Power distribution network power supply fault diagnosis and analysis system based on artificial intelligence

Abstract

This invention relates to an artificial intelligence-based power distribution network fault diagnosis and analysis system, belonging to the field of transformer data processing technology. The system first determines the inter-turn short-circuit characteristic values ​​of each winding at each time point based on the turns ratio of each winding, and then determines the corresponding inter-turn short-circuit significance after comparison. Next, based on the inter-turn short-circuit significance values ​​of any winding over a historical period, it identifies target segments of that winding that exhibit continuous evolution characteristics of inter-turn short-circuit features over that historical period. Then, based on the stability and magnitude of the inter-turn short-circuit significance of the target segments and the length of the target segments, it determines the anomaly coefficient of the target segments. The anomaly coefficients are used to cluster all target segments, and the clustering results are used to complete the early warning of transformer inter-turn short-circuit faults. This invention can provide more accurate, reliable, and timely early warnings of potential inter-turn short-circuit faults in transformers under fluctuating and highly disturbed power grid operating environments, improving the reliability and stability of power distribution networks.

Description

Technical Field

[0001] This invention relates to the field of transformer data processing technology, and in particular to an artificial intelligence-based power distribution network fault diagnosis and analysis system. Background Technology

[0002] As a crucial component of the power system directly responsible for power distribution and supply, the power distribution network's operational safety and supply reliability directly impact the power quality and stability for end users. Among these components, the distribution transformer, a key device for voltage transformation and energy distribution within the distribution network, operates under conditions of frequent load fluctuations and complex environments; therefore, its operational reliability is critical to the overall safety and reliability of the distribution network.

[0003] In actual operation of distribution networks, inter-turn short-circuit faults in distribution transformer windings are among the most common faults. These faults are typically characterized by their insidious initial occurrence, slow characteristic changes, and gradual evolution, making it difficult to trigger traditional protection devices or diagnostic rules based on fixed thresholds. Existing distribution network monitoring and analysis systems primarily focus on post-fault assessment and handling, lacking the ability to continuously monitor and comprehensively analyze the evolution of inter-turn faults in transformer windings. This makes it difficult to effectively identify and warn of risks before the fault develops to the stage of protection action or severe damage, easily leading to delayed fault detection or insufficient diagnostic accuracy, thus affecting the safety and reliability of the distribution network's power supply.

[0004] In other words, the current power distribution network faces the technical problem of being unable to meet the reliable early warning requirements for inter-turn short-circuit faults in transformer windings. Summary of the Invention

[0005] In view of this, the present invention provides an artificial intelligence-based power distribution network fault diagnosis and analysis system to solve the technical problem that the current power distribution network is unable to meet the reliable early warning requirements for inter-turn short circuit faults in transformer windings.

[0006] The present invention provides an artificial intelligence-based power distribution network fault diagnosis and analysis system, comprising:

[0007] The inter-turn short-circuit characteristic determination module is used to continuously collect the electrical parameters of each phase winding of the transformer during the operation of the distribution network, calculate the equivalent turns ratio at any time based on the electrical parameters of any phase winding at any time, and determine the inter-turn short-circuit characteristic value of any phase winding at any time based on the difference between the equivalent turns ratio and the rated turns ratio.

[0008] The target segment determination module is used to determine the inter-turn short-circuit significance of any phase winding at any time based on the magnitude of the inter-turn short-circuit characteristic value of any phase winding at any time compared with the inter-turn short-circuit characteristic values ​​of each phase winding at the same time, and to determine each target segment corresponding to any phase winding based on the inter-turn short-circuit significance in the historical period before the current time.

[0009] The reference weight determination module is used to determine the reference weight of any target segment in other phase windings relative to the current target segment based on the time distance between the current target segment of any phase winding and the previous target segment of other phase windings.

[0010] An anomaly coefficient determination module is used to determine the anomaly coefficient of the current target segment of any phase winding based on the stability difference of the change in the inter-turn short circuit significance of the current target segment of any phase winding and the time difference of the target segments of other phase windings, the mean difference of the inter-turn short circuit significance, the segment length difference, and the reference weight.

[0011] The fault diagnosis module is used to cluster all target segments according to the anomaly coefficient, and to complete the inter-turn short-circuit fault early warning of the transformer based on the clustering results.

[0012] Furthermore, the electrical parameters include the input voltage and input current of each phase primary winding in the transformer, and the output voltage and output current of each phase secondary winding in the transformer.

[0013] Furthermore, the calculation of the equivalent turns ratio at any given moment based on the electrical parameters of any phase winding includes:

[0014] The voltage equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input voltage to the output voltage of any phase winding at any given time, and the current equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input current to the output current of any phase winding at any given time.

[0015] Furthermore, determining the inter-turn short-circuit characteristic value of any phase winding at any time based on the difference between the equivalent turns ratio and the rated turns ratio includes:

[0016] The absolute value of the difference between the average of the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time and the rated turns ratio is recorded as the first inter-turn short circuit characterization item, and the absolute value of the difference between the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time is recorded as the second inter-turn short circuit characterization item.

[0017] The inter-turn short-circuit characteristic value of any phase winding at any time is constructed based on the first inter-turn short-circuit characterization term and the second inter-turn short-circuit characterization term. The inter-turn short-circuit characteristic value of any phase winding at any time is proportional to the first inter-turn short-circuit characterization term and inversely proportional to the second inter-turn short-circuit characterization term.

[0018] Furthermore, determining the significance of the inter-turn short circuit of any phase winding at that moment includes:

[0019] The ratio of the inter-turn short-circuit characteristic value of any phase winding at any time to the minimum inter-turn short-circuit characteristic value of all phase windings at any time is taken as the inter-turn short-circuit significance of any phase winding at any time.

[0020] Furthermore, determining each target segment corresponding to any phase winding includes:

[0021] The inter-turn short circuit significance at any time in the target segment is greater than 1 and not less than the inter-turn short circuit significance at the previous time, while the target segment is greater than the preset segment length.

[0022] Furthermore, determining the reference weight of any previous target segment in other phase windings relative to the current target segment includes:

[0023] The time difference between any time-preceding target segment in the other phase windings and the current target segment in any phase winding is calculated as the time distance corresponding to any time-preceding target segment in the other phase windings. The proportion of the time distance corresponding to any time-preceding target segment in the other phase windings to the sum of the time distances corresponding to all time-preceding target segments in the other phase windings is used as the reference weight of any time-preceding target segment in the other phase windings relative to the current target segment.

[0024] Furthermore, determining the stability of the change in the significance of the inter-turn short circuit includes:

[0025] After normalizing the inter-turn short-circuit significance sequence under any target segment, the corresponding second-order difference sequence is calculated. The coefficient of variation of the second-order difference sequence is calculated, and the stability of the change of the inter-turn short-circuit significance corresponding to any target segment is determined based on the coefficient of variation. The stability of the change of the inter-turn short-circuit significance corresponding to any target segment is inversely proportional to the coefficient of variation.

[0026] Furthermore, the anomaly coefficient of the current target segment of any phase winding is:

[0027]

[0028] in, This represents the anomaly coefficient of the i-th target segment of any phase winding. This represents the total number of time segments preceding the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit corresponding to the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit at time j of the i-th target segment in the preceding target segment. Let represent the mean significance of the inter-turn short circuit of the i-th target segment. This represents the mean significance of the inter-turn short circuit in the preceding target segment at the j-th time of the i-th target segment. This represents the length of the i-th target segment. This represents the length of the segment whose time preceding the i-th target segment is j-th. This represents the reference weight of the i-th target segment relative to the j-th target segment at the preceding time.

[0029] Furthermore, the step of completing the inter-turn short-circuit fault early warning for the transformer based on the clustering results includes:

[0030] Determine the proportion of the target segment corresponding to each phase winding in any cluster among all target segments in that cluster. Use the normalized value of the product of the maximum value of the proportion and the mean of the anomaly coefficients of all target segments in any cluster as the warning value of any cluster. When the warning value is greater than a preset warning threshold, provide a turn-to-turn short circuit fault warning for the winding phase corresponding to the maximum value of the proportion.

[0031] The advantages of this invention compared to the prior art are:

[0032] This invention first obtains the inter-turn short-circuit characteristic value of any winding at any time based on the turns ratio of each winding. Then, it further obtains the inter-turn short-circuit significance of any winding at any time. Next, based on the time series sequence of inter-turn short-circuit significance of any winding over a historical period, it determines the target segment corresponding to that winding with continuously evolving inter-turn short-circuit characteristics over that historical period and obtains the stability characteristics of the target segment. Based on the stability characteristics, the length of the target segment, and the magnitude of the inter-turn short-circuit significance within it, the anomaly coefficient of the target segment is obtained. Finally, based on the anomaly coefficient, all target segments are clustered, and the clustering results are used to provide early warning for the phase winding with the most likely inter-turn short-circuit fault corresponding to each cluster. This invention, through analysis, extraction, and modeling of inter-turn short-circuit related data and features, can provide more accurate, reliable, and timely early warning for potential inter-turn short-circuit faults in transformers under fluctuating and highly disturbed power grid operating environments, thereby improving the reliability and stability of power grid operation. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a structural block diagram of a power distribution network fault diagnosis and analysis system based on artificial intelligence, provided in Embodiment 1 of the present invention. Detailed Implementation

[0035] To further illustrate the technical solution of the present invention, specific embodiments are described below.

[0036] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a particular feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. Furthermore, a particular feature, structure, or characteristic in one or more embodiments may be combined in any suitable form, and the terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless otherwise specifically emphasized.

[0037] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0038] System Implementation Example:

[0039] See Figure 1 This is a structural block diagram of an artificial intelligence-based power distribution network fault diagnosis and analysis system provided in Embodiment 1 of the present invention, as shown below. Figure 1 As shown, the system includes an inter-turn short-circuit characteristic determination module 11, a target segment determination module 12, a reference weight determination module 13, an anomaly coefficient determination module 14, and a fault diagnosis module 15. The following is a detailed description of each module:

[0040] The inter-turn short-circuit characteristic determination module 11 is used to continuously collect the electrical parameters of each phase winding of the transformer during the operation of the distribution network, calculate the equivalent turns ratio at any time based on the electrical parameters of any phase winding at any time, and determine the inter-turn short-circuit characteristic value of any phase winding at any time based on the difference between the equivalent turns ratio and the rated turns ratio.

[0041] In power distribution networks, three-phase distribution transformers, as key power supply equipment, play a crucial role in power transformation and distribution. Their three-phase windings operate continuously under high voltage, high current, and high temperature conditions, making the risk of inter-turn short circuits in the windings particularly prominent. Therefore, this embodiment focuses on three-phase distribution transformers operating in a power distribution network scenario, monitoring and analyzing their inter-turn short circuit characteristics.

[0042] To monitor and analyze the transformer, it is first necessary to continuously collect the electrical parameters of each phase winding during the operation of the power distribution network. These electrical parameters include the input voltage and input current of each phase primary winding, and the output voltage and output current of each phase secondary winding. The primary and secondary windings of each phase are also known as the primary and secondary windings of each phase.

[0043] Therefore, according to the basic electromagnetic relationship of a transformer, under good structural conditions, the primary and secondary windings of any phase winding of a transformer satisfy the following proportional relationship:

[0044]

[0045] in, , These represent the input voltage and input current at the primary winding terminals of any phase winding, respectively. , These represent the output voltage and output current at the secondary winding terminals of the phase winding, respectively. , These represent the number of turns in the primary and secondary windings of the phase winding, respectively. This refers to the rated turns ratio of that phase winding. Since the rated turns ratio of each phase winding is the same when the transformer is in good working order, therefore... This is also the rated turns ratio of the transformer. Under normal operating conditions, since the number of winding turns is a structural constant, the above ratio remains relatively stable over time.

[0046] When an inter-turn short circuit occurs in a transformer winding, because this fault is a structural damage, the equivalent turns ratio of the phase winding affected by the short circuit usually remains stable or exhibits a slow, evolving characteristic over a certain period. When an inter-turn short circuit occurs, some winding turns are short-circuited, which is equivalent to a change in the effective number of turns participating in electromagnetic coupling, causing the equivalent turns ratio calculated based on operating voltage and current to deviate from its rated value. This deviation is particularly noticeable in single-phase windings and can lead to an imbalance between the three phases. However, load fluctuations during operation can also cause temporary changes in the voltage-current ratio. Therefore, it is necessary to construct inter-turn short circuit characteristic values ​​that can distinguish between inter-turn short circuits and load disturbances, in order to quantify the characteristics of inter-turn short circuits while avoiding misidentification of load disturbances.

[0047] To calculate the inter-turn short-circuit characteristic value, it is first necessary to calculate the equivalent turns ratio at any given time using the electrical parameters of any phase winding, including:

[0048] The voltage equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input voltage to the output voltage of any phase winding at any given time, and the current equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input current to the output current of any phase winding at any given time.

[0049] Then, based on the difference between the equivalent turns ratio and the rated turns ratio, the inter-turn short-circuit characteristic value of any phase winding at any time is determined, including:

[0050] The absolute value of the difference between the average of the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time and the rated turns ratio is recorded as the first inter-turn short circuit characterization item, and the absolute value of the difference between the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time is recorded as the second inter-turn short circuit characterization item.

[0051] The inter-turn short-circuit characteristic value of any phase winding at any time is constructed based on the first inter-turn short-circuit characterization term and the second inter-turn short-circuit characterization term. The inter-turn short-circuit characteristic value of any phase winding at any time is proportional to the first inter-turn short-circuit characterization term and inversely proportional to the second inter-turn short-circuit characterization term.

[0052] As a further preferred option, the characteristic value of inter-turn short circuit is:

[0053]

[0054] in, This represents the inter-turn short-circuit characteristic value of any phase winding at time t. This represents the voltage equivalent turns ratio of the phase winding at time t. This represents the equivalent turns ratio of the current in the phase winding at time t. The rated turns ratio of this phase winding is given by , and e is the natural constant. This indicates taking the absolute value.

[0055] In the formula, This represents the absolute value of the difference between the mean and rated turns ratio of the voltage and current equivalent turns ratio of the analyzed winding at time t. This value characterizes the degree of deviation of the equivalent turns ratio from the rated turns ratio under operating conditions. The larger the value, the more significant the change in the equivalent turns ratio of the winding, and the higher the probability of an inter-turn short circuit. Furthermore, since inter-turn short circuits usually have a synchronous effect on voltage and current, when... When it is smaller, that is The larger the value, the more consistent the voltage and current ratio deviation is. This characteristic is more consistent with the physical mechanism of inter-turn short circuit, rather than the load disturbance characteristic. Therefore, the larger the inter-turn short circuit characteristic value is.

[0056] The target segment determination module 12 is used to determine the inter-turn short-circuit significance of any phase winding at any time based on the magnitude of the inter-turn short-circuit characteristic value of any phase winding at any time compared with the inter-turn short-circuit characteristic values ​​of each phase winding at the same time, and to determine each target segment corresponding to any phase winding based on the inter-turn short-circuit significance in the historical period before the current time.

[0057] Since a three-phase distribution transformer consists of three sets of independent but electromagnetically coupled windings, when an inter-turn short circuit occurs in one phase winding, its corresponding inter-turn short circuit characteristic value is usually more prominent than that of the other two phases. Therefore, the three phase windings are labeled as phase a, phase b, and phase c, and taking phase a as an example, its inter-turn short circuit significance is constructed, calculated as follows:

[0058]

[0059] in, This indicates the significance of the inter-turn short circuit in phase a winding at time t. This represents the inter-turn short-circuit characteristic value of phase a winding at time t. This represents the inter-turn short-circuit characteristic value of phase b winding at time t. This represents the inter-turn short-circuit characteristic value of the c-phase winding at time t. This represents the function that takes the minimum value. This represents the ratio of the inter-turn short-circuit significance of phase a winding at time t to the minimum inter-turn short-circuit significance of the three phases at the same time. The larger this value is, the more significant the inter-turn short-circuit characteristic value of that phase at that time is, that is, the greater the inter-turn short-circuit significance is.

[0060] Through the above steps, the inter-turn short-circuit significance of each phase winding can be calculated separately in the time dimension. Based on the selected historical time period before the current time, a time-series characteristic sequence corresponding to each phase winding is formed within the historical time period. The length of the historical time period can be set according to a comprehensive consideration of fault diagnosis efficiency and fault diagnosis accuracy, such as setting its length to 1 hour. For the inter-turn short-circuit significance sequence of any phase, the time points with values ​​greater than 1 are further screened. Since when this value is greater than 1, it indicates that the inter-turn short-circuit characteristic value of that phase winding is more prominent than that of the other two phases at that time, such a time can be regarded as a potential anomaly candidate time for further analysis.

[0061] Based on this, to determine the continuous evolution characteristics of inter-turn short-circuit feature values ​​over time, the inter-turn short-circuit significance sequence is segmented. Specifically, the inter-turn short-circuit significance at the current time is compared with that at the previous time. If the inter-turn short-circuit significance at the current time is greater than or equal to that at the previous time, then that time is marked as 0; otherwise, it is marked as −1. In this way, the inter-turn short-circuit significance sequence corresponding to each phase winding can be divided into multiple time segments composed of consecutive markers. Segments consecutively marked as 0 indicate that the inter-turn short-circuit feature has continuous evolution characteristics within that time interval, while segments consecutively marked as −1 indicate that the feature does not have continuous evolution characteristics.

[0062] Furthermore, analysis is only performed on segments continuously marked as 0 and with a segment length greater than a preset segment length (e.g., more than 3 sampling points or sampling times). This is because excessively short continuous segments may be caused by random disturbances such as grid load fluctuations, making it difficult to reflect the true evolution characteristics of inter-turn short circuits. For continuous segments that meet the conditions, they are recorded as target segments, and their change stability is further evaluated. The longer the duration and the more stable the change, the higher the reliability of the inter-turn short circuit characteristics corresponding to that segment.

[0063] From the above summary of the specific process for determining the target segments, it can be seen that determining each target segment corresponding to any phase winding includes:

[0064] The inter-turn short circuit significance at any time in the target segment is greater than 1 and not less than the inter-turn short circuit significance at the previous time, while the target segment is greater than the preset segment length.

[0065] The reference weight determination module 13 is used to determine the reference weight of any target segment in other phase windings that is ahead in time to the current target segment based on the time distance between the current target segment of any phase winding and the target segment of other phase windings.

[0066] Considering that the insulation performance of the windings of a distribution transformer gradually degrades over time during operation, the risk of inter-turn short circuits usually increases over time. Therefore, it is necessary to compare and analyze the current target segment with the historical operating status to quantify the possible degree of inter-turn short circuit anomalies.

[0067] Furthermore, since the target segment with earlier operating time is less likely to experience inter-turn short circuits, its operating state is closer to the healthy baseline state; at the same time, if only historical segments of the same phase are used as comparison objects, abnormal features in the same phase may be masked by each other due to their potential aging phenomena, which is not conducive to the identification of early anomalies.

[0068] Therefore, this embodiment introduces a cross-phase comparison mechanism, which compares and analyzes any target segment of the current phase with historical target segments in other phases that are two times earlier than the target segment. This improves the comparability between different phases and helps to identify the unbalanced characteristics of the three-phase winding operation state, enabling early detection of potential anomalies.

[0069] Based on this, for any target segment in any phase, the time distance between it and historical target segments in other phases that are earlier in time is introduced as a weighting factor. The larger the time distance, the earlier the historical target segment is compared to the current target segment, the higher its reliability as a health reference benchmark, and the greater its weight in comparative analysis. Therefore, the reference weight of any earlier target segment in other phase windings relative to the current target segment is determined, including:

[0070] The time difference between any time-preceding target segment in the other phase windings and the current target segment in any phase winding is calculated as the time distance corresponding to any time-preceding target segment in the other phase windings. The proportion of the time distance corresponding to any time-preceding target segment in the other phase windings to the sum of the time distances corresponding to all time-preceding target segments in the other phase windings is used as the reference weight of any time-preceding target segment in the other phase windings relative to the current target segment.

[0071] The formula for the reference weights is as follows:

[0072]

[0073] in, This represents the reference weight of the j-th target segment corresponding to other phase windings in earlier times relative to the i-th target segment corresponding to the current phase winding. This represents the time difference between the j-th target segment corresponding to other phase windings that has an earlier time and the i-th target segment corresponding to the current phase winding. In other words, it represents the time distance corresponding to the j-th target segment corresponding to other phase windings that has an earlier time. The larger the time distance, the greater the reference weight of the earlier target segment to the current target segment. This represents the total number of time preceding target segments corresponding to the i-th target segment of the current phase winding.

[0074] Among them, it is preferable to set the sequential relationship between different target segments, and to make a judgment by comparing the initial time corresponding to each target segment.

[0075] The anomaly coefficient determination module 14 is used to determine the anomaly coefficient of the current target segment of any phase winding based on the difference in stability of the change of the inter-turn short circuit significance of the current target segment of any phase winding and the time-preceding target segment of other phase windings, the difference in the mean of the inter-turn short circuit significance, the difference in segment length, and the reference weight.

[0076] To quantitatively describe the stability of changes in the target segment, the inter-turn short-circuit significance sequence for each target segment is linearly normalized, and its second-order difference sequence is calculated. Based on this second-order difference sequence, the stability of changes in the inter-turn short-circuit significance is determined, including:

[0077] After normalizing the inter-turn short-circuit significance sequence under any target segment, the corresponding second-order difference sequence is calculated. The coefficient of variation of the second-order difference sequence is calculated, and the stability of the change of the inter-turn short-circuit significance corresponding to any target segment is determined based on the coefficient of variation. The stability of the change of the inter-turn short-circuit significance corresponding to any target segment is inversely proportional to the coefficient of variation.

[0078] Furthermore, as a preferred embodiment, the stability of the change in the significance of the inter-turn short circuit is:

[0079] Where represents the stability of the change in the inter-turn short-circuit significance of any target segment corresponding to any phase winding, e is the natural constant, and X represents the second-order difference sequence calculated after normalizing the inter-turn short-circuit significance sequence of any target segment corresponding to any phase winding. This represents the coefficient of variation of the second-order difference sequence. The coefficient of variation is used to characterize the degree of fluctuation in the sequence; the smaller the value, the more stable the change in that segment, and the greater the corresponding stability W.

[0080] Using the above method, the stability of the change in the significance of inter-turn short circuits for each target segment under any phase winding can be obtained in the time dimension.

[0081] Then, based on the stability difference of the change in the inter-turn short-circuit significance of the current target segment of any phase winding compared with the previous target segments of other phase windings, the mean difference of the inter-turn short-circuit significance, the segment length difference, and the reference weight, the anomaly coefficient of the current target segment of any phase winding can be determined:

[0082]

[0083] in, This represents the anomaly coefficient of the i-th target segment of any phase winding. This represents the total number of time segments preceding the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit corresponding to the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit at time j of the i-th target segment in the preceding target segment. The larger the value, the stronger the stability of the change in the significance of inter-turn short circuits corresponding to the i-th target segment, the greater the probability of inter-turn short circuits existing under that target segment, and the larger the anomaly coefficient. Let represent the mean significance of the inter-turn short circuit of the i-th target segment. This represents the mean significance of the inter-turn short circuit in the preceding target segment at the j-th time of the i-th target segment. The larger the value, the higher the significance of the inter-turn short circuit corresponding to the i-th target segment, the greater the probability of an inter-turn short circuit existing under that target segment, and the larger the anomaly coefficient. This represents the length of the i-th target segment. This represents the length of the segment whose time preceding the i-th target segment is j-th. This represents the reference weight of the i-th target segment at time j relative to the i-th target segment from the preceding target segment. The larger the value, the stronger the relative persistence of the significance of the inter-turn short circuit corresponding to the i-th target segment, the greater the possibility of an inter-turn short circuit under this target segment, and the larger the anomaly coefficient.

[0084] The fault diagnosis module 15 is used to cluster all target segments according to the anomaly coefficient, and to complete the inter-turn short circuit fault early warning of the transformer based on the clustering results.

[0085] After obtaining the anomaly coefficient of any target segment of any phase winding, directly using the anomaly coefficient of a single target segment as the warning threshold may have limitations in providing false warnings. The main reason is that due to load fluctuations, grid impacts, or short-term changes in operating conditions, the anomaly coefficient of a single target segment may temporarily spike, failing to reflect the true anomaly risk of continuous segments or the overall pattern. To improve the reliability of the judgment and identify anomaly patterns, this embodiment selects to perform cluster analysis on all target segments. Through clustering, target segments with similar or relatively concentrated anomaly probabilities can be automatically divided into different clusters, thereby enabling the extraction of inter-turn short-circuit anomaly patterns. Statistical analysis of each cluster yields the average anomaly coefficient, sample size, and proportion of target segments from the same phase for each cluster. When the anomaly coefficient of a certain cluster is generally high, and the proportion of target segments under a certain phase winding is significant, it can be determined that there is a potential inter-turn short-circuit risk in that phase winding, serving as a basis for early warning or further analysis. Therefore, based on the clustering results, the inter-turn short-circuit fault warning for the transformer is completed, including:

[0086] Determine the proportion of the target segment corresponding to each phase winding in any cluster among all target segments in that cluster. Use the normalized value of the product of the maximum value of the proportion and the mean of the anomaly coefficients of all target segments in any cluster as the warning value of any cluster. When the warning value is greater than a preset warning threshold, provide a turn-to-turn short circuit fault warning for the winding phase corresponding to the maximum value of the proportion.

[0087] The specific implementation steps for inter-turn short-circuit fault early warning based on clustering results are as follows:

[0088] First, the anomaly coefficients and phase information of each target segment are used to construct clustering input data. Second, the density-based HDBSCAN clustering algorithm is used for analysis, where the key parameters min_cluster_size and min_samples are both set to 3 (empirical value) to ensure that at least 3 samples form a cluster or neighborhood sample for density estimation, while avoiding the interference of isolated noise on the clustering results, thereby realizing the automatic division of different anomaly pattern clusters.

[0089] Based on the above logic, and according to the clustering results, the warning value for any cluster is calculated:

[0090]

[0091] in, This represents the warning value for the r-th cluster. Represents the normalization function. This represents the mean of the anomaly coefficients of all target segments in the r-th cluster. The larger this value is, the greater the possibility that the cluster may have inter-turn short circuits. This represents the maximum percentage of the target segment corresponding to each phase winding in the r-th cluster among all target segments in that cluster. The larger this value is, the greater the warning value of the r-th cluster.

[0092] Therefore, the warning value corresponding to each cluster can be calculated. Furthermore, a warning threshold T is set (in this embodiment, T=0.7 is preferred, but it can be adjusted based on operational experience or historical statistical results in actual applications). When the warning value of a certain cluster... When the value exceeds the preset warning threshold T, it indicates that the cluster meets the warning judgment conditions in terms of anomaly probability intensity and phase concentration. Therefore, the winding corresponding to the phase with the largest proportion of target segments in the cluster is determined to have a potential inter-turn short circuit risk, triggering the inter-turn short circuit fault warning mechanism for that phase winding. Through this method, early identification and warning of inter-turn short circuit faults can be achieved before they develop into obvious electrical anomalies, providing maintenance personnel with timely intervention basis, reducing the operational risks of the power distribution system, and improving the safety and reliability of transformer operation.

[0093] In the process of monitoring and early warning of potential inter-turn short-circuit faults in three-phase transformers, this invention first obtains the inter-turn short-circuit characteristic value of any winding at any time based on the turns ratio of each winding. Then, it further obtains the inter-turn short-circuit significance of any winding at any time. Next, based on the time series sequence of inter-turn short-circuit significance of any winding over a historical period, it determines the target segment corresponding to that winding with the continuous evolution characteristic of inter-turn short-circuit features over the historical period and obtains the stability characteristics of the target segment. Based on the stability characteristics, the length of the target segment, and the magnitude of the inter-turn short-circuit significance, the anomaly coefficient of the target segment is obtained. Finally, based on the anomaly coefficient, all target segments are clustered, and the clustering results are used to complete the early warning of the phase winding with the most likely inter-turn short-circuit fault corresponding to each cluster. This invention, through analysis, extraction, and modeling of inter-turn short-circuit related data and features, can provide more accurate, reliable, and timely early warning of potential inter-turn short-circuit faults in transformers under fluctuating and highly disturbed power grid operating environments, thereby improving the reliability and stability of power grid operation.

[0094] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A power distribution network fault diagnosis and analysis system based on artificial intelligence, characterized in that, include: The inter-turn short-circuit characteristic determination module is used to continuously collect the electrical parameters of each phase winding of the transformer during the operation of the distribution network, calculate the equivalent turns ratio at any time based on the electrical parameters of any phase winding at any time, and determine the inter-turn short-circuit characteristic value of any phase winding at any time based on the difference between the equivalent turns ratio and the rated turns ratio. The target segment determination module is used to determine the inter-turn short-circuit significance of any phase winding at any time based on the magnitude of the inter-turn short-circuit characteristic value of any phase winding at any time compared with the inter-turn short-circuit characteristic values ​​of each phase winding at the same time, and to determine each target segment corresponding to any phase winding based on the inter-turn short-circuit significance in the historical period before the current time. The reference weight determination module is used to determine the reference weight of any target segment in other phase windings relative to the current target segment based on the time distance between the current target segment of any phase winding and the previous target segment of other phase windings. An anomaly coefficient determination module is used to determine the anomaly coefficient of the current target segment of any phase winding based on the stability difference of the change in the inter-turn short circuit significance of the current target segment of any phase winding and the time difference of the target segments of other phase windings, the mean difference of the inter-turn short circuit significance, the segment length difference, and the reference weight. The fault diagnosis module is used to cluster all target segments according to the anomaly coefficient, and to complete the inter-turn short-circuit fault early warning of the transformer based on the clustering results.

2. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 1, characterized in that, The electrical parameters include the input voltage and input current of each phase primary winding in the transformer, and the output voltage and output current of each phase secondary winding in the transformer.

3. The artificial intelligence-based power distribution network fault diagnosis and analysis system according to claim 1 or 2, characterized in that, The calculation of the equivalent turns ratio at any given moment based on the electrical parameters of any phase winding includes: The voltage equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input voltage to the output voltage of any phase winding at any given time, and the current equivalent turns ratio of any phase winding at any given time is determined by the ratio of the input current to the output current of any phase winding at any given time.

4. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 3, characterized in that, Determining the inter-turn short-circuit characteristic value of any phase winding at any time based on the difference between the equivalent turns ratio and the rated turns ratio includes: The absolute value of the difference between the average of the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time and the rated turns ratio is recorded as the first inter-turn short circuit characterization item, and the absolute value of the difference between the voltage equivalent turns ratio and the current equivalent turns ratio of any phase winding at any time is recorded as the second inter-turn short circuit characterization item. The inter-turn short-circuit characteristic value of any phase winding at any time is constructed based on the first inter-turn short-circuit characterization term and the second inter-turn short-circuit characterization term. The inter-turn short-circuit characteristic value of any phase winding at any time is proportional to the first inter-turn short-circuit characterization term and inversely proportional to the second inter-turn short-circuit characterization term.

5. The artificial intelligence-based power distribution network fault diagnosis and analysis system according to claim 1 or 4, characterized in that, Determining the significance of inter-turn short circuit in any phase winding at that moment includes: The ratio of the inter-turn short-circuit characteristic value of any phase winding at any time to the minimum inter-turn short-circuit characteristic value of all phase windings at any time is taken as the inter-turn short-circuit significance of any phase winding at any time.

6. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 1, characterized in that, The determination of each target segment corresponding to any phase winding includes: The inter-turn short circuit significance at any time in the target segment is greater than 1 and not less than the inter-turn short circuit significance at the previous time, while the target segment is greater than the preset segment length.

7. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 1, characterized in that, Determining the reference weight of any previous target segment in other phase windings relative to the current target segment includes: The time difference between any time-preceding target segment in the other phase windings and the current target segment in any phase winding is calculated as the time distance corresponding to any time-preceding target segment in the other phase windings. The proportion of the time distance corresponding to any time-preceding target segment in the other phase windings to the sum of the time distances corresponding to all time-preceding target segments in the other phase windings is used as the reference weight of any time-preceding target segment in the other phase windings relative to the current target segment.

8. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 1, characterized in that, Determining the stability of the change in the significance of the inter-turn short circuit includes: After normalizing the inter-turn short-circuit significance sequence under any target segment, the corresponding second-order difference sequence is calculated. The coefficient of variation of the second-order difference sequence is calculated, and the stability of the change of the inter-turn short-circuit significance corresponding to any target segment is determined based on the coefficient of variation. The stability of the change of the inter-turn short-circuit significance corresponding to any target segment is inversely proportional to the coefficient of variation.

9. The artificial intelligence-based power distribution network fault diagnosis and analysis system according to claim 1 or 8, characterized in that, The anomaly coefficient of the current target segment of any phase winding is: in, This represents the anomaly coefficient of the i-th target segment of any phase winding. This represents the total number of time segments preceding the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit corresponding to the i-th target segment. This represents the stability of the change in the significance of the inter-turn short circuit at time j of the i-th target segment in the preceding target segment. Let represent the mean significance of the inter-turn short circuit of the i-th target segment. This represents the mean significance of the inter-turn short circuit in the preceding target segment at the j-th time of the i-th target segment. This represents the length of the i-th target segment. This represents the length of the segment whose time preceding the i-th target segment is j-th. This represents the reference weight of the i-th target segment relative to the j-th target segment at the preceding time.

10. The power distribution network fault diagnosis and analysis system based on artificial intelligence according to claim 1, characterized in that, The step of providing early warning of inter-turn short-circuit faults in the transformer based on clustering results includes: Determine the proportion of the target segment corresponding to each phase winding in any cluster among all target segments in that cluster. Use the normalized value of the product of the maximum value of the proportion and the mean of the anomaly coefficients of all target segments in any cluster as the warning value of any cluster. When the warning value is greater than a preset warning threshold, provide a turn-to-turn short circuit fault warning for the winding phase corresponding to the maximum value of the proportion.

Images