Distribution automation fault monitoring and early warning system based on the Internet of Things

Abstract

The present invention relates to the field of voltage measurement technology, and specifically to a distribution automation fault monitoring and early warning system based on the Internet of Things; obtaining a metric distance according to the data difference characteristics of the neighborhood range of any two differential data points in the differential sequence of voltage time series data; clustering according to the metric distance to obtain a cluster cluster; obtaining a normal characterization degree, a target data point and a target characterization degree according to the time interval and metric distance of the differential data points in the cluster cluster, and the number of differential data points; adding the voltage values ​​at adjacent moments to be detected to the voltage time series data for a new round of clustering. The present invention obtains an abnormality degree according to the difference characteristics of the normal characterization degree and the target characterization degree corresponding to the voltage values ​​at adjacent moments to be detected, and the metric distance between the differential data points and the target data points corresponding to the voltage values ​​at adjacent moments to be detected; performing voltage monitoring according to the abnormality degree improves the accuracy of voltage abnormality monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of voltage measurement, and in particular to a distribution automation fault monitoring and early warning system based on the Internet of Things. Background Art

[0002] With the development of Internet of Things technology, the technology can be used to achieve remote monitoring and early warning of distribution boxes and other equipment in the power grid; collecting distribution equipment operation data through the Internet of Things and judging whether it is abnormal can help maintenance personnel quickly determine the fault location and fault information, improve response speed and reduce maintenance costs. The common method for monitoring and early warning of voltage operation data of the distribution network is to set a threshold. When the voltage exceeds the set threshold, an early warning will be processed; however, the voltage in the power grid will undergo certain dynamic changes as the power equipment is connected and disconnected, which can easily cause the voltage of the distribution equipment to exceed the set threshold instantaneously, resulting in voltage monitoring based on the method of setting the threshold. It is easy to ignore the normal voltage fluctuations, resulting in false alarms or missed alarms, etc., affecting the accuracy of voltage monitoring of the distribution equipment. Summary of the invention

[0003] In order to solve the technical problem that the monitoring accuracy is low due to the above-mentioned method of setting a threshold, the purpose of the present invention is to provide a distribution automation fault monitoring and early warning system based on the Internet of Things. The technical solution adopted is as follows:

[0004] A data acquisition module is used to obtain voltage time series data of the power distribution equipment during operation;

[0005] A data processing module, used for obtaining a differential order according to fluctuation difference characteristics of different data segments in the voltage time series data; and performing adaptive differentiation on the voltage time series data according to the differential order to obtain a differential sequence;

[0006] A feature analysis module is used to obtain the metric distance between any two differential data points in the differential sequence according to the data difference characteristics of the neighborhood range of any two differential data points; perform clustering by an iterative self-organizing clustering algorithm according to the metric distance to obtain different cluster clusters; obtain a normal representation degree according to the difference characteristics of the time interval between the differential data points in the cluster cluster and the metric distance, and the quantity characteristics of the differential data points;

[0007] A voltage measurement module is used to obtain a target data point and a target characterization degree according to the normal characterization degree of the cluster cluster; add the voltage values ​​of the adjacent detection moments of the voltage time series data to the voltage time series data and perform a new round of clustering to obtain different new cluster clusters; obtain anomaly degree according to the difference characteristics of the normal characterization degree of the new cluster cluster corresponding to the voltage values ​​at the adjacent detection moments and the target characterization degree, and the metric distance between the differential data point corresponding to the voltage value at the adjacent detection moment and the target data point; and perform voltage monitoring according to the anomaly degree of the voltage value at the adjacent detection moment.

[0008] Furthermore, the step of obtaining the differential order according to the fluctuation difference characteristics of different data segments in the voltage time series data includes:

[0009] A sliding window of a preset length is constructed to slide on the voltage time series data with a preset step size, and a data segment is obtained each time it slides; the absolute value of the difference between the data segment and the standard deviation of the voltage time series data is calculated to obtain a change characteristic value; the average value of all change characteristic values ​​is calculated and normalized to obtain a fluctuation coefficient; the product of the fluctuation coefficient and a preset order is calculated and rounded down to obtain the differential order.

[0010] Furthermore, the step of obtaining the metric distance between any two differential data points in the differential sequence according to the data difference characteristics of the neighborhood range of any two differential data points includes:

[0011] A neighborhood sequence is constructed based on other differential data points in a preset adjacent historical period of the differential data point; the dynamic time warping distance of the neighborhood sequence of any two differential data points is calculated according to the dynamic time warping algorithm to obtain the difference degree; the absolute value of the difference between any two differential data points is calculated to obtain the numerical difference value; the product of the difference degree and the numerical difference value is calculated to obtain the metric distance between any two differential data points.

[0012] Furthermore, the step of obtaining the normal representation degree according to the difference characteristics and metric distance of the time intervals between the differential data points in the clusters and the quantity characteristics of the differential data points includes:

[0013] Calculate the average value of the metric distance between the differential data points in the cluster cluster to obtain the average distance characteristic value; calculate the standard deviation of the time intervals between all differential data points and the differential data point with the closest time interval in the cluster cluster to obtain the interval difference value; calculate the product of the average distance characteristic value and the interval difference value and negatively correlate and map them to obtain the intra-cluster similarity; calculate the ratio of the number of differential data points between the cluster cluster and the differential sequence to obtain the number proportion value; calculate the product of the intra-cluster similarity and the number proportion value to obtain the normal representation degree of the cluster cluster.

[0014] Furthermore, the step of obtaining target data points and target representation degrees according to the normal representation degrees of the clusters includes:

[0015] The maximum value of the normal representation degree is used as the target representation degree, and the centroid of the cluster corresponding to the target representation degree is used as the target data point.

[0016] Furthermore, the step of obtaining the abnormality degree according to the difference characteristics of the normal characterization degree of the new cluster corresponding to the voltage values ​​at the adjacent moments to be detected and the target characterization degree, and the metric distance between the differential data point corresponding to the voltage values ​​at the adjacent moments to be detected and the target data point comprises:

[0017] The ratio of the target characterization degree to the normal characterization degree of the new cluster corresponding to the voltage value at the adjacent time to be detected is calculated to obtain the abnormality coefficient; the product of the metric distance between the differential data point corresponding to the voltage value at the adjacent time to be detected and the target data point and the abnormality coefficient is calculated and normalized to obtain the abnormality of the voltage value at the adjacent time to be detected.

[0018] Furthermore, the step of performing voltage monitoring according to the abnormality of the voltage values ​​at the adjacent detection moments includes:

[0019] When the abnormality exceeds a preset abnormality threshold, the voltage values ​​at the adjacent detection moments are abnormal.

[0020] The present invention has the following beneficial effects:

[0021] In the present invention, obtaining the differential order can more accurately determine the number of differentials according to the complexity of the change of the voltage time series data; so that the differential sequence can more clearly highlight the difference between the normal voltage data and the abnormal voltage data; obtaining the differential sequence can avoid the situation where the voltage time series data presents a sinusoidal change feature, resulting in low accuracy of abnormal monitoring. Obtaining the metric distance can determine the similarity of the differential data points according to the data difference characteristics of the neighborhood range between the differential data points, and then cluster the differential data points according to the metric distance, and distinguish the normal and abnormal voltage data into different cluster clusters; improve the accuracy of subsequent voltage abnormal monitoring. Obtaining the normal characterization degree can quantify the characteristics of the differential data points corresponding to the normal voltage data in the cluster cluster, so that it can determine whether the cluster cluster is normal according to the normal characterization degree and measure the accuracy of the subsequent voltage value to be detected. Obtaining the target data point and the target characterization degree can be used to determine the accuracy of the subsequent voltage value to be detected. Obtaining the abnormality degree can characterize whether the voltage value at the adjacent detection time is abnormal, and finally, according to the abnormality degree, the abnormality monitoring can be performed through the voltage change characteristics of the distribution equipment itself, which improves the monitoring accuracy compared to monitoring by setting a threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 A block diagram of a distribution automation fault monitoring and early warning system based on the Internet of Things provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a distribution automation fault monitoring and early warning system based on the Internet of Things proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0025] Unless defined otherwise, 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 belongs.

[0026] The specific scheme of a distribution automation fault monitoring and early warning system based on the Internet of Things provided by the present invention is described in detail below with reference to the accompanying drawings.

[0027] See also Figure 1 , which shows a block diagram of a distribution automation fault monitoring and early warning system based on the Internet of Things provided by an embodiment of the present invention, the system includes the following modules:

[0028] The data acquisition module S1 is used to obtain the voltage time series data of the power distribution equipment during operation.

[0029] In an embodiment of the present invention, the implementation scenario is to perform abnormal voltage monitoring on the distribution equipment. In the distribution automation fault monitoring and early warning system, it is necessary to monitor the voltage and other data in the distribution equipment, so as to timely discover the abnormal voltage fluctuations and overload problems of the power grid line or the distribution equipment, and improve the comprehensiveness and accuracy of fault monitoring during the operation of the distribution equipment. First, the voltage time series data of the distribution equipment is obtained. In the embodiment of the present invention, the voltage within 24 hours is collected to construct the voltage time series data. The implementer can determine the collection duration and frequency of the voltage time series data according to the implementation scenario.

[0030] The data processing module S2 is used to obtain a differential order according to the fluctuation difference characteristics of different data segments in the voltage time series data; and to perform adaptive differentiation on the voltage time series data according to the differential order to obtain a differential sequence.

[0031] During the operation of the distribution equipment, the collected voltage is related to the load changes borne by the distribution equipment. The number of power-consuming equipment connected to the power grid in different time periods may be different, which makes the voltage values ​​collected in different time periods during the operation of the distribution equipment different; even when the power-consuming equipment is connected to the power grid, there may be instantaneous voltage fluctuations, which may cause false alarms or omissions when monitoring voltage anomalies by setting thresholds, and the monitoring accuracy is low. Under normal circumstances, the voltage of the power grid tends to change like a sine wave, and the change characteristics are regular and stable; when the voltage is abnormal, there is a big difference between the voltage change at the abnormal moment and the voltage change at the normal moment. Since the collected voltage data changes show a sine wave feature, the voltage data under normal circumstances will also have different values, making it difficult to highlight the difference characteristics between abnormal data and normal data in the subsequent abnormal analysis process; therefore, in order to highlight the difference characteristics between normal data and abnormal data, it is necessary to perform differential processing on the voltage time series data. The differential values ​​of the data with regular sinusoidal changes after multiple differences are small and similar, while the differential values ​​of the data with irregular abnormal changes after multiple differences are still relatively discrete; therefore, the abnormal voltage data can be highlighted after differentiation, thereby improving the accuracy of subsequent abnormal analysis. If the changes in the voltage time series data are more complex, the differential order needs to be larger and the number of differentials needs to be more, so that the data after differentiation is as stable as possible, thereby highlighting the abnormal voltage data. Therefore, the differential order is obtained according to the fluctuation difference characteristics of different data segments in the voltage time series data.

[0032] Preferably, in an embodiment of the present invention, the step of obtaining the differential order includes: constructing a sliding window of a preset length to slide on the voltage time series data with a preset step length, and obtaining a data segment each time it slides; in an embodiment of the present invention, the preset length is 7 data lengths, the preset step length is 3, and a sliding window of length 7 slides 3 data each time to obtain a data segment on the voltage time series data length, which can be determined by the implementer according to the implementation scenario. Calculate the absolute value of the difference between the data segment and the standard deviation of the voltage time series data to obtain a change characteristic value; when the change characteristic value is larger, it means that the difference between the change characteristics of the data segment and the voltage time series value is larger. Calculate the average value of all change characteristic values ​​and normalize them to obtain a fluctuation coefficient; when the fluctuation coefficient is larger, it means that the difference between the change characteristics of the local data and the overall data in the voltage time series data is larger, the data change of the voltage time series data is more complex or the abnormal data is more obvious, and multiple differential processing is more needed to highlight the abnormal data; when the fluctuation coefficient is smaller, it means that the difference between the change characteristics of the local data and the overall data in the voltage time series data is smaller, the data change of the voltage time series data is relatively stable or the possibility of abnormal data is smaller. The product of the fluctuation coefficient and the preset order is calculated and rounded down to obtain the differential order; in the embodiment of the present invention, the preset order is 5, and the implementer can determine it according to the implementation scenario.

[0033] Furthermore, the voltage time series data can be adaptively differentiated according to the differential order to obtain a differential sequence; it should be noted that differentiation belongs to the prior art, and the specific steps will not be repeated; for example, if the differential order is 3, the voltage time series data is differentiated 3 times, and the differential sequence is obtained according to the final differential value. If the voltage time series data presents a normal sinusoidal change, the value of each differential data point in the differential sequence is similar and small. If there is some abnormal voltage data, the value of the differential data point corresponding to the part is relatively discrete; and then the abnormal voltage can be analyzed based on the differential sequence.

[0034] The feature analysis module S3 is used to obtain the metric distance between any two differential data points according to the data difference characteristics of the neighborhood range of any two differential data points in the differential sequence; clustering is performed through an iterative self-organizing clustering algorithm according to the metric distance to obtain different cluster clusters; and normal representation is obtained according to the difference characteristics of the time interval between the differential data points in the cluster cluster and the metric distance, and the quantity characteristics of the differential data points.

[0035] There is a clear difference between abnormal data and normal data in the differential sequence, so clustering can be performed based on the differential data, and the clustering results can be used for subsequent abnormal monitoring of voltage data. The differential values ​​of the normal voltage period are similar and the change trends are close, while the differential values ​​of the normal period and the abnormal period are quite different and the change trends are different. Therefore, the metric distance of any two differential data points can be obtained based on the data difference characteristics of the neighborhood range of any two differential data points in the differential sequence.

[0036] Preferably, in an embodiment of the present invention, the step of obtaining the metric distance includes: constructing a neighborhood sequence based on other differential data points within a preset adjacent historical period of the differential data point; in an embodiment of the present invention, the preset adjacent historical period is the period of 50 other historical differential data points adjacent to the differential data point, and the neighborhood sequence is composed of the 50 other differential data points, which can be set by the implementer according to the implementation scenario. The dynamic time warping distance of the neighborhood sequence of any two differential data points is calculated according to the dynamic time warping algorithm to obtain the difference degree; it should be noted that the dynamic time warping algorithm belongs to the prior art, and the more similar the two neighborhood sequences are, the smaller the dynamic time warping distance and the smaller the difference degree; the greater the difference between the two neighborhood sequences, the greater the difference degree, which means that the two differential data points are more likely to be normal data and abnormal data respectively. Calculate the absolute value of the difference between any two differential data points to obtain a numerical difference value; when the numerical difference value is larger, it means that the two differential data points are more likely to represent the correct situation and the abnormal situation respectively. The product of the difference degree and the numerical difference value is calculated to obtain the metric distance of any two differential data points; when the metric distance is larger, it means that the voltage change characteristics corresponding to the two differential data points are more dissimilar, and they are less likely to be in the same cluster in the subsequent clustering process; when the metric distance is smaller, it means that the voltage change characteristics corresponding to the two differential data points are more similar.

[0037] Furthermore, after obtaining the metric distance between different differential data points, clustering can be performed according to the metric distance through an iterative self-organizing clustering algorithm to obtain different cluster clusters; different cluster clusters represent different voltage characteristics, and the voltage characteristics within the same cluster cluster are similar; it should be noted that the iterative self-organizing clustering algorithm belongs to the prior art, and the specific steps are not repeated. In order to monitor the voltage state at subsequent moments, it is necessary to determine the cluster cluster that represents the normal voltage change characteristics; there are a large number of normal voltage values ​​in the voltage time series data under normal conditions, so the number of differential data points in the cluster cluster that represents the normal voltage change characteristics is the largest and the metric distance between the differential data points is small; the voltage value under normal conditions presents a regular sinusoidal wave change characteristic, so the time intervals of the differential data points in the cluster cluster are similar; therefore, the normal characterization degree can be obtained according to the difference characteristics of the time intervals between the differential data points in the cluster cluster, the quantity characteristics of the differential data points, and the numerical characteristics.

[0038] Preferably, in an embodiment of the present invention, the step of obtaining the normal characterization degree includes: calculating the average value of the metric distance between the differential data points in the cluster cluster to obtain the average distance characteristic value; when the average distance characteristic value is smaller, it means that the voltage change characteristics corresponding to the differential data points in the cluster cluster are more similar, and the more likely it is a normal voltage change characteristic. Calculate the standard deviation of the time intervals between all differential data points and the differential data points with the closest time interval in the cluster cluster to obtain the interval difference value; when the interval difference value is smaller, it means that the time intervals between the differential data points in the cluster cluster and the differential data points with the closest time interval are more similar, and the more likely it is to represent the regularity characteristics of normal voltage changes. Calculate the product of the average distance characteristic value and the interval difference value and negatively correlate and map them to obtain the intra-cluster similarity; when the intra-cluster similarity is larger, it means that the differential data points in the cluster cluster are more likely to represent normal voltage change characteristics. Calculate the ratio of the number of differential data points of the cluster cluster and the differential sequence to obtain the number proportion value; when the data proportion value is larger, it means that the number of differential data points in the cluster cluster is larger, and the more likely it is to represent normal voltage change characteristics. Calculate the product of the similarity within the cluster and the quantity ratio to obtain the normal representation degree of the cluster; the larger the normal representation degree, the more likely the cluster is to represent the normal voltage change characteristics. The formula for obtaining the normal representation degree includes:

[0039]

[0040] In the formula, R represents the normal representation of the cluster, represents an exponential function with a natural constant as the base, X represents the average distance characteristic value, F represents the interval difference value, represents the similarity within the cluster, M represents the number of differential data points within the cluster, and N represents the number of differential data points in the differential sequence. Indicates the quantity percentage.

[0041] The voltage measurement module S4 is used to obtain the target data point and the target characterization degree according to the normal characterization degree of the clustering cluster; add the voltage values ​​of the adjacent detection moments of the voltage time series data to the voltage time series data and perform a new round of clustering to obtain different new clustering clusters; obtain the abnormality degree according to the difference characteristics of the normal characterization degree and the target characterization degree of the new clustering cluster corresponding to the voltage values ​​at the adjacent detection moments, and the measured distance between the differential data point corresponding to the voltage values ​​at the adjacent detection moments and the target data point; and perform voltage monitoring according to the abnormality degree of the voltage values ​​at the adjacent detection moments.

[0042] After obtaining the normal characterization degrees of all cluster clusters, the cluster cluster with the greatest possibility of representing the normal voltage change characteristics and the corresponding normal characterization degree can be determined, so the target data point and the target characterization degree are obtained according to the normal characterization degree of the cluster cluster; preferably, in an embodiment of the present invention, the step of obtaining the target data point and the target characterization degree includes: taking the maximum value of the normal characterization degree as the target characterization degree, which measures the cluster cluster that is most likely to represent the normal voltage change characteristics; taking the centroid of the cluster cluster corresponding to the target characterization degree as the target data point, which can be used to measure whether the subsequent voltage value is abnormal. Then, the voltage values ​​at the subsequent adjacent moments of the voltage time series data can be continuously analyzed for abnormalities, so the voltage values ​​at the adjacent moments to be detected of the voltage time series data are added to the voltage time series data and a new round of clustering is performed to obtain different new cluster clusters; after adding the voltage values ​​at the adjacent moments to be detected to the voltage time series data, adaptive difference and metric distance calculations are performed, and a new round of clustering is performed based on the results to obtain different new cluster clusters; if the voltage values ​​at the adjacent moments to be detected are abnormal data or gradually approach abnormal conditions, the differential data points corresponding to the voltage values ​​at the adjacent moments to be detected will be clustered into the more abnormal cluster clusters, and the metric distance between the differential data points and the target data points is larger; therefore, the degree of abnormality can be obtained based on the difference characteristics of the normal representation degree and the target representation degree of the new cluster cluster corresponding to the voltage values ​​at the adjacent moments to be detected, and the metric distance between the differential data points corresponding to the voltage values ​​at the adjacent moments to be detected and the target data points.

[0043] Preferably, in an embodiment of the present invention, the step of obtaining the degree of abnormality includes: calculating the ratio of the target characterization degree to the normal characterization degree of the new cluster corresponding to the voltage value at the adjacent time to be detected, and obtaining the abnormality coefficient; if the voltage value is more abnormal, the normal characterization degree of the new cluster divided is smaller, the abnormality coefficient is larger, and the metric distance between the differential data point corresponding to the voltage value and the target data point is larger. Calculate the product of the metric distance between the differential data point corresponding to the voltage value at the adjacent time to be detected and the target data point and normalize them to obtain the degree of abnormality of the voltage value at the adjacent time to be detected; when the degree of abnormality is larger, it means that the voltage value at the adjacent time to be detected is more abnormal; conversely, when the degree of abnormality is smaller, it means that the voltage value is more normal. The formula for obtaining the degree of abnormality includes:

[0044]

[0045] In the formula, Y represents the abnormality, represents the normalization function, G represents the target representation degree, and H represents the normal representation degree of the new cluster corresponding to the voltage value at the adjacent detection time; represents the abnormal coefficient, and S represents the metric distance between the differential data point corresponding to the voltage value at the adjacent detection time and the target data point.

[0046] Further, voltage monitoring can be performed according to the abnormality of the voltage value at the adjacent detection time, specifically including: when the abnormality exceeds the preset abnormal threshold, the voltage value at the adjacent detection time is abnormal and needs to be checked in time; in the embodiment of the present invention, the preset abnormal threshold is 0.5, and the implementer can determine it according to the implementation scenario. Then continue to monitor the voltage value at the next detection time for abnormality, and loop the abnormality acquisition step; as time is updated, the voltage state of the distribution equipment is monitored in real time. So far, by adaptively differentiating the voltage time series data to obtain the differential sequence, the voltage abnormal data is highlighted and the clustering accuracy is improved, the monitoring accuracy is initially improved, and the clustering cluster with normal voltage change characteristics is judged according to the clustering results; a new round of clustering is performed on the voltage values ​​at the subsequent detection time and compared with the normal voltage situation to obtain the abnormality; according to the abnormality, the voltage change characteristics of the distribution equipment itself can be used for abnormal monitoring, which improves the monitoring accuracy compared to the monitoring by setting the threshold.

[0047] In summary, the embodiment of the present invention provides a distribution automation fault monitoring and early warning system based on the Internet of Things; obtain the metric distance according to the data difference characteristics of the neighborhood range of any two differential data points in the differential sequence of voltage time series data; cluster according to the metric distance to obtain cluster clusters; obtain the normal characterization degree, target data points and target characterization degree according to the time interval and metric distance of the differential data points in the cluster clusters, and the number of differential data points; add the voltage values ​​at adjacent moments to be detected to the voltage time series data for a new round of clustering. The present invention obtains the abnormality degree according to the difference characteristics of the normal characterization degree and the target characterization degree corresponding to the voltage values ​​at adjacent moments to be detected, and the metric distance between the differential data points and the target data points corresponding to the voltage values ​​at adjacent moments to be detected; perform voltage monitoring according to the abnormality degree, thereby improving the accuracy of voltage abnormality monitoring.

[0048] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0049] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A distribution automation fault monitoring and early warning system based on the Internet of Things, characterized in that: The system includes the following modules: A data acquisition module is used to obtain voltage time series data of the power distribution equipment during operation; A data processing module, used for obtaining a differential order according to fluctuation difference characteristics of different data segments in the voltage time series data; Adaptively differentiating the voltage time series data according to the differential order to obtain a differential sequence; A feature analysis module is used to obtain the metric distance between any two differential data points in the differential sequence according to the data difference characteristics of the neighborhood range of any two differential data points; perform clustering by an iterative self-organizing clustering algorithm according to the metric distance to obtain different cluster clusters; obtain a normal representation degree according to the difference characteristics of the time interval between the differential data points in the cluster cluster and the metric distance, and the quantity characteristics of the differential data points; A voltage measurement module, used to obtain a target data point and a target characterization degree according to the normal characterization degree of the clustering cluster; add the voltage values ​​of the adjacent to-be-detected moments of the voltage time series data to the voltage time series data and perform a new round of clustering to obtain different new clustering clusters; The abnormality degree is obtained according to the difference characteristics of the normal characterization degree of the new cluster cluster corresponding to the voltage values ​​at the adjacent moments to be detected and the target characterization degree, and the metric distance between the differential data points corresponding to the voltage values ​​at the adjacent moments to be detected and the target data points; and voltage monitoring is performed according to the abnormality degree of the voltage values ​​at the adjacent moments to be detected.

2. According to the Internet of Things-based distribution automation fault monitoring and early warning system according to claim 1, it is characterized in that: The step of obtaining the differential order according to the fluctuation difference characteristics of different data segments in the voltage time series data includes: A sliding window of a preset length is constructed to slide on the voltage time series data with a preset step size, and a data segment is obtained each time it slides; the absolute value of the difference between the data segment and the standard deviation of the voltage time series data is calculated to obtain a change characteristic value; the average value of all change characteristic values ​​is calculated and normalized to obtain a fluctuation coefficient; the product of the fluctuation coefficient and a preset order is calculated and rounded down to obtain the differential order.

3. According to the Internet of Things-based distribution automation fault monitoring and early warning system of claim 1, it is characterized in that: The step of obtaining the metric distance between any two differential data points in the differential sequence according to the data difference characteristics of the neighborhood range of any two differential data points comprises: A neighborhood sequence is constructed based on other differential data points in a preset adjacent historical period of the differential data point; the dynamic time warping distance of the neighborhood sequence of any two differential data points is calculated according to the dynamic time warping algorithm to obtain the difference degree; the absolute value of the difference between any two differential data points is calculated to obtain the numerical difference value; the product of the difference degree and the numerical difference value is calculated to obtain the metric distance between any two differential data points.

4. According to the Internet of Things-based distribution automation fault monitoring and early warning system of claim 1, it is characterized in that: The step of obtaining a normal representation degree according to the difference characteristics and metric distance of the time intervals between the differential data points in the clusters and the quantity characteristics of the differential data points comprises: Calculate the average value of the metric distance between the differential data points in the cluster cluster to obtain the average distance characteristic value; calculate the standard deviation of the time intervals between all differential data points and the differential data point with the closest time interval in the cluster cluster to obtain the interval difference value; calculate the product of the average distance characteristic value and the interval difference value and negatively correlate and map them to obtain the intra-cluster similarity; calculate the ratio of the number of differential data points between the cluster cluster and the differential sequence to obtain the number proportion value; calculate the product of the intra-cluster similarity and the number proportion value to obtain the normal representation degree of the cluster cluster.

5. According to the Internet of Things-based distribution automation fault monitoring and early warning system of claim 1, it is characterized in that: The step of obtaining target data points and target representation degrees according to the normal representation degrees of the clusters comprises: The maximum value of the normal representation degree is used as the target representation degree, and the centroid of the cluster corresponding to the target representation degree is used as the target data point.

6. The distribution automation fault monitoring and early warning system based on the Internet of Things according to claim 1 is characterized in that: The step of obtaining the abnormality degree according to the difference characteristics between the normal characterization degree of the new cluster corresponding to the voltage values ​​at the adjacent detection moments and the target characterization degree and the metric distance between the differential data point corresponding to the voltage values ​​at the adjacent detection moments and the target data point comprises: The ratio of the target characterization degree to the normal characterization degree of the new cluster corresponding to the voltage value at the adjacent time to be detected is calculated to obtain the abnormality coefficient; the product of the metric distance between the differential data point corresponding to the voltage value at the adjacent time to be detected and the target data point and the abnormality coefficient is calculated and normalized to obtain the abnormality of the voltage value at the adjacent time to be detected.

7. The distribution automation fault monitoring and early warning system based on the Internet of Things according to claim 1 is characterized in that: The step of performing voltage monitoring according to the abnormality of the voltage values ​​at the adjacent detection moments comprises: When the abnormality exceeds a preset abnormality threshold, the voltage values ​​at the adjacent detection moments are abnormal.

Images