A method for identifying and diagnosing high loss causes of a transformer substation

Abstract

The application discloses a kind of high loss cause identification diagnosis methods of transformer area, belong to electric power system operation and maintenance field, solve the problem that existing method is poorly adapted to complex transformer area, high loss cause is not accurate positioning.Method includes: the basic operation data such as line operation parameter, commercial user load data etc.are collected;Determine the aging line noise feature label based on line aging related parameter matching noise feature library;Threshold is dynamically generated according to label, and get data after purification by adaptive Kalman filtering;Detect data integrity, and complete breakpoint data using LSTM model;Based on the data and transformer power supply range data, extract regional load characteristics and commercial user electricity abnormal characteristics;Fusion multiclass data, set weight using analytic hierarchy process and calculate comprehensive abnormal index;High loss cause is positioned in combination with the index and each link result, and diagnostic report is generated.The application is targeted to process line noise, data breakpoint and other interference, and improve diagnostic accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power operation and maintenance, in particular to a method for identifying and diagnosing causes of high loss in a transformer area. BACKGROUND

[0002] In the operation and maintenance management of a power system, the line loss rate of a transformer area is a core indicator for measuring power supply efficiency. High loss in a transformer area not only causes waste of power resources, but also increases the operating cost of a power supply enterprise. Therefore, accurately identifying and diagnosing the causes of high loss in a transformer area is of great significance to the efficient operation of a power system.

[0003] Currently, existing methods for identifying and diagnosing the causes of high loss in a transformer area are mostly designed based on general scenarios and lack targeted adaptation to the complex scene characteristics of different types of transformer areas. In actual applications, complex situations such as data noise caused by aging of lines in old urban areas, data collection breakpoints in remote rural transformer areas, and frequent instantaneous fluctuations in commercial user power consumption in high-density commercial transformer areas can significantly interfere with data quality and feature analysis accuracy during the diagnosis process. Existing methods are difficult to effectively handle these scene-specific interference factors, resulting in deviation in the positioning of high loss causes, insufficient accuracy and reliability of the diagnosis results, and inability to provide accurate decision-making basis for high loss management in a transformer area. SUMMARY

[0004] To solve the technical problems in the prior art of low sensitivity to changes in the relationship between a house and a transformer, difficulty in quickly capturing sudden changes in the relationship between a house and a transformer through abnormal changes in power consumption data, and lack of precise verification link combined with geographic spatial information, resulting in lag and deviation in the identification of the relationship between a house and a transformer, and further causing distortion of line loss statistical data and affecting the basic accuracy of high loss cause diagnosis, the present application provides a method for identifying and diagnosing the causes of high loss in a transformer area.

[0005] The technical solution adopted by the present application is: a method for identifying and diagnosing the causes of high loss in a transformer area, comprising the following steps:

[0006] A method for identifying and diagnosing the causes of high loss in a transformer area, comprising the following steps:

[0007] Step 1: Collecting basic operation data of the target transformer area, the basic operation data including line operation parameters, user power consumption data, transformer power supply range data and historical power consumption data, the user power consumption data covering load data of commercial users.

[0008] Step 2: Based on the aging-related parameters in the line operation parameters, matching a preset aging line noise feature library to determine the aging line noise feature label of the target transformer area.

[0009] Step 3, dynamically generate noise filtering threshold according to the aging line noise feature label, use adaptive Kalman filtering algorithm to filter noise of the basic operation data, get the purified operation data.

[0010] Step 4, integrity detection is carried out on the purified operation data, missing data is identified and continuous missing duration is counted, if the continuous missing duration exceeds the preset breakpoint threshold, it is determined as breakpoint data, LSTM time series prediction model is used, trend completion model is constructed based on the historical power consumption data of the user before the breakpoint, and the completion data is generated.

[0011] Step 5, based on the completion data and transformer power supply range data, the regional load characteristics representing the load distribution state of the transformer area are extracted, and the power consumption state characteristics of commercial users are extracted to obtain power consumption abnormal characteristics.

[0012] Step 6, fusion of the purified operation data, completion data, regional load characteristics and power consumption abnormal characteristics, the weight coefficients of each data are set by using analytic hierarchy process, and the comprehensive abnormal index is calculated.

[0013] Step 7, based on the comprehensive abnormal index and the analysis results of each link, the causes of high loss of transformer area are located, and the transformer area high loss cause diagnosis report is generated.

[0014] Preferably, the step 3 and the step 4 further comprise the following contents:

[0015] Based on the user power consumption data in the purified operation data, the ratio of current change amount in a set time interval to historical average current change amount in the same period is calculated to obtain the household transformer relationship mutation rate;

[0016] The household transformer relationship mutation rate is compared with the preset trigger threshold, if it exceeds the preset trigger threshold, the GIS system is called to locate the user power consumption address and the boundary of the transformer power supply range is compared, whether the household transformer relationship is changed is confirmed, if the change is confirmed, the household transformer association data set is updated.

[0017] Preferably, the noise filtering threshold in step 3 includes frequency cutoff threshold and amplitude threshold, the adaptive Kalman filtering algorithm realizes noise filtering by dynamically adjusting Kalman gain, and the trend similarity of the purified operation data and historical normal data in the same period is calculated after filtering.

[0018] Preferably, the current change amount is the absolute value of the difference between the current value at the end of the set time interval and the current value at the beginning, and the historical average current change amount is the average value of the current change amount of a set number of time intervals in the same period.

[0019] Preferably, the transformer power supply range in step 2 is defined by vector boundary data, and the user transformer association data set includes user identification, transformer identification, and home relationship effective timestamp.

[0020] Preferably, the LSTM time series prediction model in step 4 uses historical power consumption data of multiple complete cycles before the breakpoint as training data, and the completed data is verified for trend consistency with effective data before and after the breakpoint through the Pearson correlation coefficient.

[0021] Preferably, the step 5 includes the following contents:

[0022] Based on the user power consumption address association line segment identifier, the line segment identifier is divided into first end identifier, middle end identifier and end identifier according to power supply radius;

[0023] According to the average load data in the set time, the deviation rate of each segment average load and the overall average load of the transformer area is calculated, and the deviation rate is taken as the regional load characteristics;

[0024] The historical load data of commercial users is extracted from the completed data, the instantaneous fluctuation characteristics are obtained, the commercial user load fluctuation feature library is constructed, the dynamic instantaneous fluctuation threshold is generated based on the feature library, the dynamic instantaneous fluctuation threshold includes fluctuation amplitude threshold and fluctuation duration threshold, and the fluctuation duration is the time difference from the load deviating from the normal stable value to recovering to stable;

[0025] The real-time load fluctuation data of commercial users is compared with the dynamic instantaneous fluctuation threshold, and the electricity consumption abnormal characteristics are obtained combined with the peak-valley electricity consumption ratio.

[0026] Preferably, in step 6, each data needs to be standardized before fusion, the standardization adopts range standardization method, and the weight coefficient is calculated by comparing the influence degree of each data on the diagnosis result through analytic hierarchy process.

[0027] Preferably, in step 7, when positioning the high loss cause of transformer area, the correlation matching relationship of the numerical interval of comprehensive abnormal index and each link analysis result is combined, and the diagnosis report includes the line loss influence proportion corresponding to each cause.

[0028] Preferably, the preset trigger threshold and the preset breakpoint threshold are determined through statistical analysis of transformer operation and maintenance case data, and the preset trigger threshold and the preset breakpoint threshold are dynamically adjusted according to the type of transformer area.

[0029] The beneficial effects of the present application are: for the line aging noise of the old urban area transformer area, a dynamic filtering threshold is generated by matching the aging line noise feature library, the noise is accurately filtered by combining the adaptive Kalman filtering algorithm, the data validity is ensured by trend similarity verification, the noise filtering is more thorough and the effective data distortion is avoided; for the data breakpoint of the remote transformer area in rural areas, the historical data trend is completed based on the LSTM time series prediction model, and the rationality of the completion is verified by the Pearson correlation coefficient, the data completion accuracy is greatly improved, and the data availability is significantly improved; for the load fluctuation of high-density commercial transformer area, a dynamic instantaneous fluctuation threshold is constructed based on the historical load characteristics of commercial users, combined with the peak-valley electricity consumption ratio to judge the abnormality, and the abnormal electricity consumption misjudgment rate is effectively reduced. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 The present application is a flowchart of the method. DETAILED DESCRIPTION

[0031] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0032] Embodiment

[0033] A transformer area high loss cause identification and diagnosis method, as shown in Figure 1 , comprising the following steps:

[0034] Step 1, collecting the basic operation data of the target transformer area, the basic operation data including line operation parameters, user electricity consumption data, transformer power supply range data and historical electricity consumption data, the user electricity consumption data covering the load data of commercial users.

[0035] It should be noted that the target transformer area refers to a specific power supply area to be subjected to high loss cause identification and diagnosis, and its range is defined by the power supply coverage boundary of the corresponding transformer. The basic operation data refers to the core data set supporting the whole process analysis of the transformer area high loss cause diagnosis, and is the basic data source for subsequent steps. The line operation parameter refers to the parameter describing the operation state of the transformer area power supply line, including but not limited to line operation life, resistance value, insulation performance parameter, etc. The user electricity consumption data refers to the electricity consumption state data of all electricity users in the transformer area, including real-time current, power, voltage, electricity consumption, etc., which covers the load data of commercial users (such as supermarkets, office buildings, catering users). The transformer power supply range data refers to the vector data that clearly defines the geographical boundary of the corresponding transformer power supply, which is used to define the attribution relationship between the user and the transformer. The historical electricity consumption data refers to the electricity consumption record data of the target transformer area and users in the past set time period, including electricity consumption data of different periods (such as day, week, month) and different time periods (such as peak segment, valley segment).

[0036] The basic operation data is the core basis for diagnosing the high-loss causes of the transformer area. All subsequent steps such as noise filtering, household and transformer relationship monitoring, data completion, and load analysis depend on the data. If the basic operation data is incomplete and inaccurate, the subsequent analysis results will be biased or even invalid. Therefore, this step is the prerequisite for ensuring the feasibility of the entire diagnosis scheme.

[0037] In the specific implementation process, the line operation parameters are collected through the transformer area line operation management system. The system is connected with the sensors and operation record modules on the line to synchronize the operation state data of the line in real time. The user power consumption data is collected at a set time interval by means of smart meters and power consumption information collection terminals to ensure the real-time nature of the data. The transformer power supply range data corresponding to the target transformer area is retrieved from the power GIS system to determine the geographic boundary information. The historical power consumption data of the target transformer area and users is extracted from the power data storage center to filter out the time dimension data meeting the subsequent analysis requirements.

[0038] In step 2, based on the aging-related parameters in the line operation parameters, the preset aging line noise feature library is matched to determine the aging line noise feature label of the target transformer area.

[0039] It should be noted that the aging-related parameters refer to the parameters in the line operation parameters that are directly related to the aging degree of the line, mainly including the line operation life and resistance deviation value. The aging line noise feature library refers to a database that stores the noise features corresponding to lines in different aging states, including the noise frequency, fluctuation amplitude, and other feature information of various aging lines such as high-aging lines and unstable resistance lines. The aging line noise feature label refers to the identification information generated based on the matching results of the line aging-related parameters of the target transformer area and the noise feature library, which is used to identify the noise type of the line of the transformer area, such as "high-frequency noise + amplitude fluctuation noise".

[0040] It is considered that the lines in old urban transformer areas are prone to data noise due to aging, and the noise features corresponding to lines in different aging states are different. By matching the preset noise feature library to determine the feature label, the noise type of the line of the target transformer area can be accurately located, which provides a basis for subsequent targeted noise filtering and avoids data distortion caused by blind filtering.

[0041] In the specific implementation process, the aging-related parameters are filtered out from the line operation parameters collected in step 1 to determine the aging state of the line of the target transformer area. The preset aging line noise feature library is called to compare the filtered aging-related parameters with the parameter thresholds and corresponding noise features of various aging lines in the feature library. According to the comparison results, the noise feature type corresponding to the line of the transformer area is determined, and the corresponding aging line noise feature label is generated.

[0042] Step 3, dynamically generate noise filtering threshold according to the aging line noise feature label, use adaptive Kalman filtering algorithm to filter noise from the basic operation data, and obtain the purified operation data.

[0043] It should be noted that the noise filtering threshold refers to the critical value set according to the noise feature label to distinguish between valid data and noise data. Different noise features correspond to different filtering thresholds, such as frequency cutoff threshold, amplitude threshold, etc. The adaptive Kalman filtering algorithm refers to an algorithm that can dynamically adjust the filtering parameters (Kalman gain) according to the noise characteristics of the input data, realize adaptive filtering of different types of noise, and is suitable for data purification processing in complex noise environment. The purified operation data refers to the basic operation data after noise filtering, which removes invalid noise interference, and its data accuracy and reliability are significantly improved. The trend similarity refers to an index for judging the consistency of the filtered data and the historical normal data trend, which is calculated using the cosine similarity algorithm.

[0044] Considering that the mixed line aging noise in the basic operation data will interfere with the result accuracy of subsequent household variable relationship analysis, abnormal electricity identification, etc. Based on the noise feature label, the filtering threshold is dynamically generated, and then the adaptive Kalman filtering algorithm is used for targeted filtering, which can effectively remove noise interference while retaining valid data to the greatest extent. At the same time, the filtering effect is verified by calculating the trend similarity to ensure that the purified data conforms to the historical electricity law and provides high-quality data support for subsequent analysis.

[0045] In the specific implementation process, based on the aging line noise feature label generated in step 2, combined with the attributes of the corresponding noise in the feature library, the noise filtering threshold that adapts to the noise type is dynamically generated.

[0046] Call the adaptive Kalman filtering algorithm, take the basic operation data collected in step 1 as the algorithm input data, and take the generated noise filtering threshold as the algorithm constraint condition; start the algorithm to run, filter the noise in the basic operation data by dynamically adjusting the Kalman gain; output the filtered data set, which is the preliminary purified data; call the historical same period normal data in the near set time period, calculate the trend similarity of the preliminary purified data and the historical data using the cosine similarity algorithm; if the trend similarity reaches the pre-set confidence, the preliminary purified data is determined as the purified operation data; otherwise, reduce the fluctuation amplitude threshold and re-filter until the confidence requirement is met, if multiple adjustments still do not meet the requirement, trigger the line maintenance warning.

[0047] The core formula of the adaptive Kalman filtering algorithm is as follows:

[0048]

[0049] wherein, The filtered data, i.e., the corresponding data points in the initially purified data, This refers to the original collected data, i.e., the corresponding data points in the basic operational data. For adaptive Kalman gain, it is dynamically adjusted based on noise characteristics. For the measurement matrix, This is the predicted data from the previous moment.

[0050] The formula for calculating trend similarity (cosine similarity) is as follows:

[0051]

[0052] in, For trend similarity, The first step in the initial purification of data Data points, This is the [number]th [period] of normal data for the same period in history. Data points, This represents the total number of data points.

[0053] Considering that changes in the relationship between users and transformers (such as users illegally connecting to transformers across different areas) can cause abnormal and sudden changes in the user's electricity current, the following content is also included between steps 3 and 4:

[0054] Based on the user electricity consumption data in the purified operation data, the ratio of the current change within a set time interval to the historical average current change during the same period is calculated to obtain the household-to-electricity relationship abrupt change rate.

[0055] It should be noted that the set time interval refers to a pre-defined time period used to calculate current changes. This period is set based on the frequency of electricity data collection and the response requirements for abnormal user-transformer relationships. Current change refers to the change in the user's electricity current within the set time interval, specifically the absolute value of the difference between the current value at the end of the interval and the initial current value. The historical average current change refers to the average of multiple current changes within the same period in the past set time intervals, used as a benchmark to determine whether the current current change is abnormal. The user-transformer relationship abrupt change rate is the ratio of the current change within the current set time interval to the historical average current change rate, used to quantify the degree of abrupt change in the user's electricity current, thereby determining whether the user-transformer relationship may have changed.

[0056] Considering that changes in the relationship between a user and a transformer (such as unauthorized wiring across different transformer substations) can cause abnormal fluctuations in the user's current, this phenomenon can be quantified into a specific indicator by calculating the fluctuation rate of the user-transformer relationship. This indicator can intuitively reflect the degree of difference between the current current change and the historical normal change, providing a quantitative basis for subsequent judgment on whether the user-transformer relationship has changed.

[0057] In the specific implementation process:

[0058] Extract the current data from the post-purification operation data obtained from step 3;

[0059] At a set time interval, calculate the current change amount in each time interval, the calculation formula is as follows:

[0060]

[0061] Among them, is the current change amount of the current set time interval, is the current value at the end of the time interval, is the current value at the beginning of the time interval;

[0062] From the historical power consumption data collected in step 1, extract the current data corresponding to the user, calculate the historical average current change amount, the calculation formula is as follows:

[0063]

[0064] Among them, is the historical average current change amount, is the current change amount of the jth historical time interval, m is the total number of statistical historical time intervals;

[0065] Divide the current time interval current change amount by the historical average current change amount to get the household variable relationship mutation rate, the calculation formula is as follows:

[0066]

[0067] Among them, is the household variable relationship mutation rate.

[0068] Compare the household variable relationship mutation rate with the preset trigger threshold value, if it exceeds the preset trigger threshold value, call the GIS system to locate the user's power consumption address and compare the boundary with the transformer power supply range, confirm whether the household variable relationship has changed, if it is confirmed to change, update the household variable relationship dataset.

[0069] It should be noted that the preset trigger threshold value refers to the household variable relationship mutation rate critical value preset for judging whether the household variable relationship is likely to change, which is based on a large number of household variable relationship change case data statistical analysis.

[0070] GIS system refers to geographic information system, which is used here to obtain the geographic coordinates of the user's power consumption address and the vector boundary data of the transformer power supply range, to realize accurate comparison of spatial position. The household variable relationship dataset refers to a data set that records the ownership relationship of each user in the transformer area and the corresponding power supply transformer, including user identification, transformer identification, ownership relationship effective time, etc.

[0071] It is considered that when the household change relationship mutation rate exceeds the preset trigger threshold, it indicates that the change of the user's electricity current has exceeded the normal range, and there is a possibility of change of the household change relationship. Through spatial position comparison by the GIS system, it can be verified whether the possibility is the actual situation, and if the change is confirmed, the household change relationship data set is updated in time to ensure that all subsequent analyses based on the household change relationship are based on accurate attribution information.

[0072] In the specific implementation process, the calculated household change relationship mutation rate is compared with the preset trigger threshold; if the household change relationship mutation rate exceeds the preset trigger threshold, the GIS system is called to obtain the electricity address geographic coordinates of the user; the supply range vector boundary of the corresponding transformer is extracted from the transformer supply range data collected in step 1; the user's electricity address geographic coordinates are compared with the transformer supply range vector boundary in space, and it is judged whether the user exceeds the original transformer supply range; if the judgment result is that it exceeds, it is confirmed that the household change relationship has changed, the transformer attribution information corresponding to the user in the household change relationship data set is immediately updated, and a change timestamp is added; if the judgment result is that it does not exceed, the original household change relationship data set is maintained unchanged.

[0073] Step 4, integrity detection is performed on the purified running data, missing data is identified and continuous missing time length is counted, if the continuous missing time length exceeds the preset breakpoint threshold, it is determined as breakpoint data, an LSTM time series prediction model is used, a trend completion model is constructed based on the historical electricity data of the user before the breakpoint, and completion data is generated.

[0074] It should be noted that the integrity detection refers to checking the continuity of the purified running data, identifying whether there is data missing and the specific situation of the missing data. The continuous missing time length refers to the length of time that the same type of data (such as the current data of a certain user) is continuously not collected in the purified running data. The preset breakpoint threshold refers to the continuous missing time length critical value preset for defining ordinary data missing and data breakpoint, which is set based on the data collection period and analysis requirements. The breakpoint data refers to the missing data corresponding to the data breakpoint when the continuous missing time length exceeds the preset breakpoint threshold, which has a greater impact on analysis and needs to be completed. The LSTM time series prediction model refers to the long short-term memory neural network model, which has the ability to capture long-term dependencies of time series data and is suitable for trend prediction and data completion based on historical time series data. The completion data refers to the usable data obtained by trend completion of the breakpoint data by the LSTM time series prediction model, which is used to replace the breakpoint data for subsequent analysis. The trend consistency refers to the index used to judge the degree of consistency of the completion data with the effective data before and after the breakpoint, which is measured by the Pearson correlation coefficient.

[0075] Considering the unstable communication signal in remote rural areas, data collection breakpoints are prone to occur, leading to discontinuous data, and further affecting the subsequent load analysis, abnormal identification and other links. The trend completion model is constructed based on the historical electricity consumption data before the breakpoint using the LSTM time series prediction model. The trend correlation of time series data can be used to generate completed data that conforms to the actual electricity consumption rules. The trend consistency of the completed data and the effective data before and after the breakpoint is verified by the Pearson correlation coefficient to ensure the rationality of the completed data and restore the continuity of the data.

[0076] In the specific implementation process, the integrity of the purified operation data obtained in step 3 is detected, and various data are traversed and the missing situation and continuous missing duration are recorded; the continuous missing duration is compared with the preset breakpoint threshold to screen out breakpoint data; from the historical electricity consumption data collected in step 1, the electricity consumption data of multiple complete periods before the breakpoint of the user are extracted as the training data of the LSTM time series prediction model; the LSTM time series prediction model is constructed, the training data is input for model training, and the model parameters are determined; the last segment of effective data before the breakpoint is input into the trained model to generate the preliminary completed data corresponding to the breakpoint data; the trend consistency (Pearson correlation coefficient) of the preliminary completed data and the last segment of effective data before the breakpoint and the first segment of effective data after the breakpoint is calculated, if it meets the preset confidence requirement, the preliminary completed data is determined as the completed data; otherwise, it is marked as an uncompleted breakpoint and triggers manual verification.

[0077] The core formula of the LSTM time series prediction model is as follows:

[0078]

[0079] wherein, is the completed data, is the cell state of the LSTM unit, , is the model parameter (trained based on historical data), is the activation function, and tanh is the hyperbolic tangent function.

[0080] The formula for calculating the Pearson correlation coefficient is as follows:

[0081]

[0082] wherein, is the Pearson correlation coefficient, is the effective data before and after the breakpoint, is the preliminary completed data, is the average value of the effective data, is the average value of the preliminary completed data, is the total number of data points involved in the calculation.

[0083] Step 5, based on the completion data and transformer power supply range data, extracting the regional load characteristics representing the load distribution state of the transformer area, and extracting the power consumption state characteristics of commercial users to obtain power consumption anomaly characteristics.

[0084] Specifically, step 5 includes the following: based on the user power consumption address associated line segment identifier, the average load data in the set time is counted by segment, the deviation rate of each segment average load and the overall average load of the transformer area is calculated, and the regional load deviation result is obtained; at the same time, the historical load data of commercial users is extracted from the purified operation data, the instantaneous fluctuation characteristics are obtained to construct the commercial user load fluctuation feature library, the dynamic instantaneous fluctuation threshold is generated based on the feature library, the real-time load fluctuation data of commercial users is compared with the dynamic instantaneous fluctuation threshold, and the peak-valley power consumption ratio is combined to obtain the abnormal power consumption judgment result of commercial users.

[0085] It should be noted that the line segment identifier refers to the unique identifier information assigned to each paragraph after the transformer supply line is divided into different paragraphs (such as the first end, the middle end, and the end) according to the line supply radius, which is used to distinguish different line paragraphs.

[0086] Segment average load refers to the average power consumption load carried by a certain line segment in a set time, which is calculated by the power consumption load data of all users in the segment. The overall average load of the transformer area refers to the average power consumption load of the target transformer area in a set time, which is calculated by the power consumption load data of all users in the transformer area. The regional load deviation rate refers to a quantitative indicator reflecting the deviation degree of the average load of each line segment from the overall average load of the transformer area, which is used to judge whether the load distribution is balanced. The regional load deviation result refers to the judgment result obtained based on whether the regional load deviation rate exceeds the pre-set reasonable range. The historical load data of commercial users refers to the power consumption load record data of commercial users in the past set time period extracted from the purified operation data. The instantaneous fluctuation characteristics refer to the mutation characteristics of the power consumption load of commercial users in a short time, including the maximum fluctuation amplitude, fluctuation duration and other parameters. The commercial user load fluctuation feature library refers to a database storing the instantaneous fluctuation characteristics of different types of commercial users (such as supermarkets, office buildings, and restaurants), which is the basis for generating dynamic fluctuation thresholds. The dynamic instantaneous fluctuation threshold refers to the load fluctuation threshold dynamically adjusted according to the historical fluctuation characteristics of different types of commercial users based on the commercial user load fluctuation feature library, including the fluctuation amplitude threshold and the fluctuation duration threshold. The peak-valley power consumption ratio refers to the ratio of the power consumption in the peak segment to the power consumption in the valley segment of commercial users, which is used to assist in judging whether the load fluctuation is abnormal. The abnormal power consumption judgment result of commercial users refers to the result of judging whether the power consumption of commercial users is abnormal by comparing the real-time load fluctuation data of commercial users with the dynamic instantaneous fluctuation threshold and combining the peak-valley power consumption ratio.

[0087] Considering that the load distribution is uneven in remote rural areas due to line layout and other factors, the problem can be accurately identified by calculating the regional load deviation rate; in high-density commercial areas, the instantaneous fluctuation of commercial user power load is frequent, and if a fixed threshold is used for judgment, it is easy to produce false judgments, and based on the historical fluctuation characteristics, a dynamic threshold is generated and combined with the peak-valley power consumption ratio to effectively reduce the false judgment rate. This step solves the core problems of the two types of scenarios at the same time, providing key evidence for subsequent fusion analysis.

[0088] In the specific implementation process, the regional load deviation analysis part is based on the correspondence between user power address and line segment, and associates the corresponding line segment identifier for each user power data; the power consumption data of all users in each line segment is counted according to the set time (such as every hour), and the average load of each segment is calculated, and the calculation formula is as follows:

[0089]

[0090] Among them, is the average load of the i-th line segment (i corresponds to the first end, the middle end, and the end), is the load data of the q-th user in the i-th segment, is the total number of users in the i-th segment; the power consumption data of all users in the target area is counted, and the overall average load of the area is calculated, and the calculation formula is as follows:

[0091]

[0092] Among them, is the overall average load of the area; the deviation rate of the average load of each segment and the overall average load of the area is calculated, and the regional load deviation result is obtained according to whether the deviation rate exceeds the preset reasonable range, and the deviation rate calculation formula is as follows:

[0093]

[0094] Among them, is the regional load deviation rate of the i-th line segment.

[0095] From the purified running data obtained in step 3, the historical load data of commercial users is extracted; the historical load data is analyzed, and the instantaneous fluctuation characteristics are extracted, including the maximum single time period fluctuation amplitude, fluctuation duration, etc.; the extracted instantaneous fluctuation characteristics are classified and stored according to the type of commercial users (such as supermarkets, office buildings, and restaurants), and a commercial user load fluctuation characteristic library is constructed; based on the historical fluctuation characteristics of various types of commercial users in the characteristic library, corresponding dynamic instantaneous fluctuation thresholds are generated, including fluctuation amplitude threshold and fluctuation duration threshold, and the calculation formula is as follows:

[0096]

[0097]

[0098] wherein, is a fluctuation amplitude threshold value, is an amplitude adjustment coefficient, is the historical maximum fluctuation amplitude of the type of commercial user; is a fluctuation duration threshold value, is a time adjustment coefficient, is the historical average fluctuation duration of the type of commercial user.

[0099] extracting real-time load fluctuation data of the commercial user, calculating real-time fluctuation amplitude and real-time fluctuation duration ; calculating the peak-valley electricity consumption ratio of the user, the calculation formula is as follows:

[0100]

[0101] wherein, is the peak-valley electricity consumption ratio, is the electricity consumption amount of the peak segment in the set period, is the electricity consumption amount of the valley segment in the same period;

[0102] comparing the real-time fluctuation amplitude with the fluctuation amplitude threshold value , the real-time fluctuation duration with the fluctuation duration threshold value , if none of them exceeds the threshold value, it is determined as normal fluctuation; if the threshold value is exceeded, combined with the peak-valley electricity consumption ratio to determine whether it is abnormal electricity consumption, to obtain the abnormal electricity consumption judgment result of the commercial user.

[0103] Step 6, integrating the said purified operation data, the completed data, the regional load characteristics and the electricity consumption abnormal characteristics, using the analytic hierarchy process to set the weight coefficient of each data, calculating the comprehensive abnormal index.

[0104] It should be noted that the analytic hierarchy process refers to a method of decomposing a complex problem into multiple levels, determining the weight of each level element by pairwise comparison, and then making a comprehensive evaluation and decision, which is used here to determine the weight coefficient of each analysis data in the calculation of the comprehensive abnormal index.

[0105] The weight coefficient refers to a coefficient reflecting the influence degree of each analysis data on the diagnosis of the high loss cause of the transformer area. The greater the weight coefficient, the more significant the influence of the corresponding data on the diagnosis result. The standardization processing refers to converting the analysis data of different dimensions and different value ranges into standardized values of a unified dimension and a unified value interval, so as to eliminate the influence of dimension difference on the fusion calculation. The comprehensive abnormality index refers to a quantitative index obtained by fusing multi-dimensional analysis data and calculating according to the corresponding weight coefficient, and is used to comprehensively reflect the overall abnormality degree of the target transformer area.

[0106] Considering that the high loss cause of the transformer area is complex and the analysis result of a single dimension cannot comprehensively reflect the abnormality of the transformer area, in another optional implementation, the updated household-variable correlation data set between step 3 and step 4 is included in the fusion data range of step 6, that is, step 6 fuses the purified running data, the updated household-variable correlation data set, the completed data, the regional load deviation result and the abnormal electricity consumption judgment result of the commercial user, sets the weight coefficient of each data by using the analytic hierarchy process, and calculates the comprehensive abnormality index.

[0107] By fusing the multi-dimensional data such as the purified running data, the updated household-variable correlation data set, the completed data, the regional load deviation result and the abnormal electricity consumption judgment result of the commercial user, first, the standardization processing is performed to eliminate the dimension difference, and then the analytic hierarchy process is used to set a reasonable weight coefficient, so that the qualitative and quantitative analysis results of multiple dimensions can be converted into a unified comprehensive abnormality index, and comprehensive quantitative evaluation of the abnormality of the transformer area can be realized.

[0108] In the specific implementation process, the data sources participating in the fusion calculation are determined, including the purified running data of step 3, the updated household-variable correlation data set, the completed data of step 4, the regional load deviation result and the abnormal electricity consumption judgment result of the commercial user of step 5; the standardization processing is performed on each data source participating in the fusion, the range standardization method is used to convert the data of different dimensions into standardized values of a unified dimension, and the calculation formula is as follows:

[0109] For positive indicators (the greater the value, the higher the abnormality degree):

[0110]

[0111] For negative indicators (the smaller the value, the higher the abnormality degree):

[0112]

[0113] wherein, is the standardized value, is the original data value, is the maximum value of the index, is the minimum value of the index; an analytic hierarchy process is used to construct an evaluation system, taking the comprehensive anomaly index as a target layer and each data source participating in the fusion as a criterion layer; each element of the criterion layer is compared with each other to construct a judgment matrix and calculate the weight coefficient of each element is the number of data sources participating in the fusion, and satisfies:

[0114]

[0115] The comprehensive anomaly index is calculated according to the sum of the products of the weight coefficients of each data source and the corresponding standardized value, and the calculation formula is as follows:

[0116]

[0117] wherein, is the comprehensive anomaly index, is the weight coefficient of the i th data source, is the standardized value of the i th data source. Step 7, based on the comprehensive anomaly index and the analysis results of each link, the causes of high loss of the transformer area are located, and a high loss cause diagnosis report of the transformer area is generated.

[0118] It should be noted that the causes of high loss of the transformer area refer to the specific reasons leading to the excessively high line loss rate of the target transformer area, including line aging, abnormal relationship between households and transformers, data missing, uneven load distribution, abnormal electricity consumption of commercial users, etc. The high loss cause diagnosis report of the transformer area refers to a document recording the results of the whole process of identifying and diagnosing the causes of high loss of the transformer area, including basic data profile, analysis results of each step, comprehensive anomaly index, positioning results of high loss causes and targeted suggestions, etc. The line loss rate refers to the ratio of line loss power to total power supply, which is the core index for measuring the power supply efficiency of the transformer area, and the calculation formula is as follows:

[0119]

[0120]

[0121] wherein, is the line loss rate, is the total power supply in the set period; is the total power consumption in the period.

[0122] Considering that the comprehensive anomaly index comprehensively reflects the overall abnormality degree of the transformer area, the line loss rate intuitively reflects the power supply efficiency of the transformer area, and the analysis results of each link clearly show the specific abnormality in different dimensions. Combined with the three, the core causes leading to high loss of the transformer area can be accurately located, and a report containing complete diagnosis information is generated, providing clear decision basis for transformer area operation and maintenance personnel and guiding subsequent high loss management work.

[0123] ​​​In the specific implementation process, the line loss rate of the target transformer area is calculated to determine the overall degree of high loss of the transformer area; the overall abnormality degree of the transformer area is analyzed in combination with the calculated comprehensive abnormality index; the specific abnormality conditions of each dimension are determined by backtracking the analysis results of each link, such as line aging noise conditions, whether the relationship between a house and a transformer is changed, whether there is a data breakpoint and completion condition, whether the regional load distribution is balanced, whether there is abnormal electricity consumption of commercial users, and the like; based on the line loss rate, the comprehensive abnormality index and the specific abnormality conditions of each link, the corresponding relationship between the high loss causes and the abnormality conditions is associated to locate the core causes of the high loss of the transformer area; the related data and analysis results of the whole diagnosis process are sorted out to compile a transformer area high loss cause diagnosis report containing the basic data profile, the analysis details of each step, the line loss rate, the comprehensive abnormality index, the high loss cause positioning result and the operation and maintenance management suggestions; the diagnosis report is pushed to the transformer area operation and maintenance management platform for the operation and maintenance personnel to check and use.

[0124] The above-described embodiments only express the specific implementation of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application.

Claims

1. A method for identifying and diagnosing the cause of high loss in a transformer area, characterized in that, The method comprises the following steps: Step 1, collecting basic operation data of the target station area, wherein the basic operation data comprises line operation parameters, user power consumption data, transformer power supply range data and historical power consumption data, and the user power consumption data covers load data of commercial users; Step 2, matching a preset aging line noise feature library based on aging-related parameters in the line operation parameters to determine an aging line noise feature label of the target station area; Step 3, dynamically generating a noise filtering threshold according to the aging line noise feature label, and performing noise filtering processing on the basic operation data by using an adaptive Kalman filtering algorithm to obtain purified operation data; Step 4, performing integrity detection on the purified operation data, identifying missing data and counting continuous missing time length, if the continuous missing time length exceeds a preset breakpoint threshold, determining as breakpoint data, using an LSTM time series prediction model to construct a trend completion model based on historical power consumption data before the breakpoint of the user to generate completed data; Step 5, extracting regional load features representing the load distribution state of the station area based on the completed data and the transformer power supply range data, and extracting power consumption state features of the commercial users to obtain power consumption anomaly features; Step 6, fusing the purified operation data, the completed data, the regional load features and the power consumption anomaly features, setting weight coefficients of each data by using an analytic hierarchy process, and calculating a comprehensive anomaly index; Step 7, positioning the high-loss cause of the station area based on the comprehensive anomaly index and the analysis results of each link, and generating a station area high-loss cause diagnosis report.

2. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, Between the step 3 and the step 4, the following content is further included: Based on the user power consumption data in the purified operation data, a ratio of current change amount in a set time interval to historical same period average current change amount is calculated to obtain a household transformer relationship mutation rate; The household transformer relationship mutation rate is compared with a preset trigger threshold, if the household transformer relationship mutation rate exceeds the preset trigger threshold, the GIS system is called to locate the user power consumption address and boundary comparison is performed between the transformer power supply range to confirm whether the household transformer relationship is changed, if it is confirmed that the household transformer relationship is changed, the household transformer association data set is updated.

3. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, The noise filtering threshold in step 3 comprises a frequency cutoff threshold and an amplitude threshold, the adaptive Kalman filtering algorithm realizes noise filtering by dynamically adjusting Kalman gain, and the trend similarity of the purified operation data and historical same period normal data is calculated after filtering.

4. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 2, characterized in that, The current change amount is an absolute value of the difference between the current value at the end of the set time interval and the current value at the beginning, and the historical same period average current change amount is an average value of current change amounts of a set number of same period time intervals.

5. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 2, characterized in that, The transformer power supply range in step 2 is defined by vector boundary data, and the household transformer association data set contains user identification, transformer identification and attribution relationship effective time stamp.

6. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, The LSTM time series prediction model in step 4 takes historical power consumption data of multiple complete periods before the breakpoint as training data, and the completed data is verified for trend consistency with effective data before and after the breakpoint by using Pearson correlation coefficient.

7. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, The step 5 comprises the following content: Based on the user electricity address association line segment identification, the line segment identification is divided into first end identification, middle end identification and end identification according to power supply radius; According to the average load data in the segment statistical setting time, the deviation rate of each segment average load and the overall average load of the transformer area is calculated, and the deviation rate is taken as the regional load characteristic; The historical load data of commercial users is extracted from the completed data to obtain the instantaneous fluctuation characteristic, a commercial user load fluctuation feature library is constructed, a dynamic instantaneous fluctuation threshold is generated based on the feature library, the dynamic instantaneous fluctuation threshold includes a fluctuation amplitude threshold and a fluctuation duration threshold, and the fluctuation duration is the time difference from the load deviating from the normal stable value to recovering to the stable state. The real-time load fluctuation data of the commercial user is compared with the dynamic instantaneous fluctuation threshold, and the electricity consumption abnormal characteristic is obtained by combining the peak-valley electricity consumption ratio.

8. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, In step 6, the data needs to be standardized before fusion, the standardization processing adopts the range standardization method, and the weight coefficient is calculated by comparing the influence degree of each data on the diagnosis result through the analytic hierarchy process.

9. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 1, characterized in that, In step 7, when positioning the high loss cause of the transformer area, the numerical interval of the comprehensive abnormal index and the correlation matching relationship of each link analysis result are combined, and the diagnosis report includes the line loss influence proportion corresponding to each cause.

10. The method for identifying and diagnosing causes of high loss in a transformer area according to claim 2, characterized in that, The preset trigger threshold and the preset breakpoint threshold are determined by statistical analysis of transformer operation case data, and the preset trigger threshold and the preset breakpoint threshold are dynamically adjusted according to the type of transformer.

Images