A method for constructing electricity demand prosperity index based on power big data

Abstract

The present invention discloses a method for constructing a power demand prosperity index based on power big data, including the following steps: 1. Collecting monthly electricity data and monthly average temperature data; 2. Screening the monthly electricity data and removing noise data; 3. Converting the monthly electricity data after removing the noise data into a standard synthetic growth rate; 4. Using the principal component analysis method to select effective standard synthetic growth rate data as the independent variable of the prediction model, and inputting it into the pre-built prediction model, the output of the prediction model is the optimal predicted electricity growth rate; 5. Dynamic temperature correction: Establishing a nonlinear model of the monthly average temperature data and the standard synthetic growth rate, and correcting the optimal predicted electricity growth rate according to the average temperature data of the next three periods; 6. Verifying the predicted electricity growth rate result; 7. Compiling a synthetic index; 8. Outputting the power demand prosperity index. The present invention has stronger self-adaptability, low requirements on basic data, and no need to judge and process abnormal values.

Description

Technical Field

[0001] The present invention belongs to the technical field of power data analysis and prediction, and specifically relates to a method for constructing a power demand prosperity index based on power big data. Background Art

[0002] The severe external environment requires the power sector to respond to changes in electricity demand in a timely manner, study electricity consumption patterns, and grasp the forecast direction. By analyzing electricity data with high timeliness and accuracy and deeply exploring the value of data, it is possible to scientifically forecast electricity demand.

[0003] At present, the forecasting methods for electricity demand are mostly focused on learning from past electricity data, building a relatively single forecasting model, and realizing the forecasting of future electricity market development trends. A single forecasting model is effective for forecasting under normal electricity consumption environment, but it cannot respond quickly to sudden shocks. In response to external environmental shocks, existing forecasting methods are mostly based on post-analysis and cannot guide electricity development. In order to solve the above problems, the present invention constructs a "forecasting black box" of multiple forecasting models, selects the optimal numerical forecasting model, and sets a temperature correction module to enable it to have the ability to respond to sudden shock events and temperature changes in a timely manner and automatically adjust, and adopts the "predict quarterly, correct monthly" electricity demand prosperity index compilation model to timely reflect the changes and development trends of electricity demand. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for constructing an electricity demand prosperity index based on power big data in response to the deficiencies in the above-mentioned prior art. The method improves the timely response capability and adaptive adjustment capability of the new electricity demand prosperity index by constructing a prediction model, selecting the optimal numerical prediction model, and setting a temperature correction module. At the same time, the present invention adopts a "predicting quarterly and correcting monthly" electricity demand prosperity index compilation mode to timely predict provincial power development trends, which can provide analytical support for the operation and management of power companies and the government's macro-decision-making.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for constructing a power demand prosperity index based on power big data, characterized in that the method comprises the following steps:

[0006] Step 1: Collect monthly electricity data and monthly average temperature data;

[0007] Step 2: Screen the monthly electricity data and remove the noise data;

[0008] Step 3: Convert the monthly electricity data after removing the noise data into the standard synthetic growth rate;

[0009] Step 4: Use the principal component analysis method to select effective standard synthetic growth rate data as the prediction model independent variable, and input the prediction model independent variable into the pre-built prediction model. The output of the prediction model is the optimal predicted power growth rate;

[0010] Step 5: Dynamic temperature correction: Establish a nonlinear model of monthly average temperature data and standard synthetic growth rate, and correct the optimal predicted electricity growth rate based on the average temperature data of the next three periods;

[0011] Step 6: Verify the predicted power growth rate: When the predicted result is within the range of the average value of the real year-on-year growth rate in the most recent N periods plus or minus two standard deviations, the power growth rate of the prediction model is selected; otherwise, the prediction model is deleted and the prediction model is selected again; when all prediction models are deleted, the most stable simulation walk model is selected as the prediction model; where the value range of N is N≥12;

[0012] Step 7: Compile a composite index: Use the "predicted quarter, monthly revision" electricity demand prosperity index compilation mode to compile a composite index. First, calculate the total amount of electricity in the forecast quarter based on the forecast year-on-year growth rate of electricity, and then compile the leading index weighted year-on-year or the coincident index weighted year-on-year. Then, linearize the leading index weighted year-on-year or the coincident index weighted year-on-year to synthesize the electricity demand prosperity index.

[0013] Step 8: Output the electricity demand prosperity index.

[0014] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: after step eight, it also includes:

[0015] Step 9. Output the monthly forecast electricity and draw a fishbone diagram based on the output results of the electricity demand prosperity index and the monthly forecast electricity.

[0016] The above-mentioned method for constructing an electricity demand prosperity index based on power big data is characterized in that: when the monthly electricity data is screened in step 2 and the noise data is eliminated, the monthly electricity data of the industry containing invalid electricity consumption information is eliminated, and the influence of the Spring Festival effect on the monthly electricity data is eliminated.

[0017] The above-mentioned method for constructing an electricity demand prosperity index based on power big data is characterized in that: the invalid electricity consumption information industry includes industries whose monthly electricity data is less than 100,000 kWh and the data volume is less than 10% of the total data volume.

[0018] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: the calculation formula used in step 3 to convert the monthly power data after removing the noise data into the standard synthetic growth rate is:

[0019]

[0020] Among them, t is the data period and is a positive integer, and the monthly electricity consumption t Represents the monthly electricity data of the tth period, monthly electricity t-12 It represents the monthly electricity data of the t-12 period (i.e. the monthly electricity data of the same period last year). t-24 Indicates the monthly electricity data for the t-24 period (i.e. the monthly electricity data for the same period of the previous year), t-36 represents the monthly electricity data of the t-36th period (i.e. the monthly electricity data of the same period of the previous three years), α is the monthly electricity t-12 The weight coefficient is t-24 The weight coefficient is λ, and λ is the monthly electricity consumption. t-36 The weight coefficients are α, β, λ∈(0,1), α+β+λ=1.

[0021] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: the specific process of selecting effective standard synthetic growth rate data using the principal component analysis method as the independent variable of the prediction model in step 4 is as follows:

[0022] Step 401, eliminate the standard composite growth rate data of invalid industries: eliminate the top 20% of industries with the highest standard deviation of the latest 120 periods of data in the local power database, eliminate the industries whose current standard composite growth rate exceeds the average value of the latest 120 periods plus or minus two standard deviations, and eliminate the industries whose independent variables account for less than 1% of the total social electricity consumption in the latest 3 periods and whose correlation coefficient with the dependent variable in the latest 120 periods is less than 0.8;

[0023] Step 402: Map the n-dimensional standard synthetic growth rate data to the k-dimensional data through principal component analysis to obtain the independent variables of the prediction model; wherein the values ​​of n and k are both non-zero natural numbers and n>k.

[0024] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: in step 5, a nonlinear model of monthly average temperature data and standard synthetic growth rate is established, and the expression of the nonlinear model is:

[0025] Corrected growth rate t - Forecasted growth rate t =k 1 ×[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2

[0026] +k 2 ×[(Temperature t -High temperature thresholdt )-(temperature t-12 -High temperature threshold t )]

[0027] +k 3 ×[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )] 2

[0028] +k 4 ×[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )]+b

[0029] Among them, (temperature t -High temperature threshold t )≥0,(temperature t-12 -High temperature threshold t )≥0,(temperature t - Low temperature threshold t )≥0,(temperature t-12 - Low temperature threshold t )≥0, t is the data period and is a positive integer, temperature t Represents the monthly average temperature data for period t, temperature t-12 Represents the monthly average temperature data for the t-12th period (i.e. the monthly average temperature data for the same period last year), the high temperature threshold t Indicates the high temperature threshold and low temperature threshold of period t t represents the low temperature threshold of period t, k 1 =[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2 The coefficient, k 2 =[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )], k 3 =[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )] 2 The coefficient, k 4 =[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature thresholdt )], b is a constant.

[0030] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: the formula for calculating the predicted quarterly total power volume based on the predicted year-on-year growth rate of power volume in step 7 is:

[0031]

[0032] Among them, the predicted monthly electricity t represents the predicted monthly electricity consumption in period t, the predicted monthly electricity consumption t+1 Represents the predicted monthly electricity consumption in period t+1, predicted monthly electricity consumption t+2 It represents the predicted monthly electricity consumption in the t+2th period, where t is the data period and is a positive integer.

[0033] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: the calculation formula of the weighted year-on-year increase of the leading index in step 7 is:

[0034]

[0035] The calculation formula for the weighted year-on-year coincidence index mentioned in step 7 is:

[0036]

[0037] The time dimension of index calculation is quarterly. It is calculated once every quarter, with four quarters in a year. t represents the current quarter, t-1 represents the previous quarter, t-2 represents the quarter before the day before, and t-4 represents the same period last year. t-1-4 is the same period last year as the previous quarter, and t-2-4 is the same period last year as the quarter before the day before.

[0038] The above-mentioned method for constructing a power demand prosperity index based on power big data is characterized in that: the formula for linearizing the weighted year-on-year increase of the leading index in step 7 is: t = a × weighted year-on-year increase in the leading index t +b, t is the data period and is a positive integer, a>0, b>0, leading index t ≥0, a is the weighted year-on-year increase of the leading index t The coefficient of , b is a constant;

[0039] The formula for linearizing the weighted year-on-year increase in the coincidence index described in step 7 is: t =c×coincidence index weighted year-on-year t +d, t is the data period and is a positive integer, a>0, b>0, coincidence index t ≥0, c is the weighted year-on-year increase of the consistent index t The coefficient of , d is a constant.

[0040] Compared with the prior art, the present invention has the following advantages: the new power demand prosperity index in the present invention has stronger self-adaptability, low requirements on basic data, and no need to judge and process abnormal values. Compared with the traditional power market forecast, it can achieve timely response and automatic adjustment to sudden impact events and temperature changes, improve the reliability of power demand forecast, and broaden the application occasions of power demand forecast.

[0041] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of the method of the present invention;

[0043] Figure 2 The fishbone diagram of the electricity demand prosperity index (Jiangsu) of the present invention;

[0044] Figure 3 The fishbone diagram of the electricity demand prosperity index (primary industry) of the present invention;

[0045] Figure 4 The fishbone diagram of the electricity demand prosperity index (secondary industry) of the present invention;

[0046] Figure 5 The fishbone diagram of the electricity demand prosperity index (tertiary industry) of the present invention;

[0047] Figure 6 The fishbone diagram of the electricity demand prosperity index (urban and rural residents) of the present invention;

[0048] Figure 7 The fishbone diagram of the electricity demand prosperity index (industry) of the present invention;

[0049] Figure 8 This is the fishbone diagram of the electricity demand prosperity index (Nanjing) of the present invention. DETAILED DESCRIPTION

[0050] like Figure 1 As shown, the method for constructing the electricity demand prosperity index based on power big data of the present invention comprises the following steps:

[0051] Step 1: Collect monthly electricity data and monthly average temperature data;

[0052] In specific implementation, the monthly electricity consumption data is the electricity consumption data of all industries in various cities and provinces, and the monthly temperature data is the average temperature data of various cities in the next three months;

[0053] Step 2: Screen the monthly electricity data and remove the noise data;

[0054] In this embodiment, when the monthly electricity consumption data is screened and the noise data is removed in step 2, the monthly electricity consumption data of industries containing invalid electricity consumption information are removed, and the influence of the Spring Festival effect on the monthly electricity consumption data is eliminated;

[0055] In specific implementation, the method for eliminating the influence of the Spring Festival effect on the monthly electricity data is to add the monthly electricity data of January and February of the Gregorian calendar year to obtain the combined electricity data;

[0056] In this embodiment, the invalid electricity consumption information industry includes industries where the monthly electricity consumption data is less than 100,000 kWh and the data volume is less than 10% of the total data volume. For example, in City A from 2000 to 2021, in the catering industry of the tertiary industry, the total data volume of monthly electricity consumption data is 100 groups, of which the data volume of monthly electricity consumption data is less than 100,000 kWh is 95 groups, and only 5 groups of monthly electricity consumption data are higher than 100,000 kWh. Therefore, the catering industry in City A is judged to be an invalid electricity consumption information industry.

[0057] Step 3: Convert the monthly electricity data after removing the noise data into the standard synthetic growth rate;

[0058] In this embodiment, the calculation formula used in step 3 to convert the monthly electricity data after removing the noise data into the standard synthetic growth rate is:

[0059]

[0060] Among them, t is the data period and is a positive integer, and the monthly electricity consumption t Represents the monthly electricity data of the tth period, monthly electricity t-12 It represents the monthly electricity data of the t-12 period (i.e. the monthly electricity data of the same period last year). t-24 Indicates the monthly electricity data for the t-24 period (i.e. the monthly electricity data for the same period of the previous year), t-36 represents the monthly electricity data of the t-36th period (i.e. the monthly electricity data of the same period of the previous three years), α is the monthly electricity t-12 The weight coefficient is t-24 The weight coefficient is λ, and λ is the monthly electricity consumption. t-36 The weight coefficients are α, β, λ∈(0,1), α+β+λ=1.

[0061] The monthly electricity data is processed using the three-year historical average method, and the monthly electricity data after noise data is removed is converted into a standard synthetic growth rate, which can obtain more accurate results.

[0062] Step 4: Use the principal component analysis method to select effective standard synthetic growth rate data as the prediction model independent variable, and input the prediction model independent variable into the pre-built prediction model. The output of the prediction model is the optimal predicted power growth rate;

[0063] In this embodiment, the specific process of using the principal component analysis method to select effective standard synthetic growth rate data as the independent variable of the prediction model in step 4 is as follows:

[0064] Step 401, eliminate the standard composite growth rate data of invalid industries: eliminate the top 20% of industries with the highest standard deviation of the latest 120 periods of data in the local power database, eliminate the industries whose current standard composite growth rate exceeds the average value of the latest 120 periods plus or minus two standard deviations, and eliminate the industries whose independent variables account for less than 1% of the total social electricity consumption in the latest 3 periods and whose correlation coefficient with the dependent variable in the latest 120 periods is less than 0.8;

[0065] Step 402: Map the n-dimensional standard synthetic growth rate data to the k-dimensional data through principal component analysis to obtain the independent variables of the prediction model; wherein the values ​​of n and k are both non-zero natural numbers and n>k.

[0066] In specific implementation, the k-dimensional principal component matrix can reflect more than 80% of the information of the n-dimensional data matrix.

[0067] In specific implementation, the prediction model adopts BP neural network model, CNN convolutional neural network model, support vector machine model, VAR model, simulation walk model, ARIMAX model, machine learning model or time series model.

[0068] Step 5: Dynamic temperature correction: Establish a nonlinear model of monthly average temperature data and standard synthetic growth rate, and correct the optimal predicted electricity growth rate based on the average temperature data of the next three periods;

[0069] In this embodiment, the nonlinear model of the monthly average temperature data and the standard synthetic growth rate is established in step 5. The expression of the nonlinear model is:

[0070] Corrected growth rate t - Forecasted growth rate t =k 1 ×[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2

[0071] +k 2 ×[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )]

[0072] +k 3 ×[(Temperature t - Low temperature threshold t)-(temperature t-12 - Low temperature threshold t )] 2

[0073] +k 4 ×[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )]+b

[0074] Among them, (temperature t -High temperature threshold t )≥0,(temperature t-12 -High temperature threshold t )≥0,(temperature t - Low temperature threshold t )≥0,(temperature t-12 - Low temperature threshold t )≥0, t is the data period and is a positive integer, temperature t Represents the monthly average temperature data for period t, temperature t-12 Represents the monthly average temperature data for the t-12th period (i.e. the monthly average temperature data for the same period last year), the high temperature threshold t Indicates the high temperature threshold and low temperature threshold of period t t represents the low temperature threshold of period t, k 1 =[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2 The coefficient, k 2 =[(Temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )], k 3 =[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )] 2 The coefficient, k 4 =[(Temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )], b is a constant.

[0075] In specific implementation, the high temperature threshold t and low temperature threshold tThe temperature deviation sequence is adjusted in each period, and k1, k2, k3, k4, and b are obtained by regression fitting of the temperature deviation sequence and the corrected growth rate deviation sequence, and are adjusted in each period.

[0076] Step 6. Verify the predicted electricity growth rate: When the predicted result is within the range of the average value of the real year-on-year growth rate in the most recent N periods plus or minus two standard deviations, the electricity growth rate of the prediction model is selected; otherwise, the prediction model is deleted and the prediction model is selected again; when all prediction models are deleted, the most stable simulation walk model is selected as the prediction model; the value range of N is N≥12; in econometrics, the simulation walk model is considered to be the most stable;

[0077] Step 7: Compile a composite index: Use the "predicted quarter, monthly revision" electricity demand prosperity index compilation mode to compile a composite index. First, calculate the total amount of electricity in the forecast quarter based on the forecast year-on-year growth rate of electricity, and then compile the leading index weighted year-on-year or the coincident index weighted year-on-year. Then, linearize the leading index weighted year-on-year or the coincident index weighted year-on-year to synthesize the electricity demand prosperity index.

[0078] In specific implementation, "predict quarterly, revise monthly" means that only the electricity consumption of the current quarter is predicted in each quarter of the Gregorian calendar year, and the predicted quarterly electricity consumption will be revised once a month;

[0079] In this embodiment, the formula for calculating the predicted quarterly total electricity volume based on the predicted year-on-year growth rate of electricity volume in step 7 is:

[0080]

[0081] Among them, the predicted monthly electricity t represents the predicted monthly electricity consumption in period t, the predicted monthly electricity consumption t+1 Represents the predicted monthly electricity consumption in period t+1, predicted monthly electricity consumption t+2 It represents the predicted monthly electricity consumption in the t+2th period, where t is the data period and is a positive integer.

[0082] In this embodiment, the calculation formula of the weighted year-on-year increase of the leading index in step 7 is:

[0083]

[0084] The calculation formula for the weighted year-on-year coincidence index mentioned in step 7 is:

[0085]

[0086] The time dimension of index calculation is quarterly. It is calculated once every quarter, with four quarters in a year. t represents the current quarter, t-1 represents the previous quarter, t-2 represents the quarter before the day before, and t-4 represents the same period last year. t-1-4 is the same period last year as the previous quarter, and t-2-4 is the same period last year as the quarter before the day before.

[0087] In this embodiment, the formula for linearizing the weighted year-on-year increase in the leading index in step 7 is: t = a × weighted year-on-year increase in the leading index t +b, t is the data period and is a positive integer, a>0, b>0, leading index t ≥0, a is the weighted year-on-year increase of the leading index t The coefficient of , b is a constant;

[0088] The formula for linearizing the weighted year-on-year increase in the coincidence index described in step 7 is: t =c×coincidence index weighted year-on-year t +d, t is the data period and is a positive integer, a>0, b>0, coincidence index t ≥0, c is the weighted year-on-year increase of the consistent index t The coefficient of , d is a constant.

[0089] Step 8: Output the electricity demand prosperity index.

[0090] In this embodiment, after step eight, the following steps are further included:

[0091] Step 9. Output the monthly forecast electricity and draw a fishbone diagram based on the output results of the electricity demand prosperity index and the monthly forecast electricity.

[0092] In specific implementation, the electricity demand prosperity index is the electricity demand prosperity index of all industries in all cities and provinces, and the monthly forecast electricity volume is the forecast electricity volume of all industries in all cities and provinces in the next three months.

[0093] The fishbone diagram includes information on the consistent index of new electricity demand, the uncorrected leading index of new electricity demand, the once-corrected leading index of new electricity demand, the twice-corrected leading index of new electricity demand, and the range of the new electricity demand prosperity index.

[0094] In order to verify the technical effect that can be produced by the present invention, the method of the present invention is used to analyze the prosperity index of electricity demand in Jiangsu Province.

[0095] In step 1, the monthly electricity consumption data of Jiangsu Province and 13 cities in all industries and the average temperature forecast data of 13 cities in Jiangsu Province in the next three months are input, as shown in Table 1 and Table 2:

[0096] Table 1 Monthly data on electricity consumption by industry in Jiangsu province and 13 cities

[0097]

[0098] Table 2 Forecast data of average temperature in 13 cities in Jiangsu Province in the next three months

[0099] City Average temperature forecast for this month Average temperature forecast for next month Average temperature forecast for the next two months Nanjing 29.5 27.5 20 Xuzhou 28.5 26 18.5 Changzhou 30 28 20 Wuxi 29 27.5 20 Suzhou 30 28.5 20.5 Nantong 29 28 20 Lianyungang 28 25.5 18.5 Huaian 28.5 26 19 Yancheng 29 27 20 Yangzhou 30 28 20 Zhenjiang 29 27.5 20 Taizhou 28.5 27.5 20.5 Suqian 29.5 26.5 19

[0100] The local electricity database contains electricity consumption data for all industries in Jiangsu Province from 2005 to 2017, among which the industry electricity consumption data needs to be screened to extract valid data.

[0101] In step 2, the monthly electricity data is screened to remove noise data;

[0102] Preferably, if the number of industry electricity consumption data with null values ​​and less than 10 is greater than or equal to 15, the industry is eliminated. For the screened industries, if null values ​​appear in the data of 2018 and later, they are processed with equal intervals. In order to avoid the impact of the Spring Festival on data quality, the data of January and February of each year are merged. The industries initially screened by the basic power data are shown in Table 3:

[0103] Table 3: Industries initially screened by basic power data

[0104]

[0105] In step 3, the monthly electricity data after removing the noise data is converted into a standard synthetic growth rate;

[0106] In order to reduce the impact of sudden external shocks on electricity consumption forecasts, the three-year exponential weighted average method is used to convert it into a standard composite growth rate when predicting the electricity growth rate.

[0107] In step 4, the principal component analysis method is used to select effective standard synthetic growth rate data as the independent variable of the prediction model, and the independent variable of the prediction model is input into the pre-built prediction model. The output of the prediction model is the optimal predicted electricity growth rate. The industries for selecting the growth rate independent variable are shown in Table 4:

[0108] Table 4 Select industries with growth rate independent variables

[0109] Serial number Industry Name Serial number Industry Name 1 Total electricity consumption in society 17 11. Chemical fiber manufacturing industry 2 A. Total electricity consumption of the whole industry 18 12. Rubber and plastic products industry 3 Primary Industry 19 13. Non-metallic mineral products industry 4 Secondary Industry 20 Of which: cement manufacturing 5 B. Total electricity consumption of urban and rural residents 21 14. Ferrous metal smelting and rolling processing industry 6 Town Residents 22 15. Nonferrous metal smelting and rolling processing industry 7 Rural residents 23 16. Metal products industry 8 Electricity consumption by industry 24 17. General and special equipment manufacturing industry 9 1. Agriculture, forestry, animal husbandry and fishery 25 18. Transportation, electrical and electronic equipment manufacturing 10 2. Industry 26 Of which: Transportation equipment manufacturing industry 11 2. Manufacturing 27 19. Handicrafts and other manufacturing industries 12 1. Food, beverage and tobacco manufacturing industry 28 IV. Transportation, warehousing and postal services 13 2. Textile Industry 29 1. Transportation 14 3. Clothing, shoes, hats, leather, down and their products 30 2. Real estate industry 15 5. Paper and paper products industry 31 8. Public Utilities and Management Organizations 16 9. Chemical raw materials and chemical products manufacturing industry 32 3. Education, culture, sports and entertainment

[0110] In step 5, dynamic temperature correction: a nonlinear model of monthly average temperature data and standard composite growth rate is established, and the optimal predicted electricity growth rate is corrected according to the average temperature data of the next three periods;

[0111] Preferably, the low temperature threshold range is 8-15 degrees Celsius, and the high temperature threshold range is 16-31 degrees Celsius to scan the high and low temperature thresholds.

[0112] Scan the high (low) temperature threshold. Generate a high (low) temperature threshold two-dimensional array according to the high (low) temperature threshold range, and scan and calculate it in sequence.

[0113] Table 5 High (low) temperature threshold array table

[0114]

[0115]

[0116] According to the high (low) temperature threshold, the temperature deviation sequence and the growth rate deviation sequence are calculated respectively. The data greater than the high temperature threshold enters the high temperature sequence, and the data less than the low temperature threshold enters the low temperature sequence.

[0117] According to the above two sequences, in the nonlinear model of regression monthly average temperature and electricity consumption growth rate, k 1 , k 2 , k 3 , k 4 , b values, in this embodiment, the least squares method is used for regression.

[0118] In order to ensure that the prediction results are not "abnormally large" or "abnormally small", the prediction results are verified.

[0119] Verification of the predicted power growth rate results in step 6: When the predicted result is within the range of the average value of the real year-on-year growth rate in the most recent N periods plus or minus two standard deviations, the power growth rate of the prediction model is selected; otherwise, the prediction model is deleted and the prediction model is selected again; when all prediction models are deleted, the most stable simulation walk model is selected as the prediction model; where the value range of N is N≥12;

[0120] If the predicted growth rate is less than or equal to the average value of the last 12 periods plus or minus two standard deviations, the growth rate is output. If not, the prediction model is removed from the "Prediction Model" and a new model is selected.

[0121] Step 7: Compile a composite index: Use the "predicted quarter, monthly revision" electricity demand prosperity index compilation mode to compile a composite index. First, calculate the total amount of electricity in the forecast quarter based on the forecast year-on-year growth rate of electricity, and then compile the leading index weighted year-on-year or the coincident index weighted year-on-year. Then, linearize the leading index weighted year-on-year or the coincident index weighted year-on-year to synthesize the electricity demand prosperity index.

[0122] Step 8: Output the electricity demand prosperity index.

[0123] Finally, the electricity demand prosperity index of Jiangsu Province, 13 prefecture-level cities, and 8 categories and the predicted electricity consumption in the next three months are output.

[0124] Table 6 Output table of Jiangsu new electricity demand prosperity index

[0125]

[0126] Table 7 Predicted electricity output in Jiangsu in the next three months

[0127]

[0128]

[0129] Step 8 and beyond also include:

[0130] Step 9. Output the monthly forecast electricity and draw a fishbone diagram based on the output results of the electricity demand prosperity index and the monthly forecast electricity.

[0131] The fishbone diagram of the electricity demand prosperity index (Jiangsu) is as follows: Figure 2 As shown in the fishbone diagram of the electricity demand prosperity index (primary industry) Figure 3 As shown in the fishbone diagram of the electricity demand prosperity index (secondary industry), Figure 4 As shown in the fishbone diagram of the electricity demand prosperity index (tertiary industry), Figure 5 As shown in the fishbone diagram of electricity demand prosperity index (urban and rural residents) Figure 6 As shown in the fishbone diagram of the electricity demand prosperity index (industry), Figure 7 As shown in the fishbone diagram of the electricity demand prosperity index (Nanjing), Figure 8 shown.

[0132] The electricity demand prosperity index in the present invention has stronger adaptive capabilities, and has the ability to respond promptly to sudden impact events and temperature changes and automatically adjust. It can timely reflect the changes and development trends in electricity demand, and provide scientific statistical analysis support for corporate management and government macro-decision-making.

[0133] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0134] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, 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 generate 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 flowchart and / or block diagram. 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.

[0135] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate 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 A function specified in one or more boxes.

[0136] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0137] The foregoing description of specific exemplary embodiments of the present invention is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present invention and various different selections and changes. The scope of the present invention is intended to be limited by the claims and their equivalents.

Claims

1. A method for constructing a power demand prosperity index based on power big data, characterized in that: The method comprises the following steps: Step 1: Collect monthly electricity data and monthly average temperature data; Step 2: Screen the monthly electricity data and remove the noise data; Step 3: Convert the monthly electricity data after removing the noise data into the standard synthetic growth rate; The calculation formula used in step 3 to convert the monthly electricity data after removing the noise data into the standard synthetic growth rate is: Among them, t is the data period and is a positive integer, and the monthly electricity consumption t Represents the monthly electricity data of the tth period, monthly electricity t-12 Indicates the monthly electricity data for the t-12 period, monthly electricity t-24 Indicates the monthly electricity data for the t-24 period, monthly electricity t-36 represents the monthly electricity data of the t-36th period, α is the monthly electricity t-12 The weight coefficient is t-24 The weight coefficient is λ, and λ is the monthly electricity consumption. t-36 The weight coefficient, α, β, λ∈(0,1), α+β+λ=1; Step 4: Use the principal component analysis method to select effective standard synthetic growth rate data as the prediction model independent variable, and input the prediction model independent variable into the pre-built prediction model. The output of the prediction model is the optimal predicted power growth rate; The prediction model adopts BP neural network model, CNN convolutional neural network model, support vector machine model, VAR model, simulation walk model, ARIMAX model, machine learning model or time series model; Step 5: Dynamic temperature correction: Establish a nonlinear model of monthly average temperature data and standard synthetic growth rate, and correct the optimal predicted electricity growth rate based on the average temperature data of the next three periods; Step 6: Verify the predicted power growth rate: When the predicted result is within the range of the average value of the real year-on-year growth rate in the most recent N periods plus or minus two standard deviations, the power growth rate of the prediction model is selected; otherwise, the prediction model is deleted and the prediction model is selected again; when all prediction models are deleted, the most stable simulation walk model is selected as the prediction model; where the value range of N is N≥12; Step 7: Prepare the composite index: First, calculate the total amount of electricity for the quarter according to the predicted year-on-year growth rate of electricity, and then prepare the weighted year-on-year growth of the leading index or the weighted year-on-year growth of the coincident index. Then, linearize the weighted year-on-year growth of the leading index or the weighted year-on-year growth of the coincident index to synthesize the prosperity index of electricity demand. The calculation formula for the weighted year-on-year change of the leading index in step 7 is: The calculation formula for the weighted year-on-year coincidence index mentioned in step 7 is: The time dimension of index calculation is quarterly, which is calculated once every quarter. There are four quarters in a year. t represents the current quarter, t-1 represents the previous quarter, t-2 represents the quarter before that, and t-4 represents the same period last year. t-1-4 represents the same period last year of the previous quarter, and t-2-4 represents the same period last year of the previous quarter. Step 8: Output the electricity demand prosperity index.

2. The method for constructing a power demand prosperity index based on power big data according to claim 1 is characterized by: Step 8 and beyond also include: Step 9. Output the monthly forecast electricity and draw a fishbone diagram based on the output results of the electricity demand prosperity index and the monthly forecast electricity.

3. A method for constructing a power demand prosperity index based on power big data according to claim 1 or 2, characterized in that: When the monthly electricity consumption data is screened and the noise data is removed in step 2, the monthly electricity consumption data of industries containing invalid electricity consumption information are removed, and the influence of the Spring Festival effect on the monthly electricity consumption data is eliminated.

4. The method for constructing a power demand prosperity index based on power big data according to claim 3 is characterized by: The invalid electricity consumption information industries include industries where the monthly electricity consumption data is less than 100,000 kWh and the data volume is less than 10% of the total data volume.

5. A method for constructing a power demand prosperity index based on power big data according to claim 1 or 2, characterized in that: The specific process of using principal component analysis to select effective standard synthetic growth rate data as independent variables of the prediction model in step 4 is as follows: Step 401, eliminate the standard composite growth rate data of invalid industries: eliminate the top 20% of industries with the highest standard deviation of the latest 120 periods of data in the local power database, eliminate the industries whose current standard composite growth rate exceeds the average value of the latest 120 periods plus or minus two standard deviations, and eliminate the industries whose independent variables account for less than 1% of the total social electricity consumption in the latest 3 periods and whose correlation coefficient with the dependent variable in the latest 120 periods is less than 0.8; Step 402: Map the n-dimensional standard synthetic growth rate data to the k-dimensional data through principal component analysis to obtain the independent variables of the prediction model; wherein the values ​​of n and k are both non-zero natural numbers and n>k.

6. A method for constructing a power demand prosperity index based on power big data according to claim 1 or 2, characterized in that: In step 5, a nonlinear model of monthly average temperature data and standard composite growth rate is established. The expression of the nonlinear model is: Corrected growth rate t - Forecasted growth rate t =k1×[(temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2 +k2×[(temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] +k3×[(temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )] 2 +k4×[(temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )]+b Among them, (temperature t -High temperature threshold t )≥0,(temperature t-12 -High temperature threshold t )≥0,(temperature t - Low temperature threshold t )≥0,(temperature t-12 - Low temperature threshold t )≥0, t is the data period and is a positive integer, temperature t Represents the monthly average temperature data for period t, temperature t-12 Represents the monthly average temperature data for the t-12 period, high temperature threshold t Indicates the high temperature threshold and low temperature threshold of period t t represents the low temperature threshold of period t, k1 is [(temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )] 2 The coefficient of k2 is [(temperature t -High temperature threshold t )-(temperature t-12 -High temperature threshold t )], k3 is [(temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )] 2 The coefficient of k4 is [(temperature t - Low temperature threshold t )-(temperature t-12 - Low temperature threshold t )], b is a constant.

7. A method for constructing a power demand prosperity index based on power big data according to claim 1 or 2, characterized in that: The formula for calculating the total quarterly electricity volume based on the predicted year-on-year growth rate of electricity volume described in step 7 is: Among them, the predicted monthly electricity t represents the predicted monthly electricity consumption in period t, the predicted monthly electricity consumption t+1 Represents the predicted monthly electricity consumption in period t+1, predicted monthly electricity consumption t+2 It represents the predicted monthly electricity consumption in the t+2th period, where t is the data period and is a positive integer.

8. A method for constructing a power demand prosperity index based on power big data according to claim 1 or 2, characterized in that: The formula for linearizing the weighted year-on-year increase of the leading index in step 7 is: t = a × weighted year-on-year increase in the leading index t +b, t is the data period and is a positive integer, a>0, b>0, leading index t ≥0, a is the weighted year-on-year increase of the leading index t The coefficient of , b is a constant; The formula for linearizing the weighted year-on-year increase in the coincidence index described in step 7 is: t =c×coincidence index weighted year-on-year t +d, t is the data period and is a positive integer, a>0, b>0, coincidence index t ≥0, c is the weighted year-on-year increase of the consistent index t The coefficient of , d is a constant.

Images