Method for determining electricity selling scheme of electricity seller

Abstract

The invention discloses a method for determining an electricity selling scheme of an electricity seller, and particularly relates to the technical field of electricity market analysis, and the method comprises the following steps: predicting the future electricity demand of a user through obtaining the electricity data of the user and employing a time sequence prediction model; within a preset time period, generating a period comprehensive evaluation value according to electricity market price fluctuation, user demand change and response capability, and judging whether deep analysis is needed or not; if deep analysis is needed, market fluctuation and demand change feature analysis is carried out respectively, a price fluctuation violent index and a demand variation degree index are calculated, and a current electricity selling time period is divided into a high-risk time period and a low-risk time period; the low-risk period continues to use the existing scheme, and the high-risk period optimizes the electricity selling scheme through a multi-objective optimization model and confirms the application; according to the invention, the situation that the electricity purchasing cost is greatly increased due to the violent fluctuation of the market price is avoided, and the dynamic balance of the maximization of the electricity selling income and the minimization of the electricity purchasing cost is realized on the premise of meeting the user requirements.

Description

Technical Field

[0001] The present invention relates to the technical field of power market analysis, and more specifically, to a method for determining an electricity sales plan for an electricity seller. Background Art

[0002] With the rapid development of the global energy market, the power market is gradually transforming from the traditional centralized supply mode to the market-oriented trading mode. In this context, as a key participant in the power market, the main task of the electricity seller is to provide electricity services to users at a reasonable price, while optimizing its own electricity sales strategy to maximize profits. However, the current formulation of electricity sales plans usually relies on historical data and rule adjustments at fixed intervals, making it difficult to respond in real time to power market price fluctuations and rapid changes in user demand, and there are the following technical deficiencies: Existing electricity sales plans lack the dynamic analysis and rapid response capabilities for real-time power market price fluctuations, which easily lead to an increase in electricity purchase costs or a decline in market competitiveness. User demand is highly uncertain due to various factors such as weather, time, and activities. Traditional electricity sales plans cannot promptly capture the dynamic changes in user demand, which may result in a decrease in electricity sales revenue. At the same time, the fixed price charging may also affect user satisfaction. Therefore, the present invention proposes a method for determining an electricity sales plan for an electricity seller in order to solve the above problems. Summary of the Invention

[0003] To achieve the above object, the present invention provides the following technical solutions: A method for determining an electricity sales plan for an electricity seller, comprising the following steps: Obtain user electricity consumption data, and perform predictive analysis through a preset time series prediction model to obtain a prediction result, that is, the future electricity demand of users; Within a preset prediction time period, generate a periodic comprehensive evaluation value based on the real-time price fluctuations, user demand changes, and response capabilities of the power market, and determine whether to perform in-depth analysis according to the periodic comprehensive evaluation value; When performing in-depth analysis, perform an analysis of the power market volatility characteristics and an analysis of the user demand change characteristics respectively to obtain a price volatility severity index and a demand variation degree index, and then divide the current electricity sales period into high-risk periods and low-risk periods; For low-risk periods, use the previously determined electricity sales plan. For high-risk periods, construct a multi-objective optimization model to optimize the electricity sales plan, and use the optimization result of the multi-objective optimization model as the electricity sales plan for this time for confirmation and application.

[0004] In a preferred embodiment, the preset time series prediction model is an autoregressive moving average model or a long short-term memory network.

[0005] In a preferred embodiment, the logic for obtaining the periodic comprehensive evaluation value is as follows: Within a preset prediction time period, obtain the average value of the price fluctuation amplitude within the period , the average deviation between the actual demand and the predicted demand , and the price elasticity coefficient within the power sales area of the power seller , the price elasticity coefficient The calculation formula is: ; represents the average value of the price change amplitude of all historical time periods used in the training of the time series prediction model, represents the average change in user electricity consumption of all historical time periods used in the training of the time series prediction model, is a constant set to prevent the denominator from being zero; The calculation formula for the periodic comprehensive evaluation value is: ; , , are all preset non-zero proportionality coefficients, represents the periodic comprehensive evaluation value.

[0006] In a preferred embodiment, determining whether to perform in-depth analysis based on the periodic comprehensive evaluation value means: Compare the periodic comprehensive evaluation value with a preset standard evaluation value range. If the periodic comprehensive evaluation value falls within the preset standard evaluation value range, no in-depth analysis is performed. If the periodic comprehensive evaluation value does not fall within the preset standard evaluation value range, in-depth analysis is performed.

[0007] In a preferred embodiment, the logic for obtaining the price volatility index is as follows: Calculate the overall amplitude impact coefficient: ; represents the price change amount between adjacent sampling points t and t-1, and N represents the number of price samples within the current time period, represents the overall amplitude impact coefficient; Calculate the extreme case impact coefficient: ; represents the average value of the current time period price, represents the electricity market price at the current sampling point t, represents the maximum value of the price change amount between adjacent sampling points t and t-1, represents the extreme case impact coefficient; Calculate the dispersion impact coefficient: ; represents the standard deviation of the current time period price, reflecting the dispersion of the price, Indicates the coefficient of influence of the degree of dispersion; The calculation formula for the price volatility index is: ; , , are all preset non-zero adjustment coefficients, Indicates the price volatility index.

[0008] In a preferred embodiment, the acquisition logic for the demand mutation degree index is as follows: First, calculate the demand deviation score, and the formula is: ; Indicates the current demand power at user time point k and the predicted value The difference, is a preset constant, Indicates the standard deviation of the demand deviation in the current period, M represents the total number of time points in the current period, Indicates the demand deviation score; Then define the extreme deviation threshold as , is a preset standard constant, and count all The total number greater than the extreme deviation threshold is marked as ; Then analyze the regularity of the sequence through the change amount of adjacent demand deviations to obtain the regularity score: ; Indicates the regularity score; If , then the change direction of the current time point k and the next time point k + 1 is the same. Count all The total number, and then divide by , to obtain the direction persistence ratio value and mark it as ; Combine , , , Summarize into a demand mutation vector, and then calculate the Euclidean distance between the demand mutation vector and the preset standard mutation vector to obtain the demand mutation degree index.

[0009] In a preferred embodiment, dividing the current power selling period into high-risk periods and low-risk periods means: The demand anomaly index and price fluctuation index of the current electricity selling period are taken as input variables of fuzzy reasoning, and the division type of the current electricity selling period is taken as the output variable. The input variables are fuzzified and the values ​​of the input variables are converted into fuzzy sets. The output variables are fuzzified and the output variables are converted into fuzzy sets. Fuzzy rules are formulated to describe the adaptability of the current rules and each division type under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to obtain the division type of the current electricity selling period.

[0010] In a preferred embodiment, the multi-objective optimization model refers to constructing an optimization model with dual objectives of electricity sales price and user satisfaction by using a genetic algorithm. The fitness function of the genetic algorithm comprehensively considers the electricity sales revenue, electricity purchase cost and user satisfaction. ; represents the electricity selling price, Indicates the total electricity consumption of the user during the current electricity sales period. It represents the cost of purchasing electricity from the electricity retailer market. Indicates user satisfaction, , Both are preset optimization coefficients, and the sum of the two is one; ; Indicates the average value of the user's historical electricity charges. , , It is the preset tolerance coefficient, which indicates the price fluctuation ratio accepted by the user; The electricity selling price corresponding to the maximum value of the final fitness function is used as the electricity selling plan for the current electricity selling period.

[0011] Technical effects and advantages of the present invention: The present invention achieves a dynamic balance between maximizing electricity sales revenue and minimizing electricity purchase costs while meeting user needs through a multi-objective optimization model. The electricity sales price strategy determined by the intelligent algorithm avoids the situation where the electricity purchase cost increases significantly due to drastic market price fluctuations. The high-risk and low-risk time periods are dynamically divided, and resources are concentrated in the high-risk time periods for optimization. The existing strategy is used for the low-risk time periods, which significantly improves the utilization efficiency of computing resources.

[0012] Through quantitative analysis of the demand variation index, the present invention can accurately identify the changing trend and abnormal fluctuation of user demand, and improve the responsiveness of the power selling solution to user demand. The present invention comprehensively considers the price acceptance range and responsiveness of users, dynamically adjusts the power selling price through fuzzy reasoning, ensures that the power selling solution achieves the optimal balance between revenue and user satisfaction, and further enhances user recognition of the power selling solution services.

[0013] Through the calculation of the price volatility index, the present invention comprehensively quantifies the volatility amplitude, extreme value impact, and dispersion of electricity market prices, helping electricity sellers identify market risks and quickly adjust strategies. By comprehensively analyzing market fluctuations and user demands through a fuzzy inference method, the electricity selling periods are divided into high-risk and low-risk categories, providing a scientific basis for adjusting electricity selling plans under different risk states.

[0014] The present invention adopts a genetic algorithm. Through the design of a fitness function, it comprehensively balances revenue, cost, and user satisfaction, enabling electricity sellers to quickly find the optimal electricity selling plan in a complex market environment. The method for formulating the electricity selling plan of the present invention realizes the full-process automation from data analysis to strategy optimization, greatly reducing the complexity and time cost of manual decision-making, and improving the decision-making efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings; Figure 1 It is a schematic diagram of a method for determining an electricity selling plan for an electricity seller in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Refer to Figure 1 The following embodiments are obtained: Embodiment 1: A method for determining an electricity selling plan for an electricity seller, including the following steps: Obtain user electricity consumption data, and perform predictive analysis through a preset time series prediction model to obtain the prediction result, that is, the future electricity demand of the user; by obtaining user electricity consumption data, the electricity consumption habits, peak-valley distribution, and fluctuation characteristics of the user can be identified, providing support for the personalized customization of the electricity selling plan.

[0018] Within a preset prediction time period, based on the real-time price fluctuations, user demand changes, and response capabilities of the electricity market, generate a periodic comprehensive evaluation value, and decide whether to perform in-depth analysis according to the periodic comprehensive evaluation value; based on a time period (such as an hour or a day), comprehensively analyze the market price, demand changes, and user response capabilities to capture the overall market state. The periodic comprehensive evaluation value serves as a screening mechanism to avoid complex calculations for all time periods, saving computing resources. Through real-time evaluation, abnormal situations can be quickly identified, laying a foundation for the next in-depth analysis.

[0019] When conducting in-depth analysis, the volatility characteristics of the electricity market and the changing characteristics of user demand are analyzed separately to obtain the price volatility intensity index and the demand variation degree index, and then the current electricity selling period is divided into high-risk periods and low-risk periods; by analyzing the amplitude, frequency, and extreme changes of market price fluctuations, the instability of the current market is measured to guide the adjustment of electricity selling prices. The demand variation degree index quantifies the degree of change of the actual user demand relative to the predicted value, identifies abnormal fluctuations in demand, and provides a basis for adjusting the electricity selling strategy. Precise risk identification quantifies the deep characteristics of market price fluctuations and user demand changes, clearly differentiates different risk levels, and provides scientific support for optimizing the electricity selling plan. When the market fluctuates violently or the user demand changes significantly, it is determined as a high-risk period, indicating that optimization is required. Screening of low-risk periods: When the market and demand are relatively stable, the existing plan is directly adopted to reduce the calculation cost. Risk-driven optimization: Through risk classification, ensure that resources are concentrated on optimizing high-risk periods, improving the efficiency and benefits of the overall strategy.

[0020] The electricity selling plan determined last time is adopted for low-risk periods. For high-risk periods, a multi-objective optimization model is constructed to optimize the electricity selling plan, and the optimization result of the multi-objective optimization model is used as the electricity selling plan for this time for confirmation and application. Reusing the plan for low-risk periods: Avoid unnecessary recalculation, maintain the efficient use of computing resources, improve the stability and consistency of the plan, and avoid additional risks caused by frequent adjustments. Optimizing the plan for high-risk periods: Construct a multi-objective optimization model that comprehensively considers electricity selling revenue, electricity purchase cost, and user satisfaction, and optimize it through the genetic algorithm to quickly generate the optimal electricity selling plan that conforms to market dynamics. Enhancing the overall revenue: Avoid risks through optimization during high-risk periods and seize profit opportunities. Dynamically adjust the electricity selling strategy to adapt to changes in the market and user demand.

[0021] The preset time series prediction model is the autoregressive integrated moving average model or the long short-term memory network. The autoregressive integrated moving average model (ARIMA) is a mature and widely used existing technology in time series prediction, and its theory and application have been deeply studied in many fields and will not be elaborated here. Principle: Autoregressive (AR): Use the linear combination of historical time series data to predict future values. Moving average (MA): Correct the predicted value through the linear combination of historical prediction errors. Differencing (I): Differentiate the non-stationary time series to make it meet the stationarity assumption. The core idea of the ARIMA model is to generate the predicted value of the time series by combining AR and MA based on historical trends and error relationships.

[0022] Long short-term memory network (LSTM) is a classic recurrent neural network (RNN) extended model used to process sequence data. Its basic theory and implementation methods belong to the existing technology and will not be repeated here. Principle: LSTM can capture long-term dependencies in time series by introducing "memory units" and "gating mechanisms" (such as input gates, forget gates, and output gates). The network automatically learns the characteristics of time series during training through weight optimization and generates prediction results for future values. Compared with traditional ARIMA, LSTM is suitable for processing nonlinear, high-dimensional complex time series.

[0023] The logic for obtaining the periodic comprehensive evaluation value is as follows: Get the average price fluctuation within the preset forecast period , reflects the overall volatility of electricity market prices in the current period. The greater the price fluctuation, the higher the market uncertainty. The mean deviation between actual demand and predicted demand , which represents the degree of consistency between user demand changes and forecasts. The greater the deviation, the less accurate the forecast or the more dramatic changes in user demand. And the price elasticity coefficient of the electricity seller in the electricity sales area , which measures the responsiveness of user demand to price fluctuations. The greater the price elasticity, the more sensitive the user's electricity consumption behavior is to prices. The calculation formula is: ; It represents the average price change of all historical time periods used in the training of the time series forecasting model. It represents the average value of the change in user electricity consumption in all historical time periods used in the training of the time series prediction model. A constant set to prevent the denominator from being zero; The calculation formula for the periodic comprehensive evaluation value is: ; , , They are all preset non-zero proportional coefficients, which adjust the influence weight of each corresponding parameter in the comprehensive evaluation and can be flexibly adjusted according to specific business needs. Represents the comprehensive evaluation value of the period. It evaluates the overall status of the power market and user demand in the current period by comprehensively considering price fluctuations, demand deviation and price elasticity. Sharp price fluctuations in the power market may increase the power purchase cost or price adjustment risk of power sellers.

[0024] Deciding whether to conduct in-depth analysis based on the periodic comprehensive evaluation value means: The periodic comprehensive evaluation value is compared with the preset standard evaluation value range. If the periodic comprehensive evaluation value falls within the preset standard evaluation value range, no in-depth analysis is performed. If the periodic comprehensive evaluation value does not fall within the preset standard evaluation value range, in-depth analysis is performed.

[0025] By comparing the cycle comprehensive evaluation value (CIV) with the preset standard evaluation value range, you can quickly determine the market and demand status of the current time period. Falling into the standard range means that the market and demand changes are within an acceptable range and the risk of the current time period is low. Not falling into the standard range means that the market or demand has abnormal fluctuations and further analysis is required to formulate a more accurate power sales strategy. Avoid complex in-depth analysis of all time periods, reduce unnecessary computing resource consumption, and concentrate computing resources on high-risk periods.

[0026] High quality definition: The current time period is identified as high quality, which corresponds to a low-risk period, which means that market price fluctuations are stable, user demand forecast deviations are small, and responsiveness is good. These conditions should be reflected within the standard evaluation value range. Without in-depth analysis: If CIV falls into the standard range, it means that the market and demand conditions are within the acceptable range for electricity sellers and can be identified as high quality. High quality means: The current electricity sales plan has strong adaptability and can be used directly without optimization and adjustment. The degree of match between user demand and market price is high, and electricity sellers can obtain stable profits without additional intervention.

[0027] The logic for obtaining the price volatility index is: Calculate the overall amplitude influence coefficient: ; represents the price change between adjacent sampling points t and t-1, N represents the number of price samples in the current period, Indicates the overall amplitude impact coefficient; the fluctuation amplitude is measured by the root mean square value of the price change. It captures the overall magnitude of price fluctuations in the current time period. The greater the price fluctuation, The higher the value, the more suitable it is for measuring the comprehensive degree of price fluctuations within a cycle, and the higher the sensitivity is to sharp fluctuations.

[0028] Calculate the impact coefficient of extreme cases: ; Represents the average price of the current period, represents the electricity market price at the current sampling point t, Indicates the maximum price change between adjacent sampling points t and t-1, Indicates the extreme situation impact coefficient; measures the ratio of the maximum price fluctuation to the average value. Pay attention to the extreme fluctuations in market prices, i.e. the electricity purchase price costs of electricity sellers within a certain period of time, and emphasize the impact of high-risk points. If the price changes drastically at a certain moment, such as a sudden increase or decrease, It will increase significantly, indicating that there are abnormal fluctuations in the market.

[0029] Calculate the dispersion influence coefficient: ; Indicates the standard deviation of the price in the current period, reflecting the degree of price dispersion. Indicates the coefficient of influence of the degree of dispersion; quantified by the ratio of the standard deviation to the mean. Reflects the discreteness of prices in the current time period. The larger the standard deviation, the more dispersed the price fluctuations. The higher the value, the more the impact of discrete values ​​is smoothed through logarithmic function processing, making it more stable for the overall evaluation.

[0030] The calculation formula of the price volatility index is: ; , , All are preset non-zero adjustment coefficients. Represents the price volatility index. JPI is a comprehensive indicator calculated by weighting the overall price range, extreme changes and dispersion. It measures the volatility of market prices in the current time period. JPI takes into account the overall price volatility, extreme volatility and dispersion, and comprehensively describes the volatility characteristics of prices in the current time period. The meaning of high JPI: The market is in a state of high volatility, and the following situations may occur: drastic changes in supply and demand: sudden increase or decrease in demand or supply in the electricity market. Unstable price regulation: The market may be strongly disturbed by external factors (such as policies and weather). User responsiveness is tested: Users may adjust their electricity consumption behavior due to drastic price changes, affecting the profits of electricity sellers.

[0031] The logic for obtaining the demand variation index is: first calculate the demand deviation score, the formula is: ; Indicates the current power demand of the user at time point k With the predicted value The difference, is a preset constant, is the mean of the forecasted demand, It represents the standard deviation of the demand deviation in the current period, reflecting the degree of fluctuation. M represents the total number of time points in the current period. Represents the demand deviation score; comprehensively evaluates the total amount and volatility of demand deviation. The larger the deviation and the more drastic the fluctuation, the more uncontrollable the demand change is, and the higher the risk to the power sales strategy is. Then define the extreme deviation threshold as , is a preset standard constant. Count all total quantities greater than the extreme deviation threshold and mark them as ; Quantifies the impact of extreme values in demand deviation. The larger the standard deviation, the more significant the demand variation, and the electricity sales plan needs to improve its adaptability to extreme changes.

[0032] Then analyze the regularity of the sequence through the change amount of adjacent demand deviations to obtain a regularity score: ; represents the regularity score; The closer GL is to 1, the more regular the trend of demand deviation changes, and the more effective the electricity seller's prediction and optimization plan. When the trend is irregular (such as frequent deviation fluctuations and repeated direction changes), more refined prediction adjustments are required.

[0033] If , the change directions of the current time point k and the next time point k + 1 are the same. If the directions are the same (both positive or negative), it is regarded as "continuous change". Count all total numbers, and then divide by to obtain the direction persistence ratio value and mark it as ; Measures the direction consistency of demand deviation changes. The higher BL is, the more stable the direction of demand changes, which is beneficial to the electricity seller's optimization prediction model.

[0034] Combine , , , into a demand variation vector, and then calculate the Euclidean distance between the demand variation vector and the preset standard variation vector to obtain the demand variation degree index. If the value of the demand variation degree index is high, it indicates significant demand variation and the electricity sales plan needs to be further optimized. If the value of the demand variation degree index is low, it can be considered that the demand changes stably and the existing plan can be continued.

[0035] Dividing the current electricity sales period into high-risk periods and low-risk periods means: Take the demand variation degree index and the price volatility index of the current electricity sales period as the input variables of fuzzy inference, and take the division type of the current electricity sales period as the output variable. Perform fuzzy processing on the input variables, convert the values of the input variables into fuzzy sets, perform fuzzy processing on the output variable, convert the output variable into a fuzzy set, formulate fuzzy rules to describe the fitness of the current rules and each division type under different data combinations, and perform reasoning on the fuzzy input variables through the fuzzy rules to obtain the division type of the current electricity sales period.

[0036] The input variables include the demand mutation degree index and the price volatility index. The input variables are fuzzified, and their values are mapped to fuzzy sets, such as fuzzy linguistic sets of low, medium, high, etc., to describe the characteristics of demand changes and price fluctuations in the current period. The output variable is the classification type of the current electricity selling period, which is divided into high-risk periods and low-risk periods. The output variable is fuzzified, and the risk type is mapped to a fuzzy set to reflect the fuzziness of the risk level.

[0037] Combined with the fuzzy linguistic sets of the demand mutation degree index and the price volatility index, a rule table is formulated. The rules describe the corresponding risk classification types of the current electricity selling period under different combinations of input variables. For example: If the demand mutation degree is high and the price volatility is high, the period is classified as high-risk. If the demand mutation degree is low and the price volatility is low, the period is classified as low-risk. Other combination types are respectively set with fitness degrees to reflect the flexibility of fuzzy reasoning.

[0038] According to the fuzzy rule table, the fuzzified input variables are subjected to fuzzy reasoning. The fitness degree of each rule is calculated through rule matching, and combined with a rule reasoning model (such as the maximum membership degree method or the weighted average method), the risk classification of the current electricity selling period is obtained.

[0039] The result of fuzzy reasoning is defuzzified, and the clear classification type of the current electricity selling period is output as high-risk or low-risk. Fuzzy reasoning can handle the uncertainty and non-linear relationship of input variables, dynamically adapt to the complex changes of the market and demand, and ensure the timely adjustment of electricity selling strategies. By combining different input variable states with fuzzy rules, the risk level of the current electricity selling period is accurately identified, providing scientific support for the strategy optimization of high-risk periods. High-risk periods are given priority to enter the multi-objective optimization process, and low-risk periods follow the existing plan to achieve centralized utilization of resources and improve the calculation and management efficiency. Fuzzy reasoning still has strong judgment ability under the condition of data uncertainty, can effectively avoid misjudgment caused by traditional hard classification rules, and improve the robustness and flexibility of the decision-making system.

[0040] The multi-objective optimization model refers to an optimization model that constructs the dual objectives of the electricity selling price of the electricity seller and the user satisfaction through the genetic algorithm. The fitness function of the genetic algorithm comprehensively considers the electricity selling income, the electricity purchase cost, and the user satisfaction. ; represents the electricity selling price, represents the total electricity consumption of the user in the current electricity selling period, represents the cost of the electricity seller's market electricity purchase, represents the user satisfaction, 、 are both preset optimization coefficients, and the sum of the two is one; ; Represents the average value of the user's historical electricity bill payments. , , is a preset tolerance coefficient, representing the proportion of price fluctuations acceptable to the user. The selling electricity price corresponding to the maximum value of the final fitness function is used as the selling electricity plan for the current selling electricity period. It should be noted that if the current selling electricity period is divided into high-risk periods, then when using the selling electricity price corresponding to the maximum value of the fitness function as the selling electricity plan for the current selling electricity period, the preset time series prediction model also needs to be trained and updated to better conform to the current user's electricity consumption habits. And based on the updated time series prediction model, a preliminary power supply plan is made for the user, and power dispatching is carried out when the power supply is insufficient to meet the user's needs. When the power supply exceeds the demand, it can be stored through a preset energy storage device and supplied as the power for the next selling electricity period.

[0041] The specific steps of using the genetic algorithm are as follows: Define the population: In the optimization of the selling electricity plan, each individual represents a selling electricity price plan.

[0042] Randomly generate the initial population: Randomly assign values to the selling electricity price to generate a set of possible selling electricity price plans as the initial solution.

[0043] Individual structure: Each individual contains information such as the selling electricity price and electricity consumption. Define the fitness function: The fitness function comprehensively considers revenue, cost, and user satisfaction, and is used to measure the quality of each individual (selling electricity plan). The specific calculation formula is as described above. Revenue: Calculated based on the selling electricity price and electricity consumption. Cost: Calculated based on the purchase electricity price. Satisfaction: Calculated based on the user's acceptance of the selling electricity price to ensure a good response from the user to the current price.

[0044] Selection operation: The goal is to retain the individuals with stronger adaptability to enter the next generation. Methods such as roulette wheel selection: Allocate selection probabilities according to the fitness ratio, and individuals with higher fitness are more likely to be selected. Tournament selection: Randomly select some individuals and select the one with the highest fitness.

[0045] Crossover operation: The goal is to generate new solutions through gene recombination. Methods such as single-point crossover: Randomly select the selling electricity price as the cut point and exchange part of the genes of two individuals. Uniform crossover: Randomly exchange gene segments of the selling electricity price according to a preset probability.

[0046] Mutation operation: The goal is to introduce diversity through gene mutation and explore new solution spaces. Methods such as randomly adjusting the selling electricity price of individuals with a lower probability. The mutation range is within the set price tolerance interval to avoid exceeding the reasonable range.

[0047] Termination condition setting: The algorithm terminates when any of the following conditions is met: reaching the preset number of iterations. The fitness value of the population no longer increases significantly.

[0048] Output: Output the individual with the highest fitness, and its electricity selling price is used as the optimal solution for the current time period.

[0049] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0050] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0051] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0052] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0053] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for determining a power sales plan of a power seller, characterized in that: The following steps are involved: Obtain user electricity consumption data and perform forecast analysis through a preset time series forecasting model to obtain the forecast result, i.e. the user's future electricity demand; During the preset forecast period, based on the real-time price fluctuations, user demand changes and responsiveness of the electricity market, a comprehensive periodic evaluation value is generated, and a decision is made whether to conduct in-depth analysis based on the comprehensive periodic evaluation value; When conducting in-depth analysis, we analyze the fluctuation characteristics of the power market and the characteristics of user demand changes, obtain the price fluctuation intensity index and the demand abnormality index, and then divide the current power sales period into high-risk period and low-risk period; During low-risk periods, the power sales plan determined last time will be used. During high-risk periods, a multi-objective optimization model will be constructed to optimize the power sales plan, and the optimization results of the multi-objective optimization model will be confirmed and applied as the current power sales plan.

2. A method for determining a power sales plan of a power seller according to claim 1, characterized in that: The default time series forecasting model is the autoregressive moving average model or the long short-term memory network.

3. A method for determining a power sales plan of a power seller according to claim 2, characterized in that: The logic for obtaining the periodic comprehensive evaluation value is as follows: Get the average price fluctuation within the preset forecast period , the mean deviation between actual demand and predicted demand , and the price elasticity coefficient of the electricity seller within the electricity sales area , price elasticity coefficient The calculation formula is: ; It represents the average price change of all historical time periods used in the training of the time series forecasting model. It represents the average value of the change in user electricity consumption in all historical time periods used in the training of the time series prediction model. A constant set to prevent the denominator from being zero; The calculation formula for the comprehensive evaluation value of the cycle is: ; , , are all preset non-zero scale factors, Indicates the comprehensive evaluation value of the cycle.

4. A method for determining a power sales plan of a power seller according to claim 3, characterized in that: Deciding whether to conduct in-depth analysis based on the periodic comprehensive evaluation value means: The periodic comprehensive evaluation value is compared with the preset standard evaluation value range. If the periodic comprehensive evaluation value falls within the preset standard evaluation value range, no in-depth analysis is performed. If the periodic comprehensive evaluation value does not fall within the preset standard evaluation value range, in-depth analysis is performed.

5. A method for determining a power sales plan of a power seller according to claim 4, characterized in that: The logic for obtaining the price volatility index is: Calculate the overall amplitude influence coefficient: ; represents the price change between adjacent sampling points t and t-1, N represents the number of price samples in the current period, represents the overall amplitude influence coefficient; Calculate the impact coefficient of extreme cases: ; Represents the average price of the current period, represents the electricity market price at the current sampling point t, Indicates the maximum price change between adjacent sampling points t and t-1, Indicates the influence coefficient of extreme conditions; Calculate the dispersion influence coefficient: ; Indicates the standard deviation of the price in the current period, reflecting the degree of price dispersion. Indicates the influence coefficient of dispersion degree; The calculation formula of the price volatility index is: ; , , All are preset non-zero adjustment coefficients. An index that indicates price volatility.

6. A method for determining a power sales plan of a power seller according to claim 5, characterized in that: The logic for obtaining the demand variation index is as follows: First calculate the demand deviation score, the formula is: ; Indicates the current power demand of the user at time point k With the predicted value The difference, is a preset constant, represents the standard deviation of demand deviation in the current period, is the mean of the predicted demand, M represents the total number of time points in the current period, represents the demand deviation score; Then define the extreme deviation threshold as , is the preset standard constant, and all the values ​​greater than the extreme deviation threshold are counted. Total quantity, marked as ; Then, the regularity of the sequence is analyzed by the change in the deviation of adjacent demands, and the regularity score is obtained: ; represents the regularity score; if , then the change direction of the current time point k is consistent with that of the next time point k+1, and all The total number of , get the directional persistence ratio value and mark it as ; Will , , , The demand variation vector is summarized, and then the Euclidean distance between the demand variation vector and the preset standard variation vector is calculated to obtain the demand variation degree index.

7. A method for determining a power sales plan of a power seller according to claim 6, characterized in that: Dividing the current electricity sales period into high-risk period and low-risk period means: The demand anomaly index and price fluctuation index of the current electricity selling period are taken as input variables of fuzzy reasoning, and the division type of the current electricity selling period is taken as the output variable. The input variables are fuzzified and the values ​​of the input variables are converted into fuzzy sets. The output variables are fuzzified and the output variables are converted into fuzzy sets. Fuzzy rules are formulated to describe the adaptability of the current rules and each division type under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to obtain the division type of the current electricity selling period.

8. A method for determining a power sales plan of a power seller according to claim 7, characterized in that: The multi-objective optimization model refers to the optimization model that uses genetic algorithms to build a dual-objective optimization model for electricity sales prices and user satisfaction. The fitness function of the genetic algorithm comprehensively considers electricity sales revenue, electricity purchase costs, and user satisfaction. ; represents the electricity selling price, Indicates the total electricity consumption of the user during the current electricity sales period. It represents the cost of purchasing electricity from the electricity retailer market. Indicates user satisfaction, , Both are preset optimization coefficients, and the sum of the two is one; ; Indicates the average value of the user's historical electricity charges. , , It is the preset tolerance coefficient, which indicates the price fluctuation ratio accepted by the user; The electricity selling price corresponding to the maximum value of the final fitness function is used as the electricity selling plan for the current electricity selling period.

Images