Power purchase and sale strategy optimization aided decision-making method

Abstract

The invention discloses an electricity purchasing and selling strategy optimization aided decision-making method, and the method comprises the steps: determining an evaluation index set, and evaluating the gray correlation degree of each evaluation index through gray correlation analysis. Based on the grey correlation degree, the weight vector of the evaluation index is further determined so as to comprehensively consider multiple factors influencing the electricity market, such as market transaction share, electrical load and generating capacity, and meanwhile, the influence degree of each index on the electricity market is more accurately evaluated. Then, a fuzzy evaluation matrix of the evaluation indexes is constructed, and a fuzzy comprehensive evaluation vector is calculated in combination with the weight vectors of the evaluation indexes; the vector can comprehensively reflect the comprehensive membership degree of a plurality of evaluation indexes to high, medium and low risk levels, so that comprehensive evaluation of market transaction risks is realized. In the decision-making process, the particle swarm optimization algorithm is adopted to solve the electricity purchasing and selling strategy decision-making model, the optimal electricity purchasing and selling strategy is automatically explored, and the decision-making efficiency and accuracy are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of electricity market trading technology, and in particular to an auxiliary decision-making method for optimizing electricity purchase and sale strategies. Background Technology

[0002] With the continuous deepening and rapid development of the electricity trading market reform, the number of market participants has increased dramatically, with various power generation companies, electricity sales companies, and emerging energy service providers flocking to the market. This trend has intensified competition in the electricity market, leading to more frequent and larger price fluctuations. Consequently, the uncertainty and risks encountered in the electricity trading process have increased significantly, posing a great challenge to the decision-making of market participants.

[0003] Integrated power generation and sales groups are usually initiated by power generation companies or large energy groups, and have both power generation and sales businesses. For example, CNNC Huineng's "integrated power generation and sales" power sales company registered in Gansu can participate in power trading and power generation businesses at the same time, and has dual pricing advantages.

[0004] Currently, due to uncertainties in spot market electricity prices and next-day user electricity consumption, the traditional electricity purchase decision-making methods of integrated generation and sales companies face certain risks. Furthermore, these companies must balance power generation and sales efficiency; however, traditional power purchase and sales decision-making methods often emphasize efficiency and profitability while neglecting the significant impact of power generation on overall efficiency. Moreover, the power generation process itself is subject to uncertainties regarding the efficiency, thermal efficiency, and power generation output of the gas turbines operated by the integrated generation and sales company, further exacerbating the decision-making risks.

[0005] Therefore, this invention proposes an auxiliary decision-making method for optimizing electricity purchase and sale strategies, aiming to provide scientific, reasonable, and accurate decision support for integrated electricity sales companies, and help them achieve steady development in a complex and ever-changing market environment. Summary of the Invention

[0006] The purpose of this invention is to improve and innovate upon the shortcomings and problems existing in the background technology, and to provide an auxiliary decision-making method for optimizing the power purchase and sale strategy of power generation companies that integrate power generation and sales.

[0007] According to a first aspect of the present invention, a method for optimizing power purchase and sale strategies is provided, specifically comprising the following steps: Collect historical and real-time data on the electricity market and power purchase and sale entities, and preprocess the data; Historical and real-time data are input into the sequence model to predict the changing trends of electricity purchase and sales prices, as well as the changing trends of the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the electricity purchase and sales entities themselves. Determine the set of evaluation indicators, use grey relational analysis to evaluate the grey relational degree of each evaluation indicator, and determine the weight vector of the evaluation indicators based on the grey relational degree of the evaluation indicators. Construct a fuzzy evaluation matrix for the evaluation indicators, and determine the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix. A power purchase and sale strategy decision model is constructed. The objective function of the model is based on the predicted power purchase price, power sale price, the changing trends of the gas turbine power generation efficiency, gas turbine heat generation efficiency, and power generation of the power purchase and sale entity, as well as the fuzzy comprehensive evaluation vector. The constraints of the model are constructed based on the fuzzy comprehensive evaluation vector. The power purchase and sale strategy decision model is solved to predict the power purchase and sale strategy. The aforementioned power purchase and sale strategy decision model is constructed, wherein the objective function of the model is based on the predicted trends of power purchase price, power sale price, the gas turbine power generation efficiency, gas turbine heat generation efficiency, and power generation of the power purchase and sale entity, as well as the fuzzy comprehensive evaluation vector; the constraints of the model are constructed based on the fuzzy comprehensive evaluation vector; solving the power purchase and sale strategy decision model yields the following specific power purchase and sale strategies: Define the set of variables for the electricity purchase and sale strategy; Define the objective function; Wherein, objective function ; In the formula, = p_sell represents the electricity sales price predicted by the sequence model; p_buy represents the electricity purchase price predicted by the sequence model; x_1 represents the electricity purchase volume solved by the model; x_2 represents the electricity sales volume solved by the model; b1 represents the high-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; b2 represents the medium-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; λ_1 and λ_2 are constants. As an efficiency reward coefficient, This represents the gas turbine power generation efficiency predicted by the sequence model; This represents the gas turbine heat production efficiency predicted by the sequence model; This represents the theoretical maximum power generation efficiency of the gas turbine; This represents the theoretical maximum heat production efficiency of the gas turbine.

[0008] Set constraints; x_1*p1 <x_2<x_1*p2; Q_1max*p3≦x_1≦Q_1max*p4, where Q_1max represents the market's power generation capacity; b1≦k1; b3≧k2; Where b1 and b3 represent the high-risk and low-risk comprehensive membership degrees corresponding to the fuzzy comprehensive evaluation vector, respectively; p1, p2, p3, p4 and k1, k2 are all constants; x_1≧0, x_2≧0; The objective function is solved using the particle swarm optimization algorithm to predict the electricity purchase and sale strategy.

[0009] A further approach involves inputting historical and real-time data into a sequence model to predict trends in electricity purchase and sales prices, as well as trends in the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the electricity purchase and sales entities. Specifically, this includes: When the sequence model is an ARIMA time series model, the collected historical and / or real-time electricity purchase prices are input into the ARIMA time series model, and the trend of electricity purchase price changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time electricity sales prices are input into the ARIMA time series model, and the trend of electricity sales price changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time gas turbine power generation efficiency is input into the ARIMA time series model, and the trend of gas turbine power generation efficiency changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time power generation is input into the ARIMA time series model, and the trend of power generation changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time power generation is input into the ARIMA time series model, and the trend of power generation changes is predicted through curve fitting and parameter estimation. When the sequence model is a recurrent neural network model, the collected historical electricity purchase price and / or real-time electricity purchase price, historical electricity sales price and / or real-time electricity sales price, historical power generation and / or real-time power generation, historical electricity load and / or real-time electricity load are input into each gated recurrent unit of the trained recurrent neural network model to predict the changing trends of electricity purchase price and electricity sales price. When the sequence model is a recurrent neural network model, the collected historical and / or real-time gas turbine power generation, natural gas consumption, natural gas calorific value and gas turbine power generation efficiency are input into each gated recurrent unit of the trained recurrent neural network model to predict the changing trends of gas turbine power generation efficiency and power generation. When the sequence model is a recurrent neural network model, the collected historical and / or real-time flue gas waste heat, natural gas consumption, natural gas calorific value, and gas turbine heat production efficiency are input into each gated recurrent unit of the trained recurrent neural network model to predict the trend of gas turbine heat production efficiency.

[0010] A further approach involves determining the evaluation index set, using grey relational analysis to assess the grey relational degree of each evaluation index, and determining the weight vector of the evaluation index based on the grey relational degree. This specifically includes: Determine the set of evaluation indicators; Calculate the grey relational coefficient for each evaluation indicator; the formula for calculating the grey relational coefficient is: ; in, Represents the grey relational coefficient of the i-th evaluation index; x0(k) represents the data at time k in the reference sequence, where k∈[1,N]; This represents the data at time k in the comparison sequence corresponding to the i-th evaluation index; This represents the absolute difference between the two at time k; ; ρ is a constant; N represents the total number of elements in the reference sequence or comparison sequence; Based on the grey relational analysis of the evaluation indicators, the formula for calculating the weight vector of the evaluation indicators is as follows: ; Let represent the grey relational coefficient of the i-th evaluation index; α is a constant; exp() represents the natural exponential function; Let represent the grey relational coefficient of the j-th evaluation indicator; n is the total number of evaluation indicators.

[0011] A further approach involves constructing a fuzzy evaluation matrix for the evaluation indicators, and determining the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix. This process specifically includes: Set the fuzzy evaluation matrix R = [r ij ]n×m; where r ij represents the membership degree of the i-th evaluation indicator to the j-th evaluation level; n represents the total number of evaluation indicators, and m represents the total number of evaluation levels; Among them, the membership degree of the evaluation index is calculated based on the membership function or determined by the expert scoring method; By multiplying the weight vector of the evaluation index with the fuzzy evaluation matrix through fuzzy transformation, the fuzzy comprehensive evaluation vector B=WR=(b1,b2,…,b) is calculated. j ,…,b m ); where b j This represents the overall membership degree of each evaluation indicator to the j-th evaluation level.

[0012] A further proposed approach is to include price volatility, supply-demand ratio, and market share as evaluation indicators; and to classify evaluation levels as high risk, medium risk, and low risk. The membership degree of price fluctuations to high-risk, medium-risk, and low-risk categories is calculated as follows: Based on the purchase prices at adjacent time points, obtain the price volatility δ; Construct membership functions, where the formula for the membership function of price fluctuations is as follows: r 11 =max(0,min(1,(δ-5) / 5)); r 12 =max(0,min(1,(10-δ) / 5,δ / 5)); r 13 =max(0,min(1,(5-δ) / 5)); r 11 r represents the degree to which price volatility belongs to high risk. 12 r represents the degree of membership of price volatility to medium risk. 13 This indicates the degree to which price fluctuations belong to low risk; The membership calculation process for high-risk, medium-risk, and low-risk supply and demand comparisons is as follows: Based on the power generation and power load at the same time, the supply-demand ratio at the corresponding time is calculated; Construct membership functions, where the formula for the membership function of the supply-demand ratio is as follows: r 21 =max(0,min(1,(1-SDR) / 0.2));

[0013] r 23 =max(0,min(1,(SDR-0.8) / 0.4)); r 21 Indicates the degree of membership in a supply-demand comparison that carries high risk, r 22 r represents the degree of membership of risk in the supply and demand comparison. 23 This indicates the degree of membership in a low-risk supply-demand comparison.

[0014] A further approach is that, after solving the power purchase and sale strategy decision model and predicting the power purchase and sale strategy, the method further includes: A test dataset was constructed to verify the accuracy of the electricity purchase and sale strategy decision model, and the model was optimized based on the verification results.

[0015] A further approach involves constructing a test dataset to verify the accuracy of the electricity purchase and sale strategy decision-making model, and then optimizing the model based on the verification results. This specifically includes: Prepare the test dataset; A power purchase and sale strategy decision model is used to predict the test dataset and generate power purchase and sale strategies. Compare the forecast data with the actual data corresponding to the electricity purchase and sale strategy; Analyze the performance of the electricity purchase and sale strategy decision model on the test dataset; The power purchase and sale strategy decision model was optimized based on the verification results.

[0016] According to a second aspect of the present invention, an electronic device is provided, comprising: a memory and a processor; The memory is used to store programs; The processor is configured to invoke a program stored in the memory to execute a power purchase and sale strategy optimization auxiliary decision-making method as described in any of the preceding claims.

[0017] According to a third aspect of the present invention, a readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements a power purchase and sale strategy optimization auxiliary decision-making method as described in any of the preceding claims.

[0018] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes an auxiliary decision-making method for optimizing power purchase and sale strategies. This method determines a set of evaluation indicators and uses grey relational analysis to assess the grey relational degree of each indicator. Based on the grey relational degree, the weight vectors of the evaluation indicators are further determined to comprehensively consider multiple factors affecting the electricity market, such as market share, electricity load, and power generation, while more accurately assessing the impact of each indicator on the electricity market. Subsequently, a fuzzy evaluation matrix of the evaluation indicators is constructed, and combined with the weight vectors of the evaluation indicators, a fuzzy comprehensive evaluation vector is calculated. This vector comprehensively reflects the overall membership degree of multiple evaluation indicators to different risk levels (high, medium, and low), thereby achieving a comprehensive assessment of market transaction risks. In the decision-making process, this invention uses a particle swarm optimization algorithm to solve the power purchase and sale strategy decision model, automatically exploring the optimal power purchase and sale strategy, significantly improving the efficiency and accuracy of decision-making. The objective function of the electricity purchase and sale strategy decision model is constructed based on the predicted trends of electricity purchase price, electricity sale price, gas turbine power generation efficiency, gas turbine thermal efficiency, and power generation, as well as the fuzzy comprehensive evaluation vector. Simultaneously, the model's constraints are also set based on the fuzzy comprehensive evaluation vector, thus fully considering uncertainties such as spot market electricity prices, next-day user electricity consumption, and the power generation efficiency, thermal efficiency, and power generation of the integrated power generation and sale company's own gas turbines, avoiding transaction risks arising from these uncertainties. Through an intelligent decision-making process, this invention can significantly improve the efficiency and effectiveness of electricity trading, reduce human intervention and errors, accelerate decision-making, and thereby increase the success rate of transactions. Furthermore, this method can flexibly adapt to the needs of electricity purchase and sale decisions under different market conditions, including normal market environments, market fluctuations, and supply-demand imbalances, demonstrating strong flexibility and adaptability. In summary, this invention provides a novel, efficient, and accurate decision-making method. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an auxiliary decision-making method for optimizing electricity purchase and sale strategies provided in the first embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0023] Example 1 Please see Figure 1 This invention provides a method for optimizing power purchase and sale strategies, specifically including the following steps: Step S101: Collect historical and real-time data of the electricity market and power purchase and sale entities, and preprocess the data; Specifically, for historical data, electricity market data from the past several years is collected, including but not limited to electricity purchase prices, electricity sales prices, electricity load, power generation, market transaction volume, and weather data. This data can be obtained from electricity market operators, grid companies, and power generation enterprises.

[0024] For real-time data, the latest data from the electricity market, such as electricity purchase price, electricity sales price, electricity load, and power generation, can be obtained in real time through data interfaces.

[0025] It should be noted that the collection of historical and real-time data from the electricity market and power purchase and sale entities also includes the integrated power generation and sales company's own data, such as its own power generation, natural gas consumption, and natural gas calorific value; flue gas waste heat, flue gas mass flow rate, flue gas specific heat capacity, flue gas temperature, and ambient temperature. Among these, the flue gas waste heat is recovered and utilized by the waste heat system, and is calculated from the flue gas mass flow rate, flue gas specific heat capacity, flue gas temperature, and ambient temperature; natural gas consumption is obtained through flow meter measurement; flue gas mass flow rate is obtained through flue gas velocity meter measurement; flue gas specific heat capacity is obtained through flue gas composition analyzer measurement; flue gas temperature is obtained through thermocouple measurement; and ambient temperature can be obtained from meteorological station monitoring data.

[0026] For example, the collected electricity purchase price data, electricity load data, and power generation data are shown in Table 1 below; Table 1 Historical data on electricity purchase price, electricity load, and power generation.

[0027] It should be noted that after collecting historical and real-time data from the electricity market, data preprocessing is required. This preprocessing includes: a. Data cleaning: Remove outliers and missing values ​​to ensure data quality and consistency. For missing values, linear interpolation can be used to fill them.

[0028] b. Data conversion: Convert the data into a format suitable for analysis, such as converting date and time into a unified format. In this embodiment, date and time can be converted into a unified YYYY-MM-DD format, and text data can be converted into numerical data, etc.

[0029] c. Data Standardization: Standardization formulas are used to standardize data on electricity purchase prices, electricity sales prices, electricity load, power generation, and market transaction volume to eliminate the influence of dimensions. The standardization formulas are as follows: , where x is historical or real-time data; μ is the mean of the corresponding data; σ is the standard deviation of the corresponding data; It is standardized historical or real-time data.

[0030] Step S102: Input historical and real-time data into the sequence model to predict the changing trends of electricity purchase price and electricity sales price, as well as the changing trends of the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the electricity purchase and sales entities themselves; The sequence model can be an ARIMA time series model or a recurrent neural network model. The time series model predicts future data changes based on collected time series data through curve fitting and parameter estimation. For example, to predict the trend of electricity purchase price changes, historical and / or real-time electricity purchase prices can be used to predict future electricity purchase prices over a certain period through curve fitting and parameter estimation. Similarly, to predict the trend of electricity sales price changes, historical and / or real-time electricity sales prices can be used to predict future electricity sales prices over a certain period through curve fitting and parameter estimation. Likewise, historical and / or real-time gas turbine power generation efficiency is input into the ARIMA time series model, and the future power generation efficiency of the gas turbine is predicted through curve fitting and parameter estimation. The historical and / or real-time power generation of the integrated power generation and sales company is also input into the ARIMA time series model, and its future power generation is predicted through curve fitting and parameter estimation.

[0031] The recurrent neural network model mainly consists of an encoding module and a decoding module. In practical applications, historical or real-time data over a period of time is integrated into a sequence. Then, data from different points in time within the sequence are sequentially input into the gated recurrent units corresponding to the encoding module. After processing, the encoding module outputs a context vector C. Next, the decoding module decodes this context vector C to predict and generate the purchase and sale prices of electricity over a future period. To predict the trends in purchase and sale prices, the data input to the gated recurrent units of the encoding module is not limited to purchase and sale prices; it can also include other relevant data such as power generation and load. To predict the trends in gas turbine power generation efficiency and output, the data input to each gated loop unit of the encoding module can include historical and / or real-time gas turbine power generation, natural gas consumption, natural gas calorific value, and gas turbine power generation efficiency. Similarly, to predict the trends in gas turbine heat production efficiency, the data input to each gated loop unit of the encoding module can include historical and / or real-time flue gas waste heat, natural gas consumption, natural gas calorific value, and gas turbine heat production efficiency.

[0032] It should be noted that the historical or real-time gas turbine power generation efficiency of the integrated power generation and sales company is calculated based on power generation, natural gas consumption, and natural gas calorific value, using the following formula: ; In the formula, Indicates historical or real-time gas turbine power generation efficiency; This indicates historical or real-time gas turbine power generation. Indicates historical or real-time natural gas consumption; Indicates historical or real-time natural gas calorific value; The historical or real-time gas turbine thermal efficiency of the integrated sales company is calculated based on flue gas waste heat, natural gas consumption, and natural gas calorific value, using the following formula: ; In the formula, This indicates the historical or real-time heat production efficiency of the gas turbine; This indicates the waste heat of flue gas discharged from gas turbines in historical or real-time terms; This indicates the efficiency of the waste heat boiler (typically 80-90%).

[0033] Furthermore, the formula for calculating the waste heat of flue gas is as follows: ; In the formula, Indicates historical or real-time flue gas mass flow rate. This indicates the corresponding historical or real-time flue gas specific heat capacity; Indicates historical or real-time flue gas temperature; Indicates historical or real-time ambient temperature.

[0034] It should be further explained that predicting the changing trends of electricity purchase and sales prices using a recurrent neural network (RNN) model requires training the RNN model using historical or real-time data. For example, data such as electricity purchase price, sales price, power generation, and electricity load at times 1, 2, 3, ..., t, ..., T-1 are input into the gated recurrent units of the RNN model to be trained to predict the electricity purchase and sales prices at time T. Then, based on the predicted electricity purchase and sales prices at time T and the actual electricity purchase and sales prices at time T, the parameters of the RNN model are iteratively updated until the number of iterations reaches a preset number or the difference between the predicted electricity purchase and sales prices at time T and the actual electricity purchase and sales prices at time T is less than a preset value, thus obtaining a trained RNN model.

[0035] Step S103: Determine the set of evaluation indicators, use grey relational analysis to evaluate the grey relational degree of each evaluation indicator, and determine the weight vector of the evaluation indicators based on the grey relational degree of the evaluation indicators. First, determine the evaluation index set U = {u1, u2, ..., u...} i ,...,u n In this embodiment, U = {price fluctuation (u1), supply-demand ratio (u2), market share (u3)}; the evaluation indicators in the evaluation indicator set are selected based on experience, and the evaluation indicators are those that have a significant impact on the electricity purchase and sale strategy.

[0036] For each evaluation index, calculate its grey relational coefficient with the reference sequence; the formula for calculating the grey relational coefficient is: ; Where x0 represents the reference sequence, such as the electricity purchase price sequence, and the reference sequence also depends on experience for selection; This represents the comparison sequence corresponding to the i-th evaluation index, such as electricity load and power generation. x0(k) represents the data corresponding to time k in the reference sequence; This represents the data at time k in the comparison sequence corresponding to the i-th evaluation index; This represents the absolute difference between the two at time k; ; ; ρ represents the resolution coefficient, with a value range of 0.1≤ρ≤0.9; N represents the total number of data points in the reference sequence or comparison sequence.

[0037] The above calculation formula can be used to obtain the grey relational degree between the comparison sequence and the reference sequence corresponding to different evaluation indicators, thereby assessing the degree of impact of these evaluation indicators on the electricity market. The higher the grey relational degree, the greater the impact of the evaluation indicator on the market.

[0038] As mentioned above, after obtaining standardized historical or real-time data, the standardized historical or real-time data at each time point can be used to form a reference sequence or comparison sequence; the ratio of historical or real-time data at adjacent time points can also be used to form a reference sequence or comparison sequence; the ratio of two historical or real-time data at the same time point can also be used to form a reference sequence or comparison sequence. This invention does not impose specific limitations, all of which are within the protection scope of this invention.

[0039] In this embodiment, to assess the impact of evaluation indicators such as price fluctuations, supply-demand ratio, and market share on the electricity market, the following method can be adopted: The standardized electricity purchase price at each time point is set as a reference sequence. Then, the ratio of the standardized electricity purchase price at each time point to the standardized electricity purchase price at the previous time point is calculated sequentially, and these ratios of adjacent time points are used as the first comparison sequence. Simultaneously, the ratio of standardized electricity load to power generation at each time point is used as the second comparison sequence, and the ratio of the standardized trading volume of the target electricity trading company to the total trading volume of the entire trading market at each time point is used as the third comparison sequence. Afterwards, the grey relational coefficients between the first, second, and third comparison sequences and the reference sequence are calculated respectively, thereby determining the impact of evaluation indicators such as price fluctuations, supply-demand ratio, and market share on the electricity market.

[0040] Specifically, determining the weight vector of the evaluation indicators based on their grey relational properties includes: For each evaluation index, its corresponding weight is determined, thus obtaining the weight vector W={w1,w2,…,w i ,…,w n},in This represents the weight of the i-th evaluation indicator.

[0041] in, ; represents the grey relational coefficient of the i-th evaluation index; α represents the adjustment factor, α∈[2,3]; exp() represents the natural exponential function.

[0042] For example, if the grey relational coefficient of price fluctuations calculated in the above steps =0.85; Grey relational coefficient of supply and demand ratio =0.72; Grey relational coefficient of market share =0.63; and α=2, then the weights of the evaluation indicators price fluctuation, supply-demand ratio, and market share are calculated as follows: =exp(2×0.85)=exp(1.7)≈5.47; =exp(2×0.72)=exp(1.44)≈4.22; =exp(2×0.63)=exp(1.26)≈3.53; =5.47 / (5.47 + 4.22 + 3.53) = 5.47 / 13.22 ≈ 0.41; =4.22 / (5.47 + 4.22 + 3.53) = 4.22 / 13.22 ≈ 0.32; =3.53 / (5.47 + 4.22 + 3.53) = 3.53 / 13.22 ≈ 0.27; Therefore, the weight vector of the evaluation index is W={0.41,0.32,0.27}.

[0043] Step S104: Construct the fuzzy evaluation matrix of the evaluation indicators, and determine the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix; Specifically, the fuzzy evaluation matrix R = [r ij ]n×m; where r ij represents the membership degree of the i-th evaluation indicator to the j-th evaluation level; n represents the total number of evaluation indicators, and m represents the total number of evaluation levels.

[0044] In this embodiment, the evaluation indicators include price fluctuation, supply-demand ratio, and market share, and the evaluation levels include high risk, medium risk, and low risk; then the fuzzy evaluation matrix R=[r 11 r 12 r 13 ;r 21 r 22 r 23 ;r 31 r 32 r 33 ]; where r 11 r 12 and r 13 These represent the membership degrees of price volatility to high-risk, medium-risk, and low-risk levels, respectively; r 21 r 22 and r 23 These represent the membership degrees of high-risk, medium-risk, and low-risk in the supply-demand comparison, respectively; r 31 r32 and r 33 These represent the degree to which market share belongs to high-risk, medium-risk, and low-risk categories, respectively.

[0045] The membership degree of the evaluation indicators to the evaluation level can be calculated using membership functions or determined using expert scoring. For example, price fluctuations and the supply-demand ratio are calculated using membership functions, while market share is determined using expert scoring.

[0046] Specifically, the membership degree of price fluctuations to high-risk, medium-risk, and low-risk categories is calculated as follows: Based on the purchase prices at adjacent times, obtain the price volatility δ (unit: %, δ=|P t -P t-1 | / P t-1 ×100%, P t (This represents the purchase price at time t); adjacent times can be spaced 15 minutes apart or 1 hour apart; this can be determined by those skilled in the art based on the actual situation.

[0047] Construct membership functions, where the formula for the membership function of price fluctuations is as follows: r 11 =max(0,min(1,(δ-5) / 5)); r 12 =max(0,min(1,(10-δ) / 5,δ / 5)); r 13 =max(0,min(1,(5-δ) / 5)); It is understandable that: When δ < 5, r 11 =0, r 12 =δ / 5, r 13 =1-δ / 5; When 5 ≤ δ ≤ 10, r 11 =δ / 5-1,r 12 =2-δ / 5,r 13 =0; When δ>10, r 11 =1, r 12 =0, r 13 =0.

[0048] Specifically, the membership calculation process for the supply-demand ratio (SDR) for high-risk, medium-risk, and low-risk categories is as follows: Based on the power generation and power load at the same time, the supply-demand ratio at the corresponding time is calculated; Construct membership functions, where the formula for the membership function of the supply-demand ratio is as follows: r 21=max(0,min(1,(1-SDR) / 0.2));

[0049] r 23 =max(0,min(1,(SDR-0.8) / 0.4)); It is understandable that: When SDR < 0.6, r 21 =1, r 22 =0, r 23 =0; When 0.6 ≤ SDR < 0.8, r 21 =1, r 22 =(SDR-0.6) / 0.4, r 23 =0; When 0.8 ≤ SDR < 1, r 21 =(1-SDR) / 0.2, r 22 =1, r 23 =(SDR-0.8) / 0.4; When 1 ≤ SDR ≤ 1.2, r 21 =0, r 22 =1, r 23 =(SDR-0.8) / 0.4 When SDR > 1.2, r 21 =0, r 22 =0, r 23 =1.

[0050] Specifically, Table 2 below shows the high-risk, medium-risk, and low-risk membership of the supply-demand ratio SDR at different ranges; Table 2. High-risk, medium-risk, and low-risk membership of the supply-demand ratio SDR at different ranges.

[0051] Specifically, the membership calculation process for market share in high-risk, medium-risk, and low-risk categories is as follows: a. Establish an expert group: The expert panel consists of 5-10 experts in the field, including power market analysts, corporate executives, and policy researchers. b. Implement scoring: Direct assignment method: Experts directly fill in the membership degree, for example, experts directly fill in the membership degree [0.7, 0.2, 0.1]; Voting method: Statistically calculate the percentage of experts who selected each level. For example, if 70% of experts selected "high risk," then r... 31 =0.7; c. Data aggregation: Arithmetic mean: The average of all expert scores; Quality control: Remove extreme values ​​(such as the highest / lowest score) and then average them; Use the Delphi method (multi-round anonymous scoring) to improve consistency; Dynamically updated: Re-scoring periodically (e.g., quarterly) to reflect market changes; To avoid interference from experts, independent blind reviews were conducted.

[0052] After obtaining the high, medium, and low risk membership parameters for each evaluation indicator, the evaluation level set for each indicator is obtained. By concatenating these membership parameters, a fuzzy evaluation matrix can be constructed. The row vectors of this matrix correspond one-to-one with the evaluation level set of each evaluation indicator.

[0053] Specifically, determining the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix includes: By multiplying the weight vector of the evaluation index with the fuzzy evaluation matrix through fuzzy transformation, the fuzzy comprehensive evaluation vector B=WR=(b1,b2,…,b) is calculated. j ,…,b m ); where b j This represents the overall membership degree of each evaluation indicator to the j-th evaluation level.

[0054] By obtaining the fuzzy comprehensive evaluation vector, it is possible to determine the main types of market risks, achieve a comprehensive assessment of the risks and returns of the electricity market, and formulate a preliminary electricity purchase and sale strategy.

[0055] For example, if the weight vector of the evaluation index is W=[0.62, 0.27, 0.11], and the fuzzy evaluation matrix is ​​R={[0.7, 0.2, 0.1], [0.6, 0.3, 0.1], [0.5, 0.4, 0.1]}, then the fuzzy comprehensive evaluation vector is B=WR=[0.651, 0.249, 0.1], where 0.651 in the fuzzy comprehensive evaluation vector represents the comprehensive membership degree of each evaluation index to high risk, 0.249 in the fuzzy comprehensive evaluation vector represents the comprehensive membership degree of each evaluation index to medium risk, and 0.1 in the fuzzy comprehensive evaluation vector represents the comprehensive membership degree of each evaluation index to low risk.

[0056] Step S105: Construct a power purchase and sale strategy decision model, wherein the objective function of the model is constructed based on the predicted power purchase price, power sale price, the changing trends of the gas turbine power generation efficiency, gas turbine heat generation efficiency and power generation of the power purchase and sale entity, and the fuzzy comprehensive evaluation vector; the constraints of the model are constructed based on the fuzzy comprehensive evaluation vector; solve the power purchase and sale strategy decision model to predict the power purchase and sale strategy; Specifically, constructing a power purchase and sale strategy decision model includes: Step S1051: Set the power purchase and sale strategy variable set X={X1,X2,…,X…} i ,…,X n}, where X i This represents the i-th electricity purchase and sale strategy. The set of electricity purchase and sale strategy variables X contains all possible electricity purchase and sale strategies, where each strategy includes at least the amount of electricity purchased and the amount of electricity sold.

[0057] Step S1052: Define the objective function; The objective function is used to evaluate the performance of each electricity purchase and sale strategy. Performance evaluation can be based on multiple indicators, such as cost-benefit analysis, risk-reward balance, and market share. The design of the objective function should reflect the decision-maker's goals and preferences, such as maximizing profits, minimizing costs or risks, or a combination of maximizing profits and risks, or a combination of minimizing costs and risks. Those skilled in the art can determine the appropriate objective function based on the specific circumstances.

[0058] In this embodiment, the objective function ; In the formula, = p_sell represents the electricity sales price predicted by the sequence model; p_buy represents the electricity purchase price predicted by the sequence model; x_1 represents the electricity purchase volume solved by the model; x_2 represents the electricity sales volume solved by the model; b1 represents the high-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; b2 represents the medium-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; λ_1 and λ_2 are constants. As an efficiency reward coefficient, This represents the gas turbine power generation efficiency predicted by the sequence model; This represents the gas turbine heat production efficiency predicted by the sequence model; This represents the theoretical maximum power generation efficiency of the gas turbine; This represents the theoretical maximum heat production efficiency of the gas turbine. λ_1 represents the penalty coefficient for high-risk units, which is typically taken as 0.5 yuan / kWh; λ_2 represents the penalty coefficient for medium-risk units, which is typically taken as 0.2 yuan / kWh.

[0059] Step S1053: Set constraints; The constraints define the limitations or conditions that the power purchase and sale strategy must meet. These constraints include technical constraints (such as limitations on power generation and transmission capacity), market rules (such as limits on trading volume), and policies and regulations (such as environmental regulations). The setting of constraints ensures that the generated power purchase and sale strategy is feasible and compliant in practice.

[0060] In this embodiment, the constraints include: Policy constraints: x_1*p1 <x_2<x_1*p2; According to national policy regulations, the majority of the electricity purchased by power trading companies must be used for guaranteed power supply tasks; Power generation capacity constraints: Q_1max*p3≦x_1≦Q_1max*p4, where Q_1max represents the market's power generation capacity; Risk membership constraint: b1≦k1; b3≧k2; Where b1 and b3 represent the high-risk comprehensive membership degree and low-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector, respectively; p1, p2, p3, p4 and k1, k2 are constants; in this embodiment, p1 can be 15%, p2 can be 20%, p3 can be 30%, p4 can be 50%, k1 can be 0.7, and k2 can be 0.3.

[0061] It should be noted that the specific risk threshold can be adjusted according to the latest announcement from the local power trading center. For example, the Guangdong Power Trading Center requires that risk avoidance be initiated when b1 > 0.65; the Shanxi power market stipulates that when the proportion of new energy is > 40%, the risk threshold can be relaxed by 20%.

[0062] Nonnegativity constraint: x_1≧0, x_2≧0.

[0063] Solving the power purchase and sale strategy decision model and predicting the specific power purchase and sale strategies includes: a. Initialize the particle swarm: Set the particle swarm size N. Randomly generate the position X of each particle. i =(x i1 ,x i2 ,…,x iM ) and velocity υ i =(υ i1 ,υ i2 ,…,υ iM ), where M is the number of decision variables.

[0064] b. Evaluate particle fitness: For each particle X i Calculate its fitness value f(X) i The objective function value is used to evaluate the fitness value of each particle in the electricity purchase and sale strategy model.

[0065] c. Update the particle velocity and position: For each particle X i Its velocity υ is updated according to the following formula. i and position X i : υ i(t+1) =w*υ i(t) +c1*rand()*(pbest i -X i(t) )+c2*rand()*(gbest-X i(t) ) X i(t+1) =X i(t) +υ i(t+1) Among them, υ i(t+1) Represents particle X i The velocity at time t+1.

[0066] w represents the inertia weight, which controls the inertia that keeps the particle in its current state of motion. The inertia weight w adopts a linear decreasing strategy, decreasing from 0.9 to 0.4 during the iteration process.

[0067] υ i(t) Represents particle X i The velocity at time t.

[0068] c1 and c2 represent individual learning and social learning factors, used to control the particle's path towards the individual optimal solution pbest. i The speed at which it approaches the group's optimal solution, gbest.

[0069] rand() represents a random number between [0, 1].

[0070] pbest i Represents particle X i The optimal position of an individual.

[0071] gbest represents the globally optimal position.

[0072] X i(t) Represents particle X i The position at time t.

[0073] d. Check the termination condition: If the termination condition is met, such as the fitness value change being less than the threshold or the maximum number of iterations being reached, then stop the iteration and output the global best position gbest as the optimal electricity purchase and sale strategy. Otherwise, return to step b to continue iterating.

[0074] Therefore, through the above iterative process, the optimal power purchase and sale strategy can be found while satisfying the constraints of the power purchase and sale strategy model. This strategy can maximize the value of the objective function, thereby improving the efficiency and effectiveness of power trading while mitigating risks.

[0075] It should be noted that the present invention can use particle swarm optimization algorithm to solve the above-mentioned electricity purchase and sale strategy model, or it can use other algorithms to solve the above-mentioned electricity purchase and sale strategy model, such as genetic algorithm. Those skilled in the art can determine the appropriate algorithm based on the actual situation, and the present invention does not impose any specific limitations.

[0076] It should be further noted that if the high-risk comprehensive membership degree and low-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector fail to meet the risk membership degree constraint, then there is no need to solve the model. In this case, the power purchase and sale strategy should be implemented in accordance with the "Model Text of Medium and Long-Term Power Transaction Contract".

[0077] Step S106: Construct a test dataset to verify the accuracy of the electricity purchase and sale strategy decision model, and optimize the electricity purchase and sale strategy decision model based on the verification results; Specifically, step S106 includes the following steps: Step S1061: Prepare the test dataset; Choose an independent test dataset that contains data the model has not yet seen to ensure the objectivity and impartiality of the validation. The test dataset should cover different market scenarios, including normal market conditions, market volatility, and supply-demand imbalances, to comprehensively evaluate the model's performance.

[0078] Step S1062: Use the electricity purchase and sale strategy decision model to predict the test dataset and generate an electricity purchase and sale strategy; It should be noted that the model must be free from any human intervention during the prediction process to maintain the objectivity of the validation.

[0079] Step S1063: Compare the predicted data with the actual data; The predictions from the electricity purchase and sale strategy decision-making model were compared with the actual market results in the test dataset. The mean squared error (MSE) metric was used to measure the model's predictive performance.

[0080] For example, the comparison results between predicted data and actual data are shown in Table 3 below: Table 3 Comparison of Predicted and Actual Data

[0081] As described above, the comparison results of predicted data and actual data from January 1, 2024 to January 31, 2024 are provided. The present invention can use mean squared error to evaluate the deviation between predicted data and actual data in January 2024.

[0082] Step S1064: Analyze the performance of the electricity purchase and sale strategy decision model on the test dataset; Specifically, identify the market scenarios in which the model performs well and in which it performs poorly. Analyze the reasons for large prediction errors in the model, such as data quality issues, model structure problems, and parameter setting problems.

[0083] Step S1065: Optimize the power purchase and sale strategy decision model based on the verification results; It should be noted that if the model performance meets the requirements, the model can be considered ready for actual power trading decisions. If the model performance does not meet the requirements, and the deviation between the predicted data and the actual data exceeds a preset threshold, the model needs to be further optimized until the requirements are met. Optimization of the power purchase and sale strategy decision model includes, but is not limited to, adjusting the membership function, adjusting the expert active scoring, adjusting the reference sequence, and adjusting the evaluation index set. For example, for a power market where renewable energy accounts for more than 30%, the evaluation index set can be adjusted to include evaluation indicators such as "wind and solar forecast accuracy." Those skilled in the art can determine the appropriate optimization methods based on the actual situation, and all of the above optimization methods are within the scope of protection of this application.

[0084] The above verification process ensures the effectiveness and reliability of the power purchase and sale strategy decision-making model in practical applications, thereby improving the efficiency and effectiveness of power trading.

[0085] In summary, this invention proposes an auxiliary decision-making method for optimizing electricity purchase and sale strategies. This method determines a set of evaluation indicators and uses grey relational analysis to assess the grey relational degree of each indicator. Based on the grey relational degree, the weight vectors of the evaluation indicators are further determined to comprehensively consider multiple factors affecting the electricity market, such as market share, electricity load, and power generation, while more accurately assessing the impact of each indicator on the electricity market. Subsequently, a fuzzy evaluation matrix of the evaluation indicators is constructed, and combined with the weight vectors of the evaluation indicators, a fuzzy comprehensive evaluation vector is calculated. This vector comprehensively reflects the overall membership degree of multiple evaluation indicators to high, medium, and low risk levels, thereby achieving a comprehensive assessment of market transaction risks. In the decision-making process, this invention employs a particle swarm optimization algorithm to solve the electricity purchase and sale strategy decision model, automatically exploring the optimal electricity purchase and sale strategy, significantly improving the efficiency and accuracy of decision-making. The objective function of the electricity purchase and sale strategy decision model is constructed based on the predicted trends of electricity purchase price, electricity sale price, gas turbine power generation efficiency, gas turbine thermal efficiency, and power generation, as well as the fuzzy comprehensive evaluation vector. Simultaneously, the model's constraints are also set based on the fuzzy comprehensive evaluation vector, thus fully considering uncertainties such as spot market electricity prices, next-day user electricity consumption, and the power generation efficiency, thermal efficiency, and power generation of the integrated power generation and sale company's own gas turbines, avoiding transaction risks arising from these uncertainties. Through an intelligent decision-making process, this invention can significantly improve the efficiency and effectiveness of electricity trading, reduce human intervention and errors, accelerate decision-making, and thereby increase the success rate of transactions. Furthermore, this method can flexibly adapt to the needs of electricity purchase and sale decisions under different market conditions, including normal market environments, market fluctuations, and supply-demand imbalances, demonstrating strong flexibility and adaptability. In summary, this invention provides a novel, efficient, and accurate decision-making method.

[0086] Example 2 This invention provides an electronic device, comprising: a memory and a processor; The memory is used to store programs; The processor is used to call a program stored in the memory to execute a power purchase and sale strategy optimization auxiliary decision-making method as described in Embodiment 1.

[0087] Example 3 The present invention provides a readable storage medium storing a computer program, which, when executed by a processor, implements a power purchase and sale strategy optimization auxiliary decision-making method as described in Embodiment 1.

[0088] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A method for optimizing power purchase and sale strategies, characterized in that, Specifically, the following steps are included: Collect historical and real-time data on the electricity market and power purchase and sale entities, and preprocess the data; Historical and real-time data are input into the sequence model to predict the changing trends of electricity purchase and sales prices, as well as the changing trends of the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the electricity purchase and sales entities themselves. Determine the set of evaluation indicators, use grey relational analysis to evaluate the grey relational degree of each evaluation indicator, and determine the weight vector of the evaluation indicators based on the grey relational degree of the evaluation indicators. Construct a fuzzy evaluation matrix for the evaluation indicators, and determine the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix. A power purchase and sale strategy decision model is constructed. The objective function of the model is based on the predicted power purchase price, power sale price, the changing trends of the gas turbine power generation efficiency, gas turbine heat generation efficiency, and power generation of the power purchase and sale entity, as well as the fuzzy comprehensive evaluation vector. The constraints of the model are constructed based on the fuzzy comprehensive evaluation vector. The power purchase and sale strategy decision model is solved to predict the power purchase and sale strategy. The aforementioned power purchase and sale strategy decision model is constructed based on the predicted power purchase price, power sale price, the changing trends of the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the power purchase and sale entity, as well as the fuzzy comprehensive evaluation vector. The constraints of the model are constructed based on the fuzzy comprehensive evaluation vector; Solving the power purchase and sale strategy decision model, the predicted power purchase and sale strategies include: Define the set of variables for the electricity purchase and sale strategy; Define the objective function; Wherein, objective function ； In the formula, = p_sell represents the electricity sales price predicted by the sequence model; p_buy represents the electricity purchase price predicted by the sequence model; x_1 represents the electricity purchase volume solved by the model; x_2 represents the electricity sales volume solved by the model; b1 represents the high-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; b2 represents the medium-risk comprehensive membership degree corresponding to the fuzzy comprehensive evaluation vector; λ_1 and λ_2 are constants. As an efficiency reward coefficient, This represents the gas turbine power generation efficiency predicted by the sequence model; This represents the gas turbine heat production efficiency predicted by the sequence model; This represents the theoretical maximum power generation efficiency of the gas turbine; This represents the theoretical maximum heat production efficiency of the gas turbine. Set constraints; x_1*p1 <x_2<x_1*p2; Q_1max*p3≦x_1≦Q_1max*p4, where Q_1max represents the market's power generation capacity; b1≦k1; b3≧k2; Where b1 and b3 represent the high-risk and low-risk comprehensive membership degrees corresponding to the fuzzy comprehensive evaluation vector, respectively; p1, p2, p3, p4 and k1, k2 are all constants; x_1≧0, x_2≧0; The objective function is solved using the particle swarm optimization algorithm to predict the electricity purchase and sale strategy.

2. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 1, characterized in that: The process of inputting historical and real-time data into the sequence model to predict the changing trends of electricity purchase and sales prices, as well as the changing trends of the gas turbine power generation efficiency, gas turbine heat production efficiency, and power generation of the electricity purchase and sales entities, specifically includes: When the sequence model is an ARIMA time series model, the collected historical and / or real-time electricity purchase prices are input into the ARIMA time series model, and the trend of electricity purchase price changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time electricity sales prices are input into the ARIMA time series model, and the trend of electricity sales price changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time gas turbine power generation efficiency is input into the ARIMA time series model, and the trend of gas turbine power generation efficiency changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time power generation is input into the ARIMA time series model, and the trend of power generation changes is predicted through curve fitting and parameter estimation; the collected historical and / or real-time power generation is input into the ARIMA time series model, and the trend of power generation changes is predicted through curve fitting and parameter estimation. When the sequence model is a recurrent neural network model, the collected historical electricity purchase price and / or real-time electricity purchase price, historical electricity sales price and / or real-time electricity sales price, historical power generation and / or real-time power generation, historical electricity load and / or real-time electricity load are input into each gated recurrent unit of the trained recurrent neural network model to predict the changing trends of electricity purchase price and electricity sales price. When the sequence model is a recurrent neural network model, the collected historical and / or real-time gas turbine power generation, natural gas consumption, natural gas calorific value and gas turbine power generation efficiency are input into each gated recurrent unit of the trained recurrent neural network model to predict the changing trends of gas turbine power generation efficiency and power generation. When the sequence model is a recurrent neural network model, the collected historical and / or real-time flue gas waste heat, natural gas consumption, natural gas calorific value, and gas turbine heat production efficiency are input into each gated recurrent unit of the trained recurrent neural network model to predict the trend of gas turbine heat production efficiency.

3. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 1, characterized in that: The process of determining the evaluation index set, using grey relational analysis to evaluate the grey relational degree of each evaluation index, and determining the weight vector of the evaluation index based on the grey relational degree specifically includes: Determine the set of evaluation indicators; Calculate the grey relational coefficient for each evaluation indicator; the formula for calculating the grey relational coefficient is: ； in, Represents the grey relational coefficient of the i-th evaluation index; x0(k) represents the data at time k in the reference sequence, where k∈[1,N]; This represents the data at time k in the comparison sequence corresponding to the i-th evaluation index; This represents the absolute difference between the two at time k; ; ρ is a constant; N represents the total number of elements in the reference sequence or comparison sequence; Based on the grey relational analysis of the evaluation indicators, the formula for calculating the weight vector of the evaluation indicators is as follows: ； Let represent the grey relational coefficient of the i-th evaluation index; α is a constant; exp() represents the natural exponential function; Let represent the grey relational coefficient of the j-th evaluation indicator; n is the total number of evaluation indicators.

4. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 1, characterized in that: The construction of the fuzzy evaluation matrix for the evaluation indicators, and the determination of the fuzzy comprehensive evaluation vector based on the weight vector of the evaluation indicators and the fuzzy evaluation matrix, specifically include: Set the fuzzy evaluation matrix R = [r ij ]n×m; where r ij represents the membership degree of the i-th evaluation indicator to the j-th evaluation level; n represents the total number of evaluation indicators, and m represents the total number of evaluation levels; Among them, the membership degree of the evaluation index is calculated based on the membership function or determined by the expert scoring method; By multiplying the weight vector of the evaluation index with the fuzzy evaluation matrix through fuzzy transformation, the fuzzy comprehensive evaluation vector B=WR=(b1,b2,…,b) is calculated. j ,…,b m ); where b j This represents the overall membership degree of each evaluation indicator to the j-th evaluation level.

5. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 4, characterized in that: The evaluation indicators include price fluctuations, supply-demand ratio, and market share; the evaluation levels include high risk, medium risk, and low risk. The membership degree of price fluctuations to high-risk, medium-risk, and low-risk categories is calculated as follows: Based on the purchase prices at adjacent time points, obtain the price volatility δ; Construct membership functions, where the formula for the membership function of price fluctuations is as follows: r 11 =max(0,min(1,(δ-5) / 5)); r 12 =max(0,min(1,(10-δ) / 5,δ / 5)); r 13 =max(0,min(1,(5-δ) / 5)); r 11 r represents the degree to which price volatility belongs to high risk. 12 r represents the degree of membership of price volatility to medium risk. 13 This indicates the degree to which price fluctuations belong to low risk; The membership calculation process for high-risk, medium-risk, and low-risk supply and demand comparisons is as follows: Based on the power generation and power load at the same time, the supply-demand ratio at the corresponding time is calculated; Construct membership functions, where the formula for the membership function of the supply-demand ratio is as follows: r 21 =max(0,min(1,(1-SDR) / 0.2)); r 23 =max(0,min(1,(SDR-0.8) / 0.4)); r 21 Indicates the degree of membership in a supply-demand comparison that carries high risk, r 22 r represents the degree of membership of risk in the supply and demand comparison. 23 This indicates the degree of membership in a low-risk supply-demand comparison.

6. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 1, characterized in that: After solving the electricity purchase and sale strategy decision model and predicting the electricity purchase and sale strategy, the following steps are also included: A test dataset was constructed to verify the accuracy of the electricity purchase and sale strategy decision model, and the model was optimized based on the verification results.

7. The auxiliary decision-making method for optimizing electricity purchase and sale strategies according to claim 6, characterized in that: The construction of the test dataset to verify the accuracy of the electricity purchase and sale strategy decision model, and the optimization of the electricity purchase and sale strategy decision model based on the verification results, specifically includes: Prepare the test dataset; A power purchase and sale strategy decision model is used to predict the test dataset and generate power purchase and sale strategies. Compare the forecast data with the actual data corresponding to the electricity purchase and sale strategy; Analyze the performance of the electricity purchase and sale strategy decision model on the test dataset; The power purchase and sale strategy decision model was optimized based on the verification results.

8. An electronic device, characterized in that, include: Memory and processor; The memory is used to store programs; The processor is configured to call a program stored in the memory to execute a power purchase and sale strategy optimization auxiliary decision-making method as described in any one of claims 1-7.

9. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, implements a power purchase and sale strategy optimization auxiliary decision-making method as described in any one of claims 1-7.

Images