Electric charge payment risk assessment method and related equipment

Abstract

The invention provides an electric charge payment risk assessment method and related equipment. The electric charge payment risk assessment method comprises the steps of determining a user historical payment behavior sequence and user historical electricity utilization characteristics of electricity utilization users in a target area; determining a user payment mode according to the user historical payment behavior sequence and the user historical electricity utilization characteristics of any electricity utilization user; according to the user payment mode, payment behavior digital characteristics are determined; and according to the payment behavior digital characteristics and the comprehensive information of the power utilization user, carrying out risk assessment on the electric charge payment behavior. The user payment mode is determined according to the user historical payment behaviors and the user historical electricity consumption characteristics, the reason of risk generation can be visually reflected by analyzing the user payment mode, the interpretability of the model is improved, the electricity consumption user comprehensive information is fused, and the user experience is improved. According to the method and the device, the problems of high subjectivity and low efficiency due to dependence on subjective rules are avoided.

Description

Technical Field

[0001] The present application relates to the field of data analysis and application technology, and in particular to an electricity bill payment risk assessment method and related equipment. Background Art

[0002] Electricity revenue is the lifeblood of power companies, and the rate of electricity recovery directly affects the normal operation of the company. In the current volatile economic environment, companies are generally facing the pressure of transformation and adjustment, which makes electricity fee recovery face greater difficulties and risks. At the same time, large users usually adopt the method of "using electricity first and paying later", which also leads to the problem of electricity bill backlog. These factors work together to seriously hinder the timely recovery of electricity bills and threaten the sustainable development of enterprises. With the development of smart grids, power companies are faced with the problem of collecting and analyzing a large amount of user electricity consumption data. How to use these massive data to achieve accurate management of electricity users and improve the efficiency of electricity fee recovery is a difficult problem that power companies need to solve urgently.

[0003] In terms of risk assessment of user electricity bill recovery, traditional methods mainly rely on manual experience to make judgments, which are highly subjective and inefficient. Summary of the invention

[0004] In view of this, the purpose of this application is to propose an electricity bill payment risk assessment method and related equipment.

[0005] Based on the above purpose, this application provides a method for assessing the risk of electricity bill payment, including:

[0006] Determine the historical payment behavior sequence and historical electricity consumption characteristics of electricity users in the target area;

[0007] Determining a user payment mode according to the user's historical payment behavior sequence and the user's historical electricity usage characteristics of any electricity user;

[0008] Determining payment behavior digital features based on the user payment mode;

[0009] Based on the digital characteristics of the payment behavior and the comprehensive information of electricity users, a risk assessment of electricity bill payment behavior is conducted.

[0010] Optionally, determining the user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user includes:

[0011] Performing data mapping on the user's historical payment behavior sequence and the user's historical electricity usage characteristics through a kernel function to determine a high-dimensional user's historical payment behavior sequence and a high-dimensional user's historical electricity usage characteristics;

[0012] Determine the kernel function value of the high-dimensional user's historical payment behavior sequence and the high-dimensional user's historical electricity consumption characteristics, and determine the kernel matrix according to the kernel function value;

[0013] Low-frequency feature data is extracted from the kernel matrix, and data iteration is performed according to the low-frequency feature data to determine the user payment mode.

[0014] Optionally, extracting low-frequency feature data from the kernel matrix includes:

[0015] Performing low-rank decomposition on the kernel matrix, determining low-frequency feature data, and calculating the eigenvalue of any low-frequency feature data;

[0016] The low-frequency feature data are sorted according to the feature values, and feature data are extracted according to the sorting matrix.

[0017] Optionally, performing data iteration according to the low-frequency feature data to determine the user payment mode includes:

[0018] According to the characteristic data, data iteration is performed through a hierarchical clustering algorithm to determine the user payment mode; wherein the user payment mode includes: advance payment mode, on-time payment mode, intermittent payment mode, delayed payment mode, overdue payment mode and overdue power outage mode.

[0019] Optionally, determining the digital features of the payment behavior according to the user payment mode includes:

[0020] Determine the frequency of occurrence of any of the user payment patterns within a target time period, quantify the frequency of occurrence, and determine the digital characteristics of the payment behavior.

[0021] Optionally, the risk assessment of electricity payment behavior is performed based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user, including:

[0022] According to the digital characteristics of the payment behavior and the comprehensive information of the electricity user, the electricity bill payment risk level is determined through a pre-constructed electricity bill payment risk assessment model.

[0023] Based on the same inventive concept, the embodiment of the present application also provides an electricity bill payment risk assessment device, comprising:

[0024] A data determination module is configured to determine a user's historical payment behavior sequence and a user's historical electricity usage characteristics of electricity users in a target area;

[0025] A payment mode determination module, configured to determine a user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user;

[0026] a feature determination module, configured to determine a digital feature of a payment behavior according to the user's payment mode;

[0027] The payment risk assessment module is configured to perform risk assessment on electricity bill payment behavior based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user.

[0028] Based on the same inventive concept, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, an electricity bill payment risk assessment method as described in any one of the above items is implemented.

[0029] Based on the same inventive concept, an embodiment of the present application also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute any of the above-mentioned electricity bill payment risk assessment methods.

[0030] Based on the same inventive concept, an embodiment of the present application further provides a computer program product, including computer program instructions. When the computer program instructions are executed on a computer, the computer executes any of the above-mentioned electricity bill payment risk assessment methods.

[0031] From the above, it can be seen that the electricity bill payment risk assessment method and related equipment provided by the present application include: determining the user's historical payment behavior sequence and user's historical electricity usage characteristics of electricity users in the target area; determining the user's payment mode according to the user's historical payment behavior sequence and the user's historical electricity usage characteristics of any electricity user; determining the payment behavior digital characteristics according to the user's payment mode; and conducting electricity bill payment behavior risk assessment according to the payment behavior digital characteristics and comprehensive information of electricity users. The present application determines the user's payment mode based on the user's historical payment behavior and the user's historical electricity usage characteristics. By analyzing the user's payment mode, it can intuitively reflect the cause of the risk, improve the interpretability of the model, and at the same time integrate the comprehensive information of electricity users, that is, the user's behavior data, user attributes and other multi-source information to conduct electricity bill payment behavior risk assessment, avoiding the problem of high subjectivity and low efficiency of relying solely on subjective rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1A schematic diagram of a flow chart of an electricity bill payment risk assessment method according to an embodiment of the present application;

[0034] Figure 2 A schematic diagram of a hard-ε band hyperplane according to an embodiment of the present application;

[0035] Figure 3 A schematic diagram of the classification surface of the positive point set and the negative point set in the embodiment of the present application;

[0036] Figure 4 A schematic diagram of the flow of the Monte Carlo tree search algorithm of an embodiment of the present application;

[0037] Figure 5 This is a schematic diagram of the structure of an electricity bill payment risk assessment device according to an embodiment of the present application;

[0038] Figure 6 A schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0040] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0041] In order to facilitate understanding of the technical solutions of the present disclosure, some technical terms involved in the present disclosure are introduced below.

[0042] Hierarchical kernel PCA (Kernel Principal Component Analysis, KPCA, a nonlinear dimensionality reduction method) uses kernel techniques to map the input space into a high-dimensional feature space, and performs PCA dimensionality reduction in this high-dimensional space, thereby achieving nonlinear dimensionality reduction.

[0043] Gaussian RBF kernel (Radial Basis Function, RBF, also known as radial basis function, Gaussian Kernel) is used to map finite-dimensional data to high-dimensional space, making the originally linearly inseparable data separable.

[0044] Low-Rank Factorization (a matrix analysis technique) is used to decompose a high-dimensional matrix into the product of two or more low-rank matrices, thereby simplifying the data structure and reducing computational complexity and storage requirements.

[0045] Hierarchical clustering algorithm, an unsupervised learning method, is used to classify data samples into different categories or clusters. The algorithm calculates the distance between samples and gradually merges or splits data points to form a hierarchical clustering tree.

[0046] SMOTE, Synthetic Minority Over-sampling Technique, is artificial minority class oversampling technology.

[0047] In order to make the technical solution of the present disclosure clearer and easier to understand, the electricity fee payment risk assessment method provided by the embodiment of the present disclosure is described in detail below with reference to the accompanying drawings.

[0048] As described in the background technology section, electricity revenue is the lifeblood of power companies, and the electricity recovery rate directly affects the normal operation of the company. In the current volatile economic environment, companies are generally facing the pressure of transformation and adjustment, which makes the recovery of electricity bills face greater difficulties and risks. At the same time, large users usually adopt the method of "first use electricity, then pay", which also leads to the problem of backlog of electricity bills. These factors work together to seriously hinder the timely recovery of electricity bills and threaten the sustainable development of enterprises. With the development of smart grids, power companies are faced with the problem of collecting and analyzing a large amount of user electricity consumption data. How to use these massive data to achieve accurate management of electricity users and improve the efficiency of electricity bill recovery is a difficult problem that power companies urgently need to solve. In terms of user electricity bill recovery risk assessment, traditional methods mainly rely on manual experience for judgment, which has the problems of high subjectivity and low efficiency.

[0049] In view of this, the embodiments of the present application provide a method, device, electronic device, storage medium and program product for assessing the risk of electricity bill payment. The method for assessing the risk of electricity bill payment includes: determining the user's historical payment behavior sequence and the user's historical electricity usage characteristics of electricity users in the target area; determining the user's payment mode according to the user's historical payment behavior sequence and the user's historical electricity usage characteristics of any electricity user; determining the digital characteristics of the payment behavior according to the user's payment mode; and conducting an electricity bill payment behavior risk assessment according to the digital characteristics of the payment behavior and the comprehensive information of the electricity user. The present application determines the user's payment mode according to the user's historical payment behavior and the user's historical electricity usage characteristics. By analyzing the user's payment mode, it can intuitively reflect the cause of the risk, improve the interpretability of the model, and integrate the comprehensive information of the electricity user, that is, the user's behavior data, user attributes and other multi-source information to conduct an electricity bill payment behavior risk assessment, avoiding the problem of high subjectivity and low efficiency of relying solely on subjective rules.

[0050] like Figure 1 As shown, the electricity fee payment risk assessment method includes:

[0051] Step S102: determining the user's historical payment behavior sequence and user's historical electricity usage characteristics of the electricity users in the target area;

[0052] Step S104: determining a user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user;

[0053] Step S106: Determine the digital characteristics of the payment behavior according to the user payment mode;

[0054] Step S108: Conduct risk assessment of electricity payment behavior based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user.

[0055] This application is based on two-stage hierarchical kernel PCA hierarchical clustering and model fusion for electricity bill payment behavior pattern recognition and risk assessment. The overall process is implemented in stages, from payment behavior analysis to risk prediction. The two-stage approach realizes the modeling and risk prediction of complex patterns of user electricity consumption behavior, solving the inaccurate prediction caused by the traditional single prediction process.

[0056] In step S102, the target area may be an area divided by administrative area, including urban areas and township areas. Urban areas: including areas within administrative divisions such as municipalities, provincial capitals, prefecture-level cities, and county-level cities. These areas usually have higher electricity demand and load density. It may also be an area divided by function or a special-purpose area, wherein the area divided by function may be a residential area, a commercial area, an industrial area, or an agricultural area. Residential areas are areas dominated by residences, and electricity demand is mainly concentrated in residents' daily life, such as lighting, home appliance use, etc.; commercial areas are areas dominated by commercial activities, such as shopping centers, commercial streets, business districts, etc. These areas have large electricity demand and are greatly affected by factors such as business hours and holidays. Industrial areas are areas dominated by industrial production, including various factories, industrial parks, etc. These areas have large and stable electricity demand and high requirements for power supply reliability and power quality. Agricultural areas are areas dominated by agricultural production, including farmland, orchards, farms, etc. The electricity demand in these areas has certain seasonal characteristics, such as large electricity demand in irrigation and harvesting seasons. Special areas can be tourist attractions or areas with transportation hubs. Tourist attractions include scenic spots, theme parks, and resorts. These areas have a large increase in electricity demand during the peak tourist season, and have high requirements for power supply reliability and service quality. Transportation hubs include airports, railway stations, and bus stations. As distribution centers for people and logistics, these areas have high requirements for power supply security and emergency response capabilities.

[0057] Further, in step S102, the user's historical payment behavior sequence is a multi-dimensional user electricity bill payment feature, such as the average payment time interval, the standard deviation of the payment amount, the average electricity bill payment amount in the payment cycle, the average payment interval, the payment method of the electricity bill (cash, bank transfer, electronic channel, etc.), the number of payment channel changes, the continuous payment duration, the monthly electricity bill payment frequency factor, etc. The user's historical electricity consumption feature is a nonlinear feature, such as the electricity consumption sudden increase feature, the electricity consumption fluctuation feature, the electricity consumption step feature, the cycle abnormality feature, and the phase abnormality feature.

[0058] Among them, the characteristics of sudden increase in electricity consumption include: monthly electricity consumption increased by more than 50% compared with the same period in history, and daily electricity consumption increased by more than 20% compared with the same period last week; the characteristics of electricity consumption fluctuation include: the fluctuation range of daily electricity consumption increased by more than 50% compared with the same period in history, and the standard deviation of weekly electricity consumption increased by more than 40%; the characteristics of electricity consumption step include: the electricity consumption jumped by more than 30% in adjacent time intervals and there are multiple large jump points in the electricity consumption curve; the characteristics of periodic abnormality include: the periodic fluctuation pattern of daily electricity consumption changed significantly and the fluctuation range of weekly electricity consumption exceeded the normal range of the same period in history; the characteristics of phase abnormality include: there is an obvious imbalance in electricity consumption in different phases and the relative relationship between electricity consumption in different phases is distorted.

[0059] In some optional embodiments, determining the user payment mode based on the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user includes: performing data mapping on the user historical payment behavior sequence and the user historical electricity usage characteristics through a kernel function to determine a high-dimensional user historical payment behavior sequence and a high-dimensional user historical electricity usage characteristics; determining the kernel function values ​​of the high-dimensional user historical payment behavior sequence and the high-dimensional user historical electricity usage characteristics, and determining a kernel matrix based on the kernel function values; extracting low-frequency feature data from the kernel matrix, and performing data iteration based on the low-frequency feature data to determine the user payment mode.

[0060] Furthermore, the user's historical payment behavior sequence {x1, x2, ... xn} and the user's historical electricity usage characteristics may be data mapped by using a Gaussian RBF kernel function to determine a high-dimensional user's historical payment behavior sequence and a high-dimensional user's historical electricity usage characteristics.

[0061] In some optional embodiments, the extracting low-frequency feature data from the kernel matrix includes: performing low-rank decomposition on the kernel matrix, determining low-frequency feature data, and calculating the eigenvalues ​​of any low-frequency feature data; sorting the low-frequency feature data according to the eigenvalues, and extracting feature data according to the sorted matrix.

[0062] Specifically, the user's historical payment behavior sequence and the user's historical electricity consumption characteristics are mapped to a high-dimensional feature space through the Gaussian RBF kernel function k(x, y), the inner product is calculated, and the kernel function value K between the inner product samples is used to calculate the inner product. ij = k(x i , x j ), forming a kernel matrix K. Then perform hierarchical kernel PCA, perform low-rank decomposition on K to extract the main low-frequency components, and retain the most important p low-frequency principal components based on the extraction results to reduce the dimension. Use the low-frequency principal components as new features, calculate the kernel matrix again, implement PCA again, continue to extract fewer low-frequency components, repeat the kernel PCA multiple times, retain fewer low-frequency components each time, hierarchical iteration can extract the low-frequency and global information of the features, filter out high-frequency noise, and finally retain low-dimensional principal components that are sufficient to express nonlinear features. Evaluate the effects of different kernel functions and the number of principal components, and select the optimal mapping.

[0063] Since electricity bill payment behavior is reflected in the user's long-term electricity consumption and payment patterns, rather than short-term fluctuations. Low-frequency principal components can reflect the overall trend characteristics of electricity consumption and payment patterns, while high-frequency principal components capture more fluctuations in a short period of time. Therefore, low-frequency principal components are more relevant to users' electricity bill payment habits, payment time and other behavioral characteristics. After principal component analysis, using low-frequency principal components as features can more accurately characterize the payment behavior patterns of different users. This application fully considers key features such as nonlinear patterns, steps, and periodicity in the power load curve when modeling users, applies kernel methods to map to high-dimensional feature space, and combines nonlinear kernel technology with electricity bill feature extraction to achieve hierarchical user payment behavior modeling and risk prediction.

[0064] In some optional embodiments, performing data iteration based on the low-frequency feature data to determine the user payment mode includes: performing data iteration based on the feature data through a hierarchical clustering algorithm to determine the user payment mode; wherein the user payment modes include: advance payment mode, on-time payment mode, intermittent payment mode, delayed payment mode, overdue payment mode and overdue power outage mode.

[0065] In some optional embodiments, a low-frequency principal component is selected as the main feature input for hierarchical clustering. First, cluster only the main features, and then iteratively add more features for adjustment (i.e., add some features of p low-frequency features other than the main features). According to the clustering results, the samples are divided into modes such as early payment and delayed payment. SMOTE is used to oversample the class imbalance. Random oversampling is prone to overfitting because minority class samples are copied. SMOTE is an oversampling technique for synthesizing minority classes, which creates new synthetic samples to balance the data set. Through the risk prediction performance stage, an evaluation index that integrates multi-source heterogeneous features is constructed to provide richer risk discrimination information. At the same time, the MCTS algorithm is used to globally search for the optimal parameters of the model, which is conducive to improving the prediction performance.

[0066] Among them, "clustering only with main features" means clustering with some features among p low-frequency features as main features, and "iteratively adding more features to adjust" means adding some features among p low-frequency features other than main features. That is, if the features clustered as main features are pn features, the more features added in "iteratively adding more features to adjust" are some or all of the n features. For example, when p is 20 and n is 8, the features clustered as main features are 20-8=12, and the iteratively added features are some or all of the features included in the 8 features.

[0067] Furthermore, the working principle of SMOTE (Synthetic Minority Over-sampling Technique) is to create synthetic data using the k-nearest neighbor algorithm. The creation steps using SMOTE are: determine the feature vector and its nearest neighbor, calculate the distance between two sampling points, multiply the distance by a random number between 0 and 1, and identify new points on the line segment based on the calculated distance. The distance formula is: xsynthesized = x+rand(0,1)*(xnearest-x), and repeat this process for the identified feature vector. And further evaluate the recognition effect of different kernel functions, feature selection and category settings. Repeat the above steps to find the optimal model parameters and continuously optimize the model.

[0068] In some embodiments, the evaluation index of hierarchical clustering includes:

[0069] (1) Silhouette Coefficient: It evaluates the affinity between a sample and its category. Its value range is [-1, 1]. The closer it is to 1, the better the clustering effect.

[0070] (2) Cohesion: A measure of the similarity of samples within a class. The larger the value, the higher the degree of clustering within the class.

[0071] (3) Separation: A measure of the degree of difference between samples of different classes. The larger the value, the higher the degree of separation between classes.

[0072] (4) Dunn index: evaluates the ratio of intra-class aggregation to inter-class separation. The larger the value, the better.

[0073] (5) Davies-Bouldin Index: evaluates the internal consistency of categories and the differences between categories. The smaller the value, the better the effect.

[0074] (6) Mean Silhouette Coefficient: The mean of the silhouette coefficients of all samples, reflecting the overall clustering effect.

[0075] (7) Classification Accuracy: Use category labels to evaluate clustering accuracy.

[0076] In step S106, the digital features of payment behavior are determined according to the user payment mode, including: determining the frequency of occurrence of any user payment mode within the target time period, quantifying the frequency of occurrence, and determining the digital features of payment behavior. Specifically, it can be understood as: in the first stage, the frequency statistics of different payment modes of users are obtained, and the payment model can be clustered to identify payment behavior modes such as advance payment mode, on-time payment mode, intermittent payment mode, delayed payment mode, overdue payment mode, overdue power outage mode, etc., and the payment mode frequency is quantified to construct the digital features of payment behavior. Among them, the payment mode frequency is quantified and the digital features of payment behavior are constructed through statistical analysis, such as basic statistics, time series analysis, or cluster analysis, such as K-means clustering, DBSCAN clustering, or association rule mining, such as Apriori algorithm, or machine learning algorithm, such as logistic regression, decision tree and random forest, neural network, or survival analysis, such as Cox proportional risk model, or feature engineering, such as data preprocessing, feature construction, or text analysis, such as natural language processing (NLP, Natural Language Processing), etc.

[0077] In step S108, the risk assessment of electricity bill payment behavior is performed based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user, including: determining the risk level of electricity bill payment through a pre-built electricity bill payment risk assessment model based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user. In some embodiments, the comprehensive information of the electricity user includes: information related to electricity consumption, information related to payment data, characteristic information related to electricity consumption time, information related to user portrait, information related to account status, information related to external influence, and other relevant information.

[0078] Among them, information related to electricity consumption includes: monthly electricity consumption of users, average daily electricity consumption of users, peak-to-valley ratio of users and user load rate (such as monthly electricity consumption / contract capacity); information related to payment data includes: monthly payment amount, monthly payment duration (such as the number of days from the invoice date to the payment date), the number of monthly payments, the amount of late payment fees, the ratio of accounts receivable to actual accounts collected and the number of overdue power outages; characteristic information related to electricity usage time includes: invoice date (such as the beginning or end of the month), accounts receivable period (such as the interval between the invoice date and the payment deadline) and holiday factors (such as whether the payment date is a holiday); information related to user portraits includes: user type (such as residential, commercial, industrial users), user region (such as city level) and user age; account status information includes: user account balance, number of historical arrears of users and number of power outages and restorations of users; external influence information includes: monthly electricity price adjustment and macroeconomic situation (such as GDP growth rate, etc.); other relevant information includes: user industry category, user scale (such as business expansion), electricity category, voltage level and number of months (such as reflecting the cyclical nature of electricity consumption).

[0079] In some embodiments, since the electricity bill payment data is generated every time the user handles the electricity bill payment business, and the power market development situation perception result is monthly, in order to ensure the consistency of data frequency when unifying the algorithm modeling, the number of times the electricity bill payment data is generated is counted monthly to construct the user's monthly electricity bill payment frequency factor, which is recorded as f m .

[0080] In some embodiments, for electricity users with normal production and operation and healthy capital turnover, the interval between the time when their electricity bills are issued and the time when their electricity bills are paid is generally short, indicating that their electricity bills are paid relatively timely. p The electricity bill issuance time t r The difference between the two constructs the duration of electricity bill payment for users d p , where d p =t p -t r .

[0081] In some embodiments, the user load rate is an important indicator for evaluating the user's grid load. It measures the ratio of the user's actual monthly electricity consumption E to the rated contract capacity P. The user load rate calculation formula is: load rate = E / P. Among them, E can be the user's total electricity consumption in a month, in kilowatt-hours (kWh), and P is the user's maximum load capacity stipulated in the power supply contract signed with the power grid company, in kilowatts (kW). The load rate is calculated by dividing the user's total electricity consumption (kWh) in a month by the rated capacity (kW). Taking into account the unit difference between electricity consumption and capacity, electricity consumption needs to be converted into capacity units according to the monthly ratio.

[0082] In some embodiments, the number of times the user's electricity bill payment history is calculated is recorded as C. f .

[0083] In some embodiments, the time when the user's electricity bill payment late fee was last generated is counted through the user's electricity bill payment history data, which is recorded as t l .

[0084] In some embodiments, the user peak-to-valley power consumption ratio is an indicator for evaluating the user's power load characteristics, which reflects the ratio of the peak value to the valley value of the user's power load. The user load rate calculation formula is constructed as follows: Among them, P max It is the user's peak power consumption, that is, the maximum value of the user's power load within a certain period of time (usually within a day). min The peak-to-valley ratio is the user's valley power consumption, that is, the minimum value of the user's power load at the same time. The higher the peak-to-valley ratio, the more unstable the user's power load is and the greater the volatility. The peak-to-valley ratio of a user can be used to evaluate the fluctuation of its power load and the power demand pattern. It is an important indicator for analyzing the user's power consumption characteristics and can be used in the risk assessment of electricity fee default.

[0085] In some embodiments, the specific algorithms involved in this application are described as follows:

[0086] (1) Hierarchical clustering

[0087] Hierarchical clustering is a common clustering algorithm that can divide and classify data samples at different levels without determining the number of clustering categories in advance. Similarly, this algorithm is suitable for use when there is uncertainty about the sample or lack of domain knowledge. Generally, hierarchical clustering can be divided into two specific strategies. One is: the strategy of grouping samples (small categories) from bottom to top (large categories); the other is split hierarchical clustering: dividing large categories from the top. According to the specific conditions of the research object and data, this paper adopts the first agglomerative hierarchical clustering strategy.

[0088] The specific steps of agglomerative hierarchical clustering are to treat each sample as a single cluster with a single element, then calculate the distance (dissimilarity) between the classes, merge the class with the shortest distance (i.e. the greatest similarity), and traverse the entire process, gradually merging small classes until all samples are in the same class. Given n sample points x1, x2, ..., xn, the specific process is as follows:

[0089] 1) Treat each sample point as a class and calculate the distance dist(xi, xj) between two samples;

[0090] 2) Merge the two closest classes into a new class;

[0091] 3) Update the distance between classes;

[0092] 4) Repeat steps 2) and 3) until all samples are merged into one class / the end condition is reached.

[0093] From the process of hierarchical clustering algorithm, it can be seen that the key problem of agglomerative hierarchical clustering algorithm is to establish the distance between objects (samples) and between clusters. The distance between classes is generated from the distance between samples according to different connection functions (such as single connection and full connection). Therefore, the distance between two samples plays an important role in the algorithm. When calculating the distance between two samples, traditional hierarchical clustering methods often use Euclidean distance. For samples xi and xj, the distance metric is shown in the following formula (1):

[0094] dist(x i ,x j )=||x i -x j || 2 (1)

[0095] However, the agglomerative hierarchical clustering algorithm based on Euclidean distance is greatly affected by noise points. When the distance between two classes is close, a small number of points with close distances will be preferentially synthesized into a cluster, while most samples of the two classes are not actually close, resulting in clustering errors. The agglomerative hierarchical clustering algorithm based on Euclidean distance can be regarded as learning decision boundaries using linear models. Since it can only learn very simple linear decision boundaries, the algorithm is very sensitive to noise points and cannot effectively separate the categories. For situations that cannot be separated in linear space, the data can be increased in dimension and the classification boundary can be found in high-dimensional space to avoid the influence of noise points in the original space.

[0096] This application uses the Gaussian kernel metric method to achieve the increase in dimension, which is defined as follows:

[0097]

[0098] As shown in the following formula (3), the characteristic of the Gaussian kernel function is to transform the low-dimensional space into an infinite-dimensional space, while realizing the calculation of high-dimensional dot products in low dimensions:

[0099]

[0100] If n sample points x1, x2, ..., xn are given, the agglomerative hierarchical clustering algorithm based on Gaussian kernel is as follows:

[0101] 1) Treat each sample point as a class and calculate the distance between two samples based on formula (3);

[0102] 2) Merge the two closest classes into a new class;

[0103] 3) Update the distance between classes;

[0104] 4) Repeat 2) and 3) until all classes are merged into one class / the end condition is reached.

[0105] From the Gaussian kernel agglomerative hierarchical clustering algorithm process, we can see that the algorithm modifies the distance calculation between samples into a metric based on the Gaussian kernel function, while the other steps of the original algorithm are maintained. While ensuring the simplicity of the original hierarchical clustering algorithm, the algorithm can also improve the defect of overcoming the linear inseparable situation.

[0106] (2) SVM regression

[0107] SVM regression is the application of support vectors in the field of function regression. There is only one type of sample points in SVM regression, and the optimal hyperplane sought is to minimize the "total deviation" of all sample points from the hyperplane. In this case, the sample points are all between the two boundary lines, and finding the optimal regression hyperplane is equivalent to finding the maximum interval.

[0108] Given a dataset D: x i ∈R m ,y i ∈R, the essence of the regression problem is to find a function f(x) in order to infer the y value corresponding to any pattern x.

[0109] SVM regression is achieved by defining a hard-ε band hyperplane (such as Figure 2 As shown), that is, all sample points in the data set D satisfy: -ε≤y i -f(x i )≤ε,i=1,2,…N, the regression problem is transformed into finding the optimal hard-ε band hyperplane, as shown in the following formula:

[0110]

[0111] st-ε≤y i -f(x i )≤ε,i=1,2,…,N

[0112] Solving the above optimization problem can be transformed into solving the positive point set D + and negative class point set D - The binary classification problem (such as Figure 3 As shown), where D + and D - As shown in the following formula:

[0113]

[0114] (3) MCTS (Monte Carlo Tree Search) algorithm

[0115] Monte Carlo tree search (MCTS) is a heuristic search algorithm based on tree data structure that is effective even when the search space is huge. It can solve various complex problems by constructing random numbers that conform to certain rules. Figure 4 The Monte Carlo tree search algorithm process is shown in Figure 1, which is divided into four stages: selection, expansion, simulation, and backpropagation. The circles in the figure represent the nodes of the Monte Carlo tree.

[0116] In the selection phase of the Monte Carlo tree search algorithm, starting from the root node (i.e., the top black circle), decisions are made by comparing the values ​​of the candidate nodes, and the child node with the greatest value is selected. There are three types of candidate nodes in this phase, namely, fully expanded nodes, incompletely expanded nodes, and leaf nodes. In the expansion phase of the Monte Carlo tree search algorithm, according to the UCT value of its child nodes, a new node (i.e., the bottom white circle) will be created in the decision tree as the new child node of the node after the current node is expanded. In the simulation phase of the Monte Carlo tree search algorithm, the basis for the current node movement selection is a random strategy: that is, a child node is randomly selected from the candidate child nodes and expanded until the last layer of nodes. In the back propagation phase of the Monte Carlo tree search algorithm, Q(v) propagates in the reverse direction along the decision tree to the root node, and the node state that starts the Monte Carlo simulation is changed to visited.

[0117] (4) Logistic regression

[0118] The logistic regression model is a commonly used machine learning method in the industry. It is used to estimate the probability of something happening. For example: the probability of a user buying a certain product; the probability of a patient suffering from a certain disease. It is derived from multivariate linear regression and is a multivariate analysis method that studies the relationship between the target variable Y and a series of influencing factors X1, X2, ..., Xm. The target variable Y of logistic regression is a binary variable, which indicates the occurrence or non-occurrence of an event, usually coded as 1 and 0. The probability of an event occurring is calculated using the logistic regression model. Where: e is a natural number, approximately equal to 2.71828; β0, β1, ..., β m is the model parameter, β0 is also called the intercept.

[0119] It can be seen from the above that the electricity bill payment risk assessment method provided by this application includes: determining the user's historical payment behavior sequence and user's historical electricity usage characteristics of electricity users in the target area; determining the user payment mode according to the user's historical payment behavior sequence and the user's historical electricity usage characteristics of any electricity user; determining the digital characteristics of the payment behavior according to the user's payment mode; and conducting the risk assessment of the electricity bill payment behavior according to the digital characteristics of the payment behavior and the comprehensive information of the electricity user. This application determines the user's payment mode according to the user's historical payment behavior and the user's historical electricity usage characteristics. By analyzing the user's payment mode, it can intuitively reflect the cause of the risk, improve the interpretability of the model, and integrate the comprehensive information of the electricity user, that is, the user's behavior data, user attributes and other multi-source information to conduct the risk assessment of the electricity bill payment behavior, avoiding the problem of high subjectivity and low efficiency of relying solely on subjective rules. Improving the effect of risk identification and assessment, identifying behavior patterns and predicting risks in stages, and designing the overall process reasonably will help power companies improve operational efficiency, reduce operational risks, and improve the refinement, intelligence, and digitalization of power services and management.

[0120] It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the described method.

[0121] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0122] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides an electricity bill payment risk assessment device.

[0123] refer to Figure 5 , the electricity fee payment risk assessment device comprises:

[0124] The data determination module 502 is configured to determine the user's historical payment behavior sequence and user's historical electricity usage characteristics of the electricity users in the target area;

[0125] The payment mode determination module 504 is configured to determine a user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user;

[0126] A feature determination module 506 is configured to determine a digital feature of a payment behavior according to the user payment mode;

[0127] The payment risk assessment module 508 is configured to perform risk assessment on the electricity bill payment behavior based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user.

[0128] For the convenience of description, the above device is described in terms of functions divided into various modules. Of course, when implementing the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0129] The device of the above embodiment is used to implement the corresponding electricity bill payment risk assessment method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0130] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the electricity bill payment risk assessment method described in any of the above embodiments is implemented.

[0131] Figure 6 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0132] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0133] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0134] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0135] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0136] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).

[0137] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0138] The electronic device of the above embodiment is used to implement the corresponding electricity bill payment risk assessment method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0139] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the electricity bill payment risk assessment method as described in any of the above embodiments.

[0140] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0141] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the electricity bill payment risk assessment method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0142] Based on the same inventive concept, corresponding to the electricity bill payment risk assessment method described in any of the above embodiments, the present disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer so that the computer and / or the processor execute the electricity bill payment risk assessment method. Corresponding to the execution subject corresponding to each step in each embodiment of the electricity bill payment risk assessment method, the processor that executes the corresponding step may belong to the corresponding execution subject.

[0143] The computer program product of the above embodiment is used to enable the computer and / or the processor to execute the electricity bill payment risk assessment method as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0144] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0145] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power supply / ground connection with the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform to be implemented in the embodiments of the present application (that is, these details should be fully within the scope of understanding of those skilled in the art). In the case of elaborating specific details (e.g., circuits) to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered to be illustrative rather than restrictive.

[0146] Although the present application has been described in conjunction with specific embodiments of the present application, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0147] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.

Claims

1. A method for assessing risk of electricity bill payment, characterized in that: include: Determine the historical payment behavior sequence and historical electricity consumption characteristics of electricity users in the target area; Determining a user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user; Determining payment behavior digital features based on the user payment mode; Based on the digital characteristics of the payment behavior and the comprehensive information of electricity users, a risk assessment of the electricity bill payment behavior is conducted.

2. The method according to claim 1, characterized in that The determining of the user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user includes: Performing data mapping on the user's historical payment behavior sequence and the user's historical electricity usage characteristics through a kernel function to determine a high-dimensional user's historical payment behavior sequence and a high-dimensional user's historical electricity usage characteristics; Determine the kernel function value of the high-dimensional user's historical payment behavior sequence and the high-dimensional user's historical electricity consumption characteristics, and determine the kernel matrix according to the kernel function value; Low-frequency feature data is extracted from the kernel matrix, and data iteration is performed according to the low-frequency feature data to determine the user payment mode.

3. The method according to claim 2, characterized in that The extracting low-frequency feature data from the kernel matrix includes: Performing low-rank decomposition on the kernel matrix, determining low-frequency feature data, and calculating the eigenvalue of any low-frequency feature data; The low-frequency feature data are sorted according to the feature values, and feature data are extracted according to the sorting matrix.

4. The method according to claim 3, characterized in that The performing data iteration according to the low-frequency feature data to determine the user payment mode includes: According to the characteristic data, data iteration is performed through a hierarchical clustering algorithm to determine the user payment mode; wherein the user payment mode includes: advance payment mode, on-time payment mode, intermittent payment mode, delayed payment mode, overdue payment mode and overdue power outage mode.

5. The method according to claim 4, characterized in that Determining the digital characteristics of the payment behavior according to the user payment mode includes: Determine the frequency of occurrence of any of the user payment patterns within a target time period, quantify the frequency of occurrence, and determine the digital characteristics of the payment behavior.

6. The method according to claim 1, characterized in that The risk assessment of electricity payment behavior is performed based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user, including: The electricity bill payment risk level is determined based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user through a pre-constructed electricity bill payment risk assessment model.

7. An electricity bill payment risk assessment device, characterized in that: include: A data determination module is configured to determine a user's historical payment behavior sequence and a user's historical electricity usage characteristics of electricity users in a target area; A payment mode determination module, configured to determine a user payment mode according to the user historical payment behavior sequence and the user historical electricity usage characteristics of any electricity user; a feature determination module, configured to determine a digital feature of a payment behavior according to the user's payment mode; The payment risk assessment module is configured to perform risk assessment on the electricity bill payment behavior based on the digital characteristics of the payment behavior and the comprehensive information of the electricity user.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

9. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method according to any one of claims 1 to 6.

10. A computer program product, comprising computer program instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 6.

Images