An overhead line health monitoring method based on multi-dimensional monitoring sensors

Abstract

The present application relates to the technical field of line fault monitoring, in particular to an overhead line health monitoring method based on a multi-dimensional monitoring sensor. The method analyzes the abnormal degree of the change amplitude of the operation data of different dimensions in time sequence on each monitoring unit of the overhead line, screens out abnormal time periods, determines the fault attention factor through the distribution and duration of the abnormalities of multiple abnormal dimensions in the abnormal time periods, and obtains more accurate fault evaluation results by combining the correlation between different monitoring units. The present application analyzes the abnormal fluctuation and stability of multi-dimensional data, combines the correlation in the space of the overhead line, reduces the interference of fault evaluation, and improves the accuracy and reliability of health monitoring.

Description

Technical Field

[0001] This invention relates to the field of line fault monitoring technology, and specifically to a method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors. Background Technology

[0002] Overhead transmission lines typically cover a wide geographical area and are subject to various geographical and meteorological environmental influences. When encountering natural and external environmental factors such as lightning, heavy rain, or hurricanes, short circuits or line breaks can occur, endangering the safe operation of the power system. When an overhead transmission line fault occurs, the location and cause of the fault should be quickly identified and eliminated to rapidly restore power supply to users.

[0003] The typical monitoring process involves setting up several monitoring points and using fixed thresholds to monitor the health of the lines. However, overhead lines are highly susceptible to the influence of multimodal environmental data. Simply capturing abnormal fluctuations cannot accurately measure the root cause of the anomaly. Furthermore, since overhead lines are usually installed outdoors, they are easily affected by the natural environment and various power equipment, such as substations or transformers. This can lead to confusion between the interference and the actual faults in the lines during health monitoring, resulting in potential errors and reduced accuracy in health monitoring. Summary of the Invention

[0004] To address the technical problem in existing technologies where overhead lines are easily affected by interference from the natural environment and various power equipment, leading to confusion between interference and actual line faults in monitoring and reducing the accuracy of health monitoring, the present invention aims to provide an overhead line health monitoring method based on multi-dimensional monitoring sensors. The specific technical solution adopted is as follows: This invention provides a method for health monitoring of overhead power lines based on multi-dimensional monitoring sensors, the method comprising: Acquire operational data for each dimension in the historical time series within each monitoring unit on the overhead line; The analysis examines the degree of fluctuation anomaly in the operational data of each dimension at each time point to determine the corresponding abnormal dimension at that time point; based on the proportion of abnormal dimensions, it determines the abnormal time points that constitute abnormal time periods; based on the continuous distribution characteristics of abnormal dimensions within the abnormal time period, it determines the sub-time periods within the abnormal time period; and through the degree of fluctuation anomaly and the continuous stability of the operational data in all sub-time periods within the abnormal time period, it determines the fault concern factors for each abnormal time period. Between two monitoring units, based on the degree of overlap of abnormal dimensions and the similarity of operational data in each dimension during the intersecting abnormal periods, and combined with the degree of synchronization of the distribution of the intersecting abnormal periods, correlation evaluation indicators are obtained for the two monitoring units; and the related units of each monitoring unit are selected based on the correlation evaluation indicators. Based on the degree of correlation between each monitoring unit and related units, as well as the fault concern factors, the current fault assessment indicators for each monitoring unit are determined; and hierarchical monitoring and early warning are carried out based on the fault assessment indicators.

[0005] Furthermore, the method for obtaining the anomaly dimension includes: For any dimension at any given time, calculate the difference between the data value of that dimension at that time and the data value at adjacent times, and then take the mean of the differences between all data values ​​as the neighbor fluctuation value of that dimension at that time. The mode of the neighbor fluctuations of this dimension at all times is taken as the benchmark fluctuation value of this dimension; the difference between the neighbor fluctuation value of this dimension at that time and the benchmark fluctuation value is calculated to obtain the benchmark deviation. By combining the baseline deviation and the neighboring volatility of this dimension, the abnormal volatility of this dimension at that moment can be obtained; The dimension whose abnormal volatility exceeds the preset volatility threshold at that moment is taken as the abnormal dimension at that moment.

[0006] Furthermore, determining the abnormal time periods based on the proportion of abnormal dimensions includes: At any given time, if the proportion of abnormal dimensions among all dimensions is greater than the preset abnormal proportion, the corresponding time is designated as an abnormal time; and the time period consisting of consecutively adjacent abnormal times is designated as an abnormal time period.

[0007] Furthermore, the method for obtaining the sub-time period includes: For any abnormal time period, the set of all abnormal dimensions that exist at the abnormal time in that abnormal time period is taken as the abnormal set; the abnormal dimensions in the abnormal set are arranged and combined to obtain the dimension combination of that abnormal time period. For any combination of dimensions in the abnormal period, when there is a continuous distribution of abnormal moments corresponding to that dimension combination, the period consisting of the continuous distribution of abnormal moments corresponding to that dimension combination is regarded as a sub-period within the abnormal period.

[0008] Furthermore, the method for obtaining the fault concern factor includes: For any sub-period within any abnormal period, the ratio of the total number of abnormal dimensions corresponding to all abnormal moments in that sub-period to the total number of all dimensions is taken as the abnormal coverage percentage of that sub-period. The average of the abnormal volatility of each abnormal dimension in the sub-period at the abnormal time in the sub-period is used as the volatility factor of each abnormal dimension; the average of the volatility factors of all abnormal dimensions in the sub-period is used as the abnormal volatility index of the sub-period. By combining the duration, abnormal coverage ratio, and abnormal fluctuation indicators of the sub-period, the credibility of the local fault in the sub-period is obtained. The sum of the local fault confidence values ​​of all sub-time periods is normalized to obtain the fault concern factor for the abnormal time period.

[0009] Furthermore, the method for obtaining the correlation evaluation indicators includes: For any two monitoring units, the time period in which the abnormal time periods of the two monitoring units intersect is obtained as the intersecting time period; for any intersecting time period, the number of the same abnormal dimensions in the two monitoring units at each time point of the intersecting time period is obtained as the overlap number; the average of the overlap numbers at all times of the intersecting time period is used as the overlap index. After calculating the correlation of the data of the two monitoring units in each dimension, the average correlation of all dimensions is used as the synchronization index of the two monitoring units. The product of the overlap index and the operation synchronization index of the intersecting period is taken as the local overlap degree of the intersecting period; the mean of the local overlap degrees of all intersecting periods in the two monitoring units is taken as the abnormal overlap degree of the two monitoring units. Based on the proportion of the abnormal time periods that intersect between two monitoring units in the total abnormal time period, and combined with the location distance distribution of the two monitoring units, the temporal synchronization of the two monitoring units is obtained. By combining the degree of abnormal overlap and temporal synchronization of each pair of monitoring units, a correlation evaluation index for each pair of monitoring units is obtained.

[0010] Furthermore, the method for obtaining the timing synchronization includes: For any intersecting time period, the ratio between the duration of the intersecting time period and the maximum duration of the abnormal time period corresponding to the two monitoring units is used as the synchronization coefficient of the intersecting time period; the average of the synchronization coefficients of all intersecting time periods is used as the abnormal synchronization index of the two monitoring units. The location distance between the two monitoring units is negatively correlated and normalized to serve as the distribution coefficient of the two monitoring units. The product of the distribution coefficients of the two monitoring units and the abnormal synchronization index is taken as the temporal synchronization of the two monitoring units.

[0011] Furthermore, the method for obtaining the associated unit includes: For any given monitoring unit, monitoring units whose correlation evaluation index is greater than a preset correlation threshold are designated as associated units of that monitoring unit.

[0012] Furthermore, the method for obtaining the fault assessment indicators includes: For any monitoring unit, the product of the minimum correlation evaluation index among the associated units corresponding to that monitoring unit and the total number of associated units is used as the correlation influence coefficient of that monitoring unit. The average value of the fault concern factors for all abnormal time periods in the time series of the monitoring unit is used as the fault probability. The product of the negative correlation mapping value of the correlation influence coefficient of the monitoring unit and the fault probability is used as the fault assessment index of the monitoring unit.

[0013] Furthermore, the hierarchical monitoring and early warning system based on fault assessment indicators includes: When the fault assessment index of the monitoring unit is greater than or equal to the preset high assessment threshold, the monitoring result is recorded as high risk; When the fault assessment index of the monitoring unit is greater than the preset low assessment threshold but less than the preset high assessment threshold, the monitoring result is recorded as a warning. When the fault assessment index of the monitoring unit is less than or equal to the preset low assessment threshold, the monitoring result is recorded as normal; the preset low assessment threshold is less than the preset high assessment threshold.

[0014] The present invention has the following beneficial effects: This invention analyzes the temporal variation of operational data across different dimensions at each monitoring unit of an overhead line, identifying anomalous time periods. By combining multi-dimensional fluctuation data, it preliminarily determines these periods, screening for potential anomalies and further considering the unstable and random nature of interference. By analyzing the distribution and persistence of anomalies across multiple dimensions within the anomalous time period, it identifies fault-related factors and assesses the likelihood that the anomalies correspond to actual faults. Considering the overall interference and confusion caused by local equipment codes, it further reduces noise interference by combining the correlation between different monitoring units, resulting in more accurate fault assessment results. This invention, through multi-dimensional data anomaly fluctuation stability analysis combined with the spatial correlation of overhead lines, reduces the interference impact on fault assessment and improves the accuracy and reliability of health monitoring. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a method for monitoring the health of overhead power lines based on multi-dimensional monitoring sensors, provided in one embodiment of the present invention; Figure 2This is a flowchart illustrating a method for obtaining correlation evaluation indicators according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for an overhead line health monitoring method based on multi-dimensional monitoring sensors provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a flowchart of an overhead line health monitoring method based on multi-dimensional monitoring sensors, according to an embodiment of the present invention. The method includes the following steps: S1: Acquire operational data for each dimension in the historical time series of each monitoring unit on the overhead line.

[0021] The health status of overhead lines can be affected by multiple factors, and the impact of the external environment may vary in different distribution areas. Therefore, defining the scope of monitoring units facilitates the subsequent location of localized hidden dangers. In this embodiment of the invention, adjacent towers and the connecting conductors between them constitute a monitoring unit, and fault assessment and location are performed based on monitoring of each monitoring unit.

[0022] In this embodiment of the invention, sensors such as current sensors, voltage sensors, temperature sensors, and vibration sensors are installed on the towers of each monitoring unit to collect operational data for each dimension at each sampling time in the time series. The dimensions include current, voltage, temperature, and vibration signals, etc. The data of the previous day's time series is obtained at the current time. The sampling frequency is once every 30 seconds. It should be noted that the specific sampling settings can be adjusted by the implementer according to the specific implementation scenario, and there are no restrictions here.

[0023] It is understandable that all monitored operational data undergoes preprocessing. Preprocessing may include data standardization and time-scale normalization to facilitate unified data analysis and remove the influence of units. It should be noted that data preprocessing is a technique well-known to those skilled in the art, and will not be elaborated upon or limited here.

[0024] S2: Analyze the degree of fluctuation anomaly in the operational data of the analysis dimension at each time point to determine the corresponding abnormal dimension at that time point; determine the abnormal time points as abnormal time periods based on the proportion of abnormal dimensions; determine the sub-time periods in the abnormal time period based on the continuous distribution characteristics of abnormal dimensions within the abnormal time period; determine the fault concern factors for each abnormal time period by analyzing the degree of fluctuation anomaly and the continuous stability of the operational data in all sub-time periods within the abnormal time period.

[0025] Overhead transmission lines typically consist of multiple adjacent tower sections, each with variations in structural stress, environmental conditions, and electrical load. When the line's operating status fluctuates, local faults and system fluctuations can be quite similar from a global data perspective, leading to misjudgments. Therefore, we first analyze individual monitoring units to assess the likelihood of a true fault.

[0026] Under normal circumstances, the fluctuations in operational data across various dimensions are relatively smooth, remaining within certain limits. For example, the waveform changes of current under normal load are slow, continuous, and stable, with a relatively smooth waveform and no abnormal vibrations.

[0027] However, abnormal fluctuations are usually related to multiple factors, such as the start-up and shutdown of electrical equipment, seasonal load changes, environmental factors, and faults. For example, current and voltage data can be affected by the instantaneous impact of large equipment starting up. Such fluctuations under normal circumstances have a short duration and will return to stable values ​​within a short time. However, data fluctuations caused by faults, such as overload conditions where the current exceeds the rated capacity of the line or equipment, result in larger current changes and are caused by faults. If not handled promptly, this may lead to equipment damage.

[0028] Therefore, considering the different dimensions of fluctuation anomalies, the method for obtaining the anomaly dimension in this embodiment of the invention includes: First, for any dimension at any given time, calculate the difference between the data value of that dimension at that time and the data values ​​at adjacent times. Then, take the average of these differences as the neighbor-to-neighbor fluctuation value for that dimension at that time. The degree of difference between the data values ​​of the dimension at each time point and its adjacent times reflects the magnitude of change in the dimension's data values. A larger neighbor-to-neighbor fluctuation value indicates a greater magnitude of change and a more significant anomaly. As an example, the expression for the neighbor-to-neighbor fluctuation value is: In the formula, Represented as the first At the time of the first Neighbor-to-neighbor fluctuation values ​​in each dimension Represented as the first At the time of the first Data values ​​in each dimension Represented as with the first The total number of adjacent time intervals between each time interval. Represented as with the first The first time interval adjacent to the first At the nth adjacent time point Data values ​​in each dimension Represented as an absolute value extraction function, since each sampling time must have adjacent times, It cannot be zero. Represented as the first The dimension in the first The data value at time and the data value at time . The difference between data values ​​at adjacent time points.

[0029] Furthermore, the mode of the neighboring fluctuation values ​​at all times for this dimension is used as the baseline fluctuation value for this dimension. Since the operating state is normally normal under normal circumstances, the mode can reflect the magnitude of the change when relatively stable, that is, the magnitude of the fluctuation value under normal conditions. In other embodiments of the present invention, the mode of the neighboring fluctuation values ​​under standard normal operation experiments can also be directly used as the baseline fluctuation value, which will not be elaborated or limited here. It should be noted that if there are multiple modes among the neighboring fluctuation values, the average of the modes is used as the baseline fluctuation value; if there is no mode among the multiple neighboring fluctuation values, the average of the neighboring fluctuation values ​​is used as the baseline fluctuation value. In the embodiments of the present invention, the normalized values ​​only retain two significant figures after the decimal point.

[0030] The difference between the neighboring fluctuation value and the benchmark fluctuation value of this dimension at that time is further calculated to obtain the benchmark deviation, which reflects the degree of deviation from the normal fluctuation. Combining the benchmark deviation and the neighboring fluctuation value of this dimension, the abnormal fluctuation of this dimension at that time is obtained. In this embodiment of the invention, the normalized value of the product of the benchmark deviation and the neighboring fluctuation value of this dimension is taken as the abnormal fluctuation of this dimension at that time. The larger the benchmark deviation, the greater the difference between the dimension and the normal change range at that time. The larger the neighboring fluctuation value, the greater the change range of this dimension itself at that time. Therefore, the lower the probability that this dimension belongs to normal fluctuation at that time, the higher the probability of abnormality, and thus the greater the abnormal fluctuation.

[0031] Finally, the dimension with abnormal volatility exceeding the preset volatility threshold at that moment is taken as the abnormal dimension at that moment. The greater the abnormal volatility, the higher the degree of abnormal fluctuation in that dimension. In this embodiment of the invention, the preset volatility threshold is set to 0.8, but the implementer can adjust this value according to the actual situation. The specific value is not limited here.

[0032] At any given moment, when a significant number of dimensions exhibit abnormal behavior, it strongly suggests the possibility of a fault or anomaly. Therefore, moments are marked based on the proportion of abnormal dimensions to identify abnormal moments. In this embodiment of the invention, at each moment, when the proportion of abnormal dimensions among all dimensions exceeds a preset abnormal proportion, the corresponding moment is designated as an abnormal moment. The preset abnormal proportion is 50%. When the number of abnormal dimensions exceeds half the total number of dimensions, this moment is also designated as an abnormal moment, and the time period consisting of consecutive adjacent abnormal moments is designated as an abnormal time period. The preset abnormal proportion can be adjusted by the implementer and is not limited here.

[0033] When overhead power lines are installed outdoors, they can experience non-fault fluctuations in monitoring data due to electromagnetic signals or temporary malfunctions from surrounding power equipment. For example, a monitoring unit near a substation might normally collect a stable current of 200-300A. However, if the substation performs a switching operation, the current sensor may experience electromagnetic interference, resulting in 10 transient jumps in the current data, exhibiting pseudo-anomaly characteristics. Because these pseudo-anomalies are similar to the fluctuations of a real fault during abnormal periods, they can cause errors in fault monitoring and early warning, leading to wasted resources.

[0034] Since faults are inherent hidden dangers in lines and their impact range is relatively fixed, while interference is accidental and irregular, it can cause the affected dimensions to change randomly at different times. Therefore, we can analyze the continuous distribution of abnormal dimensions at each abnormal time in the abnormal period, and measure the degree of interference in the abnormal period to determine the degree of credible concern for the real fault.

[0035] In this embodiment of the invention, the method for obtaining sub-time periods includes: for any abnormal time period, taking the set of all abnormal dimensions present at the abnormal moments in the abnormal time period as the abnormal set, and arranging and combining the abnormal dimensions in the abnormal set to obtain the dimension combination of the abnormal time period. As an example, if the abnormal dimensions corresponding to four abnormal moments are {A, B, C, F}, {B, C}, {A, B, C}, and {B, F}, then the abnormal set is {A, B, C, F}. By arranging and combining without considering the order, the dimension combinations that can be obtained include {A}, {B}, {C}, {F}, {A, B}, {A, C}, {A, F}, {B, C}, {B, F}, {C, F}, {A, B, C}, {A, B, F}, {A, C, F}, {B, C, F}, {A, B, C, F}.

[0036] For any combination of dimensions within the abnormal time period, when there is a continuous distribution of abnormal times for that combination of dimensions, the time period consisting of the continuously distributed abnormal times corresponding to that combination of dimensions is taken as a sub-time period within the abnormal time period. As an example, in the aforementioned example, if the combination of dimensions {B, C} is continuously distributed in the first three abnormal times, then the first three times correspond to a sub-time period; if the combination of dimensions {B} is continuously distributed in four times, then the four times correspond to a sub-time period; and if the combination of dimensions {B, F} is continuously distributed in the last two times, then the last two times correspond to a sub-time period.

[0037] By observing the distribution and persistence of abnormal dimensions, the stable impact of the abnormal dimension coverage can be reflected, thereby assessing the possible scenarios of actual faults. In this embodiment of the invention, the method for obtaining fault concern factors includes: First, for any sub-period within any abnormal period, the ratio of the total number of abnormal dimensions corresponding to all abnormal moments in that sub-period to the total number of all dimensions is taken as the abnormal coverage ratio of that sub-period. The more abnormal dimensions there are in the sub-period, the greater the abnormal coverage and the wider the impact.

[0038] Furthermore, the mean of the abnormal volatility of each abnormal dimension in the sub-period is taken as the volatility factor of each abnormal dimension. The mean of the volatility factors of all abnormal dimensions in the sub-period is taken as the abnormal volatility index of the sub-period. By combining the abnormal volatility of all abnormal dimensions in the sub-period at any time, the intensity of the abnormal change amplitude in the sub-period is reflected. The larger the abnormal volatility index, the more significant the amplitude change, the more serious the abnormal state, and the higher the attention.

[0039] By combining the duration, abnormal coverage ratio, and abnormal fluctuation index of the sub-period, the credibility of the local fault in the sub-period is obtained. In this embodiment of the invention, the cumulative product of the duration, abnormal coverage ratio, and abnormal fluctuation index of the sub-period is used as the credibility of the local fault in the sub-period. The longer the duration of the sub-period and the larger the abnormal coverage ratio and abnormal fluctuation index, the more likely the anomaly is to be in a state of high sustained intensity, the more serious the abnormal change in multiple dimensions, the higher the possibility of a real fault, and the higher the attention required.

[0040] Finally, the sum of the confidence values ​​of local faults in all sub-periods is normalized to obtain the fault concern factor for the abnormal period. Combined with the distribution analysis of all sub-periods, the fault concern factor for the abnormal period is obtained in a comprehensive manner, which reflects the degree of abnormality that needs attention in the period. The larger the fault concern factor, the higher the degree of attention.

[0041] It should be noted that normalization is a technique well known to those skilled in the art. The choice of normalization can be linear normalization or standard normalization, etc., and the specific normalization method is not limited here.

[0042] S3: Between two monitoring units, based on the degree of overlap of abnormal dimensions and the similarity of operational data in each dimension during the intersecting abnormal period, and combined with the degree of synchronization of the distribution of the intersecting abnormal period, obtain the correlation evaluation index of the two monitoring units; and screen the associated units of each monitoring unit based on the correlation evaluation index.

[0043] Changes in the operation of line connection equipment can affect multiple adjacent lines. For example, peak regional electricity consumption or concentrated factory operations can lead to synchronous anomalies in multiple basic monitoring units, resulting in similar trends and timing of operational data changes. Line faults, on the other hand, often affect only specific sections of the line, such as icing on a conductor or overheating at a joint, leading to isolated anomalies in a single or a few adjacent units. When synchronous fluctuations exist, they can mask the true nature of the fault. For instance, during peak residential electricity consumption periods covering multiple lines across an area, the increased equipment load will synchronously increase the current and temperature of each unit, thus reducing the ability to locate the faulty monitoring unit.

[0044] Therefore, when abnormal fluctuations exist between monitoring units, analyzing the synchronization correlation is crucial to determine the potential correlation, thereby reducing the impact of normal fluctuations and improving the accuracy of subsequent fault assessment. Preferably, in this embodiment of the invention, the method for obtaining the correlation assessment indicators is described in [reference needed]. Figure 2 The diagram illustrates a flowchart of a method for obtaining correlation evaluation indicators according to an embodiment of the present invention. The method includes the following steps: S301: Obtain the degree of abnormal overlap between two monitoring units based on the degree of dimensional overlap between them during the intersecting abnormal period and the degree of approximation of the running data of all dimensions.

[0045] Normally, changes in the operation of line connection equipment will cause the abnormal dimensions of the two monitoring units to be highly synchronized, while faults will cause the abnormal dimensions of the two units to be random and unrelated. For example, if ice accumulates on the line of a certain monitoring unit, it will affect the vibration data and temperature data it collects, but the data of other monitoring units will not be affected.

[0046] In this embodiment of the invention, for any two monitoring units, the time period in which the abnormal periods of the two monitoring units intersect is obtained as the intersection period, and the possibility of common abnormal fluctuations is determined. It can be understood that if there is no intersection period, it reflects that there is no common abnormal fluctuation between the two monitoring units, and the correlation evaluation index can be directly set to zero, which will not be elaborated here.

[0047] Furthermore, for any intersecting time period, the number of identical abnormal dimensions in the two monitoring units at each moment of the intersecting time period is obtained as the overlap number. The average of the overlap numbers at all moments of the intersecting time period is used as the overlap index. The more abnormal dimensions that exist synchronously, the more consistent the synchronization of the abnormal state, and the lower the probability of the fault corresponding to the abnormal time period.

[0048] After further calculating the correlation of the operational data of the two monitoring units in each dimension, the average correlation of all dimensions is used as the operational synchronization index of the two monitoring units. In this embodiment of the invention, the Pearson correlation coefficient can be used to obtain the correlation between time series data. The greater the correlation, the more similar the operational states are between dimensions, and the higher the credibility of abnormal synchronization overlap. It should be noted that obtaining the correlation of time series is a technical means well known to those skilled in the art, and dynamic time rule algorithms, etc., can also be used, which are not limited or elaborated here.

[0049] Finally, the product of the overlap index and the operational synchronization index for the intersecting time period is taken as the local overlap degree for that intersecting time period, reflecting the reliability of the synchronization of abnormal states during the intersecting time period. The larger the overall local overlap degree, the higher the correlation. Therefore, the mean of the local overlap degrees for all intersecting time periods in the two monitoring units is taken as the abnormal overlap degree of the two monitoring units.

[0050] S302: Based on the proportion of the abnormal time period when two monitoring units intersect within the abnormal time period, and combined with the location distance distribution of the two monitoring units, the temporal synchronization of the two monitoring units is obtained.

[0051] Because changes in the operation of multiple devices can lead to a high degree of overlap in abnormal periods of different monitoring units, such as peak electricity consumption in the region, voltage regulation in substations, etc., while the overlap of abnormal periods of local faults is often random, such as icing in a monitoring unit or overheating of a monitoring unit's joints, we can further analyze the time synchronization of abnormalities in two monitoring units to analyze the correlation and provide data preparation for subsequent assessment of the true fault of each monitoring unit.

[0052] In this embodiment of the invention, for any intersecting time period, the ratio between the duration of the intersecting time period and the longest duration of the abnormal time period corresponding to the two monitoring units is used as the synchronization coefficient of the intersecting time period. The average of the synchronization coefficients of all intersecting time periods is used as the abnormal synchronization index of the two monitoring units. The larger the proportion of the intersecting time period in the longest period of the two abnormal time periods, the higher the complete overlap of the abnormal time periods and the more significant the temporal synchronization state.

[0053] Simultaneously considering the actual location distribution of the monitoring units, the location distance between the two monitoring units is negatively correlated and normalized, serving as the distribution coefficient for the two monitoring units. The synchronization analysis is more reliable only when the location distances are closer. It should be noted that negative correlation mapping is a technique well-known to those skilled in the art; other methods, such as using inverse proportional values ​​or negative exponential forms with a natural constant as the base, will not be elaborated upon here.

[0054] Finally, the product of the distribution coefficients of the two monitoring units and the abnormal synchronization index is taken as the temporal synchronization of the two monitoring units. The higher the temporal synchronization, the higher the correlation between the monitoring units.

[0055] S303: By combining the degree of abnormal overlap and the temporal synchronization of each pair of monitoring units, the correlation evaluation index of each pair of monitoring units is obtained.

[0056] Therefore, the correlation between monitoring units is ultimately measured by combining the degree of anomaly overlap and the temporal synchronization. In this embodiment of the invention, the product of the degree of anomaly overlap and the temporal synchronization of each pair of monitoring units is normalized to obtain a correlation evaluation index for each pair of monitoring units. The larger the correlation evaluation index, the higher the correlation between the monitoring units and the greater the impact on subsequent fault assessment.

[0057] In this embodiment of the invention, for any given monitoring unit, monitoring units whose correlation evaluation index is greater than a preset correlation threshold are considered as associated units of that monitoring unit. The preset correlation threshold can be set to 0.8, and the implementer can adjust the specific value to filter for highly correlated units. The more associated units a monitoring unit has, the more easily its abnormal fluctuations are affected, and the lower the reliability of the actual fault.

[0058] S4: Based on the degree of correlation between each monitoring unit and related units, as well as the fault concern factors, determine the current fault assessment indicators for each monitoring unit; and conduct hierarchical monitoring and early warning based on the fault assessment indicators.

[0059] By adjusting the fault assessment results based on factors with high correlation, the influence of pseudo-anomaly units is reduced, thereby improving the accuracy of real fault assessment. In this embodiment of the invention, the method for obtaining fault assessment indicators includes: For any monitoring unit, the product of the minimum correlation evaluation index among the associated units corresponding to the monitoring unit and the total number of associated units is used as the correlation influence coefficient of the monitoring unit. The more associated units a monitoring unit corresponds to, the greater the possibility of being affected. The larger the minimum correlation evaluation index, the stronger the correlation, the higher the correlation influence, and the greater the possibility of pseudo-abnormal state.

[0060] Furthermore, the average fault concern factor for all abnormal periods in the time series of the monitoring unit is used as the fault probability, reflecting the probability of faults due to abnormal fluctuations in the historical time series. By adjusting the correlation coefficient, the product of the negative correlation mapping value of the correlation influence coefficient of the monitoring unit and the fault probability is used as the fault assessment index of the monitoring unit. The lower the correlation in the historical time series and the greater the abnormal fluctuations, the higher the fault probability.

[0061] The failure probability can be assessed to provide graded early warning. In this embodiment of the invention, when the failure assessment index of the monitoring unit is greater than or equal to the preset high assessment threshold, the monitoring unit is considered to be a faulty unit, and the monitoring result is recorded as high risk. Furthermore, the alarm information, such as the span number and time, can be pushed to the dispatch center for emergency repair.

[0062] When the fault assessment index of the monitoring unit is greater than the preset low assessment threshold but less than the preset high assessment threshold, the monitoring result is recorded as a warning, which can be further pushed to the dispatch center and the mobile terminal of the operation and maintenance personnel. At the same time, the sampling frequency of this span is increased to capture more granular abnormal changes.

[0063] When the fault assessment index of the monitoring unit is less than or equal to the preset low assessment threshold, the monitoring result is recorded as normal, and subsequent multidimensional data collection and analysis can continue at the predetermined sampling frequency.

[0064] In this embodiment of the invention, the preset low evaluation threshold must be less than the preset high evaluation threshold. The preset low evaluation threshold can be set to 0.5, and the preset high evaluation threshold can be set to 0.8. The specific values ​​can be adjusted by the implementer and are not limited here.

[0065] In summary, this invention analyzes the degree of anomalous changes in the temporal variation of operational data across different dimensions at each monitoring unit of an overhead line, identifies anomalous time periods, and preliminarily determines these periods based on multi-dimensional fluctuations. It also preliminarily filters out potentially anomalous components and further considers the unstable and random nature of interference. By analyzing the distribution and persistence of anomalies across multiple dimensions within the anomalous time period, it determines fault concern factors and assesses the likelihood that the anomalies correspond to actual faults. Considering the overall interference and confusion caused by local equipment codes, it further reduces noise interference by combining the correlations between different monitoring units, resulting in more accurate fault assessment results for monitoring. This invention reduces the interference impact on fault assessment and improves the accuracy and reliability of health monitoring through multi-dimensional data anomaly fluctuation stability analysis combined with the spatial correlations of overhead lines.

[0066] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0067] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for health monitoring of overhead power lines based on multi-dimensional monitoring sensors, characterized in that, The method includes: The operation data for each dimension in the historical time series of each monitoring unit is acquired on the overhead line, with a sampling frequency of once every 30 seconds; The analysis examines the degree of fluctuation anomaly in the operational data of each dimension at each time point to determine the corresponding abnormal dimension at that time point; based on the proportion of abnormal dimensions, it determines the abnormal time points that constitute abnormal time periods; based on the continuous distribution characteristics of abnormal dimensions within the abnormal time period, it determines the sub-time periods within the abnormal time period; and through the degree of fluctuation anomaly and the continuous stability of the operational data in all sub-time periods within the abnormal time period, it determines the fault concern factors for each abnormal time period. Between two monitoring units, based on the degree of overlap of abnormal dimensions and the similarity of operational data in each dimension during the intersecting abnormal periods, and combined with the degree of synchronization of the distribution of the intersecting abnormal periods, correlation evaluation indicators are obtained for the two monitoring units; and the related units of each monitoring unit are selected based on the correlation evaluation indicators. Based on the degree of correlation between each monitoring unit and related units, as well as the fault concern factors, the current fault assessment indicators for each monitoring unit are determined; and hierarchical monitoring and early warning are carried out based on the fault assessment indicators. For any dimension at any given time, calculate the difference between the data value of that dimension at that time and the data values ​​at adjacent times. Use the mean of all differences between data values ​​as the neighbor-to-neighbor fluctuation value of that dimension at that time. Use the mode of the neighbor-to-neighbor fluctuation values ​​of that dimension across all times as the baseline fluctuation value of that dimension. Calculate the difference between the neighbor-to-neighbor fluctuation value of that dimension at that time and the baseline fluctuation value to obtain the baseline deviation. Combine the baseline deviation and the neighbor-to-neighbor fluctuation value of that dimension to obtain the abnormal fluctuation of that dimension at that time. Dimensions whose abnormal fluctuation at that time is greater than a preset fluctuation threshold are considered abnormal dimensions at that time. At each time point, when the proportion of abnormal dimensions among all dimensions is greater than the preset abnormal proportion, the corresponding time point is designated as an abnormal time point; and the time period consisting of consecutive adjacent abnormal times points is designated as an abnormal time period. The method for obtaining the sub-period includes: for any abnormal period, taking the set of all abnormal dimensions that exist at the abnormal time in the abnormal period as the abnormal set; arranging and combining the abnormal dimensions in the abnormal set to obtain the dimension combination of the abnormal period; for any dimension combination in the abnormal period, when the abnormal times of the dimension combination appear to be continuously distributed, taking the period consisting of the continuously distributed abnormal times corresponding to the dimension combination as a sub-period within the abnormal period. For any sub-period within any abnormal period, the ratio of the total number of abnormal dimensions corresponding to all abnormal moments in that sub-period to the total number of all dimensions is taken as the abnormal coverage percentage of that sub-period; the mean of the abnormal volatility of each abnormal dimension at the abnormal moments in that sub-period is taken as the volatility factor of each abnormal dimension; the mean of the volatility factors of all abnormal dimensions in that sub-period is taken as the abnormal volatility index of that sub-period; combining the duration, abnormal coverage percentage, and abnormal volatility index of that sub-period, the local fault confidence level of that sub-period is obtained; the sum of the local fault confidence levels of all sub-periods is normalized to obtain the fault concern factor of that abnormal period.

2. The method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors according to claim 1, characterized in that, The methods for obtaining the correlation evaluation indicators include: For any two monitoring units, the time period in which the abnormal time periods of the two monitoring units intersect is obtained as the intersecting time period; for any intersecting time period, the number of the same abnormal dimensions in the two monitoring units at each time point of the intersecting time period is obtained as the overlap number; the average of the overlap numbers at all times of the intersecting time period is used as the overlap index. After calculating the correlation of the data of the two monitoring units in each dimension, the average correlation of all dimensions is used as the synchronization index of the two monitoring units. The product of the overlap index and the operation synchronization index of the intersecting period is taken as the local overlap degree of the intersecting period; the mean of the local overlap degrees of all intersecting periods in the two monitoring units is taken as the abnormal overlap degree of the two monitoring units. Based on the proportion of the abnormal time periods that intersect between two monitoring units in the total abnormal time period, and combined with the location distance distribution of the two monitoring units, the temporal synchronization of the two monitoring units is obtained. By combining the degree of abnormal overlap and temporal synchronization of each pair of monitoring units, a correlation evaluation index for each pair of monitoring units is obtained.

3. The method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors according to claim 2, characterized in that, The method for obtaining timing synchronization includes: For any intersecting time period, the ratio between the duration of the intersecting time period and the maximum duration of the abnormal time period corresponding to the two monitoring units is used as the synchronization coefficient of the intersecting time period; the average of the synchronization coefficients of all intersecting time periods is used as the abnormal synchronization index of the two monitoring units. The location distance between the two monitoring units is negatively correlated and normalized to serve as the distribution coefficient of the two monitoring units. The product of the distribution coefficients of the two monitoring units and the abnormal synchronization index is taken as the temporal synchronization of the two monitoring units.

4. The method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors according to claim 1, characterized in that, The method for obtaining the associated unit includes: For any given monitoring unit, monitoring units whose correlation evaluation index is greater than a preset correlation threshold are designated as associated units of that monitoring unit.

5. The method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors according to claim 1, characterized in that, The methods for obtaining the fault assessment indicators include: For any monitoring unit, the product of the minimum correlation evaluation index among the associated units corresponding to that monitoring unit and the total number of associated units is used as the correlation influence coefficient of that monitoring unit. The average value of the fault concern factors for all abnormal time periods in the time series of the monitoring unit is used as the fault probability. The product of the negative correlation mapping value of the correlation influence coefficient of the monitoring unit and the fault probability is used as the fault assessment index of the monitoring unit.

6. The method for monitoring the health of overhead lines based on multi-dimensional monitoring sensors according to claim 1, characterized in that, The hierarchical monitoring and early warning system based on fault assessment indicators includes: When the fault assessment index of the monitoring unit is greater than or equal to the preset high assessment threshold, the monitoring result is recorded as high risk; When the fault assessment index of the monitoring unit is greater than the preset low assessment threshold but less than the preset high assessment threshold, the monitoring result is recorded as a warning. When the fault assessment index of the monitoring unit is less than or equal to the preset low assessment threshold, the monitoring result is recorded as normal; the preset low assessment threshold is less than the preset high assessment threshold.

Images