An Internet of Things box-type substation and a fault detection method

Abstract

This application relates to the technical field of substations, and particularly to an Internet of Things box-type substation and a fault monitoring method. The method includes: collecting different sequences using Internet of Things technology; performing periodic detection on the signal sequences to obtain periodicity, and combining the overall assignment and fluctuation of the sequences to obtain the internal characteristics of the signals; then obtaining the external environment characteristics based on the temperature sequence difference inside and outside the cabin, and combining the two to obtain the time-domain fault characteristics; converting the acoustic wave signal into a spectrogram, and obtaining the frequency-domain fault characteristics based on the sequence monotonicity, amplitude, data change rate, and number of peaks in the spectrogram; constructing a fault code according to the time-domain fault characteristics and the frequency-domain fault characteristics, and judging the partial discharge fault. This application makes full use of and excavates the characteristics of the monitoring data, thereby realizing more accurate fault monitoring for the box-type substation and reducing the false alarm rate.

Description

Technical Field

[0001] The present application relates to the technical field of substations, and particularly relates to an Internet of Things box-type substation and a fault detection method. Background Art

[0002] With the development of smart grids, the demand for power grid transformation has gradually increased. In this process, as a complete set of power transformation and distribution equipment, box-type substations integrate transformers, high- and low-voltage switchgear, power compensation devices, etc., and have become the mainstream equipment for terminal power transformation and distribution, and their proportion in the power supply and distribution system is also increasing. Therefore, the safety of box-type substations has become the key to ensuring the power supply and stability of the power supply and distribution system. With the application of Internet of Things technology, box-type substations are also developing in the direction of intelligence, and the online monitoring of the state of box-type substations has become the research focus.

[0003] The prior art monitors the electrical variables and non-electrical variable data during the operation of box-type substations by installing various sensors in the box-type substations, and judges whether there are faults in the box-type substations according to the abnormal changes of these data. However, the utilization of the monitoring data in the prior art is too simple. Usually, by setting thresholds, when one or more data exceed the thresholds, an alarm is given in time, and on-site investigation is required. This method does not make full use of the monitoring data and does not deeply explore the change characteristics of the data. At the same time, due to the complexity of the operation of box-type substations, the setting of a single threshold increases the probability of false alarms during the monitoring of box-type substations, which undoubtedly increases the difficulty of fault investigation, and at the same time causes a large number of ineffective investigations, increasing the difficulty of work. Therefore, there is an urgent need for a more accurate fault monitoring method to reduce the false alarm rate. Summary of the Invention

[0004] In order to solve the technical problem that false alarms in abnormal monitoring increase the difficulty of fault investigation, the present application provides an Internet of Things box-type substation and a fault detection method, and the technical solutions adopted are as follows:

[0005] In a first aspect, the present application proposes a fault detection method for an Internet of Things box-type substation, and the method includes the following steps:

[0006] Use Internet of Things technology to collect a local signal sequence, an external temperature sequence, temperature sequences and acoustic signal sequences of different compartments each time;

[0007] Process the local signal sequence based on a window of a preset size to obtain a moving average sequence; determine the periodicity based on the distances between the peaks of the moving average sequence; mark the peaks of the local signal sequence to obtain abnormal peaks and their neighborhood ranges; and obtain the fluctuation of the data based on the coefficient of variation of the data within the neighborhood range, and obtain the internal characteristics of the signal according to the fluctuation of the data, the overall amplitude size, and the periodicity in the local signal sequence; obtain the external environment characteristics according to the difference in the temperature fluctuation of the cabin and the correlation between the indoor and outdoor temperature sequences of the cabin; fuse the internal characteristics of the signal and the external environment characteristics positively to obtain the time-domain fault characteristics;

[0008] Segment the acoustic signal sequence of the cabin, and obtain the data change rate based on the slope of the line connecting the extreme values after curve fitting; convert the acoustic signal sequence into a spectrogram, perform curve fitting according to the peak sizes and corresponding frequencies in the spectrogram, and obtain the monotonicity of the acoustic signal sequence in the frequency domain based on the positive and negative differences of the preset number of derivatives therein; calculate the frequency-domain fault characteristics based on the monotonicity, data change rate, peak number, and average amplitude of the acoustic signal sequence in the frequency domain of each cabin;

[0009] Construct vectors from the time-domain fault characteristics and the frequency-domain fault characteristics for clustering, and obtain the fault code according to the differences in the fluctuation conditions of the time-domain fault characteristics and the frequency-domain fault characteristics in the clustering clusters, and judge the partial discharge fault.

[0010] In the above solution, the present application calculates the time-domain fault characteristics according to the change characteristics of the partial discharge signal and temperature in the box-type substation. This index reflects the time-series change characteristics of the partial discharge signal and temperature, which helps to improve the accuracy of subsequent partial discharge fault diagnosis and reduce the probability of misjudgment; then, according to the time-frequency domain characteristics of the acoustic signal, the frequency-domain fault characteristics are calculated, which helps to capture the specific characteristics of the partial discharge fault, improve the accuracy of identifying the partial discharge fault, and reduce the possibility of missed judgment; finally, clustering is performed according to the two indexes, and the fluctuation conditions of the data in different clustering clusters are used for fault identification to improve the accuracy of fault identification. In this way, combining the change characteristics of the box-type substation monitoring data in the time-frequency domain for fault judgment helps to capture the partial discharge characteristics more carefully and sensitively, thereby improving the accuracy of identifying the partial discharge fault. At the same time, fault judgment is performed according to the data change characteristics, avoiding the possibility of misjudgment caused only by the size change of the operation data, making full use of and mining the characteristics of the monitoring data, so as to achieve more accurate fault monitoring of the box-type substation and reduce the false alarm rate.

[0011] In one embodiment, the method for determining the periodicity based on the distances between the peaks of the moving average sequence is:

[0012] Sort the peak indexes of the moving average sequence according to their sizes to obtain a position sequence, calculate the first-order difference sequence of the position sequence, and represent the periodicity by the variance of the first-order difference sequence.

[0013] In one embodiment, the method for obtaining abnormal peaks and their neighborhood ranges from the peak markers of the local signal sequence is as follows:

[0014] Obtain all the peaks in the local signal sequence and the upper quartile of the peaks, and mark the peaks greater than the upper quartile as abnormal peaks; obtain a neighborhood range of a preset size centered on each abnormal peak.

[0015] In one embodiment, the method for obtaining the internal features of the signal according to the fluctuation of the data in the local signal sequence, the overall amplitude size, and the periodicity is as follows:

[0016] The internal feature value of the signal is positively correlated with the fluctuation of the data in the local signal sequence and the overall amplitude size, and negatively correlated with the periodicity in the local signal sequence.

[0017] In one embodiment, the method for obtaining the external environment features according to the difference between the temperature fluctuation in the cabin and the correlation of the temperature sequences inside and outside the cabin is as follows:

[0018] Calculate the variance of the temperature sequences inside different cabins, and use the variance as the temperature fluctuation metric of the corresponding temperature sequence of each cabin; calculate the correlation metric between the temperature sequence of each cabin and the external temperature sequence;

[0019] The external environment feature is negatively correlated with the correlation metric of each cabin and positively correlated with the temperature fluctuation metric.

[0020] In one embodiment, the method for obtaining the data change rate is as follows:

[0021] Perform polynomial fitting on each segmented acoustic wave sequence to obtain its corresponding fitting curve equation; sort all its corresponding extreme values in index order to form an extreme value sequence;

[0022] Calculate the slope between every two adjacent extreme points in the extreme value sequence, and use the mean value of all the slopes corresponding to the extreme value sequence as the average slope of this segmented acoustic wave sequence, and record the mean value of the average slopes of all segmented acoustic wave sequences as the data change rate.

[0023] In one embodiment, the method for performing curve fitting according to the peak size and the corresponding frequency in the spectrogram, and obtaining the monotonicity of the acoustic wave signal sequence in the frequency domain based on the positive and negative differences of a preset number of derivatives therein is as follows:

[0024] Calculate the mean value of all the peaks in the spectrogram, and mark the wave peaks with amplitudes greater than the mean value as characteristic wave peaks; perform least squares fitting with the peak value of the characteristic wave peak as the ordinate and the frequency corresponding to the characteristic wave peak as the abscissa, and output a quadratic fitting curve;

[0025] Preset a number of derivative points in the quadratic fitting curve, calculate the reciprocal of each derivative point, count the positive and negative signs of the derivatives, and use the ratio of the number of negative derivatives to the number of positive derivatives as the monotonicity of the acoustic signal sequence in the frequency domain.

[0026] In one embodiment, the method for calculating the frequency domain fault feature based on the monotonicity, data change rate, peak number, and average amplitude of the acoustic signal sequence of each compartment is as follows:

[0027] The frequency domain fault feature is positively correlated with the data change rate, peak number, and average amplitude of the acoustic signal sequence, and negatively correlated with the monotonicity of the acoustic signal sequence in the frequency domain.

[0028] In one embodiment, the method for obtaining the fault code according to the difference in the fluctuation of the time domain fault feature and the frequency domain fault feature in the clustering cluster is as follows:

[0029] Calculate the mean values of the time domain fault feature and the frequency domain fault feature in each clustering cluster; mark the clustering cluster with the largest sum of the two mean values as the suspected fault clustering cluster, and mark the clustering cluster with the smallest sum of the two mean values as the fault-free clustering cluster; the expression of the fault code is: ; is the fault code, is the sum of the coefficient of variation of the time domain fault feature and the frequency domain fault feature in the fault-free clustering cluster, is the sum of the coefficient of variation of the time domain fault feature and the frequency domain fault feature in the set composed of all elements in the fault-free clustering cluster and the suspected fault clustering cluster, is the variation threshold.

[0030] On the other hand, the embodiment of the present application also provides an IoT box-type substation, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the method for monitoring the faults of the IoT box-type substation described in any one of the above are implemented.

[0031] The beneficial effects of the present application are:

[0032] Based on the change characteristics of partial discharge signals and temperature in the box-type substation, this application calculates the time-domain fault characteristics. This index reflects the time-series change characteristics of partial discharge signals and temperature, which helps to improve the accuracy of subsequent partial discharge fault diagnosis and reduce the probability of misjudgment. Then, according to the time-frequency domain characteristics of the acoustic wave signals, the frequency-domain fault characteristics are calculated, which helps to capture the specific characteristics of partial discharge faults, improve the accuracy of identifying partial discharge faults, and reduce the possibility of missing judgments. Finally, clustering is performed based on the two indexes, and fault identification is carried out using the fluctuation conditions of the data in different clustering clusters to improve the accuracy of fault identification. In this way, combining the change characteristics of the monitoring data of the box-type substation in the time-frequency domain for fault judgment helps to capture partial discharge characteristics more carefully and sensitively, thereby improving the accuracy of identifying partial discharge faults. At the same time, fault judgment is carried out according to the data change characteristics, avoiding the possibility of misjudgment caused only by the size change of the operation data, making full use of and mining the characteristics of the monitoring data, and thus realizing more accurate fault monitoring of the box-type substation and reducing the false alarm rate. Description of the Drawings

[0033] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0034] Figure 1 It is a flowchart of a method for monitoring faults in an Internet of Things box-type substation provided by an embodiment of this application. Detailed Embodiments

[0035] To further elaborate on the technical means and effects adopted by this application to achieve the intended invention purpose, the following, in combination with the drawings and preferred embodiments, details the specific embodiments, structures, features, and effects of an Internet of Things box-type substation and a fault detection method proposed according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0037] An embodiment of an Internet of Things box-type substation and a fault detection method:

[0038] The following specifically describes the specific solution of a method for monitoring faults in an Internet of Things box-type substation provided by this application in combination with the drawings.

[0039] Please refer to Figure 1 , which shows a flowchart of a fault detection method for an IoT box-type substation provided by an embodiment of the present application. The method includes the following steps:

[0040] Step S001, collect different sequences using Internet of Things technology.

[0041] Monitor the box-type substation using Internet of Things technology. Among them, the perception layer collects data of the box-type substation through various sensors and devices installed in the box-type substation; among them, an ultrasonic PD detector is used to monitor partial discharge signals; temperature sensors and sound sensors installed in the transformer room, high-voltage room, and low-voltage room of the box-type substation are used to collect operation environment data of the box-type substation; a temperature monitoring device outside the box-type substation is used to collect the external environment temperature;

[0042] In this embodiment, the acquisition frequency of the partial discharge signal is 50KHz, the acquisition frequency of the acoustic signal is 10KHz, the temperature is acquired once per second, and the total acquisition duration is 15 minutes.

[0043] Input the various types of data collected by the perception layer respectively. Every 5 data in the same type form a group, and a median average filtering algorithm is used for preprocessing. The median average filtering algorithm used in this embodiment is a well-known technology and will not be elaborated here. The processed data is formed into corresponding data sequences according to the acquisition time sequence, namely the local signal sequence, the temperature sequence, the acoustic signal sequence, and the external temperature sequence. Since one temperature sensor and one sound sensor are installed in each of the transformer room, high-voltage room, and low-voltage room, a temperature sequence and an acoustic signal sequence are obtained in each of the transformer room, high-voltage room, and low-voltage room. Upload the collected data to the Internet of Things platform.

[0044] So far, the local signal sequence, the external temperature sequence, and the temperature sequences and acoustic signal sequences of different compartments have been obtained.

[0045] Step S002, perform periodic detection on the signal sequence to obtain periodicity, and combine the overall assignment and fluctuation of the sequence to obtain the internal characteristics of the signal; then obtain the external environment characteristics based on the temperature sequence difference inside and outside the compartment, and combine the two to obtain the time-domain fault characteristics.

[0046] As a mainstream device in the power supply and distribution system, the safety of the box-type substation has a decisive impact on the quality of power supply and distribution. Therefore, ensuring the safety of the box-type substation has become crucial. With the development of Internet of Things technology, the operating status of the box-type substation is monitored through various sensors, and empirical thresholds are set. When the operating data exceeds the threshold, timely reminders are given, improving the response and handling efficiency for faults. However, the utilization of monitoring data in this way is too simple, often only considering the magnitude change of the monitoring data, resulting in weak anti-interference ability and a high false alarm rate. For example, in the case of partial discharge faults in the box-type substation, it often leads to intermittent data anomalies. Due to the complexity of the operating environment of the box-type substation, it is also possible for the data to show accidental anomalies. Therefore, it is necessary to deeply explore the information hidden in the data, determine the change characteristics of the data, avoid making judgments only based on the magnitude of the data, improve the anti-interference ability of the monitoring system, reduce the false alarm rate, and thus enhance the accuracy of fault diagnosis for the box-type substation, which helps to ensure the safe operation of the box-type substation.

[0047] Partial discharge fault refers to the weakening of the insulation performance due to the damage or aging of the insulation layer in some areas, resulting in a decrease in the breakdown field strength required to break down the insulation material, and then the insulation material is broken down; or when the field strength distribution in the system is uneven, the local field strength is too large, causing the insulation material to be broken down; this kind of fault will gradually damage the insulation material in the corresponding area, and ultimately lead to serious hazards such as insulation failure, equipment failure or fire.

[0048] When there is no partial discharge fault in the box-type substation, each insulation layer has good insulation effect, so the partial discharge signal is weak, with small fluctuations and no obvious abnormal peaks. And due to the good insulation effect, the partial discharge signal will not strictly reflect in a timely manner with the periodic change of electric energy, that is, the periodicity of the partial discharge signal is weak or almost no periodic change; when there is a partial discharge fault in the box-type substation, due to the weakening of the insulation effect of the insulation layer, the partial discharge signal is enhanced, with obvious abnormal peaks, and the volatility of the data near the peak is enhanced. With the periodic change of electric energy, the partial discharge signal will also show obvious periodic changes.

[0049] Secondly, when there is no partial discharge fault in the box-type substation, the insulation layer has a good effect, so the temperature fluctuation is small, and the temperature change is mainly affected by the external temperature change, that is, it has a high correlation with the external temperature; when there is a partial discharge fault in the box-type substation, the breakdown of the insulation material will cause an increase in temperature, resulting in an increase in temperature fluctuation, and due to the great influence of partial discharge on temperature, the correlation with the external environmental temperature is weakened.

[0050] Input the local signal sequence and a window with a preset length, and use the moving average method to output the moving average sequence. In this embodiment, the preset length is set to 50. The moving average method is a well-known technology and will not be elaborated here. Sort the indices of the peaks in the moving average sequence according to the index size to form a position sequence. The position sequence indicates the position order of the peaks. Calculate the first-order difference sequence of the position sequence. The data values in the first-order difference sequence represent the distances between two peaks.

[0051] When there is a partial discharge fault, the amplitude of the partial discharge signal at the moment when the insulating layer is broken down reaches the peak value within the local area, which is also approximately equal to the position of the peak after moving average. Due to the periodic change of electric energy, the time intervals between the breakdowns of the insulating layer are basically the same. Therefore, the index differences between adjacent peaks after moving average are also basically the same or close. Based on this feature, the periodicity can be represented by the variance of the first-order difference sequence of the position sequence, and the periodicity can also be represented by the standard deviation of the first-order difference sequence. In this embodiment, the variance is used for calculation.

[0052] Obtain all the peaks and the upper quartile of the peaks in the local signal sequence. Mark the peaks greater than the upper quartile as abnormal peaks. At the same time, centered on each abnormal peak, obtain the number of signal values within the preset neighborhood range of the abnormal peak in the local signal sequence. In this embodiment, the size of the preset neighborhood range is set to 200.

[0053] Based on the mean value of the coefficient of variation of the data within the neighborhood range of all abnormal peaks in the local signal sequence as the fluctuation of the data within the neighborhood range of the abnormal peaks in the local signal sequence; among them, the role of the coefficient of variation is to measure the discreteness of the data. The more discrete the data, the more obvious the fluctuation of the data curve.

[0054] When the amplitude of the partial discharge signal is small, the data fluctuation between abnormal peaks is small, and the periodicity is weak, there is no partial discharge fault in the compartment substation; that is, obtain the internal characteristics of the signal based on the data fluctuation, overall amplitude size, and periodicity in the local signal sequence;

[0055] The internal characteristic value of the signal is positively correlated with the data fluctuation and overall amplitude size in the local signal sequence, and negatively correlated with the periodicity in the local signal sequence.

[0056] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the change directions of the two variables are the same. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large; the specific relationship is determined by the actual application, and this application does not make special restrictions.

[0057] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the change directions of the two variables are opposite. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small; the specific relationship is determined by actual applications, and this application does not impose special restrictions.

[0058] Preferably, in this embodiment, the expression of the internal signal feature is: , is the average amplitude of the local signal sequence, is the measure of the fluctuation of the local signal sequence, is the measure of the periodicity of the partial discharge signal, is the tuning parameter coefficient to avoid a zero denominator, and its value in this embodiment is 1, represents the internal signal feature.

[0059] Calculate the variance of the temperature sequences inside different compartments, use the variance as the temperature fluctuation measure for the corresponding temperature sequence of each compartment, and calculate the correlation measure between the temperature sequence of each compartment and the external temperature sequence;

[0060] When there is no partial discharge fault in the box-type substation, the temperature fluctuations in each compartment are small, and the temperature fluctuations are mainly affected by the external environmental temperature, and there is a strong positive correlation with the external environmental temperature. At this time, the correlation measure is large and the temperature fluctuation measure is small. Therefore, obtain the external environmental feature based on the correlation measure and temperature fluctuation measure of each compartment;

[0061] The external environmental feature has a negative correlation with the correlation measure of each compartment and a positive correlation with the temperature fluctuation measure.

[0062] Preferably, the expression of the external environmental feature is: , represents the temperature fluctuation measure of the i-th compartment, represents the correlation measure of the i-th compartment, represents the external environmental feature.

[0063] Whether it is based on temperature or electrical signals, the analysis is carried out at the time domain level. Therefore, calculate the time domain fault feature based on the internal signal feature and the external environmental feature;

[0064] The time domain fault feature has a positive correlation with the internal signal feature and the external environmental feature.

[0065] Preferably, in this embodiment, the time domain fault feature is the product of the internal signal feature and the external environmental feature.

[0066] So far, obtain the time domain fault feature within the time domain change range based on the analysis of the internal signal and the external environment.

[0067] Step S003: Convert the acoustic wave signal into a spectrogram, and obtain the frequency-domain fault features based on the sequence monotonicity, amplitude, data change rate, and number of peaks in the spectrogram.

[0068] The time-domain fault features measure the time-sequential changes of the partial discharge signal and temperature during the operation of the box-type substation. However, since the temperature is greatly affected by the external environment, and the partial discharge signal is also closely related to the power supply quality, it is difficult to comprehensively reflect the insulation quality of the internal components of the box-type substation only based on the changes of these data, which may increase the possibility of misjudgment or missed judgment of partial discharge faults. There are changes in the acoustic wave signal during the operation of the box-type substation. The acoustic wave signal is closely related to the performance of the internal components of the box-type substation. At the same time, when a partial discharge phenomenon occurs, an acoustic wave signal with specific characteristics will be generated. Therefore, identifying the characteristics of the acoustic wave signal helps to enhance the accuracy of partial discharge fault detection.

[0069] When there is no partial discharge fault in the box-type substation, the amplitude of the acoustic wave signal is small, showing periodic changes, and the monotonicity does not change frequently within one cycle, resulting in a small number of peaks within one cycle and a slow data change rate within one monotonic interval; in the frequency domain, the acoustic wave signal is mainly concentrated in the low-frequency part, and the signal amplitude in the high-frequency part is very low, so the amplitude as a whole shows a decreasing trend from low frequency to high frequency.

[0070] When a partial discharge fault occurs in the box-type substation, the breakdown of the insulating material will cause the acoustic wave signal to fluctuate frequently, that is, the acoustic wave signal still has periodic changes, but the amplitude of the acoustic wave signal is large, and the monotonicity changes frequently within one cycle, resulting in a large number of peaks within one cycle and a fast data change rate within one monotonic interval; in the frequency domain, the high-frequency components increase, the signal amplitude in the high-frequency part is high, and at the same time, the normal low-frequency components still exist, so the amplitude as a whole shows a trend of decreasing first and then increasing from low frequency to high frequency.

[0071] For the acoustic wave signal sequence of each compartment, segment the acoustic wave signal sequence with a time period of a preset length to obtain a number of segmented acoustic wave sequences, where the preset length needs to include at least one cycle. In this embodiment, the time period of the preset length is taken as 50 ms.

[0072] Perform polynomial fitting on each segmented acoustic wave sequence to obtain its corresponding fitting curve equation; take the derivative of the corresponding fitting curve equation, and form an extreme value sequence in the order of indexes for all data points with a derivative of 0. The elements in the extreme value sequence are the acoustic wave signal amplitudes of each extreme point. In the above manner, each acoustic wave signal sequence is segmented, and the extreme value sequence of each segmented acoustic wave sequence is obtained. Polynomial fitting is a well-known technology and will not be elaborated in this application.

[0073] When a partial discharge fault occurs in the box-type substation, the change rate of data is relatively fast within a monotonic interval. Therefore, based on the extreme value sequence, the slope of the line connecting two adjacent extreme points is calculated. The greater the change in this slope, the more likely a partial discharge fault occurs. The slopes are calculated for every two adjacent extreme points in the extreme value sequence, and the mean of all the slopes corresponding to the extreme value sequence is used as the average slope of this segmented acoustic wave sequence. The mean of the average slopes of all the segmented acoustic wave sequences is denoted as the data change rate.

[0074] Taking an acoustic wave signal sequence as the input, a spectrogram is obtained by Fourier transform, and the mean of all the peaks is calculated. The wave peaks with amplitudes greater than the mean are marked as characteristic wave peaks; with the peak values of the characteristic wave peaks as the ordinate and the frequencies corresponding to the characteristic wave peaks as the abscissa, least squares fitting is performed, and a quadratic fitting curve is output; H points are evenly taken on the quadratic fitting curve, and derivatives are calculated at these H points respectively. Fourier transform and least squares fitting are well-known technologies and will not be elaborated here. In this embodiment, H is taken as 100.

[0075] Count the number of points with negative derivatives and the number of points with positive derivatives among the H points taken on the quadratic fitting curve. If the quadratic fitting curve is monotonically decreasing, all the derivatives are negative; if the quadratic fitting curve is monotonically increasing, all the derivatives are positive. The ratio of the number of points with negative derivatives to the number of points with positive derivatives is used as the monotonicity of the acoustic wave signal sequence in the frequency domain.

[0076] Based on the monotonicity, data change rate, number of peaks, and average amplitude of the acoustic wave signal sequence in each compartment, frequency domain fault characteristics are obtained.

[0077] The frequency domain fault characteristics are positively correlated with the data change rate, number of peaks, and average amplitude of the acoustic wave signal sequence, and negatively correlated with the monotonicity of the acoustic wave signal sequence in the frequency domain.

[0078] Preferably, in this embodiment, the expression of the frequency domain fault characteristics is: ; represents the number of peaks of the acoustic wave signal sequence in the i-th compartment, represents the average amplitude of the acoustic wave signal sequence in the i-th compartment, represents the data change rate of the acoustic wave signal sequence in the i-th compartment, represents the monotonicity of the acoustic wave signal sequence in the frequency domain in the i-th compartment, represents the frequency domain fault characteristics.

[0079] It can be understood that when there is no partial discharge fault in the box-type substation, the number of peaks of the acoustic wave signal is small, the peak values are small, and the change rate within the monotonic interval is slow, that is 、 , are all relatively small; secondly, the quadratic fitting curve of the characteristic peak value in the frequency domain shows a monotonically decreasing trend, so the derivatives are all negative, that is, is relatively large; therefore, when there is no partial discharge fault in the box-type substation, the time-frequency domain characteristic coefficient is relatively small. On the contrary, when there is a partial discharge fault in the box-type substation, the quadratic fitting curve of the characteristic peak value in the frequency domain first decreases and then increases, the number of positive derivatives increases, and the number of negative derivatives decreases, that is, is relatively small, and the rest of the characteristics are opposite, thus making the frequency domain fault characteristics relatively large.

[0080] So far, the frequency domain fault characteristics have been obtained from the frequency domain perspective based on the acoustic signal sequence of each compartment.

[0081] Step S004, construct a fault code according to the time domain fault characteristics and the frequency domain fault characteristics, and judge the partial discharge fault.

[0082] According to the data acquisition method in Step 1, calculate the time domain fault characteristics and the frequency domain fault characteristics during each monitoring of the box-type substation; form a vector of the two as the characteristic vector of one monitoring. Take all the characteristic vectors of the box-type substation monitoring and the number of clustering clusters as the input, and use the K-means clustering algorithm to output each clustering cluster. In this embodiment, the number of clustering clusters takes the value of 4. K-means clustering is a well-known technology and will not be elaborated here.

[0083] Then calculate the mean values of the time domain fault characteristics and the frequency domain fault characteristics in each clustering cluster respectively. Mark the clustering cluster with the largest sum of the two mean values as the suspected fault clustering cluster, and mark the clustering cluster with the smallest sum of the two mean values as the fault-free clustering cluster. Determine the fault code according to the fluctuation conditions of the suspected fault clustering cluster and the fault-free clustering cluster, so as to judge whether there is a partial discharge fault in the box-type substation.

[0084] The specific expression is: ; is the fault code ( being 0 indicates that there is no partial discharge fault, being 1 indicates that there is), is the sum of the coefficients of variation of the time domain fault characteristics and the frequency domain fault characteristics in the fault-free clustering cluster. All elements in the fault-free clustering cluster and the suspected fault clustering cluster form a set, is the sum of the coefficients of variation of the time domain fault characteristics and the frequency domain fault characteristics in this set, is the variation threshold, which is used to measure the difference between the elements in the suspected fault clustering cluster and the elements in the fault-free clustering cluster, and can be selected as needed. In this embodiment takes the empirical value 1. It should be noted that The greater the value, the stricter the requirements for the differences between elements in the two clusters, which can improve the sensitivity of partial discharge fault diagnosis but also increase the probability of misjudgment.

[0085] In this way, based on the variation characteristics of each monitoring data of the box-type substation in the time-frequency domain for fault judgment, it avoids the situation in the existing detection technology that a large risk of misjudgment may be caused by only judging according to the numerical value, can relatively more accurately determine whether there are partial discharge characteristics, improves the accuracy of partial discharge fault diagnosis, helps to ensure the operation safety of the box-type substation, and reduces the adverse effects brought by misjudgment and ineffective maintenance.

[0086] Based on the same inventive concept as the above method, an embodiment of the present invention further provides an Internet of Things box-type substation, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above methods for monitoring faults of an Internet of Things box-type substation.

[0087] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

[0088] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. The key point of each embodiment is to illustrate the differences from other embodiments.

Claims

1. A fault monitoring method for an IoT box-type substation, characterized in that, The method includes the following steps: Using Internet of Things technology to collect a local signal sequence, an external temperature sequence, temperature sequences of different compartments, and acoustic signal sequences each time; Processing the local signal sequence based on a window of a preset size to obtain a moving average sequence; determining periodicity based on the distance between the peaks of the moving average sequence; marking the peaks of the local signal sequence to obtain abnormal peaks and their neighborhood ranges; and obtaining the fluctuation of the data based on the coefficient of variation of the data within the neighborhood range, and obtaining the internal characteristics of the signal according to the fluctuation of the data, the overall amplitude size, and the periodicity in the local signal sequence; obtaining the external environment characteristics according to the differences in the temperature fluctuations of the compartments and the correlation between the temperature sequences inside and outside the compartments; positively fusing the internal characteristics of the signal and the external environment characteristics to obtain time-domain fault characteristics; Segmenting the acoustic signal sequence of the compartment, obtaining the data change rate based on the slope of the line connecting the extreme values after curve fitting; converting the acoustic signal sequence into a spectrogram, performing curve fitting according to the peak sizes and corresponding frequencies in the spectrogram, and obtaining the monotonicity of the acoustic signal sequence in the frequency domain based on the positive and negative differences of a preset number of derivatives therein; calculating the frequency-domain fault characteristics based on the monotonicity, data change rate, peak number, and average amplitude of the acoustic signal sequence of each compartment in the frequency domain; Constituting vectors of the time-domain fault characteristics and the frequency-domain fault characteristics for clustering, and obtaining a fault code according to the differences in the fluctuations of the time-domain fault characteristics and the frequency-domain fault characteristics in the clustering cluster to judge the partial discharge fault.

2. The fault monitoring method of an IoT box-type substation according to claim 1, characterized in that, The method for determining periodicity based on the distance between the peaks of the moving average sequence is: Sorting the peak indices of the moving average sequence according to their magnitudes to obtain a position sequence, calculating the first-order difference sequence of the position sequence, and representing the periodicity by the variance of the first-order difference sequence.

3. The fault monitoring method of an IoT box-type substation according to claim 1, characterized in that, The method for marking the peaks of the local signal sequence to obtain abnormal peaks and their neighborhood ranges is: Obtaining all the peaks and the upper quartile of the peaks in the local signal sequence, and marking the peaks greater than the upper quartile as abnormal peaks; obtaining a neighborhood range of a preset size centered on each abnormal peak.

4. The fault monitoring method of an IoT box-type substation according to claim 1, wherein, The method for obtaining the internal characteristics of the signal according to the fluctuation of the data, the overall amplitude size, and the periodicity in the local signal sequence is: The internal characteristic value of the signal is positively correlated with the fluctuation of the data and the overall amplitude size in the local signal sequence, and negatively correlated with the periodicity in the local signal sequence.

5. The method for monitoring faults of an IoT box-type substation according to claim 1, characterized in that, The method for obtaining the external environment characteristics according to the differences in the temperature fluctuations of the compartments and the correlation between the temperature sequences inside and outside the compartments is: Calculating the variance of the temperature sequences inside different compartments, and using the variance as the temperature fluctuation measure of the corresponding temperature sequence of each compartment; calculating the correlation measure between the temperature sequence of each compartment and the external temperature sequence; The external environment characteristics are negatively correlated with the correlation measure of each compartment and positively correlated with the temperature fluctuation measure.

6. The method for monitoring faults of an IoT box-type substation according to claim 1, characterized in that, The method for obtaining the data change rate is: Performing polynomial fitting on each segmented acoustic sequence to obtain its corresponding fitting curve equation; sorting all its corresponding extreme values in index order to form an extreme value sequence; For adjacent two extreme points in the extreme value sequence, calculate their slopes, and take the mean value of all slopes corresponding to the extreme value sequence as the average slope of this segmented acoustic wave sequence. Denote the mean value of the average slopes of all segmented acoustic wave sequences as the data change rate.

7. The fault monitoring method of an IoT box-type substation according to claim 1, characterized in that, The method for curve fitting according to the peak magnitude and corresponding frequency in the spectrogram and obtaining the monotonicity of the acoustic wave signal sequence in the frequency domain based on the positive and negative differences of a preset number of derivatives therein is as follows: Calculate the mean value of all peaks in the spectrogram, and mark the wave peaks with amplitudes greater than the mean value as characteristic wave peaks; Taking the peak value of the characteristic wave peak as the ordinate and the frequency corresponding to the characteristic wave peak as the abscissa, perform least squares fitting and output a quadratic fitting curve; Preset several derivative points in the quadratic fitting curve, calculate the reciprocal of each derivative point, count the positive and negative of the derivatives, and take the ratio of the number of derivatives that are negative to the number of derivatives that are positive as the monotonicity of the acoustic wave signal sequence in the frequency domain.

8. The fault monitoring method of an IoT box-type substation according to claim 1, characterized in that, The method for calculating the frequency domain fault characteristics based on the monotonicity, data change rate, peak number, and average amplitude of the acoustic wave signal sequence in each cabin is as follows: The frequency domain fault characteristics are positively correlated with the data change rate, peak number, and average amplitude of the acoustic wave signal sequence, and negatively correlated with the monotonicity of the acoustic wave signal sequence in the frequency domain.

9. The fault monitoring method of an IoT box-type substation according to claim 1, characterized in that, The method for obtaining the fault code according to the differences in the fluctuations of the time domain fault characteristics and the frequency domain fault characteristics in the clustering cluster is as follows: Calculate the means of the time-domain fault features and frequency-domain fault features in each cluster; mark the cluster with the largest sum of the two means as the suspected fault cluster, and mark the cluster with the smallest sum of the two means as the fault-free cluster; the expression of the fault code is: ; is the fault code, is the sum of the coefficient of variation of the time-domain fault features and frequency-domain fault features in the fault-free cluster, is the sum of the coefficient of variation of the time-domain fault features and frequency-domain fault features in the set composed of all elements in the fault-free cluster and the suspected fault cluster, is the preset variation threshold.

10. An Internet of Things box-type substation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a fault monitoring method for an IoT box-type substation as described in any one of claims 1-9.

Images