Power distribution network user voltage anomaly diagnosis method and system

Abstract

The application provides a power distribution network user voltage anomaly diagnosis method and system, the method comprises the following steps: when a low voltage event occurs, extracting basic features and electrical features; under a first preset rule, obtaining an event type corresponding to the low voltage event according to at least one key factor; under a second preset rule, diagnosing the cause of the low voltage event, and marking the cause diagnosis result based on a third preset rule; constructing a final evaluation model; inputting the basic features and the electrical features into the final evaluation model according to the primary marking result, to obtain a first diagnosis result corresponding to the primary marking result; under a fourth preset rule, checking and fusing the first diagnosis result and the cause diagnosis result, to output a diagnosis report about the low voltage event according to the checking and fusing result. The application can improve the efficiency and accuracy of low voltage diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of low voltage diagnosis in distribution networks, and in particular to a method and system for diagnosing abnormal voltage among users of distribution networks. Background Art

[0002] In traditional power systems, monitoring and diagnosis of low voltage issues at users is relatively simple. This primarily relies on voltage monitoring at fixed substations and a small number of key nodes, generating alarms based on fixed thresholds. Cause analysis relies heavily on the experience and manual inspections of operations and maintenance personnel, resulting in low diagnostic efficiency and accuracy, and difficulty addressing complex and hidden faults. Furthermore, the diagnostic data source is limited, primarily steady-state operating data collected by the SCADA system. This lack of detailed user-side and distribution transformer-side measurement information further reduces diagnostic accuracy. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for diagnosing voltage anomalies among users of a distribution network, aiming to solve the problems of accuracy, real-timeness and interpretability of the cause analysis of low voltage among users in the context of new power systems, thereby improving the intelligence level of power grid operation and power supply reliability.

[0004] In a first aspect, the present invention provides a method for diagnosing abnormal voltage of users in a distribution network, the method comprising:

[0005] When a low voltage event occurs, collecting user-side electrical data and substation electrical data, and extracting basic features and electrical features from the user-side electrical data and the substation electrical data;

[0006] Acquire at least one key factor related to the low voltage event, and acquire an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule;

[0007] Performing a cause diagnosis on the low voltage event according to the basic characteristics, the electrical characteristics, and the event type under a second preset rule to obtain a cause diagnosis result, and marking the cause diagnosis result once based on a third preset rule;

[0008] Constructing an initial assessment model, obtaining multiple historical low voltage events, and performing feature extraction and cause marking on the historical low voltage events to obtain a data set, and training the initial assessment model based on the data set to obtain a final assessment model;

[0009] Inputting the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result to obtain a first diagnosis result corresponding to the primary marking result;

[0010] The first diagnosis result and the cause diagnosis result are verified and fused under a fourth preset rule, so as to output a diagnosis report on the low voltage event according to the verification and fusion results.

[0011] In a second aspect, the present invention provides a system for diagnosing abnormal voltage of distribution network users, the system comprising:

[0012] A feature extraction module is used to collect user-side electrical data and substation electrical data when a low voltage event occurs, and extract basic features and electrical features from the user-side electrical data and the substation electrical data;

[0013] an event type acquisition module, configured to acquire at least one key factor related to the low voltage event, and acquire an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule;

[0014] a marking module, configured to perform a cause diagnosis on the low voltage event based on the basic characteristics, the electrical characteristics, and the event type under a second preset rule to obtain a cause diagnosis result, and mark the cause diagnosis result once based on a third preset rule;

[0015] A model training module is used to build an initial evaluation model, obtain multiple historical low voltage events, extract features and mark causes of the historical low voltage events, obtain a data set, and train the initial evaluation model based on the data set to obtain a final evaluation model;

[0016] a machine recognition module, configured to input the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result, and obtain a first diagnosis result corresponding to the primary marking result;

[0017] The diagnosis result output module is used to verify and fuse the first diagnosis result and the cause diagnosis result under a fourth preset rule, so as to output a diagnosis report on the low voltage event according to the verification and fusion result.

[0018] In a third aspect, the present invention provides a storage medium storing one or more programs, which, when executed by a processor, implement the above-mentioned method for diagnosing abnormal voltage of distribution network users.

[0019] In a fourth aspect, the present invention provides an electronic device, comprising a memory and a processor, wherein:

[0020] The memory is used to store computer programs;

[0021] When the processor is used to execute the computer program stored in the memory, the above-mentioned method for diagnosing abnormal voltage of distribution network users is implemented.

[0022] Compared with the prior art, the present application has the following advantages:

[0023] 1、The present application constructs an electrical mechanism driven feature engineering system; designs a low voltage event type hierarchical positioning rule based on distributed state deduction; and adopts a hybrid model architecture combining rule pre-diagnosis and machine learning supplementary diagnosis; finally introduces a diagnosis result verification and conflict fusion decision mechanism based on key feature quantization review, thereby greatly improving the accuracy, real-time performance and interpretability of user low voltage cause analysis.

[0024] 2、Through deep integration of electrical mechanism, rule-based reasoning and machine learning, the recognition ability of various low voltage causes (especially complex and fuzzy scenarios) is improved; through topological relationship and distributed state deduction, the hierarchical level of low voltage event occurrence (low voltage line end, distribution transformer outlet, medium voltage line end, transformer substation outlet) can be accurately located; by combining the explicit logic of rules and the generalization ability of machine learning, and introducing electrical mechanism driven features, the diagnosis process is more transparent and the results are more reliable; for the case where rule diagnosis and machine learning model prediction are inconsistent, a fusion decision process based on key feature quantization review is designed to improve the reliability of the final diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 The flowchart of the power distribution network user voltage anomaly diagnosis method proposed for an embodiment of the present application;

[0026] Figure 2 The structural schematic diagram of the power distribution network user voltage anomaly diagnosis system proposed for an embodiment of the present application.

[0027] The following specific embodiments will further illustrate the present application in conjunction with the above drawings. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application. Unless otherwise defined, the technical terms or scientific terms used herein should be understood as the usual meanings understood by those skilled in the art in the field of the present application. The words such as "comprise" and similar words used herein mean that the elements or objects before the words cover the elements or objects listed after the words and their equivalents, without excluding other elements or objects.

[0029] The traditional analysis method of low voltage problem of power distribution network users mainly includes:

[0030] (1) Threshold-based alarm method: This is the most basic method. By monitoring user voltage data, a low voltage alarm is generated when the voltage value is lower than a preset threshold (such as -10% or -7% of the nominal voltage) for a period of time. This method can only detect the low voltage phenomenon but cannot determine its cause.

[0031] (2) Diagnostic method based on simple rules: On the basis of threshold alarm, some simple logical rules are added. For example, if low voltage occurs simultaneously in multiple users under a certain substation, it may be judged as a substation-level problem; if the voltage at the substation outlet is normal but the voltage at the user is low, it may be judged as a low-voltage line problem. These rules are usually based on experience, lack in-depth application of electrical principles, and have fixed thresholds, making it difficult to adapt to complex operating conditions and variable fault modes. For example, it may simply be judged based on the voltage drop amplitude, but it does not distinguish the voltage performance differences of transformers in different connection groups when the phase is lost, nor does it consider the voltage characteristics of specific faults such as neutral line break.

[0032] (3) Single machine learning application: Some studies have attempted to directly use machine learning (such as support vector machines (SVMs) and neural networks) to classify low voltage events. Statistical features of basic measurement data such as voltage and current are usually used as input. Although machine learning can handle nonlinear relationships, it has the following problems: ① It requires a large number of high-quality labeled samples, which are expensive to obtain; ② The model may become a "black box" and lack physical interpretability, making it difficult to gain the trust of operators; ③ It is highly dependent on feature engineering. If the key electrical mechanism features are not extracted, the model performance is limited; ④ It fails to effectively combine the topology of the power system and expert experience rules.

[0033] (4) Traditional fault recording and manual analysis: When a severe low voltage or fault occurs, the waveform is recorded by a fault recording device and manually analyzed by professional engineers. This method is highly accurate, but it is expensive, cannot cover all low voltage events, and has a slow response speed. It is not suitable for large-scale, normalized low voltage problem management.

[0034] In summary, the applicant has found the following problems with the traditional voltage anomaly detection method:

[0035] (1) Insufficient diagnostic accuracy: Simple rules have poor judgment capabilities for complex working conditions and fuzzy boundary scenarios, and are prone to misjudgment or missed judgments. For example, it is difficult to distinguish whether the voltage at the end of the low-voltage line is low or the overall voltage at the outlet of the substation is low, or it is difficult to distinguish different types of faults.

[0036] (2) Insufficient feature extraction: The electrical mechanism is not fully utilized to extract key electrical quantitative indicators that can directly reflect the specific causes of low voltage (such as phase loss in different wiring groups and neutral line break) as features, resulting in a lack of pertinence in the diagnostic model.

[0037] (3) Poor model adaptability: Rules based on fixed thresholds are difficult to adapt to dynamic scenarios such as changes in grid operation mode and load fluctuations. A single machine learning model may be sensitive to data quality and have limited generalization capabilities.

[0038] (4) Lack of hierarchical positioning: Most methods fail to effectively combine topological information to perform hierarchical and graded positioning of low-voltage problems (such as distinguishing whether the problem is at the low-voltage line, substation, medium-voltage line or substation level), which is not conducive to accurate tracing.

[0039] (5) The diagnostic results are not very interpretable: Especially for pure machine learning methods, the decision-making process is difficult to explain. Simple rules are easy to understand, but they cannot handle situations outside the rules.

[0040] (6) Lack of confidence assessment and conflict resolution: Existing methods usually only provide a single diagnostic conclusion and lack a quantitative assessment of the reliability of that conclusion. When different methods or rules conflict, there is no clear mechanism for handling conflicts.

[0041] (7) Insufficient data fusion: Failure to effectively integrate multi-source heterogeneous data such as measurement data, equipment records, user records, and event records, resulting in low information utilization.

[0042] Based on this, the present invention proposes a method for diagnosing voltage anomalies among users of distribution network, which constructs a feature engineering system driven by electrical mechanisms; designs hierarchical positioning rules for low voltage event types based on distributed state deduction; and adopts a hybrid model architecture that combines rule-based pre-diagnosis with machine learning supplementary diagnosis; finally, introduces a diagnosis result verification and conflict fusion decision-making mechanism based on quantization review of key features, realizing the diagnosis of low voltage events in various complex situations, thereby greatly improving the accuracy, real-timeness and interpretability of user low voltage cause analysis.

[0043] like Figure 1 As shown, an embodiment of the present invention provides a method for diagnosing abnormal voltage of a distribution network user, the method comprising steps S101 to S106, wherein:

[0044] Step S101: When a low voltage event occurs, user-side electrical data and substation electrical data are collected, and basic features and electrical features are extracted from the user-side electrical data and the substation electrical data;

[0045] It should be noted that the user-side electrical data includes user voltage and user current, and the substation electrical data includes substation outlet voltage and substation outlet current. In addition, in some embodiments, the collected data also includes distribution equipment ledger information: distribution transformer capacity , medium voltage line Line_MV, medium voltage line topology (superior-subordinate relationship); user ledger information: user electricity usage characteristics , TP_user of the substation to which it belongs, metering mode (single-phase / three-phase); other data: user low voltage event alarm records, line power outage and restoration records, operation records (such as load transfer) and other event information.

[0046] In some embodiments, after the above data are collected, pre-processing operations such as quality verification, outlier processing, time alignment, and data interpolation are performed on the acquired data; and pre-processing operations such as quality verification, outlier processing, time alignment, and data interpolation are performed on the acquired data.

[0047] In addition, in some embodiments, the basic features include the mean, variance, peak value, valley value, voltage change rate, current change rate, and autocorrelation coefficient of user voltage, user current, station outlet voltage, and station outlet current;

[0048] In addition, in some embodiments, key electrical quantitative indicators (electrical features) reflecting specific low voltage causes need to be calculated to serve as input features for subsequent machine learning models, specifically including:

[0049] The phase-loss characteristics of the transformer area are extracted according to the following formula:

[0050] ;

[0051] in, This is the phase-loss characteristic of the first transformer area. To return the absolute value of a number, 、 The two-phase voltage with the smaller value among the output voltages of the transformer substation, It is the single-phase voltage with the largest value among the output voltages of the transformer substation. This is the phase-loss characteristic of the second transformer area. is the nominal voltage, It is the phase-loss characteristic of the third station area;

[0052] The characteristics of low-voltage line grid problems are extracted according to the following formula:

[0053] ;

[0054] in, The first low-voltage line grid problem characteristics, The second low-voltage line grid problem characteristics, is the peak voltage at the output of the substation during peak load period, is the user voltage peak, is the valley value of the output voltage of the substation during the low load period, is the user voltage valley value;

[0055] The neutral line break feature is extracted according to the following formula:

[0056] ;

[0057] in, is the highest phase voltage, 、 、 They are the phase A, phase B, and phase C voltages at the output of the transformer substation respectively. is the minimum phase voltage, 、 These are the first and second neutral line break characteristics respectively;

[0058] The acquisition anomaly characteristics are calculated using the following formula, that is, the average and minimum values ​​of the correlation coefficients between the output voltage of the substation and the voltages of all subordinate users are calculated:

[0059] ;

[0060] in, For the first collection of abnormal features, For the second collection of abnormal features, To find the average function, is a function used to calculate the correlation coefficient between variables, is the output voltage of the substation, is the voltage of the i-th user;

[0061] The characteristics of the substation connection group are extracted according to the following formula, that is, the correlation coefficient between the change in the substation outlet voltage and the change in the current is calculated to assist in determining the connection group:

[0062] ;

[0063] in, is the characteristics of the area connection group, is the voltage change at the outlet of the substation, is the current change at the outlet of the substation;

[0064] In addition, in some embodiments, the electrical characteristics further include a fourth stage phase loss characteristic: , are the i-th phase and j-th phase of the transformer substation outlet voltage respectively, that is, the correlation coefficient between the three-phase voltages is calculated.

[0065] Step S102: obtaining at least one key factor related to the low voltage event, and obtaining an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule;

[0066] It should be noted that this step involves the hierarchical classification of user low voltage problems based on distributed state deduction. Key factors include the number of users experiencing low voltage events on the same phase in the same area. , low voltage duration , low voltage event start time difference , the minimum outlet voltage of the substation during the low voltage event The specific stratification and grading process is as follows:

[0067] (1) If ,and ,and ,and , the event type is determined to be a low voltage event at the end of the low voltage line (Event_LV_End).

[0068] (2) If ,and ,and ,and , then the event type is determined to be a distribution transformer outlet low voltage event (Event_TP_Outlet).

[0069] In addition, in some embodiments, the key factor also includes the number of low voltage events at the distribution transformer outlet under the same medium voltage line. , the starting time difference of low voltage events at each distribution transformer outlet , the stratification and grading process also includes:

[0070] (3) If ,and , the event type is determined to be a medium voltage line end low voltage event (Event_MV_End).

[0071] In addition, in some embodiments, the key factor also includes: ,and , then the event type is determined to be a low voltage event at the end of the medium voltage line. The hierarchical classification process also includes:

[0072] (4) If ,and , then the event type is determined to be a substation outlet low voltage event (Event_SS_Outlet).

[0073] Finally, assign the hierarchy: assign each user low voltage event to the highest level event it can be associated with. The priority order is: Event_SS_Outlet>Event_MV_End>Event_TP_Outlet>Event_LV_End.

[0074] Step S103: cause diagnosis of the low-voltage event according to the basic feature, the electrical feature and the event type under a second preset rule to obtain a cause diagnosis result, and marking the cause diagnosis result once based on a third preset rule;

[0075] It should be noted that the second preset rule includes multiple cases, specifically:

[0076] The first case: when the event type is a low-voltage event at the end of a low-voltage line, if , and , and , the cause diagnosis result is a low-voltage line network problem.

[0077] The second case: when the event type is a low-voltage event at the end of a medium-voltage line, if load transfer and medium-voltage single-phase disconnection are excluded, the cause diagnosis result is a medium-voltage line network problem.

[0078] The third case: when the event type is a low-voltage event at the outlet of a distribution transformer, if , and , and , and , it is a Yyn0 connection group, at this time if is normal, and , and , and , the cause diagnosis result is a high-voltage side phase loss problem.

[0079] If and / or and / or and / or , it is a Dyn11 connection group, at this time if , and two-phase voltages are both within a first preset range and the other phase voltage is less than or equal to , the cause diagnosis result is a high-voltage side phase loss problem.

[0080] The fourth case: when the event type is a low-voltage event at the outlet of a distribution transformer, the first condition is set as: , and , or two-phase voltages are both less than ; the second condition is: , or .

[0081] If the first condition and the second condition are both met, the cause diagnosis result is a neutral line disconnection problem.

[0082] The fifth case: When the event type is a low voltage event at the end of a medium voltage line, if there is a line power outage or operation record before the low voltage event, the cause diagnosis result is: load transfer;

[0083] The sixth case: When the event type is a low voltage event at the power distribution outlet, if ,and , then the cause diagnosis result is: collection abnormality;

[0084] Case 7: When the event type is a low voltage event at the end of a medium-voltage line, if at least two substations on the same line experience phase loss on the high-voltage side simultaneously, the cause diagnosis result is: single-phase disconnection of the medium-voltage line;

[0085] In addition, in some embodiments, the specific marking process is:

[0086] If there is only one cause diagnosis result related to the low voltage event, a first-level mark is performed;

[0087] If the cause diagnosis result related to the low voltage event is zero, a secondary mark is performed;

[0088] If the cause diagnosis results related to the low voltage event include at least two, a third-level marking is performed.

[0089] For example, if the user low voltage event only meets the diagnosis rules of one type of low voltage cause, it is preliminarily determined to be the cause (Cause_Rule) and marked as "high confidence cause diagnosis (Rule_High_Confidence)";

[0090] If the user low voltage event does not meet any of the diagnostic rules for the low voltage cause, it will be marked as "cause unknown (Rule_Unknown)";

[0091] If two or more low voltage cause diagnosis rules are met at the same time, it is marked as "cause ambiguity (Rule_Ambiguous)".

[0092] In summary, traditional low-voltage diagnosis methods can suffer from computational inefficiencies when processing large numbers of alarms if they directly employ complex analysis models. For typical faults with clear electrical characteristics, rule-based diagnosis is often more direct and efficient. However, existing rule-based systems can suffer from issues such as incomplete rule coverage, reliance on expert experience for threshold setting and inflexible adjustment, and inaccurate hierarchical localization of low-voltage events (especially when metering coverage is incomplete). These issues limit the diagnostic accuracy and practicality of rule-based systems. To address this, the present invention designs a distributed state-based hierarchical classification method for low-voltage issues and a rapid pre-diagnosis process based on electrical mechanisms. This step first uses distributed state-based logic to generate virtual low-voltage events for the low-voltage line end, distribution transformer exit, medium-voltage line end, and even the substation exit, based on measured data at the substation outlet and user side, combined with the spatiotemporal correlation characteristics of low-voltage events across user groups (e.g., multiple users with the same phase, same substation, duration, start time difference, etc.). Each user low-voltage event is assigned a unique, highest-priority hierarchical classification, resolving the hierarchical localization challenge caused by metering failures. Secondly, a series of clear diagnostic rules were constructed based on the basic features and electrical characteristics extracted in the above steps, targeting different hierarchical levels and known typical low voltage causes (such as low-voltage / medium-voltage grid problems, phase loss on the high-voltage side of a substation, neutral line breakage, load transfer, data collection anomalies, and single-phase medium-voltage line breakage). These rules directly utilize typical electrical manifestations of faults (for example, the characteristic of "one phase normal, two phases low, and similar voltage sum" for a phase loss in the Dyn11 distribution transformer, and "voltage imbalance with specific correlation" for a neutral line break). Finally, if a user low voltage event meets the diagnostic rules for a specific cause, it is preliminarily marked as "high-confidence cause diagnosis." If none of the rules are met, or if multiple rules are met simultaneously, it is marked as "unknown cause" or "ambiguous cause," providing preliminary screening results for subsequent steps. In addition, the layered and graded method compensates for the information loss caused by insufficient actual metering points and improves the accuracy of locating the source of low-voltage problems. At the same time, by utilizing rules containing electrical mechanisms, typical low-voltage events with obvious characteristics can be quickly and accurately identified, greatly improving diagnostic efficiency. It also provides high-value prior information and preliminary confidence assessments for subsequent diagnosis of complex causes, effectively diverting diagnostic tasks.

[0093] Step S104: constructing an initial evaluation model, obtaining multiple historical low voltage events, and performing feature extraction and cause marking on the historical low voltage events to obtain a data set, and training the initial evaluation model based on the data set to obtain a final evaluation model;

[0094] It should be noted that the evaluation model uses the XGBoost model, the input layer of which is a variety of feature vectors, which include all basic features and electrical features, and the output layer of which is the diagnosis results of each cause. The probability distribution of the probability distribution , They are the first, second, and Nth cause diagnosis results respectively.

[0095] After selecting the initial evaluation model, it needs to be trained. Specifically, a large-scale, high-quality annotated dataset is constructed. Each data entry represents a historical low voltage event, including its complete feature vector and the true cause label confirmed by experts or verified on-site. The input numerical features are then normalized. The annotated dataset is then used to train the XGBoost model, using the cross-entropy loss function to minimize the difference between the predicted probability and the true label. Finally, the model performance is evaluated on the test set. Hyperparameters are then tuned to obtain the final evaluation model.

[0096] Step S105: inputting the basic characteristics and electrical characteristics into the final evaluation model according to the primary marking result to obtain a first diagnosis result corresponding to the primary marking result;

[0097] It should be noted that, in some embodiments, the first diagnostic result is specifically:

[0098] For a low voltage event with a primary marker, inputting basic features and electrical features related to the low voltage event into the final evaluation model to output a cause diagnosis result with the highest probability and its first probability value, and outputting a second probability value corresponding to the cause diagnosis result outputted under a second preset rule for the low voltage event;

[0099] For a low voltage event with a secondary or tertiary marker, the basic features and electrical features associated with the low voltage event are input into the final evaluation model to output a cause diagnosis result with the highest probability and its third probability value.

[0100] In summary, it needs to be pointed out that although the rule-based diagnosis system is efficient, its diagnosis capability is often limited when dealing with new types of faults that are not covered by the rules, complex fault scenarios caused by multi-factor coupling, or edge cases with atypical feature manifestations, resulting in "unknown cause" or "ambiguous cause" outcomes. In addition, rule systems usually rely on fixed thresholds and idealized conditions, making it difficult to adapt to the diversity of actual working conditions and data noise. Based on this, the present application introduces a supplementary diagnosis process based on machine learning. Specifically, first, a suitable machine learning model is selected and designed, such as a gradient boosting decision tree (specifically using XGBoost), which can effectively handle complex interactions between features, is robust to data missing, and can output feature importance to assist understanding. Second, the input layer of the model is designed to extract basic statistical features and all key quantitative indicators driven by electrical mechanisms, which provide the model with rich physical information directly related to the causes of low voltage. The output layer of the model uses a Softmax function to output the probability distribution of the current low voltage event belonging to each pre-defined cause (such as low voltage network, medium voltage network, each type of open phase, neutral line breakage, etc.). Then, a large-scale, high-quality historical low voltage event dataset containing real cause labels is constructed to train the model, and strict evaluation and tuning are performed. In the application stage, the cause with the highest probability output by the model will be the supplementary diagnosis result. The effect of this is that it utilizes the powerful pattern recognition and nonlinear mapping capabilities of machine learning to effectively compensate for the shortcomings of rule systems in handling complex, atypical, and unknown fault patterns; by inputting features containing rich electrical mechanisms, the model decision-making is more physically meaningful and has potential for interpretability; it can learn from historical data the deep associations that rules cannot explicitly express, thereby improving the diagnosis accuracy and coverage of "unknown / ambiguous cause" events, providing more comprehensive information input for subsequent diagnosis result fusion, and significantly improving the intelligence level and diagnosis performance of the entire diagnosis system.

[0101] Step S106: verifying and fusing the first diagnosis result and the cause diagnosis result under the fourth preset rule to output a diagnosis report about the low voltage event according to the verification and fusion result.

[0102] In the process of verifying and fusing the cause diagnosis result identified by the rules and the first diagnosis result identified by the machine, the two identification results need to be preliminarily compared and confirmed first:

[0103] (1) For a low voltage event with a first-level label, if the first probability and the second probability are equal, it is determined that the machine identification result is the same as the rule identification result, at which point the cause diagnosis result of machine identification or rule identification is taken as the final diagnosis result, and the confidence level is set to high;

[0104] (2) If the first probability is not equal to the second probability, but the second probability is greater than or equal to the 27th threshold , the cause diagnosis result identified by the rule is used as the final diagnosis result, and the confidence rating is set to medium;

[0105] (3) If the second probability value is less than , then the first probability value is greater than or equal to the twenty-eighth threshold , then it is determined that the machine recognition result conflicts with the rule recognition result, and subsequent analysis is carried out;

[0106] (4) For a low voltage event with a secondary or tertiary marker, if the third probability value is greater than or equal to the 29th threshold , the cause diagnosis result corresponding to the third probability value identified by the machine is used as the final diagnosis result, and the confidence assessment level is set to high;

[0107] (5) If the third probability value is less than , the cause result is determined to be pending, and the confidence rating is set to low, and manual analysis is recommended.

[0108] Then, if the machine recognition results conflict with the rule recognition results, key feature review is performed:

[0109] First, obtain the first key feature related to the rule recognition result and the second key feature related to the machine recognition result, and calculate the feature deviation based on the first key feature and the second key feature. and characteristic abnormality ; In some embodiments, the first key feature is the core condition for triggering the rule recognition result, and the second key feature is a highly important feature relied upon for machine recognition, such as the top three features in the XGBoost model.

[0110] Specifically, if ,and , the cause diagnosis result corresponding to the third probability value identified by the machine is used as the final diagnosis result, and the confidence assessment level is set to high;

[0111] like ,and , the cause diagnosis result identified by the rule is used as the final diagnosis result, and the confidence rating is set to high;

[0112] like ,and , the cause is determined to be pending, and the confidence level is set to low. In this case, it is recommended to switch to manual analysis.

[0113] In some embodiments, the characteristic deviation is calculated according to the following formula:

[0114] ;

[0115] In some embodiments, the characteristic deviation is calculated according to the following formula:

[0116] ;

[0117] in, is the actual calculated or measured value of the corresponding feature, is the judgment threshold pre-set for the corresponding feature, is the first key feature, The second key feature is is the mean of the corresponding feature, is the standard deviation of the corresponding feature.

[0118] For example, assuming that in a low voltage event, the cause diagnosis result of rule identification is "low voltage line grid problem" (high confidence), while the cause diagnosis result of machine identification is confirmed to be "neutral line break" with a high probability of 0.85. At this time, it is determined that there is a diagnostic conflict. The first key feature includes the core electrical quantities it relies on, such as "user low voltage level" (assuming it meets the "low voltage line end"), "difference between the output voltage of the substation and the user voltage during the peak load period" and "the difference between the output voltage of the substation and the user voltage during the peak load period". "(Assuming the calculated value is 23.5V, and the rule threshold is >22V, the deviation is about 6.8%), "The difference between the output voltage of the substation and the user voltage during the load valley period ” (assuming the value is 10V, which meets the <22V rule) and “User off-peak voltage ” (Assuming the value is 205V, which satisfies the rule of >198V). On the other hand, through SHAP analysis, the second key feature that contributes most to the neutral line break output of the machine recognition may include: "the highest phase voltage "(Assuming the measured value is 265V, if the normal mean is 221V and the standard deviation is 5V, then its Up to 8.8), "minimum phase voltage "(Assuming the measured value is 160V, if the normal mean is 219V and the standard deviation is 4V, then its is 14.75) and " and The relative deviation relationship of the nominal voltage is shown as one phase is significantly higher by 45V and the other phase is significantly lower by 60V, which is consistent with the neutral line break feature. Although the conditions are met, the deviation of 6.8% is relatively small; and of The values ​​(8.8 and 14.75) are far above the statistical significance threshold (such as >3), indicating extremely abnormal electrical performance. Therefore, the system determines that the characteristic evidence supporting machine learning diagnosis B is stronger and more significant, overturning the rule-based diagnosis and accepting "neutral wire broken" as the final diagnosis, assigning it a high confidence level.

[0119] In some embodiments, a structured diagnostic report is output. This report clearly identifies the cause of the low voltage after verification and decision-making, along with a confidence level determined based on the decision logic. The report may also include key quantitative indicators that triggered the decision (such as feature deviation and feature anomaly) and a brief description of the decision path.

[0120] also, to are respectively the first threshold to the thirty-third threshold, illustratively, to Can be 、 、 、 、 、 ; The first preset range is to ; Since these thresholds and the first preset range are related to specific usage requirements, in some other embodiments of the present invention, the above thresholds and the first preset range can also be other values.

[0121] In summary, it needs to be pointed out that, in the existing low-voltage cause diagnosis method, when combining rule diagnosis and machine learning diagnosis, the following problems are often encountered: some methods may simply use a voting mechanism or a preset priority to handle conflicts, lacking detailed consideration of the strength of data evidence in specific events, when the applicable boundary of the rule is fuzzy or the machine learning model encounters rare working conditions not fully covered in the training data, such simple fusion strategy is easy to lead to misjudgment. Although some methods try to introduce more complex fusion models, their decision-making process may not be transparent, difficult to trace and understand, especially when the rule and model conclusions are highly conflicting, it is difficult to provide clear explanation and reliable confidence evaluation. Based on this, the present application proposes a verification and fusion decision mechanism based on key feature quantification review. Specifically, when the rule diagnosis (e.g., diagnosing cause A) and the machine learning model diagnosis (e.g., diagnosing cause B, and the probability P(B) is greater than or equal to 0.75) conflict, the present application does not simply choose one, but first identifies the first key feature and the second key feature that respectively support cause A (derived from rule definition) and cause B (derived from model explanation, such as strong correlation features obtained by SHAP value analysis). Secondly, the actual measured value or calculated value of these key features in the current event is quantitatively evaluated, and then by comparing these quantitative indicators, it is judged which side of the diagnostic conclusion is more consistent with the actual performance of the current data. The technical effect of this mechanism is that it focuses on the quantitative performance of the core electrical features directly related to the conflicting diagnosis, providing a transparent, interpretable and data evidence-based decision path to solve the diagnosis conflict between rules and machine learning models, avoiding the pitfalls of blindly trusting a certain diagnostic source or simple voting; at the same time, it can dynamically evaluate the evidence strength according to the significance of feature deviation, so as to make more robust judgments in complex and marginal cases, improve the accuracy and reliability of the final diagnosis result, and can give clear "human intervention" suggestions for difficult events, ensuring the reliability of the diagnosis.

[0122] Compared with the prior art, the present application has the following advantages:

[0123] 1、The present application constructs an electrical mechanism-driven feature engineering system; designs low-voltage event type hierarchical positioning rules based on distributed state deduction; and adopts a hybrid model architecture combining rule pre-diagnosis and machine learning supplementary diagnosis; finally introduces a diagnosis result verification and conflict fusion decision mechanism based on key feature quantification review, thereby greatly improving the accuracy, real-time performance and interpretability of user low-voltage cause analysis.

[0124] 2. By deeply integrating electrical mechanisms, rule-based reasoning, and machine learning, the ability to identify various causes of low voltage (especially complex and ambiguous scenarios) is improved. By deducing based on topological relationships and distributed states, the level at which low voltage events occur (the end of the low-voltage line, the distribution transformer outlet, the end of the medium-voltage line, and the substation outlet) can be accurately located. By combining the clear logic of rules with the generalization capabilities of machine learning and introducing features driven by electrical mechanisms, the diagnostic process is made more transparent and the results more reliable. For situations where rule-based diagnosis is inconsistent with the predictions of the machine learning model, a fusion decision-making process based on quantitative review of key features is designed to improve the reliability of the final diagnosis.

[0125] like Figure 2 As shown, an embodiment of the present invention provides a system for diagnosing abnormal voltage of users in a distribution network, the system comprising:

[0126] The feature extraction module 10 is used to collect user-side electrical data and substation electrical data when a low voltage event occurs, and extract basic features and electrical features from the user-side electrical data and the substation electrical data;

[0127] An event type acquisition module 20 is configured to acquire at least one key factor related to the low voltage event, and acquire an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule;

[0128] a marking module 30, configured to perform a cause diagnosis on the low voltage event based on the basic characteristics, the electrical characteristics, and the event type under a second preset rule to obtain a cause diagnosis result, and to mark the cause diagnosis result based on a third preset rule;

[0129] The model training module 40 is used to construct an initial evaluation model, obtain multiple historical low voltage events, extract features and mark causes of the historical low voltage events, obtain a data set, and train the initial evaluation model based on the data set to obtain a final evaluation model;

[0130] The machine recognition module 50 is configured to input the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result to obtain a first diagnosis result corresponding to the primary marking result;

[0131] The diagnosis result output module 60 is configured to verify and fuse the first diagnosis result and the cause diagnosis result under a fourth preset rule, so as to output a diagnosis report on the low voltage event according to the verification and fusion results.

[0132] On the other hand, the present invention further provides a storage medium having one or more programs stored thereon, which, when executed by a processor, implements the above-mentioned method for diagnosing abnormal voltage of distribution network users.

[0133] Another aspect of the present application also provides an electronic device comprising a memory and a processor, wherein the memory is configured to store a computer program, and the processor is configured to execute the computer program stored in the memory to implement the power grid user voltage anomaly diagnosis method described above.

[0134] Those skilled in the art can understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a list of executable instructions for implementing the logic function, which can be embodied in any computer readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch the instructions from the instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instruction execution systems, apparatus or devices. For the purpose of this specification, the "computer readable medium" can be any device that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices.

[0135] More specific examples (a non-exhaustive list) of the computer readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium can even be paper or another suitable medium upon which the program is printed, because the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by electronic means to obtain, interpret or process the program, and then store the program in a computer memory.

[0136] It should be understood that parts of the present application can be implemented in hardware, software, firmware or a combination thereof. In the above described embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and as in another embodiment, any of the following technologies known in the art or a combination thereof can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0137] While the embodiments of the present invention have been described in detail above, it will be apparent to those skilled in the art that various modifications and variations of these embodiments are possible. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as set forth in the claims. Furthermore, the invention described herein is susceptible to other embodiments and may be practiced or implemented in a variety of ways.

Claims

1. A method for diagnosing abnormal voltage of distribution network users, characterized in that: The method comprises: When a low voltage event occurs, user-side electrical data and substation electrical data are collected, and basic features and electrical features are extracted from the user-side electrical data and the substation electrical data. The electrical features are key electrical quantitative indicators that reflect the specific causes of low voltage. The electrical features include substation phase loss features, low-voltage line grid problem features, neutral line break features, and substation connection group features. Acquire at least one key factor related to the low voltage event, and acquire an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule; Performing a cause diagnosis on the low voltage event according to the basic characteristics, the electrical characteristics, and the event type under a second preset rule to obtain a cause diagnosis result, and marking the cause diagnosis result once based on a third preset rule; Constructing an initial assessment model, obtaining multiple historical low voltage events, and performing feature extraction and cause marking on the historical low voltage events to obtain a data set, and training the initial assessment model based on the data set to obtain a final assessment model; Inputting the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result to obtain a first diagnosis result corresponding to the primary marking result; The first diagnosis result and the cause diagnosis result are verified and fused under a fourth preset rule, so as to output a diagnosis report on the low voltage event according to the verification and fusion results.

2. The method for diagnosing abnormal voltage of distribution network users according to claim 1, characterized in that: When a low voltage event occurs, the steps of collecting user-side electrical data and substation electrical data, and extracting basic features and electrical features from the user-side electrical data and the substation electrical data include: The user-side electrical data includes user voltage and user current, the substation electrical data includes substation outlet voltage and substation outlet current, and the basic features include the mean, variance, peak value, valley value, voltage change rate, current change rate, and autocorrelation coefficient of user voltage, user current, substation outlet voltage, and substation outlet current; The phase-loss characteristics of the transformer area are extracted according to the following formula: ； in, This is the phase-loss characteristic of the first transformer area. To return the absolute value of a number, 、 The two-phase voltage with the smaller value among the output voltages of the transformer substation, It is the single-phase voltage with the largest value among the output voltages of the transformer substation. This is the phase-loss characteristic of the second transformer area. is the nominal voltage; The characteristics of low-voltage line grid problems are extracted according to the following formula: ； in, The first low-voltage line grid problem characteristics, The second low-voltage line grid problem characteristics, is the peak voltage at the output of the substation during peak load period, is the user voltage peak, is the valley value of the output voltage of the substation during the low load period, is the user voltage valley value; The neutral line break feature is extracted according to the following formula: ； in, is the highest phase voltage, 、 、 They are the phase A, phase B, and phase C voltages at the output of the transformer substation respectively. is the minimum phase voltage, 、 These are the first and second neutral line break characteristics respectively; The acquisition anomaly characteristics are calculated according to the following formula: ； in, For the first collection of abnormal features, For the second collection of abnormal features, To find the average function, is a function used to calculate the correlation coefficient between variables, is the output voltage of the substation, is the voltage of the i-th user; Extract the characteristics of the area connection group according to the following formula: ； in, is the characteristics of the area connection group, is the voltage change at the outlet of the substation, is the change in current at the substation outlet.

3. The method for diagnosing abnormal voltage of distribution network users according to claim 2, characterized in that: The step of obtaining at least one key factor related to the low voltage event and obtaining an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule includes: The key factors include the number of users experiencing low voltage events on the same phase in the same area , low voltage duration , low voltage event start time difference , the minimum outlet voltage of the substation during the low voltage event ; The first preset rule is: like ,and ,and ,and , then the event type is determined to be a low-voltage event at the end of the low-voltage line; like ,and ,and ,and , then the event type is determined to be a low voltage event at the distribution transformer outlet; The key factors also include the number of low voltage events at the distribution transformer outlet on the same medium voltage line. , the starting time difference of low voltage events at each distribution transformer outlet ; like ,and , then the event type is determined to be a low voltage event at the end of the medium voltage line; The key factors also include the number of low voltage events at the end of the medium voltage line at the same substation outlet. , the starting time difference of the low voltage event at the end of the medium voltage line ; like ,and , then the event type is determined to be a substation outlet low voltage event; in, to They are the first to eighth thresholds respectively.

4. The method for diagnosing abnormal voltage of distribution network users according to claim 3, characterized in that: The step of performing a cause diagnosis on the low voltage event according to the basic characteristics, the electrical characteristics, and the event type under the second preset rule to obtain a cause diagnosis result, and marking the cause diagnosis result based on a third preset rule includes: The second preset rule includes multiple situations, specifically: The first case: When the event type is a low voltage event at the end of the low voltage line, if ,and ,and , then the cause diagnosis result is: low voltage line grid problem; Case 2: When the event type is a low voltage event at the end of the medium voltage line, if load transfer and medium voltage single-phase disconnection are excluded, the cause diagnosis result is: medium voltage line grid problem; The third case: When the event type is a low voltage event at the distribution transformer outlet, if ,and ,and ,and , then it is the Yyn0 connection group, then if Normal and ,and ,and , then the cause diagnosis result is: phase loss problem occurs on the high voltage side; like and / or and / or and / or , then it is a Dyn11 connection group. ,and There are two phase voltages within the first preset range, and the other phase voltage is less than or equal to , then the cause diagnosis result is: phase loss problem occurs on the high voltage side; Case 4: When the event type is a low voltage event at the distribution transformer outlet, set the first condition to: and , or there are two phase voltages less than ; The second condition is: ,or ; If both the first and second conditions are met, the cause diagnosis result is: neutral line break problem; The fifth case: When the event type is a low voltage event at the end of a medium voltage line, if there is a line power outage or operation record before the low voltage event, the cause diagnosis result is: load transfer; The sixth case: When the event type is a low voltage event at the power distribution outlet, if ,and , then the cause diagnosis result is: collection abnormality; Case 7: When the event type is a low voltage event at the end of a medium-voltage line, if at least two substations on the same line experience phase loss on the high-voltage side simultaneously, the cause diagnosis result is: single-phase disconnection of the medium-voltage line; in, to They are the ninth to twenty-sixth thresholds respectively.

5. The method for diagnosing abnormal voltage of distribution network users according to claim 4, characterized in that: The step of marking the cause diagnosis result based on the third preset rule includes: If there is only one cause diagnosis result related to the low voltage event, a first-level mark is performed; If the cause diagnosis result related to the low voltage event is zero, a secondary mark is performed; If the cause diagnosis results related to the low voltage event include at least two, a third-level marking is performed.

6. The method for diagnosing abnormal voltage of distribution network users according to claim 5, characterized in that: The initial evaluation model adopts the XGBoost model, the feature vector in its input layer includes the basic features and electrical features, and the output of its output layer is the probability distribution of each cause diagnosis result.

7. The method for diagnosing abnormal voltage of distribution network users according to claim 6, characterized in that: The step of inputting the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result to obtain the first diagnosis result corresponding to the primary marking result includes: For a low voltage event with a primary marker, inputting basic features and electrical features related to the low voltage event into the final evaluation model to output a cause diagnosis result with the highest probability and its first probability value, and outputting a second probability value corresponding to the cause diagnosis result outputted under a second preset rule for the low voltage event; For a low voltage event with a secondary or tertiary marker, the basic features and electrical features associated with the low voltage event are input into the final evaluation model to output a cause diagnosis result with the highest probability and its third probability value.

8. The method for diagnosing abnormal voltage of distribution network users according to claim 7, characterized in that: The step of verifying and fusing the first diagnosis result and the cause diagnosis result under the fourth preset rule to output a diagnosis report on the low voltage event according to the verification and fusion result includes: For a low voltage event with a first-level flag, if the first probability and the second probability are equal, the machine recognition result and the rule recognition result are determined to be the same. In this case, the cause diagnosis result of the machine recognition or rule recognition is used as the final diagnosis result. If the first probability is not equal to the second probability, but the second probability is greater than or equal to the 27th threshold , then the cause diagnosis result identified by the rule is taken as the final diagnosis result; If the second probability value is less than , then the first probability value is greater than or equal to the twenty-eighth threshold , then it is determined that the machine recognition result conflicts with the rule recognition result; For a low voltage event with a secondary or tertiary marker, if the third probability value is greater than or equal to the 29th threshold , the cause diagnosis result corresponding to the third probability value identified by the machine is used as the final diagnosis result; If the third probability value is less than , then the cause is still to be determined.

9. The method for diagnosing abnormal voltage of distribution network users according to claim 8, characterized in that: The method further comprises: In the case of a conflict between the machine recognition result and the rule recognition result, the first key feature related to the rule recognition result and the second key feature related to the machine recognition result are obtained, and the feature deviation degree is calculated based on the first key feature and the second key feature. and characteristic abnormality ; like ,and , the cause diagnosis result corresponding to the third probability value identified by the machine is used as the final diagnosis result; like ,and , then the cause diagnosis result identified by the rules is taken as the final diagnosis result; like ,and , then the cause is determined to be undetermined. to They are the 30th to the 33rd thresholds respectively.

10. A distribution network user voltage anomaly diagnosis system, characterized in that: The system comprises: A feature extraction module is used to collect user-side electrical data and substation electrical data when a low voltage event occurs, and extract basic features and electrical features from the user-side electrical data and the substation electrical data. The electrical features are key electrical quantitative indicators that reflect the causes of specific low voltage. The electrical features include substation phase loss features, low-voltage line grid problem features, neutral line break features, and substation connection group features. an event type acquisition module, configured to acquire at least one key factor related to the low voltage event, and acquire an event type corresponding to the low voltage event according to the at least one key factor under a first preset rule; a marking module, configured to perform a cause diagnosis on the low voltage event based on the basic characteristics, the electrical characteristics, and the event type under a second preset rule to obtain a cause diagnosis result, and mark the cause diagnosis result once based on a third preset rule; A model training module is used to build an initial evaluation model, obtain multiple historical low voltage events, extract features and mark causes of the historical low voltage events, obtain a data set, and train the initial evaluation model based on the data set to obtain a final evaluation model; a machine recognition module, configured to input the basic characteristics and the electrical characteristics into the final evaluation model according to the primary marking result, and obtain a first diagnosis result corresponding to the primary marking result; The diagnosis result output module is used to verify and fuse the first diagnosis result and the cause diagnosis result under a fourth preset rule, so as to output a diagnosis report on the low voltage event according to the verification and fusion result.

Images