Method for optimizing electricity selling strategy by using load characteristic big data

Abstract

The invention relates to the technical field of intelligent electricity utilization, in particular to a method for optimizing an electricity selling strategy by means of load characteristic big data, and the method comprises the steps: firstly extracting user electricity utilization historical records and real-time monitoring data, and generating an accurate user group feature set; predicting a potential power utilization mode evolution trend through a neural network model; strategy correction is carried out on the trend exceeding the threshold value, and a preliminary electricity selling parameter set is obtained; verifying economic benefits through a decision tree algorithm, and screening parameter subsets meeting market stability and competition requirements; determining a final electricity selling scheme framework through reinforcement learning iteration in combination with user feedback; locally correcting an unmatched part to form a complete strategy model; and after the deviation is simulated and verified through the digital twinborn scene, deploying an electricity selling execution plan capable of quickly responding. According to the method, an electricity selling strategy is converted from experience driving to data intelligent driving, the refinement degree, the dynamic adaptation capability and the robustness of the strategy are remarkably improved, and the problem that an electricity selling scheme is disjointed with a real electricity utilization mode of a user is effectively solved.

Description

Technical Field

[0001] This invention belongs to the field of smart electricity technology, specifically a method for optimizing electricity sales strategies using big data on load characteristics. Background Technology

[0002] Against the backdrop of rapid energy market development, optimizing electricity sales strategies has become a crucial link in improving power system operational efficiency, enhancing user satisfaction, and ensuring market stability. With the deepening of electricity market reforms and the increasing diversification of user electricity demands, traditional methods of formulating electricity sales strategies based on simple statistical analysis of historical data and manual experience are no longer sufficient to meet the complex and ever-changing actual needs. Existing methods generally lack the ability to delve into the deep-seated characteristics of user electricity consumption behavior. Especially when facing user groups with different industry attributes and varying sizes, their strategies often exhibit significant lag and mismatch, leading to a disconnect between electricity sales plans and users' actual electricity consumption patterns. This not only affects user experience but also weakens the market competitiveness and economic benefits of electricity sales companies.

[0003] A deeper technological bottleneck lies in the high diversity and dynamic complexity of user electricity load characteristics. On the one hand, different users exhibit significant differences in electricity consumption time distribution, peak load, electricity consumption levels, and electricity consumption habits. On the other hand, these electricity consumption characteristics are continuously affected by external environmental variables such as seasonal changes, economic cycles, industry production rhythms, and even extreme weather, exhibiting a non-linear and highly time-varying evolutionary trend. This superimposed effect makes it difficult for traditional electricity sales strategies to respond quickly and accurately to new users or sudden changes in electricity consumption patterns, easily leading to problems such as imbalances in power supply resource allocation, increased electricity costs, and even decreased power supply reliability. For example, when designing electricity sales packages for newly connected industrial users, if the coupling relationship between their production cycle and peak electricity consumption cannot be accurately identified, the established electricity price structure may deviate significantly from their actual load curve, leading to user dissatisfaction or damage to enterprise profits.

[0004] Therefore, there is an urgent need for a power sales strategy generation mechanism that can deeply integrate load characteristic big data, dynamically perceive the evolution of user electricity consumption behavior, and possess adaptive optimization capabilities, in order to overcome the limitations of existing technologies in terms of flexibility, accuracy, and foresight. This invention addresses these issues by proposing an intelligent optimization method based on a combination of cluster analysis, neural network prediction, decision tree evaluation, and reinforcement learning iteration. This method aims to achieve refined, dynamic, and efficient power sales strategies, thereby effectively addressing the challenges posed by the diversity and complexity of user electricity consumption. Summary of the Invention

[0005] The purpose of this invention is to address the aforementioned shortcomings in the prior art by providing a method for optimizing electricity sales strategies using big data on load characteristics.

[0006] The objective of this invention is achieved through the following technical solution: a method for optimizing electricity sales strategies using big data on load characteristics, comprising the following steps: S1. Extract users' electricity consumption history records and real-time monitoring data from the preset database, and use a clustering algorithm to group the electricity consumption diversity and feature complexity to obtain a set of user group features after classification. S2. Based on the user group feature set after classification, obtain external environmental variables such as seasonal changes and industry type, and use a neural network model to predict and simulate demand response and scheme formulation to determine the potential electricity consumption pattern evolution trend. S3. If the potential electricity consumption pattern evolution trend exceeds the preset threshold, the strategy disconnect is corrected and adjusted by integrating historical analysis results and experience judgment knowledge base to obtain an optimized preliminary electricity sales parameter set. S4. For the optimized preliminary electricity sales parameter set, the decision tree algorithm is used to evaluate the influencing factors of market stability and competitiveness, determine whether the parameter set meets the economic benefit requirements, and obtain the verified electricity sales parameter subset. S5. Extract key indicators from the verified electricity sales parameter subset, obtain user satisfaction feedback data as supplementary input, and use an iterative update mechanism to reinforce the innovative breakthroughs and determine the final electricity sales scheme framework. S6. If there are parts in the final electricity sales scheme framework that do not match the diversity of electricity consumption, then make local corrections based on the feature complexity analysis results to obtain a complete electricity sales strategy model that adapts to changing needs. S7. Simulate the actual application scenario through the complete electricity sales strategy model, obtain the deviation value between the simulation output result and the actual demand response, determine whether the deviation value is lower than the preset threshold, and obtain a confirmed effective electricity sales optimization path. S8. Based on the confirmed effective electricity sales optimization path, integrate the logical connections between all relevant attributes, make final deployment preparations for the scheme formulation, and determine a fast-responding electricity sales execution plan.

[0007] The present invention is further configured such that: the classified user group feature set includes electricity consumption behavior clustering labels, load fluctuation pattern recognition, and typical period electricity consumption profiles; the potential electricity consumption pattern evolution trend includes seasonal migration characteristics, industry-driven response patterns, and sensitivity to external disturbances; the optimized preliminary electricity sales parameter set includes an initial electricity price structure adjustment scheme, response incentive configuration, and load transfer guidance strategy; the verified subset of electricity sales parameters includes economic compliance parameters, market stability indicators, and user acceptance thresholds; the final electricity sales scheme framework includes a dynamic pricing mechanism, differentiated service modules, and elastic response channels; the complete electricity sales strategy model includes multi-dimensional adaptation rules, local correction trigger conditions, and strategy self-healing mechanisms; the confirmed effective electricity sales optimization path includes deviation convergence judgment, scenario coverage verification, and robustness test results; and the rapidly responsive electricity sales execution plan includes strategy deployment timing, resource scheduling plans, and user communication processes.

[0008] The present invention is further configured to extract user electricity consumption history records and real-time monitoring data from a preset database, and to use a clustering algorithm to group and process the diversity and complexity of electricity consumption characteristics to obtain a classified set of user group features. Extract user electricity consumption history records and real-time monitoring data from a pre-set database, use data cleaning technology to remove outliers and missing records, retain valid data that reflects real electricity consumption behavior, and generate a standardized electricity consumption dataset. Based on the standardized electricity consumption dataset, an unsupervised clustering method is used to divide users into several groups with similar electricity consumption behaviors according to the shape of the daily load curve, peak and valley distribution characteristics and electricity consumption continuity index, generating initial clustering results. Based on the initial clustering results, and combined with basic user information such as industry attributes and capacity levels, the clustering boundaries are finely adjusted to eliminate misclassification caused by data sparsity and generate optimized user groups. Based on the optimized user grouping, typical electricity consumption characteristics of each group are extracted, including average load level, fluctuation frequency and response delay characteristics, to generate a set of user group characteristics after classification.

[0009] The present invention is further configured such that, based on the classified user group feature set, external environmental variables such as seasonal changes and industry type are obtained, and a neural network model is used to predict and simulate demand response and scheme formulation to determine the potential electricity consumption pattern evolution trend. The specific steps are as follows: Based on the user group feature set after classification, external environmental variables, including meteorological data, holiday arrangements and industrial policy guidance, are integrated to construct a multi-source input feature matrix; Based on the multi-source input feature matrix, a nonlinear mapping relationship is established using a deep feedforward structure to simulate the electricity consumption response behavior of user groups under different external conditions and generate multi-scenario electricity consumption prediction results. Based on the multi-scenario electricity consumption prediction results, the direction and intensity of the shift in electricity consumption patterns over time are identified, the degree of deviation from historical benchmark patterns is quantified, and the pattern evolution trajectory is generated. Based on the aforementioned pattern evolution trajectory, a dynamic threshold range is set. When the trajectory exceeds the normal fluctuation range, it is marked as a potential electricity consumption pattern evolution trend.

[0010] The present invention is further configured such that, if the potential electricity consumption pattern evolution trend exceeds a preset threshold, the step of correcting and adjusting the strategy disconnect by integrating historical analysis results and an experience-based knowledge base to obtain an optimized preliminary electricity sales parameter set is as follows: If the potential electricity consumption pattern evolution trend exceeds the preset threshold, retrieve the strategy adjustment records and implementation effects of similar historical scenarios to form a reference case set. Based on the reference case set and combined with the expert experience rule base, key disconnection points between the current strategy and new user needs are identified, and a strategy deviation diagnosis report is generated. Based on the strategy deviation diagnosis report, the original electricity sales parameters are corrected in a targeted manner, including adjusting the time-of-use pricing period division, optimizing the tiered electricity consumption threshold, and resetting the response incentive triggering conditions, and generating a draft parameter adjustment; Based on the aforementioned parameter adjustment draft, and considering current market constraints, feasibility screening and conflict resolution are conducted to obtain an optimized preliminary set of electricity sales parameters.

[0011] The present invention is further configured such that, for the optimized preliminary electricity sales parameter set, a decision tree algorithm is used to evaluate the influencing factors of market stability and competitiveness, determine whether the parameter set meets the economic benefit requirements, and obtain a verified subset of electricity sales parameters. The specific steps are as follows: For the optimized preliminary electricity sales parameter set, an evaluation index system including market share, user churn rate and unit electricity sales revenue is constructed. Based on the aforementioned evaluation index system, a decision tree structure is used to decompose the path of each parameter combination and identify the key branch nodes that affect economic benefits. Based on the key branch nodes, economic benefit judgment rules are set, and each parameter combination is screened layer by layer to eliminate schemes that do not meet the minimum return or stability requirements. Based on the screening results, parameter combinations that meet both market competitiveness and financial sustainability criteria are retained, resulting in a validated subset of electricity sales parameters.

[0012] The present invention is further configured to extract key indicators from the verified subset of electricity sales parameters, obtain user satisfaction feedback data as supplementary input, and perform reinforcement learning on innovative breakthroughs through an iterative update mechanism to determine the final electricity sales scheme framework. The specific steps are as follows: Key indicators were extracted from the verified subset of electricity sales parameters, including electricity price sensitivity response rate, load regulation completion rate, and billing transparency score. Based on the aforementioned key indicators, feedback data such as user satisfaction surveys, complaint records, and renewal intentions are collected simultaneously to construct a user perception evaluation vector. Based on the user perception evaluation vector, a feedback-driven iterative mechanism is adopted to fine-tune the electricity sales parameters, strengthen the high satisfaction path and suppress negative feedback correlation items. Based on the convergence results after multiple iterations, the parameter stability region and user preference peak values ​​are integrated to determine the final electricity sales scheme framework.

[0013] The present invention is further configured such that, if there are parts in the final electricity sales scheme framework that are not compatible with the diversity of electricity consumption, the step of making local corrections based on the feature complexity analysis results to obtain a complete electricity sales strategy model that adapts to changing needs is as follows: If there are parts in the final electricity sales scheme framework that do not match the diversity of electricity consumption, locate the specific user subgroups that do not match and their abnormal electricity consumption characteristics. Based on the aforementioned anomalies, we analyze the sources of their complexity in terms of time, load, and response dimensions to identify the root causes of strategy failure. Based on the aforementioned root causes, targeted local correction rules are designed, including introducing flexible pricing windows, adding industry-specific packages, or activating backup response channels. Based on the aforementioned local correction rules, an adaptive module is formed by embedding the original scheme framework, thereby obtaining a complete electricity sales strategy model that adapts to changing needs.

[0014] The present invention is further configured such that, by simulating a real-world application scenario through the complete electricity sales strategy model, obtaining the deviation value between the simulated output result and the actual demand response, and determining whether the deviation value is lower than a preset threshold, the specific steps for obtaining a confirmed effective electricity sales optimization path are as follows: Using the complete electricity sales strategy model, a virtual operation scenario with multiple users and multiple operating conditions is constructed in the digital twin environment to perform strategy simulation. Based on the simulation results, the load response curve, user participation rate and system revenue indicators of the simulation output are collected and compared with the historical real response data. Calculate the deviation values ​​of each indicator, including response delay error, power matching degree difference, and economic benefit fluctuation range; If all deviation values ​​are below the preset threshold, the strategy path is deemed valid, and a confirmed and valid electricity sales optimization path is generated.

[0015] The present invention is further configured such that, based on the confirmed effective electricity sales optimization path, the logical relationships between all relevant attributes are integrated, and the final deployment preparation for the scheme formulation is carried out to determine a rapidly responsive electricity sales execution plan. The specific steps are as follows: Based on the confirmed effective electricity sales optimization path, the coupling relationship between electricity price structure, user grouping, response mechanism and external variables is sorted out, and a strategy logic graph is constructed. Based on the aforementioned strategy logic diagram, a phased implementation roadmap is formulated, clarifying the triggering conditions, resource allocation, and risk response measures for each phase; Based on the implementation route, an automated execution script is generated to support the automatic activation of the corresponding strategy module when a specific change in electricity consumption pattern is detected. Based on automated execution scripts, combined with user notification mechanisms and system interface specifications, a fast-responding electricity sales execution plan is determined.

[0016] The beneficial effects of this invention are: I. This invention removes abnormal and missing data through data cleaning technology, and combines unsupervised clustering algorithms with information such as user industry attributes and capacity levels to achieve accurate grouping of electricity consumption diversity and feature complexity. It generates a user group feature set containing electricity consumption behavior tags, load fluctuation patterns, and typical time period profiles, which completely solves the problem of insufficient strategy targeting caused by the fuzzy user classification in traditional methods, and provides reliable data support for subsequent strategy formulation.

[0017] Second, this invention innovatively integrates multiple external environmental variables such as meteorology, holidays, and industrial policies, and constructs nonlinear mapping relationships through a neural network model to accurately predict the seasonal migration of electricity consumption patterns and the evolution of industry-driven responses, thereby identifying potential demand changes in advance. This changes the passive situation of traditional strategies that are adjusted after the fact, and ensures that the strategy can proactively adapt to external disturbances and dynamic changes in user demand.

[0018] Third, this invention constructs an evaluation system that includes market share, user churn rate, and unit electricity sales revenue through a decision tree algorithm. It filters the electricity sales parameter set layer by layer, which not only meets the financial sustainability requirements of electricity sales companies, but also ensures market stability and user acceptance. It effectively avoids the problem of benefit imbalance caused by the single goal orientation of traditional strategies, and enhances the core competitiveness of electricity sales companies.

[0019] Fourth, this invention incorporates feedback data such as user satisfaction surveys, complaint records, and renewal intentions into an optimization closed loop. It fine-tunes electricity sales parameters through a reinforcement learning iterative mechanism, while designing a dynamic pricing mechanism, differentiated service modules, and flexible response channels to specifically meet the personalized needs of different user groups. This solves the problems of poor user experience and low renewal rates caused by the one-size-fits-all approach of traditional strategies, and builds a positive power supply and consumption interaction relationship.

[0020] Fifth, this invention locates and solves the mismatch between the solution and the diversity of electricity consumption through a local correction mechanism, embeds an adaptive module to form a self-healing strategy; combined with multi-user, multi-condition simulation verification in a digital twin environment, it ensures that the deviation of the strategy in complex scenarios such as extreme weather and policy adjustments is lower than a preset threshold, effectively avoiding risks such as imbalance in power supply resource allocation, increased electricity costs, and decreased power supply reliability caused by strategy failure, thereby improving the stability of power system operation.

[0021] VI. The invention constructs a strategy logic map and phased implementation roadmap, and realizes the rapid activation of strategies after changes in electricity consumption patterns through automated script execution. Combined with resource scheduling plans and user communication processes, it ensures that strategies can be implemented quickly, helping electricity sales companies to respond quickly to market changes and seize competitive opportunities in the reform of the electricity market, while optimizing the efficiency of energy resource allocation and contributing to the achievement of dual carbon goals. Attached Figure Description

[0022] The invention will be further illustrated with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the invention. For those skilled in the art, other drawings can be obtained based on the following drawings without any creative effort.

[0023] Figure 1 This is a schematic diagram of the present invention. Detailed Implementation

[0024] The present invention will be further described in conjunction with the following embodiments.

[0025] Depend on Figure 1 As can be seen, the embodiments of the present invention provide a method for optimizing electricity sales strategies using big data on load characteristics, and its overall process is as follows: Figure 1 As shown, it includes eight steps, S1 to S8, executed sequentially. The following is combined with... Figure 1 The specific embodiments of the present invention will be described in detail below.

[0026] In practical applications, suppose a provincial power company wants to develop more competitive and adaptable electricity sales plans for industrial and commercial users within its jurisdiction to improve user satisfaction, increase market share, and ensure the stability of the power grid. First, in step S1, the system extracts users' electricity consumption history records and real-time monitoring data for the past three years from a pre-set database (e.g., SCADA system, AMI advanced measurement system, and CRM customer relationship management system). This includes fields such as load curves collected every 15 minutes, monthly electricity consumption, maximum demand, industry category, and transformer capacity. Then, a data cleaning technique based on Z-score standardization is used to remove outliers (such as negative values ​​and abrupt changes) and missing records (users with a missing rate exceeding 20% ​​are excluded), generating a standardized electricity consumption dataset D. Next, the K-means++ clustering algorithm is used to perform unsupervised grouping of D, with the objective function being to minimize the within-cluster squared error (WCSS). ;in, The preset number of clusters is 6 (determined by the elbow rule). Indicates the first User clusters, Its centroid. Clustering feature dimensions include the shape of the daily load curve (quantified by Dynamic Time Warping (DTW) distance) and the peak-to-valley difference rate. Continuous electricity usage days, etc. After the initial clustering results are generated, the boundary samples are further corrected by combining basic user information (such as industry attributes: manufacturing, service, data center; capacity level: ≤500kVA, 500–2000kVA, >2000kVA). For example, users who originally belonged to the high-fluctuation cluster but belonged to the data center industry are reclassified into the "high base load, low fluctuation" cluster, thereby eliminating misclassification caused by data sparsity. Finally, a user group feature set G is generated after classification, which includes electricity behavior clustering labels (such as G1: manufacturing high-load type during the day, G2: service flexible type at night), load fluctuation pattern recognition (standard deviation σ, coefficient of variation CV), and electricity usage profiles for typical time periods (morning peak 7–10 am, midday low 12–14 pm, etc.).

[0027] In step S2, the system constructs a multi-source input feature matrix X based on G and external environmental variables. The rows of X correspond to different user groups, and the columns include internal features (e.g., average load of G1 at 1.2MW, peak-to-valley difference rate of 0.65) and external variables (e.g., average temperature of 28℃ for the next 7 days, whether it is near the Spring Festival holiday, and local policies restricting production in high-energy-consuming industries). Subsequently, a three-layer deep feedforward neural network (DNN) model is constructed. The number of nodes in the input layer equals the dimension of X (e.g., 12 dimensions), the hidden layer uses the ReLU activation function, and the output layer predicts the daily load curve for the next 30 days. The nonlinear mapping relationship of this DNN can be expressed as: ;in, This is the weight matrix. For bias terms, This is the ReLU function. After training the model using the backpropagation algorithm, multi-scenario electricity consumption forecasts are generated (e.g., a 15% increase in G1 load under a high-temperature scenario, and a 20% decrease in G3 load under production restriction policies). The Euclidean distance between the forecast curve and the historical baseline pattern is further calculated, and a dynamic threshold is set (e.g., the historical distance mean + 2 standard deviations). If the threshold is exceeded, it is marked as a potential electricity consumption pattern evolution trend T, for example, T1: the seasonal migration of summer manufacturing loads arrives earlier in May; T2: the data center industry's sensitivity to electricity price responses increases sharply.

[0028] In step S3, if T exceeds a preset threshold (e.g., the distance deviation of T1 reaches 0.8, and the threshold is set to 0.6), the system automatically retrieves historical similar cases from the knowledge base (e.g., strategy adjustment records during the summer heat wave of 2022) to form a reference case set R. Combined with the expert rule base (e.g., "When manufacturing load migration occurs ahead of schedule and lasts for more than 10 days, peak summer time-of-use pricing should be initiated earlier"), a strategy deviation diagnosis report is generated, indicating that the current strategy is out of sync (e.g., the original time-of-use pricing off-peak period was still set at 23:00–07:00, but the actual off-peak period for users has shifted to 13:00–17:00). Based on this, the original electricity sales parameters were adjusted in a targeted manner: the time-of-use pricing periods were divided into peak (10–12, 16–18), mid-peak (8–10, 12–16, 18–20), flat (7–8, 20–23), and valley (23–7); the tiered electricity consumption threshold was increased from 3000 kWh / month to 3500 kWh / month; and the response incentive trigger condition was relaxed from "10% load reduction" to "5% load transfer." After screening by market constraints (such as the upper limit of transmission and distribution prices and the minimum guaranteed electricity consumption), the optimized preliminary electricity sales parameter set P was obtained, which includes the initial adjustment plan of the electricity price structure, the response incentive configuration (such as a subsidy of 5 yuan per kW for load transfer), and the load transfer guidance strategy (such as pushing reminders to users to charge during valley hours via the APP).

[0029] Step S4 uses a decision tree algorithm to verify the economic benefits of P. An evaluation index system I is constructed, including market share (currently 85%), user churn rate (target <3%), and unit electricity sales revenue (target >0.45 yuan / kWh). A CART decision tree is used to decompose the parameter combinations in P. For example, the root node is split according to the "time-of-use electricity price peak ratio". The average user churn rate of the left subtree (peak ratio ≤ 1.5) is 2.1%, and that of the right subtree (>1.5) is 4.7%. A judgment rule is set: only branches that simultaneously satisfy unit revenue ≥ 0.45 yuan / kWh and churn rate ≤ 3% are retained. After layer-by-layer screening, a subset of valid electricity sales parameters Q is obtained. For example, Q1: peak ratio 1.4, off-peak period extended by 2 hours, incentive subsidy 4.8 yuan / kW. Its economic performance parameters (revenue 0.47 yuan / kWh), market stability index (churn rate 2.5%), and user acceptance threshold (survey acceptance > 70%) all meet the requirements.

[0030] In step S5, key indicators K are extracted from Q, including electricity price sensitivity response rate (65% measured), load regulation completion rate (92%), and bill transparency score (4.2 / 5). User feedback data F is collected simultaneously, such as satisfaction surveys (N=1000, mean 4.0), complaint records (15 per month, mainly about complex bills), and renewal intention (88%). A user perception evaluation vector V=[0.65,0.92,4.2,4.0,0.88] is constructed and iteratively updated using a Q-learning reinforcement learning mechanism. State s is defined as the current parameter combination, action a as a fine-tuning operation (e.g., "peak electricity price reduced by 0.02 yuan"), and reward r is the weighted sum of V (weights determined by the AHP analytic hierarchy process).

[0031] Update Q-value function: ; in, For learning rate, The discount factor was used. After 20 rounds of iteration, the Q value converged, and high satisfaction paths (such as simplified bill templates and a slight reduction in peak electricity prices to 1.38 yuan / kWh) were strengthened. The final electricity sales plan framework F was finally determined, which includes a dynamic pricing mechanism (adjusted quarterly based on load forecasts), a differentiated service module (providing a dedicated customer service channel for G2 users), and a flexible response channel (supporting users to participate in demand response with one click on the APP).

[0032] Step S6 checks if there are any mismatches between F and the diversity of electricity consumption. For example, it was found that the participation rate of G4 (small catering businesses) users in the flexible response channel was only 30%, far lower than other groups. Anomalies in their electricity consumption characteristics were identified: a sudden surge in load on weekends but extremely low load on weekdays, a pattern not covered by the original plan. The complexity of the analysis stemmed from the irregular business hours (random opening between 10:00 and 22:00). Based on this, local correction rules were designed: introducing a flexible pricing window (users can choose 4 hours of off-peak electricity per day), adding a special package for catering businesses (an extra 2 hours of off-peak electricity on weekends), and activating the backup response channel (quick confirmation via SMS). These rules were embedded into F to form an adaptive module, resulting in a complete electricity sales strategy model M. This model has multi-dimensional adaptation rules covering 7 user types, local correction trigger conditions (e.g., triggering when a group's weekly participation rate is <40%), and a strategy self-healing mechanism (automatically rolling back to the previous stable version).

[0033] Step S7 constructs a virtual scenario in the digital twin platform: simulating 100,000 users (allocated according to the G ratio), superimposed with high temperature (35℃) and holiday conditions (7-day National Day holiday). After executing the M-model, the following indicators are collected and output: load response curve (predicted vs. actual deviation ±3%), user participation rate (82%), and system revenue (+5% month-on-month). The deviation value Δ is calculated: response delay error (actual 15 minutes, threshold 30 minutes), electricity matching degree difference (98.5%, threshold 95%), and economic revenue fluctuation (standard deviation 0.02 yuan / kWh, threshold 0.05). Since all Δ values ​​are below the threshold, a confirmed and effective electricity sales optimization path R is generated, including deviation convergence judgment (Δ < threshold for 7 consecutive days), scenario coverage verification (covering 95% of historical extreme events), and robustness test results (stable under ±10% load disturbance).

[0034] Finally, step S8 constructs a strategy logic graph L based on R, clarifying the coupling relationship between electricity price structure, user grouping, response mechanism, and external variables (such as a 0.8% increase in peak load for G1 for every 1°C increase in temperature). A phased implementation roadmap is formulated: Phase 1 (days 1–30) deploys automated scripts, triggering time-period adjustments when G1 experiences three consecutive days of load migration; Phase 2 (days 31–60) activates resource scheduling contingency plans (reserving 50MW of interruptible load); Phase 3 (days 61–90) optimizes user communication processes (pushing personalized electricity consumption suggestions). A rapidly responsive electricity sales execution plan E is generated, including strategy deployment timing (automatic calibration daily at 02:00), resource scheduling contingency plans (interfacing with the virtual power plant platform), and user communication processes (APP pop-ups + SMS dual channels), ensuring efficient, accurate, and user-friendly electricity sales strategy execution in a real power grid environment.

[0035] Finally, 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 scope of protection of 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 modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing electricity sales strategies using big data on load characteristics, characterized in that: Includes the following steps: S1. Extract users' electricity consumption history records and real-time monitoring data from the preset database, and use a clustering algorithm to group the electricity consumption diversity and feature complexity to obtain a set of user group features after classification. S2. Based on the user group feature set after classification, obtain external environmental variables such as seasonal changes and industry type, and use a neural network model to predict and simulate demand response and scheme formulation to determine the potential electricity consumption pattern evolution trend. S3. If the potential electricity consumption pattern evolution trend exceeds the preset threshold, the strategy disconnect is corrected and adjusted by integrating historical analysis results and experience judgment knowledge base to obtain an optimized preliminary electricity sales parameter set. S4. For the optimized preliminary electricity sales parameter set, the decision tree algorithm is used to evaluate the influencing factors of market stability and competitiveness, determine whether the parameter set meets the economic benefit requirements, and obtain the verified electricity sales parameter subset. S5. Extract key indicators from the verified electricity sales parameter subset, obtain user satisfaction feedback data as supplementary input, and use an iterative update mechanism to reinforce the innovative breakthroughs and determine the final electricity sales scheme framework. S6. If there are parts in the final electricity sales scheme framework that do not match the diversity of electricity consumption, then make local corrections based on the feature complexity analysis results to obtain a complete electricity sales strategy model that adapts to changing needs. S7. Simulate the actual application scenario through the complete electricity sales strategy model, obtain the deviation value between the simulation output result and the actual demand response, determine whether the deviation value is lower than the preset threshold, and obtain a confirmed effective electricity sales optimization path. S8. Based on the confirmed effective electricity sales optimization path, integrate the logical connections between all relevant attributes, make final deployment preparations for the scheme formulation, and determine a fast-responding electricity sales execution plan.

2. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: The categorized user group feature set includes electricity consumption behavior clustering labels, load fluctuation pattern recognition, and typical period electricity consumption profiles. The potential electricity consumption pattern evolution trend includes seasonal migration characteristics, industry-driven response patterns, and sensitivity to external disturbances. The optimized preliminary electricity sales parameter set includes an initial electricity price structure adjustment scheme, response incentive configuration, and load transfer guidance strategy. The verified subset of electricity sales parameters includes economic compliance parameters, market stability indicators, and user acceptance thresholds. The final electricity sales scheme framework includes a dynamic pricing mechanism, differentiated service modules, and flexible response channels. The complete electricity sales strategy model includes multi-dimensional adaptation rules, local correction trigger conditions, and strategy self-healing mechanisms. The confirmed effective electricity sales optimization path includes deviation convergence judgment, scenario coverage verification, and robustness test results. The rapidly responsive electricity sales execution plan includes strategy deployment timing, resource scheduling plans, and user communication processes.

3. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: The specific steps for extracting user electricity consumption history records and real-time monitoring data from a pre-set database, and using a clustering algorithm to group users based on electricity consumption diversity and feature complexity to obtain a categorized set of user group features are as follows: Extract user electricity consumption history records and real-time monitoring data from a pre-set database, use data cleaning technology to remove outliers and missing records, retain valid data that reflects real electricity consumption behavior, and generate a standardized electricity consumption dataset. Based on the standardized electricity consumption dataset, an unsupervised clustering method is used to divide users into several groups with similar electricity consumption behaviors according to the shape of the daily load curve, peak and valley distribution characteristics and electricity consumption continuity index, generating initial clustering results. Based on the initial clustering results, and combined with basic user information such as industry attributes and capacity levels, the clustering boundaries are finely adjusted to eliminate misclassification caused by data sparsity and generate optimized user groups. Based on the optimized user grouping, typical electricity consumption characteristics of each group are extracted, including average load level, fluctuation frequency and response delay characteristics, to generate a set of user group characteristics after classification.

4. The method for optimizing electricity sales strategies using load characteristic big data as described in claim 1, characterized in that: Based on the categorized user group feature set, external environmental variables such as seasonal changes and industry type are obtained. A neural network model is then used to predict and simulate demand response and solution formulation to determine the potential evolution trend of electricity consumption patterns. The specific steps are as follows: Based on the user group feature set after classification, external environmental variables, including meteorological data, holiday arrangements and industrial policy guidance, are integrated to construct a multi-source input feature matrix; Based on the multi-source input feature matrix, a nonlinear mapping relationship is established using a deep feedforward structure to simulate the electricity consumption response behavior of user groups under different external conditions and generate multi-scenario electricity consumption prediction results. Based on the multi-scenario electricity consumption prediction results, the direction and intensity of the shift in electricity consumption patterns over time are identified, the degree of deviation from historical benchmark patterns is quantified, and the pattern evolution trajectory is generated. Based on the aforementioned pattern evolution trajectory, a dynamic threshold range is set. When the trajectory exceeds the normal fluctuation range, it is marked as a potential electricity consumption pattern evolution trend.

5. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: If the potential electricity consumption pattern evolution trend exceeds a preset threshold, the specific steps for correcting and adjusting the strategy disconnect by integrating historical analysis results and an experience-based knowledge base to obtain an optimized preliminary electricity sales parameter set are as follows: If the potential electricity consumption pattern evolution trend exceeds the preset threshold, retrieve the strategy adjustment records and implementation effects of similar historical scenarios to form a reference case set. Based on the reference case set and combined with the expert experience rule base, key disconnection points between the current strategy and new user needs are identified, and a strategy deviation diagnosis report is generated. Based on the strategy deviation diagnosis report, the original electricity sales parameters are corrected in a targeted manner, including adjusting the time-of-use pricing period division, optimizing the tiered electricity consumption threshold, and resetting the response incentive triggering conditions, and generating a draft parameter adjustment; Based on the aforementioned parameter adjustment draft, and considering current market constraints, feasibility screening and conflict resolution are conducted to obtain an optimized preliminary set of electricity sales parameters.

6. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: For the optimized preliminary electricity sales parameter set, the decision tree algorithm is used to evaluate the influencing factors of market stability and competitiveness, determine whether the parameter set meets the economic benefit requirements, and obtain the validated subset of electricity sales parameters. The specific steps are as follows: For the optimized preliminary electricity sales parameter set, an evaluation index system including market share, user churn rate and unit electricity sales revenue is constructed. Based on the aforementioned evaluation index system, a decision tree structure is used to decompose the path of each parameter combination and identify the key branch nodes that affect economic benefits. Based on the key branch nodes, economic benefit judgment rules are set, and each parameter combination is screened layer by layer to eliminate schemes that do not meet the minimum return or stability requirements. Based on the screening results, parameter combinations that meet both market competitiveness and financial sustainability criteria are retained, resulting in a validated subset of electricity sales parameters.

7. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: The specific steps for extracting key indicators from the verified subset of electricity sales parameters, obtaining user satisfaction feedback data as supplementary input, and using an iterative update mechanism to reinforce innovative breakthroughs and determine the final electricity sales solution framework are as follows: Key indicators were extracted from the verified subset of electricity sales parameters, including electricity price sensitivity response rate, load regulation completion rate, and billing transparency score. Based on the aforementioned key indicators, feedback data such as user satisfaction surveys, complaint records, and renewal intentions are collected simultaneously to construct a user perception evaluation vector. Based on the user perception evaluation vector, a feedback-driven iterative mechanism is adopted to fine-tune the electricity sales parameters, strengthen the high satisfaction path and suppress negative feedback correlation items. Based on the convergence results after multiple iterations, the parameter stability region and user preference peak values ​​are integrated to determine the final electricity sales scheme framework.

8. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: If the final electricity sales scheme framework contains parts that are incompatible with the diversity of electricity consumption, the specific steps for obtaining a complete electricity sales strategy model that adapts to changing needs by making local corrections based on the feature complexity analysis results are as follows: If there are parts in the final electricity sales scheme framework that do not match the diversity of electricity consumption, locate the specific user subgroups that do not match and their abnormal electricity consumption characteristics. Based on the aforementioned anomalies, we analyze the sources of their complexity in terms of time, load, and response dimensions to identify the root causes of strategy failure. Based on the aforementioned root causes, targeted local correction rules are designed, including introducing flexible pricing windows, adding industry-specific packages, or activating backup response channels. Based on the aforementioned local correction rules, an adaptive module is formed by embedding the original scheme framework, thereby obtaining a complete electricity sales strategy model that adapts to changing needs.

9. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: The steps of simulating real-world application scenarios using the complete electricity sales strategy model, obtaining the deviation between the simulated output and the actual demand response, and determining whether the deviation is below a preset threshold to obtain a confirmed effective electricity sales optimization path are as follows: Using the complete electricity sales strategy model, a virtual operation scenario with multiple users and multiple operating conditions is constructed in the digital twin environment to perform strategy simulation. Based on the simulation results, the load response curve, user participation rate and system revenue indicators of the simulation output are collected and compared with the historical real response data. Calculate the deviation values ​​of each indicator, including response delay error, power matching degree difference, and economic benefit fluctuation range; If all deviation values ​​are below the preset threshold, the strategy path is deemed valid, and a confirmed and valid electricity sales optimization path is generated.

10. The method for optimizing electricity sales strategies using load characteristic big data according to claim 1, characterized in that: Based on the confirmed effective electricity sales optimization path, and integrating the logical relationships between all relevant attributes, the final deployment preparation for the scheme formulation is carried out. The specific steps for determining a fast-responding electricity sales execution plan are as follows: Based on the confirmed effective electricity sales optimization path, the coupling relationship between electricity price structure, user grouping, response mechanism and external variables is sorted out, and a strategy logic graph is constructed. Based on the aforementioned strategy logic diagram, a phased implementation roadmap is formulated, clarifying the triggering conditions, resource allocation, and risk response measures for each phase; Based on the implementation route, an automated execution script is generated to support the automatic activation of the corresponding strategy module when a specific change in electricity consumption pattern is detected. Based on automated execution scripts, combined with user notification mechanisms and system interface specifications, a fast-responding electricity sales execution plan is determined.

Images