A method and system for predicting total social electricity consumption based on a seasonal-accumulative air temperature index

Abstract

The application relates to a kind of whole society power consumption prediction method and system based on season-cumulative air temperature index.The method comprises the following steps: collecting daily air temperature data, daily whole society power consumption data, monthly air temperature data and monthly whole society power consumption data in historical data; the daily air temperature data is transformed by cumulative effect and season effect to obtain daily air temperature index; the monthly air temperature data is transformed by season effect to obtain monthly air temperature index; the lag period of monthly air temperature index and monthly whole society power consumption data is used as low-frequency prediction variable, and the daily air temperature index and daily whole society power consumption data are used as high-frequency prediction variable; monthly whole society power consumption is predicted by using low-frequency prediction variable and high-frequency prediction variable.The application proposes a mixed-frequency prediction model (MIDAS-MT-DT) of whole society power consumption based on "season-cumulative air temperature index", and the model has higher prediction accuracy compared with existing models.

Description

Technical Field

[0001] This invention relates to the field of macroeconomic forecasting technology, specifically to a method and system for forecasting total electricity consumption based on the "seasonal-cumulative temperature index". Background Technology

[0002] In the field of electricity consumption analysis and forecasting, numerous methods for electricity prediction have been proposed by scholars both domestically and internationally. These methods can be broadly categorized into classical forecasting methods, traditional forecasting methods, and various modern intelligent forecasting methods. Classical forecasting methods include the elasticity coefficient method and the capacity estimation method for business expansion applications. Traditional forecasting methods primarily use annual and monthly forecasts, and their main approaches include time series analysis, regression analysis, and grey forecasting. With the improvement of data processing capabilities, modern intelligent models have been widely applied in the field of electricity consumption forecasting, with forecasts primarily used monthly and daily. The main methods include neural network forecasting, support vector machines, and chaotic theory forecasting, as well as combined forecasting methods.

[0003] In recent years, frequency mixing forecasting has been introduced into the social and economic fields. Its basic principle is to extract information from early high-frequency data before official statistical data is released and to predict the future. The MIDAS model, proposed by Ghysels et al., is currently a widely used frequency mixing forecasting model. Subsequently, many scholars have proposed extended forms of the MIDAS model, such as the MS-MIDAS model and the cointegrated MIDAS model. In addition, there is the MF-VAR model, a frequency mixing forecasting model for estimating joint endogenous variables. Because many industries have a need for frequency mixing forecasting, it has been applied to various fields such as wind power forecasting, rail transit passenger flow forecasting, macroeconomic forecasting, and financial market forecasting. For example, some scholars have used multi-task learning and ensemble decomposition methods to perform frequency mixing forecasting of wind power. The frequency mixing forecasting of passenger flow by Yao Enjian et al. and Bao Lei has greatly improved the emergency response capability of transportation systems in the event of emergencies. Frequency mixing forecasting also has wide applications in macroeconomics and financial markets; for example, Zhang Wei et al. and Ghysels and Sinko have respectively performed frequency mixing forecasting of macroeconomic aggregates and financial market volatility.

[0004] In recent years, big data technology has gradually been applied to research on electricity consumption forecasting. Specifically, Xu Jun et al. (Xu Jun, Xu Wenhui, Zeng Xin, Song Le. Innovation and application of electricity consumption forecasting method based on big data mining [J]. Electric Power Big Data, 2018, 21(10): 57-61.) used meteorological big data and macroeconomic data to predict the monthly electricity consumption of a district in Huzhou City, with a prediction accuracy of 97.2%. Santamouris et al. (Santamouris M, Cartalis C, Synnefa A, et al. On the impact of urban heat island and global warming on the power demand and electricity consumption of buildings—A review [J]. Energy & Buildings, 2015, 98(jul.): 119-124.) found that for every 1 degree Celsius increase in temperature, the peak electricity consumption will increase by 0.45% to 4.6%. Similarly, using meteorological big data, Wei Xiaochuan and Wang Xingang (Wei Xiaochuan, Wang Xingang. Urban power load forecasting based on meteorological big data [J]. Electrical Measurement & Instrumentation, 2021, 58(02):90-95.) also applied deep learning algorithms such as temporal convolutional networks and recurrent convolutional networks to predict the daily electricity consumption in Shanghai, improving the prediction accuracy to 98.2%. Ayub et al (Ayub N, Irfan M, Awais M, et al. Big data analytics for short and medium term electricity load forecasting using AI techniques ensembler [J]. Energies, 2020, 13(19):5193.) applied the GRU-CNN model to predict the daily electricity consumption of the ISO-NE dataset, improving the prediction accuracy by 7% compared to the SOTA benchmark model. Some scholars have also applied some new data sources, such as nighttime illumination remote sensing data, to the electricity consumption prediction model, predicting the annual electricity consumption of an autonomous prefecture in Yunnan.

[0005] Overall, existing methods primarily rely on data from within the power system, using historical electricity price data for modeling and forecasting. Existing literature applying big data for electricity consumption forecasting largely utilizes natural environmental data such as meteorological or remote sensing big data. Most existing literature selects specific seasons and uses temperature data for local sample intervals for forecasting, failing to consider utilizing multi-source, high-frequency big data to improve the forecasting model; therefore, further improvements in forecast accuracy are possible. Summary of the Invention

[0006] To improve the accuracy of existing models in predicting total electricity consumption, this invention provides a method for predicting total electricity consumption based on the "seasonal-cumulative temperature index". This invention utilizes high-frequency temperature data to construct a "seasonal-cumulative temperature index" that considers seasonality and the inertia of electricity consumption behavior, and proposes a mixed-frequency prediction model (MIDAS-MT-DT) based on multi-source big data.

[0007] The technical solution adopted by this invention to solve its technical problem is as follows:

[0008] A method for predicting total electricity consumption based on the seasonal-cumulative temperature index includes the following steps:

[0009] Collect historical data including daily temperature data, daily total electricity consumption data, monthly temperature data, and monthly total electricity consumption data;

[0010] The daily temperature data is transformed by cumulative effect and seasonal effect to obtain the daily temperature index;

[0011] The monthly temperature data is transformed by seasonal effects to obtain the monthly temperature index;

[0012] The monthly total electricity consumption is predicted by using the lag period of the monthly temperature index and the monthly total electricity consumption data as low-frequency predictive variables, and the daily temperature index and the daily total electricity consumption data as high-frequency predictive variables.

[0013] Furthermore, the daily temperature data undergoes both cumulative effect transformation and seasonal effect transformation, wherein the formula for the cumulative effect transformation is:

[0014]

[0015] in, It is the daily temperature index on the i-th day of the m-th month after cumulative effect transformation; T i,m This is the raw daily temperature data for the i-th day of the m-th month; j represents the j-th day prior to the current day, with values ​​of 1, 2, 3, and 4; N is the total number of days in the (m-1)-th month.

[0016] Furthermore, the daily temperature data undergoes both cumulative effect transformation and seasonal effect transformation, wherein the formula for the seasonal effect transformation is:

[0017] For the southern regions south of the Qinling-Huaihe line:

[0018]

[0019] For the northern regions north of the Qinling-Huaihe line:

[0020]

[0021] Among them, SC_T i,m It is the daily temperature index of the i-th day of the m-th month after cumulative effect transformation and seasonal effect transformation, where N is the total number of days in the m-th month.

[0022] Furthermore, the monthly temperature data is transformed under seasonal effects to obtain the monthly temperature index, wherein the formula for the seasonal effect transformation is:

[0023] For the southern regions south of the Qinling-Huaihe line:

[0024]

[0025] For the northern regions north of the Qinling-Huaihe line:

[0026]

[0027] Among them, S_T m It is the seasonally adjusted monthly temperature index for the m-th month, T. m This is the monthly temperature data for the m-th month.

[0028] Furthermore, the method of using low-frequency and high-frequency predictor variables to predict monthly total electricity consumption includes: improving the MIDAS model to obtain a mixed-frequency prediction model for total electricity consumption based on the seasonal-cumulative temperature index, and using this mixed-frequency prediction model for total electricity consumption to predict monthly total electricity consumption.

[0029] Furthermore, the formula used in the mixed-frequency prediction model for total social electricity consumption is:

[0030]

[0031]

[0032] Among them, Y M,t S_T is the total monthly electricity consumption of the society in month t. t Y is the monthly temperature index of month t after seasonal changes. D,N,t SC_T is the daily total electricity consumption of the whole society on the Nth day of the tth month. N,t It is the daily temperature index on day N of month t after cumulative effect transformation and seasonal effect transformation, w i (θ) is the weight polynomial of the high-frequency variables lagged in the MIDAS model, and θ is the estimated parameter of the polynomial. μ、μ j+1 ,β,β j+1 δ are the parameters that the model needs to estimate, and p Y and p XThe optimal lag period for the model is selected using the AIC criterion, where i and j are integers in the summation operator, and u... t+1 It is the random error of the model.

[0033] A system for predicting total electricity consumption based on the seasonal-cumulative temperature index, comprising:

[0034] The data acquisition module is used to collect historical data, including daily temperature data, daily total electricity consumption data, monthly temperature data, and monthly total electricity consumption data.

[0035] The temperature index construction module is used to transform daily temperature data through cumulative effect transformation and seasonal effect transformation to obtain the daily temperature index; and to transform monthly temperature data through seasonal effect transformation to obtain the monthly temperature index.

[0036] The mixed-frequency prediction module is used to predict monthly total electricity consumption by using the lag period of monthly temperature index and monthly total electricity consumption data as low-frequency prediction variables, and daily temperature index and daily total electricity consumption data as high-frequency prediction variables.

[0037] The beneficial effects of this invention are that the MIDAS-MT-DT model has higher prediction accuracy compared to traditional methods, and the seasonal-cumulative temperature index has the ability to improve prediction accuracy. This invention utilizes high-frequency temperature big data to construct a "seasonal-cumulative temperature index" that considers seasonality and the inertia of electricity consumption behavior, which can more accurately reflect electricity consumption behavior affected by temperature. Furthermore, this invention also introduces high-frequency daily data of total social electricity consumption into the prediction model construction, supplementing it with other modes of electricity consumption behavior besides those affected by temperature. To comprehensively apply the above two types of high-frequency big data, this invention proposes a mixed-frequency prediction model for total social electricity consumption based on the "seasonal-cumulative temperature index" (MIDAS-MT-DT), which has higher prediction accuracy compared to existing models. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0039] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0040] Figure 1 This is a flowchart of the method of the present invention. The method of the present invention mainly includes three steps: data acquisition and preprocessing, temperature index construction, and frequency mixing prediction model. Specifically:

[0041] 1) The data collection and preprocessing part mainly requires collecting monthly total electricity consumption data, monthly temperature data, daily temperature data, and daily total electricity consumption data from historical data.

[0042] 2) In the temperature index construction section, monthly temperature data is transformed by seasonal effects to construct the monthly temperature index, and daily temperature data is transformed by cumulative effects and seasonal effects to construct the daily temperature index.

[0043] 3) In the mixed-frequency forecasting model, monthly total electricity consumption data is used as the predicted variable. Low-frequency forecast variables include the monthly temperature index and the lag period of the monthly total electricity consumption data, while high-frequency forecast variables include the daily temperature index and the daily total electricity consumption data. The lag period of the monthly total electricity consumption data refers to the variable value before the point in time to be predicted. For example, if the total electricity consumption for October needs to be predicted, then the total electricity consumption for September of that year would be laged by 1 period, August's by 2 periods, and so on.

[0044] The high-frequency big data used in the mixed-frequency prediction model of total social electricity consumption in this invention includes: 1) daily total social electricity consumption; 2) local temperature data, including the average of daily maximum temperature and monthly average temperature. "Averaging" refers to collecting the daily maximum temperature and monthly average temperature of several neighboring prefecture-level cities, and then averaging the daily maximum temperature and monthly average temperature of these prefecture-level cities. The average of the daily maximum temperature is used as a high-frequency variable, and the average of the monthly average temperature is used as a low-frequency variable.

[0045] Based on seasonal variations, electricity consumption behavior can be analyzed to have the following two characteristics:

[0046] (1) Cumulative effect: Electricity consumption behavior has a certain inertia. The behavior of using air conditioning in the previous few days will often continue for a short period of time. Therefore, the temperature in the previous few days has a certain impact on the electricity consumption on the current day.

[0047] (2) Seasonal effect: When the temperature is higher than the comfortable temperature, industrial production, commerce, and residential sectors all need to use air conditioning for cooling. Meanwhile, in southern regions south of the Qinling-Huaihe line (such as Fujian Province), air conditioning is also needed for heating in winter when the temperature is lower than the comfortable temperature. In summary, summer temperature should be positively correlated with electricity consumption, while winter temperature should be negatively correlated with electricity consumption.

[0048] Based on the two approaches mentioned above, firstly, a cumulative effect transformation is applied to the daily temperature data. Considering that daily electricity consumption may be influenced by the temperatures of the previous four days, and that this influence should gradually weaken over time, the daily temperature index is set as a weighted sum of the temperatures of the previous four days, where the weights decrease exponentially over time. Specifically, the cumulative effect transformation method for daily temperature data is as follows:

[0049]

[0050] in, It is the daily temperature index, T, on the i-th day of the m-th month after cumulative effect transformation. i,m This is the raw daily temperature data for day i in month m, where j represents the j days prior to the current day, and j takes values ​​of 1, 2, 3, and 4. N is the total number of days in month m-1.

[0051] Then, seasonal effects are applied to the monthly temperature data and the daily temperature data that have already undergone cumulative effect transformation. In southern regions, May to September is summer, and temperature is positively correlated with electricity consumption, so this transformation remains unchanged. January to March and October to December are winter, and temperature is negatively correlated with electricity consumption. Therefore, the original temperature data is negatively evaluated, and the negatively evaluated temperature index is shifted upwards to maintain the continuity of the temperature index. Specifically, for monthly temperature data, the following seasonal effect transformation is applied to the original temperature data (applicable to southern regions south of the Qinling-Huaihe line):

[0052]

[0053] Among them, S_T m It is the seasonally adjusted monthly temperature index for the m-th month, T. m This is the monthly temperature data for the m-th month.

[0054] The seasonal effect transformation method (formula (2)) described above can be adjusted according to the seasonal conditions of the predicted region, with the value of m adjusted accordingly. For example, summer in the northern region north of the Qinling-Huaihe line is from June to August, so formula (2) can be adjusted accordingly as follows:

[0055]

[0056] For daily temperature data, the daily temperature index that has undergone cumulative effect transformation... The following seasonal effect transformation is applied (applicable to southern regions south of the Qinling-Huaihe line):

[0057]

[0058] Among them, SC_T i,m It is the daily temperature index of the i-th day of the m-th month after cumulative effect transformation and seasonal effect transformation, where N is the total number of days in the m-th month.

[0059] Similarly, for the northern region north of the Qinling-Huaihe line, formula (3) can be adjusted accordingly:

[0060]

[0061] To comprehensively utilize high-frequency temperature data and high-frequency electricity consumption data, this invention improves upon the MIDAS model of Ghysels et al. (Ghysels E, Santa-Clara P, Valkanov R. The MIDAS touch: Mixed data sampling regression models [R]. UC Los Angeles: Finance. http: / / escholarship.org / uc / item / 9mf223rs, 2004.), proposing a mixed-frequency prediction model for total electricity consumption based on the "seasonal-cumulative temperature index" (MIDAS-MT-DT). The model's formula is as follows:

[0062]

[0063]

[0064] Among them, Y M,t S_T is the total monthly electricity consumption of the society in month t. t Y is the monthly temperature index of month t after seasonal changes. D,N,t SC_T is the daily total electricity consumption of the whole society on the Nth day of the tth month. N,t It is the daily temperature index on day N of month t, after cumulative effect and seasonal effect transformation. i (θ) is the weight polynomial of the high-frequency variables lagged in the MIDAS model, and θ is the estimated parameter of the polynomial. μ、μ j+1 ,β,β j+1 δ are the parameters that the model needs to estimate, and p Y and p X is the optimal lag number for the model selected by the AIC criterion (the AIC criterion, proposed by H. Akaike in 1974, is used to measure the goodness of fit of a statistical model), i and j are integers in the summation operator, u t+1 It is the random error of the model.

[0065] In one embodiment of the present invention, the selected lag weight polynomial is the Almon polynomial, and the specific formula is as follows:

[0066]

[0067] Where i is the order of the high-frequency data in month t in formula (4), and θ1 and θ2 are the parameters of the polynomial.

[0068] The following are the technical results verified using big data on total electricity consumption in Fujian Province and temperatures in nine prefecture-level cities in Fujian Province. Monthly total electricity consumption in Fujian Province was selected as the variable to be predicted. The high-frequency big data used in the prediction model included: 1) daily total electricity consumption in Fujian Province; 2) the temperature data collection area covered Xiamen, Putian, Fuzhou, Nanping, Quanzhou, Ningde, Longyan, Sanming, and Zhangzhou cities in Fujian Province. The average daily maximum temperature and monthly average temperature of these nine cities were taken as the original daily and monthly temperature indices, respectively.

[0069] The model description and model specification for the comparative experiment are shown in Table 1:

[0070] Table 1 Model settings for the comparative experiment

[0071]

[0072] The prediction accuracy metric used in the comparative experiment is the Mean Absolute Percentage Error (MAPE), which is calculated using the following formula:

[0073]

[0074] in, and x t These represent the predicted value and the actual value of the prediction model, respectively.

[0075] To verify the predictive power of the MIDAS-MT-DT model, the prediction accuracy of the MIDAS mixing model with and without daily and monthly temperature indices was compared. The relevant parameters of the model, estimated using empirical data, are explained in the table notes. The model prediction results are shown in Table 2. The first column of the table represents the model's testing period, corresponding to the training period from the start of the sample to the period preceding the testing period; the percentages in the table represent the corresponding model's prediction accuracy index, MAPE.

[0076] Table 2 Prediction Results of Mixing Model

[0077] Model testing period MIDAS MIDAS-MT MIDAS-DT MIDAS-MT-DT 2020.1-2020.3 83.21％ 86.15％ 90.86％ 86.15％ 2020.4-2020.6 96.54％ 93.35％ 94.00％ 98.39％ 2020.7-2020.9 94.45％ 94.91％ 94.30％ 96.29％ 2020.10-2020.11 94.23％ 95.62％ 89.58％ 94.18％

[0078] Note: The bold numbers in Table 2 indicate models that have improved prediction accuracy compared to the baseline model.

[0079] The maximum lag order for high-frequency data in the MIDAS model is 15, and the lag order for the monthly temperature index is 2.

[0080] As shown in Table 2, across all test sample periods, the prediction accuracy of the model is significantly improved compared to the traditional mixed-frequency MIDAS model, regardless of whether daily or monthly temperature indices are included. The highest prediction accuracy reached 98.39%. Furthermore, the MIDAS-MT and MIDAS-MT-DT models achieved higher prediction accuracy than the baseline model in three out of the four sample periods; however, the MIDAS-DT model only achieved higher prediction accuracy in one sample period. This indicates that the monthly temperature index has a stronger predictive ability than the daily temperature index. In summary, the results in Table 2 demonstrate that by incorporating high-frequency temperature data and high-frequency electricity consumption data, the MIDAS-MT-DT model proposed in this invention significantly improves the prediction accuracy of total electricity consumption and has good application value.

[0081] The cumulative effect transformation method (formula (1)) in the "seasonal-cumulative temperature index" of this invention can be adjusted accordingly based on the prediction effect. If this method is applied to other datasets, the value of j can be adjusted according to the training effect.

[0082] Another embodiment of the present invention provides a system for predicting total electricity consumption based on the seasonal-cumulative temperature index, comprising:

[0083] The data acquisition module is used to collect historical data, including daily temperature data, daily total electricity consumption data, monthly temperature data, and monthly total electricity consumption data.

[0084] The temperature index construction module is used to transform daily temperature data through cumulative effect transformation and seasonal effect transformation to obtain the daily temperature index; and to transform monthly temperature data through seasonal effect transformation to obtain the monthly temperature index.

[0085] The mixed-frequency prediction module is used to predict monthly total electricity consumption by using the lag period of monthly temperature index and monthly total electricity consumption data as low-frequency prediction variables, and daily temperature index and daily total electricity consumption data as high-frequency prediction variables.

[0086] For the specific implementation process of each module, please refer to the description of the method of the present invention above.

[0087] Another embodiment of the present invention provides a computer device (computer, server, smartphone, etc.) including a memory and a processor, the memory storing a computer program configured to be executed by the processor, the computer program including instructions for performing the steps of the method of the present invention.

[0088] Another embodiment of the present invention provides a computer-readable storage medium (such as ROM / RAM, disk, optical disk) storing a computer program that, when executed by a computer, implements the various steps of the method of the present invention.

[0089] The specific embodiments of the present invention disclosed above are intended to help understand the content of the present invention and to implement it accordingly. Those skilled in the art will understand that various substitutions, changes, and modifications are possible without departing from the spirit and scope of the present invention. The present invention should not be limited to the content disclosed in the embodiments of this specification; the scope of protection of the present invention is defined by the claims.

Claims

1. A method for predicting total electricity consumption based on seasonal-cumulative temperature index, characterized in that, Includes the following steps: Collect historical data including daily temperature data, daily total electricity consumption data, monthly temperature data, and monthly total electricity consumption data; The daily temperature data is transformed by cumulative effect and seasonal effect to obtain the daily temperature index; The monthly temperature data is transformed by seasonal effects to obtain the monthly temperature index; The monthly temperature index lag period and the monthly total electricity consumption data lag period are used as low-frequency forecast variables, and the daily temperature index and daily total electricity consumption data are used as high-frequency forecast variables. The monthly total electricity consumption is predicted using the low-frequency and high-frequency forecast variables. The daily temperature data is subjected to cumulative effect transformation and seasonal effect transformation, wherein the formula for the cumulative effect transformation is: in, It is the first after the cumulative effect transformation m Month 1 i The daily temperature index for the day; T i,m It is the first m Month 1 i The raw daily temperature data for the day; j Indicates the previous day j day, j The value can be 1, 2, 3, or 4. N For the first m- Total number of days in a month; The daily temperature data is subjected to cumulative effect transformation and seasonal effect transformation, wherein the formula for seasonal effect transformation is: For the southern regions south of the Qinling-Huaihe line: For the northern regions north of the Qinling-Huaihe line: in, It is the first after cumulative effect transformation and seasonal effect transformation m Month 1 i Daily temperature index for the day N For the first m Total number of days in a month; The monthly temperature data is transformed using a seasonal effect to obtain the monthly temperature index. The formula for the seasonal effect transformation is as follows: For the southern regions south of the Qinling-Huaihe line: For the northern regions north of the Qinling-Huaihe line: in, It is the first m Monthly temperature index after seasonal variations. T m It is the first m Monthly temperature data for the month.

2. The method according to claim 1, characterized in that, The method of using low-frequency and high-frequency predictor variables to predict monthly total electricity consumption includes: improving the MIDAS model to obtain a mixed-frequency prediction model for total electricity consumption based on the seasonal-cumulative temperature index, and using this mixed-frequency prediction model for total electricity consumption to predict monthly total electricity consumption.

3. The method according to claim 2, characterized in that, The formula used in the mixed-frequency prediction model for total social electricity consumption is: in, It is the first t Monthly total electricity consumption of the whole society in the month. It is the first t Monthly temperature index affected by menstrual seasonality. This is the daily total electricity consumption of the whole society on the Nth day of the tth month. It is the first after cumulative effect transformation and seasonal effect transformation t Month N Daily temperature index for the day It is the weighted polynomial of the high-frequency variables in the MIDAS model. These are the estimated parameters of the polynomial, and ; , , , , These are the parameters that the model needs to estimate. and It is the optimal lag number for the model selected by the AIC criterion. , It is an integer in the summation operator. It is the random error of the model.

4. The method according to claim 3, characterized in that, Using Almon polynomials, the specific formula is as follows: in, i It is the first t The order of the high-frequency data for the month, These are the parameters of the polynomial.

5. A system for predicting total electricity consumption based on the seasonal-cumulative temperature index, employing the method described in any one of claims 1 to 4, characterized in that, include: The data acquisition module is used to collect historical data, including daily temperature data, daily total electricity consumption data, monthly temperature data, and monthly total electricity consumption data. The temperature index construction module is used to transform daily temperature data through cumulative effect transformation and seasonal effect transformation to obtain the daily temperature index; and to transform monthly temperature data through seasonal effect transformation to obtain the monthly temperature index. The mixed-frequency prediction module is used to predict monthly total electricity consumption by using the lag period of monthly temperature index and monthly total electricity consumption data as low-frequency prediction variables, and daily temperature index and daily total electricity consumption data as high-frequency prediction variables.

6. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program configured to be executed by the processor, the computer program including instructions for performing the method of any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a computer, implements the method according to any one of claims 1 to 4.

Images