A method and system for identifying abnormal household changes based on electricity usage addresses

Abstract

The present application provides a method and system for identifying household change anomalies based on electricity usage addresses, relating to the technical field of distribution network management. The technical solution provided by the present application performs word segmentation analysis on electricity usage address text and converts it into a quantifiable address vector, then calculates the numerical value of anomaly indicators to perform anomaly judgment. Compared with the technical solutions in related technologies, accurate physical address data is used to replace incomplete latitude and longitude data for household change anomaly recognition, avoiding missed judgments caused by missing addresses. At the same time, based on the numerical analysis of address text and the quantitative judgment of anomaly indicators, a precise quantitative evaluation of different users' electricity usage addresses is achieved, thereby improving the accuracy of household change anomaly recognition.

Description

Technical Field

[0001] The present application relates to the technical field of power distribution network management, and in particular to a method and system for identifying abnormalities of household transformers based on electricity usage addresses. Background Art

[0002] In distribution network management, the relationship between users and substations (transformers) is crucial for ensuring electricity billing and stable power supply. However, this relationship can sometimes exhibit anomalies, known as household transformer anomalies. These anomalies occur when a user's electricity address deviates from the actual correspondence between the substation to which it belongs. For example, a user's address may be significantly inconsistent with the addresses of other users within the substation. These anomalies can be caused by substation divisions, equipment migrations, data entry errors, and other factors, often leading to inaccurate billing and reduced distribution network management efficiency. To ensure stable operation of the distribution system and accurate electricity usage data, accurate identification of household transformer anomalies is essential.

[0003] Related technologies often use longitude and latitude information to identify abnormal household transformers. Cluster analysis is performed on the geographic coordinates of a user's electricity address to determine whether the user falls within a normal transformer area. However, in practice, longitude and latitude data is often difficult to obtain, especially in areas with incomplete address information or incomplete geographic information system coverage. This approach relies heavily on longitude and latitude data, and when this data is incomplete or inaccurate, it can lead to inaccurate identification of abnormal household transformers. Summary of the Invention

[0004] The present application provides a method and system for identifying household change anomalies based on electricity usage addresses, which can improve the accuracy of household change anomaly identification.

[0005] In a first aspect, the present application provides a method for identifying abnormal changes in a household based on an electricity address, the method comprising:

[0006] Obtain the electricity usage addresses of multiple users in the substation area, and split the text of the electricity usage addresses into address words;

[0007] Counting whether each word in the address vocabulary table appears in the corresponding address segmentation word, and generating an address vector corresponding to the power address;

[0008] Calculating the abnormality index value corresponding to the electricity usage address according to the address vector;

[0009] When the abnormal indicator value meets the abnormality judgment condition, the user corresponding to the abnormal indicator value is determined as an abnormal user.

[0010] By adopting the above technical solution, the electricity address text is segmented and analyzed and converted into a quantifiable address vector, and then the abnormal index value is calculated to make an abnormality judgment. Compared with the technical solutions in the related art, accurate physical address data is used to replace the incomplete latitude and longitude data for household change anomaly identification, avoiding missed judgments caused by missing some addresses. At the same time, based on the numerical analysis of the address text and the quantitative judgment of the abnormal index, an accurate quantitative evaluation of the electricity addresses of different users is achieved, thereby improving the accuracy of household change anomaly identification.

[0011] Optionally, before counting whether each word in the address vocabulary table appears in the corresponding address segmentation word and generating the address vector corresponding to the power address, the method further includes:

[0012] Filtering out unique address participles from the address participles, and forming an initial vocabulary list with all the unique address participles;

[0013] Calculate the context occurrence probability of each word in the initial vocabulary using an N-gram model;

[0014] Each word in the initial vocabulary is sorted according to the context occurrence probability to obtain an address vocabulary.

[0015] By adopting the above technical solution, unique address segmentations are screened out from the address segmentations and form an initial vocabulary, which avoids the interference of repeated address segmentations on subsequent analysis, so that the generated address vector can more accurately reflect the characteristic differences of the electricity usage address, thereby improving the accuracy of household change anomaly identification; at the same time, by using the N-gram model to calculate the contextual occurrence probability between each word in the initial vocabulary, and sorting the words based on these contextual occurrence probabilities to generate an address vocabulary, it can effectively capture the contextual relevance and occurrence patterns between address words, so that the address vocabulary can better reflect the semantic characteristics and structural characteristics of the electricity usage address, thereby improving the accuracy of household change anomaly identification.

[0016] Optionally, the using an N-gram model to calculate the context occurrence probability corresponding to each word in the initial vocabulary includes:

[0017] Adding a custom starting word to each of the address segmentation words to obtain a training corpus;

[0018] An N-gram model is constructed based on the training corpus, and the context occurrence probability corresponding to each word in the initial vocabulary is calculated using the N-gram model.

[0019] By employing this technical solution, custom starting words are added to address segmentation to construct a training corpus. Using the N-gram model to analyze the contextual relationships between words, the system accurately captures the semantic relevance of words in address text. The N-gram model constructed based on the training corpus calculates the contextual probability of each word in the initial vocabulary, thereby identifying key words with significant characteristics in the address text. This contextual probability-based lexical analysis method enables the system to automatically learn and adapt to the characteristics of different address texts, improving the accuracy of identifying abnormal addresses.

[0020] Optionally, the acquiring of the electricity usage addresses of multiple users in the substation area, after splitting the text of the electricity usage addresses into address words, further includes:

[0021] Delete the head word and the tail word in the address segmentation word.

[0022] By adopting the above technical solution, we can effectively filter out common address words (such as province, city, etc.) that appear too frequently and atypical address words (such as Room 101, Room 102, etc.) that have no impact on the result judgment, so that the subsequently generated address vector can accurately reflect the key feature information of the electricity address, avoiding the interference caused by common words and eliminating the noise and high-dimensional sparsity problems that may be caused by irrelevant words, thereby improving the accuracy and discrimination of the address vector in representing the user's electricity address characteristics.

[0023] Optionally, the counting of whether each word in the address vocabulary table appears in the corresponding address segmentation word to generate the address vector corresponding to the power address includes:

[0024] If a word in the address vocabulary table appears in the corresponding address word, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 1;

[0025] If a word in the address vocabulary table does not appear in the corresponding address word, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 0.

[0026] By adopting the above technical solution, whether the address segmentation words of each user appear in the address vocabulary is converted into the address vector of the corresponding user in a binary manner, that is, it is recorded as 1 when it appears and 0 when it does not appear. This simplifies the expression of address features and reduces the interference of differences in the frequency of appearance of address segmentation words on abnormal identification.

[0027] Optionally, calculating the abnormality indicator value corresponding to the power usage address according to the address vector includes:

[0028] Obtaining a vector weight corresponding to each vector dimension in the address vector;

[0029] Substituting the vector weight and the address vector into an abnormality index calculation formula to obtain an abnormality index value corresponding to the electricity usage address;

[0030] The abnormality index calculation formula is: ;

[0031] Wherein, Z represents the abnormality indicator value, n represents the total number of vector dimensions of the address vector, i represents the i-th vector dimension, represents the vector value corresponding to the i-th vector dimension, Represents the vector weight corresponding to the i-th vector dimension.

[0032] By adopting the above technical solution, the vector weights corresponding to each vector dimension in the address vector are obtained, and the vector weights and address vectors are substituted into the anomaly index calculation formula to calculate the anomaly index value, thereby achieving differentiated treatment of the importance of different vector dimensions. At the same time, a weighted average method is used to comprehensively evaluate the degree of anomaly of the power usage address, so that the anomaly index value can more accurately reflect the abnormal characteristics of the power usage address.

[0033] Optionally, when the abnormal indicator value satisfies an abnormality judgment condition, determining the user corresponding to the abnormal indicator value as an abnormal user includes:

[0034] Filter out target users whose abnormal indicator values ​​are less than a first threshold;

[0035] If the abnormality indicator values ​​of the remaining users except the target user are all greater than the second threshold, the target user is determined to be an abnormal user, and the first threshold is less than the second threshold.

[0036] By adopting the above technical solution, target users whose abnormal indicator values ​​are less than the first threshold are screened out, and it is determined whether the abnormal indicator values ​​of the remaining users except the target users are all greater than the second threshold to determine the abnormal users. The dual-threshold judgment mechanism not only considers the abnormality level of the target user itself, but also combines the overall electricity consumption characteristics of surrounding users for comparative analysis, effectively avoiding the misjudgment that may be caused by a single threshold judgment, thereby improving the accuracy of household change abnormality identification.

[0037] In a second aspect, the present application provides a household change anomaly identification system based on electricity usage address, the system comprising:

[0038] An initialization module is used to obtain the electricity usage addresses of multiple users in the substation area and split the text of the electricity usage addresses into address words;

[0039] A vector conversion module, configured to count whether each word in the address vocabulary table appears in the corresponding address segmentation word, and generate an address vector corresponding to the power address;

[0040] A calculation module, configured to calculate an abnormality indicator value corresponding to the electricity usage address according to the address vector;

[0041] The identification module is configured to determine the user corresponding to the abnormal indicator value as an abnormal user when the abnormal indicator value meets the abnormal judgment condition.

[0042] In a third aspect, the present application provides a computer storage medium, which stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing any one of the above methods.

[0043] In a fourth aspect, the present application provides an electronic device comprising a processor, a memory and a transceiver, wherein the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs any one of the above methods.

[0044] In summary, the beneficial effects brought about by the technical solution of this application include:

[0045] By adopting the above technical solution, the electricity address text is segmented and analyzed and converted into a quantifiable address vector, and then the abnormal index value is calculated to make an abnormality judgment. Compared with the technical solutions in the related art, accurate physical address data is used to replace the incomplete latitude and longitude data for household change anomaly identification, avoiding missed judgments caused by missing some addresses. At the same time, based on the numerical analysis of the address text and the quantitative judgment of the abnormal index, an accurate quantitative evaluation of the electricity addresses of different users is achieved, thereby improving the accuracy of household change anomaly identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of a method for identifying abnormal changes in household electricity usage addresses according to an embodiment of the present application;

[0047] Figure 2 This is a structural diagram of a household change anomaly identification system based on electricity address according to an embodiment of the present application;

[0048] Figure 3 This is a structural diagram of an electronic device provided in an embodiment of the present application.

[0049] Description of reference numerals: 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0050] In order to enable people skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0051] In the description of the embodiments of this application, words such as "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0052] In the description of the embodiments of the present application, the term "multiple" means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0053] See Figure 1 This is a flow chart illustrating a method for identifying anomalies in a household power changer based on power usage addresses, provided in an embodiment of the present application. This method can be implemented using a computer program, a single-chip microcomputer, or run on a von Neumann architecture-based system for identifying anomalies in a household power changer based on power usage addresses. The computer program can be integrated into an application or run as a standalone tool application. The specific steps of the method for identifying anomalies in a household power changer based on power usage addresses are described in detail below.

[0054] S101: Obtain electricity usage addresses of multiple users in the substation area and split the electricity usage address text into address segmentation words;

[0055] The electricity address refers to the detailed geographic location of the user's electricity facilities. In this embodiment, it can be understood as a standardized text address containing a hierarchical structure such as province, city, district (county), street (township), community (village), building number, unit number, and room number. It is used to uniquely identify the power supply service location of each electricity user and serves as an important basis for user identification. Table 1 is provided below to help understand the electricity address.

[0056] Among them, address segmentation refers to the smallest semantic unit obtained by segmenting the complete text of the electricity address according to semantics and address components. In the embodiment of this application, it can be understood as dividing the standardized text address containing hierarchical structures such as province, city, district (county), street (township), community (village), building number, unit number, room number, etc. into independent address word elements, which are used to realize the normalization processing and vectorized representation of the address text.

[0057] In the specific implementation, the electricity address information of all users under the substation is first obtained from the power system database. Since the electricity address is a standardized text address with a hierarchical structure including province, city, district (county), street (township), community (village), building number, unit number, room number, etc., these addresses are usually stored in the form of continuous character strings, which is not convenient for direct abnormality analysis. Therefore, it is necessary to use Chinese word segmentation technology to split the obtained electricity address text and split each electricity address text into multiple address segmentations with independent semantics. Specifically, the Jieba word segmentation algorithm can be used, combined with the address dictionary unique to the power industry, to segment the electricity address text.

[0058] To facilitate understanding of the electricity address segmentation process, please refer to Tables 1 and 2. Table 1 shows the original electricity addresses of seven users within the distribution center. These addresses are all located in Province A, City B, District C, Street D, but are distributed across different neighborhoods, buildings, and apartment numbers. Table 2 shows the results of segmentation of these addresses. Each address is broken down into multiple independent semantic units, including the province "Province A", the city "City B", the region "District C", the street "Street D", the neighborhood name "Community A / Community B", the building number "Building 2 / Building 3", and the apartment number. This segmentation method converts continuous address text into a sequence of address segmentation words.

[0059] Table 1- Example table of electricity address:

[0060]

[0061] Table 2 - Example of address segmentation results:

[0062]

[0063] Based on the above embodiment, as an optional implementation, the step of generating an address vocabulary table specifically includes S201-S203.

[0064] S201: Filter out unique address segmentation words from the address segmentation words, and form an initial vocabulary list with all the unique address segmentation words.

[0065] Specifically, before analyzing electricity address anomalies, it is necessary to sort and remove duplicate address words for all users in the substation area. Since each user's electricity address is segmented into multiple address words, these address words are repeated between different users, such as the same province, city, district, and county administrative division information. Therefore, it is necessary to traverse the address words of all users and use a set data structure to remove duplicate address words to obtain an initial vocabulary containing all unique address words. For example, the initial vocabulary that can be constructed from the above Table 2 is: 101, 102, 2, 201, 202, 3, Community B, Street D, Building, City B, Province A, Community A, District C.

[0066] S202: Calculate the context occurrence probability of each word in the initial vocabulary using an N-gram model;

[0067] Among them, the N-gram model refers to a language model that segments text according to N consecutive characters or words. In the embodiment of the present application, it can be understood as dividing the words in the address vocabulary table into several overlapping substring sequences according to N consecutive characters, which is used to capture the local continuous features of the characters in the table.

[0068] The address vocabulary table refers to the initial vocabulary table after sorting.

[0069] In practice, the N-gram model can be used to predict the probability distribution of word sequences in text. Its core assumption is that the probability of the Nth word occurring is dependent only on the preceding N-1 words, not the entire context. The N-gram model is a commonly used statistical language model in natural language processing (NLP) and is used to predict the next element in a sequence (e.g., the next word in a text). The N-gram model can be used to sort the words in a bag-of-words vocabulary. This allows words with strong local correlations to be placed adjacent to each other, resulting in an address vocabulary that is arranged from largest to smallest region.

[0070] In specific implementation, a custom start word (such as "START") is first added to the beginning of each address segmentation word to mark the starting boundary of the address text, thereby constructing a training corpus. For example, the address segmentation word "XX community XX street" becomes "START XX community XX street" after adding the start word. Next, an N-gram model is constructed based on this training corpus. When N is 2 (i.e., a Bi-gram model), the system counts the frequency of word pairs and calculates the conditional probability of the next word appearing given the previous word. For example, the probability of "street" appearing after "community" can be calculated to obtain the contextual occurrence probability of each word in the initial vocabulary. This probabilistic analysis method captures the associations between words in the address text and effectively identifies key words in the address structure. The calculated contextual occurrence probability not only reflects the importance of a word in the address text, but also includes its positional information and semantic association information. These probabilities provide a quantitative basis for subsequent word sorting and screening, enabling the system to construct an address vocabulary containing significant characteristic words, thereby improving the expressive power of address vectors and the accuracy of anomaly detection. At the same time, this N-gram-based probability calculation method has the characteristics of strong interpretability and high computational efficiency, and is suitable for processing large-scale electricity address data.

[0071] Regarding the step of using the N-gram model to calculate the contextual occurrence probability corresponding to each word in the initial vocabulary, a specific implementation method is: adding a custom starting word to each of the address segmentation words to obtain a training corpus; constructing an N-gram model based on the training corpus, and using the N-gram model to calculate the contextual occurrence probability corresponding to each word in the initial vocabulary.

[0072] In specific implementation, the custom start word "START" is first added to the beginning of each user's address word sequence. For example, the address word "C District D Street" is processed as "START C District D Street". All address word sequences processed in this way constitute the training corpus. The purpose of adding the start word is to mark the starting boundary of each address text, enabling the system to identify and learn the characteristics of words appearing at the beginning of the address. Next, an N-gram model is constructed based on this training corpus. When using the bigram model, the system calculates the conditional probability P(word|previous word), specifically P(word|"START"), which calculates the probability of a particular word appearing as the first word in the address. For example, the system can calculate the probability of "district" appearing after "START" and before "C", thereby obtaining the probability distribution of the word in different contextual positions. This N-gram-based probability calculation method not only considers the independent occurrence frequency of words but also incorporates the positional association information between words, enabling the system to accurately grasp the occurrence patterns of words in address text.

[0073] S203: Sort each word in the initial vocabulary according to the context occurrence probability to obtain an address vocabulary.

[0074] In practice, the system uses the contextual probability calculated by the N-gram model as the basis for sorting. For example, given the calculated conditional probabilities P(word|"START") and P(word|previous word), the system sorts the words in the initial vocabulary from high to low based on these probability values. A high probability value indicates that the word is more likely to appear in a specific context and has stronger address characteristics.

[0075] Based on the above embodiment, as an optional implementation, the head word and the tail word in the address segmentation word are deleted.

[0076] To address the high-dimensional sparsity problem of address vectors caused by an overly large vocabulary, we reduce the vector dimensionality by processing the head and tail words in the address segmentation. In real-world electricity usage scenarios, normal users in the same electricity distribution area often have similar addresses. The main difference is in the final address information, such as the room number, such as "101" and "102." Since this final difference does not affect the identification of household change anomalies, words that only appear at the end can be deleted. Furthermore, given that household change anomalies typically do not cross urban areas, administrative division information at the head of the address, such as "Province A" and "City B," contributes little to anomaly detection and can be removed. This targeted dimensionality reduction method not only retains the address features that are most critical for identifying household change anomalies, but also significantly reduces the vector dimensionality, improving the algorithm's computational efficiency.

[0077] For example, after the address segmentation result example table in Table 2 is processed through the above steps, the address vector obtained is shown in Table 3. The order of the words in Table 3 has been sorted through the above steps.

[0078] Table 3 - Address vector result table:

[0079]

[0080] S102: Count whether each word in the address vocabulary table appears in the corresponding address word, and generate an address vector corresponding to the electricity address.

[0081] This application uses the Bag of Words (BoW) model to convert text data into numerical data. The Bag of Words model is a model used in natural language processing and text mining. It works by creating a vocabulary containing all unique words that appear in all texts. Then, for each text, it counts whether each word in the vocabulary appears in that text, forming a vector with the same length as the vocabulary.

[0082] Among them, the address vector refers to the conversion of the address segmentation of the user's electricity address into a numerical form. In the embodiment of the present application, it can be understood as a numerical vector obtained by establishing a vocabulary list based on the address segmentation of all users in the station area based on the bag-of-words model, and then counting whether each word in the address vocabulary list appears in the corresponding address segmentation. It is used to convert the text-based electricity address into a numerical representation that can be used for mathematical operations.

[0083] Among them, the address vocabulary refers to a non-repeating ordered vocabulary set formed by the electricity address segmentation words of all users within the substation area. In the embodiment of the present application, it can be understood as a list of all unique address segmentation words such as provinces, cities, districts, counties, streets, communities, building numbers, unit numbers, room numbers, etc. obtained after deduplication of the address segmentation words of all users in the substation area. It is used to construct a bag of words model, convert the electricity address into an address vector, and realize the mapping of address text to numerical form.

[0084] For each electricity address, the system matches its address segmentation words with words in the address vocabulary. When a word in the address vocabulary appears in the corresponding address segmentation word, the corresponding vector value of the word in the address vector is marked as 1, otherwise it is marked as 0. For example, if A's address segmentation words are 1, 3, 5 and the address vocabulary is 1, 2, 3, 4, 5, 6, then A's address vector is 1, 0, 1, 0, 1, 0.

[0085] Specifically, step S102 includes the following steps: if a word in the address vocabulary appears in the corresponding address word, the vector value of the address vector of the corresponding word in the address vocabulary is recorded as 1; if the word in the address vocabulary does not appear in the corresponding address word, the vector value of the address vector of the corresponding word in the address vocabulary is recorded as 0.

[0086] When constructing the word frequency vector, taking into account the particularity of electricity addresses, the same address segmentation word will usually only appear once in a single user's electricity address. For example, an address will not repeat "Province A" or multiple room numbers multiple times. Therefore, using a binary method to construct the word frequency vector is more in line with the actual situation. In specific implementation, for each user's electricity address, the words in the address vocabulary table are traversed, and each word is checked one by one to see if it appears in the corresponding address segmentation word: if the word in the address vocabulary table appears in the corresponding address segmentation word, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 1; if the word in the address vocabulary table does not appear in the corresponding address segmentation word, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 0.

[0087] S103: Calculating the abnormality index value corresponding to the power consumption address according to the address vector;

[0088] Among them, the abnormality index value refers to a quantitative indicator that measures the degree of difference between the electricity usage address and the addresses of other users in the substation to which it belongs. In the embodiment of the present application, it can be understood as a value calculated based on the address vector to characterize the abnormality degree of a single electricity usage address, which is used to quantitatively evaluate whether the electricity usage address is an abnormal address for the household change.

[0089] Specifically, for each address vector, we calculate its difference characteristics with other user address vectors within the same substation to obtain a set of difference values. We then construct an anomaly index by statistically analyzing the distribution characteristics of these difference values. This anomaly index calculation method based on vector differences effectively measures the degree of difference between a single electricity user address and other addresses in the substation. The greater the difference, the more different the address is from other users in the substation, and the more likely it is to be an abnormal user.

[0090] Based on the above embodiment, as an optional implementation, the abnormality indicator value may be calculated in the following manner: obtaining the vector weight corresponding to each vector dimension in the address vector;

[0091] Substitute the vector weight and address vector into the abnormal index calculation formula to obtain the abnormal index value corresponding to the electricity address;

[0092] The calculation formula of abnormal index is: ;

[0093] Among them, Z represents the abnormal index value, n represents the total number of vector dimensions of the address vector, and i represents the i-th vector dimension. Represents the vector value corresponding to the i-th vector dimension, Represents the vector weight corresponding to the i-th vector dimension.

[0094] First, the vector weights corresponding to each vector dimension in the address vector are obtained. These weights reflect the importance of each dimension in identifying household power anomalies. Then, the obtained vector weights and the address vector are substituted into the anomaly index calculation formula, where the anomaly index value is equal to the sum of the products of the vector values ​​corresponding to the vector dimensions and the vector weights divided by the sum of the vector weights. The purpose of adopting this weighted calculation method is to consider the different contributions of different dimensions in the address vector to the determination of household power anomalies. By introducing vector weights, the influence of important dimensions can be emphasized and the influence of minor dimensions can be weakened. For example, higher weights are assigned to dimensions that better reflect the core characteristics of the address, while lower weights are assigned to detailed information dimensions. This allows the calculated anomaly index value to more accurately reflect the degree of anomaly of the electricity usage address.

[0095] Regarding vector weights, they quantify the importance of each dimension of the address vector. Their setting requires comprehensive consideration of actual business scenarios and data characteristics. In practical applications, vector weights can be determined based on the following factors: First, because the address terms have been sorted, the closer the different terms appear to the front, the greater the address difference. For example, differences at the "street" level are greater than those at the "district" level, because the closer the terms appear to the front, the greater the weight. Third, based on the anomaly cases of household changes discovered in historical data, the contribution of different address dimensions to anomaly identification can be analyzed, and dimensions with good identification results can be given higher weights. In specific implementation, a data-driven approach can be adopted. By analyzing large amounts of historical data, the ability of each dimension to distinguish between normal and abnormal users can be calculated, and this ability to distinguish can be converted into corresponding weight values.

[0096] Exemplarily, the present application performs calculations on the address vectors in Table 3 and obtains the abnormality indicator value distribution table shown in Table 4.

[0097] Table 4-Abnormal index value distribution table:

[0098]

[0099] S104: When the abnormal indicator value meets the abnormality judgment condition, the user corresponding to the abnormal indicator value is determined as an abnormal user.

[0100] In this embodiment, by comparing the abnormality indicator value with the preset abnormality judgment condition, abnormal users can be effectively identified. When the calculated abnormality indicator value meets the abnormality judgment condition, the user corresponding to the abnormality indicator value can be determined as an abnormal user.

[0101] Based on the above embodiment, as an optional implementation, step S104 specifically further includes steps S301 - S302 .

[0102] The abnormal judgment conditions are divided into two judgment steps to accurately identify abnormal users.

[0103] S301: Filter out target users whose abnormality indicator values ​​are less than a first threshold;

[0104] In this embodiment, by setting a first threshold and screening target users whose abnormal index values ​​are less than the threshold, potential users with abnormal household changes can be preliminarily identified. In specific implementation, first set a suitable first threshold based on historical data and actual business experience, then compare the abnormal index values ​​of all users with the threshold, and screen out users whose abnormal index values ​​are less than the first threshold as target users. The purpose of setting the first threshold is to establish a preliminary screening standard, because the smaller the abnormal index value, the greater the difference between the user's electricity address and the addresses of other users in the substation area, and the more likely there is an abnormal household change. This threshold-based screening method can quickly locate suspicious user groups and improve the efficiency of subsequent verification work.

[0105] S302: If the abnormality indicator values ​​of the remaining users except the target user are all greater than the second threshold, the target user is determined to be an abnormal user, and the first threshold is less than the second threshold.

[0106] By introducing a second threshold and combining it with the first threshold to create a dual-threshold judgment mechanism, we can more accurately identify abnormal users with household changes. In specific implementation, we first ensure that the set first threshold is less than the second threshold. Then, we check whether the abnormality index values ​​of all users except the target user are greater than the second threshold. If this condition is met, the target user is identified as an abnormal user. This dual-threshold judgment mechanism aims to further verify the abnormality of the target user by comparing and analyzing the distribution of abnormality index values ​​between the target user and other users. When the target user's abnormality index value is significantly lower than the first threshold, while the abnormality index values ​​of other users are all higher than the second threshold, it indicates that the target user's address characteristics differ significantly from those of surrounding users, and the degree of this difference exceeds the normal fluctuation range. Therefore, it can be more reliably determined to be an abnormal user with household changes. This dual-threshold judgment method can effectively reduce the false positive rate and improve the accuracy of identifying abnormal users with household changes. For example, when a user within a substation area exhibits a significant gap in the abnormality index value distribution (such as user 7 in Table 3), that is, when the abnormality index value of a particular user is significantly lower than the first threshold, while the abnormality index values ​​of other users are all higher than the second threshold, the true abnormal user with household changes can be more accurately identified.

[0107] The following are system embodiments of the present application, which can be used to implement the method embodiments of the present application. For details not disclosed in the system embodiments of the present application, please refer to the method embodiments of the present application.

[0108] See Figure 2, which shows a schematic diagram of the structure of a household change anomaly identification system based on electricity usage address provided by an exemplary embodiment of the present application. The system can be implemented as all or part of the system through software, hardware, or a combination of both. The household change anomaly identification system based on electricity usage address includes:

[0109] The initialization module is used to obtain the electricity usage addresses of multiple users in the substation area and split the electricity usage address text into address words;

[0110] A vector conversion module, configured to count whether each word in the address vocabulary table appears in the corresponding address segmentation word, and generate an address vector corresponding to the power address;

[0111] A calculation module, used to calculate the abnormal index value corresponding to the power consumption address according to the address vector;

[0112] The identification module is used to determine the user corresponding to the abnormal indicator value as an abnormal user when the abnormal indicator value meets the abnormal judgment condition.

[0113] Based on the above embodiment, as an optional embodiment, the initialization module is further used to filter out unique address participles from the address participles, and form an initial vocabulary list with all the unique address participles; use the N-gram model to calculate the context occurrence probability corresponding to each word in the initial vocabulary; and sort each word in the initial vocabulary according to the context occurrence probability to obtain an address vocabulary list.

[0114] Based on the above embodiment, as an optional embodiment, the initialization module is further used to add a custom starting word to each of the address segmentation words to obtain a training corpus; construct an N-gram model based on the training corpus, and use the N-gram model to calculate the context occurrence probability corresponding to each word in the initial vocabulary.

[0115] Based on the above embodiment, as an optional embodiment, the initialization module is further configured to delete the head word and the tail word in the address segmentation word.

[0116] Based on the above embodiment, as an optional embodiment, the vector conversion module is also used to record the vector value of the address vector of the corresponding vocabulary in the address vocabulary as 1 if the vocabulary in the address vocabulary appears in the corresponding address participle; if the vocabulary in the address vocabulary does not appear in the corresponding address participle, then the vector value of the address vector of the corresponding vocabulary in the address vocabulary is recorded as 0.

[0117] Based on the above embodiment, as an optional embodiment, the calculation module is further used to obtain the vector weight corresponding to each vector dimension in the address vector; substitute the vector weight and the address vector into the abnormality index calculation formula to obtain the abnormality index value corresponding to the power address; the abnormality index calculation formula is: ;

[0118] Among them, Z represents the abnormal index value, n represents the total number of vector dimensions of the address vector, and i represents the i-th vector dimension. Represents the vector value corresponding to the i-th vector dimension, Represents the vector weight corresponding to the i-th vector dimension.

[0119] Based on the above embodiment, as an optional embodiment, the identification module is also used to screen out target users whose abnormal indicator values ​​are less than a first threshold; if the abnormal indicator values ​​of the remaining users except the target user are all greater than a second threshold, the target user is determined to be an abnormal user, and the first threshold is less than the second threshold.

[0120] An embodiment of the present application also provides a computer storage medium, which can store multiple instructions. The instructions are suitable for being loaded and executed by a processor as in the above-mentioned embodiment of the household change anomaly identification method based on the electricity address. The specific execution process can be found in the specific description of the embodiment and will not be repeated here.

[0121] See Figure 3 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 3 As shown, the electronic device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

[0122] The communication bus 302 is used to implement the connection and communication between these components.

[0123] The user interface 303 may include a standard wired interface or a wireless interface.

[0124] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0125] The processor 301 may include one or more processing cores. Using various interfaces and circuits, the processor 301 connects to various components within the server. It executes instructions, programs, code sets, or instruction sets stored in the memory 305, as well as accesses data stored in the memory 305, to perform various server functions and process data. Optionally, the processor 301 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 301 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing content displayed on the display screen; and the modem handles wireless communications. It is understood that the modem may not be integrated into the processor 301 but implemented as a separate chip.

[0126] Among them, the memory 305 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may also be optionally at least one storage device located away from the aforementioned processor 301. As Figure 3 As shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program of a method for identifying abnormal changes in a user's electricity address.

[0127] exist Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 301 can be used to call an application program stored in the memory 305 for a method for identifying abnormal changes in a user's electricity address based on electricity usage. When executed by one or more processors, the electronic device executes one or more methods in the above-mentioned embodiments.

[0128] An electronic device readable storage medium stores instructions, which, when executed by one or more processors, enable the electronic device to execute one or more methods in the above embodiments.

[0129] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0130] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0132] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0133] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0134] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of this application, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of this application. The aforementioned memory includes various media that can store program code, such as USB flash drives, mobile hard drives, magnetic disks, or optical disks.

[0135] The above are merely exemplary embodiments of the present disclosure and are not intended to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any variations, uses or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the field of the present disclosure that are not recorded in the present disclosure.

Claims

1. A method for identifying abnormal changes in household electricity usage based on electricity usage addresses, characterized in that: The method comprises: Obtain the electricity usage addresses of multiple users in the substation area, and split the text of the electricity usage addresses into address words; Counting whether each word in the address vocabulary table appears in the corresponding address segmentation word, and generating an address vector corresponding to the power address; Before counting whether each word in the address vocabulary table appears in the corresponding address segmentation word and generating the address vector corresponding to the power address, the method further includes: Filtering out unique address segmentations from the address segmentations, and forming an initial vocabulary list with all the unique address segmentations; calculating a contextual occurrence probability corresponding to each word in the initial vocabulary list using an N-gram model; and sorting each word in the initial vocabulary list according to the contextual occurrence probability to obtain an address vocabulary list; Calculating the abnormality index value corresponding to the electricity usage address according to the address vector; Calculating the abnormality index value corresponding to the power usage address based on the address vector includes: obtaining a vector weight corresponding to each vector dimension in the address vector; substituting the vector weight and the address vector into an abnormality index calculation formula to obtain an abnormality index value corresponding to the power usage address; the abnormality index calculation formula is: ; Wherein, Z represents the abnormality indicator value, n represents the total number of vector dimensions of the address vector, and i represents the i-th vector dimension, represents the vector value corresponding to the i-th vector dimension, represents the vector weight corresponding to the i-th vector dimension; When the abnormal indicator value meets the abnormality judgment condition, the user corresponding to the abnormal indicator value is determined as an abnormal user.

2. The method according to claim 1, characterized in that The N-gram model is used to calculate the context occurrence probability corresponding to each word in the initial vocabulary, including: Adding a custom starting word to each of the address segmentation words to obtain a training corpus; An N-gram model is constructed based on the training corpus, and the context occurrence probability corresponding to each word in the initial vocabulary is calculated using the N-gram model.

3. The method according to claim 1, characterized in that The method of obtaining the electricity usage addresses of multiple users in the substation area and splitting the text of the electricity usage addresses into address words further includes: Delete the head word and the tail word in the address segmentation word.

4. The method according to claim 1, wherein The counting of whether each word in the address vocabulary table appears in the corresponding address segmentation word to generate an address vector corresponding to the power address includes: If a word in the address vocabulary table appears in the corresponding address segment, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 1; If a word in the address vocabulary table does not appear in the corresponding address word, the vector value of the address vector of the corresponding word in the address vocabulary table is recorded as 0.

5. The method according to claim 1, wherein When the abnormal indicator value satisfies the abnormality judgment condition, determining the user corresponding to the abnormal indicator value as an abnormal user includes: Filter out target users whose abnormal indicator values ​​are less than a first threshold; If the abnormality indicator values ​​of the remaining users except the target user are all greater than the second threshold, the target user is determined to be an abnormal user, and the first threshold is less than the second threshold.

6. A household change anomaly identification system based on electricity address, characterized in that: The system comprises: An initialization module is used to obtain the electricity usage addresses of multiple users in the substation area and split the text of the electricity usage addresses into address words; A vector conversion module is configured to count whether each word in the address vocabulary table appears in the corresponding address segmentation word and generate an address vector corresponding to the power address; before counting whether each word in the address vocabulary table appears in the corresponding address segmentation word and generating the address vector corresponding to the power address, the module further includes: Filtering out unique address segmentations from the address segmentations, and forming an initial vocabulary list with all the unique address segmentations; calculating a contextual occurrence probability corresponding to each word in the initial vocabulary list using an N-gram model; and sorting each word in the initial vocabulary list according to the contextual occurrence probability to obtain an address vocabulary list; A calculation module is configured to calculate an abnormality index value corresponding to the power usage address based on the address vector; the calculation of the abnormality index value corresponding to the power usage address based on the address vector includes: obtaining a vector weight corresponding to each vector dimension in the address vector; substituting the vector weight and the address vector into an abnormality index calculation formula to obtain an abnormality index value corresponding to the power usage address; the abnormality index calculation formula is: ; Wherein, Z represents the abnormality indicator value, n represents the total number of vector dimensions of the address vector, and i represents the i-th vector dimension, represents the vector value corresponding to the i-th vector dimension, represents the vector weight corresponding to the i-th vector dimension; The identification module is configured to determine the user corresponding to the abnormal indicator value as an abnormal user when the abnormal indicator value meets the abnormal judgment condition.

7. A computer storage medium, characterized in that The computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method according to any one of claims 1 to 5.

8. An electronic device, characterized in that: The electronic device comprises a processor, a memory and a transceiver, wherein the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of claims 1 to 5.

Images