Interaction method and system for electricity marketing

Abstract

The invention provides an interaction method and system for electricity marketing, and the method comprises the steps: collecting the electricity consumption data of a user through an Internet of Things device, and carrying out the preprocessing of the data; a power consumption behavior prediction model is constructed according to the preprocessed data to predict the future power demand of the user, and a response is made to sudden changes; constructing a dynamic electric power pricing model according to the prediction result, and adjusting the electric power pricing according to the actual demand of the user and the load change of the power grid; market conditions are monitored through real-time data streams, and potential anomalies are found in time in combination with a trained power consumption behavior prediction model and a dynamic power pricing model; and an optimal power package is recommended for each user through an optimization algorithm, and real-time adjustment is carried out. According to the invention, the marketing accuracy and effect of the electric power company are improved, the interactivity and energy-saving awareness of the user are enhanced through a social mechanism, and the transformation of the electric power industry to the direction of intelligence, individuation and socialization is promoted.

Description

Technical Field

[0001] The present invention belongs to the field of power marketing, and in particular relates to an interactive method and system for power marketing. Background Art

[0002] As a key component in the energy industry, power marketing has traditionally relied on static pricing models and periodic promotions to attract consumers. However, with the increasingly fierce competition in the power market and the increasing diversification of consumer demand, traditional power marketing methods have gradually exposed their limitations. First, power demand itself has significant seasonal and temporal characteristics. For example, electricity consumption is higher in winter and summer, and power demand in the evening and morning is often higher than during the daytime off-peak period. In addition, due to factors such as consumers' lifestyles, the number of family members, and the area where they live, power consumption behavior often presents personalized and dynamic changes. Therefore, it is difficult for a single power pricing model and unified marketing activities to respond quickly to these changes, resulting in poor marketing results, reduced customer satisfaction, and loss of potential profits for power companies.

[0003] On the other hand, with the rapid development of smart home and Internet of Things (IoT) technologies, consumers' electricity consumption behavior has gradually become more intelligent and traceable. Traditional electricity marketing models cannot fully utilize the large amount of real-time data provided by these smart devices, thus missing the opportunity for personalized recommendations and dynamic pricing. Although some power companies have tried to adopt technologies based on big data analysis and machine learning to predict users' electricity consumption behavior, these technologies still face some challenges. For example, although existing smart meters and home appliances can provide accurate electricity consumption data, there is a lack of effective ways to convert these data into business intelligence that can drive accurate recommendations for power products and services. In addition, although personalized recommendations have become an important means for many industries to improve user experience, personalized recommendations in electricity marketing still lack in-depth analysis of consumers' electricity consumption behavior, resulting in recommendations that are usually only rough guesses based on historical consumption data, lacking real-time and accuracy. Therefore, the existing electricity marketing system is still insufficient in responding to user needs and providing accurate marketing solutions.

[0004] At the same time, the rise of social media and social platforms has also brought new challenges and opportunities to the power marketing model. Many industries have promoted product sales through social recommendations and user interactions. Although the power industry is also exploring this direction, most attempts remain at the level of simple promotion and interaction, and fail to truly combine the social recommendation mechanism with the personalized needs of power consumption behavior. The consumption behavior, energy-saving experience and product recommendations shared by users on social platforms have certain reference value for other users. However, most existing power marketing systems fail to effectively integrate this social data with personalized recommendation algorithms, resulting in insignificant social marketing effects, low user participation, and the recommendation and promotion of power products failing to give full play to the advantages of socialization.

[0005] Therefore, the existing power marketing technology faces many problems, which are mainly reflected in the following aspects: First, the traditional pricing and recommendation methods cannot be adjusted according to the user's real-time power consumption behavior and time pattern; second, the data generated by smart devices has not been effectively utilized, and it is impossible to provide personalized and dynamic marketing suggestions for power companies; third, the application of social recommendation mechanism in power marketing is not yet mature, user participation and interactivity are low, and social recommendation cannot effectively promote product sales. Therefore, how to combine advanced technologies such as smart Internet of Things, big data, and machine learning to provide accurate marketing recommendations for personalized power needs, and at the same time encourage user participation and interaction through social platforms, has become a technical problem that needs to be solved in the field of power marketing. Summary of the invention

[0006] The purpose of this invention is to propose an interactive method and system for power marketing, which collects power consumption data generated by smart home devices in real time, combines the user's time perception behavior, and predicts the user's future power demand through advanced machine learning algorithms, thereby providing users with personalized power product recommendations. This technological breakthrough can effectively solve the problems of insufficient personalized recommendations and untimely dynamic responses in existing power marketing systems.

[0007] In order to achieve the above object, a first aspect of the present invention provides an interactive method for power marketing, the method comprising:

[0008] S1. Collecting the user's power consumption data through IoT devices, preprocessing the data, obtaining the preprocessed power consumption data, introducing the user's social behavior as the user behavior feature, and integrating the diversity of home devices and the changes in environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp;

[0009] S2. Construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes; wherein the power consumption behavior prediction model adopts a time series convolutional neural network, extracts high-order features from the time series data of power consumption through the convolution layer, and identifies long-term dependencies. During the training process, the convolution layer and the pooling layer work together to extract the potential rules of user power consumption, and the fully connected layer integrates the time series features with the user behavior features and external environmental factors to predict power demand;

[0010] Wherein, the power consumption behavior prediction model is designed as a multi-task learning framework based on the use of the time series convolutional neural network, and two objectives are optimized respectively: power demand prediction and user behavior prediction, wherein the loss function of the power consumption behavior prediction model is designed as a weighted combination of the mean square error loss of power demand prediction and the cross entropy loss of user behavior classification;

[0011] S3. A dynamic electricity pricing model is constructed based on the prediction results, and electricity pricing is adjusted according to the actual needs of users and changes in grid load; wherein the dynamic electricity pricing model is constructed as follows:

[0012] Design a dynamic electricity pricing mechanism based on electricity demand forecasting and user behavior forecasting to flexibly adjust electricity supply according to real-time demand and price;

[0013] Design multiple constraints, including supply-demand balance constraints and fairness constraints;

[0014] Use a deep Q learning algorithm based on reinforcement learning to dynamically adjust the pricing strategy; wherein the deep Q learning algorithm is expressed as:

[0015] Q(s t , a t )←Q(s t , a t )+α[r t+1 +γmax a′ Q(s t+1 , a′)-Q(s t , a t )]

[0016] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual return, γ is the discount factor;

[0017] S4, real-time data stream monitoring of market conditions, combined with the trained power consumption behavior prediction model and dynamic power pricing model, timely detect potential anomalies, and issue intelligent warnings if anomalies are found;

[0018] S5. Combining historical data and real-time data of electricity marketing, we recommend the best electricity package for each user through optimization algorithm and make real-time adjustments to ensure that the most suitable electricity plan is provided to users under different conditions.

[0019] Furthermore, the preprocessing includes:

[0020] Interpolation based on historical consumption patterns of users: Assuming that the electricity consumption patterns of users in a specific time period have certain regularity, the missing values ​​are predicted through historical data;

[0021] A standardization method based on equipment category is adopted: During the standardization process, the mean and standard deviation of each equipment are calculated based on its historical power consumption data, and then the power consumption data at the current moment is standardized according to the following formula;

[0022] Multi-time scale processing: For intraday time periods, intraweek time periods and holidays, an algorithm based on time series feature nested weighted average is used to adjust the impact of historical data on the current moment through different weights to obtain pre-processed power consumption data;

[0023] Data feature extension is introduced to collect user social behaviors as user behavior features, and the diversity of home devices and changes in environmental factors as external environmental factors.

[0024] Furthermore, for the power demand forecasting regression task, the loss function is:

[0025]

[0026] in, is the actual power demand, To forecast electricity demand;

[0027] For the user behavior classification task, the loss function is:

[0028]

[0029] Among them, y i is the actual label of the user behavior, is the predicted behavior probability;

[0030] Finally, the loss function of the electricity consumption behavior prediction model is for:

[0031]

[0032] Among them, λ1 and λ2 are the weight coefficients between tasks.

[0033] Furthermore, in order to improve the generalization ability of the power consumption behavior prediction model and avoid overfitting, an adaptive regularization mechanism is introduced. The adaptive regularization mechanism dynamically adjusts the strength of the regularization term based on the changes in user behavior data to avoid imposing the greatest constraints on those users whose behavior patterns change the most. It is expressed as:

[0034]

[0035] in, is the L2 norm of user behavior features, α t is the regularization strength at the current moment. β1 and β2 are adjustment parameters;

[0036] The Adam optimizer is used to train the model and the back propagation algorithm is used to minimize the loss function. , and adopt the learning rate decay strategy to improve the stability of the training process;

[0037] After training, the model will output the user's future electricity demand forecast and the probability of user behavior change, and then use the RMSE and AUC indicators to evaluate the accuracy of electricity demand forecast and the effectiveness of user behavior forecast, respectively.

[0038] For power demand forecasting, the RMSE calculation formula is:

[0039]

[0040] in, Forecast future electricity demand for users, is the actual power demand of the user, N is the total number of samples, and i is the current user i;

[0041] For user behavior prediction, the AUC calculation formula is:

[0042]

[0043] Among them, TPR is the true positive rate, FPR is the false positive rate; the AUC value is closest to 1, indicating that the model performs best in the classification task.

[0044] Furthermore, the dynamic electricity pricing mechanism includes a basic electricity price and a dynamic adjustment factor, wherein the basic electricity price is a basic pricing based on the current market electricity cost and long-term trend forecast, and the dynamic adjustment factor is a price fluctuation factor adjusted based on the electricity demand forecast and user behavior forecast. The dynamic electricity pricing mechanism is expressed as:

[0045] P t =P base +ΔP t

[0046] Among them, Pt is the market electricity price, P base is the basic electricity price, ΔP t It is the dynamic adjustment part based on the power demand forecast and user behavior forecast, and its calculation method is as follows:

[0047]

[0048] Among them, γ1 and γ2 are adjustment coefficients, For user behavior prediction, E avg is the average electricity demand during a certain period of time;

[0049] The supply-demand balance constraint is used to maintain the market supply-demand balance and is expressed as:

[0050]

[0051] Among them, S t is the electricity supply at time t, ensuring that pricing matches demand; T is the total duration;

[0052] The fairness constraint is used to set the price fluctuation range to not exceed a certain ratio, which is expressed as:

[0053] ΔP t ≤α·P base

[0054] Among them, α is the preset maximum price fluctuation ratio, which is used to prevent excessive price fluctuations from affecting low-income users.

[0055] Furthermore, the S3 further includes:

[0056] Two types of evaluation indicators are used to evaluate the effect of the dynamic electricity pricing model, and the evaluation indicators include supply-demand matching and user satisfaction; the supply-demand matching Mismatch t The difference between electricity demand and supply at each moment is expressed as:

[0057]

[0058] The user satisfaction is based on the user's response behavior, calculates the user's total consumption under the adjusted pricing, and analyzes their acceptance of the pricing strategy.

[0059] Furthermore, the S4 specifically includes:

[0060] S401, obtaining operation data from each power user and the power grid through a real-time data interface; wherein the operation data includes real-time power demand, power supply, power market price and historical user behavior data;

[0061] S402, monitor and compare current power demand and supply to identify potential anomalies;

[0062] S403, based on real-time feedback, set optimization target optimization pricing strategy according to power demand and supply situation, adjust market electricity price, and achieve supply and demand balance; wherein the objective function of the optimization target is expressed as:

[0063]

[0064] Among them, E t and S t are electricity demand and supply, P t is the market electricity price, is the electricity price predicted by the model, and λ is the weight factor that balances supply and demand with electricity price fluctuations.

[0065] Furthermore, the S402 specifically includes:

[0066] Anomalies are identified by calculating the deviation between the current actual demand and the predicted demand, expressed as:

[0067]

[0068] Among them, E t represents the actual power demand, It is the expected demand based on historical data and forecasting models. t If the set threshold is exceeded, it is considered as abnormal demand;

[0069] Similarly, if the supply (S t ) and the supply capacity predicted by the model ( ) The difference exceeds the threshold, which triggers a supply anomaly alert:

[0070]

[0071] When the deviation value ΔE t or ΔS t When the preset threshold is exceeded, an intelligent warning is automatically triggered.

[0072] Furthermore, the S5 specifically includes:

[0073] S501. Design a personalized electricity package recommendation model, expressed as:

[0074]

[0075] Among them, E t is the actual power demand of the user in time period t, is the user power demand forecast based on the forecast model, P rec(t) is the recommended electricity package price, P t is the actual electricity market price, A i (t) is the user’s behavior pattern in time period t, A rec (t) is the recommended user behavior pattern, γ and λ are adjustment coefficients used to control the balance between electricity price fluctuations and behavior patterns;

[0076] S502. After the user accepts the recommended package, the power consumption behavior prediction model is adjusted to optimize the power package by real-time monitoring of user feedback, so that the recommended package is more adaptable; wherein the objective function of adjusting the power consumption behavior prediction model to optimize the power package is expressed as:

[0077]

[0078] Among them, U t is the historical data of the user’s actual electricity consumption behavior in time period t, U rec The user's electricity consumption history recommended by the system based on the prediction results. δ, μ and θ are dynamic adjustment coefficients used to balance different optimization objectives.

[0079] S503. When the user actually uses the electricity package, his electricity usage behavior will be collected in real time through feedback, and the effect of the recommended package will be automatically evaluated, and the subsequent recommendation strategy will be adjusted accordingly; wherein, the effect of the automatically evaluated recommended package is adaptively adjusted through a reinforcement learning model, and the recommendation process is optimized based on historical feedback data, and the objective function is:

[0080]

[0081] Among them, α, β and γ are dynamic adjustment coefficients, which are used to control the role of feedback results in the optimization process.

[0082] In another aspect of the present invention, an interactive system for power marketing is provided, the system comprising:

[0083] The power consumption data collection unit is used to collect the power consumption data of users through the Internet of Things devices, and pre-process the data to obtain the pre-processed power consumption data, and introduce the user's social behavior as the user behavior feature, and integrate the diversity of home devices and the changes in environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp;

[0084] The power demand prediction unit is used to construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes; wherein the power consumption behavior prediction model adopts a time series convolutional neural network, extracts high-order features from the time series data of power consumption through the convolution layer, and identifies long-term dependencies. During the training process, the convolution layer and the pooling layer work together to extract the potential rules of the user's power consumption, and the fully connected layer integrates the time series features with the user's behavior features and external environmental factors to predict power demand;

[0085] Wherein, the power consumption behavior prediction model is designed as a multi-task learning framework based on the use of the time series convolutional neural network, and two objectives are optimized respectively: power demand prediction and user behavior prediction, wherein the loss function of the power consumption behavior prediction model is designed as a weighted combination of the mean square error loss of power demand prediction and the cross entropy loss of user behavior classification;

[0086] The dynamic pricing unit is used to construct a dynamic power pricing model according to the prediction results, and adjust the power pricing according to the actual needs of users and the changes in the power grid load; wherein the dynamic power pricing model is constructed as follows:

[0087] Design a dynamic electricity pricing mechanism based on electricity demand forecasting and user behavior forecasting to flexibly adjust electricity supply according to real-time demand and price;

[0088] Design multiple constraints, including supply-demand balance constraints and fairness constraints;

[0089] Use a deep Q learning algorithm based on reinforcement learning to dynamically adjust the pricing strategy; wherein the deep Q learning algorithm is expressed as:

[0090] Q(s t , a t )←Q(s t , a t )+α[r t+1 +γmax a′ Q(s t+1 , a′)-Q(s t , a t )]

[0091] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual return, γ is the discount factor;

[0092] The power consumption data abnormal alarm unit is used to monitor the market situation with real-time data streams, and combines the trained power consumption behavior prediction model with the dynamic power pricing model to timely detect potential abnormalities. If abnormalities are found, intelligent early warning will be issued;

[0093] The personalized recommendation unit is used to combine historical data and real-time data of power marketing, recommend the best power package for each user through optimization algorithm, and make real-time adjustments to ensure that the most suitable power plan is provided to users under different conditions.

[0094] The beneficial technical effects of the present invention are at least as follows:

[0095] The present invention constructs a dynamic electricity consumption model by deeply analyzing the data collected by IoT devices such as smart meters, smart appliances, and sensors with the user's historical consumption behavior. By applying machine learning algorithms, the platform can analyze users' electricity consumption data in real time, accurately predict future electricity demand, and customize personalized electricity product recommendations for different users. For example, when it is predicted that a user will use a lot of electricity in the next few days, the system will push electricity price packages with time discounts in advance, or recommend energy-saving equipment to help users reduce electricity costs during peak periods. This effectively overcomes the limitations of static pricing and unified recommendations in traditional electricity marketing.

[0096] Combined with the user's electricity consumption time pattern, the present invention proposes a pricing mechanism based on time perception. The system can predict the user's future electricity demand based on the user's electricity consumption behavior in different time periods, and dynamically adjust the electricity price plan. For example, for users who use more electricity during the peak period in the evening, the system will push the night valley electricity price plan to encourage users to use more electricity during the valley period of electricity demand to achieve the purpose of balancing the load. This time-based pricing strategy enables power companies to flexibly adjust electricity prices to reduce fluctuations in the electricity market, while providing users with more attractive electricity packages and improving users' electricity experience.

[0097] The present invention also combines social recommendation technology to promote interaction between users and the promotion of power products through social platforms. Users can not only share their own power consumption data and energy-saving tips, but also choose power products based on the recommendations of their social circles. This social recommendation not only increases the interactivity and sense of participation of users, but also improves the credibility and marketing effect of power product recommendations based on the trust and evaluation mechanism between users. Through methods such as point systems and social challenges, the platform can also encourage users to share their energy-saving experiences, thereby promoting the popularization of energy-saving behaviors. This innovation effectively makes up for the shortcomings of traditional power marketing in terms of user participation and social recommendations, and improves the long-term participation of users and platform stickiness.

[0098] In summary, the present invention provides a new method for predicting and marketing electricity consumption behavior by combining intelligent IoT data, time-aware pricing, and social recommendations, solving the problems of insufficient personalized recommendations, untimely responses, and low user participation in the prior art. This technological innovation not only improves the accuracy and effectiveness of power companies' marketing, but also enhances users' interactivity and energy-saving awareness through social mechanisms, and promotes the transformation of the power industry towards intelligence, personalization, and socialization. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0100] Figure 1 The present invention is a flowchart of an interactive method for power marketing according to an embodiment of the present invention.

[0101] Figure 2 This is a framework diagram of an interactive system for power marketing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0102] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0103] like Figure 1 As shown, an interactive method for power marketing provided by an embodiment of the present invention includes the following steps:

[0104] S1. Collect the user's power consumption data through the Internet of Things devices, and pre-process the data to obtain the pre-processed power consumption data, and introduce the user's social behavior as the user behavior characteristics, and the diversity of home devices and changes in environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp.

[0105] Specifically, electricity consumption data is collected in real time through multiple data sources such as smart meters, environmental sensors, and equipment status sensors. Each data point consists of the following core variables:

[0106] Electricity consumption (E t ): The total amount of electricity consumed by the user at time point t, in kilowatt-hours (kWh).

[0107] Device Status (S t): Record the switch status of the user's home appliances (such as air conditioners, refrigerators, washing machines, etc.) at time point t, binary (0 means off, 1 means on).

[0108] Timestamp (t): The time point when the data was collected, usually in minutes or hours.

[0109] Furthermore, the collected data includes not only the power consumption, but also the factors affecting power consumption, such as the usage status of the appliance, environmental variables (such as temperature, humidity), etc. Through these data, the present invention can fully understand the user's power consumption behavior.

[0110] Furthermore, since missing values ​​and noise are inevitable in the data collection process, the present invention designs an adaptive interpolation method to handle missing data. Different from the common mean or linear interpolation methods, the present invention adopts interpolation based on the user's historical consumption pattern. The present invention assumes that the user's electricity consumption pattern in a specific time period has a certain regularity, so the missing values ​​can be predicted through historical data.

[0111] For example, if the consumption data at a certain time t is missing, the present invention will consider the weighted average of the most recent historical data points (t-1, t-2, ...):

[0112]

[0113] Among them, α i The specific calculation method takes into account the time difference and changes in user behavior patterns. The design of the weighting coefficient is not only based on the time proximity of historical data, but also combined with the device usage pattern. For example, the frequency of air conditioning use is higher at night, and the weighting coefficient will be adjusted accordingly.

[0114] Furthermore, since the power consumption of different power equipment varies greatly (such as the power difference between air conditioners and refrigerators), standardization is an important step to ensure data consistency and fairness. In order to avoid the disproportionate impact of high-power equipment on the model, the present invention standardizes the power consumption of various types of equipment.

[0115] A standardization method based on equipment category is adopted. In the standardization process, the present invention calculates the mean and standard deviation of each equipment based on its historical power consumption data, and then standardizes the power consumption data at the current moment according to the following formula:

[0116]

[0117] Among them, μ Et It is device E t The mean of historical electricity consumption data, σ Et is the standard deviation, This is the standardized power consumption data. This standardization method can remove the impact of power differences between devices, so that the subsequent behavior prediction model can handle the power requirements of different devices more fairly.

[0118] Furthermore, the power consumption data has obvious time series, that is, the power consumption pattern is significantly affected by time factors. In this step, the present invention processes the data at multiple time scales to capture the power consumption changes at different levels such as within the day, within the week, and on holidays.

[0119] Intraday session (D t ): Daily electricity consumption behavior is affected by morning peak hours, evening peak hours and other periods.

[0120] Weekly time period (W t ): Electricity usage patterns may differ on different weekdays and weekends.

[0121] Holidays (H t ): Holidays or special events will affect users' electricity usage patterns.

[0122] Furthermore, to handle these temporal changes, the present invention adopts a technique based on "nested weighted average of temporal features". This method takes into account the characteristics of each time period and adjusts the impact of historical data on the current moment through different weights:

[0123]

[0124] in, is the time-series power consumption data, w i is the weight adjusted according to the time period (such as working days, holidays, etc.), E t-i It is the consumption data for the historical period.

[0125] This approach can ensure that the temporal nature of electricity consumption is fully considered in the model, especially the ability to identify special situations such as intraday peaks, weekends and holidays.

[0126] Furthermore, in addition to standardization and timing processing, the present invention also designs the expansion of user behavior features. The introduction of these features can help the present invention more accurately capture the user's electricity usage patterns and provide additional contextual information for subsequent demand forecasting. These features include:

[0127] Diversity of household devices: the switching frequency and duration of different household devices (such as air conditioners, washing machines, etc.).

[0128] Changes in environmental factors: such as the impact of indoor temperature and humidity on the use of air conditioning and heating equipment.

[0129] User social behavior: For example, the electricity consumption behavior shared by users on social platforms may reflect certain specific electricity consumption patterns.

[0130] These features are fused with basic electricity consumption data to generate a comprehensive feature set to provide more comprehensive information.

[0131] S2. Construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes.

[0132] Specifically, in order to capture the time series characteristics of user power consumption behavior, the present invention designs a time series convolutional neural network (TCN), which extracts high-order features from the time series data of power consumption through the convolution layer and identifies long-term dependencies. In this process, the convolution layer and the pooling layer work together to extract the potential patterns of user power consumption. The fully connected layer further combines these time series features with user behavior features (B t ) and environmental factors (H t ) to forecast electricity demand.

[0133] Input data: B t , H t

[0134] Output prediction: (Users' future electricity demand)

[0135] The output of the network is the forecast of future electricity demand and prediction of user behavior related to electricity demand, which will be further processed in the subsequent multi-task learning stage.

[0136] Furthermore, this model adopts a multi-task learning framework to optimize two objectives: power demand prediction and user behavior prediction. The loss function is designed as a weighted combination of the mean square error (MSE) loss for power demand prediction and the cross entropy loss for user behavior classification.

[0137] For the power demand forecasting regression task, the loss function is:

[0138]

[0139] in, is the actual power demand, To forecast electricity demand.

[0140] For the user behavior classification task, the loss function is:

[0141]

[0142] Among them, y iis the actual label of the user behavior, is the predicted behavior probability.

[0143] The final loss function is:

[0144]

[0145] Among them, λ1 and λ2 are the weight coefficients between tasks.

[0146] Furthermore, in order to improve the generalization ability of the model and avoid overfitting, the present invention introduces an adaptive regularization mechanism. This mechanism dynamically adjusts the strength of the regularization term based on the changes in user behavior data to avoid imposing too strong constraints on users whose behavior patterns change greatly. In particular, the strength of the regularization term depends on the behavior feature B. t The volatility of is calculated using the following formula:

[0147]

[0148] in, is the L2 norm of user behavior features, α t is the regularization strength at the current moment. β1 and β2 are adjustment parameters. In this way, when user behavior patterns change dramatically, the model can adapt to these changes while avoiding over-regularization.

[0149] Furthermore, the present invention uses the Adam optimizer to train the model and combines the back propagation algorithm to minimize the loss function , and adopts a learning rate decay strategy to improve the stability of the training process. In order to prevent overfitting, the present invention combines cross-validation with early stopping technology to dynamically adjust the model's hyperparameters, including the convolution kernel size and the parameters of the pooling layer. Through multiple rounds of training, the model will gradually converge to the optimal state.

[0150] Furthermore, after training is completed, the model will output the user's future electricity demand forecast and the probability of user behavior changes In order to evaluate the performance of the model, the present invention uses two indicators, RMSE (root mean square error) and AUC (area under the curve), to evaluate the accuracy of power demand prediction and the effectiveness of user behavior prediction, respectively.

[0151] For power demand forecasting, the RMSE calculation formula is:

[0152]

[0153] For user behavior prediction, the AUC calculation formula is:

[0154]

[0155] Among them, TPR is the true positive rate and FPR is the false positive rate. The closer the AUC value is to 1, the better the model performs in the classification task.

[0156] Furthermore, the innovative design of this step fully combines the time series convolutional neural network (TCN) and the multi-task learning framework, effectively solving the multi-objective optimization problem of power demand forecasting and user behavior forecasting. In particular, the introduction of the adaptive regularization mechanism enables the model to adjust the regularization strength according to the dynamic changes of user behavior, avoid overfitting and improve the model's adaptability to new changes. This innovation effectively improves the prediction accuracy and generalization ability of the model, and meets the complex timing and behavior fluctuation problems in power consumption forecasting. Through the above steps, the model can not only accurately predict power demand, but also provide reliable predictions under complex user behavior and external environment changes, thereby providing strong support for power demand response and personalized power marketing strategies.

[0157] S3. Build a dynamic electricity pricing model based on the prediction results and adjust electricity pricing according to the actual needs of users and changes in grid load.

[0158] Specifically, in order to solve the problem of imbalance between supply and demand in the electricity market, the present invention designs a dynamic electricity pricing mechanism, which combines the predicted demand and behavioral characteristics of users and can flexibly adjust the electricity supply according to real-time demand and price. t , which consists of two parts:

[0159] Basic electricity price P base : This is base pricing based on current market electricity costs and long-term trend forecasts.

[0160] Dynamic adjustment factor ΔP t :Power demand forecast based on users and behavioral prediction Adjusted price volatility factor.

[0161] The specific dynamic electricity pricing formula is:

[0162] P t =P base +ΔP t

[0163] Among them, ΔP t It is the dynamic adjustment part based on demand and behavior prediction, and its calculation method is as follows:

[0164]

[0165] Among them, γ1 and γ2 are adjustment coefficients, E avg The average electricity demand during a period of time.

[0166] in, The predicted value of the user's future electricity demand comes from the prediction result of step 2. avg It is the average power demand of the entire user group and is used to determine whether the current demand exceeds the normal level. It is the user behavior prediction, which indicates the user's sensitivity or responsiveness to changes in electricity prices and is used to dynamically adjust pricing strategies. γ1 and γ2 are adjustment coefficients that control the impact of demand and behavior on pricing.

[0167] Furthermore, in order to ensure that electricity pricing not only takes into account user demand but also takes into account system stability and fairness, the model designs multiple constraints, including:

[0168] Supply and demand balance constraints: To avoid excessive fluctuations in electricity supply and demand, electricity pricing needs to maintain a balance between market supply and demand, namely:

[0169]

[0170] Among them, S t is the amount of electricity supplied at time t, ensuring that pricing matches demand.

[0171] Fairness constraint: In order to ensure fairness among users, the price fluctuation range is set to no more than a certain percentage:

[0172] ΔP t ≤α·P base

[0173] Among them, α is the preset maximum price fluctuation ratio, which is used to prevent excessive price fluctuations from affecting low-income users.

[0174] Furthermore, this model uses an optimization algorithm based on reinforcement learning to dynamically adjust the pricing strategy. Specifically, a time series-based deep Q learning (DQN) algorithm is designed, which can optimize the electricity pricing strategy under the constraints of electricity supply and demand and user behavior. Through this algorithm, the model can adjust the electricity price in real time at each time step to ensure that users maximize their utility while optimizing the overall benefits of the power system. Q value function update:

[0175]

[0176] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual benefit (such as the matching degree of electricity demand), and γ is the discount factor.

[0177] Furthermore, during the training process, the model continuously adjusts the adjustment coefficients such as γ1 and γ2 through the reinforcement learning algorithm based on historical power demand data, user behavior data, and supply and demand balance constraints to achieve the goal of optimizing power pricing. Each time it is updated, the model will evaluate the impact of power price adjustments on power market stability, user satisfaction, etc., and adjust the strategy in real time to ensure that the model converges to the optimal pricing solution.

[0178] Furthermore, in order to evaluate the effectiveness of the electricity pricing model, two types of evaluation indicators are used:

[0179] Supply and demand matching: Calculate the difference between electricity demand and supply at each moment:

[0180]

[0181] User satisfaction: Based on the user's response behavior, calculate the total amount of consumption under the adjusted pricing and analyze their acceptance of the pricing strategy.

[0182] Through comprehensive evaluation of these indicators, the model will adjust the pricing strategy based on feedback to make it more in line with actual needs.

[0183] Furthermore, the innovative design of this step combines the dynamic electricity pricing mechanism with reinforcement learning to solve the pricing challenges brought by the uncertainty of electricity demand forecasting and user behavior. In particular, the adaptive pricing factor ΔP is adopted. t , can dynamically adjust electricity prices according to real-time electricity demand and user behavior, avoiding the delayed response to demand fluctuations in traditional pricing methods. In addition, the design of constraints and optimization objectives ensures that the pricing mechanism can effectively balance market supply and demand while taking into account system fairness.

[0184] This dynamic electricity pricing model can not only accurately match user demand, but also improve the overall economic benefits of the power system, reduce the risk of power supply surplus or shortage, and provide power suppliers with flexible and adjustable pricing strategies.

[0185] S4. Real-time data stream monitoring of market conditions, combined with the trained electricity consumption behavior prediction model and dynamic electricity pricing model, can promptly detect potential anomalies and provide intelligent warnings if anomalies are found.

[0186] Specifically, in the actual application of the electricity pricing model, the real-time monitoring and intelligent early warning system plays a vital role, especially in the face of uncertainties such as fluctuations in electricity demand and supply, which can effectively control and respond to risks. This step aims to monitor market conditions through real-time data streams, combine the trained electricity consumption behavior prediction model with the dynamic pricing model, and promptly detect potential anomalies (such as a surge in electricity demand, insufficient supply, and sharp fluctuations in market prices, etc.), and provide intelligent early warnings.

[0187] Furthermore, at the beginning of this step, the system first obtains important operating data from various power users and the power grid through a real-time data interface, including but not limited to:

[0188] Real-time electricity demand (E t ): Indicates the current electricity consumption, including the electricity consumption at the user end (households, industries, etc.).

[0189] Power supply (S t ): Represents the current supply capacity of the power grid, including the power generation of power plants and the available capacity of power transmission.

[0190] Electricity market price (P t ): reflects the current price fluctuations in the electricity market and makes dynamic adjustments based on supply and demand.

[0191] Historical user behavior data (U t ): Includes cyclical changes in electricity consumption, seasonal fluctuations, etc., used in conjunction with demand forecasts.

[0192] After the real-time data stream is cleaned and standardized, the system inputs the data into subsequent models to ensure that the models can make intelligent decisions based on the latest information.

[0193] Furthermore, based on the output of the power consumption behavior prediction model established in the previous step, the present invention needs to compare the current power demand and supply to identify potential anomalies. The system identifies anomalies by calculating the deviation between the current actual demand and the predicted demand:

[0194]

[0195] Among them, E t represents the actual power demand, It is the expected demand based on historical data and forecasting models. t If the demand exceeds the set threshold (such as a multiple of the standard deviation), the system will determine that the demand is abnormal.

[0196] Likewise, monitoring of the power supply is critical. t ) and the supply capacity predicted by the model ( ) is too different, the system will trigger a supply anomaly alarm:

[0197]

[0198] By calculating these deviations in real time, the system can promptly detect deviations in electricity supply and demand and initiate an early warning mechanism.

[0199] It can be understood that when the deviation value ΔE t or ΔS t When the preset threshold is exceeded, the system will automatically trigger an intelligent warning. In order to more accurately evaluate the warning triggering conditions, a weighted threshold parameter can be introduced to take into account the sensitivity and market volatility of different time periods. For example, the system can dynamically adjust the threshold δ according to the volatility of historical data. E and δ S , in order to improve the accuracy of early warning.

[0200] When an alert is generated, the system logic is as follows:

[0201] Abnormal power demand warning: When |ΔE t |>δ E When , it indicates that the electricity demand deviates significantly from expectations and the system triggers a demand warning.

[0202] Power supply abnormality warning: When |ΔS t |>δ S , it indicates that the power supply is significantly lower than expected and the system triggers a supply warning.

[0203] Among them, δ E and δ S It is an adaptive threshold learned based on historical data and can be dynamically adjusted according to different situations.

[0204] Furthermore, in order to cope with changes in electricity supply and demand, the real-time monitoring system will dynamically adjust the pricing of the electricity market. Based on real-time feedback, the system will optimize the pricing strategy according to electricity demand and supply, and adjust the market electricity price (P t ), to achieve a balance between supply and demand. At this time, the optimization goal not only includes minimizing the error, but also needs to comprehensively consider the impact of price fluctuations and design a dynamic pricing model to adjust the electricity price.

[0205] The present invention handles the tasks of electricity price adjustment and supply-demand balance through weighted optimization of the objective function. The objective function is in the following form:

[0206]

[0207] Among them, E t and S t are electricity demand and supply, P t is the market electricity price, is the electricity price predicted by the model, and λ is the weight factor that balances supply and demand with electricity price fluctuations. The optimization goal is to minimize the difference between actual electricity demand and supply and limit electricity price fluctuations to an acceptable range.

[0208] Furthermore, after the early warning system issues an alarm, the feedback mechanism plays a vital role. The system responds quickly based on real-time data and makes real-time adjustments to the price, supply, and demand of the power market. After each feedback, the predicted values ​​of power demand and supply are and The model is dynamically updated based on new data. This real-time feedback mechanism ensures that the model can adapt to the ever-changing market environment and avoid imbalances in electricity supply and demand or excessive market fluctuations.

[0209] Through this closed-loop adjustment mechanism, electricity market pricing can respond more flexibly to emergencies and abnormal fluctuations, ensuring stable operation of the system.

[0210] This step is closely integrated with the aforementioned forecasting model, the former providing accurate demand forecasts, while the latter optimizing electricity pricing in real time based on changes in demand and supply. This dynamic adjustment mechanism is the core of intelligent dispatching in the power market, and effectively solves the limitations of traditional power pricing models in real-time response.

[0211] S5. Combining historical data and real-time data of electricity marketing, we recommend the best electricity package for each user through optimization algorithm and make real-time adjustments to ensure that the most suitable electricity plan is provided to users under different conditions.

[0212] Specifically, personalized electricity package recommendation and real-time optimization are crucial parts of the system. By combining the electricity demand forecast, market electricity price fluctuations and user behavior patterns in the previous steps, this step recommends the best electricity package for each user through an optimization algorithm and makes real-time adjustments to ensure that the most suitable electricity plan is provided to users under different conditions. The core goal of this step is to optimize the allocation of power resources, reduce users' electricity costs, and improve the efficiency and response speed of the power grid.

[0213] Furthermore, the input of this step comes from the output of the previous step, especially the electricity demand forecast from the dynamic electricity pricing model ( ), real-time power demand (E t ), electricity market price (P t ) and the user's electricity usage behavior pattern (A t ). Specifically include:

[0214] User power consumption history (U t ): User historical electricity consumption data, reflecting the periodicity and seasonality of user electricity consumption.

[0215] Real-time electricity demand (E t ): User’s current electricity demand data collected from the monitoring system.

[0216] Electricity market price (Pt ): Real-time electricity price fluctuation data obtained from the market.

[0217] Electricity package price forecast (P rec (t)): The recommended electricity price given by the system based on real-time demand and electricity price forecast results.

[0218] User behavior pattern analysis (A t ): By analyzing the user's historical electricity consumption data, the user's behavior pattern (such as whether they tend to consume during peak or off-peak periods) is extracted.

[0219] Furthermore, in order to optimize the recommendation of electricity packages, the goal is to design an optimal electricity package recommendation model based on the user's electricity demand, market electricity price fluctuations and behavior patterns. To this end, the present invention first quantifies the effect of the recommended package by optimizing the following objective function:

[0220]

[0221] Among them, E t is the actual power demand of the user in time period t, is the user power demand forecast based on the forecast model, P rec (t) is the recommended electricity package price (recommended by the system based on real-time demand and electricity price fluctuations), P t is the actual electricity market price, A i (t) is the user's behavior pattern in time period t (such as the tendency to use electricity during peak or off-peak periods), A rec (t) is the recommended user behavior pattern, and γ and λ are adjustment coefficients used to control the balance between electricity price fluctuations and behavior patterns. Through the above optimization function, the present invention makes the recommended electricity package as close as possible to the actual needs of the user, while taking into account electricity price fluctuations and user behavior preferences, thereby achieving personalized recommendations.

[0222] Furthermore, after the user accepts the recommended package, the system still needs to dynamically adjust the electricity package to cope with the real-time changes in electricity demand, market electricity prices and user behavior patterns. At this time, the present invention optimizes the electricity package through real-time monitoring and prediction models, with the goal of making the recommended package more adaptable.

[0223] Taking into account the fluctuation of electricity prices and changes in user demand, the present invention introduces a dynamically adjusted objective function:

[0224]

[0225] Among them, U t is the historical data of the user’s actual electricity consumption behavior in time period t, U recThe system recommends the user's electricity usage history (i.e., the recommended electricity usage pattern) based on the prediction results. δ, μ, and θ are dynamic adjustment coefficients used to balance different optimization objectives. The additional term θ in this objective function introduces the influence of the user's historical electricity usage data, aiming to adjust the recommended plan to better suit the user's electricity usage habits. Through this adjustment, the system can respond to market changes and user demand fluctuations more accurately, ensuring that users always have the most suitable electricity package at different time periods.

[0226] Furthermore, in order to ensure the continuous optimization and self-adaptation of the electricity package recommendation system, the present invention adds an intelligent feedback mechanism. When users actually use electricity packages, their electricity usage behavior will be collected in real time through the feedback system. The system will automatically evaluate the effect of the recommended package and adjust the subsequent recommendation strategy accordingly.

[0227] This feedback mechanism is adaptively adjusted through a reinforcement learning model to optimize the recommendation process based on historical feedback data. The optimization objective function is:

[0228]

[0229] Among them, α, β and γ are dynamic adjustment coefficients, which are used to control the role of feedback results in the optimization process.

[0230] The core of this feedback mechanism is to fine-tune the electricity package through real-time feedback from users, thereby achieving continuous optimization of personalized packages and the system's adaptive capabilities.

[0231] Furthermore, in order to enable users to better understand and manage their electricity packages, the system provides an easy-to-interact control interface. Users can view their electricity packages, real-time electricity demand, current electricity prices, and estimated costs through the interface. If users think the existing packages are not suitable, they can also manually adjust their electricity package settings. User behavior feedback will directly affect the system's optimization process, helping the system to more accurately recommend suitable electricity packages.

[0232] Through personalized recommendations and real-time optimization, it provides users with economical and flexible electricity consumption solutions. For power suppliers, these optimization measures can effectively reduce power waste and improve user satisfaction, while helping power companies achieve more refined power management.

[0233] like Figure 2 As shown, in another embodiment of the present invention, an interactive system for power marketing is provided, the system comprising:

[0234] The power consumption data collection unit 301 is used to collect the power consumption data of the user through the Internet of Things device, and pre-process the data to obtain the pre-processed power consumption data, and introduce the user's social behavior as the user behavior feature, and integrate the diversity of home devices and the changes of environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp;

[0235] The power demand prediction unit 302 is used to construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes; wherein the power consumption behavior prediction model adopts a time series convolutional neural network, extracts high-order features from the time series data of power consumption through the convolution layer, and identifies long-term dependencies. During the training process, the convolution layer and the pooling layer work together to extract the potential rules of the user's power consumption, and the fully connected layer integrates the time series features with the user's behavior features and external environmental factors to predict power demand;

[0236] Wherein, the power consumption behavior prediction model is designed as a multi-task learning framework based on the use of the time series convolutional neural network, and two objectives are optimized respectively: power demand prediction and user behavior prediction, wherein the loss function of the power consumption behavior prediction model is designed as a weighted combination of the mean square error loss of power demand prediction and the cross entropy loss of user behavior classification;

[0237] The dynamic pricing unit 303 is used to construct a dynamic power pricing model according to the prediction results, and adjust the power pricing according to the actual needs of users and the changes in grid load; wherein the dynamic power pricing model is constructed as follows:

[0238] Design a dynamic electricity pricing mechanism based on electricity demand forecasting and user behavior forecasting to flexibly adjust electricity supply according to real-time demand and price;

[0239] Design multiple constraints, including supply-demand balance constraints and fairness constraints;

[0240] Use a deep Q learning algorithm based on reinforcement learning to dynamically adjust the pricing strategy; wherein the deep Q learning algorithm is expressed as:

[0241]

[0242] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual return, γ is the discount factor;

[0243] The power consumption data abnormality alarm unit 304 is used to monitor the market situation with real-time data streams, and timely detect potential abnormalities by combining the trained power consumption behavior prediction model with the dynamic power pricing model. If an abnormality is found, an intelligent early warning is issued;

[0244] The personalized recommendation unit 305 is used to combine the historical data and real-time data of power marketing, recommend the best power package for each user through the optimization algorithm, and make real-time adjustments to ensure that the most suitable power plan is provided to the user under different conditions.

[0245] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0246] In addition, for technical details that are not described in detail in this embodiment, reference can be made to the parameter operation method provided in any embodiment of the present invention, and will not be repeated here.

[0247] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0248] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0249] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0250] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An interactive method for power marketing, characterized in that: The method comprises: S1. Collecting the user's power consumption data through the Internet of Things device, and preprocessing the data to obtain the preprocessed power consumption data, and introducing the user's social behavior as the user behavior feature, and the diversity of home devices and the changes in environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp; S2. Construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes; wherein the power consumption behavior prediction model adopts a time series convolutional neural network, extracts high-order features from the time series data of power consumption through the convolution layer, and identifies long-term dependencies. During the training process, the convolution layer and the pooling layer work together to extract the potential rules of user power consumption, and the fully connected layer integrates the time series features with the user behavior features and external environmental factors to predict power demand; Wherein, the power consumption behavior prediction model is designed as a multi-task learning framework based on the use of the time series convolutional neural network, and two objectives are optimized respectively: power demand prediction and user behavior prediction, wherein the loss function of the power consumption behavior prediction model is designed as a weighted combination of the mean square error loss of power demand prediction and the cross entropy loss of user behavior classification; S3. A dynamic electricity pricing model is constructed based on the prediction results, and electricity pricing is adjusted according to the actual needs of users and changes in grid load; wherein the dynamic electricity pricing model is constructed as follows: Design a dynamic electricity pricing mechanism based on electricity demand forecasting and user behavior forecasting to flexibly adjust electricity supply according to real-time demand and price; Design multiple constraints, including supply-demand balance constraints and fairness constraints; Use a deep Q learning algorithm based on reinforcement learning to dynamically adjust the pricing strategy; wherein the deep Q learning algorithm is expressed as: Q(s t ,a t )←Q(s t ,a t )+α[r t+1 +γmax a′ Q(s t+1 ,a′)-Q(s t ,a t )] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual return, γ is the discount factor; S4, real-time data stream monitoring of market conditions, combined with the trained power consumption behavior prediction model and dynamic power pricing model, timely detect potential anomalies, and issue intelligent warnings if anomalies are found; S5. Combining historical data and real-time data of electricity marketing, we recommend the best electricity package for each user through optimization algorithm and make real-time adjustments to ensure that the most suitable electricity plan is provided to users under different conditions.

2. The interactive method for power marketing according to claim 1, characterized in that: The pre-processing comprises: Interpolation based on historical consumption patterns of users: Assuming that the electricity consumption patterns of users in a specific time period have certain regularity, the missing values ​​are predicted through historical data; A standardization method based on equipment category is adopted: During the standardization process, the mean and standard deviation of each equipment are calculated based on its historical power consumption data, and then the power consumption data at the current moment is standardized according to the following formula; Multi-time scale processing: For intraday time periods, intraweek time periods and holidays, an algorithm based on time series feature nested weighted average is used to adjust the impact of historical data on the current moment through different weights to obtain pre-processed power consumption data; Data feature extension is introduced to collect user social behaviors as user behavior features, and the diversity of home devices and changes in environmental factors as external environmental factors.

3. The interactive method for power marketing according to claim 1, characterized in that: For the power demand forecasting regression task, the loss function is: in, is the actual power demand, To forecast electricity demand; For the user behavior classification task, the loss function is: Among them, y i is the actual label of the user behavior, is the predicted behavior probability; Finally, the loss function of the electricity consumption behavior prediction model is for: Among them, λ1 and λ2 are the weight coefficients between tasks.

4. The interactive method for power marketing according to claim 3, characterized in that: In order to improve the generalization ability of the power consumption behavior prediction model and avoid overfitting, an adaptive regularization mechanism is introduced. The adaptive regularization mechanism dynamically adjusts the strength of the regularization term based on the changes in user behavior data to avoid imposing the greatest constraints on those users whose behavior patterns change the most. It is expressed as: in, is the L2 norm of user behavior features, α t is the regularization strength at the current moment. β1 and β2 are adjustment parameters; The Adam optimizer is used to train the model and the back propagation algorithm is used to minimize the loss function. And adopt the learning rate decay strategy to improve the stability of the training process; After training, the model will output the user's future electricity demand forecast and the probability of user behavior change, and then use the RMSE and AUC indicators to evaluate the accuracy of electricity demand forecast and the effectiveness of user behavior forecast, respectively. For power demand forecasting, the RMSE calculation formula is: in, Forecast future electricity demand for users, is the actual power demand of the user, N is the total number of samples, and i is the current user i; For user behavior prediction, the AUC calculation formula is: Among them, TPR is the true positive rate, FPR is the false positive rate; the AUC value is closest to 1, indicating that the model performs best in the classification task.

5. The interactive method for power marketing according to claim 1, characterized in that: The dynamic electricity pricing mechanism includes a basic electricity price and a dynamic adjustment factor. The basic electricity price is a basic price based on the current market electricity cost and long-term trend forecast. The dynamic adjustment factor is a price fluctuation factor adjusted based on the electricity demand forecast and user behavior forecast. The dynamic electricity pricing mechanism is expressed as: P t =P base +ΔP t Among them, P t is the market electricity price, P base is the basic electricity price, ΔP t It is the dynamic adjustment part based on the power demand forecast and user behavior forecast, and its calculation method is as follows: Among them, γ1 and γ2 are adjustment coefficients, For user behavior prediction, E avg is the average electricity demand during a certain period of time; The supply-demand balance constraint is used to maintain the market supply-demand balance and is expressed as: Among them, S t is the electricity supply at time t, ensuring that pricing matches demand; T is the total duration; The fairness constraint is used to set the price fluctuation range to not exceed a certain ratio, which is expressed as: ΔP t ≤α·P base Among them, α is the preset maximum price fluctuation ratio, which is used to prevent excessive price fluctuations from affecting low-income users.

6. The interactive method for power marketing according to claim 5, characterized in that: The S3 further includes: Two types of evaluation indicators are used to evaluate the effect of the dynamic electricity pricing model, and the evaluation indicators include supply-demand matching and user satisfaction; the supply-demand matching Mismatch t The difference between electricity demand and supply at each moment is expressed as: The user satisfaction is based on the user's response behavior, calculates the user's total consumption under the adjusted pricing, and analyzes their acceptance of the pricing strategy.

7. The interactive method for power marketing according to claim 1, characterized in that: The S4 specifically includes: S401, obtaining operation data from each power user and the power grid through a real-time data interface; wherein the operation data includes real-time power demand, power supply, power market price and historical user behavior data; S402, monitor and compare current power demand and supply to identify potential anomalies; S403, based on real-time feedback, set optimization target optimization pricing strategy according to power demand and supply situation, adjust market electricity price, and achieve supply and demand balance; wherein the objective function of the optimization target is expressed as: Among them, E t and S t are electricity demand and supply, P t is the market electricity price, is the electricity price predicted by the model, and λ is the weight factor that balances supply and demand with electricity price fluctuations.

8. The interactive method for power marketing according to claim 7, characterized in that: The S402 specifically includes: Anomalies are identified by calculating the deviation between the current actual demand and the predicted demand, expressed as: Among them, E t represents the actual power demand, It is the expected demand based on historical data and forecasting models. t If the set threshold is exceeded, it is considered as abnormal demand; Similarly, if the supply (S t ) and the supply capacity predicted by the model If the difference exceeds the threshold, a provisioning anomaly alert will be triggered: When the deviation value ΔE t or ΔS t When the preset threshold is exceeded, an intelligent warning is automatically triggered.

9. The interactive method for power marketing according to claim 1, characterized in that: The S5 specifically includes: S501. Design a personalized electricity package recommendation model, expressed as: Among them, E t is the actual power demand of the user in time period t, is the user power demand forecast based on the forecast model, P rec (t) is the recommended electricity package price, P t is the actual electricity market price, A i (t) is the user’s behavior pattern in time period t, A rec (t) is the recommended user behavior pattern, γ and λ are adjustment coefficients used to control the balance between electricity price fluctuations and behavior patterns; S502. After the user accepts the recommended package, the power consumption behavior prediction model is adjusted to optimize the power package by real-time monitoring of user feedback, so that the recommended package is more adaptable; wherein the objective function of adjusting the power consumption behavior prediction model to optimize the power package is expressed as: Among them, U t is the historical data of the user’s actual electricity consumption behavior in time period t, U rec The user's electricity consumption history recommended by the system based on the prediction results. δ, μ and θ are dynamic adjustment coefficients used to balance different optimization objectives. S503. When the user actually uses the electricity package, his electricity usage behavior will be collected in real time through feedback, and the effect of the recommended package will be automatically evaluated, and the subsequent recommendation strategy will be adjusted accordingly; wherein, the effect of the automatically evaluated recommended package is adaptively adjusted through a reinforcement learning model, and the recommendation process is optimized based on historical feedback data, and the objective function is: Among them, α, β and γ are dynamic adjustment coefficients, which are used to control the role of feedback results in the optimization process.

10. An interactive system for power marketing, characterized in that: The system comprises: The power consumption data collection unit is used to collect the power consumption data of users through the Internet of Things devices, and pre-process the data to obtain the pre-processed power consumption data, and introduce the user's social behavior as the user behavior feature, and integrate the diversity of home devices and the changes in environmental factors as external environmental factors; wherein the power consumption data includes power consumption, device status and timestamp; The power demand prediction unit is used to construct a power consumption behavior prediction model based on the preprocessed data to predict the user's future power demand and respond to sudden changes; wherein the power consumption behavior prediction model adopts a time series convolutional neural network, extracts high-order features from the time series data of power consumption through the convolution layer, and identifies long-term dependencies. During the training process, the convolution layer and the pooling layer work together to extract the potential rules of the user's power consumption, and the fully connected layer integrates the time series features with the user's behavior features and external environmental factors to predict power demand; Wherein, the power consumption behavior prediction model is designed as a multi-task learning framework based on the use of the time series convolutional neural network, and two objectives are optimized respectively: power demand prediction and user behavior prediction, wherein the loss function of the power consumption behavior prediction model is designed as a weighted combination of the mean square error loss of power demand prediction and the cross entropy loss of user behavior classification; The dynamic pricing unit is used to construct a dynamic power pricing model according to the prediction results, and adjust the power pricing according to the actual needs of users and the changes in the power grid load; wherein the dynamic power pricing model is constructed as follows: Design a dynamic electricity pricing mechanism based on electricity demand forecasting and user behavior forecasting to flexibly adjust electricity supply according to real-time demand and price; Design multiple constraints, including supply-demand balance constraints and fairness constraints; Use a deep Q learning algorithm based on reinforcement learning to dynamically adjust the pricing strategy; wherein the deep Q learning algorithm is expressed as: Q(s t ,a t )←Q(s t ,a t )+α[r t+1 +γmax a′ Q(s t+1 ,a′)-Q(s t ,a t )] Among them, s t is the current state, a t is the selected pricing strategy, r t+1 is the actual return, γ is the discount factor; The power consumption data abnormal alarm unit is used to monitor the market situation with real-time data streams, and combines the trained power consumption behavior prediction model with the dynamic power pricing model to timely detect potential abnormalities. If abnormalities are found, intelligent early warning will be issued; The personalized recommendation unit is used to combine historical data and real-time data of power marketing, recommend the best power package for each user through optimization algorithm, and make real-time adjustments to ensure that the most suitable power plan is provided to users under different conditions.

Images