Four-in-one terminal fault detection system based on sensor analysis

Abstract

The present invention relates to the field of electrical fault detection technology, and in particular to a four-in-one terminal fault detection system based on sensor analysis. The system includes: a data acquisition module for collecting voltage data and current data of a preset time period; converting voltage data into spectrum data; a scatter plot acquisition module for using a window to slide on the spectrum data and current data respectively, obtain the harmonic influence degree and coefficient of variation of the window, and then construct a scatter plot; a weight calculation module for obtaining the confidence and weight of the data points in the scatter plot; a reference straight line construction module for establishing a minimum error square sum objective function based on the weight of each data point, and calculating the first parameter and the second parameter, and then obtaining the reference straight line corresponding to the linear regression model; an anomaly detection module for constructing a boundary line parallel to the reference straight line, and then detecting harmonic anomalies. The present invention can accurately detect faults caused by harmonic influence.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical fault detection, and in particular to a four-in-one terminal fault detection system based on sensor analysis. Background Art

[0002] The four-in-one terminal includes an integrated miniature circuit breaker device that can measure, transmit, exchange, analyze and control data in the power grid. The device integrates multiple related electrical data acquisition modules and data transmission modules. The real-time collected electrical related data can be transmitted to the analysis and control terminal through the transmission module. The analysis and control terminal performs real-time detection and analysis of the operating status of the power grid. When a fault occurs, the circuit breaker can be controlled through the remote control system to provide timely protection for the power grid and its related equipment, thereby improving the operating stability and efficiency of the power grid and its access equipment, and reducing the failure rate and maintenance cost.

[0003] The access of nonlinear devices in the power grid will cause certain harmonic components with non-fundamental frequency content to appear in the voltage waveform, and these waveforms will cause overheating and increased loss of power grid equipment, causing equipment failure or increased probability of failure. The traditional method usually sets a fixed threshold for harmonic detection. When the harmonic components in the power grid are detected to exceed the set threshold, an early warning prompt is issued. However, this method may ignore the occurrence and change trend of harmonics and the normal dynamic change characteristics in the power grid, resulting in insufficient detection or an increased probability of false warnings. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a four-in-one terminal fault detection system based on sensor analysis, and the technical solutions adopted are as follows:

[0005] An embodiment of the present invention provides a four-in-one terminal fault detection system based on sensor analysis, the system comprising:

[0006] The data acquisition module is used to collect voltage data and current data of a preset period during the operation of the power grid; and convert the voltage data into spectrum data;

[0007] A scatter plot acquisition module is used to slide the window on the spectrum data and the current data respectively, calculate the harmonic influence degree of the window based on the non-fundamental frequency points with non-zero amplitude in the window, and obtain the coefficient of variation of the window after performing first-order difference on the current data in the window; a scatter plot is constructed according to the harmonic influence degree and coefficient of variation of each window;

[0008] A weight calculation module, used to calculate the confidence of a data point according to the harmonic influence degree and coefficient of variation of the data point in the scatter plot and the harmonic influence degree and coefficient of variation of other data points, and obtain the weight of the data point;

[0009] A reference straight line construction module is used to establish a minimum error square sum objective function when constructing a reference straight line based on the weight of each data point, and calculate a first parameter and a second parameter; a linear regression model is constructed according to the first parameter and the second parameter, and the straight line corresponding to the linear regression model is the reference straight line;

[0010] The anomaly detection module is used to construct boundary lines parallel to the reference straight line on both sides of the reference straight line according to the minimum distance from each data point to the reference straight line; and detect harmonic anomalies using the boundary lines and the reference straight line.

[0011] Preferably, the preset time period includes:

[0012] The preset time period is a time period before the current time point and a set time length from the current time point.

[0013] Preferably, using a window to slide on the spectrum data and the current data respectively includes:

[0014] When the window slides on the spectrum data and the current data respectively, the length of the window is a preset length, and the sliding step of the window is a preset step.

[0015] Preferably, calculating the harmonic influence degree of the window based on the non-fundamental frequency points whose amplitudes in the window are not zero includes:

[0016] When the window is sliding on the spectrum data, non-fundamental frequency points with non-zero amplitudes in the window are obtained, and the ratios of the amplitudes corresponding to each non-fundamental frequency point with non-zero amplitude in the window to the amplitudes corresponding to the fundamental frequency point in the window are calculated respectively, which are recorded as amplitude ratios; the product of the amplitude ratio corresponding to a non-fundamental frequency point with non-zero amplitude in a window and the inverse of the frequency of the non-fundamental frequency point with non-zero amplitude is obtained, and the products are summed to obtain the degree of harmonic influence of the window.

[0017] Preferably, a scatter plot is constructed according to the harmonic influence degree and the coefficient of variation of each window, including:

[0018] A two-dimensional plane coordinate system is established, in which the vertical axis is the harmonic influence degree of each window, and the horizontal axis is the variation coefficient corresponding to each window; the harmonic influence degree and variation coefficient corresponding to each window in the preset time period are mapped to the two-dimensional plane coordinate system according to the corresponding position to obtain the corresponding scatter plot.

[0019] Preferably, the confidence of a data point is calculated according to the harmonic influence degree and coefficient of variation of the data point and the harmonic influence degree and coefficient of variation of other data points in the scatter plot, and the weight of the data point is obtained, including:

[0020] Calculate the absolute value of the difference between the harmonic influence degree of a data point in the scatter plot and the harmonic influence degree of any other data point, and record it as the influence degree difference; calculate the absolute value of the difference between the coefficient of variation of a data point in the scatter plot and the coefficient of variation of any other data point and add it to the constant term to obtain the coefficient of variation difference; the ratio of the influence degree difference and the coefficient of variation difference corresponding to a data point in the scatter plot and one of the other data points is the data point difference; average and normalize the differences of each data point corresponding to a data point in the scatter plot to obtain the difference sum of the data point; use the preset value to negatively correlate the difference sum of the data point to obtain the confidence of the data point;

[0021] The weight of a data point is the ratio of the confidence of the data point to the sum of the confidences of all data points.

[0022] Preferably, the method for establishing a minimum square error objective function when constructing a reference straight line based on the weight of each data point and calculating the first parameter and the second parameter includes:

[0023] The minimum error square objective function is constructed based on the harmonic influence degree, coefficient of variation and weight of each data point. The calculation formula of the minimum error square objective function is as follows:

[0024] ,

[0025] in, represents the minimum error square sum objective function; Represents the weight corresponding to the jth data point in the scatter plot; Indicates the degree of harmonic influence of the jth data point in the scatter plot; represents the coefficient of variation of the jth data point in the scatter plot; a and b represent the first parameter and the second parameter respectively; n represents the number of data points in the scatter plot;

[0026] The first parameter and the second parameter are obtained by solving the minimum value of the minimum square error objective function.

[0027] Preferably, boundary lines parallel to the reference straight line are constructed on both sides of the reference straight line according to the minimum distance from each data point to the reference straight line, including:

[0028] Taking the reference straight line as the central axis, the minimum distance from each data point on one side of the reference straight line to the reference straight line is obtained, and the minimum distances corresponding to the data points on this side form a set, which is recorded as the distance set; in the distance set, the absolute value of the difference between a minimum distance and each other minimum distance is calculated, and then the average value is obtained to obtain the average difference of the minimum distance, and the minimum distance with the smallest average difference is obtained as the normal distance of the distance set; on the side corresponding to the distance set, a boundary line with a normal distance from the reference straight line and parallel to the reference straight line is made, which is recorded as the first boundary line; similarly, the boundary line on the other side of the reference straight line is obtained, which is recorded as the second boundary line.

[0029] Preferably, the harmonic anomaly is detected using the reference straight line and the boundary line, including:

[0030] Obtain the data point in the scatter plot corresponding to the window where the voltage data at the current time point is located, and record it as the current detection point; calculate the minimum distance between the current detection point and the reference straight line, and record it as the current detection minimum distance; if the current detection point is between the first boundary line and the second boundary line, obtain a preset number of data points before the current time point in time series, and record them as auxiliary detection data points; arrange the minimum distance between each auxiliary detection data point and the reference straight line and the current detection minimum distance in time series to form a distance sequence, and calculate the slope of the distance sequence. If the slope is greater than zero, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the first threshold, then the harmonic is abnormal;

[0031] If the current detection point is outside the first boundary line and the second boundary line, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the second threshold, the harmonic is abnormal.

[0032] The embodiment of the present invention has at least the following beneficial effects: when analyzing and monitoring the harmonics of the voltage data in the power grid, in order to obtain the accuracy of its monitoring, the present application first sets a sliding window to obtain the value of the degree of harmonic influence on the voltage data in each window, that is, the degree of harmonic influence. Since the influence of voltage harmonics is mainly caused by nonlinear loads in the power grid, etc., the influence of voltage by harmonics has a certain relationship with the nonlinear change of current. Therefore, the change characteristics of the current data in each window, that is, the coefficient of variation, are obtained, and then a coordinate system is established to select the characteristics corresponding to each window in a preset time period, and a scatter plot is obtained through the constructed coordinate system, and the voltage data and the current data are corresponding in each window. The characteristics are combined for comprehensive analysis to improve the accuracy of detection; further analysis is performed through the scatter plot to obtain the weight of each data point in the scatter plot when constructing the linear regression model, and then the relevant objective function when constructing the reference line is obtained, and then a more accurate reference line corresponding to the linear regression model is obtained, and then the boundary lines are constructed on both sides of the reference line through the data points distributed around the reference line to obtain the distribution range of normal data points, and finally the boundary lines and the reference lines are used to detect harmonic anomalies, so as to accurately warn of the voltage harmonics and related anomalies in the current power grid, so as to improve the accuracy of the operation of related equipment such as circuit breakers, which is beneficial to the stable operation of the power grid system equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] 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.

[0034] Figure 1 A system block diagram of a four-in-one terminal fault detection system based on sensor analysis provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] 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, structure, features and effects of a four-in-one terminal fault detection system based on sensor analysis 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.

[0036] 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.

[0037] The following is a detailed description of a specific scheme of a four-in-one terminal fault detection system based on sensor analysis provided by the present invention in conjunction with the accompanying drawings.

[0038] Example:

[0039] The main application scenarios of the present invention are:

[0040] Harmonics in the power grid may cause certain equipment failures or increase the probability of failure, so they need to be detected. Traditional threshold-based detection methods may ignore the normal dynamic changes of harmonics in the power grid, resulting in insufficient detection or false warnings.

[0041] See also Figure 1 , which shows a system block diagram of a four-in-one terminal fault detection system based on sensor analysis provided by an embodiment of the present invention, the system includes the following modules:

[0042] The data acquisition module is used to collect voltage data and current data of a preset period during the operation of the power grid; and convert the voltage data into spectrum data.

[0043] When the voltage harmonics in the power grid are detected by the four-in-one integrated terminal - the circuit breaker, it may ignore the normal dynamic changes and trends of the harmonics in the power grid, resulting in a high probability of false operation of the traditional circuit breaker.

[0044] Therefore, it is necessary to use the four-in-one terminal in this example, that is, the voltage data acquisition module in the new intelligent circuit breaker equipment, to collect the time-series voltage data and current data in the operating power grid, and then send the collected voltage time-series data to the remote control terminal system through its built-in communication module. The remote terminal control system analyzes the voltage data collected by the circuit breaker in real time, and then obtains the real-time analysis results, and sends relevant instructions to the circuit breaker to control the opening and closing actions of the circuit breaker, etc.

[0045] It should be noted that, for the detection of the impact of harmonics on voltage data at the current time point, it is necessary to analyze the voltage data and current data in historical time. Therefore, the voltage data and current data of a preset time period during the operation of the power grid are collected, wherein the preset time period is a time period before the current time point with a set time length from the current time point. Preferably, the set time length in the present invention is 2 hours, which can be set by the implementer according to actual conditions; the frequency of data collection can also be set by the implementer according to actual conditions. Since harmonic analysis is required, it is necessary to use Fourier transform to decompose the voltage data and convert it into spectrum data, thereby completing the data preparation work and then proceeding to the next step of data analysis.

[0046] The scatter plot acquisition module is used to use a window to slide on the spectrum data and the current data respectively, calculate the harmonic influence degree of the window based on the non-fundamental frequency points in the window whose amplitude is not zero, and obtain the coefficient of variation of the window after performing a first-order difference on the current data in the window; and construct a scatter plot according to the harmonic influence degree and coefficient of variation of each window.

[0047] Due to nonlinear loads, power electronic equipment, and internal and external interference in the power grid, harmonics are generated in the power grid, causing distortion of the voltage waveform. Harmonics increase power grid losses, heat transmission lines and transformers, reduce equipment efficiency, and even cause power grid resonance, resulting in abnormal increases in voltage and current, which in turn cause safety accidents such as equipment burning and protection device malfunction. Therefore, it is necessary to detect harmonics in the power grid more accurately to prevent electrical equipment from malfunctioning.

[0048] Since the generation and change of harmonics in actual power grids are often dynamic, they may be affected by multiple factors such as load fluctuations, changes in equipment operating status, and system disturbances. The fixed threshold detection method cannot capture the continuous change trend of harmonics, and only triggers an early warning when the harmonic exceeds the threshold, so it is easy to ignore some important change trends. For example, although the harmonics do not exceed the threshold, their gradually rising trend may indicate potential problems; if this trend is not identified, it may cause delays in harmonic governance and maintenance work. At the same time, the dynamic characteristics of the power grid make harmonics fluctuate greatly in a short period of time. For example, when the load is started or switched, the harmonics may exceed the threshold momentarily, but it does not necessarily mean abnormality or harm. Fixed threshold detection may mistakenly identify these short-term normal fluctuations as problems, triggering unnecessary early warnings and increasing the complexity of system management and maintenance. Moreover, the changing rules and patterns of harmonics in different power grids are not the same.

[0049] Therefore, in order to increase the accuracy of harmonic detection, it is necessary to first set a window, which is a time window, in which the length of the window is a preset length, which is 10 seconds, and the sliding step of the window is a preset step, which is 5 seconds. These are all set based on experience. In actual use, the implementer can make adjustments according to actual conditions, and then obtain the harmonic component information in the window when the window slides.

[0050] Since the equipment loads in the power grid may be different in different time periods, the harmonic content in the voltage may also be different. Therefore, in order to accurately analyze the degree of harmonic influence on the voltage waveform in any window, it is necessary to calculate the harmonic influence of the window based on the non-fundamental frequency points with non-zero amplitude in the window.

[0051] Specifically, when the window is sliding on the spectrum data, the non-fundamental frequency points in the window with non-zero amplitude are obtained, and the ratio of the amplitude corresponding to each non-fundamental frequency point in the window with non-zero amplitude to the amplitude corresponding to the fundamental frequency point in the window is calculated, and recorded as the amplitude ratio; the product of the amplitude ratio corresponding to a non-fundamental frequency point with non-zero amplitude in a window and the reciprocal of the frequency of the non-fundamental frequency point with non-zero amplitude is obtained, and the product is summed to obtain the harmonic influence degree of the window. The calculation formula for the harmonic influence degree is:

[0052] ,

[0053] Among them, Q represents the degree to which the voltage data in a window corresponding to a window is affected by harmonics, that is, the degree of harmonic influence; m represents the number of frequency points corresponding to the non-fundamental frequency points in the spectrum data corresponding to the voltage data in the window when the amplitude is not 0, that is, the number of non-fundamental frequency points in the window with non-zero amplitude; Indicates the amplitude of the i-th non-fundamental frequency point among the non-fundamental frequency points whose amplitude is not 0 in the window; Indicates the amplitude corresponding to the fundamental frequency point in the window; Indicates the frequency corresponding to the i-th non-fundamental frequency point among the non-fundamental frequency points whose amplitude is not 0 in the window.

[0054] It represents the ratio between the amplitude corresponding to the i-th non-fundamental frequency point in the analyzed window and the amplitude corresponding to the fundamental frequency point in the analyzed window, that is, the amplitude ratio. The larger the value, the more the non-fundamental frequency point contains compared to the fundamental component in the current window, which means that it has a greater impact on the waveform of the current voltage data. It indicates the reciprocal of the frequency corresponding to the i-th non-fundamental frequency point whose corresponding amplitude is not 0 in the analyzed window. The larger the value, the lower the frequency of the current non-fundamental frequency point. The low-frequency harmonics or interharmonics usually have a global distortion on the voltage waveform, which often causes the overall distortion of the waveform. The propagation distance and influence are more significant. Therefore, the larger the value, the greater the weight of the non-fundamental component of the frequency point on the voltage waveform. Therefore, It represents the influence of the i-th non-fundamental frequency point with non-zero amplitude in the analyzed window on the voltage waveform in the time period. The sum of them can represent the influence of harmonics on the voltage data in the analyzed time window as a whole.

[0055] In actual use, the frequent switching of equipment in the power grid and the changes in major nonlinear loads will increase the generation and propagation of harmonics in the power grid, causing the voltage waveform to be more significantly affected by harmonics, manifested as waveform distortion, high-frequency oscillation and harmonic amplification effects.

[0056] Therefore, there is a certain correlation between the change of the voltage waveform in the power grid and the nonlinear change of the current in the power grid.

[0057] If the current data in a certain window is relatively stable or in a linear change, and the voltage data waveform is more affected by harmonics, it may indicate that the probability of a fault in the power grid is greater at this time. For example: in a wind farm or photovoltaic power station, the inverter converts DC power into AC power and connects it to the grid. Such systems usually have a relatively stable current output, but when the inverter fails or the filter circuit fails, the voltage waveform will be significantly distorted. If the current fluctuation is not large at this time, but the voltage harmonic content increases, it may be necessary to check the health of the inverter or the grid connection point.

[0058] Therefore, further, when the window slides on the current data, it is necessary to obtain the corresponding effective current data in each window and perform a first-order difference on it, so as to effectively remove its linear change trend, and then obtain the coefficient of variation of the corresponding current data in the window, which is used as the change characteristic value of the corresponding current in the window.

[0059] Finally, the harmonic influence degree of the corresponding voltage data in each sliding window is used as the vertical axis of the two-dimensional plane coordinate system, and the variation coefficient of the effective current after processing in the corresponding window is used as the horizontal axis of the two-dimensional plane coordinate system. Then, the harmonic influence degree and variation coefficient corresponding to each window in the preset time period are mapped to the two-dimensional plane coordinate system according to the corresponding position to obtain the corresponding scatter plot. Each data point in the scatter plot corresponds to each window, and the harmonic influence degree and variation coefficient of each data point correspond to the harmonic influence degree and variation coefficient of the corresponding window.

[0060] The weight calculation module is used to calculate the confidence of a data point according to the harmonic influence degree and variation coefficient of the data point in the scatter plot and the harmonic influence degree and variation coefficient of other data points, and obtain the weight of the data point.

[0061] Through the above, a scatter plot is obtained. Since the harmonic influence of voltage has a certain correlation with the change of current, a linear correlation model is further established by combining the change relationship between the harmonic influence of the current power grid and the current. This will facilitate more accurate detection and prediction of harmonic anomalies of voltage in the power grid in the future.

[0062] When establishing the relevant model, in order to avoid the possibility that the voltage harmonics corresponding to some time periods have certain randomness, which may lead to certain large differences from the overall harmonic change law in the power grid. Therefore, in order to understand the law and pattern of the overall normal change of harmonics in the current power grid. Further combine the distribution of each point in the scatter plot to obtain its confidence in the distribution close to the overall change.

[0063] The confidence of a data point is calculated based on the harmonic influence degree and coefficient of variation of the data point in the scatter plot and the harmonic influence degrees and coefficients of variation of other data points. Specifically, the absolute value of the difference between the harmonic influence degree of a data point in the scatter plot and the harmonic influence degree of any one of the other data points is calculated, and recorded as the influence degree difference; the absolute value of the difference between the coefficient of variation of a data point in the scatter plot and the coefficient of variation of any one of the other data points is calculated and added to the constant term to obtain the coefficient of variation difference; the ratio of the influence degree difference and the coefficient of variation difference corresponding to a data point in the scatter plot and one of the other data points is the data point difference; the differences of each data point corresponding to a data point in the scatter plot are averaged and normalized to obtain the difference sum of the data point; the difference sum of the data point is negatively correlated with the preset value to obtain the confidence of the data point.

[0064] The specific calculation formula of confidence is:

[0065] ,

[0066] in, Represents the confidence of the j-th data point in the scatter plot; norm() represents the normalization operation; n represents the number of data points in the scatter plot, n-1 represents the number of data points other than the j-th data point; 1 represents the preset value; Indicates the degree of harmonic influence of the jth data point in the scatter plot, Represents the rth data point in the scatter plot except the jth data point; represents the coefficient of variation of the jth data point in the scatter plot, It represents the coefficient of variation of the rth data point in the scatter plot except the jth data point; u represents a constant term, which is taken as 0.01 in the present invention to prevent the denominator from being 0.

[0067] The numerator in represents the difference in the degree of influence between the jth data point in the scatter plot and the rth data point among the data points other than the jth data point, and the denominator represents the difference in the coefficient of variation between the jth data point in the scatter plot and the rth data point among the data points other than the jth data point. The overall representation of the data point difference between the jth data point in the scatter plot and the rth data point among the data points other than the jth data point. Because if the coefficient of variation corresponding to these two points is closer, then the change degree of the current information corresponding to the local window of these two points is closer, and the degree of harmonic influence on the power grid should be close in theory. Therefore, the size of this value represents the difference between the jth data point and the rth data point among the data points other than the jth data point, and then the average is calculated to obtain the difference and sum of the jth data point. The smaller the value, the closer the distribution characteristic difference between this point and the other points is, that is, The larger the value

[0068] Through the above, we get the confidence level of each point in the distribution of harmonic change rules. The higher the confidence level, the more it can show the overall rules and patterns of voltage harmonic changes and current changes in the current power grid. Therefore, the greater its weight is when participating in the establishment of the relevant linear model.

[0069] Therefore, based on the confidence of each data point in the above scatter plot, the weight of each data point is calculated, where the weight of a data point is the ratio of the confidence of the data point to the sum of the confidences of all data points. The specific calculation formula is:

[0070] ,

[0071] in, Represents the weight corresponding to the jth data point in the scatter plot; represents the confidence of the jth data point in the scatter plot; n represents the number of data points in the scatter plot. From this, the weight of each data point can be obtained.

[0072] The reference straight line construction module is used to establish the minimum error square sum objective function when constructing the reference straight line based on the weight of each data point, and calculate the first parameter and the second parameter; a linear regression model is constructed according to the first parameter and the second parameter, and the straight line corresponding to the linear regression model is the reference straight line.

[0073] Furthermore, it is necessary to use the least squares method to construct a linear regression model using the data points in the scatter plot. First, the minimum error square objective function for constructing the reference line is established based on the weight of each data point. Specifically, the minimum error square objective function is constructed based on the harmonic influence degree, coefficient of variation and weight of each data point. The calculation formula of the minimum error square objective function is as follows:

[0074] ,

[0075] in, represents the minimum error square sum objective function; Represents the weight corresponding to the jth data point in the scatter plot; Indicates the degree of harmonic influence of the jth data point in the scatter plot; represents the coefficient of variation of the jth data point in the scatter plot; a and b represent the first parameter and the second parameter respectively; n represents the number of data points in the scatter plot.

[0076] The first parameter and the second parameter are obtained by solving the minimum value of the minimum square error objective function.

[0077] Then, a linear regression model is constructed according to the first parameter a and the second parameter b. The specific expression of the linear regression model is: Q=a×Y+b; the straight line corresponding to the linear regression model is the reference straight line. The first parameter a and the second parameter b as the parameters of the linear regression model are obtained based on the least squares method through the above method, and then the linear regression model is obtained, and then the reference straight line corresponding to the linear regression model is obtained.

[0078] The anomaly detection module is used to construct boundary lines parallel to the reference straight line on both sides of the reference straight line according to the minimum distance from each data point to the reference straight line; and detect harmonic anomalies using the reference straight line and the boundary lines.

[0079] By obtaining the reference straight line of the linear regression model of the scatter plot distribution, the influence of harmonics on the voltage under normal conditions will be distributed near the constructed reference straight line, that is, within a certain range close to the reference straight line. However, when the power grid is abnormal, the abnormal harmonic point may exceed the range.

[0080] Therefore, further, taking the reference straight line as the central axis, the minimum distance from each data point on one side of the reference straight line to the reference straight line is obtained, and the minimum distances corresponding to the data points on this side form a set, which is recorded as the distance set; in the distance set, the absolute value of the difference between a minimum distance and each other minimum distance is calculated, and then the average value is obtained to obtain the average difference of the minimum distance, and the minimum distance with the smallest average difference is obtained as the normal distance of the distance set; on the side corresponding to the distance set, a boundary line with a normal distance from the reference straight line and parallel to the reference straight line is made, which is recorded as the first boundary line; similarly, the boundary line on the other side of the reference straight line is obtained, which is recorded as the second boundary line.

[0081] Get the data point in the scatter plot corresponding to the window where the voltage data at the current time point is located, and record it as the current detection point; calculate the minimum distance between the current detection point and the reference straight line, and record it as the current detection minimum distance; if the current detection point is between the first boundary line and the second boundary line, obtain a preset number of data points before the current time point in time series, and record them as auxiliary detection data points; arrange the minimum distance between each auxiliary detection data point and the reference straight line and the current detection minimum distance in time series to form a distance sequence, and calculate the slope of the distance sequence. If the slope is greater than zero, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the first threshold, it means that the harmonic is abnormal. It means that the degree of influence of harmonics in the power grid shows a serious upward trend, and then an early warning is issued, otherwise it is normal.

[0082] If the current detection point is outside the first boundary line and the second boundary line, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the second threshold, it means that the harmonic is abnormal.

[0083] If the current detection point is between the first boundary line and the second boundary line, and the slope of the distance sequence is greater than zero, it means that the degree of harmonic influence is on the rise, wherein the values ​​of the first threshold and the second threshold are 0.4 and 0.7 respectively, and the implementer can adjust them according to the actual situation.

[0084] Through the above, the detection result of the harmonics in the current power grid is obtained. When it meets the above judgment conditions, the circuit breaker is controlled by the remote control module to break the circuit or take measures such as harmonic suppression to avoid damage to electrical equipment by harmonics in the power grid.

[0085] 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 above is a description of a specific embodiment of this specification. In addition, 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.

[0086] 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.

[0087] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A four-in-one terminal fault detection system based on sensor analysis, characterized in that: The system includes: The data acquisition module is used to collect voltage data and current data of a preset period during the operation of the power grid; and convert the voltage data into spectrum data; A scatter plot acquisition module is used to slide the window on the spectrum data and the current data respectively, calculate the harmonic influence degree of the window based on the non-fundamental frequency points with non-zero amplitude in the window, and obtain the coefficient of variation of the window after performing first-order difference on the current data in the window; a scatter plot is constructed according to the harmonic influence degree and coefficient of variation of each window; A weight calculation module, used to calculate the confidence of a data point according to the harmonic influence degree and coefficient of variation of the data point in the scatter plot and the harmonic influence degree and coefficient of variation of other data points, and obtain the weight of the data point; A reference straight line construction module is used to establish a minimum error square sum objective function when constructing a reference straight line based on the weight of each data point, and calculate a first parameter and a second parameter; a linear regression model is constructed according to the first parameter and the second parameter, and the straight line corresponding to the linear regression model is the reference straight line; The anomaly detection module is used to construct boundary lines parallel to the reference straight line on both sides of the reference straight line according to the minimum distance from each data point to the reference straight line; and detect harmonic anomalies using the boundary lines and the reference straight line.

2. According to the sensor analysis-based four-in-one terminal fault detection system of claim 1, it is characterized in that: The preset time period includes: The preset time period is a time period before the current time point and a set time length from the current time point.

3. According to the sensor analysis-based four-in-one terminal fault detection system of claim 1, it is characterized in that: The using window to slide on the spectrum data and the current data respectively includes: When the window slides on the spectrum data and the current data respectively, the length of the window is a preset length, and the sliding step of the window is a preset step.

4. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The calculating the harmonic influence degree of the window based on the non-fundamental frequency points whose amplitudes in the window are not zero includes: When the window is sliding on the spectrum data, non-fundamental frequency points with non-zero amplitudes in the window are obtained, and the ratios of the amplitudes corresponding to each non-fundamental frequency point with non-zero amplitude in the window to the amplitudes corresponding to the fundamental frequency point in the window are calculated respectively, which are recorded as amplitude ratios; the product of the amplitude ratio corresponding to a non-fundamental frequency point with non-zero amplitude in a window and the inverse of the frequency of the non-fundamental frequency point with non-zero amplitude is obtained, and the products are summed to obtain the degree of harmonic influence of the window.

5. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The step of constructing a scatter plot according to the harmonic influence degree and the coefficient of variation of each window includes: A two-dimensional plane coordinate system is established, in which the vertical axis is the harmonic influence degree of each window, and the horizontal axis is the variation coefficient corresponding to each window; the harmonic influence degree and variation coefficient corresponding to each window in the preset time period are mapped to the two-dimensional plane coordinate system according to the corresponding position to obtain the corresponding scatter plot.

6. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The step of calculating the confidence of a data point according to the harmonic influence degree and coefficient of variation of the data point and the harmonic influence degrees and coefficients of variation of other data points in the scatter plot, and obtaining the weight of the data point includes: Calculate the absolute value of the difference between the harmonic influence degree of a data point in the scatter plot and the harmonic influence degree of any other data point, and record it as the influence degree difference; calculate the absolute value of the difference between the coefficient of variation of a data point in the scatter plot and the coefficient of variation of any other data point and add it to the constant term to obtain the coefficient of variation difference; the ratio of the influence degree difference and the coefficient of variation difference corresponding to a data point in the scatter plot and one of the other data points is the data point difference; average and normalize the differences of each data point corresponding to a data point in the scatter plot to obtain the difference sum of the data point; use the preset value to negatively correlate the difference sum of the data point to obtain the confidence of the data point; The weight of a data point is the ratio of the confidence of the data point to the sum of the confidences of all data points.

7. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The method for establishing a minimum error square sum objective function when constructing a reference straight line based on the weight of each data point and calculating the first parameter and the second parameter includes: The minimum error square objective function is constructed based on the harmonic influence degree, coefficient of variation and weight of each data point. The calculation formula of the minimum error square objective function is as follows: ， in, represents the minimum error square sum objective function; Represents the weight corresponding to the jth data point in the scatter plot; Indicates the degree of harmonic influence of the jth data point in the scatter plot; represents the coefficient of variation of the jth data point in the scatter plot; a and b represent the first parameter and the second parameter respectively; n represents the number of data points in the scatter plot; The first parameter and the second parameter are obtained by solving the minimum value of the minimum square error objective function.

8. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The step of constructing boundary lines parallel to the reference straight line on both sides of the reference straight line according to the minimum distance between each data point and the reference straight line includes: Taking the reference straight line as the central axis, the minimum distance from each data point on one side of the reference straight line to the reference straight line is obtained, and the minimum distances corresponding to the data points on this side form a set, which is recorded as the distance set; in the distance set, the absolute value of the difference between a minimum distance and each other minimum distance is calculated, and then the average value is obtained to obtain the average difference of the minimum distance, and the minimum distance with the smallest average difference is obtained as the normal distance of the distance set; on the side corresponding to the distance set, a boundary line with a normal distance from the reference straight line and parallel to the reference straight line is made, which is recorded as the first boundary line; similarly, the boundary line on the other side of the reference straight line is obtained, which is recorded as the second boundary line.

9. The four-in-one terminal fault detection system based on sensor analysis according to claim 1 is characterized in that: The method of detecting harmonic anomalies by using the reference straight line and the boundary line includes: Obtain the data point in the scatter plot corresponding to the window where the voltage data at the current time point is located, and record it as the current detection point; calculate the minimum distance between the current detection point and the reference straight line, and record it as the current detection minimum distance; if the current detection point is between the first boundary line and the second boundary line, obtain a preset number of data points before the current time point in time series, and record them as auxiliary detection data points; arrange the minimum distance between each auxiliary detection data point and the reference straight line and the current detection minimum distance in time series to form a distance sequence, and calculate the slope of the distance sequence. If the slope is greater than zero, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the first threshold, then the harmonic is abnormal; If the current detection point is outside the first boundary line and the second boundary line, and the ratio of the current detection minimum distance to the normal distance corresponding to the boundary line on the side where the current detection point is located is greater than the second threshold, the harmonic is abnormal.

Images