Four-fusion integrated terminal with electric energy quality monitoring function

Abstract

The invention relates to the technical field of data processing, in particular to a four-fusion integrated terminal with an electric energy quality monitoring function, and the terminal specifically comprises a data collection module which is used for obtaining the parameter value of each monitoring index at each monitoring point; the data analysis module is used for obtaining the data stability degree of each monitoring point according to the change characteristics of each parameter value; the feature acquisition module is used for combining the parameter value of each monitoring index with the data stability degree of each monitoring point to obtain the distribution uniformity degree of each monitoring index; and the abnormity monitoring module is used for screening out target monitoring indexes according to the distribution uniformity, clustering parameter values of the target monitoring indexes by using a K-means algorithm, and monitoring the power quality according to abnormal data in each cluster, so that feature dimension reduction is realized, and the monitoring accuracy is improved. The effect and accuracy of monitoring the power quality by using the K-means algorithm are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a four-in-one terminal with a power quality monitoring function. Background Art

[0002] Power quality monitoring is a key link in the efficient and stable operation of power systems. It is directly related to the stability, reliability and normal operation of power systems. Especially for large-scale power supply and distribution plants, the stable operation of power systems is extremely important to ensure the output, benefits and safety of enterprises.

[0003] In the traditional way, the four-in-one terminal is used to monitor the power quality. The four-in-one terminal refers to an electric power measurement and control terminal that integrates all photovoltaic four-in-one devices, takes the circuit breaker as the basic carrier, and combines the interface converter and current transformer to form an observable, measurable, controllable and adjustable function. The four-in-one terminal uses a metering chip to measure all monitoring indicators of power quality, and uses the K-means algorithm to cluster the monitoring data of all monitoring indicators of power quality within the preset time period to obtain abnormal values, and then realizes the monitoring and analysis of power quality based on the abnormal values.

[0004] However, since there are many monitoring indicators in the monitoring of power quality, such as voltage deviation, frequency deviation, voltage flicker, etc., there are many data features when clustering using the K-means algorithm. The final clustering result may not reflect the optimal data distribution, resulting in poor monitoring effect on abnormal data and inaccurate monitoring results on power quality.

[0005] Therefore, how to perform feature dimensionality reduction to improve the accuracy of power quality monitoring using the K-means algorithm in the four-in-one terminal has become an urgent problem to be solved. Summary of the invention

[0006] In view of this, an embodiment of the present invention provides a four-in-one terminal with power quality monitoring function to solve the problem of how to perform feature dimensionality reduction to improve the accuracy of power quality monitoring using the K-means algorithm in the four-in-one terminal.

[0007] An embodiment of the present invention provides a four-in-one terminal with a power quality monitoring function, the terminal comprising the following steps: The data acquisition module is used to obtain the parameter values ​​of each monitoring indicator at each monitoring point in the power system, and obtain the time series of the parameter values ​​of each monitoring indicator at each monitoring point at the current sampling time; A data analysis module, used to obtain a change characteristic value of each parameter value in each parameter value time series according to the data fluctuation in each parameter value time series, and obtain the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in all parameter value time series corresponding to each monitoring point; A feature acquisition module is used to obtain a target sequence for any monitoring indicator according to the parameter value of any monitoring indicator at each monitoring point at the current sampling time, and obtain the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation amplitude of the data points in the target sequence; The abnormality monitoring module is used to obtain the distribution uniformity of each of the monitoring indicators, screen out the target monitoring indicators according to all the distribution uniformity, respectively form sample data with the parameter values ​​of the target monitoring indicators at each of the monitoring points at the current sampling time, cluster the sample data using the K-means algorithm to obtain at least one cluster, obtain the abnormality degree of each data point in each of the clusters, and monitor the power quality of the power system according to all the abnormality degrees.

[0008] Preferably, obtaining the change characteristic value of each parameter value in each parameter value time series sequence according to the data fluctuation in each parameter value time series sequence includes: For any parameter value in any parameter value time series sequence, calculating the deviation of the any parameter value from the mean in the any parameter value time series sequence to obtain the degree of deviation of the any parameter value; If the any parameter value is not the first parameter value or the last parameter value in the any parameter value time series sequence, respectively obtaining a left adjacent parameter value and a right adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating a first difference between the any parameter value and the left adjacent parameter value, and a second difference between the any parameter value and the right adjacent parameter value, and taking an average of the absolute values ​​between the first difference and the second difference as a degree of change of the any parameter value; If the any parameter value is the first parameter value in the any parameter value time series sequence, obtaining the right adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating the absolute value of the difference between the any parameter value and the right adjacent parameter value, and obtaining the degree of change of the any parameter value; If the any parameter value is the last parameter value in the any parameter value time series sequence, obtaining the left adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating the absolute value of the difference between the any parameter value and the left adjacent parameter value, and obtaining the degree of change of the any parameter value; The sum of the variation degree and the deviation degree of any parameter value is calculated to obtain a variation characteristic value of any parameter value in the time series of any parameter value.

[0009] Preferably, obtaining the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in the time series of all parameter values ​​corresponding to each monitoring point includes: Record any one of the monitoring points as a target monitoring point, record other parameter value time series sequences corresponding to the target monitoring point as reference sequences for any parameter value time series sequence corresponding to the target monitoring point, obtain all peak data in any one of the parameter value time series sequences, and obtain the weight of each peak data in any one of the parameter value time series sequences according to the change characteristic value of each peak data and the change characteristic value of each data in the reference sequence; According to the weight of each of the peak data, weighted sum is performed on all the peak data in the any parameter value time series sequence to obtain the similarity between the any parameter value time series sequence and all reference sequences; The similarity between each parameter value time series sequence corresponding to the target monitoring point and other parameter value time series sequences is calculated respectively, all similarities are accumulated, and the accumulated results are inversely normalized to obtain the data stability of the target monitoring point.

[0010] Preferably, obtaining the weight of each peak data in any parameter value time series sequence according to the change characteristic value of each peak data and the change characteristic value of the data in each reference sequence includes: For any peak data in any parameter value time series sequence, the sampling moment of any peak data is recorded as the target moment, the change characteristic value of the parameter value at the target moment is obtained in each reference sequence, recorded as the reference change characteristic value, and the average value of the absolute value of the difference between the change characteristic value of any peak data and each reference change characteristic value is calculated to obtain the weight of any peak data.

[0011] Preferably, obtaining the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation range of the data points in the target sequence includes: Recording the other data in the target sequence except the first data and the last data as data to be analyzed, and obtaining a first variable in the target sequence according to the difference between each data to be analyzed and the data before and after it, and the data stability of the monitoring point corresponding to each data to be analyzed; In the target sequence, a second variable is obtained according to the data stability of the monitoring points corresponding to every two adjacent data; The first variable and the second variable are accumulated to obtain the distribution uniformity of any monitoring indicator.

[0012] Preferably, obtaining the first variable according to the difference between each of the data to be analyzed and the data before and after it, and the data stability of the monitoring point corresponding to each of the data to be analyzed, comprises: For any data to be analyzed, in the target sequence, the left adjacent target data and the right adjacent target data of the data to be analyzed are obtained, a first absolute value of the difference between the data to be analyzed and the left adjacent target data, and a second absolute value of the difference between the data to be analyzed and the right adjacent target data are calculated, the absolute value of the difference between the first absolute value of the difference and the second absolute value of the difference is taken as the degree of difference between the data to be analyzed and its preceding and succeeding data, the product of the inverse of the degree of difference and the degree of data stability of the monitoring point corresponding to the data to be analyzed is calculated, and the degree of local stability of the data to be analyzed is obtained; The accumulated value of the local stability degree of each of the data to be analyzed is linearly normalized to obtain a first variable.

[0013] Preferably, in the target sequence, obtaining the second variable according to the data stability of the monitoring points corresponding to every two adjacent data includes: For any data in the target sequence, if the any data is not the first data or the last data in the target sequence, respectively obtain the left adjacent data and the right adjacent data of the any data in the target sequence, calculate the third absolute difference value of the data stability of the monitoring point corresponding to the any data and the left adjacent data, and calculate the fourth absolute difference value of the data stability of the monitoring point corresponding to the any data and the right adjacent data, calculate the average value between the third absolute difference value and the fourth absolute difference value, and obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the first data in the target sequence, then the right adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the right adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the last data in the target sequence, the left adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the left adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; The accumulated values ​​of the stability similarity indexes of the monitoring points corresponding to all the data in the target sequence are normalized in inverse proportion to obtain a second variable.

[0014] Preferably, the target monitoring indicators are screened out according to the uniformity of all distributions, including: The distribution uniformity of all monitoring indicators is sorted in descending order to obtain a distribution uniformity sequence, and the monitoring indicators corresponding to the first preset number of data are selected from the distribution uniformity sequence and recorded as target monitoring indicators.

[0015] Preferably, the obtaining of the abnormality degree of each data point in each of the clusters and monitoring the power quality of the power system according to all abnormality degrees includes: Record any cluster as a target cluster, and for any data point in the target cluster, respectively calculate the Euclidean distance between the data point and each other data point in the target cluster, perform linear normalization on the average of all Euclidean distances, and obtain the abnormality degree of the data point; If the abnormality level is greater than or equal to a preset abnormality level threshold, any of the data points is regarded as an abnormal data point; Acquire all abnormal data points in all clusters, accumulate the abnormality degrees of all abnormal data points, and obtain the power quality abnormality value of the power system at the current sampling moment. If the power quality abnormality value is greater than or equal to the preset power quality abnormality threshold, it is determined that the power quality of the power system is abnormal at the current sampling moment.

[0016] Preferably, obtaining the target sequence according to the parameter value of any monitoring indicator at each monitoring point at the current sampling time includes: The parameter values ​​of any monitoring indicator at each monitoring point at the current sampling moment are sorted in ascending order to obtain a target sequence.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention includes a data acquisition module, which is used to respectively obtain the parameter values ​​of various monitoring indicators at each monitoring point in the power system, and obtain the parameter value time series sequence of various monitoring indicators at each monitoring point at the current sampling time; a data analysis module, which is used to obtain the change characteristic value of each parameter value in each parameter value time series sequence according to the data fluctuation in each parameter value time series sequence, and obtain the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in all parameter value time series sequences corresponding to each monitoring point; a feature acquisition module, which is used to obtain the target value according to the parameter value of any monitoring indicator at each monitoring point at the current sampling time. A sequence, according to the data stability of the monitoring point corresponding to each data in the target sequence and the change amplitude of the data points in the target sequence, the distribution uniformity of any monitoring indicator is obtained; an abnormality monitoring module is used to obtain the distribution uniformity of each of the monitoring indicators, select the target monitoring indicator according to all the distribution uniformity, and respectively form sample data with the parameter value of the target monitoring indicator at each monitoring point at the current sampling time, cluster the sample data using the K-means algorithm to obtain at least one cluster, obtain the abnormality of each data point in each cluster, and monitor the power quality of the power system according to all the abnormality. Among them, the degree of interference of each monitoring point with noise is judged according to the data stability of each monitoring point. In the target sequence corresponding to each monitoring indicator, the distribution uniformity index of each monitoring indicator is obtained according to the stability of the data and the data stability of the monitoring point corresponding to each data, which is used to evaluate the clustering effect of each monitoring indicator, so as to screen out the target monitoring indicators with stable data changes, good clustering effects and less noise interference at the current sampling moment. Then, according to the data of the target monitoring indicators of each monitoring point at the current sampling moment, the K-means algorithm is used for clustering, and the power quality of the power system is monitored according to the clustering results to realize feature dimensionality reduction and improve the accuracy of power quality monitoring using the K-means algorithm in the four-in-one terminal. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be 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 labor.

[0019] Figure 1 It is a structural block diagram of a four-in-one terminal with power quality monitoring function provided by Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0020] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0021] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0022] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0023] The specific scenario targeted by the present invention is: when using the K-means algorithm to perform abnormal monitoring on the power quality data of the power system, due to the presence of multiple dimensions in the power quality data, the amount of calculation is large in the K-means clustering process, and the clustering effect of high-dimensional clustering is not good. The final cluster cluster cannot reflect the optimal data distribution, resulting in poor monitoring effect on abnormal data in the cluster cluster and inaccurate power quality monitoring results. Therefore, the present invention analyzes the data distribution of each monitoring indicator to achieve dimensionality reduction of the cluster data and improves the accuracy of the four-in-one terminal using the K-means algorithm to monitor power quality.

[0024] See also Figure 1 , is a structural block diagram of a four-in-one terminal with power quality monitoring function provided by the first embodiment of the present invention, such as Figure 1 As shown, the terminal may include: The data acquisition module 11 respectively obtains the parameter values ​​of various monitoring indicators at each monitoring point in the power system, and obtains the time series of the parameter values ​​of various monitoring indicators at each monitoring point at the current sampling time.

[0025] The sampling interval is set to 10 minutes and the sampling time is set to 24 hours. There is no restriction here. The implementer can set it according to the specific scenario. Through the four-in-one terminal, the parameter values ​​of the power quality-related monitoring indicators at each monitoring point in the power system within 24 hours, such as voltage deviation, frequency deviation, voltage flicker, three-phase voltage imbalance, harmonic distortion rate, etc., are obtained, and the parameter value time series sequence of each monitoring indicator at each monitoring point is obtained, which is convenient for the subsequent analysis of each monitoring indicator. Among them, the last parameter value in the parameter value time series sequence is the parameter value at the current sampling moment. At the same time, the sampling moment, monitoring indicator and monitoring point are respectively used as a dimension, so that the collected parameter values ​​are stored in the database of the four-in-one terminal in the form of a three-dimensional matrix. Among them, the three-dimensional matrix is ​​a prior art and will not be repeated here.

[0026] The data analysis module 12 obtains the change characteristic value of each parameter value in each parameter value time series sequence according to the data fluctuations in each parameter value time series sequence, and obtains the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in all parameter value time series sequences corresponding to each monitoring point.

[0027] At present, when the power quality of the power system is monitored abnormally, the parameter values ​​of all monitoring indicators at each monitoring point of the power system are usually clustered using the K-means algorithm to detect abnormal data according to the clustering results, thereby realizing abnormal monitoring of the power quality. However, due to the large number of monitoring indicators of the power quality, the time complexity and space complexity of the K-means algorithm will be increased, and the K-means algorithm has stronger robustness in a data set with fewer features (monitoring indicators), so that the effect of clustering the data of all monitoring indicators using the K-means algorithm is not good, and the final clustering cluster may not reflect the best data distribution, resulting in poor monitoring effect of abnormal data in the clustering cluster and inaccurate monitoring results of the power quality. Therefore, in the embodiment of the present invention, the data of various monitoring indicators of the power quality are analyzed to obtain the best monitoring indicators to achieve feature dimensionality reduction and improve the accuracy of monitoring the power quality.

[0028] Affected by the environment and power system equipment or lines, the data at the monitoring point may contain certain noise, resulting in different degrees of change in the data of the monitoring indicators at different monitoring points, which in turn affects the subsequent results of power quality monitoring. Therefore, it is necessary to judge the degree of noise interference at each monitoring point in order to reflect the authenticity of the parameter values ​​of each monitoring indicator at each monitoring point.

[0029] In the embodiment of the present invention, all parameter values ​​in all parameter value time series sequences corresponding to each monitoring point are first analyzed to obtain the change characteristic values ​​of all parameter values, which are used to determine the noise possibility of each parameter value. Because the fluctuation degree of noise data in the parameter value time series sequence is greater than the deviation of normal fluctuation, and the difference between noise data and previous and subsequent data is large, the change characteristic value of each data in each parameter value time series sequence can be obtained according to the data fluctuation in each parameter value time series sequence under each monitoring point.

[0030] Taking the qth parameter value in the mth parameter value time series sequence as an example, the specific process of obtaining the change characteristic value of the qth parameter value in the mth parameter value time series sequence is: Calculate the mean deviation of the qth parameter value in the mth parameter value time series to obtain the degree of deviation of any parameter value, wherein the mean deviation is a prior art and will not be described in detail here; If the qth parameter value is not the first parameter value or the last parameter value in the mth parameter value time series sequence, respectively obtaining the left adjacent parameter value and the right adjacent parameter value of the qth parameter value in the mth parameter value time series sequence, calculating a first difference between the qth parameter value and the left adjacent parameter value, and a second difference between the qth parameter value and the right adjacent parameter value, and taking the average of the absolute values ​​between the first difference and the second difference as the degree of change of the qth parameter value; If the qth parameter value is the first parameter value in the mth parameter value time series sequence, then obtain the right adjacent parameter value of the qth parameter value in the mth parameter value time series sequence, calculate the absolute value of the difference between the qth parameter value and the right adjacent parameter value, and obtain the degree of change of the qth parameter value; If the qth parameter value is the last parameter value in the mth parameter value time series, then obtain the left adjacent parameter value of the qth parameter value in the mth parameter value time series, calculate the absolute value of the difference between the qth parameter value and the left adjacent parameter value, and obtain the degree of change of the qth parameter value; The sum of the change degree and the deviation degree of the qth parameter value is calculated to obtain the change characteristic value of the qth parameter value in the mth parameter value time series sequence.

[0031] In one embodiment, the calculation formula for the change characteristic value of the qth parameter value in the mth parameter value time series is:

[0032] in, Represents the change characteristic value of the qth parameter value in the mth parameter value time series. represents the qth parameter value, represents the average value of all parameter values ​​in the mth parameter value time series, represents the left adjacent parameter value of the qth parameter value, represents the right adjacent parameter value of the qth parameter value, Represents the absolute value symbol.

[0033] It should be noted that the average value of all parameter values ​​in the mth parameter value time series is It is used to characterize the overall characteristics of the time series of the mth parameter value. Therefore, the average value As a reference benchmark. That is, the deviation degree of the qth parameter value. The greater the deviation degree, the greater the deviation between the qth parameter value and the average value. The larger the value is, the greater the possibility that the qth parameter value is noise; considering that if the qth parameter value is real abnormal data, the difference between the left adjacent parameter value and the right adjacent parameter value is large, and the left adjacent parameter value, the qth parameter value and the right adjacent parameter value are likely to show an increasing trend, if the qth parameter value is noise, the left adjacent parameter value and the right adjacent parameter value may be equal, and the qth parameter value is much larger than the left adjacent parameter value and the right adjacent parameter value. In order to highlight the fluctuation of the qth parameter value, that is, to more accurately judge the possibility that the qth parameter value is noise, in the embodiment of the present invention, the calculation formula for the degree of change of the qth parameter value is : If the qth parameter value is a true abnormal data, assume , , , when the calculation formula for the change degree of the qth parameter value is hour, ; When the calculation formula for the change degree of the qth parameter value is When , the qth parameter value is the real abnormal data, which is less likely to be noise and its degree of change should be small, so we use To characterize the degree of change of the qth parameter value, the greater the degree of change, the greater the difference between the qth parameter value and its two adjacent data before and after it, and the more the qth parameter value deviates from the average value, and thus The larger it is, the more likely it is that the qth parameter value is noise.

[0034] Similarly, the change characteristic values ​​of all parameter values ​​in all parameter value time series sequences corresponding to each monitoring point are obtained.

[0035] Because the multi-dimensional data of power quality are interrelated, for example, the time point of abnormal voltage fluctuation is often accompanied by the change of abnormal voltage deviation and three-phase voltage imbalance at the same time point, so there is a certain correlation between multiple parameter value time series. Therefore, in the embodiment of the present invention, taking the u-th monitoring point as an example, the peak data in each parameter value time series under the u-th monitoring point is obtained by the AMPD peak search algorithm. These peak data are likely to be noise data or abnormal data. The data stability of the u-th monitoring point is obtained by the similarity between the parameter values ​​of different monitoring indicators at the sampling time where the peak data is located, so as to distinguish whether these peak data are noise data or real abnormal value data, and more accurately judge the degree of noise interference of each monitoring point: if the similarity is large, it means that the peak data is more likely to be real abnormal data, and the u-th monitoring point is less affected by noise; otherwise, it means that the peak data is more likely to be noise data, and the u-th monitoring point is more affected by noise. Among them, the AMPD peak search algorithm is a prior art and will not be repeated here.

[0036] The specific process of obtaining the data stability of the u-th monitoring point is: Record the uth monitoring point as the target monitoring point, and for any parameter value time series sequence corresponding to the target monitoring point, record other parameter value time series sequences corresponding to the target monitoring point as reference sequences, obtain all peak data in any parameter value time series sequence, and obtain the weight of each peak data in any parameter value time series sequence according to the change characteristic value of each peak data and the change characteristic value of each data in the reference sequence; According to the weight of each of the peak data, weighted sum is performed on all the peak data in the any parameter value time series sequence to obtain the similarity between the any parameter value time series sequence and all reference sequences; The similarity between each parameter value time series sequence corresponding to the target monitoring point and other parameter value time series sequences is calculated respectively, all similarities are accumulated, and the accumulated results are inversely normalized to obtain the data stability of the target monitoring point.

[0037] Wherein, according to the change characteristic value of each of the peak data and the change characteristic value of the data in each of the reference sequences, the weight of each of the peak data in any parameter value time series sequence is obtained, including: For any peak data in any parameter value time series sequence, the sampling moment of any peak data is recorded as the target moment, the change characteristic value of the parameter value at the target moment is obtained in each reference sequence, recorded as the reference change characteristic value, and the average value of the absolute value of the difference between the change characteristic value of any peak data and each reference change characteristic value is calculated to obtain the weight of any peak data.

[0038] In one embodiment, the calculation formula for the data stability of the u-th monitoring point is:

[0039] in, Indicates the data stability of the u-th monitoring point, represents the number of all monitoring indicators under the u-th monitoring point, Represents the number of all peak data in the i-th parameter value time series at the u-th monitoring point, It represents the change characteristic value of the jth peak data in the i-th parameter value time series at the u-th monitoring point. It represents the change characteristic value of the parameter value at the target time (the sampling time of the jth peak data in the i-th parameter value time series sequence) in the z-th parameter value time series sequence (that is, the reference sequence of the i-th parameter value time series sequence) under the u-th monitoring point. Represents an exponential function with the natural constant e as the base, used for inverse normalization, Represents the absolute value symbol.

[0040] It should be noted that The larger the value, the greater the fluctuation of the jth peak data in the i-th parameter value time series at the u-th monitoring point. The smaller it is, the more unstable the data of the u-th monitoring point is, that is, the greater the degree of noise interference to the u-th monitoring point is; That is, the weight of the j-th peak data in the i-th parameter value time series at the u-th monitoring point, The larger the value is, the greater the difference between the multiple parameter value time series sequences at the u-th monitoring point at the sampling time (target time) to which the j-th peak data in the i-th parameter value time series sequence at the u-th monitoring point belongs, that is, the lower the similarity between the multiple parameter value time series sequences at the u-th monitoring point, the greater the possibility that the j-th peak data in the i-th parameter value time series sequence at the u-th monitoring point is noise, and thus The smaller it is, the more unstable the data of the u-th monitoring point is, that is, the greater the degree to which the u-th monitoring point is disturbed by noise.

[0041] Similarly, the data stability of all monitoring points in the power system is obtained.

[0042] The feature acquisition module 13 obtains, for any monitoring indicator, a target sequence according to the parameter value of any monitoring indicator at each monitoring point at the current sampling moment, and obtains the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation amplitude of the data points in the target sequence.

[0043] In an embodiment of the present invention, taking the sth monitoring indicator as an example, the parameter values ​​of the sth monitoring indicator at each monitoring point at the current sampling time are sorted in ascending order to obtain a target sequence, that is, each parameter value in the target sequence corresponds to a monitoring point, which is convenient for analyzing the parameter value of the sth monitoring indicator and evaluating the clustering effect of the sth monitoring indicator, so as to screen out the data of the monitoring indicators with good clustering effects, realize feature dimensionality reduction, and improve the accuracy of monitoring the power quality.

[0044] Because the K-means algorithm has certain assumptions about the shape and distribution of clusters, that is, the shape of each cluster is circular or spherical (that is, the uniform distribution of data points within the cluster), and the variance of each cluster is roughly the same, if the shape of each cluster is irregular, or the variance difference between each cluster is large, then the clustering effect of K-means is considered to be affected. However, in actual application, when the clustering effect of K-means is not affected, the shape of each cluster may be regular, but the variance difference between different clusters may be large, so the variance cannot be directly used to represent the uniformity of the distribution of data points within the cluster, that is, the clustering effect of the data of the sth monitoring indicator cannot be evaluated based on the variance of the target sequence.

[0045] Therefore, in an embodiment of the present invention, the distribution uniformity of the sth monitoring indicator is obtained by the variation amplitude of each two adjacent data points in the target sequence, that is, the variation amplitude of the parameter value of the sth monitoring indicator under each two adjacent monitoring points, and then the clustering effect of the data of the sth monitoring indicator is evaluated according to the distribution uniformity of the sth monitoring indicator, and at the same time, the degree to which the target sequence is interfered by noise can be more accurately reflected: when the parameter value of the sth monitoring indicator under each monitoring point is more similar to the parameter value of the sth monitoring indicator under its left and right adjacent monitoring points, and the data stability of the monitoring point itself is higher, it means that the frequency and amplitude of the change of the parameter value of the sth monitoring indicator under each monitoring point are smaller, the data in the target sequence corresponding to the sth monitoring indicator changes smoothly, the noise interference is smaller, the clustering effect is better, and the abnormal data is easier to highlight within the cluster cluster.

[0046] At the same time, because the monitoring points with low data stability are more affected by noise and may contain more noise, it is easier to identify noise as abnormal data when clustering using the K-means algorithm. If the data under the monitoring points with low data stability are clustered into one cluster as much as possible, the impact of noise on other clusters can be greatly reduced, and the number of noises mistaken for abnormal data can be reduced. Therefore, in the target sequence corresponding to the s-th monitoring indicator, if the data stability of the monitoring points corresponding to each two adjacent data in the target sequence is more similar, then when clustering using the K-means algorithm, the data under the monitoring points with low data stability are more likely to be clustered into one cluster, the better the clustering effect of the s-th monitoring indicator, and the easier it is to highlight abnormal data in the cluster.

[0047] Therefore, combining the change amplitude of each two adjacent data points in the target sequence and the similarity of the data stability of the monitoring points corresponding to each two adjacent data, the distribution uniformity of the sth monitoring indicator is obtained to evaluate the clustering effect of the target sequence corresponding to the sth monitoring indicator. Specifically: Record the other data except the first data and the last data in the target sequence as data to be analyzed; for any data to be analyzed, obtain the left adjacent target data and the right adjacent target data of the data to be analyzed in the target sequence; calculate the first absolute value of the difference between the any data to be analyzed and the left adjacent target data, and the second absolute value of the difference between the any data to be analyzed and the right adjacent target data; take the absolute value of the difference between the first absolute value of the difference and the second absolute value of the difference as the degree of difference between the any data to be analyzed and the data before and after it; calculate the product of the inverse of the degree of difference and the data stability degree of the monitoring point corresponding to the any data to be analyzed, and obtain the local stability degree of the any data to be analyzed; Linearly normalizing the accumulated value of the local stability degree of each of the data to be analyzed to obtain a first variable; For any data in the target sequence, if the any data is not the first data or the last data in the target sequence, respectively obtain the left adjacent data and the right adjacent data of the any data in the target sequence, calculate the third absolute difference value of the data stability of the monitoring point corresponding to the any data and the left adjacent data, and calculate the fourth absolute difference value of the data stability of the monitoring point corresponding to the any data and the right adjacent data, calculate the average value between the third absolute difference value and the fourth absolute difference value, and obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the first data in the target sequence, then the right adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the right adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the last data in the target sequence, the left adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the left adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; The accumulated values ​​of the stability similarity indexes of the monitoring points corresponding to all the data in the target sequence are normalized in inverse proportion to obtain a second variable.

[0048] The first variable and the second variable are accumulated to obtain the distribution uniformity of any monitoring indicator.

[0049] In one embodiment, the calculation formula for the distribution uniformity of the sth monitoring indicator is:

[0050] in, Indicates the distribution uniformity of the sth monitoring indicator, represents the number of all data in the target sequence corresponding to the sth monitoring indicator, that is, the number of all monitoring points. It represents the absolute value of the difference between the kth data (data to be analyzed) and the k-1th data in the target sequence. Indicates the absolute value of the difference between the kth data (data to be analyzed) and the k+1th data in the target sequence. Indicates the data stability of the monitoring point corresponding to the kth data (data to be analyzed) in the target sequence, Indicates the data stability of the monitoring point corresponding to the k-1th data in the target sequence. Indicates the data stability of the monitoring point corresponding to the k+1th data in the target sequence, represents the linear normalization function, Represents an exponential function with the natural constant e as the base, used for inverse normalization, Represents the absolute value symbol.

[0051] It should be noted that That is, the local stability of the kth data in the target sequence corresponding to the sth monitoring indicator. The larger the basis, The smaller it is, the more similar the k-th data in the target sequence corresponding to the s-th monitoring indicator is to its adjacent data at the current sampling time, and the less the monitoring point corresponding to the k-th data in the target sequence corresponding to the s-th monitoring indicator is affected by noise, the higher the credibility of the k-th data is, and thus The larger it is, the more evenly the data of the sth monitoring indicator is distributed at the current sampling time, and the better the clustering effect is. That is, it is the stability similarity index of the monitoring point corresponding to the kth data in the target sequence corresponding to the sth monitoring indicator. The larger the stability similarity index, the more similar the data stability of the monitoring point corresponding to each data is under the target sequence corresponding to the sth monitoring indicator, and the more the data in the target sequence presents a clustered distribution, and thus The larger it is, the more uniform the distribution of the data of the sth monitoring indicator at the current sampling time is, and the better the clustering effect is.

[0052] Similarly, the distribution uniformity of all monitoring indicators in the power system is obtained to screen out the data of several monitoring indicators with good clustering effect, realize dimensionality reduction analysis of power quality, and improve the accuracy of monitoring power quality.

[0053] The abnormal monitoring module 14 obtains the distribution uniformity of each of the monitoring indicators, selects the target monitoring indicators according to all the distribution uniformity, and respectively composes sample data with the parameter values ​​of the target monitoring indicators at each of the monitoring points at the current sampling time, clusters the sample data using the K-means algorithm to obtain at least one cluster, obtains the abnormality degree of each data point in each of the clusters, and monitors the power quality of the power system according to all the abnormality degrees.

[0054] The preset number is set to 3, and there is no limit here. The implementer can set it according to the specific scenario, sort the distribution uniformity of all monitoring indicators in descending order, obtain the distribution uniformity sequence, and select the first 3 monitoring indicators in the distribution uniformity sequence as the target monitoring indicators. Assuming that the target detection indicators are voltage deviation, frequency deviation and voltage flicker, the parameter values ​​of voltage deviation, frequency deviation and voltage flicker at one of the monitoring points at the current sampling time are combined into a feature data group, and then the feature data group corresponding to each monitoring point is used as sample data. The data in the sample data is less affected by noise and is evenly distributed. When the K-means algorithm is used to cluster the sample data, the obtained clusters can better highlight the abnormal data and reduce the interference caused by noise, so as to more accurately screen out the abnormal data of power quality.

[0055] Furthermore, the K-means algorithm is used to cluster the sample data to obtain at least one cluster, and each data point in each cluster is traversed to obtain the abnormality of each data point, and then the data points that may have abnormalities in each cluster are obtained. Taking the cth data point in the fth cluster as an example, the specific method for obtaining the abnormality of the cth data point in the fth cluster is: For the cth data point in the fth cluster, the Euclidean distance between the cth data point and each other data point in the fth cluster is calculated respectively, and the average value of all Euclidean distances is linearly normalized to obtain the abnormality degree of the cth data point, wherein the Euclidean distance is a prior art and will not be described in detail here.

[0056] In one embodiment, the calculation formula for the abnormality degree of the cth data point in the fth cluster is:

[0057] in, Indicates the abnormality of the cth data point in the fth cluster. represents the number of all data points in the fth cluster, represents the Euclidean distance between the cth data point and the xth data point in the fth cluster, represents the linear normalization function.

[0058] It should be noted that The larger the value, the greater the difference between the cth data point in the fth cluster and the xth data point, and the more outliers the cth data point in the fth cluster is in the fth cluster. The larger it is, the greater the possibility that the cth data point in the fth cluster is abnormal.

[0059] The preset abnormality threshold is set to 0.6. There is no limit here. The implementer can set it according to the specific scenario. If the abnormality of the cth data point in the fth cluster is , then the cth data point in the fth cluster is considered to have abnormal risk and may be abnormal electric energy data, so the cth data point in the fth cluster is taken as an abnormal data point.

[0060] Similarly, the abnormal data points in each cluster are obtained, and then according to the abnormal degree of all abnormal data points, the abnormal value of the power quality of the power system at the current sampling time is obtained, which is used to monitor the power quality of the power system at the current sampling time. The specific method of obtaining the abnormal value of the power quality of the power system at the current sampling time is: Acquire all abnormal data points in all clusters, accumulate the abnormality degrees of all abnormal data points, and obtain the power quality abnormality value of the power system at the current sampling moment. If the power quality abnormality value is greater than or equal to the preset power quality abnormality threshold, it is determined that the power quality of the power system is abnormal at the current sampling moment.

[0061] In one embodiment, the calculation formula of the power quality abnormal value of the power system at the current sampling time is:

[0062] Among them, Q represents the abnormal value of power quality of the power system at the current sampling time, represents the number of all clusters, G represents the cluster number, represents the number of all abnormal data points in the Gth cluster, Indicates the number of the abnormal data point, Indicates the Gth cluster in the The abnormal degree of an abnormal data point.

[0063] It should be noted that the more abnormal data points there are, the greater the degree of abnormality of each abnormal data point. The larger it is, the stronger the abnormality of the power quality of the power system at the current sampling moment.

[0064] The preset power quality abnormality threshold is set to 0.8. There is no limit here. The implementer can set it according to the specific scenario. , it means that there is an abnormality in the power quality of the power system at the current sampling moment. At this time, the four-in-one terminal marks the abnormal data in the database and issues an alarm. At the same time, all abnormal monitoring indicators and the locations of monitoring points, as well as the degree of abnormality, are sent to the relevant maintenance personnel of the power system.

[0065] In summary, the present invention includes a data acquisition module, which is used to respectively obtain the parameter values ​​of various monitoring indicators at each monitoring point in the power system, and obtain the parameter value time series sequence of various monitoring indicators at each of the monitoring points at the current sampling moment; a data analysis module, which is used to obtain the change characteristic value of each parameter value in each of the parameter value time series sequence according to the data fluctuations in each of the parameter value time series sequence, and obtain the data stability of each of the monitoring points according to the change characteristic values ​​of all parameter values ​​in all parameter value time series sequences corresponding to each of the monitoring points; a feature acquisition module, which is used to obtain, for any monitoring indicator, according to the parameter value of any monitoring indicator at each of the monitoring points at the current sampling moment. to the target sequence, and obtain the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation amplitude of the data points in the target sequence; an abnormal monitoring module is used to obtain the distribution uniformity of each monitoring indicator, screen out the target monitoring indicator according to all the distribution uniformity, and respectively form sample data with the parameter value of the target monitoring indicator at each monitoring point at the current sampling moment, cluster the sample data using the K-means algorithm to obtain at least one cluster, obtain the abnormality of each data point in each cluster, and monitor the power quality of the power system according to all the abnormality. Among them, the degree of interference of each monitoring point with noise is judged according to the data stability of each monitoring point. In the target sequence corresponding to each monitoring indicator, the distribution uniformity index of each monitoring indicator is obtained according to the stability of the data and the data stability of the monitoring point corresponding to each data, which is used to evaluate the clustering effect of each monitoring indicator, so as to screen out the target monitoring indicators with stable data changes, good clustering effects and less noise interference at the current sampling moment. Then, according to the data of the target monitoring indicators of each monitoring point at the current sampling moment, the K-means algorithm is used for clustering, and the power quality of the power system is monitored according to the clustering results to realize feature dimensionality reduction and improve the accuracy of power quality monitoring using the K-means algorithm in the four-in-one terminal.

[0066] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A four-in-one terminal with power quality monitoring function, characterized in that: The four-in-one terminal with power quality monitoring function includes: The data acquisition module is used to obtain the parameter values ​​of each monitoring indicator at each monitoring point in the power system, and obtain the time series of the parameter values ​​of each monitoring indicator at each monitoring point at the current sampling time; A data analysis module, used to obtain a change characteristic value of each parameter value in each parameter value time series according to the data fluctuation in each parameter value time series, and obtain the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in all parameter value time series corresponding to each monitoring point; A feature acquisition module is used to obtain a target sequence for any monitoring indicator according to the parameter value of any monitoring indicator at each monitoring point at the current sampling time, and obtain the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation amplitude of the data points in the target sequence; The abnormality monitoring module is used to obtain the distribution uniformity of each of the monitoring indicators, screen out the target monitoring indicators according to all the distribution uniformity, respectively form sample data with the parameter values ​​of the target monitoring indicators at each of the monitoring points at the current sampling time, cluster the sample data using the K-means algorithm to obtain at least one cluster, obtain the abnormality degree of each data point in each of the clusters, and monitor the power quality of the power system according to all the abnormality degrees.

2. The four-in-one terminal with power quality monitoring function according to claim 1 is characterized in that: The step of obtaining a change characteristic value of each parameter value in each parameter value time series sequence according to the data fluctuation in each parameter value time series sequence comprises: For any parameter value in any parameter value time series sequence, calculating the deviation of the any parameter value from the mean in the any parameter value time series sequence to obtain the degree of deviation of the any parameter value; If the any parameter value is not the first parameter value or the last parameter value in the any parameter value time series sequence, respectively obtaining a left adjacent parameter value and a right adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating a first difference between the any parameter value and the left adjacent parameter value, and a second difference between the any parameter value and the right adjacent parameter value, and taking an average of the absolute values ​​between the first difference and the second difference as a degree of change of the any parameter value; If the any parameter value is the first parameter value in the any parameter value time series sequence, obtaining the right adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating the absolute value of the difference between the any parameter value and the right adjacent parameter value, and obtaining the degree of change of the any parameter value; If the any parameter value is the last parameter value in the any parameter value time series sequence, obtaining the left adjacent parameter value of the any parameter value in the any parameter value time series sequence, calculating the absolute value of the difference between the any parameter value and the left adjacent parameter value, and obtaining the degree of change of the any parameter value; The sum of the variation degree and the deviation degree of any parameter value is calculated to obtain a variation characteristic value of any parameter value in the time series of any parameter value.

3. The four-in-one terminal with power quality monitoring function according to claim 1 is characterized in that: The step of obtaining the data stability of each monitoring point according to the change characteristic values ​​of all parameter values ​​in the time series of all parameter values ​​corresponding to each monitoring point includes: Record any one of the monitoring points as a target monitoring point, record other parameter value time series sequences corresponding to the target monitoring point as reference sequences for any parameter value time series sequence corresponding to the target monitoring point, obtain all peak data in any one of the parameter value time series sequences, and obtain the weight of each peak data in any one of the parameter value time series sequences according to the change characteristic value of each peak data and the change characteristic value of each data in the reference sequence; According to the weight of each of the peak data, weighted sum is performed on all the peak data in the any parameter value time series sequence to obtain the similarity between the any parameter value time series sequence and all reference sequences; The similarity between each parameter value time series sequence corresponding to the target monitoring point and other parameter value time series sequences is calculated respectively, all similarities are accumulated, and the accumulated results are inversely normalized to obtain the data stability of the target monitoring point.

4. The four-in-one terminal with power quality monitoring function according to claim 3 is characterized in that: The step of obtaining the weight of each of the peak data in any parameter value time series sequence according to the change characteristic value of each of the peak data and the change characteristic value of the data in each of the reference sequences comprises: For any peak data in any parameter value time series sequence, the sampling moment of any peak data is recorded as the target moment, the change characteristic value of the parameter value at the target moment is obtained in each reference sequence, recorded as the reference change characteristic value, and the average value of the absolute value of the difference between the change characteristic value of any peak data and each reference change characteristic value is calculated to obtain the weight of any peak data.

5. The four-in-one terminal with power quality monitoring function according to claim 1 is characterized in that: The obtaining of the distribution uniformity of any monitoring indicator according to the data stability of the monitoring point corresponding to each data in the target sequence and the variation range of the data points in the target sequence includes: Recording the other data in the target sequence except the first data and the last data as data to be analyzed, and obtaining a first variable in the target sequence according to the difference between each data to be analyzed and the data before and after it, and the data stability of the monitoring point corresponding to each data to be analyzed; In the target sequence, a second variable is obtained according to the data stability of the monitoring points corresponding to every two adjacent data; The first variable and the second variable are accumulated to obtain the distribution uniformity of any monitoring indicator.

6. The four-in-one terminal with power quality monitoring function according to claim 5 is characterized in that: The first variable is obtained according to the difference between each of the data to be analyzed and the data before and after it, and the data stability of the monitoring point corresponding to each of the data to be analyzed, including: For any data to be analyzed, in the target sequence, the left adjacent target data and the right adjacent target data of the data to be analyzed are obtained, a first absolute value of the difference between the data to be analyzed and the left adjacent target data, and a second absolute value of the difference between the data to be analyzed and the right adjacent target data are calculated, the absolute value of the difference between the first absolute value of the difference and the second absolute value of the difference is taken as the degree of difference between the data to be analyzed and its preceding and succeeding data, the product of the inverse of the degree of difference and the degree of data stability of the monitoring point corresponding to the data to be analyzed is calculated, and the degree of local stability of the data to be analyzed is obtained; The accumulated value of the local stability degree of each of the data to be analyzed is linearly normalized to obtain a first variable.

7. The four-in-one terminal with power quality monitoring function according to claim 5 is characterized in that: The second variable is obtained according to the data stability of the monitoring points corresponding to each two adjacent data in the target sequence, including: For any data in the target sequence, if the any data is not the first data or the last data in the target sequence, respectively obtain the left adjacent data and the right adjacent data of the any data in the target sequence, calculate the third absolute difference value of the data stability of the monitoring point corresponding to the any data and the left adjacent data, and calculate the fourth absolute difference value of the data stability of the monitoring point corresponding to the any data and the right adjacent data, calculate the average value between the third absolute difference value and the fourth absolute difference value, and obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the first data in the target sequence, then the right adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the right adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; If the any data is the last data in the target sequence, the left adjacent data of the any data is obtained in the target sequence, and the absolute value of the difference between the data stability of the monitoring point corresponding to the any data and the left adjacent data is calculated to obtain the stability similarity index of the monitoring point corresponding to the any data; The accumulated values ​​of the stability similarity indexes of the monitoring points corresponding to all the data in the target sequence are normalized in inverse proportion to obtain a second variable.

8. The four-in-one terminal with power quality monitoring function according to claim 1 is characterized in that: The target monitoring indicators are screened out according to the uniformity of all distributions, including: The distribution uniformity of all monitoring indicators is sorted in descending order to obtain a distribution uniformity sequence, and the monitoring indicators corresponding to the first preset number of data are selected from the distribution uniformity sequence and recorded as target monitoring indicators.

9. The four-in-one terminal with power quality monitoring function according to claim 1 is characterized in that: The obtaining of the abnormality degree of each data point in each of the clusters and monitoring the power quality of the power system according to all abnormality degrees includes: Record any cluster as a target cluster, and for any data point in the target cluster, respectively calculate the Euclidean distance between the data point and each other data point in the target cluster, perform linear normalization on the average of all Euclidean distances, and obtain the abnormality degree of the data point; If the abnormality level is greater than or equal to a preset abnormality level threshold, any data point is regarded as an abnormal data point; Acquire all abnormal data points in all clusters, accumulate the abnormality degrees of all abnormal data points, and obtain the power quality abnormality value of the power system at the current sampling moment. If the power quality abnormality value is greater than or equal to the preset power quality abnormality threshold, it is determined that the power quality of the power system is abnormal at the current sampling moment.

10. The four-in-one terminal with power quality monitoring function according to claim 1, characterized in that: The step of obtaining a target sequence according to the parameter value of any monitoring indicator at each monitoring point at the current sampling time includes: The parameter values ​​of any monitoring indicator at each monitoring point at the current sampling moment are sorted in ascending order to obtain a target sequence.

Images