Transformer area line loss simulation method based on multi-source data

Abstract

The invention discloses a transformer area line loss simulation method based on multi-source data, and relates to the technical field of power system data analysis, and the method comprises the steps: collecting the electric energy data of each user in a transformer area and the electric energy data of a transformer area general meter, and extracting the power utilization behavior feature vector and the first behavior stability index of each user, a user behavior feature set is obtained and clustered, and a plurality of homogeneous user groups and the group deviation degree of each user are obtained; based on the transformer area general meter electric energy data, a transformer area line loss abnormal event is identified, the occurrence time period of the abnormal event is recorded, the association strength of the power consumption behavior of each user and the transformer area line loss abnormal event is calculated as the event association degree, and the dynamic abnormal confidence coefficient is calculated in combination with the first behavior stability index of the user; and calculating a comprehensive abnormal score of each user based on the group deviation degree and the dynamic abnormal confidence coefficient of each user, and screening out abnormal users. The technical problem that in the prior art, transformer area line loss analysis efficiency is low is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system data analysis, and particularly relates to a simulation method for line loss of a transformer area based on multi-source data. BACKGROUND

[0002] The existing line loss analysis method of a transformer area is highly dependent on accurate power grid topology for physical simulation. However, the topology of a distribution network, especially a low-voltage transformer area, changes frequently, and the archive data often does not match the actual situation, resulting in inherent deviation in the theoretical line loss calculation based on simulation. When the theoretical line loss and the actual statistical line loss differ, the existing technology cannot accurately locate the specific reasons for the difference and the abnormal users, and still needs to rely on manual on-site investigation, which is low in efficiency and high in cost. SUMMARY

[0003] The present application provides a simulation method for line loss of a transformer area based on multi-source data, which is used to solve the technical problems of low efficiency and inaccurate results of the line loss analysis method of a transformer area in the prior art.

[0004] In view of the above problems, the present application provides a simulation method for line loss of a transformer area based on multi-source data, which comprises: Collecting electric energy data of each user in the transformer area and total meter electric energy data of the transformer area, wherein the electric energy data at least includes time series data of voltage, current and active power, extracting a power consumption behavior feature vector and a first behavior stability index of each user based on the electric energy data of each user, obtaining a user behavior feature set, clustering the user behavior feature set based on a preset clustering algorithm, and obtaining a plurality of homogeneous user groups and a group deviation degree of each user; Based on the total meter electric energy data of the transformer area, identifying a line loss abnormal event of the transformer area and recording an abnormal event occurrence period, calculating a power consumption behavior of each user and an association strength of the line loss abnormal event in the abnormal event occurrence period as an event association degree, and calculating a dynamic abnormal confidence degree in combination with the first behavior stability index of the user; Based on the group deviation degree and the dynamic abnormal confidence degree of each user, calculating a comprehensive abnormal score of each user, and screening out an abnormal user based on a preset comprehensive abnormal score threshold.

[0005] One or more technical solutions provided in the present application have at least the following technical effects or advantages: The application provides a transformer area line loss simulation method based on multi-source data, constructs a complete technical scheme from feature extraction, behavior clustering to abnormal event correlation analysis by comprehensively utilizing time sequence data such as voltages, currents and active powers of users and total meters in the transformer area, and significantly improves the accuracy and reliability of transformer area line loss abnormal user identification. Compared with the traditional method, the technical scheme provided by the application significantly overcomes the limitation of only relying on simple data comparison and static threshold judgment, realizes deep mining and dynamic evaluation of user power consumption behavior, achieves the technical effect of comprehensively improving the fine and intelligent level of transformer area line loss management, can provide clear and specific abnormal user positioning report and operation work order guidance for operation and maintenance personnel, and greatly improves the pertinence and efficiency of on-site troubleshooting, thereby providing strong technical support for guaranteeing the stable operation of the power grid and reducing power loss. BRIEF DESCRIPTION OF DRAWINGS

[0006] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiments description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0007] Figure 1 A flowchart of a transformer area line loss simulation method based on multi-source data provided by an embodiment of the application.

[0008] Figure 2 A flowchart of obtaining a user behavior feature set in the method provided by the embodiment of the application. DETAILED DESCRIPTION

[0009] The application provides a transformer area line loss simulation method based on multi-source data, which is used to solve the technical problems of low efficiency and inaccurate results of the existing transformer area line loss analysis method.

[0010] The technical solutions in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0011] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units does not have to be limited to those clearly listed steps or units, but can include other steps or modules that are not clearly listed or inherent to the process, method, product or device.

[0012] Examples, such as Figure 1 As shown, this application provides a method for simulating line loss in transformer substations based on multi-source data, wherein the method includes: S10: Collect the electricity data of each user in the transformer area and the electricity data of the transformer area's main meter. The electricity data includes at least time series data of voltage, current, and active power. Based on the electricity data of each user, extract the electricity consumption behavior feature vector and the first behavior stability index of each user to obtain a user behavior feature set. Based on a preset clustering algorithm, cluster the user behavior feature set to obtain multiple homogeneous user groups and the group deviation of each user.

[0013] In transformer substation line loss analysis, traditional methods typically treat users as isolated individuals or perform simple classifications, lacking in-depth analysis of the similarities and differences in electricity consumption behavior within user groups. Because user electricity consumption behavior is complex and variable, relying on only a single or a few electricity consumption indicators is insufficient to comprehensively and accurately depict users' true electricity consumption patterns, leading to unreliable baselines in subsequent anomaly analysis.

[0014] Step S10 in the method provided in this application embodiment includes: The electricity information collection system acquires time-series data of voltage, current, and active power uploaded by smart meters of each user in the distribution area. The distribution automation system acquires the voltage, current, and active power time-series data of the transformer area's main meter for the same time period. The collected time-series data is cleaned and time-series aligned to form a multi-source data set; Among these, obtaining user behavior feature sets, such as Figure 2 As shown, it includes: Extract load pattern characteristics, including calculating the dynamic time-normalized distance vector of the user's daily load curve, and statistical characteristics including maximum value, minimum value, average value, standard deviation, and peak-to-valley difference; Extract voltage-current coupling characteristics, including calculating the average and standard deviation of the power factor of users at different power levels, and calculating the windowed cross-correlation coefficient between the voltage series and the current series; Extracting electricity consumption entropy features, including calculating the sample entropy of the user's active power sequence; Based on cosine similarity, obtain the first behavior stability index for each user; The stability index of each user's first behavior is obtained, including: Based on the user's current period's electricity consumption behavior feature vector, a cosine similarity is calculated with the average of its historical feature vectors over the past M periods, and the value of the similarity is used as the first behavior stability index. The load form features, voltage-current coupling relationship features, and power consumption entropy features are combined to form a power consumption behavior feature vector of each user, and the first behavior stability index is used to form a user behavior feature set together with the first behavior stability index; The K-means++ clustering algorithm is used to cluster the power consumption behavior feature vectors of all users to obtain K homogeneous user groups. For each user, the Mahalanobis distance of the power consumption behavior feature vector of the user to the cluster center to which the user belongs is calculated as an initial group deviation of the corresponding user. The initial group deviation is corrected using the first behavior stability index of the user to obtain a group deviation of the corresponding user. The initial group deviation is corrected using the first behavior stability index of the user, and the correction method is: The group deviation = the initial group deviation / (the first behavior stability index + a), where a is a preset smoothing factor.

[0015] In the embodiments of the present application, the power data of each user in the transformer area and the total meter power data of the transformer area are collected, and the power data at least includes time series data of voltage, current, and active power. Based on the power data of each user, the power consumption behavior feature vector and the first behavior stability index of each user are extracted, the user behavior feature set is obtained, and the user behavior feature set is clustered based on a preset clustering algorithm to obtain multiple homogeneous user groups and the group deviation of each user.

[0016] Specifically, first, the time series data of voltage, current, and active power uploaded by the smart meters of each user in the transformer area is obtained through the power consumption information collection system. For example, the time series data of voltage, current, and active power recorded by each smart meter of each user in the transformer area every 15 minutes in the past month is collected.

[0017] Further, the time series data of voltage, current, and active power of the total meter in the same time period is obtained through the power distribution automation system. For example, the time series data of voltage, current, and active power every 15 minutes in the past month is also collected.

[0018] Further, the collected time series data is processed for data cleaning and time series alignment to form a multi-source data set. Specifically, data cleaning mainly eliminates null values or obvious error outliers, and time series alignment ensures that the user data and the total meter data completely correspond in time stamp, and finally forms a complete and consistent multi-source data set.

[0019] Further, load shape features are extracted, including calculating dynamic time warping distance vector of user daily load curve, and statistical features including maximum, minimum, average, standard deviation, peak-valley difference. For example, a user's load maximum can be 2.1 kW, minimum 0.3 kW, average 0.8 kW, standard deviation 0.5 kW, peak-valley difference 1.8 kW, and the dynamic time warping distance vector of the user's daily load curve is a vector representing the local difference in shape between two curves. For example, a typical working day is selected as the reference daily load curve, and the user's different daily load curves are obtained. The optimal matching path between the user's load curve and the reference curve is calculated using the dynamic time warping algorithm, and the sequence of distance values at each point on the matching path constitutes the dynamic time warping distance vector.

[0020] Further, voltage-current coupling relationship features are extracted, including calculating the average and standard deviation of the power factor of the user at different power levels, and calculating the windowed cross-correlation coefficient of the voltage sequence and the current sequence. Specifically, by analyzing the power factor of the user at different power levels, the average and standard deviation are calculated, and the cross-correlation coefficient such as the Pearson coefficient between the voltage change sequence and the current change sequence in the sliding time window is calculated.

[0021] Further, the power consumption entropy feature is extracted, and the sample entropy of the user's active power sequence is calculated as an example. The higher the sample entropy value, the more complex and unpredictable the sequence is. For example, the power consumption sequence of a user with extremely regular work and rest habits may have a low sample entropy, such as 0.3, while the power consumption sequence of a user with irregular work and rest habits and frequent appliance start-stop may have a high sample entropy, such as 0.8.

[0022] Further, based on the cosine similarity, the first behavior stability index of each user is obtained. Specifically, based on the user's current period power consumption behavior feature vector, the cosine similarity is calculated with the average of its historical feature vectors in the past M periods, and the value of the similarity is taken as the first behavior stability index. For example, the power consumption behavior feature vector of the user in the current period, such as this week, is obtained, and the cosine similarity between it and the average of the historical feature vectors in the past M periods, such as the past four weeks, is calculated as the first behavior stability index. The value of the similarity ranges from 0 to 1, and the closer the value is to 1, the more similar the user's current behavior is to the past habits, and the more stable it is.

[0023] Further, the load shape features, voltage-current coupling relationship features, and power consumption entropy features are combined to form the power consumption behavior feature vector of each user, and together with the first behavior stability index, they form the user behavior feature set.

[0024] Further, the K-means++ clustering algorithm is used to cluster the electricity consumption behavior feature vectors of all users to obtain K homogeneous user groups. For example, users are divided into groups such as "low energy consumption stable users" and "high energy consumption fluctuating users", and each homogeneous user group has similar electricity consumption characteristics.

[0025] Further, for each user, the Mahalanobis distance of the electricity consumption behavior feature vector of the user to the cluster center to which the user belongs is calculated as the initial group deviation of the corresponding user. The Mahalanobis distance is used to represent the distance between a point and a distribution, and is used to preliminarily measure the difference between the behavior of a user and the average behavior of the homogeneous user group to which the user belongs.

[0026] Further, the initial group deviation is corrected using the first behavior stability index of the user to obtain the group deviation of the corresponding user.

[0027] Specifically, the group deviation = initial group deviation / (first behavior stability index + a), where a is a preset smoothing factor to prevent the denominator from being 0 and being unable to calculate, and the value of a can be set to a = 0.1, for example. For example, the initial deviation of user A = 2.5, the stability index = 0.95, and the final group deviation = 2.5 / (0.95+0.1) = 2.38. The corrected group deviation reflects both the static difference between the user and the group and the dynamic instability of the user's own behavior.

[0028] By extracting multi-dimensional features including load shape, voltage-current coupling relationship and electricity consumption entropy to form the electricity consumption behavior feature vector, and calculating the first behavior stability index reflecting the time sequence stability of the user behavior, a user behavior feature set that can comprehensively and deeply describe the user electricity consumption mode is constructed. On this basis, a clustering algorithm is used to divide users with similar electricity consumption behaviors into homogeneous user groups, and a dynamic and accurate reference benchmark is established for each user. Further, by calculating the group deviation of each user relative to the group center to which the user belongs, the difference between the user behavior and the group normal is effectively quantified, and the transition from extensive individual analysis to fine group intelligence analysis is realized, which lays a scientific and reliable benchmark for subsequent anomaly discrimination.

[0029] S20: Based on the total meter electric energy data of the transformer area, a transformer area line loss anomaly event is identified, and the anomaly event occurrence period is recorded. In the anomaly event occurrence period, the association strength of the electricity consumption behavior of each user with the transformer area line loss anomaly event is calculated as the event association degree, and the dynamic anomaly confidence is calculated in combination with the first behavior stability index of the user.

[0030] After identifying that the overall line loss anomaly event occurs in the transformer area, the main problem faced by the prior art is how to accurately correlate the macro transformer area anomaly with the micro specific user behavior in the time dimension. The traditional method often stops at the simple comparison of user power consumption in the abnormal period, and cannot deeply reveal the internal cause or strong correlation between user power consumption behavior change and transformer line loss anomaly event, resulting in the inability to effectively lock the root cause of the anomaly. In addition, user behavior itself has inherent instability, and the power consumption change of some users in the abnormal period may be due to their inherent random behavior habits, rather than the direct cause of the line loss anomaly. If the difference in behavior stability is ignored and only based on the correlation strength for judgment, it is easy to misjudge those users with irregular habits as the source of the anomaly, causing false positives.

[0031] The step S20 in the method provided by the embodiment of the present application comprises: calculating the daily line loss rate of the transformer area, and establishing a line loss rate baseline based on historical line loss rate data; when the daily line loss rate continuously exceeds a preset threshold or suddenly increases relative to the baseline, marking it as a line loss anomaly event; recording the start time, end time and anomaly intensity of the anomaly event; extracting the current sequence of each user and the total abnormal power sequence of the transformer area in the anomaly event period; for each user, calculating the Granger causality between the current sequence and the total abnormal power sequence of the transformer area, and taking the F test statistic as the event correlation degree; based on the event correlation degree and the first behavior stability index of the user, calculating the dynamic anomaly confidence of the corresponding user, wherein the dynamic anomaly confidence = event correlation degree x (1-first behavior stability index).

[0032] In the embodiment of the present application, based on the transformer area total meter electric energy data, the transformer area line loss anomaly event is identified, and the anomaly event period is recorded. In the anomaly event period, the correlation strength between the power consumption behavior of each user and the transformer line loss anomaly event is calculated as the event correlation degree, and the dynamic anomaly confidence is calculated in combination with the first behavior stability index of the user.

[0033] Specifically, first, the daily line loss rate of the transformer area is calculated, and a line loss rate baseline is established based on historical line loss rate data. The total daily active power sum of the transformer area total meter and the total daily active power sum of all users are calculated. Further, the formula: daily line loss rate = (transformer area total meter daily active power sum - all user daily active power sum) / transformer area total meter daily active power sum * 100% is used for calculation. For example, the total daily active power sum of a certain transformer area total meter is 1000 kWh, and the total daily active power sum of all users is 950 kWh, then the daily line loss rate is (1000-950) / 1000*100%=5%. The historical daily line loss rate data of the past 30 days is obtained, and the arithmetic mean value thereof is calculated as the baseline, for example, the obtained baseline value is 4.5%.

[0034] Further, when the daily line loss rate continuously exceeds the preset threshold or suddenly increases relative to the baseline, it is marked as a line loss abnormal event. For example, a preset threshold of 7% and a relative sudden increase threshold are set. For example, the relative sudden increase threshold is a sudden increase of 50% relative to the baseline. Continuously monitor the daily line loss rate: if the daily line loss rate exceeds 7%, it is marked as abnormal. If the daily line loss rate does not exceed 7%, but the increase relative to the baseline 4.5% exceeds 50% (i.e. line loss rate > 4.5%*(1+50%)=6.75%), it is also marked as abnormal.

[0035] Further, the start time, end time and abnormal intensity of the abnormal event are recorded, for example, from October 5, the line loss rate of the next three days is 6.8%, 7.5%, 8.0% respectively. 6.8% exceeds the sudden increase threshold, and the subsequent days all exceed the absolute threshold (7%), so a abnormal event starting from October 5 and ending on October 7 is marked. The abnormal intensity of the event is exemplarily represented by the difference between the average line loss rate during the event and the baseline: (6.8+7.5+8.0) / 3-4.5=2.43%.

[0036] Further, the current sequence of each user and the total abnormal power sequence of the transformer area in the period of the abnormal event are extracted. For example, for the identified abnormal event from October 5 to October 7, the current time series data of user A and user B in these three days, the total abnormal power time series data of the transformer area, and the Granger causality test of the current time series data of user A and user B and the total abnormal power time series data of the transformer area are performed, respectively. The Granger causality test is a statistical hypothesis test for determining whether the past values of a variable help predict the current values of another variable. The grangercausalitytests function in the statsmodels library of Python can be used as an example to complete the test. An F test statistic is obtained, and the test outputs an F statistic. The larger the F value, the greater the possibility that the current sequence of the user is the Granger cause of the total abnormal power sequence of the transformer area. For example, the F value of the test result of user A can be 9.5, and the F value of user B can be 1.2. This indicates that the correlation between the power consumption behavior of user A and the line loss abnormal event is much stronger than that of user B.

[0037] Further, based on the event correlation degree of the user and the first behavior stability index, the dynamic abnormal confidence of the corresponding user is calculated, wherein the dynamic abnormal confidence = event correlation degree x (1-first behavior stability index). For example, the event correlation degree of user A = 9.5, and the first behavior stability index = 0.65. Then the dynamic abnormal confidence = 9.5 x (1-0.65) = 9.5 x 0.35 = 3.325. If a user with an unstable behavior itself, i.e., a low first behavior stability index, is highly correlated with the abnormal event, the credibility of the user causing the abnormality, i.e., the dynamic abnormal confidence, is higher.

[0038] By first accurately identifying the transformer area line loss abnormal event and its occurrence period, the focus of analysis is concentrated on the key time window. On this basis, by calculating the event correlation degree between the power consumption behavior of each user and the transformer area line loss abnormal event, effective tracing from macro abnormality to micro user behavior is realized, and users with suspicious changes in behavior synchronously appearing at the time of abnormality can be captured. More importantly, the first behavior stability index is integrated into the analysis, and the dynamic abnormal confidence is calculated to correct the credibility of the event correlation degree. This mechanism effectively reduces the risk of misjudgment of users with unstable power consumption habits accidentally showing high correlation degree in the abnormal period, so that the dynamic abnormal confidence obtained finally can more truly reflect the potential causal possibility between the user behavior and the line loss abnormality.

[0039] S30: Based on the group deviation degree and the dynamic abnormal confidence of each user, a comprehensive abnormal score of each user is calculated, and based on a preset comprehensive abnormal score threshold, an abnormal user is screened out.

[0040] After calculating the group deviation degree and dynamic abnormal confidence in two dimensions of static deviation of user behavior from the group and dynamic association of user behavior with specific abnormal events respectively, how to scientifically and reasonably integrate the two types of indicators with different properties but complementary to each other becomes the final key to accurately locate abnormal users.

[0041] The step S30 in the method provided by the embodiment of the application comprises: The group deviation degree and the dynamic abnormal confidence are normalized respectively; The normalized group deviation degree and the normalized dynamic abnormal confidence are weighted and summed using preset weight coefficients to obtain a comprehensive abnormal score; A threshold value of the comprehensive abnormal score is set; A user whose comprehensive abnormal score exceeds the threshold value is marked as an abnormal user; An abnormal user positioning report is generated, including an abnormal user list, a comprehensive abnormal score, and a main abnormal feature description; Based on the abnormal user positioning report, an operation and maintenance work order is output to guide on-site troubleshooting.

[0042] In the embodiment of the application, the comprehensive abnormal score of each user is calculated based on the group deviation degree and the dynamic abnormal confidence of each user, and the abnormal users are screened based on the preset threshold value of the comprehensive abnormal score.

[0043] Specifically, first, the group deviation degree and the dynamic abnormal confidence are normalized respectively. Exemplarily, the minimum-maximum normalization method is used for normalization, and the formula is: normalized value=(original value of a certain user-minimum value among all users) / (maximum value among all users-minimum value among all users). The normalized group deviation degree and the normalized dynamic abnormal confidence are obtained by calculating the group deviation degree and the dynamic abnormal confidence of all users respectively.

[0044] Further, the normalized group deviation degree and the normalized dynamic abnormal confidence are weighted and summed using preset weight coefficients to obtain a comprehensive abnormal score. For example, comprehensive abnormal score=(weight coefficient W1*normalized group deviation degree)+(weight coefficient W2*normalized dynamic abnormal confidence). The weight coefficients W1 and W2 are preset values, and satisfy W1+W2=1. For example, W1=0.4 and W2=0.6 are set, which means that more attention is paid to the dynamic performance of the user in the specific abnormal event. For example, the normalized group deviation degree of user A is 0.625, and the normalized dynamic abnormal confidence is 0.82. Then the comprehensive abnormal score of user A is (0.4*0.625)+(0.6*0.82)=0.25+0.492=0.742. The higher the score is, the greater the possibility that the user causes the line loss abnormality is.

[0045] Further, a comprehensive abnormal score threshold is set, for example, the comprehensive abnormal score threshold is set to 0.7 according to historical abnormal analysis experience.

[0046] Further, the user whose comprehensive abnormal score exceeds the threshold is marked as an abnormal user.

[0047] Further, an abnormal user positioning report is generated, including an abnormal user list, a comprehensive abnormal score and a main abnormal feature description. For example, for user A, the comprehensive abnormal score is 0.742, and the abnormal feature description is "the user's behavior pattern deviates significantly from the group to which he belongs, and shows high correlation in recent line loss abnormal events, and the stability of his own power consumption behavior is low". The positioning report of all abnormal users is generated.

[0048] Further, based on the abnormal user positioning report, an operation and maintenance work order is output to guide on-site troubleshooting. The operation and maintenance work order clearly indicates the target user and the core abnormal features that need to be checked on site, and the operation and maintenance personnel can prioritize and target on-site inspection accordingly, greatly improving the efficiency and accuracy of troubleshooting.

[0049] By constructing a comprehensive abnormal score, the information of group deviation degree and dynamic abnormal confidence is effectively fused. By normalizing and weighted summing the two indexes with preset weights, a multi-angle and quantitative comprehensive evaluation of the user abnormality possibility is realized, avoiding the arbitrariness of relying on a single criterion, so that the final judgment result is more comprehensive and objective. Based on the preset comprehensive abnormal score threshold, abnormal users can be screened, which can more accurately identify users who deviate from the norm in behavior patterns and show high correlation in specific abnormal events, thereby greatly improving the accuracy and reliability of the positioning result.

[0050] In summary, the embodiments of the present application have at least the following technical effects: The application provides a transformer area line loss simulation method based on multi-source data. By comprehensively utilizing the time series data such as voltage, current and active power of each user and total meter in the transformer area, a complete technical solution from feature extraction, behavior clustering to abnormal event correlation analysis is constructed, which significantly improves the accuracy and reliability of transformer area line loss abnormal user identification. Specifically, by extracting multi-dimensional features such as load form, voltage-current coupling relationship and power consumption entropy, and generating a user behavior feature set, the application can more comprehensively represent the real power consumption mode of the user; then, by using clustering algorithm to divide homogeneous user groups and calculate group deviation, the application can effectively identify abnormal individuals whose behavior mode deviates from the normal group; at the same time, by correlating the line loss abnormal event with the user behavior in time sequence, calculating the event correlation degree and dynamic abnormal confidence, the application can sensitively capture the user with suspicious behavior in a specific abnormal period; finally, the application generates a comprehensive abnormal score by comprehensively considering the group deviation and dynamic abnormal confidence of the user, which provides a scientific and quantitative basis for accurately screening abnormal users. The synergistic effect of a series of technical means enables the application to effectively distinguish normal load fluctuation from potential electricity stealing, metering failure and other management line loss, greatly reducing the false positives and false negatives commonly seen in traditional methods, and improving the credibility of the identification results. Compared with the traditional method, the technical solution provided by the application significantly overcomes the limitations of relying only on simple data comparison and static threshold judgment, realizes the deep mining and dynamic evaluation of user power consumption behavior, and achieves the technical effect of comprehensively improving the fine and intelligent level of transformer area line loss management, which can provide clear and specific abnormal user positioning report and operation work order guidance for operation and maintenance personnel, thereby greatly improving the pertinence and efficiency of on-site investigation, providing strong technical support for ensuring the stable operation of power grid and reducing power loss.

[0051] It should be noted that the above sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0052] The above only describes the preferred embodiments of the application and does not limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

[0053] The specification and drawings are, of course, to be regarded in an illustrative rather than a restrictive sense. It is to be understood that any such modifications, variations, combinations or equivalents that fall within the scope of the application are intended to be embraced herein.

Claims

1. A method for simulating the line loss of a transformer area based on multi-source data, characterized in that, The method comprises: Collecting the electric energy data of each user in the transformer area and the total meter electric energy data, the electric energy data at least including time series data of voltage, current and active power, extracting the electricity behavior feature vector and the first behavior stability index of each user based on the electric energy data of each user, obtaining the user behavior feature set, clustering the user behavior feature set based on the preset clustering algorithm, obtaining a plurality of homogeneous user groups and the group deviation degree of each user, wherein the first behavior stability index of each user is extracted by: Based on the electricity behavior feature vector of the user in the current period and the average value of the historical feature vector in the past M periods, the cosine similarity is calculated, and the value of the similarity is taken as the first behavior stability index; Based on the total meter electric energy data, identifying the transformer area line loss abnormal event and recording the abnormal event period, calculating the association strength of the electricity behavior of each user with the transformer area line loss abnormal event as the event association degree in the abnormal event period, and calculating the dynamic abnormal confidence degree combining the first behavior stability index of the user, including: Extracting the current sequence of each user and the total abnormal power sequence in the abnormal event period; For each user, the Granger causality relationship between the current sequence and the total abnormal power sequence is calculated, and the F test statistic is taken as the event association degree; Based on the event association degree and the first behavior stability index of the user, the dynamic abnormal confidence degree of the corresponding user is calculated, wherein the dynamic abnormal confidence degree = event association degree × (1-first behavior stability index); Based on the group deviation degree and the dynamic abnormal confidence degree of each user, the comprehensive abnormal score of each user is calculated, and based on the preset comprehensive abnormal score threshold, the abnormal user is screened out.

2. The method of claim 1, wherein, Collecting the electric energy data of each user in the transformer area and the total meter electric energy data, the electric energy data at least including time series data of voltage, current and active power, including: Through the electricity information collection system, the time series data of voltage, current and active power uploaded by the smart meters in the transformer area is obtained; Through the distribution automation system, the time series data of voltage, current and active power of the total meter in the same time period is obtained; The collected time series data is subjected to data cleaning and time series alignment processing to form a multi-source data set. 3.The method of claim 1, wherein, Based on the electric energy data of each user, the electricity behavior feature vector and the first behavior stability index of each user are extracted to obtain the user behavior feature set, including: Extracting load shape features, including calculating the dynamic time warping distance vector of the user daily load curve, and statistical features including maximum value, minimum value, average value, standard deviation and peak-valley difference; Extracting voltage-current coupling relationship features, including calculating the average value and standard deviation of the power factor of the user at different power levels, and calculating the windowed cross-correlation coefficient of the voltage sequence and the current sequence; Extracting electricity entropy features, including calculating the sample entropy of the active power sequence of the user; Based on the cosine similarity, the first behavior stability index of each user is obtained; The load form feature, the voltage-current coupling relationship feature, and the electricity consumption entropy feature are combined to form an electricity consumption behavior feature vector of each user, and the first behavior stability index is used to form a user behavior feature set.

4. The method of claim 1, wherein, Based on a preset clustering algorithm, the user behavior feature set is clustered to obtain a plurality of homogeneous user groups and a group deviation degree of each user, including: The K-means++ clustering algorithm is used to cluster the electricity consumption behavior feature vectors of all users to obtain K homogeneous user groups. For each user, the Mahalanobis distance of the electricity consumption behavior feature vector of the user to the clustering center to which the user belongs is calculated as an initial group deviation degree of the corresponding user. The first behavior stability index of the user is used to correct the initial group deviation degree to obtain the group deviation degree of the corresponding user.

5. The method of claim 4, wherein, The first behavior stability index of the user is used to correct the initial group deviation degree, and the correction method is: Group deviation degree = initial group deviation degree / (first behavior stability index + α), wherein α is a preset smoothing factor.

6. The method of claim 1, wherein, Based on the total meter electric energy data of the transformer area, a transformer area line loss abnormal event is identified, and the abnormal event occurrence period is recorded, including: The daily line loss rate of the transformer area is calculated, and a line loss rate baseline is established based on historical line loss rate data. When the daily line loss rate continuously exceeds a preset threshold or suddenly increases relative to the baseline, it is marked as a line loss abnormal event. The start time, end time, and abnormal intensity of the abnormal event are recorded.

7. The method of claim 1, wherein, Based on the group deviation degree and the dynamic abnormal confidence degree of each user, a comprehensive abnormal score of each user is calculated, including: The group deviation degree and the dynamic abnormal confidence degree are normalized. The normalized group deviation degree and the dynamic abnormal confidence degree are weighted and summed using a preset weight coefficient to obtain the comprehensive abnormal score.

8. The method of claim 1, wherein, Based on a preset comprehensive abnormal score threshold, abnormal users are screened out, including: The comprehensive abnormal score threshold is set. Users with a comprehensive abnormal score exceeding the threshold are marked as abnormal users. An abnormal user positioning report is generated, including an abnormal user list, a comprehensive abnormal score, and a main abnormal feature description. Based on the abnormal user positioning report, an operation and maintenance work order is output to guide on-site troubleshooting.

Images