Multi-time scale power consumption dynamic analysis method, device and equipment

Abstract

The application discloses a multi-time-scale power consumption dynamic analysis method, device and equipment, relates to the technical field of power load analysis, and comprises the following steps: acquiring a historical daily power consumption sequence and corresponding multi-source covariants; decomposing the historical daily power consumption sequence to obtain a trend component, a seasonal component and a residual component; constructing a correction operator to perform primary correction on the residual component; quantifying a lag effect through a kernel function based on industry operation capacity, and reconstructing the trend component; constructing an impact variable based on a period application capacity to perform secondary correction on the primary corrected residual component; and fusing prediction results of a first prediction model and a second prediction model to obtain a full-society power consumption prediction value, and mapping the full-society power consumption prediction value into power supply and power sale business indexes. The application realizes accurate quantification of industrial structural changes and collaborative analysis of multi-time-scale influencing factors, and significantly improves the accuracy of medium-and short-term power consumption prediction and the degree of connection with power grid business.

Description

Technical Field

[0001] This invention relates to the field of power load analysis technology, and more specifically, to a method, apparatus, and equipment for dynamic analysis of electricity consumption at multiple time scales. Background Technology

[0002] Electricity load forecasting is a core foundation for ensuring the safe and stable operation of the power grid, conducting scientific power dispatch, and making market transaction decisions. With the large-scale grid connection of new energy sources and the rapid development of emerging industries such as high-end manufacturing and data centers, my country's electricity load structure is undergoing profound changes. The electricity consumption growth momentum of emerging industries is strong and its patterns are unique. Their capacity ramp-up cycles have a significant and increasingly powerful structural impact on total electricity consumption, especially short- and medium-term changes. Traditional forecasting methods often treat the total electricity load as a homogeneous whole, relying on historical total data for trend extrapolation. This makes it difficult to quantify and respond to this new growth pattern brought about by the dramatic changes in the internal industrial structure, leading to a severe challenge to forecast accuracy.

[0003] The main drawback lies in the mismatch between the existing model construction logic and the actual dynamics of the current power system. On the one hand, mainstream forecasting models (whether classical time series models or complex machine learning models) generally lack the ability to model the specific development drivers of emerging industries, failing to depict their unique electricity consumption ramp-up patterns based on applied capacity and growth cycles, thus causing forecasting bias at the source. On the other hand, existing methods are insufficient in coordinating the influence of multiple time dimensions, failing to finely coordinate the effects of key driving factors such as industry growth (long-term trend), seasons and holidays (medium-term cycle), and meteorological temperature differences (short-term fluctuations) within a unified framework, making it difficult for forecasting results to simultaneously possess trend stability and fluctuation sensitivity. In addition, the connection between forecast output and actual grid operations (such as power supply and sales plans) is often insufficient. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method, apparatus, and equipment for multi-timescale dynamic analysis of electricity consumption, in order to solve the problems of insufficient modeling of industrial structure changes, low prediction accuracy, and disconnection from power supply and sales operations in traditional methods.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The multi-timescale electricity consumption dynamic analysis method includes the following steps: obtaining historical daily electricity consumption sequences and corresponding multi-source covariates; performing time-series decomposition on the historical daily electricity consumption sequences to obtain trend components, seasonal components, and residual components; constructing a correction operator coupled with holiday and meteorological effects to perform a first correction on the residual components; based on the industry-specific operating capacity, quantifying the lag effect of capacity ramp-up through kernel functions and embedding it into the trend component model to obtain the reconstructed trend components; constructing an impact variable based on periodic capacity requests to perform a second correction on the residuals after the first correction; based on the reconstructed trend components, seasonal components, and the residuals after the second correction, constructing a first prediction model for total social electricity consumption; and based on emerging industry-related data selected from the multi-source covariates, constructing a second prediction model for emerging industry electricity consumption; and integrating the prediction results of the first and second prediction models to obtain the predicted electricity consumption value and mapping it to electricity supply and sales business indicators.

[0007] In a preferred embodiment, the step of obtaining the historical daily electricity consumption sequence and the corresponding multi-source covariates includes: the multi-source covariates include holiday data, meteorological effect data, industry-specific operating capacity, and periodic application capacity; obtaining historical daily electricity consumption and aligning it with a unified daily time base to form a basic electricity consumption sequence; obtaining and encoding holiday status identifiers and meteorological effect data corresponding to the time base; performing spatial aggregation and daily-scale sampling on industry-specific operating capacity data to construct an industry-specific operating capacity sequence; and constructing a periodic application capacity sequence aligned with the time base based on the periodic application capacity.

[0008] In a preferred embodiment, the step of performing time-series decomposition on the historical daily electricity consumption sequence to obtain trend components, seasonal components, and residual components specifically involves: decomposing the basic electricity consumption sequence to obtain initial trend components, seasonal components, and initial residual components; based on the industry-specific operating capacity sequence, verifying the stationarity of the initial trend components to identify periods of trend abrupt changes caused by industrial restructuring; performing robust decomposition during these periods to separate the incremental trend components dominated by the capacity ramp-up effect; merging the incremental trend components with the initial trend components to form the final trend components; and removing the final trend components and seasonal components from the basic electricity consumption sequence to obtain the residual components to be corrected.

[0009] In a preferred embodiment, the construction of the correction operator coupling holidays and meteorological effects to correct the residual components involves: converting the encoded holiday status identifiers and meteorological effect data into dummy variables and meteorological variables, respectively; interacting the dummy variables and meteorological variables to generate an interaction term for quantifying meteorological sensitivity during holidays; using the dummy variables, meteorological variables, and interaction term as joint inputs, performing regularized regression fitting on the residual components to be corrected, and simultaneously calibrating to form the correction operator; and using the correction operator to perform calculations on the residual components to be corrected, correcting and outputting the first corrected residual.

[0010] In a preferred embodiment, the step of quantifying the lag effect of capacity ramp-up based on industry-specific operating capacity using a kernel function and embedding it into trend component modeling specifically includes: calculating the month-on-month growth rate based on the industry-specific operating capacity sequence to obtain a month-on-month growth sequence; performing a convolution operation on the growth sequence using a log-normal-exponential hybrid kernel function to obtain a driving sequence characterizing the smooth ramp-up process of capacity; embedding the driving sequence as an exogenous variable into the modeling equation used to generate the trend component; and simultaneously determining the weights of the modeling equation and the exogenous variable through joint parameter estimation to generate a reconstructed trend component that incorporates the lag effect of industry-specific capacity ramp-up.

[0011] In a preferred embodiment, the step of constructing an impact variable based on the periodic application capacity and performing a secondary correction on the residuals after the first correction specifically includes: calculating the cumulative amount of the periodic application capacity sequence within a preset statistical period to generate an initial impact index; analyzing the time-shift correlation between the initial impact index and the historical residual sequence, constructing a first-order lag transfer function, and inputting the initial impact index into the first-order lag transfer function to output a delayed impact sequence; using the delayed impact sequence as an explanatory variable, performing a constrained linear regression on the first-corrected residuals to estimate and eliminate its systematic bias, thereby obtaining the second-corrected residuals.

[0012] In a preferred embodiment, the construction of the second prediction model specifically involves: based on a preset emerging industry classification, extracting the operating capacity subsequence and application capacity subsequence corresponding to emerging industries from the industry-specific operating capacity sequence and the periodic application capacity sequence; using the operating capacity subsequence and application capacity subsequence as input data, generating trend components, seasonal components, and secondary correction residual components corresponding to emerging industries according to the time-series decomposition, residual correction, and trend reconstruction rules consistent with the first prediction model; and constructing a second prediction model for outputting the predicted electricity consumption value of emerging industries based on the trend components, seasonal components, and secondary correction residual components corresponding to emerging industries.

[0013] In a preferred embodiment, mapping it to electricity supply and electricity sales business indicators specifically involves: establishing linear regression relationships between the predicted total electricity consumption and the actual electricity supply and actual electricity sales based on historical data, and calibrating the electricity supply conversion coefficient and the electricity sales conversion coefficient; multiplying the final predicted total electricity consumption by the electricity supply conversion coefficient and the electricity sales conversion coefficient, and simultaneously outputting the predicted electricity supply and the predicted electricity sales.

[0014] An apparatus for a multi-timescale dynamic analysis method of electricity consumption includes: a data module for acquiring historical daily electricity consumption sequences and corresponding multi-source covariates; a decoupling module for performing time-series decomposition on the historical daily electricity consumption sequences to obtain trend components, seasonal components, and residual components; a correction module for constructing a correction operator coupling holiday and meteorological effects to perform a first correction on the residual components; based on the industry-specific operating capacity, quantifying the lag effect of capacity ramp-up through kernel functions and embedding it into the trend component model to obtain the reconstructed trend components; constructing an impact variable based on periodic capacity requests to perform a second correction on the residuals after the first correction; a modeling module for constructing a first prediction model for total social electricity consumption based on the reconstructed trend components, seasonal components, and the residuals after the second correction, and constructing a second prediction model for electricity consumption in emerging industries based on emerging industry-related data selected from the multi-source covariates; and an output module for integrating the prediction results of the first and second prediction models to obtain electricity consumption prediction values ​​and mapping them to electricity supply and sales business indicators.

[0015] A multi-timescale electricity consumption dynamic analysis device includes a memory and a processor: the memory is used to store a program; the processor is used to execute the program to implement the various steps of the multi-timescale electricity consumption dynamic analysis method.

[0016] This invention constructs a correction operator coupling holidays and meteorological effects to perform a primary correction on the residual components. This directly identifies and removes short-term complex fluctuations caused by the interplay of social activity patterns and natural climate conditions, resulting in a purer and more stable residual sequence. This eliminates noise barriers for subsequent accurate analysis and addresses the prediction biases caused by mixed factors and response lags in traditional methods from a model mechanism perspective. Furthermore, by constructing an impact variable based on periodic application capacity, a secondary correction is performed on the residual components after the primary correction. This identifies and eliminates systemic biases driven by potential growth demand, significantly improving the purity of the prediction results. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the multi-timescale dynamic electricity consumption analysis method of the present invention.

[0018] Figure 2This is a schematic diagram of the device structure for the multi-timescale dynamic power consumption analysis method of the present invention;

[0019] Figure 3 This is a structural block diagram of an exemplary electronic device provided for implementing embodiments of the present disclosure. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1, Figure 1 The present invention provides a method for dynamic analysis of electricity consumption at multiple time scales, comprising the following steps:

[0022] S1, obtain the historical daily electricity consumption sequence and the corresponding multi-source covariates;

[0023] In this embodiment, obtaining the historical daily electricity consumption sequence and the corresponding multi-source covariates includes:

[0024] The multi-source covariates include holiday data, meteorological effect data, industry-specific operating capacity, and periodic application capacity.

[0025] S11, forming the basic electricity consumption sequence. Obtain the raw historical daily electricity consumption data for the past N years from the energy management system of the target area. Use the forward imputation method to repair missing daily values ​​and ensure that all data points are strictly arranged according to the Gregorian calendar date, forming a continuous basic electricity consumption sequence with the day as the basic time unit. , where t is the date index.

[0026] S12, acquire and encode holiday status identifiers and meteorological effect data corresponding to the time base. Based on the published statutory holiday schedule, generate... A complete sequence of holiday status identifiers corresponding to the dates The specific coding rules are as follows: 0 for ordinary weekdays, 1 for weekends, 2 for statutory holidays, and 3 for adjusted workdays. Simultaneously, the daily maximum temperature data for the target area during the same period is extracted from climate monitoring data. To quantify its marginal impact (rather than absolute temperature), it is standardized and graded: the percentiles of temperature over the entire period are calculated (e.g., 25%, 50%, 75%), dividing daily temperatures into four levels, as shown in Table 1, thereby generating a meteorological grade sequence. .

[0027] Table 1

[0028]

[0029] S13, Construct an industry-specific operating capacity sequence. Retrieve monthly operating capacity data for each industry from relevant databases, and sum the operating capacities of sub-sectors within the same major industry category to obtain the total monthly operating capacity for that aggregated industry; for example, aggregate the operating capacity for "high-end manufacturing." Because... The data is daily, while the operational capacity data is monthly, requiring time scale alignment. A linear interpolation method is used to convert the aggregated monthly operational capacity sequence into a smooth daily sequence. Specifically, the operational capacity value at the end of the Mth month is known. Then the operating capacity value for each day of the month. The interpolation is a linearly increasing sequence from the beginning to the end of the month. Therefore, a sequence is constructed that is... The industry-specific operating capacity sequence with consistent time granularity is denoted as , where i represents the i-th aggregation industry.

[0030] S14, Construct the periodic application capacity sequence. Extract the original electricity application records from the power grid company. Each record contains the application date, voltage level, and applied capacity. Using the application date as the baseline, sum the capacities of all low-voltage electricity applications for each day to obtain the daily low-voltage application capacity; similarly, sum them to obtain the daily high-voltage application capacity. Add the daily low-voltage application capacity and the daily high-voltage application capacity to obtain the total daily application capacity. This sequence constitutes the original periodic application capacity sequence.

[0031] S2, perform time-series decomposition on the historical daily electricity consumption series to obtain trend components, seasonal components and residual components;

[0032] In this embodiment, the time-series decomposition of the historical daily electricity consumption series to obtain trend components, seasonal components, and residual components is specifically as follows:

[0033] S21, Decomposition of the basic electricity series. A seasonal decomposition method is used to decompose the basic electricity series. Initial processing is performed. In a preferred embodiment, the STL (Seasonal and Trend decomposition using Loess) method is employed, setting the seasonal cycle parameters S=7 (to capture the weekly cycle) and S=365 (to capture the annual cycle). Through STL decomposition, the... It can be decomposed into three additive components:

[0034]

[0035] in, This represents the initial trend component, reflecting the long-term direction of electricity consumption changes. As a seasonal component, it reflects fixed weekly and annual cyclical fluctuations. The initial residual components contain residual fluctuations and noise that are not explained by trend and seasonality.

[0036] S22, Identification of periods of sudden trend change. Industry-specific operating capacity sequences. Perform first-order difference operations to obtain its daily variation sequence, and calculate the 7-day moving average sequence of the daily variation sequence. This smooths out random fluctuations and forms a smoothed indicator sequence that reflects the short-term momentum of industrial expansion.

[0037] Will As an external reference signal, for the initial trend component Perform change point detection. In a preferred embodiment, the Bai-Perron multi-structure breakpoint test is employed. This test uses... As the dependent variable, with Or its lagged terms are covariates, which are identified through statistical tests. One or more significant structural abrupt changes in the sequence, and the time interval between these abrupt changes, are identified as the period of trend abrupt change caused by industrial restructuring.

[0038] S23, Robust decomposition during trend abrupt change periods. Within each identified trend abrupt change period, the basic electrical quantity sequence... Robust decomposition is performed on this subsequence. Traditional STL decomposition may vaguely assign structural changes to the trend and residuals during abrupt changes. This step aims to more cleanly separate the increments driven by the specific cause of "capacity ramp-up." Decomposition techniques insensitive to outliers are typically employed, such as using robust local regression based on the M-estimate to fit the trend, or iteratively fitting the trend line using the RANSAC (Random Sample Consensus) regression framework, thus resisting the interference of anomalous fluctuations that may exist during abrupt changes. Through robust decomposition, a purer incremental trend component within the abrupt trend period is obtained. .

[0039] S24, Fusion of trend components. Incremental trend components obtained from robustness decomposition. With initial trend components The components are added together to form a final trend component that can comprehensively reflect long-term changes (including stationary changes and structural abrupt changes). During non-mutation periods, the initial trend component This is the final trend component. .

[0040] S25, Generation of the residual components to be corrected. (Based on the basic electrical quantity sequence) By removing the final trend component and the seasonal component, the residual component to be corrected can be obtained. :

[0041]

[0042] S3, construct a correction operator that couples holidays and meteorological effects to correct the residual components once;

[0043] In this embodiment, the construction of the correction operator coupling holidays and meteorological effects to correct the residual components is specifically as follows:

[0044] S31, holiday status identifier sequence Converted into dummy variables that can be used in a linear model. Because It includes multiple categories and uses a k-1 dummy variable encoding scheme to avoid multicollinearity. For example, taking a normal working day (0) as the baseline, the following three dummy variables are created as shown in Table 2. For simplicity, the set of these three dummy variables is denoted as a vector. .

[0045] Table 2

[0046]

[0047] To capture the nonlinear effects of different temperature levels, a meteorological grade sequence will be used. It is treated as an ordered categorical variable and numerically represented. In a preferred embodiment, the grade value is directly used as the meteorological variable, i.e., let... .

[0048] S32, Generation of Interaction Terms. Recognizing that social behavior patterns during holidays alter people's sensitivity to weather conditions (e.g., high temperatures during holidays may lead to a greater increase in air conditioning load), this invention constructs interaction terms to explicitly quantify this coupling effect. Specifically, dummy variables representing the core state of holidays are used... With meteorological variables Perform multiplication interaction to generate interactive items. .

[0049] S33 forms the correction operator. For historical periods... To N, construct the characteristic matrix F, where each row... Includes: holiday dummy variables Meteorological variables and interactive items Historical residual components to be corrected. Using the feature matrix F as the target variable and regularized linear regression as input, a ridge regression is preferably performed, with the following optimization objective:

[0050]

[0051] in, Let be the coefficient vector to be estimated. The regularization strength hyperparameter is determined through cross-validation.

[0052] The coefficient vector obtained after fitting That is, a correction operator is defined. This operator is a linear function. For any given feature vector Its output is This refers to the predicted residuals that should be explained by holidays and weather effects.

[0053] The calibrated correction operator is applied to the residual components of the entire time series to be corrected. For each time point t, a corresponding feature vector is generated based on its holiday and meteorological information. Calculate the predicted effect value at this point. :

[0054]

[0055] From the original residual components to be corrected Subtract the predicted effect value from the middle to obtain the first-corrected residual. .

[0056] S4, based on the operating capacity of different industries, quantifies the lag effect of capacity ramp-up through kernel functions and embeds it into the trend component model to obtain the reconstructed trend component;

[0057] In this embodiment, the step of quantifying the lag effect of capacity ramp-up based on industry-specific operating capacity using a kernel function and embedding it into trend component modeling specifically includes:

[0058] S41, Calculate the month-on-month growth sequence. Based on the industry-specific operating capacity baseline sequence. Calculate the month-on-month growth rate of operating capacity at each time point t relative to 30 days prior (approximately one month prior). This leads to the month-on-month growth sequence. The month-on-month growth rate mentioned The specific calculation formula is as follows:

[0059]

[0060] S42, Generation of the driving sequence. Since the corresponding electricity load does not immediately reach full capacity after production capacity is put into operation, but is gradually released as the production line is debugged and capacity ramps up, this invention uses a specific convolution kernel function to simulate this lag distribution. In this embodiment, a log-normal-exponential hybrid kernel function is used, its mathematical definition being:

[0061]

[0062] in, The components are log-normally distributed to simulate the time lag during the main construction and commissioning phases of capacity ramp-up. The location parameter is a logarithmic scale, representing the typical time scale (median) of the capacity ramp-up process. It is a logarithmic-scale shape parameter that characterizes the uncertainty or dispersion of the climbing cycle; As an exponentially distributed component, it simulates the process by which the impact of peak capacity gradually diminishes with technological, market, or equipment factors. The decay rate parameter characterizes the rate at which the productivity effect decays. It is a mixed weight.

[0063] Month-on-month growth sequence Viewed as a series of "capacity expansion pulses" occurring at different times, the effects of these pulses spreading over time are simulated using convolution operations. The driving sequence is then calculated. :

[0064]

[0065] Where L is the effective length of the kernel function (e.g., 180 days). Sequence This is a smooth, continuous driving sequence that includes standard lag effects, representing the potential pressure on electricity demand from "effective capacity ramp-up".

[0066] S43, the driving sequence is embedded as an exogenous variable in the modeling equation for the trend component. An extended autoregressive integral moving average (ARIMAX) model is used as the basic equation for trend modeling. Let the trend component to be generated be... Construct the following ARIMAX(p,d,q) model structure:

[0067]

[0068] in, These are the autoregressive and moving average operator polynomials, respectively. For difference operators, It is a white noise sequence. This is a new item. As the estimated coefficient of the driving sequence, this term directly introduces the lagged driving effect of industry ramp-up into the trend generation equation.

[0069] S44, Joint parameter estimation and trend component generation. The final trend components... As a training objective, all parameters of the ARIMAX model are simultaneously optimized using maximum likelihood estimation or conditional least squares methods.

[0070] Using the estimated complete model, trend calculations are performed for both the in-sample and forecast periods. For the forecast period, based on known or predicted industry operating capacity data, steps S41-S42 are repeated to generate future driving sequences, which are then substituted into the model to generate a reconstructed trend component that incorporates the lag effect of industry-specific capacity ramp-up. .

[0071] S5, constructing an impact variable based on the periodic application capacity, and performing a second correction on the residual after the first correction;

[0072] In this embodiment, the step of constructing an impact variable based on the periodic application capacity and performing a second correction on the residual after the first correction specifically includes:

[0073] S51, Generate initial impact indicators. Based on the aforementioned periodic application capacity sequence, obtain the daily total application capacity. This has already been reflected in S14 above and will not be repeated here. To characterize the accumulated potential demand pressure, a preset statistical window is defined to reflect the typical cycle of investment intentions transforming into actual demand. In a preferred embodiment, The term "day" is used to characterize the annual investment and construction cycle. The rolling cumulative amount within this window is calculated to generate the initial impact indicator. :

[0074]

[0075] Initial shock index Indicates the deadline t, in the past The total potential demand accumulated within a day that has been submitted but not yet fully converted into electricity load is a leading indicator of future pressure on electricity consumption growth.

[0076] S52, Construction of the first-order hysteresis transfer function and output of the delayed shock sequence. Calculation of the initial shock index. Residual components to be corrected in history (Output from S25) at different lag orders Cross-correlation coefficients :

[0077]

[0078] in, The correlation coefficient calculation function, usually referring to the Pearson correlation coefficient, is used to calculate the degree of linear correlation between two time series.

[0079] Analyze cross-correlation coefficients The changing trend reveals that its impact remains significant even after a lag of tens to hundreds of days, and the correlation strength decays exponentially over time, confirming a delayed and decaying transmission effect on the impact of requested capacity on electricity consumption residuals. Therefore, a first-order lag transfer function is constructed to quantify this transmission law. The core of this function is a decay coefficient. Its physical meaning is to control the retention and release rate of the impact over time. The initial impact index... Input this function, and through iterative computation, generate a delayed impulse sequence. , Essentially The exponentially weighted moving average smooths out the noise in the original indicator and reasonably delays and diffuses its effects over time, thus more accurately representing the dynamic release process of "potential demand pressure" over time.

[0080] S53, obtain the second-corrected residual. Then, use the first-corrected residual... As the dependent variable, the delayed shock sequence As explanatory variables, a linear regression model is established:

[0081]

[0082] in, The intercept is... for coefficient, This is the new residual.

[0083] During model fitting, for regression coefficient Applying nonnegativity constraints ensures the logical direction of correction and improves the robustness of the model. Parameter estimation is performed using least squares with this inequality constraint. Using the estimated model, parameters that can be obtained from... The systematic bias portion of the explanation is removed from the primary correction residual to obtain the secondary correction residual. .

[0084] S6. Based on the reconstructed trend component, seasonal component and residual after secondary correction, a first prediction model for total electricity consumption is constructed. Based on emerging industry-related data selected from multi-source covariates, a second prediction model for emerging industry electricity consumption is constructed.

[0085] In this embodiment, the construction of the first prediction model and the second prediction model is specifically as follows:

[0086] S61, Construct the first prediction model. Then, reconstruct the trend components... Seasonal portion and secondary correction residuals By summing the results, a training sequence of daily electricity consumption for the entire society, optimized throughout the entire process, is reconstructed. This sequence is then used as the training target to train a final prediction model for total electricity consumption, i.e., the first prediction model for total electricity consumption. This model can employ a lightweight machine learning model (such as a gradient boosting tree) or a time series model, and its output is the first predicted value. .

[0087] S62, Construct the second prediction model. Based on the preset emerging industry classification (e.g., "Computer, Communication and Other Electronic Equipment Manufacturing" and "Electrical Machinery and Equipment Manufacturing"), extract the corresponding operating capacity subsequence and application capacity subsequence from the industry-specific operating capacity sequence and periodic application capacity sequence. Use the operating capacity subsequence and application capacity subsequence as input data, and generate the trend component, seasonal component, and secondary correction residual component corresponding to the emerging industry according to the time series decomposition, residual correction, and trend reconstruction rules consistent with the first prediction model. Based on the trend component, seasonal component, and secondary correction residual component corresponding to the emerging industry, reconstruct its electricity consumption sequence, and train an independent prediction model, namely the second prediction model for emerging industry electricity consumption, to obtain the second predicted value. The detailed process of this step is the same as that of constructing the first prediction model and outputting the first predicted value, and will not be repeated here.

[0088] S7 integrates the prediction results of the first prediction model and the second prediction model to obtain the predicted electricity consumption value and maps it into the business indicators of electricity supply and electricity sales.

[0089] In this embodiment, the method of fusing the prediction results of the first prediction model and the second prediction model to obtain the electricity consumption prediction value specifically involves: calculating the rolling average absolute percentage error of the first prediction model and the second prediction model over the most recent K days (e.g., K=90), denoted as... and The model with the smaller the error should receive a higher weight; calculate the dynamic weights of the first prediction model. :

[0090]

[0091] Final forecast of total electricity consumption :

[0092]

[0093] In this embodiment, mapping this to electricity supply and sales volume business indicators specifically involves: collecting the final predicted total electricity consumption sequence and the actual daily electricity supply and sales volume sequences for the same historical period (e.g., the past three years), establishing two simple linear regression relationships, fitting them using the least squares method, and calibrating the electricity supply conversion coefficient, electricity sales conversion coefficient, and intercept. For future forecast days, inputting the final predicted total electricity consumption value into the calibrated linear relationship yields the predicted electricity supply and electricity sales volume values.

[0094] Example 2, Figure 2 An apparatus for multi-timescale dynamic analysis of electricity consumption is presented, including:

[0095] The data module is used to obtain historical daily electricity consumption sequences and corresponding multi-source covariates;

[0096] The decoupling module is used to perform time-series decomposition on historical daily electricity consumption sequences to obtain trend components, seasonal components, and residual components.

[0097] The correction module is used to construct a correction operator that couples holidays and meteorological effects, and performs a correction on the residual components.

[0098] Based on the operating capacity of different industries, the lag effect of capacity ramp-up is quantified by kernel function and embedded into the trend component model to obtain the reconstructed trend component.

[0099] Based on the periodic application capacity, an impact variable is constructed, and the residual after the first correction is then corrected a second time.

[0100] The modeling module is used to construct a first prediction model for total electricity consumption based on the reconstructed trend component, seasonal component, and residual after secondary correction, and to construct a second prediction model for electricity consumption of emerging industries based on emerging industry-related data selected from multi-source covariates.

[0101] The output module is used to integrate the prediction results of the first prediction model and the second prediction model to obtain the electricity consumption prediction value and map it into the electricity supply and sales business indicators.

[0102] Example 3: A multi-timescale power consumption dynamic analysis device, such as... Figure 3 As shown, it includes a memory and a processor: the memory is used to store a program; the processor is used to execute the program to implement any of the embodiments in Example 1.

[0103] Since the multi-timescale power consumption dynamic analysis device described in this embodiment is the same device used to implement the method in Embodiment 1 of this invention, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the method described in Embodiment 1 of this application. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any device used by those skilled in the art to implement the method in this application embodiment falls within the scope of protection of this application.

[0104] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0108] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0109] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic analysis of electricity consumption across multiple time scales, characterized in that, Includes the following steps: Obtain historical daily electricity consumption sequences and corresponding multi-source covariates, including holiday data, meteorological effect data, industry-specific operating capacity, and periodic application capacity. The historical daily electricity consumption series is decomposed into trend components, seasonal components, and residual components. A correction operator coupling holidays and meteorological effects is constructed to correct the residual components once, resulting in a first-corrected residual component. Based on the operating capacity of different industries, the lag effect of capacity ramp-up is quantified by kernel function to obtain the driving sequence characterizing the smooth ramp-up process of capacity. The driving sequence is then embedded into the modeling equation of the trend component to obtain the reconstructed trend component. The shock variable is constructed based on the periodic application capacity. The first-correction residual component is then corrected a second time to obtain the second-correction residual component. The shock variable is the delayed shock sequence obtained by the periodic application capacity after cumulative and time-shift correlation analysis. Based on the reconstructed trend component, seasonal component, and secondary correction residual component, a first prediction model for total electricity consumption is constructed. Based on relevant data of emerging industries selected from multi-source covariates, a second prediction model for electricity consumption of emerging industries is constructed. By integrating the prediction results of the first prediction model and the second prediction model, the predicted electricity consumption value is obtained and the predicted value is mapped into the business indicators of electricity supply and electricity sales.

2. The multi-timescale dynamic analysis method for electricity consumption according to claim 1, characterized in that, The acquisition of historical daily electricity consumption sequences and corresponding multi-source covariates includes: Historical daily electricity consumption is obtained and aligned according to a unified daily time base to form a basic electricity consumption sequence; Acquire and encode holiday status identifiers and meteorological effect data corresponding to the time base; Spatial aggregation and daily-scale sampling are performed on industry-specific operational capacity data to construct industry-specific operational capacity sequences. Based on the periodic application capacity, a periodic application capacity sequence aligned with the time base is constructed.

3. The multi-timescale dynamic electricity consumption analysis method according to claim 2, characterized in that, The time-series decomposition of historical daily electricity consumption sequences yields trend components, seasonal components, and residual components, specifically: The basic electricity sequence is decomposed to obtain the initial trend component, seasonal component, and initial residual component; Based on the industry-specific operating capacity sequence, the stationarity of the initial trend component is tested to identify the period of trend abrupt change caused by industrial restructuring. During periods of abrupt trend change, robustness decomposition is performed to separate the incremental trend component dominated by the capacity ramp-up effect. The incremental trend component is merged with the initial trend component to form the final trend component; The final trend component and seasonal component are removed from the basic electricity series to obtain the residual component to be corrected.

4. The multi-timescale dynamic electricity consumption analysis method according to claim 3, characterized in that, The aforementioned modified operator that couples holidays and meteorological effects performs a correction on the residual components, specifically as follows: The holiday status identifiers and meteorological effect data obtained from the coding are converted into dummy variables and meteorological variables, respectively. Interacting dummy variables with meteorological variables generates interactive terms for quantifying meteorological sensitivity during holidays; Using dummy variables, meteorological variables, and interaction terms as joint inputs, the correction operator is formed by regularizing regression fitting of the residual components to be corrected and simultaneously calibrating. The correction operator is used to perform calculations on the residual components to be corrected, and the corrected residual components are then output.

5. The multi-timescale dynamic electricity consumption analysis method according to claim 4, characterized in that, The reconstructed trend components specifically include: Calculate the month-on-month growth rate of the industry-specific operating capacity sequence to obtain the month-on-month growth sequence; The growth sequence is convolved using a log-normal-exponential hybrid kernel function to obtain a driving sequence characterizing the smooth ramp-up process of production capacity; The driving sequence is embedded as an exogenous variable into the modeling equation used to generate the trend component; By simultaneously determining the role weights of the modeling equations and exogenous variables through joint parameter estimation, a reconstructed trend component that integrates the lag effect of capacity ramp-up in different industries is generated.

6. The multi-timescale dynamic analysis method for electricity consumption according to claim 5, characterized in that, The process of constructing impact variables based on periodic application capacity and performing secondary corrections on the residual components of the first correction specifically includes: The cumulative amount of the periodic application capacity sequence within a preset statistical period is calculated to generate an initial impact indicator. The time-shift correlation between the initial shock index and the historical residual sequence is analyzed, a first-order lag transfer function is constructed, and the initial shock index is input into the first-order lag transfer function to output the delayed shock sequence. Using the delayed shock sequence as the explanatory variable, a constrained linear regression is performed on the first-corrected residual component to estimate and remove the systematic bias of the first-corrected residual component, thereby obtaining the second-corrected residual component.

7. The multi-timescale dynamic electricity consumption analysis method according to claim 6, characterized in that, The construction of the second prediction model is as follows: Based on the pre-defined classification of emerging industries, extract the corresponding operating capacity subsequence and application capacity subsequence from the industry-specific operating capacity sequence and the periodic application capacity sequence; Using the running capacity subsequence and the requested capacity subsequence as input data, and following the time series decomposition, residual correction and trend reconstruction rules consistent with the first prediction model, the trend component, seasonal component and secondary correction residual component corresponding to the emerging industry are generated. Based on the trend component, seasonal component, and secondary correction residual component corresponding to the emerging industries, a second prediction model is constructed to output the predicted electricity consumption value of the emerging industries.

8. The multi-timescale dynamic analysis method for electricity consumption according to claim 7, characterized in that, The process of mapping the predicted values ​​to electricity supply and electricity sales business indicators specifically involves: Based on historical data, linear regression relationships were established between the predicted total electricity consumption and the actual electricity supply and sales, respectively, and the electricity supply conversion coefficient and the electricity sales conversion coefficient were calibrated. The predicted electricity consumption value is multiplied by the power supply conversion coefficient and the power sales conversion coefficient respectively, and the predicted power supply value and the predicted power sales value are output simultaneously.

9. An apparatus for using the multi-timescale dynamic electricity consumption analysis method as described in any one of claims 1-8, characterized in that, include: The data module is used to obtain historical daily electricity consumption sequences and corresponding multi-source covariates, which include holiday data, meteorological effect data, industry-specific operating capacity, and periodic application capacity. The decoupling module is used to perform time-series decomposition on historical daily electricity consumption sequences to obtain trend components, seasonal components, and residual components. The correction module is used to construct a correction operator that couples holidays and meteorological effects, and performs a correction on the residual components to obtain a first-corrected residual component. Based on the operating capacity of different industries, the lag effect of capacity ramp-up is quantified by kernel function to obtain the driving sequence characterizing the smooth ramp-up process of capacity. The driving sequence is then embedded into the modeling equation of the trend component to obtain the reconstructed trend component. The shock variable is constructed based on the periodic application capacity. The first-correction residual component is then corrected a second time to obtain the second-correction residual component. The shock variable is the delayed shock sequence obtained by the periodic application capacity after cumulative and time-shift correlation analysis. The modeling module is used to construct a first prediction model for total electricity consumption based on the reconstructed trend component, seasonal component, and secondary correction residual component, and to construct a second prediction model for electricity consumption of emerging industries based on relevant data of emerging industries selected from multi-source covariates. The output module is used to integrate the prediction results of the first prediction model and the second prediction model to obtain the predicted electricity consumption value and map the predicted value into the business indicators of electricity supply and electricity sales.

10. A multi-timescale dynamic power consumption analysis device, characterized in that, Including memory and processor: The memory is used to store programs; The processor is used to execute the program to implement each step of the multi-timescale dynamic electricity consumption analysis method as described in any one of claims 1-8.

Images