Carbon neutralization-oriented peak-valley electricity price dynamic excitation mechanism design and user response prediction system

Abstract

The invention relates to the technical field of power systems, in particular to a carbon neutralization-oriented peak-valley electricity price dynamic excitation mechanism design and user response prediction system, which comprises a data acquisition module, a load characteristic analysis module, a dynamic regulation and control module, a sample evaluation module and an execution module. The system obtains power grid operation, user power consumption behaviors and new energy power generation data through a multi-source data access interface, and realizes partition load optimization and new energy adaptation in combination with a load classification model, a dynamic excitation framework and an energy storage optimization scheduling scheme. The method can accurately predict the user response, optimizes the peak-valley electricity price strategy, improves the new energy consumption capability, and assists the realization of a carbon neutralization target.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization and demand-side management, and specifically to the design of a dynamic incentive mechanism for peak and valley electricity prices and a user response prediction system for carbon neutrality. Background Art

[0002] With the advancement of carbon neutrality goals and the increasing share of renewable energy generation in the power system, dynamic incentive mechanisms based on user behavior and grid supply and demand balance have become a key issue in achieving a low-carbon power system. Current mainstream peak-valley electricity pricing mechanisms often rely on fixed time periods, making them inadequate for adapting to regional load characteristics and fluctuating renewable energy output. With the widespread adoption of distributed energy resources (DESS) and smart meters, user electricity consumption behavior has become increasingly diverse, making traditional, single-source incentive models inflexible in guiding user participation in demand response. Furthermore, in practice, due to the lack of accurate user response prediction models, existing incentive mechanisms often struggle to fully address multiple objectives, including carbon emission control, grid operating cost optimization, and user satisfaction. Furthermore, building high-precision prediction models faces challenges such as complex data dimensions and uneven sample distribution. Traditional rule-based control methods also have limited support for real-time regulation, resulting in a certain lag in the implementation of dynamic regulation strategies, impacting the adaptability and efficiency of the overall system. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a dynamic incentive mechanism design for peak and valley electricity prices and a user response prediction system for carbon neutrality.

[0004] In order to solve the above technical problems, the technical solution of the present invention is: a design of a dynamic incentive mechanism for peak and valley electricity prices and a user response prediction system for carbon neutrality:

[0005] It includes a data acquisition module, a load characteristics analysis module, and a dynamic control module. The data acquisition module is configured with a multi-source data access interface. The data acquisition module is used to obtain power grid operation data, user electricity consumption behavior data, and new energy power generation data through the multi-source data access interface, and perform structured processing on the data to generate a unified data format;

[0006] The load characteristic analysis module is used to generate a load characteristic curve based on historical power consumption data and real-time monitoring data, and to establish a load classification model based on regional environmental characteristics and user power consumption patterns. The load classification model is used to classify different types of loads into steady-state loads, fluctuating loads, and intermittent loads;

[0007] The dynamic control module is used to build a dynamic incentive framework. The dynamic incentive framework generates a time-based electricity price adjustment strategy by introducing a price elasticity coefficient matrix and a user preference weight vector, and performs rolling optimization based on real-time feedback data.

[0008] The sample evaluation module is equipped with a bias correction algorithm that uses exponential smoothing to correct the bias. The formula is:

[0009]

[0010] The smoothing coefficient α is 0.2, and when the deviation rate is greater than 10%, the model reconstruction is triggered. It is used to calculate the deviation between the load forecast result and the actual load, and to modify the forecast model based on the deviation value to generate a high-precision load forecast sub-model;

[0011] The first control unit includes a regional coordination component and a load distribution component. The regional coordination component generates geographical partition characteristics based on preset regional division rules and generates partition load constraints based on meteorological data and user electricity consumption behavior data. The load distribution component distributes steady-state load and fluctuating load according to the partition load constraints to generate a partition load optimization plan.

[0012] The second control unit 6 includes a new energy adaptation component and an energy storage scheduling component. The new energy adaptation component generates a new energy consumption priority list by analyzing the matching relationship between the new energy power generation output characteristics and the load demand. The energy storage scheduling component schedules the charging and discharging of the energy storage equipment according to the new energy consumption priority list to generate an energy storage optimization scheduling plan;

[0013] The execution module includes inputting the partition load optimization plan, the energy storage optimization scheduling plan and the high-precision load prediction sub-model into the dynamic incentive framework, and inputting the fluctuating load data into the first control unit or the second control unit 6.

[0014] Further: it also includes an anomaly detection module, the anomaly detection module is configured with an abnormal behavior database, the abnormal behavior database stores abnormal power consumption behavior features, the abnormal power consumption behavior features have a behavior index set, the behavior index set includes a number of behavior sub-features, each behavior sub-feature can be indexed to a corresponding abnormal power consumption behavior pattern, each abnormal power consumption behavior pattern corresponds to a behavior risk value, the anomaly detection module is configured with a behavior analysis strategy, the behavior analysis strategy is used to extract the behavior sub-features in the fluctuating load data and call a number of corresponding abnormal power consumption behavior patterns, and replace the abnormal power consumption behavior pattern with the fluctuating load data through abnormal power consumption constraints to generate abnormal power consumption samples, the abnormal power consumption constraints are that the total behavior risk value of the abnormal power consumption sample falls within a preset risk threshold range;

[0015] The execution module brings the abnormal electricity usage samples into the dynamic incentive framework.

[0016] Furthermore: it also includes a data verification subsystem, which is used to collect electricity consumption data of each partition to generate historical electricity consumption samples; the data collection module is configured with a data correction unit and a data completion unit, the data correction unit is used to extract distortion features in historical electricity consumption samples that are judged to be abnormal electricity consumption samples, and the data completion unit generates corresponding data completion instructions based on the distortion features, and sends the data completion instructions to the data verification subsystem.

[0017] Furthermore: the data verification subsystem is configured with a virtual load unit, which is used to access the electrical equipment and execute corresponding data completion instructions. The data completion instructions include load simulation parameters, and the load simulation parameters are used to configure the corresponding virtual load unit.

[0018] Furthermore: the load characteristic analysis module is configured with a timing mapping unit, which generates a corresponding time series classification model by controlling the electrical equipment to work in different time periods, analyzes historical power consumption samples through the time series classification model to generate a corresponding electrical equipment working sequence, and controls the working state of the electrical equipment according to the electrical equipment working sequence to generate a real-time load sample corresponding to the historical power consumption sample.

[0019] Furthermore: the area division rules include

[0020] Step S1: Obtaining geographic location information of electrical equipment corresponding to historical electricity usage samples;

[0021] Step S2: Retrieving geographically relevant electric device numbers from a pre-built electric device distribution network model based on the geographical location information;

[0022] Step S3: Obtain corresponding relevant environmental characteristics according to the electrical equipment number;

[0023] Step S4: identifying key geographical parameters in relevant environmental features through a preset environmental adaptability model to generate the geographical partition features.

[0024] Furthermore: the new energy adapter component is configured with a new energy matching strategy, the new energy matching strategy includes

[0025] Step A1: Calculate the difference between the renewable energy power generation output curve and the load demand curve in the same time period to obtain an adaptive difference curve;

[0026] Step A2: classify the adaptation difference curve using a preset feature decomposition strategy to obtain the adaptation difference feature corresponding to each new energy type item;

[0027] Step A3: Repeat step A1 until all new energy power generation output curves are collected;

[0028] Step A4: Mark the fitness difference characteristics of each new energy type item with fitness using a preset fitness calculation method;

[0029] Step A5: performing mean processing on the adapted difference feature through the adaptation degree mark to obtain the corresponding difference mean feature;

[0030] Step A6: Combine different new energy type items to obtain a new energy adaptation cluster;

[0031] Step A7: In the new energy adaptation cluster set, cluster analysis is performed on the difference mean features using a cluster analysis algorithm to obtain a corresponding new energy adaptation index cluster.

[0032] Furthermore: the first control unit includes a partition optimization strategy, the partition optimization strategy includes

[0033] Step B1: Split the geographical partition feature into a number of geographical sub-feature links, obtain geographical feature response items, and configure a corresponding response weight value for each geographical feature response item corresponding to the geographical sub-feature;

[0034] Step B2: construct a geographic feature network using geographic sub-features as nodes;

[0035] Step B3: Calculate the comprehensive weight value of each geographic sub-feature using a preset response weight algorithm;

[0036] Step B4: Use a preset neighborhood expansion algorithm to determine whether any adjacent geographic sub-features meet the expansion conditions. If the expansion conditions are met, adjust the positions of the two geographic sub-features in the geographic feature network until all geographic sub-features do not meet the expansion conditions.

[0037] Furthermore: the second control unit 6 is configured with a random energy storage strategy, which includes generating corresponding random variable items and corresponding random variable intervals according to the new energy adaptation characteristics, generating random variable values ​​of each random variable item according to the random variable intervals, generating a random energy storage scheduling plan according to the random variable values, replacing the random energy storage scheduling plan with the energy storage optimization scheduling plan to generate the new energy adaptation load sample, and correcting the corresponding new energy adaptation characteristics.

[0038] The technical effects of the present invention are mainly reflected in the following aspects: the system obtains grid operation, user electricity consumption behavior and new energy power generation data through a multi-source data access interface, and combines the load classification model, dynamic incentive framework and energy storage optimization scheduling scheme to realize partition load optimization and new energy adaptation. This application can accurately predict user response, optimize peak and valley electricity price strategies, improve new energy absorption capacity, and help achieve carbon neutrality goals. The system obtains the grid operation data, user electricity consumption behavior data and new energy power generation data in the area through the data acquisition module, and generates load characteristic curves and load classification models through the load characteristic analysis module. The dynamic control module formulates a time-based electricity price adjustment strategy based on the analysis results, and inputs the partition load optimization scheme and energy storage optimization scheduling scheme into the dynamic incentive framework through the execution module. At the same time, the anomaly detection module performs anomaly detection on the fluctuating load data and generates abnormal electricity consumption samples to verify the stability of the system. Ultimately, the system achieves effective guidance of user electricity consumption behavior through the dynamic incentive framework, significantly improves the utilization rate of new energy and reduces carbon emissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 : A dynamic incentive mechanism design for peak-valley electricity prices towards carbon neutrality and the overall structural block diagram of the user response prediction system;

[0040] Figure 2 : A design of a dynamic incentive mechanism for peak-valley electricity prices for carbon neutrality and a workflow diagram of the load characteristic analysis module of the user response prediction system;

[0041] Figure 3 :A flow chart of the new energy matching strategy in the new energy adaptation component of a carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system;

[0042] Figure 4 :A flow chart of the first control unit partition optimization strategy for the design of a dynamic incentive mechanism for peak-valley electricity prices and a user response prediction system for carbon neutrality;

[0043] Figure 5 : A design of a dynamic incentive mechanism for peak and valley electricity prices for carbon neutrality and a working principle diagram of the anomaly detection module of the user response prediction system;

[0044] Figure 6 : An algorithm data interaction flow chart of the design of a dynamic incentive mechanism for peak and valley electricity prices for carbon neutrality and a user response prediction system.

[0045] Figure numerals: 1. Data acquisition module; 2. Load characteristic analysis module; 3. Dynamic control module; 4. Execution module; 5. Timing mapping unit; 6. Second control unit; 7. First control unit; 8. Abnormality detection module. DETAILED DESCRIPTION

[0046] The specific embodiments of the present invention are further described below in conjunction with the accompanying drawings to make the technical solutions of the present invention easier to understand and grasp.

[0047] The present invention provides a carbon neutral peak-valley electricity price dynamic incentive mechanism design and user response prediction system, the overall structure of which is as follows: Figure 1 As shown, the system includes a data acquisition module 1, a load characteristics analysis module 2, a dynamic control module 3, and an execution module 4. These modules form a complete system through data flow and logical connections, which is used to implement the design of a dynamic incentive mechanism for peak and valley electricity prices and user response prediction. The following describes the specific implementation of the present invention in detail with reference to the accompanying drawings and specific examples.

[0048] In practical applications, the data acquisition module 1 serves as the system's entry point and is equipped with a multi-source data access interface for acquiring multi-source data from grid operating equipment, user power terminals, and renewable energy power generation equipment. This data includes, but is not limited to, grid operating status parameters, user power consumption records, and renewable energy power generation output data. It also includes a data verification subsystem, which collects power consumption data for each partition to generate historical power consumption samples. The data acquisition module is equipped with a data correction unit and a data completion unit. The data correction unit extracts distortion features from historical power consumption samples identified as abnormal. The data completion unit generates corresponding data completion instructions based on the distortion features and sends these instructions to the data verification subsystem. The data verification subsystem is equipped with a virtual load unit, which connects to the power consumption equipment and executes corresponding data completion instructions. The data completion instructions include load simulation parameters, which are used to configure the corresponding virtual load unit. To ensure data consistency and availability, the data acquisition module 1 preprocesses the acquired data, including data cleaning, format conversion, and outlier removal, ultimately generating a unified data format and passing it to subsequent modules. This process relies on the collaborative work of the data correction unit and the data completion unit, where the data correction unit is responsible for identifying and correcting distorted data, while the data completion unit generates completion instructions based on the context information of the missing data. For example, when data is missing due to equipment failure within a certain period of time, the data completion unit will call the virtual load unit to simulate the corresponding load parameters to fill the data gaps, thereby ensuring data integrity. The unified data format adopts the JSON Schema standard, including fields such as timestamp (timestamp), node number (node_id), data type (type), and value (value); the data cleaning rule uses the Z-score method to eliminate outliers (threshold ±3σ), and missing values ​​are completed through a bidirectional LSTM network.

[0049] The data preprocessing unit uses a bidirectional LSTM network to fill missing values. The specific process is as follows:

[0050] 1. For the original data sequence X=[x1,x2,\dots,x n ] for standardization;

[0051] 2. Input a bidirectional LSTM network (hidden layer dimension = 32) and use forward and backward propagation to capture contextual dependencies;

[0052] 3. Output the padded sequence The error is evaluated using the mean absolute error (MAE).

[0053] The load characteristic analysis module 2 receives the structured data from the data acquisition module 1 and generates a load characteristic curve based on the historical power consumption data and real-time monitoring data. The module is internally configured with a time sequence mapping unit 5, and its workflow is as follows: Figure 2 As shown. The load characteristics analysis module 2 is configured with a time series mapping unit. The time series mapping unit generates a corresponding time series classification model by controlling the operation of electrical equipment in different time periods. The time series classification model analyzes historical power consumption samples to generate corresponding power equipment operation sequences. The operating state of the electrical equipment is controlled according to the power equipment operation sequence to generate real-time load samples corresponding to the historical power consumption samples. The time series mapping unit 5 first generates a time series classification model by controlling the operating state of the electrical equipment in different time periods. On this basis, the time series mapping unit 5 analyzes the power equipment operation sequences in the historical power consumption samples and maps them to real-time power consumption scenarios to generate corresponding real-time load samples. For example, if the historical power consumption data of a certain area shows that air conditioning equipment is concentratedly turned on during the peak period in summer, the time series mapping unit 5 will replicate this pattern in the current period, thereby providing a basis for subsequent load forecasting. In addition, the load characteristics analysis module 2 also establishes a load classification model based on regional environmental characteristics and user power consumption patterns, dividing the load into steady-state load, fluctuating load, and intermittent load. Steady-state loads: load fluctuations ≤ 5% and lasting ≥ 1 hour; fluctuating loads: load fluctuations 5%-30% with a periodic pattern; and intermittent loads: loads that start and stop randomly and have no fixed pattern (such as electric vehicle charging). The model is trained using the K-means algorithm, with k = 3 clusters and the Euclidean distance metric. This classification method effectively distinguishes the characteristics of different load types, supporting subsequent dynamic control.

[0054] The dynamic control module 3 is the core part of the system. Its main function is to build a dynamic incentive framework and generate a time-based electricity price adjustment strategy. The dynamic incentive framework introduces a price elasticity coefficient matrix and a user preference weight vector, comprehensively considers the user's electricity consumption behavior characteristics and the market supply and demand relationship, and formulates a reasonable electricity price adjustment plan. The price elasticity coefficient matrix is ​​E = [E ij ] m×n , where m is the number of electricity price periods, n is the number of load types (such as peak / flat / valley load); E ij The load transfer rate (%) for period i when the electricity price changes by 1% during period j is obtained through regression analysis of historical electricity consumption data or machine learning model training. The specific formula is:

[0055]

[0056] where ΔL i is the load change during period i, ΔP j The user preference weight vector is ω=[ω1,ω2,,...,ω k ], where k is the user preference dimension (such as electricity price sensitivity, electricity usage time preference, device type), ω i ∈[0,1] represents the importance weight of the i-th dimension, which is determined by the analytic hierarchy process (AHP) of user survey data or historical behavior data and is used to adjust the priority of time-of-use electricity prices for user responses.

[0057] For example, when the renewable energy power generation output in a certain area is high and the user electricity demand is low, the dynamic incentive framework will lower the electricity price during this period to encourage users to increase electricity consumption; otherwise, the electricity price will be increased to suppress electricity demand. At the same time, the dynamic control module 3 also has rolling optimization capabilities, which can dynamically adjust the electricity price strategy based on real-time feedback data. The optimization cycle is 15 minutes, and the objective function is a multi-objective optimization model, which includes minimizing carbon emissions, minimizing grid operation costs, and maximizing user satisfaction. It is solved using the non-dominated sorting genetic algorithm (NSGA-II), with a population size of 100 and 50 iterations. For example, when the actual load in a certain area exceeds the predicted value, the dynamic control module 3 will re-evaluate the current electricity price strategy and make corresponding adjustments to ensure the stable operation of the system. The specific contents of the dynamic adjustment module are as follows: Dynamic control module

[0058] The dynamic incentive framework is built based on a multi-objective optimization model. The objective function includes carbon neutrality, grid operation costs, and user satisfaction. The specific definitions are as follows: Carbon neutrality objective function:

[0059]

[0060] Among them, L tis the grid load in period t (MW), G t is the renewable energy power generation in period t (MW), e t is the carbon emission per unit of electricity (kg / kWh), and the goal is to minimize carbon emissions.

[0061] Grid operation cost objective function:

[0062]

[0063] Among them, C p is the unit electricity generation cost (yuan / kWh), C s is the grid regulation cost (yuan / MW), and the goal is to minimize the grid operation cost and the regulation loss caused by load fluctuations.

[0064] User electricity satisfaction objective function:

[0065]

[0066] Among them, α i is the electricity price sensitivity coefficient of user i (obtained through training of historical electricity consumption data), P t is the electricity price in period t (yuan / kWh), is the expected electricity price of user i in time period t, and the goal is to maximize the user's acceptance of the electricity price.

[0067] Constraints:

[0068] 1. Power balance constraint: L t =G t +S t , where S t is the power supply of traditional energy (MW), and S t ≥0;

[0069] 2. Electricity price range constraints: P min ≤P t ≤P max , where P min 、P max The upper and lower limits of electricity prices (determined by electricity market policies);

[0070] 3. User load constraint: L i,min ≤L i,t ≤L i,max , where L i,t is the electricity load of user i in time period t (MW).

[0071] Optimization algorithm: Non-dominated sorting genetic algorithm (NSGA-II) is used to solve multi-objective optimization problems. c =0.8, mutation probability p m=0.1, population size N p =100 for iterative optimization, and finally obtain the Pareto optimal solution set.

[0072] User response prediction algorithm

[0073] The user response prediction module adopts a random forest-neural network hybrid model, and its specific structure is as follows:

[0074] 1. Feature Engineering: User electricity consumption characteristics: daily electricity consumption (kWh), peak-offset electricity consumption ratio (%), equipment power (kW); user attribute characteristics: user type (residential / industrial / commercial), household size, average monthly electricity bill (yuan); incentive characteristics: peak-offset electricity price difference (yuan / kWh), subsidy intensity (yuan / kWh).

[0075] 1. Random forest model: This model contains 100 decision trees, which are used to capture nonlinear feature interactions and output preliminary predictions of user load transfer (MW).

[0076] 2. Neural network model: A 3-layer fully connected network (input layer dimension = 15, hidden layer neurons = 64, output layer dimension = 3) was used. The activation function was ReLU to optimize the prediction results. The loss function was the root mean square error (RMSE).

[0077] Dynamic adjustment algorithm of incentive mechanism: The dynamic adjustment module is based on model predictive control (MPC) and is implemented with the following core steps:

[0078] 1. State variable definition: Real-time grid load (MW); User's actual load response (MW), in For no incentive

[0079] Price elasticity coefficient matrix:

[0080]

[0081] Among them, E ij Indicates the percentage of impact of a 1% change in electricity price during period j on the load during period i (e.g., E pv represents the shifting effect of valley electricity price changes on peak electricity load), which is obtained through historical data training.

[0082] 1. Rolling optimization model: Objective function (optimization time domain N = 24 hours):

[0083]

[0084] Among them, λ1=0.6 (carbon emission deviation weight), λ2=0.3 (load response weight), and λ3=0.1 (regulation smoothness weight) are determined through expert experience and data training.

[0085] Adjustment strategy: Short-term adjustment (minute level): When |∈ L When (t)|>θ (θ=5% is the deviation threshold), the electricity price fine-tuning is triggered:

[0086] P t+1 =P t ×(1+α×sign(∈ L (t))×|∈ L (t)|)

[0087] Wherein, α=0.1 is the adjustment intensity coefficient. t is the electricity price at time t.

[0088] Mid-term adjustment (hourly): based on new energy output forecast and user response prediction Adjust the duration of peak and valley periods:

[0089]

[0090] Among them, β = 0.01 is the time period adjustment coefficient, and T = 4 hours is the prediction period.

[0091] The execution module 4 is responsible for inputting the partition load optimization plan, energy storage optimization scheduling plan and high-precision load prediction sub-model into the dynamic incentive framework, and calling the first control unit 7 or the second control unit 6 for further processing according to the fluctuating load data.

[0092] The area division rules include

[0093] Step S1: Obtaining geographic location information of electrical equipment corresponding to historical electricity usage samples;

[0094] Step S2: Retrieving geographically relevant electric device numbers from a pre-built electric device distribution network model based on the geographical location information;

[0095] Step S3: Obtain corresponding relevant environmental characteristics according to the electrical equipment number;

[0096] Step S4: Identify key geographical parameters in relevant environmental characteristics through a preset environmental adaptability model to generate the geographical partition characteristics for generating a partition load optimization solution.

[0097] The new energy adapter component is configured with a new energy matching strategy, which includes

[0098] Step A1: Calculate the difference between the renewable energy power generation output curve and the load demand curve in the same time period to obtain an adaptive difference curve;

[0099] Step A2: classify the adaptation difference curve using a preset feature decomposition strategy to obtain the adaptation difference feature corresponding to each new energy type item; the feature decomposition strategy uses principal component analysis (PCA) to project the time domain features (such as mean, variance, and peak) of the adaptation difference curve into the principal component space, and retain the principal component with a cumulative contribution rate greater than 85% as the adaptation difference feature.

[0100] Step A3: Repeat step A1 until all new energy power generation output curves are collected;

[0101] Step A4: Mark the fitness difference characteristics of each new energy type item with fitness using a preset fitness calculation method;

[0102] Step A5: performing mean processing on the adapted difference feature through the adaptation degree mark to obtain the corresponding difference mean feature;

[0103] Step A6: Combine different new energy type items to obtain a new energy adaptation cluster;

[0104] Step A7: In the new energy adaptation cluster set, cluster analysis is performed on the difference mean features using a cluster analysis algorithm to obtain a corresponding new energy adaptation index cluster, which is used to generate an energy storage optimization scheduling plan.

[0105] The first control unit includes a partition optimization strategy, which includes

[0106] Step B1: Split the geographical partition feature into a number of geographical sub-feature links, obtain geographical feature response items, and configure a corresponding response weight value for each geographical feature response item corresponding to the geographical sub-feature;

[0107] Step B2: construct a geographic feature network using geographic sub-features as nodes;

[0108] Step B3: Calculate the comprehensive weight value of each geographic sub-feature using a preset response weight algorithm; the response weight algorithm calculates the comprehensive weight value of the geographic sub-feature based on the entropy weight method, and the formula is:

[0109]

[0110] where e i The neighborhood expansion condition is: the load correlation coefficient of adjacent geographical sub-features is greater than 0.7 and the geographical distance is less than 5 kilometers. If the conditions are met, they are merged into the same partition.

[0111] Step B4: Use a preset neighborhood expansion algorithm to determine whether any adjacent geographic sub-features meet the expansion conditions. If the expansion conditions are met, adjust the positions of the two geographic sub-features in the geographic feature network until all geographic sub-features do not meet the expansion conditions. The second control unit 6 is configured with a random energy storage strategy, which includes generating corresponding random variable items and corresponding random variable intervals according to the new energy adaptation characteristics, generating random variable values ​​for each random variable item according to the random variable interval, and generating a random energy storage scheduling plan according to the random variable value. The random variable item includes the new energy output prediction error (obeying the normal distribution N 0, σ 2 ), load fluctuation amplitude (uniformly distributed U[-L_max, L_max]); the random variable interval is determined based on historical data statistics; the correction rule is: when the energy storage charge and discharge error is greater than 5%, the mean and variance of the new energy adaptation characteristics are updated through Kalman filtering. The random energy storage scheduling plan is replaced with the energy storage optimization scheduling plan to generate the new energy adaptation load sample, and the corresponding new energy adaptation characteristics are corrected.

[0112] The first control unit 7 includes a regional coordination component and a load distribution component, and its working principle is as follows: Figure 4 As shown. The regional coordination component generates geographical zoning features through preset regional division rules. The specific steps include obtaining the geographical location information of the power equipment, extracting relevant environmental features from the power equipment distribution network model, and identifying key geographical parameters through the environmental adaptability model. Subsequently, the load distribution component reasonably distributes the steady-state load and the fluctuating load according to the partition load constraints. For example, when the steady-state load in a certain area is high, the load distribution component will give priority to allocating the fluctuating load to other areas to balance the overall load distribution. The second control unit 6 includes a new energy adaptation component and an energy storage scheduling component, and its workflow is as follows Figure 3 As shown in the figure, the new energy adapter component generates a priority list of new energy consumption by analyzing the matching relationship between the output characteristics of new energy generation and load demand. The energy storage scheduling component then schedules the charging and discharging of energy storage devices based on this list to maximize the utilization of new energy. For example, when the photovoltaic power generation output is high, the energy storage scheduling component will prioritize storing excess electricity in the energy storage device for subsequent use. The dynamic incentive framework and the execution module achieve real-time communication through a message queue (MQTT protocol).

[0113] Anomaly detection module, the anomaly detection module is configured with an abnormal behavior database, the abnormal behavior database stores abnormal electricity consumption behavior features, the abnormal electricity consumption behavior features have a behavior index set, the behavior index set includes several behavior sub-features, each behavior sub-feature can be indexed to the corresponding abnormal electricity consumption behavior pattern, each abnormal electricity consumption behavior pattern corresponds to a behavior risk value, the anomaly detection module is configured with a behavior analysis strategy, the behavior analysis strategy is used to extract the behavior sub-features in the fluctuating load data and call several corresponding abnormal electricity consumption behavior patterns, and replace the abnormal electricity consumption behavior pattern with the fluctuating load data through abnormal electricity consumption constraints to generate abnormal electricity consumption samples, the abnormal electricity consumption constraints are that the total behavior risk value of the abnormal electricity consumption sample falls within the preset risk threshold range; the execution module brings the abnormal electricity consumption sample into the dynamic incentive framework. The anomaly detection module 8 is an important supplementary part of the system, and its working principle is as follows Figure 5 As shown. The anomaly detection module 8 is configured with an abnormal behavior database for storing abnormal electricity consumption behavior characteristics and their corresponding behavioral risk values. In actual operation, the anomaly detection module 8 extracts behavioral sub-features from the fluctuating load data through a behavioral analysis strategy, and retrieves the corresponding abnormal electricity consumption behavior pattern. The behavioral sub-features include quantitative indicators such as load fluctuation frequency (times / hour), power mutation amplitude (kW), and time period deviation (%); the behavioral index set realizes rapid mapping of features to patterns through a hash table structure. The behavioral risk value adopts a 5-level scoring system (1-low risk, 5-high risk), which is determined according to the degree of impact of abnormal electricity consumption behavior on the stability of the power grid. For example, a load mutation amplitude of >20kW corresponds to a risk value of 4. For example, when a user's electricity consumption behavior shows abnormal high-frequency fluctuations, the anomaly detection module 8 will mark it as a potential abnormal electricity consumption behavior and generate abnormal electricity consumption samples through abnormal electricity consumption constraints. These samples are then input into the dynamic incentive framework to verify the robustness and adaptability of the system.

[0114] The data verification subsystem serves as an auxiliary unit of the system, which is used to collect electricity consumption data for each partition and generate historical electricity consumption samples. The data verification subsystem is internally configured with a virtual load unit to execute data completion instructions. For example, when there is missing data in the historical electricity consumption samples of a certain area, the virtual load unit will generate corresponding virtual load data based on the load simulation parameters to fill the missing part. The load simulation parameters include rated power (kW), power factor, start-stop cycle, etc., and virtual load simulation is achieved through MATLAB / Simulink software modeling with an error range of ≤±3%. This process ensures the integrity and accuracy of the historical electricity consumption samples, and provides a reliable data basis for subsequent load forecasting and dynamic regulation.

[0115] Through the collaborative work of the above modules, this system can realize the design of dynamic incentive mechanism for peak and valley electricity prices and user response prediction for carbon neutrality. For example, in a typical application scenario, a city's power grid faces greater pressure on electricity consumption during the peak period in summer. The system obtains the power grid operation data, user electricity consumption behavior data and new energy power generation data in the area through the data acquisition module 1, and generates load characteristic curves and load classification models through the load characteristic analysis module 2. The dynamic control module 3 formulates a time-based electricity price adjustment strategy based on the analysis results, and inputs the partition load optimization plan and energy storage optimization scheduling plan into the dynamic incentive framework through the execution module 4. At the same time, the anomaly detection module 8 performs anomaly detection on the fluctuating load data and generates abnormal electricity consumption samples to verify the stability of the system. Ultimately, the system achieves effective guidance of user electricity consumption behavior through the dynamic incentive framework, significantly improving the utilization rate of new energy and reducing carbon emissions.

[0116] In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is supplemented below with reference to a specific application scenario.

[0117] During the peak period of summer in a certain city, the power grid faces great pressure on electricity consumption. The system obtains the power grid operation data, user electricity consumption behavior data and renewable energy power generation data in the area through the data acquisition module 1. The data acquisition module 1 first extracts raw data from the power grid operation equipment, user electricity terminals and distributed photovoltaic power generation equipment through the multi-source data access interface. These data include the real-time load curve of the power grid, the user's electricity consumption records and the output data of photovoltaic power generation. Since some data are missing or have abnormal values ​​due to equipment transmission problems, the data correction unit identifies distorted data through preset rules, and the data completion unit generates completion instructions. For example, when the power generation data is missing due to a photovoltaic equipment failure in a certain time period, the data completion unit calls the virtual load unit to simulate the photovoltaic output parameters of the corresponding time period to fill the data gaps, thereby ensuring data integrity.

[0118] Subsequently, the load characteristic analysis module 2 receives the structured data from the data acquisition module 1 and generates a load characteristic curve based on historical power consumption data and real-time monitoring data. The time series mapping unit 5 generates a time series classification model by controlling the operating status of electrical appliances such as air conditioners and water heaters during different time periods. Based on this, the time series mapping unit 5 maps the operating sequences of electrical appliances in historical power consumption samples to the current real-time power consumption scenario, generating corresponding real-time load samples. The electrical appliance operating sequences are linked to historical samples and real-time control instructions using a rule engine. The rule is: if the probability of the appliance being turned on during time period t in the historical data is greater than 70%, the real-time control instruction is set to "on." The validation criterion for real-time load samples is a root mean square error (RMSE) of less than 5%. For example, if historical data shows that air conditioners in a region are concentrated during peak summer hours, the time series mapping unit 5 replicates this pattern in the current time period, providing a basis for subsequent load forecasting. Furthermore, the load characteristic analysis module 2 classifies loads into steady-state, fluctuating, and intermittent loads based on regional environmental characteristics and user power consumption patterns. This classification method effectively distinguishes the characteristics of different load types, supporting subsequent dynamic control.

[0119] Based on the load characteristic curve and classification model generated by the load characteristic analysis module 2, the dynamic control module 3 constructs a dynamic incentive framework and generates a time-based electricity price adjustment strategy. The dynamic incentive framework comprehensively considers the user's electricity consumption behavior and market supply and demand by introducing a price elasticity coefficient matrix and a user preference weight vector. For example, when the photovoltaic power generation output in a certain area is high and the user's electricity demand is low, the dynamic incentive framework will lower the electricity price during that period to encourage users to increase electricity consumption; conversely, it will increase the electricity price to suppress electricity demand. At the same time, the dynamic control module 3 has a rolling optimization capability and can dynamically adjust the electricity price strategy based on real-time feedback data. For example, when the actual load in a certain area exceeds the predicted value, the dynamic control module 3 re-evaluates the current electricity price strategy and makes corresponding adjustments to ensure the stable operation of the system.

[0120] The execution module 4 is responsible for inputting the zoning load optimization plan, energy storage optimization scheduling plan, and high-precision load prediction sub-model into the dynamic incentive framework. The regional coordination component of the first control unit 7 generates geographical zoning features based on preset regional division rules. Specifically, the regional coordination component first obtains the geographic location information of the power equipment and extracts relevant environmental features from the pre-built power equipment distribution network model. It then identifies key geographical parameters through the environmental adaptability model. The input parameters of the environmental adaptability model include meteorological data (temperature, humidity), terrain data (altitude, slope), and population density data; the key geographical parameters are extracted through principal component analysis to extract the first two principal components, with a cumulative contribution rate of >90%. For example, if the key geographical parameters of a certain area show that its load carrying capacity is strong, the load distribution component will prioritize allocating fluctuating loads to other areas to balance the overall load distribution. The second control unit 6 analyzes the matching relationship between photovoltaic power generation output characteristics and load demand through the new energy adaptation component to generate a new energy consumption priority list. The energy storage scheduling component schedules the charging and discharging of energy storage devices based on this list. For example, when the photovoltaic power generation output is high, the energy storage scheduling component will prioritize storing excess electricity in the energy storage device for subsequent use.

[0121] In actual operation, the anomaly detection module 8 uses behavioral analysis strategies to extract behavioral sub-features from fluctuating load data and retrieve corresponding abnormal power usage patterns. For example, if a user's power usage exhibits abnormally high-frequency fluctuations, the anomaly detection module 8 marks it as potentially abnormal and generates abnormal power usage samples based on abnormal power usage constraints. These samples are then input into the dynamic incentive framework to verify the system's robustness and adaptability.

[0122] Through the collaborative work of the above modules, the system implements the design of a dynamic incentive mechanism for peak and valley electricity prices and user response prediction for carbon neutrality. For example, in a typical application scenario, the system significantly improved the utilization rate of new energy and reduced carbon emissions through a dynamic incentive framework. Specifically, when photovoltaic power generation output is high, the system guides users to increase electricity consumption by lowering electricity prices, while using energy storage devices to store excess electricity, thereby maximizing the utilization rate of new energy. In addition, by detecting and processing abnormal electricity consumption behavior, the system further improves operational stability and ensures a balance between electricity supply and demand.

[0123] Of course, the above are only typical examples of the present invention. In addition, the present invention may also have many other specific implementation methods. Any technical solutions formed by equivalent replacement or equivalent transformation fall within the scope of protection required by the present invention.

Claims

1. A carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system, characterized by: The system comprises a data acquisition module (1), a load characteristics analysis module (2), a dynamic control module (3) and an execution module (4), wherein the data acquisition module (1) is configured with a multi-source data access interface for acquiring power grid operation data, user electricity consumption behavior data and new energy generation data, and performing structured processing on the data to generate a unified data format; The load characteristic analysis module (2) is used to generate a load characteristic curve based on historical power consumption data and real-time monitoring data, and to establish a load classification model in combination with regional environmental characteristics and user power consumption patterns. The load classification model is used to classify different types of loads into steady-state loads, fluctuating loads, and intermittent loads. The dynamic control module (3) is used to build a dynamic incentive framework. The dynamic incentive framework generates a time-based electricity price adjustment strategy by introducing a price elasticity coefficient matrix and a user preference weight vector, and performs rolling optimization based on real-time feedback data. The execution module (4) is used to input the partition load optimization plan, the energy storage optimization scheduling plan and the high-precision load prediction sub-model into the dynamic incentive framework, and call the first control unit or the second control unit (6) according to the fluctuating load data.

2. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The system further includes an anomaly detection module (8), which is configured with an abnormal behavior database. The abnormal behavior database stores abnormal electricity consumption behavior features. The abnormal electricity consumption behavior features have a behavior index set. The behavior index set includes a number of behavior sub-features. Each behavior sub-feature can be indexed to a corresponding abnormal electricity consumption behavior pattern. Each abnormal electricity consumption behavior pattern corresponds to a behavior risk value. The anomaly detection module (8) is configured with a behavior analysis strategy. The behavior analysis strategy is used to extract behavior sub-features from fluctuating load data and retrieve a number of corresponding abnormal electricity consumption behavior patterns. The abnormal electricity consumption behavior pattern is replaced with the fluctuating load data through abnormal electricity consumption constraints to generate abnormal electricity consumption samples. The abnormal electricity consumption constraints are that the total behavior risk value of the abnormal electricity consumption sample falls within a preset risk threshold range. The execution module (4) brings the abnormal electricity consumption sample into the dynamic incentive framework.

3. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The invention also includes a data verification subsystem, which is used to collect the electricity consumption data of each partition to generate a historical electricity consumption sample; the data collection module (1) is configured with a data correction unit and a data completion unit, the data correction unit is used to extract the distortion features in the historical electricity consumption samples judged as abnormal electricity consumption samples, and the data completion unit generates a corresponding data completion instruction according to the distortion features and sends the data completion instruction to the data verification subsystem.

4. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 3 is characterized by: The data verification subsystem is configured with a virtual load unit, which is used to access the electrical equipment and execute the corresponding data completion instructions. The data completion instructions include load simulation parameters, which are used to configure the corresponding virtual load unit.

5. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The load characteristic analysis module (2) is configured with a time sequence mapping unit (5). The time sequence mapping unit (5) generates a corresponding time sequence classification model by controlling the electric equipment to work in different time periods, analyzes historical power consumption samples through the time sequence classification model to generate a corresponding electric equipment working sequence, and controls the working state of the electric equipment according to the electric equipment working sequence to generate a real-time load sample corresponding to the historical power consumption sample.

6. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The zoning rules include the following steps: Step S1: Obtaining geographic location information of electrical equipment corresponding to historical electricity usage samples; Step S2: Retrieving geographically relevant electric device numbers from a pre-built electric device distribution network model based on the geographical location information; Step S3: Obtain corresponding relevant environmental characteristics according to the electrical equipment number; Step S4: identifying key geographical parameters in relevant environmental features through a preset environmental adaptability model to generate geographical partition features.

7. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The new energy adapter component is configured with a new energy matching strategy, which includes the following steps: Step A1: Calculate the difference between the renewable energy power generation output curve and the load demand curve in the same time period to obtain an adaptive difference curve; Step A2: classify the adaptation difference curve using a preset feature decomposition strategy to obtain the adaptation difference feature corresponding to each new energy type item; Step A3: Repeat step A1 until all new energy power generation output curves are collected; Step A4: Mark the fitness difference characteristics of each new energy type item with fitness using a preset fitness calculation method; Step A5: performing mean processing on the adapted difference feature through the adaptation degree mark to obtain the corresponding difference mean feature; Step A6: Combine different new energy type items to obtain a new energy adaptation cluster; Step A7: In the new energy adaptation cluster set, cluster analysis is performed on the difference mean features using a cluster analysis algorithm to obtain a corresponding new energy adaptation index cluster.

8. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The first control unit is configured with a partition optimization strategy, which includes the following steps: Step B1: Split the geographical partition feature into a number of geographical sub-feature links, obtain geographical feature response items, and configure a corresponding response weight value for each geographical feature response item corresponding to the geographical sub-feature; Step B2: construct a geographic feature network using geographic sub-features as nodes; Step B3: Calculate the comprehensive weight value of each geographic sub-feature using a preset response weight algorithm; Step B4: Use a preset neighborhood expansion algorithm to determine whether any adjacent geographic sub-features meet the expansion conditions. If the expansion conditions are met, adjust the positions of the two geographic sub-features in the geographic feature network until all geographic sub-features do not meet the expansion conditions.

9. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1 is characterized by: The second control unit (6) is configured with a random energy storage strategy, which includes generating corresponding random variable items and corresponding random variable intervals according to the new energy adaptation characteristics, generating random variable values ​​of each random variable item according to the random variable intervals, generating a random energy storage scheduling plan according to the random variable values, replacing the random energy storage scheduling plan with the energy storage optimization scheduling plan to generate a new energy adaptation load sample, and correcting the corresponding new energy adaptation characteristics.

10. The carbon neutrality-oriented peak-valley electricity price dynamic incentive mechanism design and user response prediction system according to claim 1, characterized in that: The sample evaluation module is equipped with a deviation correction algorithm, which is used to calculate the deviation between the load forecast result and the actual load, and to correct the forecast model according to the deviation value to generate a high-precision load forecast sub-model.

Images