Modeling method and system for influence of energy price fluctuation on industrial energy consumption structure

Abstract

According to the modeling method for the influence of the energy price fluctuation on the industrial energy consumption structure and the system constructed based on the method, a multi-model integrated technical framework is constructed, quantitative modeling technologies of a VAR model, a CGE model and an integrated learning prediction model are integrated, the economic interpretation and macroscopic analysis ability of the VAR model and the CGE model are reserved, and the energy consumption performance of the industrial energy consumption structure is improved. The method can deeply analyze a macroeconomic conduction mechanism of price impact and inter-industry resource allocation changes, and can also improve prediction precision and flexibility by using a machine learning method, so as to meet the actual demands of power insurance and supply policy formulation and energy consumption structure optimization in a complex market environment.

Description

Technical Field

[0001] The present invention relates to the field of energy economic modeling, and particularly to a modeling method and system for the impact of energy price fluctuations on industrial energy use structure. Background Art

[0002] In the context of global energy transition and climate change response, the prices of primary energy and electricity are constantly fluctuating, which has a significant impact on the energy use structure of the industrial sector. In recent years, with the adjustment of China's economic structure and the promotion of green and low-carbon policies, how to quantitatively analyze and predict the impact of energy price shocks on industrial energy use structure has become a key task faced by energy and economic decision-making departments. Traditional models (such as simple regression analysis or single equilibrium models) often have difficulty in comprehensively depicting the complex non-linear relationship between price and structural changes, and decision-makers urgently need high-precision and scalable modeling tools.

[0003] Although single econometric models (such as simple regression or vector autoregressive VAR models) can capture the dynamic relationship between prices, they are difficult to include the changes in industrial structure at the macro level and lack a panoramic description of the changes in economic equilibrium under multi-dimensional policy shocks.

[0004] CGE models can better simulate the interactions and equilibrium changes among various sectors in the macro economy, but they have limitations in parameter setting and non-linear processing, and at the same time have insufficient ability for short-term non-linear complex relationships and high-precision predictions.

[0005] Machine learning models such as random forests have improved in prediction accuracy, but their economic interpretability is not strong and it is difficult to independently become the main tool for economic policy evaluation.

[0006] The deficiencies of the above-mentioned existing technologies lie in the lack of a comprehensive method that organically combines economic interpretation, macro equilibrium analysis and high-precision prediction models, so that it is impossible to fully meet the decision-making needs when facing complex energy price fluctuations. Summary of the Invention

[0007] Aiming at the problems of single models, insufficient interpretation and prediction capabilities in the existing technologies, the present invention proposes a modeling method and system for the impact of energy price fluctuations on industrial energy use structure, and specifically adopts the following technical solutions:

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] The first aspect of the present invention discloses a modeling method for the impact of energy price fluctuations on industrial energy use structure, including the following steps:

[0010] S1. Collect historical energy price data, electricity price data and relevant macroeconomic indicator data, and perform data cleaning, standardization and seasonal adjustment to construct a data set;

[0011] S2. Based on the constructed dataset, use the VAR model to analyze the dynamic conduction relationship between energy prices and electricity prices, and obtain the time-lag effect and conduction path of price shocks;

[0012] S3. Take the conduction path of price shocks output by the VAR model as exogenous shock conditions and input them into the CGE model. Based on the social accounting matrix and inter-sector input-output data, conduct a macro-equilibrium solution for the resource reallocation among industries, changes in energy demand, and output impacts, and obtain the preliminary simulation results of the changes in the energy consumption structure of each industrial sector under price shocks;

[0013] S4. Using the preliminary simulation results of the CGE and historical data as feature inputs, adopt the integrated learning method of the random forest model and the XGBoost model to conduct refined prediction and error correction on the proportions of various energies in the industrial energy consumption structure and electricity demand;

[0014] S5. Introduce policy parameters, simulate various policy combinations based on the integrated model, and quantitatively evaluate the impacts of different policy superpositions on the industrial energy consumption structure and the balance between electricity supply and demand;

[0015] S6. Synthesize the results of each model and output the prediction results of the industrial energy consumption structure under price shocks and policy combinations and a report on policy recommendations.

[0016] Furthermore, the historical energy price data includes coal price data, natural gas price data, and oil price data, and the relevant macroeconomic indicator data includes industrial output data, GDP data, and energy consumption data.

[0017] Furthermore, step S2 specifically includes:

[0018] Conduct a stationarity test on the energy price data series and relevant macroeconomic indicator data in the dataset, and perform stationarity processing on the non-stationary data by means of differencing or logarithmic transformation;

[0019] Select the lag order of the VAR model through information criteria;

[0020] Based on the determined order, use the least squares method to estimate the parameters of the VAR model, and conduct residual autocorrelation test, normality test, and stability test on the estimated model;

[0021] Through impulse response function analysis, obtain the dynamic response of electricity prices in multiple periods when a certain energy price is shocked, quantitatively evaluate the lagging impact on electricity prices, and obtain the conduction path and duration of price shocks;

[0022] Plot the conduction path and lag effect of the obtained electricity price on the energy price shock over time as an electricity price response curve, which together serve as the output of the VAR model.

[0023] Further, step S3 specifically includes:

[0024] Establish a social accounting matrix, use the SAM matrix containing information on the output, input, factor distribution, and income distribution of each department as the benchmark database, determine the elastic parameters, and set various external parameters under the benchmark year or benchmark scenario;

[0025] Input the conduction path of the price shock output by the VAR model as an exogenous variable into the electricity sector or the entire energy market to obtain the impacts on production costs, factor inputs, and product prices;

[0026] Use the numerical iteration method to solve the CGE model; during the iteration process, the model continuously adjusts the sectoral output, factor inputs, and prices until the supply - demand balance conditions in the commodity market and factor market are simultaneously satisfied, and the system converges to a general equilibrium solution;

[0027] Extract the changes in energy consumption or the proportion of energy input of each industrial sector under the price shock from the model output results as the basic data for the subsequent integrated learning prediction module.

[0028] Further, step S4 specifically includes:

[0029] Integrate the output of the CGE model with the historically observed actual energy - use data, construct time - series features, macro - economic features, and policy - variable features as needed, and perform data cleaning, missing - value imputation, standardization, or normalization on all features and the target variable;

[0030] Divide the processed data into a training set, a validation set, and a test set, and adopt a cross - validation method to reduce the risk of overfitting and improve the robustness of the model;

[0031] For the random forest model, obtain the prediction value through the voting results of multiple decision trees, and use the number of trees, maximum depth, and minimum sample split number as adjustable hyperparameters; for the XGBoost model, generate weak learners sequentially based on the gradient - boosting framework and perform weighted combination, and use the learning rate, max_depth, and subsample as adjustable hyperparameters; adopt grid search, Bayesian optimization, or evolutionary algorithms to optimize the hyperparameters of the above models;

[0032] Perform weighted averaging or superposition correction on the prediction results of the random forest model and the XGBoost model, and perform further correction using a residual correction or a dynamic correction strategy based on a rolling window;

[0033] Evaluate the prediction effect of the integrated model through preset evaluation indicators, and use the passed integrated model to output the refined prediction ratios of various energies in the industrial energy consumption structure, the electricity demand, and the possible peak-valley characteristics.

[0034] Further, step S5 specifically includes:

[0035] Clarify the specific parameter values of each policy, and set multiple groups of policy combinations according to user needs;

[0036] Integrate the set policy combinations into the CGE model and the integrated model, and adjust the cost function and constraint conditions to evaluate the changes in the future energy consumption structure or demand under different policy levels;

[0037] Run the CGE model for each group of policy combinations to obtain the results of changes in inter-industry energy demand and adjustments of macroeconomic indicators; input the data output by the CGE model into the integrated model for refined prediction to obtain a finer-grained allocation result of each energy in industrial sector consumption;

[0038] Perform iterative adjustment or sensitivity analysis on each group of policy combinations to find the optimal or sub-optimal policy combination.

[0039] Further, step S6 specifically includes:

[0040] Summarize the dynamic response analysis of the VAR model to electricity prices, the equilibrium results of the CGE model on industrial structure and energy demand, and the refined prediction results of the integrated model on the future energy consumption structure;

[0041] Present, in the form of charts or indicator lists, the changes in the energy usage ratio of the industrial sector, the peak-valley characteristics of electricity demand, and the economic benefits and emission reduction effects under different price shocks and policy combinations;

[0042] Give differentiated policy combination suggestions for different sectors or regions.

[0043] The second aspect of the present invention discloses a system for the impact of energy price fluctuations on the industrial energy consumption structure, which is constructed based on the modeling method described in the first aspect above, and includes:

[0044] A data acquisition module, which is used to collect historical energy price data, electricity price data, and relevant macroeconomic indicator data, and perform data cleaning, standardization, and seasonal adjustment to construct a data set;

[0045] A VAR model module, which is used to analyze the dynamic conduction relationship between energy prices and electricity prices based on the constructed data set, and obtain the time-lag effect and conduction path of price shocks;

[0046] The CGE model module is used to take the transmission path of the price shock output by the VAR model as an exogenous shock condition, and based on the social accounting matrix and inter-sector input-output data, conduct a macro-equilibrium solution for the resource reallocation among industries, energy demand changes, and output impacts, and obtain the preliminary simulation results of the energy consumption structure changes of each industrial sector under the price shock;

[0047] The integrated model module is used to take the preliminary simulation results of the CGE and historical data as feature inputs, and adopt the integrated learning method of the random forest model and the XGBoost model to conduct refined prediction and error correction on the proportion of each energy in the industrial energy consumption structure and the electricity demand;

[0048] The simulation module is used to introduce policy parameters, simulate multiple policy combinations based on the integrated model, and quantitatively evaluate the impact of different policy superpositions on the industrial energy consumption structure and the balance of electricity supply and demand;

[0049] The structure output module is used to synthesize the results of each model and output the prediction results of the industrial energy consumption structure under the price shock and policy combination and the strategic recommendation report.

[0050] The beneficial effects of the present invention are as follows:

[0051] By constructing a technical framework integrating multiple models and combining the quantitative modeling techniques of the VAR model, CGE model, and integrated learning prediction model, the present invention not only retains the economic interpretability and macro-analysis ability of the VAR model and CGE model, and can deeply analyze the macroeconomic transmission mechanism of price shocks and the changes in resource allocation among industries, but also can use machine learning methods to improve the prediction accuracy and flexibility to meet the actual needs of power supply guarantee policy formulation and energy consumption structure optimization in a complex market environment. Description of the Drawings

[0052] Figure 1 It is a schematic flowchart of an embodiment of the modeling method for the impact of energy price fluctuations on the industrial energy consumption structure of the present invention. Detailed Embodiments

[0053] The embodiments of the present invention will be described in more detail below with reference to the drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not used to limit the protection scope of the present invention.

[0054] See Figure 1 , the embodiments of the present invention provide a modeling method for the impact of energy price fluctuations on the industrial energy consumption structure, including the following steps.

[0055] S1. Collect historical energy price data, electricity price data, and relevant macroeconomic indicator data, perform data cleaning, standardization, and seasonal adjustment, and construct a dataset.

[0056] As a preferred implementation, in this embodiment, the historical energy price data includes coal price data, natural gas price data, and oil price data, and the relevant macroeconomic indicator data includes industrial output data, GDP data, and energy consumption data.

[0057] S2. Based on the constructed dataset, use the VAR model to analyze the dynamic conduction relationship between energy prices and electricity prices, and obtain the time-lag effect and conduction path of price shocks.

[0058] As a preferred implementation, in this embodiment, this step specifically includes:

[0059] Conduct a stationarity test on the energy price data series (coal, natural gas, oil, and electricity prices) and relevant macroeconomic indicator data (industrial output, GDP, and energy consumption data) in the dataset. Common methods include the ADF (Augmented Dickey–Fuller) test, PP (Phillips–Perron) test, etc. If the data is non-stationary, methods such as differencing or logarithmic transformation can be used for stationary processing.

[0060] Select the lag order of the VAR model through information criteria (AIC, BIC, HQ, etc.) to ensure that the dynamic relationship between price series can be captured while avoiding overfitting.

[0061] Based on the determined order, use the least squares method (OLS) or other appropriate estimation methods to estimate the parameters of the VAR model, and conduct residual autocorrelation tests (Ljung–Box test), normality tests (Jarque–Bera test), and stability tests (Roots test) on the estimated model to ensure the rationality and stability of the model.

[0062] Through impulse response function (IRF) analysis, obtain the dynamic response of electricity prices in multiple periods when a certain energy price is shocked, quantitatively evaluate the lagging impact on electricity prices, and obtain the conduction path and duration of price shocks.

[0063] If necessary, the model can also be decomposed by variance to further quantify the relative proportion of the contribution of each energy price shock to the electricity price fluctuation, helping to identify the most critical price factors.

[0064] Plot the conduction path and lag effect of the obtained electricity price on the energy price shock over time as the electricity price response curve, which is used as the output of the VAR model together.

[0065] S3. Take the conduction path of the price shock output by the VAR model as the exogenous shock condition and input it into the CGE model. Based on the social accounting matrix (SAM) and inter-sectoral input-output data, conduct a macro equilibrium solution for the resource reallocation among industries, changes in energy demand, and output impact, and obtain the preliminary simulation results of the energy consumption structure changes of each industrial sector under the price shock.

[0066] As a preferred implementation, in this embodiment, this step specifically includes:

[0067] 1) Model structure design

[0068] In this embodiment, divide the economic system into several production sectors (such as coal, oil, natural gas, electricity, steel, chemical industry, manufacturing, etc.), and include the household sector, government sector, and foreign trade sector. For the production activities of each sector, usually adopt production functions in the form of Cobb-Douglas, CES (Constant Elasticity of Substitution), or Leontief, etc., to characterize the substitution or complementary relationship of different factors (capital, labor, energy factors, etc.).

[0069] Consumption and investment. In the utility function or consumption function of the household sector, introduce consumption preferences and budget constraints; in the investment equation, handle capital accumulation and depreciation.

[0070] International trade. Adopt the Armington hypothesis or other trade hypotheses to distinguish imports, exports, and domestic demand, and set the substitution elasticity between imports and domestic products in the model.

[0071] Market equilibrium condition. Require the simultaneous achievement of supply-demand balance in the commodity market and factor market; the regulatory role of prices in the market is determined by the internal mechanism of the model or policy settings.

[0072] 2) Data preparation and calibration

[0073] Establish a social accounting matrix and use the SAM matrix containing information on the output, input, factor distribution, and income distribution of each sector as the benchmark database. Determine the elastic parameters: calibrate the price elasticities, substitution elasticities, etc. required for the production function, trade function, consumption function, etc. through literature or empirical methods. At the same time, set various external parameters in the benchmark year or benchmark scenario, such as population, technological progress rate, international energy price level, etc., to ensure that the model can accurately reproduce the benchmark data.

[0074] 3) Introduce exogenous shock (electricity price path)

[0075] According to the electricity price fluctuation path output by the VAR module, use it as an exogenous variable (or parameter change curve) and input it into the power sector or the entire energy market to obtain the impacts on production costs, factor inputs, and product prices. Electricity price paths with different time series can be set (such as short-term shocks, medium-term adjustments, and long-term stabilization) to simulate the industrial responses under different scenarios.

[0076] 4) Solving and iterative algorithms

[0077] Use numerical iterative methods (such as the Newton-Raphson algorithm) to solve the CGE model; during the iterative process, the model continuously adjusts sectoral outputs, factor inputs, and prices until the supply-demand balance conditions in the commodity market and factor market are simultaneously satisfied, and the system converges to a general equilibrium solution.

[0078] 5) Result output and analysis

[0079] The model outputs include the output changes of each sector, factor usage, changes in the energy demand structure, and macro indicators (such as GDP, employment, trade balance, etc.). Extract the changes in energy consumption or the proportion of energy input of each industrial sector under price shocks from the model output results as the basic data for the subsequent integrated learning prediction module.

[0080] S4. Using the preliminary simulation results of the CGE and historical data as feature inputs, adopt the integrated learning method of the random forest model and the XGBoost model to conduct refined prediction and error correction on the proportion of each energy in the industrial energy consumption structure and electricity demand.

[0081] As a preferred implementation, in this embodiment, this step specifically includes:

[0082] Integrate the energy consumption data, output change data, and key macro indicators output by the CGE model with the historically observed energy usage data; construct time series features, macroeconomic features (such as GDP, investment, consumption level, etc.), and policy variable features (such as electricity price subsidy level, whether the demand response mechanism is enabled) as needed; perform data cleaning, missing value imputation, standardization, or normalization processing on all features and target variables.

[0083] Divide the processed data into a training set, a validation set, and a test set, and adopt a cross-validation method to reduce the risk of overfitting and improve the robustness of the model.

[0084] For the random forest model, the predicted value is obtained through the voting results of multiple decision trees. The number of trees, maximum depth, and minimum sample split number are used as adjustable hyperparameters. For the XGBoost model, weak learners are sequentially generated based on the gradient boosting framework and weighted combined. The learning rate, max_depth, and subsample are used as adjustable hyperparameters. Grid Search, Bayesian Optimization, or evolutionary algorithms are used to optimize the hyperparameters of the above models.

[0085] The predicted results of the random forest model and the XGBoost model are weighted averaged or superimposed and corrected to obtain a more robust predicted value. According to the historical prediction error pattern, residual correction or a dynamic correction strategy based on a rolling window can be used for further correction to improve the prediction accuracy.

[0086] The prediction effect of the integrated model is evaluated through preset evaluation metrics such as MAE (Mean Absolute Error), RMSE (Root Mean Squared Error), and MAPE (Mean Absolute Percentage Error). The integrated model that passes the evaluation is used to output the refined prediction ratios, electricity demand, and possible peak-valley characteristics of each energy source in the industrial energy consumption structure.

[0087] S5. Introduce policy parameters, simulate multiple policy combinations based on the integrated model, and quantitatively evaluate the impact of different policy superpositions on the industrial energy consumption structure and the balance of electricity supply and demand.

[0088] Optionally, the policy parameters include the activation of the demand response mechanism, the level of electricity price subsidy, the reserve power release plan, etc.

[0089] As a preferred implementation, in this embodiment, this step specifically includes:

[0090] Clarify the specific parameter values of each policy, such as: electricity price subsidy: subsidy rate or absolute subsidy amount; demand response mechanism, incentive method of demand response, responsive user scale, response depth; reserve power release: release timing, release scale, duration, etc. According to the needs of the decision-making department or the research party, set multiple groups of policy combinations (such as electricity price subsidy, demand response mechanism, reserve power release plan, etc.);

[0091] Integrate the set policy combinations into the CGE model and the integrated model. When introducing policy parameters into the CGE model, the cost functions and constraint conditions of the production sectors or end - energy users need to be adjusted accordingly. For example, if there is an electricity price subsidy, the actual expenditure cost of the power sector or the using sector decreases, thus affecting their factor input and output decisions; if there is demand response, the electricity demand curve is adjusted during peak hours, changing the peak - valley electricity consumption. In the integrated model, the policy parameters are also regarded as new features input to evaluate the changes in the future energy use structure or demand under different policy levels.

[0092] Run the CGE model for each group of policy combinations to obtain the changes in inter - industry energy demand and the adjustment results of macro - economic indicators; input the data output by the CGE model into the integrated model for refined prediction to obtain a more granular distribution result of each energy in industrial sector consumption;

[0093] On this basis, conduct iterative adjustment or sensitivity analysis on each group of policy combinations to find the optimal or sub - optimal policy combination.

[0094] Furthermore, it is also possible to conduct quantitative evaluation on the indicators output by different policy combinations, such as the change range of industrial energy use structure, emission reduction effect, power sector supply - demand balance, the impact on GDP or employment, etc. Compare the results of each policy combination horizontally to analyze the synergy effect or potential conflict under policy superposition, providing multi - dimensional references for decision - makers.

[0095] Finally, based on the evaluation results, summarize the advantages and disadvantages of each policy combination, and propose a policy combination or optimization plan suitable for the actual situation.

[0096] S6. Synthesize the results of each model and output the prediction results of the industrial energy use structure under price shocks and policy combinations and a report on policy recommendations.

[0097] As an optimal implementation plan, in this embodiment, this step specifically includes:

[0098] Summarize the dynamic response analysis of the VAR model to electricity prices, the equilibrium results of the CGE model for industrial structure and energy demand, and the refined prediction results of the integrated model for the future energy use structure;

[0099] In the form of charts or indicator lists, display the changes in the energy use ratio of the industrial sector, the peak - valley characteristics of electricity demand, economic benefits, and emission reduction effects under different price shocks and policy combinations;

[0100] Give differentiated policy combination suggestions for different sectors or regions, such as whether to increase electricity price subsidies, improve the participation rate of demand response, guide enterprises to replace energy, etc. It is also possible to give a more operable schedule or implementation roadmap, and propose corresponding response plans in combination with risk assessment.

[0101] The second embodiment of the present invention discloses a system for the impact of energy price fluctuations on industrial energy use structures, which is constructed based on the modeling method described in the first embodiment above and includes:

[0102] A data acquisition module, which is used to collect historical energy price data, electricity price data, and relevant macroeconomic indicator data, and perform data cleaning, standardization, and seasonal adjustment to construct a data set;

[0103] A VAR model module, which is used to analyze the dynamic conduction relationship between energy prices and electricity prices based on the constructed data set, and obtain the time-lag effect and conduction path of price shocks;

[0104] A CGE model module, which uses the conduction path of price shocks output by the VAR model as exogenous shock conditions, and based on the social accounting matrix and inter-sector input-output data, conducts macro equilibrium solutions for resource reallocation, energy demand changes, and output impacts among industries to obtain preliminary simulation results of changes in energy use structures of various industrial sectors under price shocks;

[0105] An integrated model module, which uses the preliminary simulation results of the CGE and historical data as feature inputs, and adopts an integrated learning method of a random forest model and an XGBoost model to perform refined prediction and error correction on the proportions of various energies in the industrial energy use structure and electricity demand;

[0106] A simulation module, which is used to introduce policy parameters, simulate multiple strategy combinations based on the integrated model, and quantitatively evaluate the impacts of different policy superpositions on industrial energy use structures and the balance of electricity supply and demand;

[0107] A structure output module, which is used to synthesize the results of each model and output a prediction result of the industrial energy use structure under price shocks and policy combinations and a strategy recommendation report.

[0108] It should be noted that the method of the embodiment of the present invention can be executed by a single device, such as a computer or a server, etc. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In this distributed scenario, one of these multiple devices can only execute one or more steps of the method of the embodiment of the present invention, and these multiple devices will interact with each other to complete the described method.

[0109] It should be noted that some embodiments of the present invention have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the above-described embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0110] Embodiments of the present invention are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of the present invention shall be included within the protection scope of the present invention.

Claims

1. A modeling method for the impact of energy price fluctuations on the industrial energy consumption structure, characterized in that, It includes the following steps: S1. Collect historical energy price data, electricity price data, and relevant macroeconomic indicator data, perform data cleaning, standardization, and seasonal adjustment, and construct a dataset; S2. Based on the constructed dataset, use the VAR model to analyze the dynamic conduction relationship between energy prices and electricity prices, and obtain the time-lag effect and conduction path of price shocks; S3. Take the conduction path of price shocks output by the VAR model as an exogenous shock condition and input it into the CGE model. Based on the social accounting matrix and inter-sectoral input-output data, conduct a macro equilibrium solution for the resource reallocation, energy demand changes, and output impacts among industries to obtain the preliminary simulation results of the energy consumption structure changes of each industrial sector under price shocks; S4. Take the preliminary simulation results of the CGE and historical data as feature inputs, and use the integrated learning methods of the random forest model and the XGBoost model to conduct refined prediction and error correction on the proportions of various energies in the industrial energy consumption structure and electricity demand; S5. Introduce policy parameters, simulate multiple policy combinations based on the integrated model, and quantitatively evaluate the impacts of different policy superpositions on the industrial energy consumption structure and the balance of electricity supply and demand; S6. Synthesize the results of each model, and output the prediction results of the industrial energy consumption structure under price shocks and policy combinations and a report on policy recommendations.

2. The modeling method for the impact of energy price fluctuations on industrial energy consumption structure according to claim 1, characterized in that, The historical energy price data includes coal price data, natural gas price data, and oil price data, and the relevant macroeconomic indicator data includes industrial output data, GDP data, and energy consumption data.

3. The modeling method for the impact of energy price fluctuations on the industrial energy consumption structure according to claim 2, wherein, Step S2 specifically includes: Conduct a stationarity test on the energy price data series and relevant macroeconomic indicator data in the dataset, and perform stationarity processing on the non-stationary data by means of differencing or logarithmic transformation; Select the lag order of the VAR model through information criteria; Based on the determined order, use the least squares method to estimate the parameters of the VAR model, and conduct residual autocorrelation test, normality test, and stability test on the estimated model; Through impulse response function analysis, obtain the dynamic response of electricity prices in multiple periods when a certain energy price is shocked, quantitatively evaluate the lagging impact on electricity prices, and obtain the conduction path and duration of price shocks; Plot the conduction path and lag effect of electricity prices on energy price shocks over time as an electricity price response curve, which is used as the output of the VAR model together.

4. The modeling method for the impact of energy price fluctuations on industrial energy consumption structure according to claim 3, characterized in that Step S3 specifically includes: Establish a social accounting matrix, use the SAM matrix containing information on the output, input, factor distribution, and income distribution of each department as the benchmark database, determine the elastic parameters, and set various external parameters in the benchmark year or benchmark scenario; Take the conduction path of price shocks output by the VAR model as an exogenous variable and input it into the electricity sector or the entire energy market to obtain the impacts on production costs, factor inputs, and product prices; Use the numerical iteration method to solve the CGE model; during the iteration process, the model continuously adjusts the sector output, factor inputs, and prices until the supply and demand balance conditions in the commodity market and factor market are simultaneously satisfied, and the system converges to a general equilibrium solution. Extract the changes in the energy consumption or the proportion of energy input of each industrial sector under price shocks from the model output results as the basic data for the subsequent integrated learning prediction module.

5. The modeling method for the impact of energy price fluctuations on the industrial energy consumption structure according to claim 4, characterized in that, Step S4 specifically includes: Integrate the output of the CGE model with the historically observed actual energy usage data, construct time series features, macroeconomic features, and policy variable features as needed, and perform data cleaning, missing value imputation, standardization, or normalization on all features and target variables; Divide the processed data into a training set, a validation set, and a test set, and adopt a cross-validation method to reduce the risk of overfitting and improve the robustness of the model; For the random forest model, obtain the predicted values through the voting results of multiple decision trees, and use the number of trees, the maximum depth, and the minimum sample split number as adjustable hyperparameters; for the XGBoost model, generate weak learners sequentially based on the gradient boosting framework and perform weighted combination, and use the learning rate, max_depth, and subsample as adjustable hyperparameters; use grid search, Bayesian optimization, or evolutionary algorithms to optimize the hyperparameters of the above models; Perform weighted averaging or superposition correction on the predicted results of the random forest model and the XGBoost model, and use residual correction or a dynamic correction strategy based on a rolling window for further correction; Evaluate the prediction effect of the integrated model through preset evaluation indicators, and use the integrated model that passes the evaluation to output the refined prediction ratio of each energy in the industrial energy consumption structure, the electricity demand, and the possible peak and valley characteristics.

6. The modeling method for the impact of energy price fluctuations on industrial energy consumption structure according to claim 5, wherein Step S5 specifically includes: Clarify the specific parameter values of each policy, and set multiple sets of policy combinations according to user needs; Integrate the set policy combinations into the CGE model and the integrated model, and adjust the cost function and constraint conditions to evaluate the changes in the future energy consumption structure or demand under different policy levels; Run the CGE model for each set of policy combinations to obtain the results of changes in inter-industry energy demand and macroeconomic indicator adjustments; input the data output by the CGE model into the integrated model for refined prediction to obtain a finer-grained allocation result of each energy in industrial sector consumption; Perform iterative adjustment or sensitivity analysis on each set of policy combinations to find the optimal or sub-optimal policy combination.

7. The modeling method for the impact of energy price fluctuations on the industrial energy consumption structure according to claim 6, wherein Step S6 specifically includes: Summarize the dynamic response analysis of the VAR model to electricity prices, the equilibrium results of the CGE model for industrial structure and energy demand, and the refined prediction results of the integrated model for the future energy consumption structure; Present, in the form of charts or indicator lists, the changes in the energy usage ratio, the peak and valley characteristics of electricity demand, economic benefits, and emission reduction effects of the industrial sector under different price shocks and policy combinations; Give differentiated policy combination suggestions for different sectors or regions.

8. A system for the impact of energy price fluctuations on the industrial energy consumption structure, constructed based on the modeling method described in any one of claims 1-7, characterized in that, Including: A data collection module, which is used to collect historical energy price data, electricity price data, and relevant macroeconomic indicator data, perform data cleaning, standardization, and seasonal adjustment, and construct a data set; A VAR model module, which is used to analyze the dynamic conduction relationship between energy prices and electricity prices based on the constructed data set, and obtain the time lag effect and conduction path of price shocks; The CGE model module is used to take the transmission path of the price shock output by the VAR model as an exogenous shock condition, and based on the social accounting matrix and inter-sectoral input-output data, conduct a macro equilibrium solution for the resource reallocation, energy demand change and output impact among industries, and obtain the preliminary simulation results of the energy consumption structure change of each industrial sector under the price shock; The integrated model module is used to take the preliminary simulation results of the CGE and historical data as feature inputs, and adopt the integrated learning method of the random forest model and the XGBoost model to conduct refined prediction and error correction on the proportion of each energy in the industrial energy consumption structure and the electricity demand; The simulation module is used to introduce policy parameters, simulate various policy combinations based on the integrated model, and quantitatively evaluate the impact of different policy superpositions on the industrial energy consumption structure and the balance of electricity supply and demand; The structure output module is used to synthesize the results of each model and output the prediction results of the industrial energy consumption structure under the price shock and policy combination and the strategic recommendation report.

Images