Early warning processing method for pole-mounted circuit breaker reclosing control data

Abstract

The application belongs to the technical field of power distribution network, and particularly relates to a pre-warning processing method for reclosing control data of a pole-mounted circuit breaker, comprising the following steps: obtaining reclosing control data of a target pole-mounted circuit breaker, extracting an action time, a reclosing interval, a steady-state recovery time length and a reclosing result marker, and constructing a cross-action time sequence matrix in chronological order; based on the cross-action time sequence matrix, respectively calculating a trend slope of the reclosing interval sequence, a trend slope of the steady-state recovery time length sequence and a trend slope of a sliding window success rate sequence, and combining to obtain a three-dimensional deterioration feature vector; and extracting historical feature data of a normal operation period of the target pole-mounted circuit breaker to construct an individualized baseline parameter set. The application can effectively identify an implicit deterioration mode of continuous deterioration of the cross-action time sequence trend, and advance the pre-warning opportunity to before the occurrence of a permanent fault.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network technology, specifically relating to a method for early warning processing of reclosing control data of pole-mounted circuit breakers. Background Technology

[0002] Pole-mounted circuit breakers are widely deployed automated switching devices in 10kV distribution lines. Their reclosing function is a standard means of ensuring the continuity of power supply to the distribution network. When a transient fault occurs on the line, the circuit breaker trips and then automatically recloses according to preset logic to restore power supply. During each reclosing operation, the circuit breaker terminal generates control data covering dimensions such as the timing of the operation, the current recovery pattern, the reclosing interval, and the result of the operation, and uploads this data to the master station system.

[0003] Patent application CN109904927A discloses a method and device for the operation and maintenance of a pole-mounted switch, including: acquiring operation and maintenance parameters of the pole-mounted switch through a sensor group; sending the operation and maintenance parameters to an FTU; acquiring the operation and maintenance parameters of the FTU; and determining the operation and maintenance status of the pole-mounted switch based on the operation and maintenance parameters. This method eliminates the need for manual inspection and monitoring of the pole-mounted switch. Instead, it utilizes a sensor group to acquire the operation and maintenance parameters. Since acquiring operation and maintenance parameters through a sensor group and transmitting them through the FTU is more convenient than manual methods, the operation and maintenance parameters acquired by the sensor group are more comprehensive and in-depth than those obtained through manual inspection or monitoring. Furthermore, since the operation and maintenance parameters are acquired under the same testing standards, they are more objective and reliable than those obtained through manual inspection or monitoring.

[0004] For early warning processing of reclosing control data, existing technologies typically set fixed or statistical adaptive thresholds for control data of a single reclosing action. An alarm is triggered when the current value, voltage value, or number of reclosing failures exceeds the threshold for a given action. Some systems introduce anomaly detection based on historical averages, comparing the current single-action data with recent historical averages; if the difference exceeds a set multiple, an alarm is issued. All of the above solutions use a snapshot of single-action data as the judgment unit, belonging to a static, point-by-point comparison framework.

[0005] Because in the critical evolution stage of power distribution lines tending towards permanent faults, the data values ​​of a single reclosing action are often still within the threshold range. If early warning is based only on a single snapshot, the continuous shrinking trend of the reclosing interval between adjacent reclosing actions, the monotonically increasing drift of the time required for the line current to recover to a steady state, and the decreasing trend of the reclosing power within the sliding window will be ignored. This will cause the system to be unable to detect the hidden signals of equipment and line degradation, thus missing the early warning window period before permanent faults. Summary of the Invention

[0006] This invention provides an early warning processing method for reclosing control data of pole-mounted circuit breakers, in order to solve the technical problem that existing early warning methods for reclosing control data cannot detect the gradual deterioration trend across the action time sequence because they use snapshots of single action data as the judgment unit.

[0007] In a first aspect, the present invention provides a method for early warning processing of reclosing control data of pole-mounted circuit breakers, comprising the following steps: S1. Obtain the reclosing control data of the target pole-mounted circuit breaker, extract the action time, reclosing interval, steady-state recovery time and reclosing result mark, and construct a cross-action timing matrix in chronological order. S2, based on the cross-action time series matrix, calculate the trend slope of the overlapping interval sequence, the trend slope of the steady-state recovery time sequence, and the trend slope of the sliding window success rate sequence, and combine them to obtain a three-dimensional degradation feature vector. S3. Extract historical feature data of the target pole-mounted circuit breaker during normal operation to construct a personalized baseline parameter set. Combine the deviation of each trend slope in the three-dimensional degradation feature vector from the physical direction of degradation to calculate the comprehensive degradation score. S4, in response to the update of the cross-action time sequence matrix, recalculates the comprehensive degradation score to form a score time series, calculates the score growth rate and acceleration of the score time series, and triggers the corresponding graded warning output in combination with the preset graded warning boundary.

[0008] Its effect is that by reconstructing the dispersed reclosing control data into a cross-action timing matrix, and by extracting the reclosing interval shrinkage slope, steady-state recovery time growth slope, and sliding window success rate decline slope, the reclosing operation status is comprehensively evaluated. This ensures the accuracy of early warning while effectively identifying the hidden degradation mode of the cross-action timing trend, so that the early warning time can be moved forward to before the occurrence of permanent faults.

[0009] Furthermore, the steady-state recovery time is extracted, including: Multiple consecutive sampling points are selected as the steady-state confirmation window; If the maximum deviation of the current value within the steady-state confirmation window does not exceed a preset proportion of the rated current, the line is determined to have entered a steady state and the timing is terminated. The steady-state recovery time is defined as the time it takes for the line current to recover to a steady state from the moment of action after reclosing.

[0010] Its effect is that by selecting multiple consecutive sampling points as steady-state confirmation windows and adaptively capturing the steady-state starting point, it avoids deviation of the steady-state recovery time estimation from the true value due to improper selection of the truncation time, thereby improving the accuracy of feature extraction.

[0011] Furthermore, a cross-action time sequence matrix is ​​constructed in chronological order, including: Going back from the current moment to the past, extract the number of records of all overlapping actions that have occurred within a preset time span; If the number of actions within the preset time span is less than the preset lower limit, then extend forward until the preset lower limit is met; The action time, overlap interval, steady-state recovery time, and overlap result markers corresponding to the number of records are arranged in rows to obtain the cross-action time sequence matrix.

[0012] Its effects are as follows: It provides adaptive time series matrix construction rules, first backtracks to overlapping records of a fixed duration, and expands forward to meet the lower limit of the number of actions when the data is insufficient. Then, it organizes multiple features in chronological order to form a structured matrix, which not only ensures that the time series analysis has a sufficient sample size to avoid the distortion of trend fitting, but also unifies the data storage format of overlapping actions in multiple rounds, providing a standardized data carrier for subsequent batch extraction of time series sequences, sliding window calculation, and linear slope fitting, and avoiding the problem of deterioration trend identification failure caused by missing data fragments and disordered sorting.

[0013] Further, the trend slope of the sliding window success rate sequence is calculated, including: The sliding window size is determined based on the number of rows in the cross-action timing matrix; Slide the cross-action time sequence matrix at a preset step size, calculate the recombining power within each sliding window, and obtain the sliding window success rate sequence; By fitting a linear slope to the success rate sequence of the sliding window, the trend slope of the success rate sequence of the sliding window is finally obtained.

[0014] Its effects are as follows: To address the issue of high noise from discrete 0 / 1 markings in the overlap results, a sliding window matching the number of matrix rows is set, and the window overlap success rate is calculated by sliding the window with a fixed step size. Then, a linear slope is fitted to the smoothed success rate sequence. The data fluctuations caused by a single random overlap failure are filtered out by the window mean, smoothing the fitting interference brought by discrete binary data. This makes the slope result of the success rate decline trend more consistent with the actual degradation process of the line, and improves the characterization ability of the equipment overlap reliability dimension in the three-dimensional degradation feature vector.

[0015] Furthermore, historical characteristic data of the target pole-mounted circuit breaker during normal operation are extracted to construct a personalized baseline parameter set, including: Extract the normal operating periods during which no permanent faults occurred from the historical operating data of the target pole-mounted circuit breaker; Calculate the historical mean and historical standard deviation of the overlap interval sequence, steady-state recovery time sequence, and sliding window success rate sequence during normal operation periods, respectively. By combining historical mean and historical standard deviation, a personalized set of baseline parameters is obtained.

[0016] Its effect is as follows: Based on the normal historical data of the target circuit breaker without permanent faults, the historical mean and standard deviation of the three types of time-series slopes are calculated as exclusive baselines. This is different from the globally unified fixed threshold. It fully adapts to the individual differences of the circuit breaker's own operating conditions, line load, and fault frequency characteristics, eliminates the baseline judgment deviation caused by individual equipment differences, and makes the subsequent calculation of deviation and comprehensive deterioration score fit the normal operating level of the equipment, greatly reducing false alarms caused by fixed thresholds.

[0017] Furthermore, by combining the deviation of each trend slope in the three-dimensional degradation feature vector from the physical direction of degradation, a comprehensive degradation score is calculated, including: Determine whether the trend slope of the overlapping interval sequence, the trend slope of the steady-state recovery time sequence, and the trend slope of the sliding window success rate sequence conform to their respective physical deterioration directions. For the trend slope that conforms to the direction of physical degradation, the deviation is calculated using the historical mean and historical standard deviation corresponding to the personalized baseline parameter set; The deviations are summed up to obtain the final comprehensive degradation score.

[0018] Its effect is that by introducing directional indicator constraints into the improved normalized deviation superposition score, the scoring interference caused by occasional data fluctuations shifting towards the non-deterioration direction is eliminated, and the ability of the comprehensive score to reflect the true deterioration trend is improved.

[0019] Furthermore, the rate of increase and acceleration of the rating time series are calculated, including: Calculate the first-order difference sequence of scores between adjacent batches over the recent consecutive time series of scores; The rate of increase in scores is obtained by averaging the first-order difference sequence. The linear trend slope is calculated for the first-order difference sequence, and the acceleration of the scoring time series is finally obtained.

[0020] Its effects are as follows: by comprehensively analyzing the first-order difference mean of the degradation score time series, the growth rate is obtained; by fitting the slope of the difference sequence, the degradation acceleration is obtained; the speed of equipment degradation and the trend of accelerated deterioration are quantitatively characterized; no longer relying solely on a single static score value to judge the risk of failure, it can capture the key feature of sudden acceleration of the degradation process, provide a basis for judging dynamic changes in the dimension of graded early warning, and improve the response sensitivity to rapidly deteriorating lines.

[0021] Furthermore, the steps for obtaining the preset graded early warning boundaries include: A sample set is constructed by extracting the comprehensive degradation score time series of events in history that have been confirmed to have permanent failures during the period preceding the failure. The first preset quantile of the comprehensive degradation score of the sample set in the early stage before the failure is taken as the lower boundary of the yellow warning. The median of the comprehensive degradation score of the sample set in the early and middle stages of the fault is taken as the lower boundary of the orange warning. The second preset quantile of the comprehensive degradation score of the sample set in the pre-fault critical stage is taken as the lower boundary of the red warning.

[0022] Furthermore, based on preset tiered early warning boundaries, corresponding tiered early warning outputs are triggered, including: A yellow alert is triggered when the overall degradation score exceeds the lower boundary of a yellow alert and the score growth rate is greater than 0. An orange alert is triggered when the overall degradation score exceeds the lower boundary of the orange alert and the acceleration is greater than 0. A red alert is triggered when the overall degradation score exceeds the lower boundary of the red alert and the increment of a single batch is greater than the judgment benchmark.

[0023] Its effect is that by using the growth rate and acceleration of the score sequence as the criteria for triggering graded warnings, it has higher sensitivity to the acceleration phase of the deterioration process compared to the judgment method that simply relies on the absolute value of the score, thus achieving accurate classification of warning levels.

[0024] Furthermore, after triggering the corresponding tiered early warning output, it also includes: The warning results, the current comprehensive degradation score, the contribution ratio of each trend slope in the three-dimensional degradation feature vector, and the acceleration are packaged into the warning message. The early warning message is sent to the main station system.

[0025] The beneficial effects of this invention are: Breaking away from the limitations of traditional static judgments based on snapshots of single-action data, this method constructs a cross-action time-series matrix and extracts multi-dimensional trend slopes, enabling dynamic perception of the gradual deterioration trend of power distribution lines. This time-series feature extraction method can capture the hidden evolutionary patterns between adjacent overlapping actions.

[0026] Meanwhile, this invention combines personalized baseline parameter sets with directional constraints to calculate a comprehensive degradation score, and uses the growth rate and acceleration of the score to trigger graded early warnings. This not only effectively filters out false alarms caused by occasional disturbances, but also significantly improves the response sensitivity to the degradation acceleration stage, providing reliable technical support for the preventive maintenance of the distribution network. Attached Figure Description

[0027] Figure 1 This is a flowchart of the early warning processing method for reclosing control data of pole-mounted circuit breakers.

[0028] Figure 2 This is a time-series comparison chart of the comprehensive degradation scoring of existing methods and the method of this invention. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] An embodiment of the early warning processing method for reclosing control data of pole-mounted circuit breakers provided by the present invention: like Figure 1 As shown, the method for processing early warning data of reclosing control data of pole-mounted circuit breakers includes the following steps: S1: Acquire control data and construct a cross-action timing matrix.

[0031] For the target pole-mounted circuit breaker, the original data source is the control record generated by each reclosing action within its historical operating cycle. From the... Four features are extracted sequentially from the control data of the secondary overlapping motion: motion timing. The interval between the previous action and the current action. The unit is minutes; the time it takes for the line current to recover to a steady state from the moment of action after reclosing. The unit is seconds; and the current overlap result is marked. If the overlap is successful, take 1; if it fails, take 0.

[0032] against The steady-state determination adopts a fixed window confirmation method. The number of sampling points is determined based on the power frequency sampling frequency and transient component decay time constant of the 10kV distribution line. For example, the number of sampling points can be 10. Within the steady-state confirmation window composed of continuous sampling points, if the maximum deviation of the current value within the window does not exceed a preset proportion of the rated current, it is determined that the line has entered a steady state and the timing is terminated. The time taken for the line current to recover to a steady state from the moment of action after reclosing is taken as the steady-state recovery time. For example, this preset proportion can be 2%. Since the transient decay rate after different reclosing actions varies due to the nature of the fault, using a sliding window point-by-point inspection method instead of a fixed delay cutoff can adaptively capture the actual steady-state starting point of each action, avoiding errors caused by improper selection of the cutoff time. The estimate deviates from the true value.

[0033] For continuous The overlapping action will be the timing of the action. Overlapping interval Steady-state recovery time and overlapping result marking The four features are arranged in rows to construct a cross-action time series matrix. , where X represents the real number field, and all elements of the matrix are real numbers. Behavior Vector Cross-action timing matrix The line order strictly corresponds to the chronological order in which the actions occur.

[0034] The value is determined adaptively, by tracing back from the current moment to the past and extracting the number of records that have had overlapping actions within a preset time span. For example, the preset time span can be the most recent 72 hours. If the number of actions within the preset time span is less than a preset lower limit, the process extends forward until the preset lower limit is met; for example, the preset lower limit can be 5 actions. It should be noted that those skilled in the art can appropriately adjust the preset time span and the preset lower limit according to the actual operation and maintenance response cycle to adapt to the operation and management rhythm of different power distribution networks.

[0035] The original overlapping action records are organized into a cross-action time sequence matrix. This provides a structured data carrier for extracting progressive degradation trends across action dimensions.

[0036] S2, calculate the trend slope to obtain the three-dimensional degradation feature vector.

[0037] Based on cross-action time sequence matrix For the overlapping interval sequences respectively Steady-state recovery time sequence and sliding window success rate sequence Extract the slope of the linear trend to obtain the three-dimensional degradation feature vector. ,in, This indicates the slope of the trend in the overlapping interval sequence. This represents the slope of the steady-state recovery time series trend. This represents the slope of the composite power trend during sliding window recombination.

[0038] For overlapping interval sequences By action number For independent variable, Using least squares linear fitting as the dependent variable, the trend slope of the overlapping interval sequence is calculated. A smaller value indicates a systematic contraction in the overlap interval between adjacent overlapping actions, suggesting a continuously accelerating fault triggering frequency; a value close to 0 or positive indicates that the overlap interval remains stable or lengthens, indicating no deterioration signal in the short term. (Regarding steady-state recovery time sequences...) The trend slope of the steady-state recovery time series was calculated in the same way. The larger this value, the longer the time required for the line to return to a steady state after re-collision, reflecting the gradual accumulation of damage to the line.

[0039] For the recombining power, the sliding window size is determined according to a preset ratio based on the number of rows in the cross-action timing matrix. For example, Can be taken Round the value to 1 / 3. In the cross-action timing matrix... The sliding window is slid upwards by a preset step size, and the re-combination power within each sliding window is calculated to obtain a sliding window success rate sequence. In this embodiment, the preset step size for sliding the window is 1, meaning that the window moves downwards by only one re-combination action record each time. Those skilled in the art can adjust the step size (e.g., 2 or 3) according to the total amount of time-series data and the requirements for warning response sensitivity, all of which are equivalent alternative embodiments of the present invention. The formula for the recombinant power within each window is:

[0040] This represents the recombination power of the j-th sliding window, due to the single recombination result. A binary quantity that is either 0 or 1, if directly applied to Linear slope fitting is susceptible to interference from discrete noise. Therefore, a sliding window averaging method is used before slope estimation to smooth out the impact of single random failures on trend judgment. This is applied to the sliding window success rate sequence. Calculate the slope of the linear trend .

[0041] The above three slopes are combined into a three-dimensional degradation feature vector. .in, Characterizing the fault triggering rhythm, Characterizing the accumulation of line damage, The three characteristics of resynthesis, failure, and deterioration together constitute an independent dimension of degradation.

[0042] By extracting the linear trend slopes of the three types of time-series features and combining them into a three-dimensional degradation feature vector, the three degradation paths of overlapping interval shrinkage, steady-state recovery extension, and success rate decline are uniformly incorporated into the same feature space, providing a multi-dimensional perception basis for subsequent baseline deviation calculation.

[0043] S3, calculate the overall deterioration score by combining the deviation amount.

[0044] For the same circuit breaker, extract the normal operating periods without permanent faults from its historical operating data, calculate the historical mean and historical standard deviation for the three slope features respectively, and construct a personalized baseline parameter set. ,in, Indicates the trend slope of the overlapping interval sequence under normal operating conditions. Historical average, Indicates the trend slope of the overlapping interval sequence under normal operating conditions. The historical standard deviation; This represents the trend slope of the steady-state recovery time series under normal operating conditions. Historical average, This represents the trend slope of the steady-state recovery time series under normal operating conditions. The historical standard deviation; This represents the slope of the recombined power trend under normal operating conditions. Historical average, This represents the slope of the recombined power trend under normal operating conditions. The historical standard deviation; this personalized baseline parameter set is updated after the completion of on-site operation and maintenance inspection.

[0045] Overall degradation score An orientation constraint is introduced based on the standard Z-score 3D stacking. The original Z-score stacking form is:

[0046] Compared to the original Z-score, which includes potential false alarm noise for both positive and negative deviations, this invention determines whether the trend slopes of the overlapping interval sequence, the steady-state recovery time sequence, and the sliding window success rate sequence conform to their respective degradation physical directions. It only accumulates deviations that conform to these degradation physical directions, achieving directional filtering and avoiding scoring interference caused by occasional slope shifts towards non-degradation directions. The improved scoring formula is:

[0047] in, This is an indicator function; it takes the value 1 if the condition within the parentheses is true, and 0 otherwise. In this relation, This represents the overall degree to which the degradation trend feature vector deviates from the normal baseline. The larger the value, the greater the deviation of the three components from the normal baseline in the direction of degradation, and the deeper the line is approaching a permanent fault; the smaller the value, the less the slope of the three dimensions deviates significantly from the normal baseline range, and the better the current operating condition of the equipment.

[0048] Thus, the improved normalized deviation superposition score, which introduces directional indication constraints, eliminates the scoring interference caused by occasional data fluctuations shifting towards non-deterioration directions, and improves the ability of the comprehensive score to reflect the true deterioration trend.

[0049] S4, calculates the rate and acceleration to trigger graded early warning output.

[0050] In response to triggering new overlapping actions and completing the cross-action timing matrix Update, recalculate overall degradation score To form a scoring time series ,in Calculate batch numbers for scoring. In the most recent consecutive... Calculate the first-order difference sequence of scores between adjacent batches over the time series of scores for each batch. For first-order difference sequences The average value is used to obtain the rate of increase in scores. ; and for first-order difference sequences Calculate the slope of the linear trend The acceleration of the rating time series was obtained. The value is determined based on the actual operation and maintenance response cycle. For example, You can retrieve the number of score update batches within the last 24 hours.

[0051] From historical events where permanent faults have been confirmed in the circuit breaker or a group of circuit breakers of the same model, a sample set is constructed by extracting the comprehensive degradation score time series during the pre-fault period of each fault. The pre-fault period is divided into three stages based on the time elapsed since the fault occurred. The specific time ranges for each stage are: early pre-fault stage (72h–24h before fault occurrence); mid-pre-fault stage (24h–6h before fault occurrence); and critical pre-fault stage (6h before fault occurrence to the time of fault occurrence). The duration of each stage can be adaptively scaled according to the distribution network maintenance cycle and fault evolution rate. This time division is only an exemplary implementation and does not limit the scope of protection of this invention. The first preset quantile of the comprehensive degradation score value in the sample set during the earlier pre-fault stage is taken as the lower boundary of the yellow warning. For example, this first preset quantile can be the 25th percentile. The median of the comprehensive degradation score value in the sample set during the mid-pre-fault stage is taken as the lower boundary of the orange warning. The second preset quantile of the comprehensive degradation score value in the sample set during the critical pre-fault stage is taken as the lower boundary of the red warning. For example, this second preset quantile can be the 75th percentile. For first-order difference The single-step jump determination threshold is statistically analyzed during historical normal operation periods. mean of the sequence and standard deviation ,by As a criterion for judgment ,in The preset multiplier is determined based on the required false alarm rate tolerance. For example, You can choose 2.

[0052] Response to overall degradation score Exceeding the lower boundary of the yellow alert and the rate of increase in the score This triggers a yellow alert; in response to the overall degradation score. Exceeding the lower boundary of the orange alert and accelerating This triggers an orange alert; in response to the overall degradation score. Exceeding the lower boundary of the red alert and single batch increment This triggered a red alert.

[0053] The warning results and the current overall degradation score will be used to assess the situation. The contribution ratio of each trend slope component in the three-dimensional degradation feature vector and The alert is packaged into an early warning message and sent to the main system. It should be noted that those skilled in the art can flexibly adjust the specific field format and transmission protocol of the early warning message according to the interface specifications of the main system.

[0054] Thus, by using the growth rate and acceleration of the score as additional criteria for triggering graded early warnings, the system exhibits higher response sensitivity to the accelerated phase of the degradation process than simply relying on the absolute value of the score, thereby achieving accurate classification of early warning levels.

[0055] like Figure 2 As shown, it can be observed that the scoring curve of the method of the present invention has crossed the warning boundaries of each level before the fault occurs, while the scoring curve of the existing method crosses the boundary at a significantly delayed time, indicating that the method of the present invention has a stronger ability to perceive in advance in terms of cross-action time-series trend analysis.

[0056] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for early warning processing of reclosing control data of pole-mounted circuit breakers, characterized in that, Includes the following steps: S1. Obtain the reclosing control data of the target pole-mounted circuit breaker, extract the action time, reclosing interval, steady-state recovery time and reclosing result mark, and construct a cross-action timing matrix in chronological order. S2, based on the cross-action time series matrix, calculate the trend slope of the overlapping interval sequence, the trend slope of the steady-state recovery time sequence, and the trend slope of the sliding window success rate sequence, and combine them to obtain a three-dimensional degradation feature vector; S3. Extract historical feature data of the target pole-mounted circuit breaker during normal operation to construct a personalized baseline parameter set. Combine the deviation of each trend slope in the three-dimensional degradation feature vector from the physical direction of degradation to calculate the comprehensive degradation score. S4, in response to the update of the cross-action time sequence matrix, recalculate the comprehensive degradation score to form a score time series, calculate the score growth rate and acceleration of the score time series, and trigger the corresponding graded warning output in combination with the preset graded warning boundary; By combining the deviation of the slopes of each trend in the three-dimensional degradation feature vector from the physical direction of degradation, a comprehensive degradation score is calculated, including: Determine whether the trend slope of the overlapping interval sequence, the trend slope of the steady-state recovery time sequence, and the trend slope of the sliding window success rate sequence conform to their respective physical deterioration directions. For the trend slope that conforms to the direction of physical degradation, the deviation is calculated using the historical mean and historical standard deviation corresponding to the personalized baseline parameter set; The deviations are summed up to obtain the final comprehensive degradation score.

2. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Extract the steady-state recovery time, including: Multiple consecutive sampling points are selected as the steady-state confirmation window; If the maximum deviation of the current value within the steady-state confirmation window does not exceed a preset proportion of the rated current, the line is determined to have entered a steady state and the timing is terminated. The steady-state recovery time is defined as the time it takes for the line current to recover to a steady state from the moment of action after reclosing.

3. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Construct a cross-action time series matrix in chronological order, including: Going back from the current moment to the past, extract the number of records of all overlapping actions that have occurred within a preset time span; If the number of actions within the preset time span is less than the preset lower limit, then extend forward until the preset lower limit is met; The action time, overlap interval, steady-state recovery time, and overlap result markers corresponding to the number of records are arranged in rows to obtain the cross-action time sequence matrix.

4. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Calculate the trend slope of the sliding window success rate sequence, including: The sliding window size is determined based on the number of rows in the cross-action timing matrix; Slide the cross-action time sequence matrix at a preset step size, calculate the recombining power within each sliding window, and obtain the sliding window success rate sequence; By fitting a linear slope to the success rate sequence of the sliding window, the trend slope of the success rate sequence of the sliding window is finally obtained.

5. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Historical characteristic data of the target pole-mounted circuit breaker during normal operation are extracted to construct a personalized baseline parameter set, including: Extract the normal operating periods during which no permanent faults occurred from the historical operating data of the target pole-mounted circuit breaker; Calculate the historical mean and historical standard deviation of the overlap interval sequence, steady-state recovery time sequence, and sliding window success rate sequence during normal operation periods, respectively. By combining historical mean and historical standard deviation, a personalized set of baseline parameters is obtained.

6. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Calculate the rate of increase and acceleration of ratings over a time series, including: Calculate the first-order difference sequence of scores between adjacent batches over the recent consecutive time series of scores; The rate of increase in scores is obtained by averaging the first-order difference sequence. The linear trend slope is calculated for the first-order difference sequence, and the acceleration of the scoring time series is finally obtained.

7. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, The steps for obtaining the preset graded early warning boundaries include: A sample set is constructed by extracting the comprehensive degradation score time series of events in history that have been confirmed to have permanent failures during the period preceding the failure. The first preset quantile of the comprehensive degradation score of the sample set in the early stage before the failure is taken as the lower boundary of the yellow warning. The median of the comprehensive degradation score of the sample set in the early and middle stages of the fault is taken as the lower boundary of the orange warning. The second preset quantile of the comprehensive degradation score of the sample set in the pre-fault critical stage is taken as the lower boundary of the red warning.

8. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 1, characterized in that, Trigger corresponding tiered warning outputs based on preset tiered warning boundaries, including: A yellow alert is triggered when the overall degradation score exceeds the lower boundary of a yellow alert and the score growth rate is greater than 0. An orange alert is triggered when the overall degradation score exceeds the lower boundary of the orange alert and the acceleration is greater than 0. A red alert is triggered when the overall degradation score exceeds the lower boundary of the red alert and the increment of a single batch is greater than the judgment benchmark.

9. The method for early warning processing of reclosing control data of pole-mounted circuit breakers according to claim 8, characterized in that, After triggering the corresponding tiered early warning output, it also includes: The warning results, the current comprehensive degradation score, the contribution ratio of each trend slope in the three-dimensional degradation feature vector, and the acceleration are packaged into the warning message. The early warning message is sent to the main station system.

Images