraints between current, voltage, and active power telemetry measurements within the same interval for coefficient inversion verification. When systematic deviations exist in the telemetry coefficients, existing methods struggle to identify them using a single telemetry value, and they lack a linkage verification mechanism based on the time period of the switch remote signaling tripping position, resulting in a significant risk of missed detections of abnormal coefficients.

[0006] The existing monitoring information table verification method lacks a dynamic screening and iterative update mechanism for benchmark samples. It does not select samples that meet both the integrity and consistency scores from historical monitoring information tables as benchmarks, nor does it automatically incorporate information table samples that have been manually verified into the benchmark sample set to update the verification criteria. As a result, the representativeness and timeliness of the benchmark samples are difficult to guarantee. Summary of the Invention

[0007] Therefore, embodiments of the present invention provide a method for intelligent analysis and verification of substation monitoring information tables. Includes the following steps: S1. Obtain the monitoring information table and historical monitoring information table sample of the substation to be checked, perform word segmentation and named entity recognition on the signal entries in each information table, and extract the information elements of each signal entry. The information elements include the bay to which it belongs, equipment name, voltage level, signal type, point number and telemetry coefficient. Each signal entry is normalized into a structured record according to its information elements; S2. Calculate the integrity score and consistency score of each historical monitoring information table sample. The consistency score is determined by the frequency of occurrence of the segment combination of each signal item in the sample in all historical monitoring information table samples. Based on the integrity score and consistency score, a benchmark sample set is formed by selecting benchmark samples from each historical monitoring information table sample. S3. Divide the names of signal entries in the benchmark sample set into segments, and calculate the transition probabilities of words in adjacent segments to obtain the transition probability matrix of the name segment sequence. The signal entries in the baseline sample set are clustered according to equipment type and voltage level. The configuration frequency of each signal type in each cluster is counted. The appropriate signal set for each type of equipment is determined based on the configuration frequency. S4. Calculate the sequence probability of each signal entry name in the monitoring information table of the substation to be checked based on the transition probability matrix, and determine the signal entries with abnormal naming. Compare the signal entries of each device in the monitoring information table of the substation to be checked with the corresponding signal set of the corresponding device type to detect missing signals, redundant signals, and misaligned segments in the signal sequence. Based on the set of signals to be matched, determine the current telemetry, voltage telemetry and active power telemetry within the same interval. Use the electrical constraint relationship between the active power telemetry value, voltage telemetry value and current telemetry value to invert the telemetry coefficients. Compare the inverted telemetry coefficients with the telemetry coefficients filled in the monitoring information table of the substation to be checked. Determine if the coefficients are abnormal and correct them with the inverted telemetry coefficients. S5. Extract the time period in which the switch remote signal is in the open position during the interval of the completed coefficient correction, and perform a linkage verification of the coefficient correction based on the comparison results of the current telemetry conversion value and the zero value interval during this time period. The zero value interval is determined based on the statistical distribution of the current telemetry conversion value during the time period in which the switch remote signal is in the open position in the benchmark sample set. Summarize the judgment results of naming anomalies, configuration anomalies, and coefficient anomalies, coefficient correction records, and linkage review conclusions, and output the verification results.

[0008] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: This invention obtains a sample of the monitoring information table and historical monitoring information table of the substation to be verified, performs word segmentation and named entity recognition on the signal entries in each information table, and extracts information elements including the bay, equipment name, voltage level, signal type, point number, and telemetry coefficient. Each signal entry is then normalized into a structured record according to the information elements. This solves the technical problems of existing technologies where the verification of monitoring information tables relies on manual item-by-item comparison, which is inefficient and prone to omissions. It achieves the technical effect of realizing structured parsing and automated processing of monitoring information tables and improving verification efficiency.

[0009] This invention calculates the integrity score and consistency score of each historical monitoring information table sample. The integrity score is determined by the ratio of the number of all non-empty signal entries to the total number of signal entries. The consistency score is determined by the frequency of occurrence of the segment combination of each signal entry in the sample in all historical samples. Based on the upper quartiles of the two scores, benchmark samples are selected to form a benchmark sample set. This solves the technical problems of the lack of a benchmark sample selection mechanism based on historical sample statistical learning and the insufficient representativeness of benchmark samples in the prior art. It achieves the technical effect of obtaining high-quality benchmark samples through dual screening of integrity score and consistency score, and improving the accuracy of subsequent verification.

[0010] This invention solves the technical problems of existing technologies that use fixed rule templates for name format checking and cannot determine anomalies based on the historical sample term transition probability distribution. It achieves the technical effect of establishing a naming norm assessment model based on historical sample statistical learning and improving the naming anomaly detection rate.

[0011] This invention compares the signal entries of each device in the information table to be verified with the corresponding signal set of the device type to detect missing signals, redundant signals, and misaligned segments in the signal sequence. It uses the electrical constraint relationship between active power telemetry values, voltage telemetry values, and current telemetry values ​​to invert telemetry coefficients. The inverted telemetry coefficients are compared with the telemetry coefficients filled in the information table to determine coefficient anomalies and corrected with the inverted values. This solves the technical problems in the prior art that the difference between device type and voltage level is not considered, resulting in a high false alarm rate of configuration anomalies, and that the coefficient inversion verification is not used with electrical constraint relationships. It achieves the technical effect of realizing adaptive signal detection based on device type and telemetry coefficient inversion verification based on electrical constraints, and improving the accuracy of configuration anomaly and coefficient anomaly detection. Attached Figure Description

[0012] Figure 1 A flowchart provided for an embodiment of this application. Detailed Implementation

[0013] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0014] like Figure 1 As shown, a method for intelligent analysis and verification of substation monitoring information tables includes the following steps: S1. Obtain the monitoring information table and historical monitoring information table sample of the substation to be checked, perform word segmentation and named entity recognition on the signal entries in each information table, and extract the information elements of each signal entry. The information elements include the bay to which it belongs, equipment name, voltage level, signal type, point number and telemetry coefficient. Each signal entry is normalized into a structured record according to its information elements; S2. Calculate the integrity score and consistency score of each historical monitoring information table sample. The consistency score is determined by the frequency of occurrence of the segment combination of each signal item in the sample in all historical monitoring information table samples. Based on the integrity score and consistency score, a benchmark sample set is formed by selecting benchmark samples from each historical monitoring information table sample. S3. Divide the names of signal entries in the benchmark sample set into segments, and calculate the transition probabilities of words in adjacent segments to obtain the transition probability matrix of the name segment sequence. The signal entries in the baseline sample set are clustered according to equipment type and voltage level. The configuration frequency of each signal type in each cluster is counted. The appropriate signal set for each type of equipment is determined based on the configuration frequency. S4. Calculate the sequence probability of each signal entry name in the monitoring information table of the substation to be checked based on the transition probability matrix, and determine the signal entries with abnormal naming. Compare the signal entries of each device in the monitoring information table of the substation to be checked with the corresponding signal set of the corresponding device type to detect missing signals, redundant signals, and misaligned segments in the signal sequence. Based on the set of signals to be matched, determine the current telemetry, voltage telemetry and active power telemetry within the same interval. Use the electrical constraint relationship between the active power telemetry value, voltage telemetry value and current telemetry value to invert the telemetry coefficients. Compare the inverted telemetry coefficients with the telemetry coefficients filled in the monitoring information table of the substation to be checked. Determine if the coefficients are abnormal and correct them with the inverted telemetry coefficients. S5. Extract the time period in which the switch remote signal is in the open position during the interval of the completed coefficient correction, and perform a linkage verification of the coefficient correction based on the comparison results of the current telemetry conversion value and the zero value interval during this time period. The zero value interval is determined based on the statistical distribution of the current telemetry conversion value during the time period in which the switch remote signal is in the open position in the benchmark sample set. Summarize the judgment results of naming anomalies, configuration anomalies, and coefficient anomalies, coefficient correction records, and linkage review conclusions, and output the verification results.

[0015] It should be noted that the consistency score in step S2, the transition probability matrix and allocation frequency in step S3, the sequence probability and coefficient inversion value in step S4, and the zero value interval in step S5 are all obtained by statistical calculation of the sample data. The calculation process is explained in the following descriptions. In step S4, the detection of misaligned segments is based on the matching signal set obtained in step S3. The inversion of telemetry coefficients is based on the synchronous sampling sequence of active power telemetry, voltage telemetry and current telemetry within the same interval. The input of each step is the output of the previous step. The electrical constraints are as follows: In a three-phase AC circuit, the active power P, line voltage U, line current I, and power factor cosφ satisfy the following equation, which is determined by circuit laws: In the formula: P, U, I have the same physical meaning as the active power telemetry value, voltage telemetry value, and current telemetry value in step S4, and cosφ is the power factor; The linkage verification in step S5 is based on the circuit characteristic that the circuit current is zero after the switch is opened. The coefficient correction results in step S4 are verified and the verification conclusion is incorporated into the check results.

[0016] As an optional embodiment, step S1 specifically involves: A signal domain dictionary covering four types of signals—telemetry, remote signaling, remote control, and remote adjustment—is constructed. The signal domain dictionary includes entries for interval names, equipment names, and signal names. The names of signal entries are sequentially segmented by forward maximum matching and backward maximum matching using a signal domain dictionary. When the forward segmentation results are inconsistent with the backward segmentation results, the segmentation result with more entries is taken as the segmentation result, thus obtaining the entry sequence. Named entity recognition is performed on each term in the term sequence, and the segment category to which each term belongs is labeled. The segment categories include voltage level segment, interval segment, equipment segment, and signal segment. Each signal entry is normalized into a structured record containing voltage level segments, interval segments, equipment segments, signal segments, and parameter fields. The parameter fields record the point number, telemetry coefficient, and alarm level.

[0017] It should be noted that: forward maximum matching segmentation starts from the beginning of the name and matches the longest word in the signal domain dictionary word by word, while reverse maximum matching segmentation starts from the end of the name and performs the same matching in reverse. When the two segmentation results are inconsistent, the incorrectly segmented long terms will generate more short terms. The segmentation result with more terms has a more complete coverage of the signal domain dictionary. Therefore, the segmentation result with more terms is taken as the segmentation result. The execution method of named entity recognition is as follows: using the historical monitoring information table that has been verified to be correct on-site, each word is labeled with its segment category, the labeled corpus is used to train the sequence labeling model, the sequence of words to be identified is input into the sequence labeling model, and the segment category of each word is output. An example of a structured record format is as follows: voltage level segment 110kV, bay segment I busbar, equipment segment PT, signal segment A phase voltage, parameter field records point number and telemetry coefficient; The segment segmentation and transition statistics in step S3 and the sequence probability calculation in step S4 are both performed on the normalized five-segment structure.

[0018] As an optional embodiment, in step S2, the integrity score is denoted as C and calculated using the following formula: Where: m is the number of signal entries in a single sample where all information elements are non-empty; M is the total number of signal entries in a single sample; The consistency score, denoted as N, is calculated using the following formula: In the formula: ρ j The relative frequency of the segment combination of the j-th signal entry in the sample is calculated using the following formula: Where: g j This represents the number of times the segment combination appears in the signal entries of all historical monitoring information table samples; G represents the total number of signal entries in all historical monitoring information table samples; The integrity score and consistency score of each historical monitoring information table sample are sorted in ascending order. The upper quartile of the integrity score distribution is taken as the integrity benchmark threshold, and the upper quartile of the consistency score distribution is taken as the consistency benchmark threshold. Historical monitoring information samples in which both types of scores are not lower than the corresponding benchmark threshold are used as benchmark samples.

[0019] It should be noted that the calculation steps for the consistency score are as follows: First, take the combination of terms from the voltage level segment, interval segment, equipment segment, and signal segment in the structured record of each signal entry as the segment combination for that entry; Second, count the number of times (g) the segment combination appears in the signal entries of all historical monitoring information table samples. jDivide by the total number of signal entries G to obtain the relative frequency ρ. j ; Third, the arithmetic mean of the relative frequencies of the M signal items in the sample is taken to obtain the consistency score of the sample. The upper quartile is calculated as follows: sort the Z rating values ​​in ascending order, take the value at position 0.75×(Z+1), and if the position is not an integer, linearly interpolate the values ​​of the two adjacent positions. Z is the number of rating values. Example: After sorting 100 rating values ​​in ascending order, the calculated position is 75.75. The 75th and 76th rating values ​​are then interpolated using weighted averages of 0.25 and 0.75, respectively.

[0020] As an optional embodiment, in step S3, the transition probability matrix is ​​obtained in the following way: The names of each signal entry in the benchmark sample set are segmented into segment sequences according to segment category; Count the number of occurrences of any word in combination with words in its adjacent segments, and the transition probability q of word u appearing in the adjacent segment. uv Calculate using the following formula: Where: n uv This represents the number of combinations of word v appearing in adjacent segments after word u in the baseline sample set. n u The total number of times term u appears in the preceding segment of the baseline sample set; The transition probability matrix of the name segment sequence is constructed using the transition probabilities between each pair of terms.

[0021] It should be noted that the segment sequence consists of voltage level segment, interval segment, equipment segment, and signal segment arranged in that order. Statistical example: In the benchmark sample set, the number of times the term "PT" appears in the adjacent equipment segment after the busbar of segment I in the interval segment is 86. The term "busbar of segment I" appears as an interval segment term 90 times. The transition probability of this term pair is 86 / 90. Counting and normalizing are performed on all terms in the benchmark sample set one by one to obtain the transition probability matrix. The size of the matrix is ​​determined by the actual number of terms appearing in the benchmark sample set.

[0022] As an optional embodiment, in step S4, the sequence probability p is calculated using the following formula: In the formula: q i Let i be the transition probability corresponding to the i-th adjacent segment transition in the signal entry name, where i ranges from 1 to L; L represents the number of segment transitions for the signal entry name; probability threshold p* Take the lower quartile of the probability distribution of all signal entry name sequences in the benchmark sample set; Signal entries with a sequence probability lower than the probability threshold are identified as signal entries with abnormal naming. The location of the adjacent segment pair with the lowest transition probability in the signal entry name is output as the abnormal segment location.

[0023] It should be noted that: the number of rank transfers L is equal to the number of ranks in the name minus one; The sequence probability is the Lth root of the product of the transition probabilities, used to normalize the probabilities of sequences with different names to the same numerical range. probability threshold p * The calculation method is as follows: sort the sequence probabilities of all signal entries in the reference sample set in ascending order, take the value at position 0.25×(Z+1), and when the position is not an integer, interpolate linearly between two adjacent positions. Z is the number of sequence probability values. The abnormal segment location is determined by finding the adjacent segment transition with the lowest transition probability in the signal entry name. The two segments involved in this transition are the abnormal segment locations. Example: The probability of a name transitioning from the device segment to the signal segment is the minimum among all transitions of that name. The anomaly is located in the combination of the device segment and the signal segment.

[0024] As an optional embodiment, in step S3, the method for determining the appropriate signal set is as follows: Signal entries with the same device type and voltage level are grouped into the same cluster; The configuration frequency f of each signal type within a statistical cluster s f s Calculate using the following formula: Where: H s The number of reference samples configured with signal type s within a cluster; D is the total number of reference samples within the cluster; Arrange the signal types within each cluster in descending order of their configuration frequencies, and calculate the difference Δf between adjacent configuration frequencies using the following formula. s : In the formula: f s+1 The configuration frequency for the signal type that is one position after signal type s in descending order; Take Δf s The position with the highest frequency is used as the frequency boundary position. Signal types with a configured frequency not lower than the configured frequency at the frequency boundary position are classified into the appropriate signal set for this type of device.

[0025] It should be noted that the criteria for clustering are that the equipment types are the same and the voltage levels are the same. Statistical example: A cluster of 110kV line bays contains 50 baseline samples, of which 48 samples are equipped with circuit breaker location remote signaling. The configuration frequency of this signal type is 48 / 50, or 0.96. The process of determining the frequency boundary position is as follows: After arranging the configuration frequencies of each signal type in the cluster in descending order, calculate the difference between adjacent configuration frequencies one by one, and the position with the largest difference is the frequency boundary position. The configuration frequency of each signal type in the matching signal set shall not be lower than the configuration frequency at the frequency boundary position, and signal types after the boundary position shall not be included in the matching signal set.

[0026] As an optional embodiment, in step S4, detecting missing signals, redundant signals, and misaligned segments in the signal sequence specifically involves: Missing signal detection: Record the signal types that are not present in the signal entries of the equipment to be checked from the corresponding equipment type's required signal set into the missing signal list; Redundancy detection: Record the signal types that appear in the signal entries of the device to be verified and whose configured frequency is lower than the configured frequency at the frequency boundary position into the redundant signal list; Pairing verification: Within the same cluster, count the co-occurrence combinations of remote control signals and remote signaling signals that appear simultaneously in the same reference sample. When the device to be verified has a remote control signal but lacks the corresponding remote signaling signal in the co-occurrence combination, or when the point numbers of the corresponding remote control signals and remote signaling signals are not continuous, the pairing is determined to be abnormal. Misalignment detection: The sequence of signal entries of the device to be checked is compared bit by bit with the arrangement of signal types in the matching signal set. When two or more consecutive signal entries are offset from the matching signal set sequence in the same direction and by the same amount, they are determined to be misaligned segments. The range and offset of the misaligned segments are output.

[0027] It should be noted that the correspondence in the co-occurrence combination is as follows: the remote control closing signal corresponds to the remote signal closing position signal, and the remote control opening signal corresponds to the remote signal opening position signal. The step-by-step comparison process for misalignment detection is as follows: starting from the first entry of the signal entry sequence of the device to be checked, the signal type is compared with the corresponding signal set sequence bit by bit, and the offset direction and offset quantity of each bit are recorded. When two or more consecutive entries have the same offset direction and offset amount, the consecutive segment is determined to be a misaligned segment; The offset of a single entry is included in the missing signal list or the redundant signal list, but is not involved in the misalignment determination.

[0028] As an optional embodiment, in step S4, the inversion and correction of the telemetry coefficients specifically involves: Determine the current telemetry, voltage telemetry, and active power telemetry within the same interval based on the appropriate signal set; Collect the synchronous sampling sequence of each telemetry device in the interval during the reference operating period. The reference operating period is the normal operating period of the substation to which the interval belongs after the monitoring information table has been verified to be correct by on-site transmission. The target current value at sampling time t is calculated using the following formula. t : In the formula: P t The active telemetry value at sampling time t; U t The voltage telemetry value at sampling time t; μ r For each sampling time r within the reference runtime period t The arithmetic mean, r t Calculate using the following formula: In the formula: I t The current telemetry value at sampling time t; Calculate the coefficient inversion value k using the following formula * : In the formula: x t The code value of the current telemetry at sampling time t; Î t The target current value at sampling time t; T is the total number of sampling times; The relative deviation δ of the coefficient is calculated using the following formula: In the formula: k0 is the telemetry coefficient filled in the monitoring information table of the substation to be checked; The baseline operating period is divided into time periods by natural days. The relative fluctuation of the coefficient inversion value relative to k0 is calculated for each day. The upper quartile of the relative fluctuation distribution for each day is taken as the deviation threshold δ. * ; When δ is greater than δ * The time-determining coefficient is abnormal, and k is used to determine the abnormality. * Replace k0 to complete the correction.

[0029] It should be noted that: the coefficient inversion value k * The solution process is as follows: the telemetry converted value is equal to the product of the code value and the coefficient. An expression is established for the sum of squared residuals between the target current value and the converted current value at each sampling time. The coefficient is differentiated, and the derivative is set to zero to obtain k. * It equals the sum of the products of the code value and the target current value at each time point, divided by the sum of the squares of the code values ​​at each time point; μr It reflects the power factor level of this interval during the reference operating period; Deviation threshold δ * The calculation steps are as follows: Divide the reference running period into natural days, calculate the relative deviation of the coefficient inversion value of the day relative to k0, arrange the relative deviations of each day in ascending order, take the value at position 0.75×(Z+1), and when the position is not an integer, perform linear interpolation between two adjacent positions, where Z is the number of relative deviation values. The correction method is based on k * Replace the telemetry coefficient k0 entered in the monitoring information table of the substation to be verified.

[0030] As an optional embodiment, in step S5, the zero-value interval is determined by the following formula: In the formula: μ0 is the mean of the current telemetry conversion value sequence within the time period of the reference sample centralized switch telemetry position. σ0 is the standard deviation of the sequence; The method for extracting the time period of the opening position is as follows: read the change record of the remote signal of the switch, take the time when the remote signal value changes to the opening position as the start of the time period, and take the time when the remote signal value changes to the closing position as the end of the time period. When the current telemetry conversion value of the interval that has completed coefficient correction is greater than the upper limit of the zero value interval during the opening position period, it is determined that there is a telemetry point number misalignment or telemetry association error in the interval. When the converted value of the current telemetry is not greater than the upper limit of the zero range, the correction coefficient is confirmed to be effective.

[0031] It should be noted that the calculation process for the upper limit of the zero value interval is as follows: summarize the current telemetry conversion values ​​of all switching periods in the benchmark sample set, calculate the mean μ0 and standard deviation σ0 of the sequence, and add three times the standard deviation to the mean to obtain the upper limit of the interval. This interval covers the distribution range of measurement noise and line capacitance current. The verification and judgment process is as follows: After the coefficient correction is completed, the current telemetry conversion value of the opening period of the interval is extracted. When the conversion value is less than or equal to the upper limit of the zero value interval, the coefficient correction is confirmed to be effective. When the conversion value is greater than the upper limit of the zero value interval, the telemetry point number is determined to be misaligned or the telemetry association is incorrect. After the switch is opened, the circuit current is zero. During the opening period, the telemetry value of the current is greater than the upper limit of the zero value range, indicating that the remote signaling of the switch position does not correspond to the telemetry of the current.

[0032] As an optional embodiment, in step S5, the output verification result is specifically as follows: The verification results are classified according to the degree of impact of the anomalies on power grid monitoring. Pairing anomalies and remote signaling point number misalignment are recorded as Level 1 anomalies, coefficient anomalies and misaligned segments are recorded as Level 2 anomalies, and naming anomalies and redundant signals are recorded as Level 3 anomalies. Generate a verification form, which records the anomaly level, the signal entry where the anomaly occurred, the anomaly type, the coefficient correction value, and the linkage review conclusion; The information table samples that have been manually verified and confirmed to be correct are incorporated into the benchmark sample set. The integrity score, consistency score, transition probability matrix, configuration frequency and each distribution threshold are recalculated according to the calculation process of steps S2 to S4 to complete the iterative update of the benchmark sample set and the verification criteria.

[0033] It should be noted that the impact of anomalies is classified as follows: Level 1 anomalies involve remote control operation and switch position monitoring; Level 2 anomalies involve the correspondence between remote measurement and information; and Level 3 anomalies involve the standardization of information table writing. The verification sheet records the anomaly level, anomaly type, coefficient correction value, and linkage review conclusion for each signal item; The execution process of iterative update is as follows: the information table samples that have been manually verified and confirmed to be correct are merged into the benchmark sample set, and the statistical calculations of steps S2 to S4 are re-executed. The integrity benchmark threshold, consistency benchmark threshold, probability threshold, frequency boundary position and deviation threshold are recalculated according to the expanded sample distribution.

[0034] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for intelligent analysis and verification of substation monitoring information tables, characterized in that, Includes the following steps: S1. Obtain the monitoring information table and historical monitoring information table sample of the substation to be checked, perform word segmentation and named entity recognition on the signal entries in each information table, and extract the information elements of each signal entry. The information elements include the bay to which it belongs, equipment name, voltage level, signal type, point number and telemetry coefficient. Each signal entry is normalized into a structured record according to its information elements; S2. Calculate the integrity score and consistency score of each historical monitoring information table sample. The consistency score is determined by the frequency of occurrence of the segment combination of each signal item in the sample in all historical monitoring information table samples. Based on the integrity score and consistency score, a benchmark sample set is formed by selecting benchmark samples from each historical monitoring information table sample. S3. Divide the names of signal entries in the benchmark sample set into segments, and calculate the transition probabilities of words in adjacent segments to obtain the transition probability matrix of the name segment sequence. The signal entries in the baseline sample set are clustered according to equipment type and voltage level. The configuration frequency of each signal type in each cluster is counted. The appropriate signal set for each type of equipment is determined based on the configuration frequency. S4. Calculate the sequence probability of each signal entry name in the monitoring information table of the substation to be checked based on the transition probability matrix, and determine the signal entries with abnormal naming. Compare the signal entries of each device in the monitoring information table of the substation to be checked with the corresponding signal set of the corresponding device type to detect missing signals, redundant signals, and misaligned segments in the signal sequence. Based on the set of signals to be matched, determine the current telemetry, voltage telemetry and active power telemetry within the same interval. Use the electrical constraint relationship between the active power telemetry value, voltage telemetry value and current telemetry value to invert the telemetry coefficients. Compare the inverted telemetry coefficients with the telemetry coefficients filled in the monitoring information table of the substation to be checked. Determine if the coefficients are abnormal and correct them with the inverted telemetry coefficients. S5. Extract the time period in which the switch remote signal is in the open position during the interval of the completed coefficient correction, and perform a linkage verification of the coefficient correction based on the comparison results of the current telemetry conversion value and the zero value interval during this time period. The zero value interval is determined based on the statistical distribution of the current telemetry conversion value during the time period in which the switch remote signal is in the open position in the benchmark sample set. Summarize the judgment results of naming anomalies, configuration anomalies, and coefficient anomalies, coefficient correction records, and linkage review conclusions, and output the verification results.

2. The intelligent analysis and verification method for substation monitoring information tables according to claim 1, characterized in that, Step S1 is as follows: A signal domain dictionary covering four types of signals—telemetry, remote signaling, remote control, and remote adjustment—is constructed. The signal domain dictionary includes entries for interval names, equipment names, and signal names. The names of signal entries are sequentially segmented by forward maximum matching and backward maximum matching using a signal domain dictionary. When the forward segmentation results are inconsistent with the backward segmentation results, the segmentation result with more entries is taken as the segmentation result, thus obtaining the entry sequence. Named entity recognition is performed on each term in the term sequence, and the segment category to which each term belongs is labeled. The segment categories include voltage level segment, interval segment, equipment segment, and signal segment. Each signal entry is normalized into a structured record containing voltage level segments, interval segments, equipment segments, signal segments, and parameter fields. The parameter fields record the point number, telemetry coefficient, and alarm level.

3. The intelligent analysis and verification method for substation monitoring information tables according to claim 2, characterized in that, In step S2, the integrity score is denoted as C and calculated using the following formula: Where: m is the number of signal entries in a single sample where all information elements are non-empty; M is the total number of signal entries in a single sample; The consistency score, denoted as N, is calculated using the following formula: In the formula: ρ j The relative frequency of the segment combination of the j-th signal entry in the sample is calculated using the following formula: Where: g j This represents the number of times the segment combination appears in the signal entries of all historical monitoring information table samples; G represents the total number of signal entries in all historical monitoring information table samples; The integrity score and consistency score of each historical monitoring information table sample are sorted in ascending order. The upper quartile of the integrity score distribution is taken as the integrity benchmark threshold, and the upper quartile of the consistency score distribution is taken as the consistency benchmark threshold. Historical monitoring information samples in which both types of scores are not lower than the corresponding benchmark threshold are used as benchmark samples.

4. The intelligent analysis and verification method for substation monitoring information tables according to claim 3, characterized in that, In step S3, the transition probability matrix is ​​obtained as follows: The names of each signal entry in the benchmark sample set are segmented into segment sequences according to segment category; Count the number of occurrences of any word in combination with words in its adjacent segments, and the transition probability q of word u appearing in the adjacent segment. uv Calculate using the following formula: Where: n uv This represents the number of combinations of word v appearing in adjacent segments after word u in the baseline sample set. n u The total number of times term u appears in the preceding segment of the baseline sample set; The transition probability matrix of the name segment sequence is constructed using the transition probabilities between each pair of terms; Counting and normalizing are performed on all terms in the benchmark sample set one by one to obtain the transition probability matrix. The size of the matrix is ​​determined by the actual number of terms appearing in the benchmark sample set.

5. The intelligent analysis and verification method for substation monitoring information tables according to claim 4, characterized in that, In step S4, the sequence probability is calculated using the following formula: In the formula: Let q be the sequence probability. i Let i be the transition probability corresponding to the i-th adjacent segment transition in the signal entry name, where i ranges from 1 to L; L represents the number of segment transitions for the signal entry name; probability threshold p * Take the lower quartile of the probability distribution of all signal entry name sequences in the benchmark sample set; Signal entries with a sequence probability lower than the probability threshold are identified as signal entries with abnormal naming. The location of the adjacent segment pair with the lowest transition probability in the signal entry name is output as the abnormal segment location.

6. The intelligent analysis and verification method for substation monitoring information tables according to claim 3, characterized in that, In step S3, the method for determining the appropriate signal set is as follows: Signal entries with the same device type and voltage level are grouped into the same cluster; The configuration frequency f of each signal type within a statistical cluster s f s Calculate using the following formula: Where: H s The number of reference samples configured with signal type s within a cluster; D is the total number of reference samples within the cluster; Arrange the signal types within each cluster in descending order of their configuration frequencies, and calculate the difference Δf between adjacent configuration frequencies using the following formula. s : In the formula: f s+1 The configuration frequency for the signal type that is one position after signal type s in descending order; Take Δf s The position with the highest frequency is used as the frequency boundary position. Signal types with a configured frequency not lower than the configured frequency at the frequency boundary position are classified into the appropriate signal set for this type of device.

7. The intelligent analysis and verification method for substation monitoring information tables according to claim 6, characterized in that, In step S4, the detection of missing signals, redundant signals, and misaligned segments in the signal sequence specifically involves: Missing signal detection: Record the signal types that are not present in the signal entries of the equipment to be checked from the corresponding equipment type's required signal set into the missing signal list; Redundancy detection: Record the signal types that appear in the signal entries of the device to be verified and whose configured frequency is lower than the configured frequency at the frequency boundary position into the redundant signal list; Pairing verification: Within the same cluster, count the co-occurrence combinations of remote control signals and remote signaling signals that appear simultaneously in the same reference sample. When the device to be verified has a remote control signal but lacks the corresponding remote signaling signal in the co-occurrence combination, or when the point numbers of the corresponding remote control signals and remote signaling signals are not continuous, the pairing is determined to be abnormal. Misalignment detection: The sequence of signal entries of the device to be checked is compared bit by bit with the arrangement of signal types in the matching signal set. When two or more consecutive signal entries are offset from the matching signal set sequence in the same direction and by the same amount, they are determined to be misaligned segments. The range and offset of the misaligned segments are output.

8. The intelligent analysis and verification method for substation monitoring information tables according to claim 7, characterized in that, In step S4, the inversion and correction of the telemetry coefficients specifically involves: Determine the current telemetry, voltage telemetry, and active power telemetry within the same interval based on the appropriate signal set; Collect the synchronous sampling sequence of each telemetry device in the interval during the reference operating period. The reference operating period is the normal operating period of the substation to which the interval belongs after the monitoring information table has been verified to be correct by on-site transmission. The target current value at sampling time t is calculated using the following formula. t : In the formula: P t The active telemetry value at sampling time t; U t The voltage telemetry value at sampling time t; μ r For each sampling time r within the reference runtime period t The arithmetic mean, r t Calculate using the following formula: In the formula: I t The current telemetry value at sampling time t; Calculate the coefficient inversion value k using the following formula. * : In the formula: x t The code value of the current telemetry at sampling time t; Î t The target current value at sampling time t; T is the total number of sampling times; The relative deviation δ of the coefficient is calculated using the following formula: In the formula: k0 is the telemetry coefficient filled in the monitoring information table of the substation to be checked; The baseline operating period is divided into time periods by natural days. The relative fluctuation of the coefficient inversion value relative to k0 is calculated for each day. The upper quartile of the relative fluctuation distribution for each day is taken as the deviation threshold δ. * ; When δ is greater than δ * The time-determining coefficient is abnormal, and k is used to determine the abnormality. * Replace k0 to complete the correction.

9. The intelligent analysis and verification method for substation monitoring information tables according to claim 8, characterized in that, In step S5, the zero-value interval is determined by the following formula: In the formula: μ0 is the mean of the current telemetry conversion value sequence within the time period of the reference sample centralized switch telemetry position. σ0 is the standard deviation of the sequence; The method for extracting the time period of the opening position is as follows: read the change record of the remote signal of the switch, take the time period starting when the remote signal value changes to the opening position, and take the time period ending when the remote signal value changes to the closing position. When the current telemetry conversion value of the interval that has completed coefficient correction is greater than the upper limit of the zero value interval during the opening position period, it is determined that there is a telemetry point number misalignment or telemetry association error in the interval. When the converted value of the current telemetry is not greater than the upper limit of the zero range, the correction coefficient is confirmed to be effective.

10. The intelligent analysis and verification method for substation monitoring information tables according to claim 9, characterized in that, In step S5, the output verification result is specifically as follows: The verification results are classified according to the degree of impact of the anomalies on power grid monitoring. Pairing anomalies and remote signaling point number misalignment are recorded as Level 1 anomalies, coefficient anomalies and misaligned segments are recorded as Level 2 anomalies, and naming anomalies and redundant signals are recorded as Level 3 anomalies. Generate a verification form, which records the anomaly level, the signal entry where the anomaly occurred, the anomaly type, the coefficient correction value, and the linkage review conclusion; The information table samples that have been manually verified and confirmed to be correct are incorporated into the benchmark sample set. The integrity score, consistency score, transition probability matrix, configuration frequency and each distribution threshold are recalculated according to the calculation process from step S2 to step S4, and the benchmark sample set and verification criteria are iteratively updated.

Images