Intelligent sensing method for identifying circuit faults of a semiconductor hybrid solid-state circuit breaker

Abstract

The application relates to the technical field of circuit monitoring, in particular to a smart sensing knowledge identification method for circuit faults of a semiconductor hybrid solid-state circuit breaker; the fluctuation abnormal moment is obtained according to the amplitude feature and the duration feature of current fluctuation in a current time sequence curve; the suspected current abnormal moment is obtained according to the difference feature of the fluctuation abnormal degree of the same moment in history; the different clustering clusters are obtained by clustering the current features and the fluctuation features of all suspected current abnormal moments in the current time sequence curve; the current abnormal degree of the clustering cluster is obtained according to the number feature of the suspected current abnormal moment in the clustering cluster and the distance feature in the clustering space; the current abnormal moment is obtained according to the current abnormal degree; the fault feature value is obtained according to the distribution feature of the current abnormal moment in the current time sequence curve and the change feature of the current abnormal degree, and circuit fault identification is carried out, so that the stability of the circuit breaker work is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of circuit monitoring, in particular to a smart knowledge identification method for circuit fault of a semiconductor hybrid solid-state circuit breaker. BACKGROUND

[0002] The semiconductor hybrid solid-state circuit breaker is a power protection device combining semiconductor devices and mechanical switches, which has the advantages of fast circuit breaking capacity and low conduction loss, and is mainly used in medium and high voltage DC circuit systems. Usually, the line current is monitored by a current sensor, and when the current exceeds the safety value, it is determined that the line is faulty, and the semiconductor hybrid solid-state circuit breaker device cuts off the current instantaneously to achieve the purpose of fast cutting and protecting the circuit. However, this method cannot provide early warning of potential faults, and only monitors the current when it is too large to directly cut off the circuit. When the power is cut off, the fault current is too large, which can easily cause the device to overheat due to the high load of the circuit breaker, causing further circuit failure and increasing the risk of failure. SUMMARY

[0003] In order to solve the technical problem that when only the current size is used for circuit monitoring, the fault current is too large when the power is cut off, which can easily cause the device to overheat due to the high load of the circuit breaker, causing further circuit failure and increasing the risk of failure, the purpose of the present application is to provide a smart knowledge identification method for circuit fault of a semiconductor hybrid solid-state circuit breaker, and the technical solution adopted is as follows:

[0004] Obtain the current time sequence curve of the circuit where the semiconductor hybrid solid-state circuit breaker is located.

[0005] Obtain the fluctuation abnormality degree of the fluctuation point according to the amplitude feature and the duration feature of the current fluctuation in the current time sequence curve; obtain the fluctuation abnormal time according to the fluctuation abnormality degree; obtain the fluctuation regularity degree according to the difference feature of the fluctuation abnormality degree of the fluctuation abnormal time and the historical same time; obtain the suspected current abnormal time according to the fluctuation regularity degree;

[0006] Cluster the current features and fluctuation features of all suspected current abnormal times in the current time sequence curve to obtain different clustering clusters; obtain the current abnormality degree of the clustering cluster according to the number feature of the suspected current abnormal time in the clustering cluster and the distance feature in the clustering space; obtain the current abnormal time according to the current abnormality degree.

[0007] Obtain the fault feature value according to the distribution feature of the current abnormal time and the change feature of the current abnormality degree in the current time sequence curve; and identify the circuit fault according to the fault feature value.

[0008] Further, the step of obtaining the fluctuation abnormality degree of the fluctuation point according to the amplitude feature and the duration feature of the current fluctuation in the current time sequence curve comprises:

[0009] The extreme point in the current time sequence curve is taken as a fluctuation point; the absolute value of the slope change rate of the data points in the current time sequence curve is calculated and normalized to obtain a change characteristic value of the data points; the data point whose change characteristic value exceeds a preset inflection point threshold is taken as an inflection point; the average value of the absolute value of the difference between the fluctuation point and the adjacent inflection points before and after the fluctuation point in the current time sequence curve is calculated to obtain a fluctuation amplitude of the fluctuation point; the ratio of the fluctuation amplitude to the maximum value in all fluctuation amplitudes is calculated to obtain an amplitude characteristic value; the time interval between the adjacent inflection points before and after the fluctuation point in the current time sequence curve is calculated to obtain a fluctuation duration of the fluctuation point; the ratio of the fluctuation duration to the maximum value in all fluctuation durations is calculated and negatively correlated to obtain a duration characteristic value; and the sum of the amplitude characteristic value and the duration characteristic value is calculated to obtain a fluctuation abnormality degree of the fluctuation point.

[0010] Further, the step of obtaining a fluctuation abnormality time according to the fluctuation abnormality degree comprises:

[0011] When the fluctuation abnormality degree of the fluctuation point exceeds a preset fluctuation threshold, the time at which the fluctuation point is located is a fluctuation abnormality time.

[0012] Further, the step of obtaining a fluctuation rule degree according to the difference feature of the fluctuation abnormality time and the fluctuation abnormality degree of the historical same time comprises:

[0013] The average value of the absolute value of the difference between the fluctuation abnormality time and the fluctuation abnormality degree of all historical same times is calculated and negatively correlated to obtain a fluctuation rule degree of the fluctuation abnormality time.

[0014] Further, the step of obtaining a suspected current abnormality time according to the fluctuation rule degree comprises:

[0015] When the fluctuation rule degree of the fluctuation abnormality time does not exceed a preset rule threshold, the fluctuation abnormality time is a suspected current abnormality time.

[0016] Further, the step of clustering according to the current feature and the fluctuation feature of all suspected current abnormality times in the current time sequence curve to obtain different clustering clusters comprises:

[0017] The K-means clustering algorithm is used to cluster according to the current value corresponding to all suspected current abnormality times, the fluctuation amplitude and the fluctuation duration to obtain different clustering clusters.

[0018] Further, the step of obtaining a current abnormality degree of a clustering cluster according to the number feature of the suspected current abnormality times in the clustering cluster and the distance feature in the clustering space comprises:

[0019] Calculate the average of the Euclidean distances between all suspected current abnormal time points in the clustering cluster and other suspected current abnormal time points in the nearest neighbor in the clustering space, to obtain a distribution dispersion value; calculate the product of the reciprocal of the number of suspected current abnormal time points in the clustering cluster and the distribution dispersion value and normalize it to obtain the current abnormality degree of the clustering cluster.

[0020] Further, the step of obtaining current abnormal time points according to the current abnormality degree comprises:

[0021] Suspected current abnormal time points in the clustering cluster whose current abnormality degree exceeds the preset abnormal threshold are taken as current abnormal time points.

[0022] Further, the step of obtaining a fault feature value according to the distribution characteristics and the change characteristics of the current abnormality degree of the current abnormal time points in the current time sequence curve comprises:

[0023] The current time sequence curve is divided into a first half and a second half from the middle time point; the ratio of the number of current abnormal time points in the second half to the first half is calculated to obtain a first fault trend value; the ratio of the average of the current abnormality degrees of the current abnormal time points in the second half to the first half is calculated to obtain a second fault trend value; the product of the first fault trend value and the second fault trend value is calculated to obtain the fault feature value.

[0024] Further, the step of identifying a circuit fault according to the fault feature value comprises:

[0025] When the fault feature value exceeds the preset fault threshold, a circuit fault warning is performed.

[0026] The present application has the following beneficial effects:

[0027] In the present application, since the fluctuation characteristics of the current can reflect whether the circuit current is abnormal, the fluctuation abnormality degree can represent the fluctuation characteristics of the current in the circuit, and then the fluctuation abnormality time of the current can be preliminarily determined according to the fluctuation abnormality degree. Since the factors causing the fluctuation abnormality time are not only circuit faults, but also large load switch changes, the fluctuation regularity degree can determine whether the fluctuation abnormality degree of the fluctuation abnormality time is regular, so as to distinguish the fluctuation abnormality time caused by the load switch change from the real circuit fault according to the fluctuation regularity degree. The suspected current abnormality time can determine the current fluctuation time caused by the non-fixed time load switch change. The clustering cluster can distinguish the current change characteristics corresponding to different suspected current abnormality times. Since the current change characteristics caused by the circuit fault have low similarity and low frequency, the current abnormality degree can represent the triggering factors of the suspected current abnormality time in the clustering cluster. The current abnormality time can represent the time of the real circuit fault. The fault feature value can represent whether the circuit fault has an increasing risk trend. Finally, the circuit fault is identified according to the fault feature value, which improves the accuracy of the circuit fault identification and the stability of the circuit breaker work, and avoids the damage of the circuit breaker caused by the overheat of the circuit breaker under multiple loads. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, the following will briefly introduce the drawings needed in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0029] Figure 1 A flow chart of a smart knowledge identification method for circuit fault of a semiconductor hybrid solid-state circuit breaker is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined application purpose, the following will combine the drawings and the preferred embodiments to specifically describe a smart knowledge identification method for circuit fault of a semiconductor hybrid solid-state circuit breaker according to the present application, the specific implementation, structure, features and effects thereof, as follows. 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.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0032] Specifically, the application provides a method for intelligently identifying circuit faults of a semiconductor hybrid solid-state circuit breaker.

[0033] Please refer to Figure 1 which shows a flowchart of a method for intelligently identifying circuit faults of a semiconductor hybrid solid-state circuit breaker according to an embodiment of the application. The method comprises the following steps:

[0034] In step S1, the current time series curve of the circuit in which the semiconductor hybrid solid-state circuit breaker is located is obtained.

[0035] The current time series curve of the circuit in which the semiconductor hybrid solid-state circuit breaker is located is obtained. In the embodiment of the application, the current monitoring and collection frequency is 10 Hz, and 10 seconds of current data is collected each time. Since the current will fluctuate slightly in a normal state, in order to improve the accuracy of fault identification, the collected current data is smoothed to remove the slight fluctuation characteristics. The smoothed current data is projected into a two-dimensional rectangular coordinate system to construct the current time series curve, with the horizontal axis representing time and the vertical axis representing current. The current data within 10 seconds before the current time is taken as the latest current time series curve. The implementer can determine the collection frequency and the length of the time series curve according to the implementation scenario.

[0036] In step S2, the fluctuation abnormality degree of the fluctuation point is obtained according to the amplitude characteristics and duration characteristics of the current fluctuation in the current time series curve. The fluctuation abnormality time is obtained according to the fluctuation abnormality degree. The fluctuation rule degree is obtained according to the difference characteristics of the fluctuation abnormality degree between the fluctuation abnormality time and the historical same time. The suspected current abnormality time is obtained according to the fluctuation rule degree.

[0037] First, the latest current time series curve is analyzed to determine whether there is an abnormality in the real-time monitoring period. Abnormal fluctuation of the current means that there may be a risk of circuit failure. In a normal situation, the circuit current will fluctuate slightly, such as power supply ripple or slow and slight fluctuation caused by temperature. Abnormal fluctuation is usually larger in amplitude and faster in appearance and end than normal fluctuation. Therefore, the fluctuation abnormality degree of the fluctuation point can be obtained according to the amplitude characteristics and duration characteristics of the current fluctuation in the current time series curve.

[0038] Preferably, in the embodiments of the present application, the step of obtaining the fluctuation abnormality degree comprises: taking the extreme point in the current time series curve as a fluctuation point; it should be noted that the extreme point can be obtained by using the existing AMPD peak finding algorithm, and the extreme point is a wave peak point and a wave trough point in the current time series curve, in which the fluctuation characteristics are more obvious. The absolute value of the slope change rate of the data point in the current time series curve is calculated and normalized to obtain the change characteristic value of the data point; the slope change rate is the slope difference between the data point and the previous data point, and under normal circumstances, the current data is relatively stable, and the change characteristic value is small; when the current fluctuates, the inflection points at the beginning and the end of the fluctuation have obvious slope changes, and the change characteristic value is large. In the embodiments of the present application, the normalization means is linear normalization, the change characteristic value of the data point with consistent current change trend tends to 0, and the change characteristic value of the data point at the place where the current change trend appears difference tends to 1; therefore, the preset inflection point threshold is set to 0.5, and the data point with a change characteristic value exceeding the preset inflection point threshold is taken as an inflection point; the inflection point represents the starting position of the current fluctuation. The average value of the absolute value of the difference between the fluctuation point and the adjacent inflection points before and after the fluctuation point in the current time series curve is calculated to obtain the fluctuation amplitude of the fluctuation point; the greater the fluctuation amplitude, the more obvious the fluctuation, and the more likely it is an abnormal fluctuation. The ratio of the fluctuation amplitude to the maximum value in all fluctuation amplitudes is calculated to obtain the amplitude characteristic value; the purpose is to make the value range of the fluctuation amplitude between 0 and 1, remove the dimensional influence, and the greater the amplitude characteristic value, the greater the fluctuation degree, and the more abnormal the fluctuation. The time interval between the adjacent inflection points before and after the fluctuation point in the current time series curve is calculated to obtain the fluctuation duration of the fluctuation point; the shorter the fluctuation duration, the faster the fluctuation, and the less likely it is a slow fluctuation. The ratio of the fluctuation duration to the maximum value in all fluctuation durations is calculated and negatively related to map to obtain the duration characteristic value; the purpose is to make the value range of the fluctuation duration between 0 and 1, and the smaller the fluctuation duration, the closer the value of the duration characteristic value to 1, and the more likely it is an abnormal fluctuation. The sum of the amplitude characteristic value and the duration characteristic value is calculated to obtain the fluctuation abnormality degree of the fluctuation point; the greater the fluctuation abnormality degree, the more likely it is that the fluctuation is caused by the abnormal circuit; the smaller the fluctuation abnormality degree, the more likely it is that the fluctuation is under the normal condition of the circuit. The formula for obtaining the fluctuation abnormality degree comprises:

[0039]

[0040] In the formula, R i represents the fluctuation abnormality degree of the i th fluctuation point, S i represents the fluctuation amplitude of the i th fluctuation point, S max represents the maximum value of the fluctuation amplitude, represents the amplitude characteristic value, T i represents the fluctuation duration of the i th fluctuation point, T max represents the maximum value of the fluctuation duration, exp() represents the exponential function with the natural constant as the base, representing a duration characteristic value.

[0041] Further, the fluctuation abnormal time can be obtained according to the fluctuation abnormal degree, specifically including: when the fluctuation abnormal degree of the fluctuation point exceeds a preset fluctuation threshold, the time at which the fluctuation point is located is the fluctuation abnormal time; in the embodiment of the present application, the preset fluctuation threshold is 0.6, and when the preset fluctuation threshold is exceeded, it means that the current fluctuation feature at this time is greatly different from that under normal circumstances, the fluctuation feature is abnormal, and then the time at which the fluctuation point is located is taken as the fluctuation abnormal time, and the implementer can determine the preset fluctuation threshold according to the implementation scene. After obtaining the fluctuation abnormal time, it is necessary to further analyze whether it is caused by the circuit abnormality; some large loads in the circuit automatically change at fixed time points, and the time points of such regular load changes may be determined as the fluctuation abnormal time, but they are not caused by the real circuit fault; therefore, the fluctuation regularity degree is obtained according to the difference feature of the fluctuation abnormal degree of the fluctuation abnormal time and the historical same time.

[0042] Preferably, in the embodiment of the present application, the step of obtaining the fluctuation regularity degree includes: calculating the average value of the absolute value of the difference between the fluctuation abnormal time and the fluctuation abnormal degree of all historical same time, and negatively correlating mapping to obtain the fluctuation regularity degree of the fluctuation abnormal time, in the embodiment of the present application, the exp() function is used for negatively correlating mapping, and the range of the fluctuation regularity degree is 0 to 1; when the fluctuation abnormal degree of the historical same time and the fluctuation abnormal time is similar, it means that the current fluctuation at this time is caused by the load switch change of the large load at the fixed time. Therefore, the greater the fluctuation regularity degree is, the more likely the fluctuation abnormal time is caused by the normal load change; the smaller the fluctuation regularity degree is, the more likely the fluctuation abnormal time is caused by the real circuit fluctuation abnormality. Then the suspected current abnormal time can be obtained according to the fluctuation regularity degree; preferably, in the embodiment of the present application, the step of obtaining the suspected current abnormal time includes: when the fluctuation regularity degree of the fluctuation abnormal time does not exceed a preset regularity threshold, the fluctuation abnormal time is the suspected current abnormal time; since the fluctuation regularity degree corresponding to the normal load change tends to 1, and the fluctuation regularity degree corresponding to the real circuit abnormality tends to 0, the preset regularity threshold in the embodiment of the present application is 0.5. The suspected current abnormal time means that this time has a greater possibility of being caused by the circuit fault.

[0043] Step S3, clustering according to the current features and fluctuation features of all suspected current abnormal times in the current time sequence curve to obtain different clustering clusters; obtaining the current abnormal degree of the clustering cluster according to the number features of the suspected current abnormal times in the clustering cluster and the distance features in the clustering space; obtaining the current abnormal time according to the current abnormal degree.

[0044] Since there are occasionally large loads controlled by manual switch in the circuit, such time-variable load switch changes can also cause suspected current abnormal moments; therefore, further analysis is needed according to the difference between the current change characteristics of the manual control load switch and the current change characteristics of the circuit fault abnormality. The manual control load switch does not have regularity characteristics, but the current change mode caused by the change of the manually controlled load is similar, and with the accumulation of the number of manual regulation and control, the current change characteristics with high similarity will repeatedly appear in the historical current time sequence curve; and each time the circuit fault such as overload, electric leakage, component aging and electric arc occurs, the environment and the fault degree are different, and the current change characteristics generated are relatively random, and the possibility of appearing similar current change mode in the historical current time sequence curve is low. Then, the current characteristics and fluctuation characteristics of all suspected current abnormal moments in the current time sequence curve can be clustered to obtain different clustering clusters; preferably, in the embodiments of the present application, the step of obtaining different clustering clusters comprises: clustering according to the current value, fluctuation amplitude and fluctuation duration corresponding to all suspected current abnormal moments by K-means clustering algorithm to obtain different clustering clusters; it should be noted that the K-means clustering algorithm belongs to the prior art, the number of clustering clusters is obtained by elbow method, and the specific clustering steps will not be described again. Through clustering, suspected current abnormal moments with different fluctuation characteristics can be divided.

[0045] Further, the more the number of suspected current abnormal moments in the clustering cluster, the more the number of suspected current abnormal moments with similar current change characteristics, and the more likely the suspected current abnormal moments in the clustering cluster are caused by manual control load change. The fewer the number of suspected current abnormal moments in the clustering cluster, the fewer the number of suspected current abnormal moments with similar current change characteristics, and the more likely the suspected current abnormal moments in the clustering cluster are caused by circuit fault. In the clustering process, the more similar the current change characteristics of two suspected current abnormal moments, the shorter the Euclidean distance in the clustering space; therefore, when the Euclidean distance of the suspected current abnormal moments in the clustering cluster in the clustering space is shorter, it means that the current change characteristics corresponding to the suspected current abnormal moments in the clustering cluster are more similar, and are more likely to be caused by manual control load change; on the contrary, when the Euclidean distance is larger, it means that the current change characteristics are less similar, and the suspected current abnormal moments in the clustering cluster are more likely to be caused by circuit fault. Therefore, the current abnormality degree of the clustering cluster is obtained according to the number characteristics of the suspected current abnormal moments in the clustering cluster and the distance characteristics in the clustering space.

[0046] Preferably, in the embodiments of the present application, the step of obtaining the current abnormality degree comprises: calculating the average of the Euclidean distance between all suspected current abnormal time points in the cluster cluster and other suspected current abnormal time points with the nearest neighbor in the cluster space, obtaining the distribution dispersion value; the greater the distribution dispersion value, the lower the similarity of the current change characteristics of the suspected current abnormal time points in the cluster cluster, and the more likely it is to represent the circuit failure. Calculate the product of the reciprocal of the number of suspected current abnormal time points in the cluster cluster and the distribution dispersion value and normalize it to obtain the current abnormality degree of the cluster cluster; the greater the current abnormality degree, the more likely the suspected current abnormal time points in the cluster cluster are caused by circuit failure. Further, the current abnormal time points can be obtained according to the current abnormality degree, specifically including: the suspected current abnormal time points in the cluster cluster whose current abnormality degree exceeds the preset abnormal threshold value are taken as the current abnormal time points; since the value range of the current abnormality degree after linear normalization is 0 to 1, the current abnormality degree of the cluster cluster corresponding to the manual control load change tends to 0, while the current abnormality degree of the cluster cluster corresponding to the real current failure tends to 1, both of which are at both ends of the value range; therefore, in the embodiments of the present application, the preset abnormal threshold value is 0.5, and the implementer can determine it according to the implementation scene.

[0047] Step S4, obtaining a fault feature value according to the distribution characteristics of the current abnormal time points and the change characteristics of the current abnormality degree in the current time series curve; and performing circuit fault identification according to the fault feature value.

[0048] The current abnormal time points represent the time points of significant current fluctuations caused by non-load switch changes, and the more current abnormal time points, the greater the possibility of current failure. Therefore, a fault feature value is obtained according to the distribution characteristics of the current abnormal time points and the change characteristics of the current abnormality degree in the current time series curve; preferably, in the embodiments of the present application, the step of obtaining the fault feature value comprises: dividing the current time series curve into a first half and a second half from the middle time point; calculating the ratio of the number of current abnormal time points in the second half to the first half to obtain a first fault trend value; when the number of current abnormal time points in the second half is more than that in the first half, the first fault trend value is greater than 1, which means that the current abnormal time points have a trend of increasing, and the circuit failure trend is obvious. Calculate the ratio of the average of the current abnormality degrees corresponding to the current abnormal time points in the second half to the first half to obtain a second fault trend value; when the average of the current abnormality degrees of the current abnormal time points in the second half is greater than that in the first half, the second fault trend value is greater than 1, which means that the risk degree of circuit failure is larger. Calculate the product of the first fault trend value and the second fault trend value to obtain the fault feature value. The greater the fault feature value, the more obvious the trend of circuit failure, and the greater the risk degree of failure. The formula for obtaining the fault feature value comprises:

[0049]

[0050] In the formula, W represents a fault characteristic value, D1 represents the number of current abnormal time in the first half, D2 represents the number of current abnormal time in the second half, represents a first fault trend value, H1 represents the average value of the current abnormal degree corresponding to the current abnormal time in the first half, and H2 represents the average value of the current abnormal degree corresponding to the current abnormal time in the second half, represents a second fault trend value.

[0051] Further, after obtaining the fault characteristic value, circuit fault recognition can be performed according to the fault characteristic value, specifically including: when the fault characteristic value exceeds a preset fault threshold, a circuit fault warning is performed; in the embodiment of the present application, the preset fault threshold is 1, and when the fault characteristic value exceeds 1, it means that the current abnormal characteristics in the second half are more obvious than those in the first half, the trend of circuit fault increases, and the risk rises; in order to avoid the situation that the semiconductor hybrid solid-state circuit breaker directly breaks when the fault current is too large, causing the overload and overheating of the circuit breaker and damage, timely warning is needed, thereby improving the stability of the circuit breaker work.

[0052] To sum up, the embodiment of the present application provides a smart and intelligent recognition method for circuit fault of a semiconductor hybrid solid-state circuit breaker; the fluctuation abnormal time is obtained according to the amplitude characteristics and duration characteristics of current fluctuation in a current time sequence curve; the suspected current abnormal time is obtained according to the difference characteristics of fluctuation abnormal degree of the fluctuation abnormal time and the historical same time; the different clustering clusters are obtained by clustering according to the current characteristics and fluctuation characteristics of all suspected current abnormal times in the current time sequence curve; the current abnormal degree of the clustering cluster is obtained according to the number characteristics of suspected current abnormal times in the clustering cluster and the distance characteristics in the clustering space; the current abnormal time is obtained according to the current abnormal degree; the fault characteristic value is obtained according to the distribution characteristics of the current abnormal time and the change characteristics of the current abnormal degree in the current time sequence curve, and circuit fault recognition is performed, thereby improving the stability of the circuit breaker work.

[0053] It should be noted that: the above-mentioned embodiment of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0054] Each embodiment in the specification is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

Claims

1. A smart sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker, characterized in that, The method includes the following steps: Obtain the current timing curve of the circuit containing the semiconductor hybrid solid-state circuit breaker; The degree of fluctuation anomaly at the fluctuation point is obtained based on the amplitude and duration characteristics of the current fluctuation in the current time series curve; the time of fluctuation anomaly is obtained based on the degree of fluctuation anomaly; the degree of fluctuation regularity is obtained based on the difference between the degree of fluctuation anomaly at the time of fluctuation anomaly and the same time in history; and the suspected time of current anomaly is obtained based on the degree of fluctuation regularity. Clustering is performed based on the current characteristics and fluctuation characteristics of all suspected current anomalies in the current time series curve to obtain different clusters; the current anomaly degree of the cluster is obtained based on the number of suspected current anomalies in the cluster and the distance characteristics in the cluster space; and the current anomaly moment is obtained based on the current anomaly degree. Fault characteristic values ​​are obtained based on the distribution characteristics of abnormal current moments and the variation characteristics of current anomaly degree in the current time series curve; circuit fault identification is performed based on the fault characteristic values. The step of obtaining the degree of fluctuation anomaly at the fluctuation point based on the amplitude and duration characteristics of the current fluctuation in the current time series curve includes: The extreme points in the current time series curve are taken as fluctuation points; the absolute value of the rate of change of the slope of the data points in the current time series curve is calculated and normalized to obtain the change characteristic value of the data points; the data points whose change characteristic value exceeds the preset inflection point threshold are taken as inflection points; the average value of the absolute value of the difference between the fluctuation point and the adjacent inflection points in the current time series curve is calculated to obtain the fluctuation amplitude of the fluctuation point; the ratio of the fluctuation amplitude to the maximum value among all fluctuation amplitudes is calculated to obtain the amplitude characteristic value; the time interval between the adjacent inflection points of the fluctuation point in the current time series curve is calculated to obtain the fluctuation duration of the fluctuation point; the ratio of the fluctuation duration to the maximum value among all fluctuation durations is calculated and negatively correlated to obtain the duration characteristic value; the sum of the amplitude characteristic value and the duration characteristic value is calculated to obtain the fluctuation anomaly degree of the fluctuation point.

2. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining the moment of fluctuation anomaly based on the degree of fluctuation anomaly includes: When the degree of fluctuation abnormality of the fluctuation point exceeds the preset fluctuation threshold, the time when the fluctuation point is located is the time of fluctuation abnormality.

3. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining the degree of fluctuation regularity based on the difference in the degree of fluctuation anomaly between the time of the fluctuation anomaly and the historical time of the same anomaly includes: Calculate the average of the absolute values ​​of the differences between the abnormal fluctuation time and the fluctuation anomaly degree of all historical times of the same time, and perform a negative correlation mapping to obtain the fluctuation regularity degree of the abnormal fluctuation time.

4. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining the suspected moment of current anomaly based on the degree of fluctuation pattern includes: When the degree of fluctuation regularity at the time of the abnormal fluctuation does not exceed a preset regularity threshold, the time of the abnormal fluctuation is considered a suspected current abnormality.

5. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of clustering based on the current characteristics and fluctuation characteristics of all suspected current anomalies in the current time series curve to obtain different clusters includes: Based on the current value, fluctuation amplitude, and fluctuation duration corresponding to all suspected current anomalies, different clusters are obtained by using the K-means clustering algorithm.

6. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining the current anomaly degree of the cluster based on the quantitative characteristics of suspected current anomalies in the cluster and the distance characteristics in the cluster space includes: In the cluster space, calculate the average Euclidean distance between all suspected current anomaly moments in the cluster and their nearest neighboring suspected current anomaly moments to obtain the discrete distribution value; calculate the product of the reciprocal of the number of suspected current anomaly moments in the cluster and the discrete distribution value, and normalize it to obtain the current anomaly degree of the cluster.

7. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining the time of current anomaly based on the current anomaly degree includes: The suspected current anomaly moments in the clusters where the current anomaly exceeds the preset anomaly threshold are defined as current anomaly moments.

8. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of obtaining fault characteristic values ​​based on the distribution characteristics of current anomalies and the variation characteristics of current anomalies in the current time-series curve includes: The current time-series curve is divided into a first half and a second half from the middle time point; the ratio of the number of current abnormal moments in the second half to the number of current abnormal moments in the first half is calculated to obtain a first fault trend value; the ratio of the average current abnormality degree corresponding to the current abnormal moments in the second half to the average current abnormality degree in the first half is calculated to obtain a second fault trend value; the product of the first fault trend value and the second fault trend value is calculated to obtain the fault characteristic value.

9. The intelligent sensing and identification method for circuit faults in a semiconductor hybrid solid-state circuit breaker according to claim 1, characterized in that, The step of identifying circuit faults based on the fault characteristic values ​​includes: When the fault characteristic value exceeds the preset fault threshold, a circuit fault warning is issued.

Images