Power distribution network user characteristic portrait generation method based on electric energy metering

Abstract

The invention discloses a power distribution network user characteristic portrait generation method based on electric energy metering, which overcomes the problems of tedious calculation in a user portrait analysis clustering process and low accuracy of a user portrait classification result in the prior art, and comprises the following steps: collecting user electric energy metering data, standardizing the data, and removing redundant attribute characteristics and samples; dimensionality reduction is carried out on the electric energy metering data, power utilization characteristics and power utilization curves of users are extracted, and individual user characteristic portraits of the users are generated; on the basis of individual user feature portraits, a feature portrait library is established, and according to demand tags, an improved fuzzy C-means algorithm is adopted to obtain user groups of different behavior modes under the demand tags; and enabling the power utilization characteristics to correspond to the behavior modes, gathering the number of user groups, and generating a group user characteristic portrait. The dimension of the electric energy metering data is reduced, the data set is simplified, the calculation efficiency is improved, and the accuracy of the output result of the user feature portrait is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of user power quality demand tendency analysis, and in particular to a method for generating characteristic portraits of distribution network users based on power metering. Background Art

[0002] With the continuous construction and development of smart grids, the degree of grid informatization has been continuously improved. The information integration system of the power sector has collected relatively detailed user power information and accumulated a large amount of power consumption data. These data also have certain correlations and similarities. At the same time, users have high requirements for the quality of power services. In this context, how to make full use of existing data resources, deeply explore the potential needs of users, improve the quality of power supply services, and enhance user stickiness is crucial to the future development of power grid companies.

[0003] When analyzing power user data, existing technologies mainly focus on clustering user electricity usage characteristics, but fail to connect user portraits, power user electricity usage data, processing analysis and applications. Therefore, the value of user electricity usage information data has not been fully utilized, and it is difficult to play a role in improving the energy efficiency of power users, thereby supporting the guidance and decision-making of user service business on the grid side. In addition, user portrait analysis is based on massive power usage data, which makes the clustering process cumbersome and time-consuming, and the output of user feature portrait classification results is inaccurate. Summary of the invention

[0004] The purpose of the present invention is to overcome the problems in the prior art that the user portrait analysis clustering process is cumbersome in calculation and the user portrait classification results have low accuracy, and provide a distribution network user feature portrait generation method based on electric energy metering. By removing redundant attribute features and samples in the collected electric energy metering data, the extracted electricity consumption features are reduced in dimension, the data set is simplified, the calculation efficiency is improved, and the accuracy of the user feature portrait output results is improved.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A method for generating a characteristic profile of a distribution network user based on electric energy metering, characterized in that it comprises the following steps: Collect user electricity metering data and standardize it to remove redundant attribute features and samples; Reduce the dimension of electric energy metering data, extract the user's electricity consumption characteristics and electricity consumption curve, and generate the user's individual user characteristic portrait; Based on the feature portrait of individual users, a feature portrait library is established. According to the demand label, the improved fuzzy C-means algorithm is used to obtain the user groups with different behavior patterns under the demand label; Match electricity consumption characteristics with behavioral patterns, aggregate the number of user groups, and generate group user characteristic portraits.

[0006] When generating user feature portraits, redundant attribute features and samples in the collected power metering data are removed, and the power metering data is reduced in dimension, so that the data set is simplified, the calculation efficiency is improved, and the accuracy of the output results of the user feature portrait is improved. After generating individual user feature portraits, staff can quickly understand user behavior preferences and provide targeted power services to individual users. At the same time, according to classification needs, user feature portraits of a certain group can be generated. Staff can observe the characteristics of this type of users and provide targeted power services to this type of users, thereby promoting the development of power companies.

[0007] Preferably, the dimensionality reduction of the electric energy metering data includes: reducing the noise of the electric energy metering data by using a wavelet threshold denoising method, reducing the dimensionality of the electric energy metering data by using a segmented aggregation approximation algorithm, and optimizing the compression ratio of the segmented aggregation approximation algorithm by using an accelerated particle swarm optimization algorithm.

[0008] Preferably, when the dimension reduction is performed on the electricity usage feature data set, the time series characteristics of the data fluctuation changes in the electricity usage features are retained.

[0009] Preferably, extracting the electricity consumption curve includes: dividing the daily electricity consumption data into m segments, calculating the mean of each segment after the daily electricity consumption data is reduced in dimension; symbolizing each segment according to the mean to obtain a discrete symbol sequence; counting the frequency of occurrence of each symbol sequence, screening the electricity consumption curve form, and proposing abnormal electricity consumption data; superimposing the daily electricity consumption curves with the highest frequency and taking the average to obtain the daily electricity consumption curve.

[0010] Preferably, the extraction of electricity consumption characteristics includes: screening characteristic indicators belonging to the user's electricity consumption level description type from the electricity metering data; calculating the user's actual load level utilization rate based on the relationship between the user's expected reported capacity and the terminal load when the user's actual electricity load reaches saturation after connection; calculating the load development rate after the user is connected to electricity based on the proportional relationship between the actual load in each year and the actual load in the terminal year after the user is connected to electricity.

[0011] Preferably, the improved fuzzy C-means algorithm is used to obtain user groups with different behavior patterns under the demand label, including: performing dimensionality reduction processing on the electric energy metering data to retain the time series characteristics of data fluctuations; setting initial clustering parameters and initializing the clustering parameters according to the demand label; calculating the distance from each data point to each cluster center, calculating the fuzziness of each data point belonging to each cluster, and updating the cluster center according to the fuzziness; traversing the number of clusters until the cluster center is determined or the maximum number of iterations is reached.

[0012] Preferably, the removing of redundant attribute features and samples comprises: merging data with the same attribute features in the standardized electric energy metering data; and removing redundant samples in the standardized data by object reduction.

[0013] Preferably, the electric energy metering data includes static data and dynamic data, the static data includes user attribute data and industry information identification, and the dynamic data includes user behavior data and dynamic change trends.

[0014] Preferably, the silhouette coefficient index is used to determine the optimal number of clusters, and each cluster center obtained by this clustering is normalized.

[0015] Preferably, the collected electric energy metering data is preprocessed, multiple time series are modeled, abnormal data is monitored, and the missing data is filled by the mean substitution method.

[0016] Therefore, the present invention has the following beneficial effects: 1. The present invention reduces the dimension of the collected electric energy metering data by removing redundant attribute features and samples, simplifies the data set, improves the calculation efficiency, and improves the accuracy of the output results of the user feature portrait.

[0017] 2. The present invention analyzes the user's electricity consumption characteristics and electricity consumption curves to generate individual user characteristic portraits. The staff can quickly understand the user's behavioral preferences and provide targeted electricity services to individual users.

[0018] 3. The present invention analyzes and clusters the overall electricity consumption characteristics of the region, and can generate a characteristic portrait of a certain group of users according to classification needs. The staff can observe the characteristics of this type of users to facilitate the formulation of corresponding demand strategies for different types of users, thereby achieving efficient and refined power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The figure is a flow chart of the overall steps of the method of the present invention.

[0020] Figure 2 This is a flow chart of the steps of processing electric energy metering data in the second embodiment. DETAILED DESCRIPTION

[0021] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments: Embodiment 1: like Figure 1In the illustrated embodiment, a method for generating a user characteristic portrait of a distribution network based on electric energy metering can be seen, and its operation process is as follows: Step 1, collecting and standardizing the user's electric energy metering data, and removing redundant attribute features and samples; Step 2, reducing the dimension of the electric energy metering data, extracting the user's electricity consumption characteristics and electricity consumption curve, and generating the user's individual user characteristic portrait; Step 3, based on the individual user characteristic portrait, establishing a characteristic portrait library, and according to the demand label, using the fuzzy C-type mean algorithm to obtain user groups with different behavior patterns under the demand label; Step 4, matching the electricity consumption characteristics with the behavior patterns, collecting the number of user groups, and generating a group user characteristic portrait.

[0022] The method for generating a characteristic profile of a distribution network user based on electric energy metering provided in this embodiment uses technical means such as big data analysis and data visualization analysis to deeply study the electricity consumption characteristics and patterns of distribution network users. Based on the main electricity consumption behaviors and guided by the theory of electric energy metering, a characteristic profile of distribution network users is established. When generating a user characteristic profile, the redundant attribute features and samples in the collected electric energy metering data are removed, and the dimension of the electric energy metering data is reduced, the data set is simplified, the calculation efficiency is improved, and the accuracy of the output results of the user characteristic profile is improved. After generating an individual user characteristic profile, the staff can quickly understand the user behavior preferences and provide targeted electricity services to individual users. At the same time, according to the classification needs, a characteristic profile of a certain group of users can be generated. The staff can observe the characteristics of this type of users, provide targeted electricity services to this type of users, and promote the development of power companies.

[0023] The technical scheme and technical effects of the present invention will be further described below through specific examples. The following examples are explanations of the present invention and the present invention is not limited to the following examples.

[0024] Step 1: Collect user electricity metering data and standardize it to remove redundant attribute features and samples.

[0025] The standardization in this embodiment may be to normalize the electric energy metering data using a linear proportional normalization method.

[0026] Step 2: Reduce the dimension of the electricity metering data, extract the user's electricity consumption characteristics and electricity consumption curve, generate the user's individual user feature portrait, and establish a feature portrait library.

[0027] The feature portrait library contains the user ID and the corresponding individual user feature portrait. The individual user feature portrait includes the user's electricity consumption characteristics and electricity consumption curve. According to the feature portrait library, one or two electricity consumption characteristics can be selected to extract user groups with certain commonalities from the feature portrait library for analysis.

[0028] Extracting electricity consumption characteristics includes: screening characteristic indicators belonging to the user's electricity consumption level description type from the electricity metering data; calculating the user's actual load level utilization rate based on the relationship between the user's expected reported capacity and the terminal load when the user's actual electricity load reaches saturation after connection; calculating the load development rate after the user is connected to electricity based on the proportional relationship between the actual load in each year and the actual load in the terminal year after the user is connected to electricity.

[0029] Extracting the electricity consumption curve includes: using the wavelet threshold denoising method to reduce the noise of the energy metering data, using the segmented aggregation approximation algorithm to reduce the dimension of the energy metering data, and using the accelerated particle swarm optimization algorithm to optimize the compression ratio of the segmented aggregation approximation algorithm.

[0030] The daily electricity consumption data is divided into m segments, and the mean of each segment after the dimension reduction of the daily electricity consumption data is calculated; each segment is symbolized and assigned a value according to the mean to obtain a discrete symbol sequence; the frequency of occurrence of each symbol sequence is counted, the form of electricity consumption curve is screened, and abnormal electricity consumption data is proposed; the daily electricity consumption curves with the highest frequency are superimposed and the average is taken to obtain the daily electricity consumption curve.

[0031] In other embodiments, the electricity consumption curve can be extracted by selecting a certain day within a period of time as a typical day, taking the electricity consumption curve of the typical day as a benchmark, distributing the electricity metering data of the remaining days near the electricity consumption curve of the typical day, removing the electricity metering data with a large offset distance, and averaging the remaining data to obtain the user's electricity consumption curve.

[0032] Step 3: According to the demand label, the improved fuzzy C-means algorithm is used to obtain the user groups with different behavior patterns under the demand label.

[0033] The demand label is the standard for clustering users. For example, if users are clustered according to industry, the demand label is industry.

[0034] like Figure 1 As shown in Figure 1, the improved fuzzy C-means algorithm combines the data dimensionality reduction method, including: (1) Obtain the feature portraits of individual users in the feature portrait library, perform dimensionality reduction processing on the electricity consumption curves and electricity consumption characteristics in the feature portraits of individual users, retain the temporal characteristics of data fluctuations, reduce the computational complexity of subsequent clustering algorithms, and improve the clustering rate.

[0035] Specifically: for the set of individual user feature portraits, set the number of nearest neighbors and the dimension after dimensionality reduction; use the Euclidean distance as the standard to calculate the s nearest neighbors of each individual user feature portrait; for a single individual user feature portrait, calculate the local covariance matrix and the corresponding weight coefficient vector; the weight coefficient vectors of all individual user feature portraits form a weight matrix, and use the weight matrix to perform dimensionality reduction.

[0036] (2) Set the initial clustering parameters according to the required labels and initialize the clustering parameters.

[0037] (3) Calculate the distance from each data point to each cluster center, calculate the fuzziness of each data point belonging to each cluster, and update the cluster center based on the fuzziness.

[0038] (4) Traverse the number of clusters until the cluster center is determined or the maximum number of iterations is reached, and clustering is completed.

[0039] Step 4: Match electricity consumption characteristics with behavioral patterns, combine with the number of user groups, and generate group user characteristic portraits.

[0040] In the third step, users have been divided into user groups with different behavior patterns according to demand labels, including user group 1, user group 2, ..., user group n.

[0041] For each user group, the group user characteristics and group power consumption curve of the group are extracted. The group user characteristics include the long-term development characteristics of the load, so as to obtain the group user characteristic portrait. The group user characteristic portrait includes the group user characteristics and group power consumption curve, the number of groups, and the power consumption characteristics and power consumption curve of individual users in the group.

[0042] At the same time, based on the individual user feature portrait obtained in the second step, individual users in each user group can also be analyzed.

[0043] The method for generating a distribution network user characteristic profile based on electric energy metering provided in this embodiment can be used to solve the following problems: 1. When analyzing power user data, existing technologies do not connect user portraits, power user electricity consumption data, processing analysis and applications. Therefore, the value of user electricity consumption information data has not been fully utilized, making it difficult to play a role in improving the energy efficiency of power users, thereby supporting the guidance and decision-making of user service business on the grid side.

[0044] 2. User portrait analysis is based on massive electricity consumption data, which makes the clustering process cumbersome and time-consuming, and the output of user feature portrait classification results is inaccurate.

[0045] The following beneficial effects exist: 1. The present invention reduces the dimension of the extracted electricity usage features by removing redundant attribute features and samples in the collected electricity metering data, simplifies the data set, improves the calculation efficiency, and improves the accuracy of the output results of the user feature portrait.

[0046] 2. The present invention analyzes the user's electricity consumption characteristics and the electricity consumption curve to generate an individual user characteristic portrait. At the same time, it analyzes and clusters the overall electricity consumption characteristics of the region to generate a group user characteristic portrait, which is convenient for formulating corresponding demand strategies for different types of users and realizing efficient and refined power grid operation.

[0047] 3. User portrait is an effective technical means for enterprises to achieve personalized user analysis and assist in refined operational decision-making. With the in-depth promotion of information construction and the rapid development of power business, power grid enterprises have also accumulated rich and valuable data resources. Deeply mining existing data and making full use of data analysis results to assist decision-making, and then studying the laws of power grid development and customer service, has become one of the important ways to drive the innovative development of power grid enterprises.

[0048] 4. The power user portrait analysis is based on massive electricity consumption data. It takes power users as the main body, conducts information mining on user personal characteristics and user electricity consumption behavior data, assigns user portrait labels based on user characteristics, analyzes the differences in electricity consumption behaviors of different users, and then mines the differences in user behaviors, providing a theoretical basis for the intelligent management of power companies.

[0049] Embodiment 2: This embodiment provides a method for generating a user characteristic profile of a distribution network based on electric energy metering. Figure 1 As shown in the figure, the operation process is as follows: Step 1, collect user electricity metering data and standardize it to remove redundant attribute features and samples; Step 2, reduce the dimension of the electricity metering data, extract the user's electricity consumption characteristics and electricity consumption curve, and generate the individual user feature portrait of the user; Step 3, based on the individual user feature portrait, establish a feature portrait library, and according to the demand label, use the fuzzy C-type mean algorithm to obtain user groups with different behavior patterns under the demand label; Step 4, correspond the electricity consumption characteristics with the behavior pattern, collect the number of user groups, and generate the group user feature portrait.

[0050] The method for generating a characteristic profile of a distribution network user based on electric energy metering provided in this embodiment uses technical means such as big data analysis and data visualization analysis to deeply study the electricity consumption characteristics and patterns of distribution network users. Based on the main electricity consumption behaviors and guided by the theory of electric energy metering, a characteristic profile of distribution network users is established. When generating a user characteristic profile, the redundant attribute features and samples in the collected electric energy metering data are removed, and the dimension of the electric energy metering data is reduced, the data set is simplified, the calculation efficiency is improved, and the accuracy of the output results of the user characteristic profile is improved. After generating an individual user characteristic profile, the staff can quickly understand the user behavior preferences and provide targeted electricity services to individual users. At the same time, according to the classification needs, a characteristic profile of a certain group of users can be generated. The staff can observe the characteristics of this type of users, provide targeted electricity services to this type of users, and promote the development of power companies.

[0051] The technical scheme and technical effects of the present invention will be further described below through specific examples. The following examples are explanations of the present invention and the present invention is not limited to the following examples.

[0052] Step 1: Collect user electricity metering data and standardize it to remove redundant attribute features and samples.

[0053] Electricity metering data includes static data and dynamic data. Static data includes user attribute data and industry information identification, where user attribute data includes name, gender, age, place of origin, user category, electricity type, voltage level and equipment information, etc. Dynamic data includes user behavior data and dynamic change trends. User behavior data includes user electricity behavior, payment behavior, demand behavior, power generation behavior, power consumption, power consumption trend, power consumption risk, number of power outages, payment time, payment frequency, payment channel preference, power generation, power generation anomalies, complaint frequency, interaction frequency, user evaluation and other data generated by user behavior.

[0054] The collected electric energy metering data is big data with a large amount of data, which may contain a large number of outliers and missing values. Therefore, it is necessary to preprocess the collected electric energy metering data, identify the outliers and missing values, process the missing values, and correct the outliers.

[0055] like Figure 2 As shown, the specific process includes: dividing the collected daily user power metering data into several time periods, and the power metering data of the same time period on different dates is used as a data sample. (In other embodiments, the power metering data of each day can also be used as a data sample. The specific division method can be selected according to the specific data type or user type. If the user's electricity consumption time is relatively fixed and the daily electricity consumption time does not fluctuate much, the data sample division method of this embodiment can be selected.) Detect whether there are abnormal values ​​or missing values ​​in the electric energy metering data of each time period, and determine whether the sum of the number of abnormal values ​​and missing values ​​reaches a first threshold. If the sum of the number reaches the first threshold, the data sample is discarded, and the data of the missing time period is supplemented by the data of the adjacent time period.

[0056] If the sum of the number does not reach the first threshold, the outliers are removed as missing value data. For missing values, h normal values ​​in the time period adjacent to the point and h normal values ​​of the point at the same time but on different dates are selected, and the Lagrange difference method is used to fill them. (The Lagrange method can be used to calculate two filling values ​​using two types of data, and the average of the two filling values ​​is used as the final filling value. If the difference between the two filling values ​​exceeds the third threshold, the data point is reselected for calculation. If the difference between the two filling values ​​still exceeds the third threshold, the filling value calculated by the h normal values ​​in the time period adjacent to the point is selected as the final filling value.) When judging abnormal values, if a data value is negative, it means that the value is abnormal; if the data value is positive, calculate the difference between all values ​​in the data sample and the average value A of the data sample, and then calculate the average value B of the average value difference. If the difference f between the data value and the data value at the same time of the previous day and the difference r between the data value and the data value at the same time of the next day are both between [A-Be, A+Be], it means that the value is normal, otherwise the value is abnormal. Among them, e represents the fluctuation factor of electric energy metering data.

[0057] Standardization of electric energy metering data can utilize data reconstruction to convert the collected electric energy metering data from one format to another, so as to unify the structure, format and type of electric energy metering data and facilitate subsequent processing.

[0058] For the standardized data, merge the data with the same attribute features, remove unnecessary attribute features, and remove redundant attribute features; use object reduction to remove redundant samples in the standardized data.

[0059] Step 2: Reduce the dimension of the electricity metering data, extract the user's electricity consumption characteristics and electricity consumption curve, and generate the user's individual user characteristic portrait.

[0060] The characteristic portrait of an individual user includes the user's electricity consumption characteristics and the user's typical electricity consumption curve. The user's typical electricity consumption curve includes a load curve, which can be drawn using the average value of the electricity metering data at each moment over a period of time.

[0061] Extracting electricity consumption characteristics includes: screening characteristic indicators belonging to the user's electricity consumption level description type from the electricity metering data; calculating the user's actual load level utilization rate based on the relationship between the user's expected reported capacity and the terminal load when the user's actual electricity load reaches saturation after connection; calculating the load development rate after the user is connected to electricity based on the proportional relationship between the actual load in each year and the actual load in the terminal year after the user is connected to electricity.

[0062] User electricity consumption characteristics also include: reported contract capacity, daily peak-to-valley difference rate (reflecting the degree of load fluctuation and the peak-shaving capacity of the power grid), daily load rate (reflecting the degree of load change), daily maximum load time, daily minimum load time, peak-to-valley difference time (reflecting the peak-to-valley time span), daily minimum load rate (reflecting the size of load fluctuation), daily maximum load rate, peak load rate, flat load rate, valley load rate, user payment characteristics and abnormal time periods of user electricity consumption, etc.

[0063] Step 3: Based on the feature portrait of individual users, a feature portrait library is established. According to the demand label, the fuzzy C-mean algorithm is used to obtain user groups with different behavior patterns under the demand label.

[0064] Step 4: Match the electricity consumption characteristics with the behavior patterns, combine them with the number of user groups, generate group user characteristic portraits and display them visually.

[0065] The group user characteristic portrait, taking the demand label as industry as an example, the group user characteristic portrait includes each industry and the corresponding industry electricity consumption pattern and industry electricity consumption characteristics.

[0066] Or you can divide the big industry into the first-level label, subdivide the big industry into various small industries, set the small industry as the second-level label under the first-level label, and set the weight. Each label has its corresponding industry electricity consumption mode and industry electricity consumption characteristics. Under each industry, there are individual user feature portraits of all users under the industry label.

[0067] It enables the staff to intuitively understand the electricity consumption of each industry, making it easier to regulate the electricity consumption of each industry.

[0068] In other embodiments, the electricity usage type may also be set as an industry label, such as ordinary users, corporate users, and government, etc.

[0069] Embodiment three: This embodiment provides a method for generating characteristic portraits of distribution network users based on electric energy metering, and implements the method for generating characteristic portraits of distribution network users based on electric energy metering by introducing a specific application scenario.

[0070] In the new distribution network, on the one hand, the integration of large-scale distributed photovoltaics has indeed solved the electricity consumption problem of users, and can achieve on-site compensation and clean energy utilization; on the other hand, photovoltaic grid connection has also changed the network structure of the traditional distribution network and its "single source - one-way" energy flow characteristics, and will also increase the randomness on the user side. If no measures are taken, some power quality problems may occur, including three-phase load imbalance, voltage problems, harmonic pollution, etc.

[0071] Photovoltaic power generation, as one of the types of renewable energy generation, has been increasingly popularized throughout the country, and the importance of photovoltaic power generation has been recognized both at home and abroad, and it has strong economic development value. Photovoltaic power generation can also improve energy utilization efficiency and reduce problems caused by environmental pollution, but photovoltaic power generation must focus on safety maintenance during the power generation process and accurately control the impact of photovoltaic power access to the distribution network.

[0072] In view of the large-scale access of distributed photovoltaics, the power network needs to be uniformly designed in the early stage. It should be designed according to the development needs of related industries, and then a series of plans should be carried out to ensure reasonable operation and reasonable control of some inevitable factors. The development of low-voltage distribution networks cannot meet the needs of social development in the new stage. It is necessary to continuously upgrade the network structure and optimize the original operation plan so that users can have a better experience in the process of use and participation. Not only can it better meet the needs of the development of the power market, but it can also make full use of renewable energy and improve people's quality of life. The large-scale access of distributed photovoltaics can better improve the planning and design of the distribution network, make the operation more scientific and reasonable, and also reduce the risks of power production and transmission.

[0073] Based on the above, a local power company needs to analyze user characteristics and generate group user characteristic portraits based on industry electricity consumption, so that it can design the power network based on the user characteristic portraits to ensure the reasonable operation of the distribution network.

[0074] Therefore, a method for generating characteristic profiles of distribution network users based on electric energy metering is adopted, and technical means such as big data analysis and data visualization analysis are used to deeply study the electricity consumption characteristics and patterns of distribution network users. Based on the main electricity consumption behaviors and guided by the electric energy metering theory, a characteristic profile of distribution network users is established.

[0075] This enables: Divide user groups and plan the power distribution network: Divide demand labels into manufacturing, construction, mining, etc., and then further subdivide them, such as dividing manufacturing into machinery manufacturing, plastic manufacturing, metal products manufacturing, and textiles, to generate a panoramic portrait of the power consumption behavior of the industry in the region. According to the characteristics of users in each industry, the power network needs to be uniformly designed in the early stage. The design should be based on the development needs of related industries shown in the group user characteristic portrait, and then a series of plans should be carried out to ensure the reasonable operation of the distribution network and to reasonably control some inevitable factors.

[0076] Segment user groups to achieve accurate information push: Segment power user groups based on user feature portraits, divide demand tags into groups such as general users, industrial production, government departments, small businesses, etc., and push notifications and suggestions to different service groups.

[0077] Promote differentiated power supply services and improve user satisfaction: Through user feature profiling, power companies can quickly understand the user's interaction with the power company in historical data, including the user's historical electricity consumption, historical payment status, historical demands, and user temperament, so that they can timely adjust the attitude and service strategy suitable for the user and provide users with high-quality services.

[0078] Help promote brands and improve marketing effectiveness: Carry out targeted marketing for specific users based on user profiles to improve marketing success rate and avoid user disgust. In power user services, targeted marketing service configuration is carried out for each category of users, which not only improves user satisfaction and power consumption perception, but also improves the service quality and power supply quality of power companies.

[0079] At present, the service system of power companies has begun to have the characteristics of technical integration, and can provide services to users from multiple angles. However, from a detailed perspective, the accuracy of its services has always been flawed. In the analysis of power quality needs of multiple users, current research mainly focuses on the acquisition and application of user power quality information data, as well as the establishment and quantitative analysis of power quality demand indicators. Since power quality problems will reduce the utilization efficiency of electricity on the demand side, it will bring economic losses to the whole society. Moreover, power supply companies have not yet established an intuitive user power quality demand tendency analysis model based on the intelligent perception information data of multiple user power quality.

[0080] Therefore, the method for generating a user characteristic profile of a distribution network based on electric energy metering provided in this embodiment can help power companies grasp the energy consumption characteristics of users, accurately identify user needs, and provide personalized customized services through user characteristic profile technology, so as to better meet the increasingly diverse customer service requirements. From multiple perspectives such as the diversity of power user needs and the technical requirements of sensitive equipment for power quality, the power quality requirements of users are analyzed, the differentiated needs of different users for power quality are understood and mastered, and targeted solutions are given to ensure high-quality power supply to users.

[0081] Embodiment 4: This embodiment provides a distribution network user characteristic profile generation system based on electric energy metering, which is used to execute the distribution network user characteristic profile generation method based on electric energy metering in Embodiment 1 or Embodiment 2.

[0082] A distribution network user feature profile generation system based on electric energy metering, the system framework includes: (1) Data source layer.

[0083] The sources of electricity metering data collection mainly include the power system's internal information system database (such as marketing business application system or user service system), as well as external public data such as the Internet, power department official website, and corporate official website.

[0084] (2) Data collection layer.

[0085] The electric energy metering data is collected from the data source layer and transmitted to the data storage and processing layer.

[0086] (3) Data storage and processing layer.

[0087] The collected electricity metering data is cleaned, merged, converted and standardized to remove redundant attribute features and samples. The electricity metering data is reduced in dimension based on particle swarm optimization algorithm, segmented aggregation approximate analysis algorithm and other methods for subsequent processing and analysis.

[0088] The data storage and processing layer includes: The first memory is used to store the collected electric energy metering data and the processed electric energy metering data, the related application programs for processing the electric energy metering data, and the configuration files. It can be a volatile memory, such as a random access memory RAM; or a non-volatile memory, such as a read-only memory ROM, a flash memory, a hard disk HDD or a solid state drive SSD; or a combination of the above types of memory.

[0089] The second processor is the control center for processing electric energy metering data, which may be a central processing unit CPU, or a specific integrated circuit ASIC, or one or more integrated circuits configured to implement the embodiments of the present invention. It performs various functions of processing electric energy metering data by running or executing software programs and / or modules stored in the memory, and calling data stored in the memory.

[0090] (3) Data analysis and service layer.

[0091] Based on statistical analysis, data mining, feature extraction, power consumption curve extraction, fuzzy C-mean algorithm and other methods, the power data processed by the data storage and processing layer is analyzed to generate user feature images. The user feature portrait is provided as a data service to the outside world in the form of RESTAPI.

[0092] The data analysis and service layer includes: The second memory is used to store the extracted user power consumption characteristics, power consumption curves, and generated user feature portraits, related applications for generating user feature portraits, and configuration files.

[0093] The second processor is the control center for generating user feature portraits, and performs various functions of electric energy metering data processing by running or executing software programs and / or modules stored in the memory and calling data stored in the first memory.

[0094] (5) Business application layer.

[0095] Provides visual labeling applications, including labeling system (creating labels for user profiles), group characteristics, and approval management functions.

[0096] The business application layer includes: a screen that provides an output interface between the communication module and the staff, and displays the generated individual user feature portraits and group user feature portraits, as well as a panoramic portrait of electricity consumption behavior in a certain area to the staff. If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the staff. The touch panel can also detect the duration and pressure associated with the touch or sliding operation.

[0097] (6) Terminal layer.

[0098] Carry out user classification, operation promotion and other applications through fixed terminals, mobile terminals and related data interface platforms.

[0099] The terminal layer includes a display screen and a front camera and / or a rear camera for receiving external multimedia data.

[0100] The terminal layer also includes a microphone, which can receive external audio signals and send instructions when the communication module is in operation mode, such as call mode, recording mode, and voice recognition mode. The terminal layer also includes a speaker for outputting audio signals, such as broadcasting the generated user feature portrait by the staff.

[0101] The above-described embodiment is only a preferred solution of the present invention and does not limit the present invention in any form. There are other variations and modifications without exceeding the technical solution described in the claims.

Claims

1. A method for generating a characteristic profile of a distribution network user based on electric energy metering, characterized in that: include: Collect user electricity metering data and standardize it to remove redundant attribute features and samples; Reduce the dimension of electric energy metering data, extract the user's electricity consumption characteristics and electricity consumption curve, generate the user's individual user feature portrait, and establish a feature portrait library; According to the demand label, the improved fuzzy C-means algorithm is used to obtain the user groups with different behavior patterns under the demand label; Match the electricity consumption characteristics of user groups with their behavior patterns, and combine them with the number of user groups to generate a group user characteristic portrait.

2. The method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, characterized in that: The dimensionality reduction of the electric energy metering data includes: reducing the noise of the electric energy metering data by using a wavelet threshold denoising method, reducing the dimensionality of the electric energy metering data by using a segmented aggregation approximation algorithm, and optimizing the compression ratio of the segmented aggregation approximation algorithm by using an accelerated particle swarm optimization algorithm.

3. A method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1 or 2, characterized in that: When reducing the dimension of electric energy metering data, the time series characteristics of data fluctuations in electricity consumption characteristics are retained.

4. The method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 2 is characterized in that: The power consumption curve extraction method includes: dividing the daily power consumption data into m segments, calculating the mean value of each segment after the daily power consumption data is reduced in dimension; symbolizing each segment according to the mean value to obtain a discrete symbol sequence; counting the frequency of occurrence of each symbol sequence, screening the power consumption curve form, and proposing abnormal power consumption data; superimposing the daily power consumption curves with the highest frequency and taking the average value to obtain the daily power consumption curve.

5. A method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, 2 or 4, characterized in that: The extraction of electricity consumption characteristics includes: screening characteristic indicators belonging to the user's electricity consumption level description type from the electric energy metering data; calculating the user's actual load level utilization rate based on the relationship between the user's expected reported capacity and the terminal load when the user's actual electricity load reaches saturation after connection; calculating the load development rate after the user is connected to electricity based on the proportional relationship between the actual load in each year and the actual load in the terminal year after the user is connected to electricity.

6. The method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, characterized in that: The improved fuzzy C-means algorithm is used to obtain user groups with different behavior patterns under the demand label, including: performing dimensionality reduction processing on the electric energy metering data to retain the time series characteristics of data fluctuation changes; setting initial clustering parameters according to the demand label and initializing the clustering parameters; calculating the distance from each data point to each cluster center, calculating the fuzziness of each data point belonging to each cluster, and updating the cluster center according to the fuzziness; traversing the number of clusters until the cluster center is determined or the maximum number of iterations is reached.

7. A method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, 2, 4 or 6, characterized in that: The method of removing redundant attribute features and samples includes: merging data with the same attribute features in the standardized electric energy metering data; and removing redundant samples in the standardized data by object reduction.

8. A method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, 2 or 6, characterized in that: The electric energy metering data includes static data and dynamic data. The static data includes user attribute data and industry information identification, and the dynamic data includes user behavior data and dynamic change trends.

9. The method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 6, characterized in that: The silhouette coefficient indicator is used to determine the optimal number of clusters, and the cluster centers obtained in this clustering are normalized.

10. A method for generating a characteristic profile of a distribution network user based on electric energy metering according to claim 1, 2, 4 or 6, characterized in that: The collected electric energy metering data is preprocessed, multiple time series are modeled, abnormal data is monitored, and the missing data is filled using the mean substitution method.

Images