Time-varying price elastic matrix calculation method and system for constructing similar days based on double screening mechanism

Abstract

The invention relates to a time-varying price elastic matrix calculation method and system for constructing similar days based on a double screening mechanism, and the method comprises the following steps: obtaining a power load and feature category data affecting the power load, carrying out the feature screening, obtaining the feature category data strongly related to the load change, and forming an initial load data set; performing similar day screening on the initial load data set based on a grey relational model to obtain a final load data set related to the electricity price intensity; and calculating a power time-varying price elastic coefficient based on the final load data set, and forming a time-varying price elastic matrix. Compared with the prior art, the method has the advantages that price elasticity matrix calculation has higher rationality and accuracy and the like.

Description

Technical Field

[0001] The present invention relates to the field of electric power economic technology, and in particular to a method and system for calculating a time-varying price elasticity matrix of similar days constructed based on a double screening mechanism. Background Art

[0002] With the deepening of electricity market reforms and the rapid development of smart grids, demand-side response has become a key means of regulating the balance between electricity supply and demand and improving the absorption capacity of new energy. As a core tool for quantifying the sensitivity of users' electricity consumption behavior to electricity prices, the price elasticity matrix's calculation accuracy directly affects key decisions such as time-of-use electricity price setting and demand response resource scheduling. However, traditional elasticity matrices are mostly constructed based on static historical data and do not fully consider the dynamic impact of external factors such as meteorological conditions and date types. As a result, problems such as poor timeliness of elasticity coefficients and insufficient representation of industry differences in complex market environments have become increasingly prominent, making it difficult to meet the needs of refined power system regulation in the context of a high proportion of renewable energy access.

[0003] In economics, price elasticity of demand is often used to reflect the degree to which demand responds to price changes. It is mathematically expressed as the percentage change in demand divided by the percentage change in price. Existing research has mainly used the following three methods to calculate the electricity price elasticity matrix:

[0004] The core concept of the traditional elasticity matrix is ​​to construct an elasticity matrix based on the traditional demand curve model to reflect the direct impact of electricity price changes on load during different time periods and the cross-period transmission effect. The calculation method is to calculate the relative rate of change of electricity price and load during different time periods and fit the elasticity matrix using multivariate regression. The disadvantage of the traditional elasticity matrix is ​​that it relies on the assumption of stationarity of historical data and has difficulty handling nonlinear relationships in dynamic market environments.

[0005] 2. An improved price elasticity matrix model based on elastic effect weights: According to the load characteristics, the change in electricity prices in each period has a significantly greater impact on the electricity consumption in this period than in other periods. Therefore, in order to make more accurate predictions, this method introduces elastic influence weight factors to distinguish the importance of electricity prices in different periods, thereby reducing the effects on time periods outside the current period.

[0006] 3. Price elasticity matrix considering multidimensional factors: This method draws on the concept of multidimensional elastic coefficient in material mechanics. It believes that each type of power load indicator will be similar to the force on a material coming from multiple directions and will be affected by multidimensional power price indicators. Based on the traditional power demand price elasticity coefficient, a multidimensional price elasticity coefficient matrix is ​​proposed. The changes in peak and valley power demand are refined into multidimensional load indicators, and the changes in time-of-use electricity prices are refined into multidimensional price indicators. The multidimensional price elasticity coefficient matrix is ​​constructed by mapping the influence between multidimensional load indicators such as the peak-valley difference reduction rate, peak power transfer rate, valley power filling rate, and peak-valley power transfer rate and multidimensional price indicators. This allows for a more comprehensive and targeted exploration of the sensitivity of peak and valley power demand changes in various regions and industries to multidimensional price indicators.

[0007] In existing technologies, the construction of price elasticity matrix faces three challenges:

[0008] First, the explosion of load feature dimensions leads to data redundancy, and interference from non-critical features distorts the elasticity coefficient.

[0009] Second, the selection of similar days relies on manual experience or a single indicator (such as similar temperature), which makes it difficult to accurately capture the spatiotemporal heterogeneity of the electricity price-load dynamic relationship;

[0010] Third, electricity consumption patterns vary significantly across industries, but traditional methods mostly use overall elasticity coefficients, which cannot support the precise design of differentiated electricity price policies. Especially in a highly volatile spot market environment, static elasticity models are difficult to adapt to the nonlinear changes in load response caused by sharp fluctuations in electricity prices. There is an urgent need to establish a data-driven dynamic multi-period elasticity calculation system. Summary of the Invention

[0011] The purpose of the present invention is to provide a method and system for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism to improve the accuracy of elasticity matrix calculation.

[0012] The purpose of the present invention can be achieved by the following technical solutions:

[0013] A method for calculating a time-varying price elasticity matrix of similar days based on a double screening mechanism includes the following steps:

[0014] Obtaining characteristic category data of power load and factors affecting power load, performing feature screening, obtaining characteristic category data that is strongly correlated with load changes, and forming an initial load data set;

[0015] Performing similar day screening on the initial load data set based on a grey correlation model to obtain a final load data set that is strongly correlated with electricity prices;

[0016] Based on the final load data set, a time-varying price elasticity coefficient of electricity is calculated to form a time-varying price elasticity matrix.

[0017] Furthermore, the characteristic category data affecting power load includes meteorological condition characteristics, economic factor characteristics, political factor characteristics and other factor characteristics, wherein the meteorological condition characteristics include temperature, humidity, air pressure, precipitation, light intensity, geographical location, and weather type sub-characteristics.

[0018] The economic factor characteristics include national economy, GDP, and electricity price sub-characteristics.

[0019] The political factor features include social events and incentive policy sub-features.

[0020] The other factor characteristics include seasonal conditions, holiday types, time information, and electricity usage habit sub-characteristics.

[0021] Furthermore, the step of obtaining characteristic category data that is strongly correlated with load changes includes:

[0022] The outliers in the power load and the characteristic category data affecting the power load are replaced by a linear interpolation method, and are processed by a maximum / minimum value normalization method to obtain normalized power load and characteristic category data, wherein the maximum / minimum value normalization expression is:

[0023]

[0024] Where, is the normalized value, x m,n is the original data, x min 、x max are the minimum and maximum values;

[0025] The Pearson correlation coefficient method is used to calculate the Pearson correlation coefficient between each sub-feature in the normalized feature category data and the normalized power load, wherein the calculation expression of the Pearson correlation coefficient is:

[0026]

[0027] Where r is the Pearson correlation coefficient, x i is the i-th sub-feature, is the average value of the sub-features, is the average power load, y i is the power load, n is the number of sub-features;

[0028] The Pearson correlation coefficient is compared with a preset coefficient range to determine feature category data that is strongly correlated with load changes.

[0029] Furthermore, the initial load data set includes temperature, electricity price, whether it is a holiday and power load data.

[0030] Furthermore, the step of obtaining a final load data set that is strongly correlated with electricity prices includes:

[0031] Divide the year into seasons;

[0032] Dividing the initial load data set according to the seasonal division solution to obtain a plurality of sub-data sets, wherein the sub-data sets include the power load of the corresponding quarter and feature category data that is strongly correlated with load changes;

[0033] For each sub-dataset, the data of the last day of the corresponding quarter is used as the comparison sequence x * =[x it,1 ,…,x it,2 ,…,x it,m ], and the rest of the data is used as the reference sequence X=[x t,1 ,…,x t,2 ,…,x t,m ];

[0034] Calculate the characteristic category data of the reference sequence X and the comparison sequence x * The correlation coefficient between the corresponding feature category data in ;

[0035] Calculate the average value of the correlation coefficient at different times, and further calculate the grey correlation degree between the temperature characteristics of the i-th day before the day to be predicted and the temperature characteristics of the day to be predicted;

[0036] The grey correlation degree is compared with a preset correlation degree range, and similar days are selected. The load data set of similar days is further selected as the final load data set that is strongly correlated with the electricity price. If the day to be predicted is a working day, the selected similar day is also a working day.

[0037] Furthermore, the seasonal division results include summer, winter and other seasons, wherein the summer includes July and August, the winter includes December and January, and the other seasons include February, March, April, May, June, September, October and November.

[0038] Furthermore, the calculation expression of the correlation coefficient is:

[0039]

[0040] Where, ξ it is the correlation coefficient at the xth moment on the i-th day before the day to be calculated, max it,m max t,m |x t,m -x it,m | and min it,m min t,m |xt,m -x it,m | are the maximum and minimum values ​​of the difference between the load value of the reference sequence at time t and all the features corresponding to the comparison sequence, λ∈[0,1] is the resolution, the smaller λ is, the greater the difference between the correlation coefficients and the stronger the discrimination ability. t,m is the eigenvalue of the mth characteristic factor at the tth moment, x it,m is the eigenvalue corresponding to the mth characteristic factor at time t on the i-th day before the day to be calculated.

[0041] Furthermore, the calculation expression of the grey relational degree is:

[0042]

[0043] Where R i is the grey relational degree, ξ it is the correlation coefficient.

[0044] Furthermore, the calculation expression of the time-varying price elasticity matrix is:

[0045]

[0046] in:

[0047]

[0048] Where Q i With P i represents the electricity consumption and electricity price in the i-th time period, n represents the total number of time periods in a given time period, ρ i,i The elastic coefficient of electricity price, i.e. the user's single-period response to a certain electricity price, ρ i,j It represents the cross-elasticity coefficient of electricity price, that is, the user's response to a certain electricity price in other time periods. i and j represent different moments, ρ is the time-varying price elasticity coefficient of electricity, Q represents user demand, P represents electricity price, and Δ represents the change in demand and electricity price.

[0049] The present invention also provides a system for calculating a time-varying price elasticity matrix of similar days based on a double screening mechanism, comprising:

[0050] Preliminary screening module: used to obtain characteristic category data of power load and factors affecting power load, perform feature screening, obtain characteristic category data that is strongly correlated with load changes, and form an initial load data set;

[0051] Second screening module: used for screening the initial load data set for similar days based on the grey correlation model to obtain the final load data set that is strongly correlated with the electricity price;

[0052] Elasticity matrix calculation module: used to calculate the time-varying price elasticity coefficient of electricity based on the final load data set to form a time-varying price elasticity matrix.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] (1) The present invention uses a dual screening mechanism of feature screening and grey correlation to effectively eliminate the interference of non-related variables and accurately identify the load response pattern dominated by electricity prices. At the same time, it introduces time-varying similar day construction technology and combines the multi-dimensional feature correction elasticity matrix calculation data set to ultimately reflect the price elasticity differences under different loads. Compared with the traditional price elasticity matrix calculation, it has stronger rationality and accuracy.

[0055] (2) The present invention constructs a similar day plan based on a double screening mechanism, which can accurately eliminate the interference of non-core variables such as meteorological types, and solves the problem of load-electricity price correlation distortion caused by data redundancy in traditional methods.

[0056] (3) The present invention calculates the time-varying price elasticity matrix under different industrial and commercial sectors based on the double-screened data set, and divides the time scale into four periods: peak, peak, flat and valley under seasons and days. The time-varying price elasticity matrix calculated in this way can accurately quantify the spatiotemporal heterogeneity of price sensitivity, and provide a highly reliable decision-making basis for the power market to formulate electricity prices and optimize the scheduling of demand response resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Schematic diagram of the method flow of the present invention;

[0058] Figure 2 This is a flow chart of the Pearson correlation coefficient calculation model of the present invention;

[0059] Figure 3 This is a heat map of the Pearson correlation coefficient results of the present invention;

[0060] Figure 4 Constructing a similar day flow chart for the color correlation calculation model of the present invention;

[0061] Figure 5 This is the electricity consumption in July and August of the present invention. DETAILED DESCRIPTION

[0062] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0063] Example 1

[0064] This embodiment provides a method for calculating a time-varying price elasticity matrix based on a dual screening mechanism to construct similar days. This method is an innovative method that combines a dual screening mechanism with dynamic matching of similar days. Through two-stage data screening using the Pearson correlation coefficient and grey correlation, it effectively removes interference from noise data and accurately identifies load response patterns dominated by electricity prices. At the same time, it introduces time-varying similar day construction technology and combines multi-dimensional features such as meteorology and industry type to correct the elasticity matrix calculation data set. This method can improve the accuracy of calculating the price elasticity matrix.

[0065] Specifically, if Figure 1 As shown, the method includes the following steps:

[0066] Step 1: Collect all feature categories and data that affect the load, and build a feature screening model based on the Pearson correlation coefficient to screen out feature categories that are strongly correlated with load changes.

[0067] First, we collected the characteristic categories and data that affect the load, as shown in Table 1. The factors that affect the load include meteorological conditions, economic factors, political factors, and other factors.

[0068] Table 1 Influencing load characteristics

[0069]

[0070] The characteristic factors affecting load include the above four categories. However, in the sensitivity analysis of electricity price and load, taking all characteristic factors into account in the model will not only not improve the accuracy, but will also introduce weakly correlated factors to reduce the accuracy of the model. Therefore, the Pearson correlation coefficient is introduced to screen the characteristics affecting load.

[0071] When calculating the correlation coefficient between load and features, it is important to consider whether the data is readily available. Comprehensively considering all sub-features of the four major aspects is unrealistic. This embodiment, after referring to relevant materials, determines that the sub-feature data for calculating the correlation in step 1 are electricity price, temperature, humidity, and whether it is a holiday.

[0072] After determining the data type of the input correlation model, all data are preprocessed, and the outliers in the data are replaced by linear interpolation. All data are normalized by maximum / minimum values. Data normalization is to eliminate the dimension of the data and map it to the range of 0 to 1. If m = 1 represents temperature, then the temperature time series of the day to be predicted is x t,1 =[x 1,1 ,x 2,1 ,…,x 24,1 ], with x 1,1 For example, the data normalization can be expressed as the following formula:

[0073]

[0074] Where, is the normalized value, x max and x min They are x t,1 The maximum and minimum values ​​of .

[0075] After all data are normalized, we start to build a feature screening model based on the Pearson correlation coefficient. The flow chart is as follows: Figure 2 shown.

[0076] Factors affecting load include meteorology, historical load, electricity prices, and economic factors. Specifically, they include temperature, humidity, air pressure, rainfall, light intensity, seasonal conditions, weather type, geographic location, holiday type, time of day information, electricity prices, incentive policies, electricity usage habits, social events, and the national economy. However, in the sensitivity analysis of electricity prices and load, taking all factors into account in the model will not only fail to improve accuracy, but will also introduce weakly correlated factors, reducing model accuracy. Therefore, the Pearson correlation coefficient is introduced to screen the characteristics that affect load. The Pearson correlation coefficient formula is as follows:

[0077]

[0078] In the formula, r is the correlation coefficient, x i is the influencing factor, y i is the load. The correlation classification is shown in Table 2:

[0079] Table 2 Correlation coefficients Correlation classification

[0080]

[0081] Compare the result r with the table and select the feature x that is strongly correlated with the load. i Complete step 1; after feature screening, calculate the Pearson correlation coefficient and express it in the form of a heat map, such as Figure 3 shown.

[0082] Step 2: Obtain load, electricity price, and temperature data of sufficient time scale based on the screening results of step 1, and remove non-working day load data to complete the construction of the initial load data set.

[0083] according to Figure 3Correlation coefficient analysis revealed significantly different association patterns between the three industries and various influencing factors. The catering industry exhibited strong positive correlations with holidays (0.832) and temperature (0.742), indicating significant impacts of seasonal demand and holiday consumption, but was significantly inhibited by rising electricity prices (-0.652). The industrial sector exhibited a unique negative correlation with holidays (-0.793), reflecting the impact of shutdowns and production cuts. It was also sensitive to electricity prices (-0.613), while rising temperatures could improve the efficiency of some production processes (0.674). The transportation industry was most affected by electricity price fluctuations (-0.73), but holiday travel demand (0.679) and suitable temperatures (0.622) had a positive impact. All three industries showed that electricity price was the most critical inhibitory factor, while humidity generally had a weaker impact. These differences reflect the fundamental differences in the response mechanisms of different industries to economic, climatic, and temporal factors, providing a quantitative basis for the formulation of differentiated industry policies. Finally, the three categories of temperature, electricity price, weekday status, and industrial and commercial load were selected to form the dataset for Step 2.

[0084] Step 3: Use the grey correlation model to filter the initial load data set to obtain the final load data set that is only strongly correlated with the target electricity price.

[0085] Collect 8760 hours of industrial and commercial load data throughout the year. The seasons are divided into summer (July, August), winter (December, January) and other months (February, March, April, May, June, September, October, November). Taking summer load as an example, the loads of July and August are taken as data set data1. Data set data1 contains load data of all months and is related to the filtered related feature data x i Merge into data set data2, use the last day of data22 as the reference data, and select all feature data x in data2 i,j (j∈[1,24]), use the grey relational analysis (GRA) to filter the date data in the data set data2 that are similar to the reference data, and save their load and feature data to data3. The steps of constructing similar date data using the grey relational analysis model are as follows: Figure 4 shown.

[0086] The specific steps are as follows:

[0087] (1) Constructing a sequence matrix

[0088] This patent uses summer data as an example for calculation, and takes the characteristic factors of the last day of summer as the reference sequence X=[x t,1 ,…,x t,2 ,…,x t,m ], taking other summer day-related characteristic factors as the comparison sequence x * =[xit,1 ,…,x it,2 ,…,x it,m ], which includes temperature, humidity, and whether it is a holiday, m is the type of characteristic factor; x it,m is the eigenvalue corresponding to the tth moment of the mth characteristic factor on the i-th day before the day to be calculated; t is the tth moment of the i-th day, the data collection interval of the characteristic factor time series is 1 hour, t∈[1,24]; x t,m is the eigenvalue corresponding to the tth moment of the mth factor on the day to be determined.

[0089] (2) Calculation of correlation coefficient

[0090] Based on the normalized time series, the correlation coefficient between each characteristic factor in the reference series and the corresponding feature in the comparison series is calculated, which is expressed as follows:

[0091]

[0092] Where: it is the correlation coefficient at time t on the i-th day before the day to be calculated; max it,m max t,m |x t,m -x it,m | and min it,m min t,m |x t,m -x it,m | are the maximum and minimum values ​​of the difference between the load value of the reference sequence at time t and all the features corresponding to the comparison sequence; λ∈[0,1] is the resolution. The smaller λ is, the greater the difference between the correlation coefficients and the stronger the discrimination ability. In this experiment, λ is set to 0.5.

[0093] (3) Calculate the correlation

[0094] By calculating the average value of the correlation coefficient at different times, the grey correlation degree R between the meteorological characteristics of the i-th day before the forecast day and the meteorological characteristics of the forecast day is obtained. i It is expressed as the following formula:

[0095]

[0096] (4) Select similar day feature variables

[0097] Calculate the correlation of day i according to the above steps 1 to 4, and take the correlation R i The load data of the day with a value greater than 0.8 is used as the similar day load variable. Finally, if the day to be predicted is a weekday, only the similar day characteristic variables of the weekday are retained, and vice versa.

[0098] After completing step 3 and obtaining a dataset that is only strongly correlated with the target electricity price characteristics as the final load dataset, the price elasticity coefficient matrix for different industries is calculated below.

[0099] Step 4: Calculate the final price elasticity coefficient.

[0100] The price elasticity matrix coefficient quantifies the ratio of the percentage change in quantity demanded to the percentage change in price. The absolute value of the elasticity coefficient also represents the sensitivity. The elasticity coefficient is expressed as follows:

[0101]

[0102] Where: ρ represents the elasticity coefficient; Q represents the user demand; P represents the electricity price; Δ represents the change in demand and electricity price. Electricity demand is generally affected by prices in different periods, as shown in the following formula:

[0103]

[0104] Where: Q i With P i represents the amount of electricity and the price of electricity in the i-th time period; n represents the total number of time periods in a given time period; it is generally related to the divided time periods; ρ i,i The elastic coefficient of electricity price, i.e. the user's single-period response to a certain electricity price; ρ i,j represents the cross-elasticity coefficient of electricity price, i.e. the response of users to a certain electricity price in other periods; i and j represent different moments. Therefore, the price elasticity matrix of electricity demand in multiple periods is calculated as:

[0105]

[0106] The above is the solution process for calculating the price elasticity coefficient based on the double screening mechanism in step 4. Now we calculate the industrial and commercial price elasticity matrix based on the actual data of a certain city.

[0107] First, industry accounts for a large proportion in industry and commerce, so the industrial load is selected to represent the industrial and commercial load of a city for calculation. The industrial load of a city in July and August in the summer of 2023 is as follows: Figure 5 As shown;

[0108] The 2023 10kV agency electricity purchase price data is used for calculation. The electricity price data is shown in Table 3.

[0109] Table 3: Electricity purchase price for agents in July and August in summer

[0110]

[0111] According to Table 3, the electricity price change rate in each season can be calculated as shown in Table 4:

[0112] Table 4 Electricity price change rate in July and August in summer

[0113]

[0114] Using the measured load data and electricity price data for calculation, we finally obtained the electricity price elasticity matrix under three seasons. The calculation results for a typical summer day for industry and commerce are shown in Table 5.

[0115] Table 5 Price elasticity coefficients at different time periods on a typical day

[0116]

[0117] According to the summer price elasticity matrix, it can be found that the self-price elasticity (diagonal elements) in the price elasticity matrix are all negative, which means that an increase in electricity prices in each period will reduce demand in that period; from the absolute value of the diagonal: peak (-0.258) > peak (-0.206) > flat (-0.179) > valley (-0.129), indicating that users are most sensitive to prices during peak hours and least sensitive during valley hours.

[0118] Cross-price elasticities (off-diagonal elements) are all positive, reflecting a substitution effect between time periods: a price increase in one time period leads to increased demand in other time periods. Substitution intensity decreases: demand increases are greater in adjacent time periods than in peak-time price increases. For example, price increases during peak hours have the greatest impact on peak hours (0.036), followed by off-peak and off-peak hours. The self-elasticity effect is significantly stronger than the cross-elasticity, indicating that users are more inclined to reduce electricity consumption during the current time period rather than shifting electricity consumption en masse to other times.

[0119] The embodiment of the present invention improves data analysis accuracy by fusing multidimensional data, and effectively eliminates interference from non-related variables by constructing a feature screening model and a gray correlation dual screening mechanism. Ultimately, it can reflect the price elasticity differences under different industry loads, and has stronger rationality and accuracy than traditional price elasticity matrix calculation.

[0120] The above method can be summarized into two stages. The first stage proposes a dual-screening mechanism to construct similar day scenarios. This accurately eliminates interference from non-core variables such as weather types, resolving the problem of distorted load-price relationships caused by data redundancy in traditional methods. The second stage calculates a time-varying price elasticity matrix for different industrial and commercial sectors based on the double-screened dataset, dividing the time scale into four categories: peak, flat, and valley periods, based on seasons and days. This calculated time-varying price elasticity matrix accurately quantifies the spatiotemporal heterogeneity of price sensitivity, providing a highly reliable decision-making basis for power market pricing and optimal resource scheduling for demand response.

[0121] The calculation results of the price elasticity matrix show that the self-price elasticity (diagonal elements) are all negative, which means that an increase in electricity prices in each period will reduce demand in that period; from the absolute value of the diagonal: peak (-0.258) > peak (-0.206) > flat (-0.179) > valley (-0.129), indicating that users are most sensitive to prices during peak hours and least sensitive during valley hours; cross-price elasticity (non-diagonal elements) are all positive, reflecting the existence of a substitution effect between time periods: an increase in electricity prices in a certain period leads to an increase in demand in other periods.

[0122] Diminishing substitution intensity: Demand increases are greater during periods adjacent to price fluctuations. For example, price increases during peak hours have the greatest impact on peak hours (0.036), followed by flat and off-peak hours. The self-elasticity effect is significantly stronger than the cross-elasticity, indicating that users are more inclined to reduce electricity consumption during the current period rather than shifting electricity consumption en masse to other times.

[0123] The above calculation results show that the results are in line with expectations and are relatively reasonable.

[0124] Example 2

[0125] This embodiment provides a system for calculating a time-varying price elasticity matrix for similar days based on a dual screening mechanism, including:

[0126] Preliminary screening module: used to obtain characteristic category data of power load and factors affecting power load, perform feature screening, obtain characteristic category data that is strongly correlated with load changes, and form an initial load data set;

[0127] Second screening module: used for screening the initial load data set for similar days based on the grey correlation model to obtain the final load data set that is strongly correlated with the electricity price;

[0128] Elasticity matrix calculation module: used to calculate the time-varying price elasticity coefficient of electricity based on the final load data set to form a time-varying price elasticity matrix.

[0129] The rest is the same as Example 1.

[0130] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0131] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0132] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0133] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0135] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0136] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism, characterized in that: The following steps are involved: Obtaining characteristic category data of power load and factors affecting power load, performing feature screening, obtaining characteristic category data that is strongly correlated with load changes, and forming an initial load data set; Performing similar day screening on the initial load data set based on a grey correlation model to obtain a final load data set that is strongly correlated with electricity prices; Based on the final load data set, a time-varying price elasticity coefficient of electricity is calculated to form a time-varying price elasticity matrix.

2. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 1, characterized in that: The characteristic category data affecting power load includes meteorological condition characteristics, economic factor characteristics, political factor characteristics and other factor characteristics, wherein the meteorological condition characteristics include temperature, humidity, air pressure, precipitation, light intensity, geographical location, and weather type sub-characteristics. The economic factor characteristics include national economy, GDP, and electricity price sub-characteristics. The political factor features include social events and incentive policy sub-features. The other factor characteristics include seasonal conditions, holiday types, time information, and electricity usage habit sub-characteristics.

3. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 2, characterized in that: The step of obtaining characteristic category data that is strongly correlated with load changes includes: The outliers in the power load and the characteristic category data affecting the power load are replaced by a linear interpolation method, and are processed by a maximum / minimum value normalization method to obtain normalized power load and characteristic category data, wherein the maximum / minimum value normalization expression is: Where, is the normalized value, x m,n is the original data, x min 、x max are the minimum and maximum values; The Pearson correlation coefficient method is used to calculate the Pearson correlation coefficient between each sub-feature in the normalized feature category data and the normalized power load, wherein the calculation expression of the Pearson correlation coefficient is: Where r is the Pearson correlation coefficient, x i is the i-th sub-feature, is the average value of the sub-features, is the average power load, y i is the power load, n is the number of sub-features; The Pearson correlation coefficient is compared with a preset coefficient range to determine feature category data that is strongly correlated with load changes.

4. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 1, characterized in that: The initial load data set includes temperature, electricity price, whether it is a holiday and power load data.

5. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 1, characterized in that: The step of obtaining a final load data set that is strongly correlated with electricity prices comprises: Divide the year into seasons; Dividing the initial load data set according to the seasonal division solution to obtain a plurality of sub-data sets, wherein the sub-data sets include the power load of the corresponding quarter and feature category data that is strongly correlated with load changes; For each sub-dataset, the characteristic data of the last day of the corresponding quarter is used as the comparison sequence x * =[x it,1 ,…,x it,2 ,…,x it,m ], other daily characteristic data in the same season are used as reference sequence X=[x t,1 ,…,x t,2 ,…,x t,m ]; Calculate the characteristic category data of the reference sequence X and the comparison sequence x * The correlation coefficient between the corresponding feature category data in ; Calculate the average value of the correlation coefficient at different times, and further calculate the grey correlation degree between the temperature characteristics of the i-th day before the day to be predicted and the temperature characteristics of the day to be predicted; The grey correlation degree is compared with a preset correlation degree range, and similar days are selected. The load data set of similar days is further selected as the final load data set that is strongly correlated with the electricity price. If the day to be predicted is a working day, the selected similar day is also a working day.

6. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 5, characterized in that: The seasonal division results include summer, winter and other seasons, wherein the summer includes July and August, the winter includes December and January, and the other seasons include February, March, April, May, June, September, October and November.

7. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 5, characterized in that: The calculation expression of the correlation coefficient is: Where, ξ it is the correlation coefficient at time t on the i-th day before the day to be calculated, max it,m max t,m |x t,m -x it,m | and min it, m min t,m |x t,m -x it,m | are the maximum and minimum values ​​of the difference between the load value of the reference sequence at time t and all the features corresponding to the comparison sequence, λ∈[0,1] is the resolution, the smaller λ is, the greater the difference between the correlation coefficients and the stronger the discrimination ability. t,m is the eigenvalue of the mth characteristic factor at the tth moment, x it,m is the eigenvalue corresponding to the mth characteristic factor at time t on the i-th day before the day to be calculated.

8. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 5, characterized in that: The calculation expression of the grey relational degree is: Where R i is the grey relational degree, ξ it is the correlation coefficient.

9. The method for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism according to claim 1, characterized in that: The calculation expression of the time-varying price elasticity matrix is: in: Where Q i With P i represents the electricity consumption and electricity price in the i-th time period, n represents the total number of time periods in a given time period, ρ i,i The elastic coefficient of electricity price, i.e. the user's single-period response to a certain electricity price, ρ i,j It represents the cross-elasticity coefficient of electricity price, that is, the user's response to a certain electricity price in other time periods. i and j represent different moments, ρ is the time-varying price elasticity coefficient of electricity, Q represents user demand, P represents electricity price, and Δ represents the change in demand and electricity price.

10. A system for calculating the time-varying price elasticity matrix of similar days based on a double screening mechanism, characterized in that: include: Preliminary screening module: used to obtain characteristic category data of power load and factors affecting power load, perform feature screening, obtain characteristic category data that is strongly correlated with load changes, and form an initial load data set; Second screening module: used for screening the initial load data set for similar days based on the grey correlation model to obtain the final load data set that is strongly correlated with the electricity price; Elasticity matrix calculation module: used to calculate the time-varying price elasticity coefficient of electricity based on the final load data set to form a time-varying price elasticity matrix.

Images