Regional energy consumption monitoring method and system based on ARDL model

Abstract

The invention provides a regional energy consumption monitoring method and system based on an ARDL model, and the monitoring method is applied to a monitoring system comprising an energy consumption data processing module, a model construction module, an energy consumption prediction module and a visual display module, and specifically comprises the steps: constructing an energy consumption variable according to the historical multi-source energy consumption data of a to-be-monitored region, data stability verification is carried out; an ARDL model is trained according to the energy consumption variables passing the data stability verification, and a regional energy consumption prediction model is obtained; and inputting the multi-source energy consumption data of the to-be-monitored area in the current monitoring period into the area energy consumption prediction model, obtaining energy consumption prediction data of the next monitoring period, and performing visual display. According to the method, the pseudo-regression phenomenon of the ARDL model in the training process is avoided by means of energy consumption variable construction and stability verification, the accuracy of energy consumption prediction is guaranteed, and the regional energy consumption monitoring effect is improved.

Description

Technical Field

[0001] This invention relates to the field of energy consumption monitoring technology, and in particular to a regional energy consumption monitoring method and system based on the ARDL model. Background Technology

[0002] Most current regional energy consumption monitoring platforms utilize comprehensive energy consumption evaluation methods to assess regional energy consumption and then conduct subsequent energy consumption monitoring to achieve refined regional energy management. However, when calculating comprehensive energy consumption, these evaluation methods only include direct energy consumption such as primary energy, secondary energy, and energy media, failing to intuitively reflect energy waste and making it difficult to accurately predict energy consumption.

[0003] The ARDL (Autoregressive Distributed Lag) model, as an effective time series analysis method, can consider both long-term and short-term impacts. In the context of regional energy consumption monitoring, the lagged values ​​of energy consumption variables themselves, as well as other relevant influencing factors such as economic indicators, the proportion of clean energy, and temperature, can be incorporated into the ARDL model as explanatory variables to analyze the long-term and short-term impacts of these factors on regional energy consumption. This effectively solves the problems existing in current comprehensive energy consumption evaluation methods, achieves more accurate prediction of energy consumption trends, and ensures the accuracy of energy consumption prediction.

[0004] However, the effective application of the ARDL model relies on the stationarity of the data. In the field of energy consumption forecasting, many time series data, such as monthly and annual regional energy consumption data, often exhibit complex fluctuations. In such cases, directly applying the ARDL model for energy consumption forecasting is highly likely to lead to biased model estimation results and spurious regression, making it impossible for the model to accurately capture the true changing patterns behind the energy consumption data. This, in turn, affects the accuracy of the forecast and makes it difficult to achieve effective regional energy consumption monitoring. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies that use ARDL models for energy consumption prediction, which are prone to prediction deviations due to fluctuations in regional energy consumption data, resulting in low accuracy of energy consumption prediction and poor regional energy consumption monitoring. This invention provides a regional energy consumption monitoring method and system based on ARDL models. By constructing energy consumption variables and verifying stationarity, the invention avoids spurious regression phenomena in the ARDL model during training, ensuring the accuracy of energy consumption prediction and improving the regional energy consumption monitoring effect.

[0006] The objective of this invention is achieved through the following technical solution: Regional energy consumption monitoring methods based on the ARDL model include: Energy consumption variables are constructed based on historical multi-source energy consumption data of the area to be monitored, and the data stability of the energy consumption variables is verified. An ARDL model is trained using energy consumption variables that have passed data stationarity verification to obtain a regional energy consumption prediction model; The system acquires multi-source energy consumption data for the area to be monitored during the current monitoring period, inputs the acquired multi-source energy consumption data into the regional energy consumption prediction model, obtains the energy consumption prediction data for the next monitoring period for the area to be monitored, and visualizes the energy consumption prediction data.

[0007] By constructing multi-source energy consumption variables and conducting stationarity checks, non-stationary data interference is eliminated at the source, avoiding the risk of spurious regression in ARDL model training and improving the accuracy and reliability of subsequent regional energy consumption predictions. The trained regional energy consumption prediction model can accurately capture the regional energy consumption change patterns. Combined with the monthly breakdown of multi-source energy consumption data for the current monitoring period, the data granularity is further refined, making the model input more closely match the actual energy consumption fluctuation characteristics. This results in accurate output of energy consumption prediction data for the next monitoring period. With the visualization of energy consumption prediction data, refined regional energy consumption monitoring can be effectively achieved.

[0008] Furthermore, the step of constructing energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and performing data stationarity verification on the energy consumption variables, includes: Energy consumption variables are constructed by taking the total energy consumption in historical multi-source energy consumption data as the dependent variable and the remaining data in historical multi-source energy consumption data as the independent variable. The stationarity of energy consumption variables was verified using unit root test and cointegration test methods.

[0009] Furthermore, the step of verifying the stationarity of energy consumption variables using unit root tests and cointegration tests includes: The stationarity of energy consumption variables is verified by unit root test. Based on the stationarity verification results, energy consumption variables are divided into stationary energy consumption variables and non-stationary energy consumption variables. Based on the stationarity verification results, the initial integration order of the non-stationary energy consumption variables is obtained, and non-stationary energy consumption variables whose initial integration order does not exceed a preset threshold are selected. The selected non-stationary energy consumption variables are subjected to difference processing to obtain the corresponding stationary difference sequences; The selected non-stationary energy consumption variables are subjected to cointegration tests. The non-stationary energy consumption variables that pass the cointegration test, their stationary difference sequences, and stationary variables are used as energy consumption variables that pass the data stationarity verification.

[0010] Furthermore, the step of training an ARDL model based on energy consumption variables that have passed data stationarity verification to obtain a regional energy consumption prediction model includes: Using total energy consumption, which has passed the data stationarity test, as the dependent variable and the remaining variables as independent variables, the variable combination of the ARDL model is set. The optimal lag order of the ARDL model is selected based on the energy consumption variable, which is verified by data stationarity, and combined with the information criterion optimization method. Based on the selected lag order, the parameters of the ARDL model are estimated using the least squares method to obtain the trained ARDL model. Perform parameter significance and residual sequence tests on the trained ARDL model; When the test is passed, the trained ARDL model is used as the regional energy consumption prediction model. If the test fails, adjust the variable combination or lag order, and re-estimate and test the parameters until the test is passed.

[0011] Furthermore, before inputting the acquired multi-source energy consumption data into the regional energy consumption prediction model, the following steps are also performed: Construct a quadratic objective function for the monthly energy consumption prediction error and set total energy consumption constraints; The optimal combination of monthly energy consumption estimates is obtained by solving a quadratic objective function based on multi-source energy consumption data using the Lagrange multiplier method. The multivariate energy consumption data is split into monthly data based on the optimal combination of monthly energy consumption estimates.

[0012] Furthermore, the visualization of energy consumption prediction data includes: Set monitoring indicators according to the monitoring needs of the area to be monitored, and match the corresponding chart type according to the indicator type of the monitoring indicator; Based on the chart type, the energy consumption prediction data is split and features are extracted to construct corresponding visualization images.

[0013] Furthermore, the historical multi-source energy consumption data includes at least the region's historical electricity consumption, regional GDP, the proportion of clean energy in the region, and historical total energy consumption.

[0014] A regional energy consumption monitoring system based on the ARDL model is used to execute the aforementioned regional energy consumption monitoring method, including: The energy consumption data processing module is used to construct energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and to perform data stability verification on the energy consumption variables. The model building module, connected to the energy consumption data processing module, is used to train the ARDL model based on energy consumption variables that have passed the data stationarity verification, and obtain a regional energy consumption prediction model. The energy consumption prediction module, connected to the model building module, is used to split the acquired multi-source energy consumption data into monthly data and combine it with the regional energy consumption prediction model to obtain the energy consumption prediction data of the area to be monitored. The visualization module, connected to the energy consumption prediction module, is used to visualize the acquired energy consumption prediction data.

[0015] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned regional energy consumption monitoring method based on the ARDL model.

[0016] A non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the aforementioned regional energy consumption monitoring method based on the ARDL model.

[0017] The beneficial effects of this invention are: (1) By constructing multi-source energy consumption variables and conducting stationarity verification, non-stationary data interference is eliminated from the source, avoiding the risk of spurious regression in ARDL model training, and improving the accuracy and reliability of subsequent regional energy consumption prediction. The trained regional energy consumption prediction model can accurately capture the regional energy consumption change pattern. Combined with the monthly splitting of multi-source energy consumption data in the current monitoring period, the data granularity is further refined, making the model input more consistent with the actual energy consumption fluctuation characteristics, thereby accurately outputting the energy consumption prediction data for the next monitoring period. With the visualization of the energy consumption prediction data, the refined supervision of regional energy consumption can be effectively realized.

[0018] (2) A variable system is constructed with total energy consumption as the dependent variable and the remaining data as independent variables. Then, the unit root test is used to distinguish between stationary and non-stationary variables. After screening non-stationary variables by combining the order of integration, the difference processing is carried out. Finally, the long-term equilibrium relationship is explored by cointegration test to determine the energy consumption variables that pass the verification, so as to ensure their data stationarity. On the basis of ensuring the amount of training data, the risk of spurious regression in the subsequent ARDL model training is effectively avoided.

[0019] (3) Construct a quadratic objective function for monthly energy consumption prediction error and set total constraints. Then, use the Lagrange multiplier method to solve for the optimal monthly energy consumption estimate, realize the splitting of energy consumption data in the monthly dimension, refine the data granularity, ensure that the data input to the prediction model can better match the actual fluctuations, and improve the prediction accuracy and timeliness. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of a process of the present invention; Figure 2 This is a schematic diagram of a structure according to an embodiment of the present invention.

[0021] The module includes: 1. Energy consumption data processing module; 2. Model building module; 3. Energy consumption prediction module; and 4. Visualization module. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Example: Regional energy consumption monitoring methods based on ARDL models, such as Figure 1 As shown, it includes: Energy consumption variables are constructed based on historical multi-source energy consumption data of the area to be monitored, and the data stability of the energy consumption variables is verified. A regional energy consumption prediction model is obtained by training an ARDL model based on energy consumption variables that have passed data stationarity verification. The system acquires multi-source energy consumption data for the area to be monitored during the current monitoring period, inputs the acquired multi-source energy consumption data into the regional energy consumption prediction model, obtains the energy consumption prediction data for the next monitoring period for the area to be monitored, and visualizes the energy consumption prediction data.

[0024] Regional energy consumption is influenced by multiple factors such as industrial production, residential life, seasonal climate, and energy prices, exhibiting trend-like and seasonal fluctuations. Relying solely on a single energy consumption data point cannot accurately reflect the patterns of energy consumption changes within a region. Therefore, it is necessary to incorporate data including historical regional electricity consumption. Regional GDP Regional clean energy ratio and historical total energy consumption Multi-source energy consumption data serves as the data foundation for regional energy consumption prediction, ensuring the accuracy of subsequent predictions.

[0025] Such multi-source energy consumption data covers numerous collection sources and is characterized by multiple sources, heterogeneous structure, and inconsistent granularity. Therefore, when acquiring historical multi-source energy consumption data, the corresponding data is extracted, transformed, and loaded through an ETL process.

[0026] In the data extraction process, a data sampling strategy combining periodic collection and event-triggered collection is adopted. The sampling period is mainly monthly, but some data is sampled quarterly or annually. Meanwhile, breakpoint resume mechanisms, data reporting format protocols, collection frequencies, and incremental identifiers are set up for data collection to ensure the integrity and continuity of historical multi-source energy consumption data during the extraction process.

[0027] Furthermore, to ensure security during energy consumption monitoring, historical energy consumption data is transmitted via an HTTPS encrypted channel, and corresponding IP whitelists and access control policies are implemented to restrict interface access permissions. The extracted historical multi-source energy consumption data undergoes data cleaning and format conversion to integrate the data and store it in a unified database, forming a multi-dimensional time series dataset for subsequent modeling, which can be defined as... .

[0028] Historical multi-source energy consumption data often exhibits trends and seasonal fluctuations. If directly used for model training, the ARDL model is prone to spurious regression due to data non-stationarity. This means the model may appear statistically significant but fails to accurately reflect the causal relationships between variables, impacting the accuracy of subsequent predictions. Therefore, it is necessary to first construct a comprehensive energy consumption variable system and eliminate data interference through stationarity checks.

[0029] Specifically, the step of constructing energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and performing data stationarity verification on the energy consumption variables, includes: Energy consumption variables are constructed by taking the total energy consumption in historical multi-source energy consumption data as the dependent variable and the remaining data in historical multi-source energy consumption data as the independent variable. The stationarity of energy consumption variables was verified using unit root test and cointegration test methods.

[0030] Total energy consumption directly reflects the final scale of regional energy consumption and is the core objective of energy consumption monitoring and forecasting. Therefore, total energy consumption is used as the dependent variable. The remaining multi-source energy consumption data are used as influencing factors affecting energy consumption changes and are set as corresponding independent variables. By setting the dependent and independent variables, a variable system can be constructed, thereby accurately covering the core influencing factors of energy consumption and avoiding model bias caused by missing dependent variables.

[0031] Data stationarity is a prerequisite for the reliable operation of the ARDL model. Therefore, the constructed energy consumption variables are further verified and processed through data stationarity verification.

[0032] Specifically, the method of verifying the stationarity of energy consumption variables using unit root tests and cointegration tests includes: The stationarity of energy consumption variables is verified by unit root test. Based on the stationarity verification results, energy consumption variables are divided into stationary energy consumption variables and non-stationary energy consumption variables. Based on the stationarity verification results, the initial integration order of the non-stationary energy consumption variables is obtained, and non-stationary energy consumption variables whose initial integration order does not exceed a preset threshold are selected. The selected non-stationary energy consumption variables are subjected to difference processing to obtain the corresponding stationary difference sequences; The selected non-stationary energy consumption variables are subjected to cointegration tests. The non-stationary energy consumption variables that pass the cointegration test, their stationary difference sequences, and stationary variables are used as energy consumption variables that pass the data stationarity verification.

[0033] The basic form of the unit root test method is as follows: ; in, Let be any time series variable to be tested, i.e., energy consumption variable. It is a first-order difference. For constant terms, This is the time trend coefficient. The coefficients of the lagged terms of the original variable. The lag order is... For the lagged difference term in period i coefficient, This is the residual term.

[0034] Based on the above test format, a unit root test is performed individually for each energy consumption variable to determine whether the time series of each variable satisfies the stationarity characteristic that the mean and variance do not change over time. Furthermore, the specific parameters are determined accordingly. The significance of these parameters is used to identify stationary energy consumption variables. If the statistic is significant, the null hypothesis that the energy consumption variable has a unit root can be rejected, and the variable can be determined to be a stationary energy consumption variable. Otherwise, the variable can be determined to be a non-stationary energy consumption variable.

[0035] For non-stationary energy consumption variables, their initial integration order is extracted using unit root tests, i.e., the number of differencing steps required to transform the variable into a stationary sequence. This initial order is then used for selection. Since the ARDL model requires that the integration order of a variable not exceed 1, non-stationary energy consumption variables with an integration order greater than or equal to 2 are first filtered out before differencing. These variables, even after multiple differencing steps, are unlikely to meet the model's basic requirement for data stationarity; forcibly including them would only lead to the failure of parameter estimation during subsequent model training. After filtering, only non-stationary energy consumption variables with an integration order of 1 are retained, ensuring their potential to be corrected into a stationary sequence through differencing.

[0036] Then, the selected non-stationary energy consumption variables are differencing to obtain the corresponding stationary differencing sequences. Specifically, each selected non-stationary energy consumption variable is first-differenced, i.e., the difference between the current period's value and the previous period's value is calculated. After obtaining the differencing sequences, their stationarity must be verified again using a unit root test. Only when the differencing sequence is confirmed to be stationary is the correction considered complete. Furthermore, after correcting the non-stationary energy consumption variables, both the original non-stationary energy consumption variables and the differencing stationary sequences must be retained simultaneously. The former is used for subsequent analysis of long-term relationships between variables, while the latter is used to capture short-term fluctuation characteristics.

[0037] Finally, cointegration tests are performed on the selected non-stationary energy consumption variables. The core of the cointegration test is to verify whether there is a long-term stable equilibrium relationship among these non-stationary variables. Specifically, the cointegration test can be performed by constructing regression equations between the variables and testing whether their residual sequences are stationary. If the residuals are stationary, it indicates that there is a cointegration relationship among the variables.

[0038] If the cointegration test is passed, it indicates that although these non-stationary energy consumption variables fluctuate individually over time, they exhibit a stable long-term correlation and can be included in the ARDL model for analyzing long-term effects. If the cointegration test is not passed, variable selection or combination needs to be re-evaluated until a cointegration relationship is confirmed.

[0039] Ultimately, the non-stationary energy consumption variables that pass the cointegration test and their stationary difference sequences, together with the original stationary energy consumption variables, are used as energy consumption variables that pass the data stationarity check. This provides the ARDL model with input data that meets both mathematical requirements and has economic significance, preserving key correlations between variables while avoiding spurious regression problems during ARDL model training, thus ensuring prediction accuracy from the source.

[0040] Once the relevant training data is prepared, the corresponding ARDL model can be built and the corresponding training work can be carried out to obtain a regional energy consumption prediction model.

[0041] Specifically, the step of training an ARDL model based on energy consumption variables that have passed data stationarity verification to obtain a regional energy consumption prediction model includes: Using total energy consumption, which has passed the data stationarity test, as the dependent variable and the remaining variables as independent variables, the variable combination of the ARDL model is set. The optimal lag order of the ARDL model is selected based on the energy consumption variable, which is verified by data stationarity, and combined with the information criterion optimization method. Based on the selected lag order, the parameters of the ARDL model are estimated using the least squares method to obtain the trained ARDL model. Perform parameter significance and residual sequence tests on the trained ARDL model; When the test is passed, the trained ARDL model is used as the regional energy consumption prediction model. If the test fails, adjust the variable combination or lag order, and re-estimate and test the parameters until the test is passed.

[0042] Considering that the training objective of the ARDL model is to predict regional energy consumption, the total energy consumption that has passed the stationarity check is designated as the dependent variable in the ARDL model. This is the core indicator reflecting the regional energy consumption level and the final object the model needs to predict. Then, variables with an economic correlation to total energy consumption are selected from the remaining variables that have passed the stationarity check as independent variables in the ARDL model, such as regional GDP, regional electricity consumption, and the proportion of clean energy.

[0043] Based on this, the expression of the ARDL model established in this embodiment is: ; in, For constant terms, This represents the total energy consumption. Indicates battery level. Indicates supplementary variables, These are the lag orders of the dependent variable, independent variable, and supplementary variable, respectively. The coefficient of the lagged term of the dependent variable. The coefficient of the lagged term of the independent variable. To supplement the coefficients of the lagged terms of the variables, This is the random error term.

[0044] These lag orders directly affect the ARDL model's ability to capture short-term dynamic effects and long-term equilibrium relationships. To ensure the accuracy of the ARDL model, the optimal lag order is further selected using the information criterion optimization method.

[0045] The formula for calculating the optimal lag coefficient using the information criterion selection method is as follows: ; in, For residual variance, To estimate the number of parameters for the model, This represents the number of samples.

[0046] Using the above calculation formula, for each possible combination of lag orders, a corresponding ARDL model is constructed and the AIC value is calculated. After traversing all combinations, the combination with the smallest AIC value is selected as the optimal lag termination.

[0047] After combining the energy consumption variables and obtaining the optimal lag order, the parameters are further estimated using the least squares method to convert them into a specific model.

[0048] Specifically, based on the selected optimal lag order, the dependent and independent variables are first lagged to ensure that all lagged data are time-aligned with the current variable data, forming a complete modeling dataset. Then, all parameters of the ARDL model are estimated using the least squares method. The optimal estimates of all parameters are obtained by minimizing the sum of squared residuals between the model's predicted values ​​and the actual observed values, ultimately forming an ARDL model containing the specific parameter values.

[0049] The trained ARDL model is then subjected to parameter significance and residual sequence tests to ensure that the constructed model satisfies the basic assumptions. The residual sequence tests specifically include residual independence tests, residual normality tests, and residual variance tests. Both the parameter significance test and the residual sequence test can be implemented using existing algorithms, which will not be elaborated upon here.

[0050] If the model passes all tests, it indicates that the model parameters are significant, the residuals meet the assumptions, and the statistical reliability and fit are satisfactory. In this case, the ARDL model can be selected as the final regional energy consumption prediction model. If the model fails the tests, it needs to be adjusted and remodeled to ultimately create an ARDL model that accurately captures the patterns of regional energy consumption changes and meets business forecasting needs.

[0051] Since energy statistics are mostly based on an annual unit, when making energy consumption forecasts, the annual total needs to be broken down into monthly data to meet the needs of regional energy consumption trend analysis.

[0052] Therefore, before inputting the acquired multi-source energy consumption data into the regional energy consumption prediction model, the following steps are also performed: Construct a quadratic objective function for the monthly energy consumption prediction error and set total energy consumption constraints; The optimal combination of monthly energy consumption estimates is obtained by solving a quadratic objective function based on multi-source energy consumption data using the Lagrange multiplier method. The multi-source energy consumption data is split into monthly data based on the optimal combination of monthly energy consumption estimates.

[0053] The expression for the quadratic objective function of the monthly energy consumption prediction error is as follows: ; in, The objective function value, This is the estimated energy consumption value for month m. This is a function based on the predicted electricity consumption for the current month.

[0054] The expression for the total energy consumption constraint is as follows: ; in, This represents the total annual energy consumption for the corresponding year.

[0055] Then, the objective function is solved by using the Lagrange multiplier method to minimize the error and satisfy the total amount constraint, thereby obtaining the optimal monthly estimate and realizing the monthly splitting of multi-source energy consumption data.

[0056] In the actual forecasting process, the latest monthly electricity consumption, GDP forecast, and clean energy ratio obtained in the current monitoring period can be substituted into the regional energy consumption forecasting model to output the energy consumption forecast data for the next monitoring period.

[0057] To achieve intuitive monitoring of regional energy consumption, further visualization of energy consumption prediction data is needed, including: Set monitoring indicators according to the monitoring needs of the area to be monitored, and match the corresponding chart type according to the indicator type of the monitoring indicator; Based on the chart type, the energy consumption prediction data is split and features are extracted to construct corresponding visualization images.

[0058] To improve energy monitoring efficiency, monitoring indicators should first be set based on the core needs of the area to be monitored. If the focus of the area is on industrial energy consumption supervision, monitoring indicators may include total industrial energy consumption, the proportion of energy consumption in high-energy-consuming industries, and energy consumption per unit of industrial output. If the focus is on energy consumption analysis for people's livelihood, indicators such as residential energy consumption, energy consumption of public buildings, and seasonal fluctuations in energy consumption should be included.

[0059] Next, the appropriate chart type is matched according to the type of monitoring indicator. For time series indicators, such as monthly energy consumption trends and year-on-year changes in energy consumption, line charts or area charts are preferred for display. Then, based on the determined chart type, the energy consumption forecast data is subjected to targeted data splitting and feature extraction to determine the corresponding visualization content.

[0060] Another aspect of this embodiment also provides a regional energy consumption monitoring system based on the ARDL model, such as... Figure 2 As shown, it includes: Energy consumption data processing module 1 is used to construct energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and to perform data stability verification on the energy consumption variables. Model building module 2 is connected to the energy consumption data processing module and is used to train the ARDL model based on the energy consumption variables that have passed the data stationarity verification to obtain the regional energy consumption prediction model. Energy consumption prediction module 3, connected to the model building module, is used to split the acquired multi-source energy consumption data into monthly data and combine it with the regional energy consumption prediction model to obtain the energy consumption prediction data of the area to be monitored. The visualization module 4 is connected to the energy consumption prediction module and is used to visualize the acquired energy consumption prediction data.

[0061] The energy consumption data processing module, model building module, energy consumption prediction module, and visualization module are all data processing components with corresponding processing algorithms, such as computers and microprocessors. They can be deployed on private servers or cloud platforms, and can all interface with data centers via RESTful APIs to periodically call new data for model calculation and rolling prediction.

[0062] Furthermore, to ensure data security and traceability, the monitoring system is equipped with corresponding access control and encryption mechanisms. All output pages automatically generate digital watermarks and timestamps to prevent data leakage.

[0063] This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the various processes of the above-described embodiment of the regional energy consumption monitoring method based on the ARDL model and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0064] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0065] This application also provides a non-transitory computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the various processes of the above-described embodiment of the regional energy consumption monitoring method based on the ARDL model and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0066] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0067] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Other variations and modifications are possible without departing from the technical solutions described in the claims.

Claims

1. A regional energy consumption monitoring method based on the ARDL model, characterized in that, include: Energy consumption variables are constructed based on historical multi-source energy consumption data of the area to be monitored, and the data stability of the energy consumption variables is verified. An ARDL model is trained using energy consumption variables that have passed data stationarity verification to obtain a regional energy consumption prediction model; The system acquires multi-source energy consumption data for the area to be monitored during the current monitoring period, inputs the acquired multi-source energy consumption data into the regional energy consumption prediction model, obtains the energy consumption prediction data for the next monitoring period for the area to be monitored, and visualizes the energy consumption prediction data.

2. The regional energy consumption monitoring method based on the ARDL model according to claim 1, characterized in that, The process of constructing energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and verifying the data stationarity of the energy consumption variables, includes: Energy consumption variables are constructed by taking the total energy consumption in historical multi-source energy consumption data as the dependent variable and the remaining data in historical multi-source energy consumption data as the independent variable. The stationarity of energy consumption variables was verified using unit root test and cointegration test methods.

3. The regional energy consumption monitoring method based on the ARDL model according to claim 2, characterized in that, The method of verifying the stationarity of energy consumption variables using unit root tests and cointegration tests includes: The stationarity of energy consumption variables is verified by unit root test. Based on the stationarity verification results, energy consumption variables are divided into stationary energy consumption variables and non-stationary energy consumption variables. Based on the stationarity verification results, the initial integration order of the non-stationary energy consumption variables is obtained, and non-stationary energy consumption variables whose initial integration order does not exceed a preset threshold are selected. The selected non-stationary energy consumption variables are subjected to difference processing to obtain the corresponding stationary difference sequences; The selected non-stationary energy consumption variables are subjected to cointegration tests. The non-stationary energy consumption variables that pass the cointegration test, their stationary difference sequences, and stationary variables are used as energy consumption variables that pass the data stationarity verification.

4. The regional energy consumption monitoring method based on the ARDL model according to claim 2, characterized in that, The step of training an ARDL model based on energy consumption variables that have passed data stationarity verification to obtain a regional energy consumption prediction model includes: Using total energy consumption, which has passed the data stationarity test, as the dependent variable and the remaining variables as independent variables, the variable combination of the ARDL model is set. The optimal lag order of the ARDL model is selected based on the energy consumption variable, which is verified by data stationarity, and combined with the information criterion optimization method. Based on the selected lag order, the parameters of the ARDL model are estimated using the least squares method to obtain the trained ARDL model. Perform parameter significance and residual sequence tests on the trained ARDL model; When the test is passed, the trained ARDL model is used as the regional energy consumption prediction model. If the test fails, adjust the variable combination or lag order, and re-estimate and test the parameters until the test is passed.

5. The regional energy consumption monitoring method based on the ARDL model according to claim 1, characterized in that, Before inputting the acquired multi-source energy consumption data into the regional energy consumption prediction model, the following steps are also performed: Construct a quadratic objective function for the monthly energy consumption prediction error and set total energy consumption constraints; The optimal combination of monthly energy consumption estimates is obtained by solving a quadratic objective function based on multi-source energy consumption data using the Lagrange multiplier method. The multivariate energy consumption data is split into monthly data based on the optimal combination of monthly energy consumption estimates.

6. The regional energy consumption monitoring method based on the ARDL model according to claim 1, characterized in that, The visualization of energy consumption prediction data includes: Set monitoring indicators according to the monitoring needs of the area to be monitored, and match the corresponding chart type according to the indicator type of the monitoring indicator; Based on the chart type, the energy consumption prediction data is split and features are extracted to construct corresponding visualization images.

7. The regional energy consumption monitoring method based on the ARDL model according to any one of claims 1 to 6, characterized in that, The historical multi-source energy consumption data includes at least the region's historical electricity consumption, regional GDP, the proportion of clean energy in the region, and historical total energy consumption.

8. A regional energy consumption monitoring system based on the ARDL model, used to execute the regional energy consumption monitoring method according to any one of claims 1 to 7, characterized in that, include: The energy consumption data processing module is used to construct energy consumption variables based on historical multi-source energy consumption data of the area to be monitored, and to perform data stability verification on the energy consumption variables. The model building module, connected to the energy consumption data processing module, is used to train the ARDL model based on energy consumption variables that have passed the data stationarity verification, and obtain a regional energy consumption prediction model. The energy consumption prediction module, connected to the model building module, is used to split the acquired multi-source energy consumption data into monthly data and combine it with the regional energy consumption prediction model to obtain the energy consumption prediction data of the area to be monitored. The visualization module, connected to the energy consumption prediction module, is used to visualize the acquired energy consumption prediction data.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the regional energy consumption monitoring method based on the ARDL model as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the regional energy consumption monitoring method based on the ARDL model as described in any one of claims 1 to 7.

Images