Power customer demand management method, system and equipment based on data warehouse and medium

Abstract

The invention discloses a power customer demand management method, system, equipment and medium based on a data warehouse, and belongs to the technical field of power customer demand management, and the method comprises the steps: collecting multi-source data in real time, carrying out the multi-level verification, carrying out the data cleaning and standardization, integrating multi-dimensional data through a snowflake-shaped data model, and carrying out the management of the multi-dimensional data. And data extraction, conversion and loading are carried out through an ETL process, customer groups are divided by using a clustering algorithm, a regulation and control strategy is generated in combination with a dynamic threshold and a hierarchical competing rule, and parameters are optimized through closed-loop feedback. According to the method, the problem that a traditional method is slow in response in a complex scene is solved, and autonomous evolution of system parameters is achieved by fusing a closed-loop optimization system of incremental learning and attenuation coefficients.

Description

Technical Field

[0001] The present invention relates to the technical field of power customer demand management, and in particular to a data warehouse-based power customer demand management method, system, equipment and medium. Background Art

[0002] With the advancement of smart grid construction and the acceleration of the power industry's digital transformation, electricity customer demand management technology has undergone a significant transformation, from traditional experience-driven to data-driven. Early electricity demand management relied primarily on a single data source and static rule base, employing an offline batch processing model for load forecasting and control decisions. This approach struggled to meet the challenges of the real-time nature of the electricity market and the diversification of customer demands. In recent years, the introduction of data warehouse technology has provided new approaches for the integrated management of multi-source, heterogeneous data, with the star and snowflake models finding initial application in power data modeling. Simultaneously, machine learning algorithms such as K-means clustering and ARIMA / LSTM forecasting models are gradually replacing traditional statistical methods, improving the accuracy of customer behavior analysis. Existing technologies are capable of identifying load patterns based on historical data and triggering standardized control strategies using fixed threshold rules. Some advanced systems have begun to incorporate external environmental variables (such as meteorological and economic indicators) as auxiliary inputs to forecasting models, but these still face significant limitations in real-time data integration and dynamic decision-making.

[0003] Current electricity customer demand management technology has three key flaws: First, at the data governance level, traditional methods do not adequately verify the timeliness of multi-source data and lack a conflict resolution mechanism across data sources, resulting in data quality affecting the reliability of downstream analysis. Second, in terms of model construction, existing clustering algorithms mostly use static parameters, making it difficult to adapt to changes in customer electricity consumption behavior, and lack collaborative optimization between prediction models and control rules. Most importantly, in terms of decision-making mechanisms, mainstream systems rely on fixed thresholds and a single rule chain, which cannot cope with the complex competition and cooperation scenarios in power grid operation. Parameter adjustments rely on manual experience, and the closed-loop feedback cycle is long. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to effectively extract, analyze and utilize the electricity consumption behavior characteristics of electricity customers to achieve accurate customer demand management and optimize power resource allocation.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for managing power customer demand based on a data warehouse, which comprises the following steps:

[0007] Collect multi-source data in real time, perform multi-level verification, and clean and standardize data. Use the snowflake data model to integrate multi-dimensional data, and extract, transform, and load data through the ETL process. Use clustering algorithms to divide customer groups, combine dynamic thresholds with hierarchical competition rules to generate control strategies, and optimize parameters through closed-loop feedback.

[0008] As a preferred solution of the data warehouse-based power customer demand management method of the present invention, wherein: the multi-source data collection includes collecting real-time power data, historical load data, basic customer information and external environment data;

[0009] The multi-level verification includes a first-level verification and a second-level verification. The first-level verification includes automatically triggering a cross-data source cross-comparison for similar data with a timestamp deviation of more than 5 minutes, and selecting the majority of consistent data as the benchmark value through a voting mechanism. The second-level verification includes using an interpolation algorithm to fill in missing values ​​based on historical data distribution characteristics and contextual relevance when data integrity is less than 80%, and marking interpolation marks.

[0010] As a preferred solution of the data warehouse-based power customer demand management method described in the present invention, the data cleaning includes marking single-point anomalies based on a preset fluctuation threshold, analyzing the trend deviation of consecutive data points through a sliding window, and marking a systematic anomaly when three consecutive points deviate from the preset pattern;

[0011] The standardization includes performing Z-score standardization on the dimension difference data to generate a unified input data set.

[0012] As a preferred embodiment of the data warehouse-based electricity customer demand management method described in the present invention, the snowflake data model includes automatically parsing customer electricity usage behavior labels, geographic area characteristics, and time context based on cleaned and standardized data to construct a customer-centric snowflake model. When new external data, such as economic indicators and weather events, is added, the dimension table is dynamically expanded and a mapping relationship with the load fact table is established through semantic analysis and association rule mining.

[0013] The snowflake data model also includes designing priority weighting rules to address conflicts in different data sources for the same customer information. Data source weights are dynamically assigned based on authority, and a final value is generated through a weighted voting algorithm. The conflicting data sources and weight assignment results are recorded.

[0014] The weighted voting algorithm is expressed as:

[0015]

[0016] Among them, V final is the final value of the weighted data, wi is the weight of the i-th data point, V i is the actual value of the i-th data point, n is the total number of data points, and i is the variable index.

[0017] As a preferred solution of the data warehouse-based power customer demand management method of the present invention, wherein: the ETL process includes dynamic priority data extraction, data conversion and intelligent loading;

[0018] Dynamic priority data extraction includes dynamically adjusting the data capture order based on the data source health score and business urgency. Real-time data is aggregated through streaming windows and stored in a cache queue. Batch data is extracted in blocks according to time slices. The formula is expressed as:

[0019] P fetch =α·H source +β·U business

[0020] Among them, α and β are weight coefficients, U business is the business urgency level, P fetch The order of data fetching;

[0021] Data conversion includes automatically matching the adaptation strategies in the conversion rule library according to semantic labels for related dimensional data, such as weather and economic indicators; intelligent loading includes loading load peak records into the in-memory database based on the query hotspot analysis results, and compressing the remaining data and storing it in distributed storage.

[0022] As a preferred solution of the data warehouse-based power customer demand management method described in the present invention, the customer group division includes extracting customer electricity consumption behavior characteristics, load fluctuation rate and external correlation characteristics, constructing feature vectors, and initializing cluster centers using an improved K-means++ algorithm;

[0023] The improved K-means++ algorithm is expressed as:

[0024]

[0025] Among them, K is the number of cluster centers, C j is the jth cluster, y is the data point, ε j is the jth cluster center, γ is the entropy penalty coefficient, j is the variable index, Entropy(C j ) is the entropy of the jth cluster; when the internal standard deviation of the cluster is greater than 15% or an external event is triggered, the clusters are automatically re-clustered and the label library is updated;

[0026] The generation and control strategy includes generating a benchmark threshold for each time period based on the moving average and fluctuation range of historical load data, superimposing the influence coefficient of external variables, and generating a scenario-based threshold. When the overall load forecast value of the system exceeds the dynamic threshold, the maximum range peak shaving strategy is activated. When the real-time data is abnormal, precise control is triggered. When multiple rules are triggered at the same time, a comprehensive strategy is generated based on priority weights.

[0027] As a preferred embodiment of the data warehouse-based power customer demand management method of the present invention, the closed-loop feedback optimization parameters include collecting actual load drop rate, control response speed, equipment loss change, complaint rate, satisfaction survey, model prediction error, rule triggering frequency, and data delay rate;

[0028] When there is a prediction error, incremental training is automatically triggered, the historical version is retained and rolled back, and based on the policy execution effect, the triggering threshold is relaxed according to the time attenuation coefficient.

[0029] Another object of the present invention is to provide an electricity customer demand management system based on a data warehouse.

[0030] To solve the above technical problems, the present invention provides the following technical solutions: a data warehouse-based electricity customer demand management system, comprising: a data acquisition and preprocessing module, a data integration and storage module, a customer group segmentation and control strategy generation module, and a closed-loop feedback and parameter optimization module.

[0031] The data acquisition and preprocessing module is used for real-time electricity data, historical load data, basic customer information and external environment data, and performs multi-level verification, marks single-point anomalies and systemic anomalies, and generates a unified input data set.

[0032] The data integration and storage module is used to analyze customer electricity usage behavior labels, geographical area characteristics and time context based on cleaned standardized data, build a customer-centric snowflake model, and dynamically expand the dimension table and establish a mapping relationship with the load fact table through semantic analysis and association rule mining.

[0033] The customer group division and control strategy generation module is used to extract customer electricity consumption behavior characteristics, initialize cluster centers, divide customer groups, and dynamically re-cluster based on internal standard deviations or external events. Based on historical load data and the influence of external variables, it generates time-based benchmark thresholds and scenario-based thresholds, and formulates peak shaving strategies and precise control strategies.

[0034] The closed-loop feedback and parameter optimization module is used to collect feedback data. When there is a prediction error, it automatically triggers incremental training, retains historical versions and rolls back. According to the policy execution effect, the trigger threshold is optimized and the restrictions are relaxed according to the time attenuation coefficient.

[0035] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a data warehouse-based power customer demand management method when executing the computer program.

[0036] The present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for managing power customer demand based on a data warehouse are implemented.

[0037] The beneficial effects of the present invention are as follows: by collecting multi-source data in real time and performing multi-level verification, data cleaning and standardization, high-quality data integration is achieved, the accuracy and reliability of analysis are ensured, and a solid data foundation is provided for power customer demand management. By utilizing the snowflake data model and ETL process, multi-dimensional data is effectively integrated, the efficiency and accuracy of data processing are improved, and a structured and easy-to-query data environment is provided for subsequent analysis. By dividing customer groups through clustering algorithms and combining dynamic thresholds with hierarchical competition rules to generate control strategies, accurate identification and differentiated management of customer needs are achieved, and the ability to optimize the allocation of power resources is improved. The closed-loop feedback mechanism continuously optimizes parameters, enhances the adaptability and robustness of the system, and ensures the continued effectiveness and flexibility of the control strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] 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.

[0039] Figure 1 An overall flow chart of a data warehouse-based power customer demand management method provided by one embodiment of the present invention.

[0040] Figure 2 A system solution module diagram of a data warehouse-based power customer demand management system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0041] 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.

[0042] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a method for managing power customer demand based on a data warehouse, comprising:

[0043] S1: Real-time collection of multi-source data, multi-level verification, and data cleaning and standardization.

[0044] The multi-source data collection includes collecting real-time electricity data, historical load data, basic customer information and external environment data;

[0045] The multi-level verification includes a first-level verification and a second-level verification. The first-level verification includes automatically triggering a cross-data source cross-comparison for similar data with a timestamp deviation of more than 5 minutes, and selecting the majority of consistent data as the benchmark value through a voting mechanism. The second-level verification includes using an interpolation algorithm to fill in missing values ​​based on historical data distribution characteristics and contextual relevance when data integrity is less than 80%, and marking interpolation marks.

[0046] It should be noted that the data cleaning includes marking single-point anomalies based on a preset fluctuation threshold, analyzing the trend deviation of consecutive data points through a sliding window, and marking it as a systematic anomaly when three consecutive points deviate from the preset pattern;

[0047] Through a cross-data source voting mechanism and an interpolation algorithm based on historical distribution, the time asynchrony and missing value issues of multi-source data in the power industry are resolved, ensuring data consistency. The use of sliding window analysis combined with dual threshold (single point anomaly and systematic anomaly) detection significantly improves the accuracy of abnormal data identification. Standardization processing eliminates the interference of data of different dimensions on subsequent analysis.

[0048] The standardization includes performing Z-score standardization on the dimension difference data to generate a unified input data set, which is expressed as follows:

[0049]

[0050] Among them, x is the original data, μ is the mean of the data, σ is the standard deviation of the data, and z is the standardized data.

[0051] S2: Use the snowflake data model to integrate multidimensional data and extract, transform and load data through the ETL process.

[0052] Furthermore, the snowflake data model includes automatically parsing customer electricity usage behavior labels, geographic area characteristics, and time context based on cleaned and standardized data to build a customer-centric snowflake model. When new external data, such as economic indicators and weather events, is added, the dimension table is dynamically expanded and a mapping relationship with the load fact table is established through semantic analysis and association rule mining.

[0053] The snowflake data model also includes designing priority weighting rules to address conflicts in different data sources for the same customer information. Data source weights are dynamically assigned based on authority, and a final value is generated through a weighted voting algorithm. The conflicting data sources and weight assignment results are recorded.

[0054] The weighted voting algorithm is expressed as:

[0055]

[0056] Among them, V final is the final value of the weighted data, w i is the weight of the i-th data point, V i is the actual value of the i-th data point, n is the total number of data points, and i is the variable index.

[0057] The dynamic dimension expansion mechanism enables the system to automatically absorb new types of data (such as indicators of sudden weather events). The priority weight rules effectively solve the common problem of data conflicts in multiple meters for one household in the power industry. The complete conflict handling records provide support for data traceability.

[0058] Furthermore, the ETL process includes dynamic priority data extraction, data transformation and intelligent loading;

[0059] Dynamic priority data extraction includes dynamically adjusting the data capture order based on the data source health score and business urgency. Real-time data is aggregated through streaming windows and stored in a cache queue. Batch data is extracted in blocks according to time slices. The formula is expressed as:

[0060] P fetch =α·H source +β·U business

[0061] Among them, α and β are weight coefficients, U business is the business urgency level, P fetch The order of data fetching;

[0062] Data conversion includes automatically matching the adaptation strategies in the conversion rule library according to semantic labels for related dimensional data, such as weather and economic indicators; intelligent loading includes loading load peak records into the in-memory database based on the query hotspot analysis results, and compressing the remaining data and storing it in distributed storage.

[0063] S3: Use clustering algorithms to divide customer groups, combine dynamic thresholds with hierarchical competition rules to generate control strategies, and optimize parameters through closed-loop feedback.

[0064] Furthermore, the division of customer groups includes extracting customer electricity consumption behavior characteristics, load fluctuation rate and external correlation characteristics, constructing feature vectors, and initializing cluster centers using an improved K-means++ algorithm;

[0065] The improved K-means++ algorithm is expressed as:

[0066]

[0067] Among them, K is the number of cluster centers, C j is the jth cluster, y is the data point, ε j is the jth cluster center, γ is the entropy penalty coefficient, j is the variable index, Entropy(C j ) is the entropy of the jth cluster; when the internal standard deviation of the cluster is greater than 15% or an external event is triggered, the clusters are automatically re-clustered and the label library is updated;

[0068] The entropy penalty term effectively solves the noise sensitivity problem of traditional clustering in power customer grouping. The multivariable dynamic threshold increases the warning accuracy by 35% compared with the fixed threshold. The rule competition mechanism properly handles the contradiction between global optimization and local power supply protection in power grid operation.

[0069] The generation and control strategy includes generating a benchmark threshold for each time period based on the moving average and fluctuation range of historical load data, superimposing the influence coefficient of external variables, and generating a scenario-based threshold. When the overall load forecast value of the system exceeds the dynamic threshold, the maximum range peak shaving strategy is activated. When the real-time data is abnormal, precise control is triggered. When multiple rules are triggered at the same time, a comprehensive strategy is generated based on priority weights.

[0070] Furthermore, the parameters optimized through closed-loop feedback include collecting actual load reduction rate, control response speed, equipment loss change, complaint rate, satisfaction survey, model prediction error, rule triggering frequency and data delay rate;

[0071] When there is a prediction error, incremental training is automatically triggered, the historical version is retained and rolled back, and based on the policy execution effect, the triggering threshold is relaxed according to the time attenuation coefficient.

[0072] Example 2, reference Figure 2 , which is the second embodiment of the present invention, provides an electricity customer demand management system based on a data warehouse, including: a data acquisition and preprocessing module, a data integration and storage module, a customer group division and regulation strategy generation module, and a closed-loop feedback and parameter optimization module.

[0073] The data acquisition and preprocessing module is used for real-time electricity data, historical load data, basic customer information and external environment data, and performs multi-level verification, marks single-point anomalies and systemic anomalies, and generates a unified input data set.

[0074] The data integration and storage module is used to analyze customer electricity usage behavior labels, geographical area characteristics and time context based on cleaned standardized data, build a customer-centric snowflake model, and dynamically expand the dimension table and establish a mapping relationship with the load fact table through semantic analysis and association rule mining.

[0075] The customer group division and control strategy generation module is used to extract customer electricity consumption behavior characteristics, initialize cluster centers, divide customer groups, and dynamically re-cluster based on internal standard deviations or external events. Based on historical load data and the influence of external variables, it generates time-based benchmark thresholds and scenario-based thresholds, and formulates peak shaving strategies and precise control strategies.

[0076] The closed-loop feedback and parameter optimization module is used to collect feedback data. When there is a prediction error, it automatically triggers incremental training, retains historical versions and rolls back. According to the policy execution effect, the trigger threshold is optimized and the restrictions are relaxed according to the time attenuation coefficient.

[0077] 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 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.

[0078] Example 3 is the third embodiment of the present invention, which differs from the first two embodiments in that:

[0079] 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.

[0080] 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.

[0081] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0082] 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.

Claims

1. A method for managing power customer demand based on a data warehouse, characterized by: include, Real-time collection of multi-source data, multi-level verification, and data cleaning and standardization; Use snowflake data model to integrate multidimensional data, and extract, transform and load data through ETL process; Use clustering algorithms to divide customer groups, combine dynamic thresholds with hierarchical competition rules to generate control strategies, and optimize parameters through closed-loop feedback.

2. The method for managing power customer demand based on a data warehouse according to claim 1, characterized in that: The multi-source data collection includes collecting real-time electricity data, historical load data, basic customer information and external environment data; The multi-level verification includes a first-level verification and a second-level verification. The first-level verification includes automatically triggering a cross-data source cross-comparison for similar data with a timestamp deviation of more than 5 minutes, and selecting the majority of consistent data as the benchmark value through a voting mechanism. The second-level verification includes using an interpolation algorithm to fill in missing values ​​based on historical data distribution characteristics and contextual relevance when data integrity is less than 80%, and marking interpolation marks.

3. The method for managing power customer demand based on a data warehouse according to claim 2, characterized in that: The data cleaning includes marking single point anomalies based on a preset fluctuation threshold, analyzing the trend deviation of consecutive data points through a sliding window, and marking it as a systematic anomaly when three consecutive points deviate from the preset pattern; The standardization includes performing Z-score standardization on the dimension difference data to generate a unified input data set.

4. The method for managing power customer demand based on a data warehouse according to claim 3, characterized in that: The snowflake data model includes automatically parsing customer electricity usage behavior tags, geographic area characteristics, and time context based on cleaned and standardized data to build a customer-centric snowflake model. When new external data, such as economic indicators and weather events, is added, the dimension table is dynamically expanded and a mapping relationship with the load fact table is established through semantic analysis and association rule mining; The snowflake data model also includes designing priority weighting rules to address conflicts in different data sources for the same customer information. Data source weights are dynamically assigned based on authority, and a final value is generated through a weighted voting algorithm. The conflicting data sources and weight assignment results are recorded. The weighted voting algorithm is expressed as: Among them, V final is the final value of the weighted data, w i is the weight of the i-th data point, V i is the actual value of the i-th data point, n is the total number of data points, and i is the variable index.

5. The method for managing power customer demand based on a data warehouse according to claim 4, characterized in that: The ETL process includes dynamic priority data extraction, data transformation and intelligent loading; Dynamic priority data extraction includes dynamically adjusting the data capture order based on the data source health score and business urgency. Real-time data is aggregated through streaming windows and stored in a cache queue. Batch data is extracted in blocks according to time slices. The formula is expressed as: P fetch =α·H source +β·U business Among them, α and β are weight coefficients, U business is the business urgency level, P fetch The order of data fetching; Data conversion includes automatically matching the adaptation strategies in the conversion rule library according to semantic labels for related dimensional data, such as weather and economic indicators; intelligent loading includes loading load peak records into the in-memory database based on the query hotspot analysis results, and compressing the remaining data and storing it in distributed storage.

6. The method for managing power customer demand based on a data warehouse according to claim 5, characterized in that: The customer group division includes extracting customer electricity consumption behavior characteristics, load fluctuation rate and external correlation characteristics, constructing feature vectors, and initializing cluster centers using an improved K-means++ algorithm; The improved K-means++ algorithm is expressed as: Among them, K is the number of cluster centers, C j is the jth cluster, y is the data point, ε j is the jth cluster center, γ is the entropy penalty coefficient, j is the variable index, Entropy(C j ) is the entropy of the jth cluster; when the internal standard deviation of the cluster is greater than 15% or an external event is triggered, the clusters are automatically re-clustered and the label library is updated; The generation and control strategy includes generating a benchmark threshold for each time period based on the moving average and fluctuation range of historical load data, superimposing the influence coefficient of external variables, and generating a scenario-based threshold. When the overall load forecast value of the system exceeds the dynamic threshold, the maximum range peak shaving strategy is activated. When the real-time data is abnormal, precise control is triggered. When multiple rules are triggered at the same time, a comprehensive strategy is generated based on priority weights.

7. The method for managing power customer demand based on a data warehouse according to claim 6, characterized in that: The parameters optimized through closed-loop feedback include collecting actual load reduction rate, control response speed, equipment loss change, complaint rate, satisfaction survey, model prediction error, rule triggering frequency and data delay rate; When there is a prediction error, incremental training is automatically triggered, the historical version is retained and rolled back, and based on the policy execution effect, the triggering threshold is relaxed according to the time attenuation coefficient.

8. A power customer demand management system based on a data warehouse, applying the power customer demand management method based on a data warehouse according to any one of claims 1 to 7, characterized in that: include: Data acquisition and preprocessing module, data integration and storage module, customer group segmentation and control strategy generation module, and closed-loop feedback and parameter optimization module; The data acquisition and preprocessing module is used to collect real-time electricity data, historical load data, basic customer information and external environment data, and perform multi-level verification, mark single-point anomalies and systemic anomalies, and generate a unified input data set; The data integration and storage module is used to analyze customer electricity usage behavior tags, geographical area characteristics and time context based on the cleaned and standardized data, build a customer-centric snowflake model, and dynamically expand the dimension table and establish a mapping relationship with the load fact table through semantic analysis and association rule mining; The customer group segmentation and control strategy generation module is used to extract customer electricity consumption behavior characteristics, initialize cluster centers, divide customer groups, and dynamically re-cluster based on internal standard deviations or external events. Based on historical load data and the influence of external variables, it generates time-based benchmark thresholds and scenario-based thresholds to formulate peak shaving strategies and precise control strategies. The closed-loop feedback and parameter optimization module is used to collect feedback data. When there is a prediction error, it automatically triggers incremental training, retains historical versions and rolls back. According to the policy execution effect, the trigger threshold is optimized and the restrictions are relaxed according to the time attenuation coefficient.

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 steps of the power customer demand management method based on a data warehouse according to any one of claims 1 to 7 are implemented.

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 data warehouse-based power customer demand management method according to any one of claims 1 to 7 are implemented.

Images