Power customer demand prediction method and system based on data mining

Abstract

The invention relates to the field of power customer demand prediction based on data mining, in particular to a power customer demand prediction method and system based on data mining, and the method comprises the steps: collecting multi-source data of a target customer group; carrying out power consumption behavior clustering on the clients based on pattern mining, and generating user behavior tags; extracting a multi-dimensional feature vector of the customer based on the behavior tag; a prediction model is constructed according to the multi-dimensional feature vectors, and accurate prediction and optimal management of power customer demands are realized by combining a data mining technology and power demand prediction. Firstly, a multi-source data acquisition technology is adopted, rich input data is provided for a prediction model, and power consumption behaviors of various clients can be reflected more accurately. And secondly, through mode mining and power consumption behavior clustering, the clients are divided into different power consumption groups, personalized features of the clients are generated based on behavior tags, accurate input feature vectors are provided for a subsequent prediction model, and the prediction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of power customer demand prediction based on data mining, and in particular to a power customer demand prediction method and system based on data mining. Background Art

[0002] With the continuous development of the power industry, traditional power demand forecasting methods are no longer able to meet the precise forecasting and flexible regulation requirements of modern power systems. In the past, power demand forecasting primarily relied on historical electricity consumption data and simple statistical models, such as regression analysis and time series analysis. However, these methods suffer from insufficient accuracy, fail to reflect individual customer differences, and fail to fully consider external factors. Therefore, the use of data mining and machine learning techniques for power demand forecasting has become crucial for improving forecast accuracy and optimizing resource allocation.

[0003] Data mining technology can analyze large amounts of multi-source data to reveal underlying patterns in electricity consumption, thereby providing power companies with more accurate demand forecasts. In this process, customer usage patterns, weather data, holidays, and other external factors can all serve as important data sources. To improve the accuracy and adaptability of forecasting models, this multi-source data must be effectively collected, preprocessed, and integrated to construct a unified time series matrix, providing high-quality data support for subsequent analysis.

[0004] Clustering electricity usage behavior is a crucial step in data mining. Clustering algorithms can group customers with similar electricity usage patterns, enabling customized prediction models for different customer groups. Based on these clustering results, user behavior labels (such as "peak electricity user" or "energy-saving") are generated, forming the basis for subsequent feature extraction. These labels reflect typical customer electricity usage behavior and provide more representative input features for prediction models.

[0005] Furthermore, based on these behavioral labels and other numerical features (such as historical electricity consumption and maximum load), multi-dimensional feature vectors can be extracted for each customer. These feature vectors comprehensively describe the customer's electricity demand and are suitable for prediction using machine learning algorithms. By building appropriate prediction models (such as regression models, support vector machines, and neural networks), power companies can accurately predict customers' future electricity loads and flexibly adjust power supply and design differentiated electricity pricing plans based on the predictions, thereby optimizing the allocation and efficiency of power resources.

[0006] Therefore, the electricity customer demand forecasting method based on data mining can provide power companies with accurate electricity load forecasts and personalized electricity services by deeply mining customer behavior data, and promote the intelligent and green development of the power system. Summary of the Invention

[0007] In view of the above problems in the prior art, the present invention is proposed.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: a method for predicting electricity customer demand based on data mining, comprising: collecting multi-source data of a group of target customers;

[0009] Cluster customers' electricity usage behaviors based on pattern mining and generate user behavior labels;

[0010] Extract customer multi-dimensional feature vectors based on behavioral tags;

[0011] Construct a prediction model based on multi-dimensional feature vectors.

[0012] As a preferred solution of the method for predicting electricity customer demand based on data mining described in the present invention, wherein: the multi-source data collected from the target customer group includes:

[0013] Collecting initial data through a collection device;

[0014] Perform data preprocessing on the initial data and establish a unified data time series matrix.

[0015] As a preferred solution of the method for predicting electricity customer demand based on data mining according to the present invention, wherein: the clustering of customers' electricity consumption behaviors based on pattern mining includes:

[0016] Acquire multi-source data and construct a transaction set of customer electricity usage behavior;

[0017] Based on the customer's electricity consumption behavior transaction set, a first-level algorithm is used to perform pattern mining to obtain frequent electricity consumption behaviors;

[0018] The electricity usage behavior of customer groups is matched with their frequent electricity usage behavior to cluster electricity usage behavior.

[0019] As a preferred solution of the method for predicting power customer demand based on data mining described in the present invention, the generating of user behavior tags includes:

[0020] Analyze the clustered customer groups and identify common behavioral characteristics within the clusters;

[0021] Generate user behavior tags based on common behavioral characteristics.

[0022] As a preferred solution of the method for predicting power customer demand based on data mining described in the present invention, wherein: the extraction of customer multi-dimensional feature vectors based on behavior tags includes:

[0023] Use a secondary algorithm to convert customer behavior tags into numerical features;

[0024] Combine the customer's label features and numerical features to form a multi-dimensional feature vector.

[0025] As a preferred solution of the data mining-based power customer demand forecasting method of the present invention, wherein: said constructing a forecasting model based on a multi-dimensional feature vector includes converting the metadata data into a form recognizable by a machine learning model through feature engineering;

[0026] Select and train models based on target tasks;

[0027] After the model training is completed, the model is evaluated and put into practical application.

[0028] As a preferred solution of the data mining-based power customer demand forecasting method of the present invention, the practical application includes:

[0029] Based on the customer's historical electricity consumption data, predict future electricity load and adjust power supply in different time periods;

[0030] Based on the customer's forecast results, customers are divided into different categories to optimize the allocation of power resources.

[0031] Design differentiated electricity price plans based on customers' electricity usage behavior.

[0032] As a preferred solution of the method for predicting electricity customer demand based on data mining described in the present invention, wherein: a system for predicting electricity customer demand based on data mining includes: an acquisition module, an analysis module, an extraction module, and a prediction module;

[0033] The acquisition module collects multi-source data of target customer groups;

[0034] The analysis module clusters customers' electricity usage behaviors based on association rule mining and generates user behavior labels;

[0035] Extraction module, extracts customer multi-dimensional feature vectors based on behavioral tags;

[0036] The prediction module builds a prediction model based on the multi-dimensional feature vector.

[0037] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, it implements the steps of the above-mentioned method for predicting electricity customer demand based on data mining.

[0038] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the above-mentioned method for predicting electricity customer demand based on data mining are implemented.

[0039] Beneficial effects of the present invention: By combining data mining technology with electricity demand forecasting, the present invention achieves accurate forecasting and optimized management of electricity customer demand. First, multi-source data acquisition technology is used, including historical electricity consumption data, weather data, and basic customer attributes, to provide rich input data for the prediction model, which can more accurately reflect the electricity consumption behavior of various customers. Second, through pattern mining and electricity consumption behavior clustering, customers are divided into different electricity consumption groups, and personalized features of customers are generated based on behavioral labels, providing accurate input feature vectors for subsequent prediction models, thereby improving the accuracy of predictions.

[0040] Through machine learning model training, power supply can be intelligently adjusted according to different time periods and demand fluctuations, avoiding waste or shortages of power resources. Furthermore, based on customer electricity usage behavior and forecast results, differentiated electricity pricing strategies can be designed to encourage users to use electricity during off-peak hours, balance power loads, and optimize power system efficiency.

[0041] The present invention can provide power companies with personalized and accurate power demand forecasting and resource allocation solutions, improve the intelligence level of the power system, reduce operating costs, and promote the efficient use and sustainable development of power resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 A flowchart of a method for predicting electricity customer demand based on data mining is provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0044] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0045] Example 1, with reference to Figure 1 , is an embodiment of the present invention, which provides a method for predicting power customer demand based on data mining, comprising:

[0046] S1: Collect multi-source data of target customer groups;

[0047] S2: Cluster customers’ electricity usage behaviors based on pattern mining and generate user behavior labels;

[0048] S3: Extract customer multi-dimensional feature vectors based on behavioral tags;

[0049] S4: Build a prediction model based on the multi-dimensional feature vector.

[0050] It should be noted that traditional forecasting methods ignore individual differences among customers, resulting in low forecasting accuracy. This present invention clusters customer electricity usage behavior and generates personalized behavioral labels, enabling the forecasting model to more accurately reflect the electricity demand of different customer groups, thereby improving forecasting accuracy. Traditional methods fail to fully consider the impact of external factors such as weather and holidays on electricity demand. This present invention integrates multi-source data, such as weather and basic customer information, to ensure that these factors are fully considered in the forecasting model, making the forecast results more realistic. Traditional methods rely on historical data for forecasting and are unable to cope with rapidly changing electricity demand. The forecasting model constructed by this present invention, based on multi-dimensional feature vectors, can accurately predict electricity load in different time periods, helping power companies to precisely allocate electricity resources and avoid waste and supply shortages. Traditional electricity pricing models are relatively simple and cannot be flexibly adjusted to meet customer needs. This present invention analyzes customer electricity usage behavior and designs differentiated electricity pricing strategies to encourage users to use electricity during off-peak hours, optimize electricity load, and improve power system operating efficiency.

[0051] Therefore, by collecting historical customer electricity usage data, weather data, and basic customer information, we can fully understand the various factors affecting electricity demand and ensure the comprehensiveness and timeliness of the data. Using pattern mining technology, we conduct cluster analysis on customer electricity usage behavior, identify groups with similar electricity usage characteristics, and generate personalized behavior labels for each group. This process helps to develop more accurate forecasting strategies for different customer groups. Based on the behavioral labels, we extract the customer's multi-dimensional feature vector, including multiple indicators such as electricity usage frequency and load fluctuation, to ensure that the feature vector can fully reflect the customer's electricity usage habits and provide accurate input for subsequent forecasting models. Based on the multi-dimensional feature vector, a machine learning forecasting model is constructed, and appropriate algorithms are selected for training and optimization. Through this model, power companies can accurately predict customer electricity demand, help optimize the allocation and scheduling of power resources, and design differentiated electricity pricing strategies.

[0052] Example 2, reference Figure 1 , which is an embodiment of the present invention, provides a method for predicting electricity customer demand based on data mining based on the above embodiment.

[0053] In the embodiment of the present application, collecting multi-source data of the target customer group in step S1 includes the following steps A1-A2:

[0054] A1: Collect initial data through the acquisition device;

[0055] A2: Preprocess the initial data and establish a unified data time series matrix.

[0056] In this embodiment, the collection device refers to a hardware and software system used to acquire multi-source initial data from electricity customers. It primarily includes components such as smart meters, environmental monitoring sensors, user terminals, a data collection gateway, and a customer information system interface. Its purpose is to collect raw data from various sources, including historical electricity usage data, weather information, and basic customer attributes, in real time or periodically, to provide a comprehensive and accurate data foundation for subsequent analysis.

[0057] The data acquisition device exchanges data with the master station platform via built-in data communication modules (such as RS485, NB-IoT, and 4G), and has preliminary processing capabilities such as protocol parsing, data caching, time synchronization, and identifier completion. For heterogeneous data output by different devices, the data acquisition device first converts to a standard protocol and encapsulates it in a unified format.

[0058] After completing the initial collection, the system enters the data preprocessing stage, which specifically includes: filling missing values ​​in the collected data, removing or correcting outliers. Unifying the data time granularity (such as 15 minutes / hour) and performing time series completion. Mapping various types of raw data into data fields with a unified structure. Attaching index information such as customer number, equipment type, and geographic location to each piece of data. Based on a unified time axis, integrate multi-source data to generate a standardized time series data matrix. This preprocessing process ensures the consistency, integrity, and analyzability of multi-source data, providing a solid data foundation for subsequent electricity consumption behavior clustering and predictive modeling.

[0059] In an optional embodiment, the multi-source data of the target customer group collected in step S1 can also be indirectly collected by integrating an external data platform or an enterprise information system interface to collect historical data, environmental data and business attribute data of the target customer group, which is mainly suitable for large-scale or cross-regional customer data acquisition.

[0060] In another optional fact method, the multi-source data of the target customer group collected in step S1 is used to obtain more detailed personalized data through user-side mobile devices, interactive questionnaires or APP authorized uploads, thereby enhancing customer behavior understanding and label modeling.

[0061] In the embodiment of the present application, step S2 clusters the customer's electricity usage behavior based on pattern mining to generate user behavior labels, including the following steps B1-B5:

[0062] B1: Obtain multi-source data and construct a customer electricity usage transaction set;

[0063] B2: Based on the customer's electricity usage transaction set, a first-level algorithm is used to perform pattern mining to obtain frequent electricity usage behaviors;

[0064] B3: Match the customer group's historical electricity usage behavior with their frequent electricity usage behavior to cluster electricity usage behavior.

[0065] B4: Analyze the clustered customer groups and identify common behavioral characteristics within the clusters;

[0066] B5: Generate user behavior tags based on common behavior characteristics.

[0067] The specific expression of obtaining multi-source data and constructing the customer electricity consumption behavior transaction set in step B1 is:

[0068] Consider each customer's electricity usage behavior for a period of time (such as one day or one week) as a "transaction";

[0069] Discretize behavioral features (such as "power consumption greater than the threshold between 7 and 9 o'clock" and "peak on weekend nights") into several "power consumption label items";

[0070] Example: Customer A's behavior record for the day → {high consumption during peak hours, peak hours on weekends, and prominent electricity consumption at night}.

[0071] In an embodiment of the present application, the primary algorithm in step B2 adopts the FP-Growth frequent pattern mining algorithm, which has the advantages of efficiently constructing a frequent item set tree and avoiding the problem of explosion of candidate item combinations. It is suitable for processing large-scale customer electricity consumption behavior transaction data, especially under high concurrency and massive data, and can maintain high mining efficiency and stability.

[0072] In an optional embodiment, the primary algorithm in step B2 may also adopt a PrefixSpan sequence pattern mining algorithm.

[0073] In another optional embodiment, the primary algorithm in step B2 may also adopt the Apriori association rule mining algorithm.

[0074] In the embodiment of the present application, extracting the customer multi-dimensional feature vector based on the behavior tag in step S3 includes the following steps: C1-C2:

[0075] C1: Use a secondary algorithm to convert customer behavior tags into numerical features;

[0076] C2: Combine the customer's label features and numerical features to form a multi-dimensional feature vector.

[0077] In this embodiment of the present application, the secondary algorithm in step C1 uses a one-hot encoding algorithm. One-hot encoding converts each customer's behavior label (such as "peak high consumption" or "weekend peak", etc.) into a binary feature to generate a sparse matrix. For example, "peak high consumption" may be encoded as [1,0,0], indicating that the customer belongs to this behavior label during a specific time period.

[0078] In an optional embodiment, the secondary algorithm may use an embedded learning algorithm (e.g., Word2Vec or GloVe). Such algorithms can perform vectorization based on the semantic relevance of labels, thereby representing each label as a continuous vector and capturing the semantic similarity between labels. It is suitable for scenarios where there is a certain semantic association between labels, especially when the labels are complex or multi-dimensional. For example, "peak consumption" and "weekend peak" may show similarities in the behavior of some customers, and the embedded algorithm can create a more expressive feature representation by learning the similarities between labels.

[0079] In another optional embodiment, the secondary algorithm can use a frequency mapping algorithm, which converts the frequency of occurrence of labels into numerical features. For example, if a customer label (such as "weekend peak") appears frequently in the data set, a higher numerical value is assigned to the label. Labels with lower frequencies are assigned lower numerical values. This is suitable for situations where the frequency of labels is more obvious. In this way, labels can be converted into numerical features with certain weights, further improving the accuracy of the prediction model.

[0080] It should be noted that by efficiently mining frequent patterns in customer electricity usage, we can accurately identify representative patterns, such as high electricity consumption during peak hours and peak weekend usage. This not only improves the accuracy of electricity usage clustering but also provides a reliable data foundation for subsequent personalized forecasts, significantly improving the accuracy of electricity demand forecasts.

[0081] Furthermore, customer behavior labels are converted into numerical features, making discrete electricity usage data compatible with machine learning models. One-hot encoding allows labels to be easily and efficiently converted into numerical features, facilitating subsequent analysis and modeling. This conversion enhances the expressiveness of features, providing a solid foundation for accurate predictions and the design of differentiated electricity pricing strategies.

[0082] In the embodiment of the present application, constructing a prediction model based on the multi-dimensional feature vector in step S4 includes the following steps D1-D3:

[0083] D1: Convert metadata into a form recognizable by machine learning models through feature engineering;

[0084] The raw data and the multi-dimensional feature vectors obtained by transforming the behavior labels will be processed by feature engineering. The purpose of feature engineering is to convert the raw data into a format that can be effectively processed by machine learning algorithms. This includes:

[0085] Normalize numerical features to a uniform dimension to prevent certain features from having a significant impact on model training. Select features that have a significant impact on prediction results from a large number of features, removing redundant or irrelevant features to improve model efficiency. Enhance nonlinear relationships between features and enrich data representation through methods such as cross-features or polynomial features.

[0086] D2: Select and train models based on target tasks;

[0087] Select an appropriate machine learning model based on the nature of the prediction task (such as regression task, classification task). For example:

[0088] Regression models: such as linear regression and support vector regression (SVR), are used to predict specific power load values.

[0089] Classification models: such as decision trees, random forests, and XGBoost, are used to classify customers into different electricity load categories.

[0090] During the training process, a multi-dimensional feature vector is used as input to train the model to fit the target variable (such as electricity load). Model hyperparameters are optimized through cross-validation and other techniques to avoid overfitting or underfitting and ensure the model's ability to generalize to new data.

[0091] D3: After model training is completed, the model is evaluated and put into practical application.

[0092] In the embodiments of the present application, the practical application includes:

[0093] Furthermore, based on the model's predictions, power companies can accurately forecast electricity load for various time periods. By analyzing historical electricity usage data and multi-dimensional feature vectors for each customer group, the model can predict customer electricity demand over the next period of time. Based on these predictions, power companies can proactively allocate power resources to avoid overloads or undersupply, thereby improving grid stability and operational efficiency. For example, if a significant increase in electricity load in a particular area is predicted within a specific time period, the power company can increase power supply to that area in advance, mitigating power outages caused by sudden demand.

[0094] Furthermore, based on power load forecasts, the model can classify customers into different categories, such as high-load users, low-load users, and users during peak and off-peak periods. Based on these classifications, power companies can design differentiated electricity pricing plans for different customer groups. For example:

[0095] High-load users: Provide higher electricity prices during peak hours to encourage them to reduce their electricity load.

[0096] Low-load users: Provide preferential electricity prices during off-peak hours to encourage them to increase electricity consumption during off-peak hours.

[0097] This can guide users to use electricity during reasonable periods through price incentives, thereby balancing the load and optimizing the use of electricity resources.

[0098] Furthermore, by updating customer electricity usage data and behavior tags in real time, the model can perform dynamic load forecasting. This real-time forecast data can provide power companies with more flexible load control solutions. Based on accurate load forecasting, power companies can monitor the operating status of the power grid in real time and implement intelligent control to ensure stable and efficient operation.

[0099] Furthermore, power companies can conduct targeted marketing based on predicted customer electricity usage. For example, they can offer discounts on energy-saving products or equipment to high-power users, or promote customized energy-saving plans based on their electricity usage habits, thereby improving customer satisfaction. Furthermore, by accurately predicting the electricity usage patterns of different customer groups, power companies can provide more personalized electricity services to each customer.

[0100] Example 2 is an embodiment of the present invention, which provides a power customer demand prediction system based on data mining, including: a collection module, an analysis module, an extraction module, and a prediction module;

[0101] The acquisition module collects multi-source data of target customer groups;

[0102] The analysis module clusters customers' electricity usage behaviors based on association rule mining and generates user behavior labels;

[0103] Extraction module, extracts customer multi-dimensional feature vectors based on behavioral tags;

[0104] The prediction module builds a prediction model based on the multi-dimensional feature vector.

[0105] This embodiment further provides a computing device applicable to a method for predicting power customer demand based on data mining, including:

[0106] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement a method for predicting electricity customer demand based on data mining as proposed in the above embodiment.

[0107] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the method for predicting electricity customer demand based on data mining as proposed in the above embodiment is implemented.

[0108] The storage medium proposed in this embodiment and the method for predicting electricity customer demand based on data mining proposed in the above embodiment belong to the same inventive concept. Technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0110] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0111] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for predicting electricity customer demand based on data mining, characterized by: include: Collect multi-source data of target customer groups; Cluster customers' electricity usage behaviors based on pattern mining and generate user behavior labels; Extract customer multi-dimensional feature vectors based on behavioral tags; Construct a prediction model based on multi-dimensional feature vectors.

2. The method for predicting electricity customer demand based on data mining according to claim 1, characterized in that: The multi-source data collected from the target customer group includes: Collecting initial data through a collection device; Perform data preprocessing on the initial data and establish a unified data time series matrix.

3. The method for predicting electricity customer demand based on data mining according to claim 2, characterized in that: The clustering of customers' electricity consumption behaviors based on pattern mining includes: Acquire multi-source data and construct a transaction set of customer electricity usage behavior; Based on the customer's electricity consumption behavior transaction set, a first-level algorithm is used to perform pattern mining to obtain frequent electricity consumption behaviors; The electricity usage behavior of customer groups is matched with their frequent electricity usage behavior to cluster electricity usage behavior.

4. The method for predicting electricity customer demand based on data mining according to claim 3, characterized in that: Generating user behavior tags includes: Analyze the clustered customer groups and identify common behavioral characteristics within the clusters; Generate user behavior tags based on common behavioral characteristics.

5. The method for predicting electricity customer demand based on data mining according to claim 4, characterized in that: The method of extracting customer multi-dimensional feature vectors based on behavior tags includes: Use a secondary algorithm to convert customer behavior tags into numerical features; Combine the customer's label features and numerical features to form a multi-dimensional feature vector.

6. The method for predicting electricity customer demand based on data mining according to claim 5, characterized in that: The constructing of a prediction model based on the multi-dimensional feature vector includes converting the metadata data into a form recognizable by the machine learning model through feature engineering; Select and train models based on target tasks; After the model training is completed, the model is evaluated and put into practical application.

7. The method for predicting electricity customer demand based on data mining according to claim 6, characterized in that: The practical application includes: Based on the customer's historical electricity consumption data, predict future electricity load and adjust power supply in different time periods; Based on the customer's forecast results, customers are divided into different categories to optimize the allocation of power resources. Design differentiated electricity price plans based on customers' electricity usage behavior.

8. A system based on the data mining-based electricity customer demand forecasting method according to any one of claims 1 to 7, characterized in that: include: Acquisition module, analysis module, extraction module, prediction module; The acquisition module collects multi-source data of target customer groups; The analysis module clusters customers' electricity usage behaviors based on association rule mining and generates user behavior labels; Extraction module, extracts customer multi-dimensional feature vectors based on behavioral tags; The prediction module builds a prediction model based on the multi-dimensional feature vector.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the processor implements the steps of a method for predicting electricity customer demand based on data mining according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for predicting electricity customer demand based on data mining according to any one of claims 1 to 7 are implemented.

Images