A power user load forecasting method and software system considering carbon trading

Abstract

The application discloses a kind of considering carbon transaction electric power user load prediction method and software system, method includes: establishing the nonlinear panel regression model considering carbon transaction, to describe the nonlinear relationship between price and carbon price;The nonlinear panel regression model considering carbon transaction is carried out regression analysis, and model parameters are obtained, so as to obtain the carbon price of the day to be predicted according to the price of the day to be predicted;Using min-max normalization method, the price of the day to be predicted, historical load data and the carbon price of the day to be predicted are normalized;Based on the parameter optimization method of bamboo joint insect population evolution algorithm, the electric power user load prediction model of improved LSTM is established;The historical load data, the price and carbon price of the day to be predicted are input into improved LSTM model for load prediction, and load prediction result is output.The influence of price and historical data on load prediction is considered, and the influence of carbon trading price is also considered, and electric power user load prediction under the background of carbon trading market is realized.

Description

Technical Field

[0001] This application relates to the field of load forecasting technology, and in particular to a method and related apparatus for forecasting electricity user load considering carbon trading. Background Technology

[0002] The carbon trading market aims to control total carbon emissions through the trading of carbon emission rights. In the power industry, the introduction of the carbon trading market will incorporate carbon emission costs into the power generation costs of power generation companies, affecting the clearing order of the wholesale electricity market and thus changing the market clearing price, i.e., the electricity price after supply and demand are matched. This price change will be further transmitted to users' electricity prices through transactions in the retail electricity market, affecting users' electricity consumption behavior and thus changing their electricity load.

[0003] Current load forecasting technologies typically use historical electricity load data and electricity trading prices as samples, employing intelligent algorithms to predict and analyze future load curves for electricity users. Many of these algorithms require manual setting of time characteristics and careful consideration of data temporal correlations, and their predictive ability for long-term load data is limited. Because Long Short-Term Memory (LSTM) neural networks can effectively learn the temporal and nonlinear relationships of data based on historical information, they are increasingly being widely applied in the field of electricity load forecasting. However, LSTM model parameters often rely on human experience for selection, and different model parameters exhibit significant differences in fitting ability, training speed, and prediction performance, making the rational selection of LSTM model parameters a major challenge.

[0004] In summary, existing research has the following shortcomings: First, current load forecasting methods do not take into account the impact of the carbon trading market. In practical applications, they cannot capture the mechanism by which the carbon trading market affects users' electricity load and the nonlinear relationship between electricity price and carbon price, resulting in insufficient accuracy of forecast results. Second, even if the impact of the carbon trading market is taken into account, current load forecasting methods rely heavily on human experience to select model parameters, resulting in insufficient stability of forecast results. Summary of the Invention

[0005] This invention provides a method and software system for predicting electricity user load considering carbon trading, which addresses the technical problem that existing load forecasting does not take into account the influence of the carbon trading market, resulting in inaccurate forecasting results.

[0006] In view of this, the first aspect of the present invention provides a method for forecasting electricity user load considering carbon trading, comprising:

[0007] S1. Obtain the basic data required for power user load forecasting;

[0008] S2. Establish a nonlinear panel regression model that considers carbon trading to describe the nonlinear relationship between electricity prices and carbon prices;

[0009] S3. Perform regression analysis on the nonlinear panel regression model considering carbon trading to obtain the model parameters, and then obtain the predicted daily carbon price based on the predicted daily electricity price.

[0010] S4. The min-max normalization method is used to normalize the predicted daily electricity price, historical load data, and predicted daily carbon price;

[0011] S5. Based on the parameter optimization method of stick insect population evolution algorithm, establish an improved LSTM power user load prediction model;

[0012] S6. Input historical load data, electricity price and carbon price for the forecast date into the improved LSTM model for load forecasting, and output the load forecast results.

[0013] Preferably, the acquisition of basic data required for electricity user load forecasting includes:

[0014] Obtain the electricity price for the forecast date; obtain historical load data; obtain data on the core explanatory variables affecting electricity price stability, namely historical carbon prices; obtain data on control variables affecting electricity price stability, including coal prices, natural gas prices, economic activity index, temperature, thermal power generation, and electricity market maturity.

[0015] The economic activity index uses monthly GDP data, which is converted from quarterly GDP data and the industrial value growth rate index; the maturity of the electricity market is represented by the ratio of the number of provincial power generation enterprises to the number of electricity sales enterprises.

[0016] Preferably, the nonlinear panel regression model considering carbon trading specifically includes:

[0017]

[0018] In the formula, i represents the province, t represents time, and c0, c1, c2, c3, and c4 are regression parameters, which can be obtained by least squares regression estimation of the model. C it It is the carbon trading price, E it This represents the electricity trading price, which in the actual forecasting process is the daily electricity price to be forecasted. This represents exogenous control variables, including the prices of coal and natural gas. These are demand variables among the exogenous control variables, including the economic activity index and temperature. The supply variable representing the exogenous control variable, namely thermal power generation, γ i ε represents the maturity of the electricity market in the corresponding province. itIt is the fitted variable, which usually fluctuates randomly within a certain range;

[0019] This model reflects that power generation companies transmit the impact of carbon prices to electricity prices in a non-linear manner, that is, there is a non-linear relationship between carbon prices and electricity prices, and the current electricity trading price can be calculated based on the carbon trading price.

[0020] Preferably, the specific implementation process of obtaining the carbon price of the predicted day based on the predicted daily electricity price is as follows:

[0021] The predicted daily electricity price is obtained from the local power trading center, and a nonlinear panel regression model considering carbon trading is used for inverse solution. Based on the predicted daily electricity price and the control variable data affecting the stability of the electricity price, the predicted daily carbon price is obtained.

[0022] Preferably, the normalization process is as follows:

[0023] Each type of data is divided into three parts: training set, validation set, and test set. Normalization is performed using the training set as the standard, retaining the maximum and minimum values ​​in the training set. The maximum and minimum values ​​in the training set are then used to normalize the validation set and test set.

[0024] The data with uniform dimensions obtained after normalization can be represented as X = [X1, X2, ..., X...]. j ,…,X n ], where X j This indicates retrieving data for day j, where n represents the total number of days of historical data to be retrieved.

[0025] Preferably, the improved LSTM power user load forecasting model specifically comprises:

[0026] The load prediction model of Long Short-Term Memory Network (LSTM) optimized based on the stick insect population evolution algorithm (PPE) uses the number of units in the two hidden layers of LSTM, the learning rate, and the number of training iterations as features for PPE algorithm optimization. The PPE algorithm is used to adjust and optimize the LSTM model. In the hidden layers of LSTM, the output of each layer is used as the input of the next layer, and finally the data is output through a fully connected layer.

[0027] Preferably, the parameter optimization method based on the stick insect population evolution algorithm specifically involves the following process:

[0028] 1): A vector consisting of the electricity price, carbon price, and the equivalent amount of historical daily electricity load data before that day is considered as a sample, and multiple samples constitute a sample set;

[0029] 2): Initialize the evolutionary parameters of the stick insect population, including determining the population size p, the current evolutionary trend ev, the current population growth rate a, and the number of historical best solutions k;

[0030] 3): Construct an LSTM power load forecasting model, determine the optimization parameters and optimization range; randomly initialize N. p Each solution represents a stick insect population. The number of units in the two hidden layers of the LSTM, the learning rate, and the number of training iterations are used as key parameter variables to determine the range of parameter optimization. The sample set is divided into training set, validation set, and test set.

[0031] 4): Determine the evaluation function for the stick insect population; define the individual fitness value f(x) of the population with LSTM model parameters as:

[0032]

[0033] Where P and Q are the number of training samples and validation samples, respectively; x p , These are the true and predicted values ​​of the training samples, respectively; x q , These are the true values ​​and predicted values ​​of the validation samples, respectively;

[0034] 5): Calculate the fitness value of the stick insect population, evaluate based on the size of the fitness value, determine the global optimal solution gbest of the stick insect population, and record it as its historical best position.

[0035] 6): Based on the stick insect population evolution algorithm, for the number of solutions N... p Update the population size p, current evolutionary trend ev, and current population growth rate a of stick insects;

[0036] 7): Determine if the iteration termination condition is met; if the maximum number of iterations is reached, assign the optimal parameters to the LSTM and perform prediction, outputting the short-term power load prediction value; otherwise, return to 5) and continue execution until the termination condition is met.

[0037] A second aspect of the present invention provides a power user load forecasting software system that takes carbon trading into account, comprising:

[0038] The data acquisition module is used to acquire historical electricity load data of electricity users, the daily electricity price to be predicted, and the core explanatory and control variables that affect the stability of electricity prices.

[0039] The data association module is used to associate electricity prices with various variables that affect the stability of electricity prices, establish and calculate the parameters of a nonlinear panel regression model that considers carbon trading, and facilitate the prediction of carbon prices by utilizing the nonlinear relationship between electricity prices and carbon prices.

[0040] The carbon price prediction module is used to predict carbon prices. Based on the predicted daily carbon price obtained from the power trading center, the predicted daily carbon price is obtained through a nonlinear panel regression model that takes into account the carbon trading price.

[0041] The data processing module is used to normalize electricity price, carbon price, and historical load data of electricity users, so as to facilitate training and improving the LSTM electricity user load prediction model.

[0042] The model building module is used to build an LSTM load forecasting model optimized based on the PPE algorithm.

[0043] The load forecasting module is used to obtain the load forecast results of electricity users on the day to be forecasted and to visualize the results.

[0044] A third aspect of the present invention provides a computer-readable storage medium for storing program code for executing the electricity user load forecasting method considering carbon trading as described in the first aspect.

[0045] As can be seen from the above technical solutions, the present invention has the following beneficial effects compared with the prior art:

[0046] (1) The load forecasting method for electricity users considering carbon trading provided by the present invention uses a nonlinear panel regression model considering carbon trading to predict the carbon price of the day to be predicted based on the obtained daily electricity price to be predicted. Then, load forecasting is performed based on the daily electricity price to be predicted, historical load data and carbon price of the day to be predicted to obtain the load forecasting result. In this process, not only the influence of the daily electricity price to be predicted and historical load data on the forecasting result is considered, but also the influence of carbon trading price is considered. This realizes the load forecasting of electricity users under the background of carbon trading market and can ensure the accuracy and reliability of the forecasting result.

[0047] (2) The electricity user load forecasting method considering carbon trading provided by this invention establishes an improved LSTM electricity user load forecasting model based on the parameter optimization method of stick insect population evolution algorithm. In this process, the PPE algorithm is used to avoid the adverse effects of manually selecting LSTM model parameters, which can solve the problem of the difficulty in selecting LSTM model parameters and effectively improve the stability of electricity user load forecasting. Attached Figure Description

[0048] Figure 1 A flowchart illustrating a method for predicting electricity user load considering carbon trading, provided as an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of a power user load forecasting software system that takes carbon trading into account, provided as an embodiment of the present invention. Detailed Implementation

[0050] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] For easier understanding, please refer to Figure 1 This invention provides an embodiment of a power user load forecasting method that takes carbon trading into account, comprising:

[0052] Step S1: Obtain the basic data required for power user load forecasting.

[0053] Further, step S1 specifically includes:

[0054] Obtain the electricity price for the forecast date; obtain historical load data; obtain data on the core explanatory variables affecting electricity price stability, namely historical carbon prices; obtain data on control variables affecting electricity price stability, including coal prices, natural gas prices, economic activity index, temperature, thermal power generation, and electricity market maturity.

[0055] It should be noted that historical load data can be obtained directly from the power system, while the predicted daily electricity price is obtained from relevant electricity price data at the local power trading center and then predicted. The specific prediction method is a mature existing technology and will not be elaborated here. The economic activity index uses monthly GDP data, converted from quarterly GDP data and the industrial value growth rate index; the maturity of the electricity market is represented by the ratio of the number of provincial power generation enterprises to the number of electricity sales enterprises.

[0056] Step S2: Establish a nonlinear panel regression model that considers carbon trading to describe the nonlinear relationship between electricity price and carbon price.

[0057] It should be noted that in step S2, a nonlinear panel regression model for carbon trading is considered, which can be specifically described as follows:

[0058]

[0059] In the formula, i represents the province, t represents time, and c0, c1, c2, c3, and c4 are regression parameters, which can be obtained by least squares regression estimation of the model. C it It is the carbon trading price, E it This represents the electricity trading price, which in the actual forecasting process is the daily electricity price to be forecasted. This represents exogenous control variables, including the prices of coal and natural gas. These are demand variables among the exogenous control variables, including the economic activity index and temperature. The supply variable representing the exogenous control variable, namely thermal power generation, γ i ε represents the maturity of the electricity market in the corresponding province. it It is the fitted variable, which usually fluctuates randomly within a certain range;

[0060] This model reflects how power generation companies transmit the impact of carbon prices to electricity prices in a non-linear manner, meaning there is a non-linear relationship between carbon prices and electricity prices. The current electricity trading price can be calculated based on the carbon trading price. This embodiment uses this model for inverse solving, estimating the carbon price for the predicted day based on the predicted daily electricity price.

[0061] Step S3: Perform regression analysis on the nonlinear panel regression model considering carbon trading to obtain the model parameters, and then obtain the predicted daily carbon price based on the predicted daily electricity price.

[0062] It should be noted that while the method for obtaining the predicted daily electricity price is relatively mature in step S3, the method for predicting the carbon price is somewhat lacking and cannot achieve fine time granularity. Therefore, this invention employs a nonlinear panel regression model that considers carbon trading to obtain the predicted daily carbon price, ensuring the reliability of the prediction results. By substituting the obtained predicted daily electricity price and control variable data into the above equation and combining the model parameters obtained after regression analysis, the predicted daily carbon price can be obtained.

[0063] Step S4: Use the min-max normalization method to normalize the predicted daily electricity price, historical load data, and predicted daily carbon price.

[0064] It should be noted that, in step S4, the data with uniform dimensions obtained after the min-max normalization calculation can be represented as X = [X1, X2, ..., X...]. j ,…,X n ], where X j This indicates retrieving data for day j, where n represents the total number of days of historical data retrieved. The specific normalization process is as follows: divide each type of data into three parts: training set, validation set, and test set; perform normalization processing using the training set as the standard, retaining the maximum and minimum values ​​in the training set; and use the maximum and minimum values ​​in the training set to normalize the validation set and test set.

[0065] Step S5: Based on the parameter optimization method of stick insect population evolution algorithm, establish an improved LSTM power user load prediction model.

[0066] This embodiment uses an improved LSTM model for load prediction. Specifically, the model is a Long Short-Term Memory (LSTM) load prediction model optimized based on the Stick Insect Population Evolution (PPE) algorithm. The number of units in the two hidden layers of the LSTM, the learning rate, and the number of training iterations are used as features for the PPE algorithm to optimize the LSTM model. In the hidden layers of the LSTM, the output of each layer is used as the input of the next layer, and finally the data is output through a fully connected layer.

[0067] The specific process of adjusting and optimizing the key parameters of the LSTM model using the PPE algorithm is as follows:

[0068] 1): A vector consisting of the electricity price, carbon price, and the equivalent amount of historical daily electricity load data before that day is considered as a sample, and multiple samples constitute a sample set;

[0069] 2): Initialize the evolutionary parameters of the stick insect population, including determining the population size p, the current evolutionary trend ev, the current population growth rate a, and the number of historical best solutions k;

[0070] 3): Construct an LSTM power load forecasting model and determine the optimization parameters and optimization range. Randomly initialize N. p Each solution represents a stick insect population. The number of units in the two hidden layers of the LSTM, the learning rate, and the number of training iterations are used as key parameter variables to determine the range of parameter optimization. The sample set is divided into training set, validation set, and test set.

[0071] 4) Determine the evaluation function for the stick insect population. The fitness value f(x) of the individual population with LSTM model parameters is defined as:

[0072]

[0073] Where P and Q are the number of training samples and validation samples, respectively; x p , These are the true and predicted values ​​of the training samples, respectively; x q , These are the true values ​​and predicted values ​​of the validation samples, respectively;

[0074] 5): Calculate the fitness value of the stick insect population, evaluate based on the size of the fitness value, determine the global optimal solution gbest of the stick insect population, and record it as its historical best position.

[0075] 6): Based on the stick insect population evolution algorithm, for the number of solutions N... p Update the population size p, current evolutionary trend ev, and current population growth rate a of stick insects;

[0076] 7): Determine if the iteration termination condition is met. If the maximum number of iterations has been reached, assign the optimal parameters to the LSTM and perform prediction, outputting the short-term power load forecast value; otherwise, return to 5) and continue execution until the termination condition is met.

[0077] The present invention provides a load forecasting method for electricity users that considers carbon trading. It employs a nonlinear panel regression model that considers carbon trading to calculate the carbon price for the predicted day based on the acquired daily electricity price, historical load data, and the carbon price for the predicted day. Then, load forecasting is performed based on the daily electricity price, historical load data, and the carbon price for the predicted day to obtain the load forecasting result. This process considers not only the impact of the daily electricity price and historical load data on the forecasting result but also the impact of carbon trading prices, enabling load forecasting for electricity users in the context of a carbon trading market and ensuring the accuracy and reliability of the forecasting results. Therefore, the present invention solves the technical problem that existing load forecasting methods do not consider the influence of the carbon trading market, leading to a lack of accuracy in the forecasting results.

[0078] For easier understanding, please refer to Figure 2 This invention provides an embodiment of a power user load forecasting software system that considers carbon trading, comprising:

[0079] The data acquisition module is used to acquire historical electricity load data of electricity users, the daily electricity price to be predicted, and the core explanatory and control variables that affect the stability of electricity prices.

[0080] The data association module is used to associate electricity prices with various variables that affect the stability of electricity prices, establish and calculate the parameters of a nonlinear panel regression model that considers carbon trading, and facilitate the prediction of carbon prices by utilizing the nonlinear relationship between electricity prices and carbon prices.

[0081] The carbon price prediction module is used to predict carbon prices. Based on the predicted daily carbon price obtained from the power trading center, the predicted daily carbon price is obtained through a nonlinear panel regression model that takes into account the carbon trading price.

[0082] The data processing module is used to normalize electricity price, carbon price, and historical load data of electricity users, so as to facilitate training and improving the LSTM electricity user load prediction model.

[0083] The model building module is used to build an LSTM load forecasting model optimized based on the PPE algorithm.

[0084] The load forecasting module is used to obtain the load forecast results of electricity users on the day to be forecasted and to visualize the results.

[0085] The present invention also provides a computer-readable storage medium for storing program code for executing the electricity user load forecasting method considering carbon trading in the above method embodiments.

[0086] In the embodiments provided by this invention, it should be understood that the disclosed software systems and methods can be implemented in other ways. For example, the software system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between software systems or modules may be electrical, mechanical, or other forms.

[0087] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional modules in the various embodiments of the present invention 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. The integrated modules described above can be implemented in hardware or as software functional modules.

[0089] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods described in the various embodiments of the present invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0090] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting electricity user load considering carbon trading, characterized in that, include: S1. Obtain the basic data required for power user load forecasting; S2. Establish a nonlinear panel regression model that considers carbon trading to describe the nonlinear relationship between electricity prices and carbon prices; S3. Perform regression analysis on the nonlinear panel regression model that considers carbon trading prices to obtain model parameters, and then obtain the predicted daily carbon price based on the predicted daily electricity price; S4. The min-max normalization method is used to normalize the predicted daily electricity price, historical load data, and predicted daily carbon price; S5. A parameter optimization method based on the stick insect population evolution algorithm is used to establish an improved LSTM power user load forecasting model. Specifically, the improved LSTM power user load forecasting model is as follows: The load prediction model of Long Short-Term Memory Network (LSTM) optimized based on the Population Evolution Algorithm (PPE) of stick insects uses the number of units in the two hidden layers of LSTM, the learning rate, and the number of training iterations as features for PPE algorithm optimization. The PPE algorithm is used to adjust and optimize the LSTM model. In the hidden layers of LSTM, the output of each layer is used as the input of the next layer, and finally the data is output through a fully connected layer. In step S5, the parameter optimization method based on the stick insect population evolution algorithm is specifically as follows: 1): Consider the vector consisting of the electricity price, carbon price and the historical daily electricity load data of the previous day corresponding to a certain day as a sample, and multiple samples constitute a sample set; 2): Initialize the stick insect population evolution parameters, including determining the stick insect population size p, the current evolutionary trend ev, the current population growth rate a, and the number of historical best solutions k; 3): Construct an LSTM power load forecasting model, determine the optimization parameters and optimization range; randomly initialize N. p Each solution represents a stick insect population. The number of units in the two hidden layers of the LSTM, the learning rate, and the number of training iterations are used as key parameter variables to determine the range of parameter optimization. The sample set is divided into training set, validation set, and test set. 4): Determine the evaluation function for the stick insect population; define the individual fitness value f(x) of the population with LSTM model parameters as: （1） Where P and Q are the number of training samples and validation samples, respectively; , These are the true and predicted values ​​of the training samples, respectively. , These represent the true value and the predicted value of the validation sample, respectively. 5): Calculate the fitness value of the stick insect population, evaluate based on the size of the fitness value, determine the global optimal solution gbest of the stick insect population, and record it as its historical best position. 6): Based on the stick insect population evolution algorithm, for the number of solutions N... p Update the population size p, current evolutionary trend ev, and current population growth rate a of stick insects; 7): Determine if the iteration termination condition is met; if the maximum number of iterations is reached, assign the optimal parameters to the LSTM and perform prediction, outputting the short-term power load prediction value; otherwise, return to 5) and continue execution until the termination condition is met. S6. Input historical load data, electricity price and carbon price for the forecast date into the improved LSTM model for load forecasting, and output the load forecast results.

2. The electricity user load forecasting method considering carbon trading according to claim 1, characterized in that, The basic data required for load forecasting in step S1 includes: Electricity price on the forecast date; historical load data; data on the core explanatory variables affecting electricity price stability, namely historical carbon price; data on control variables affecting electricity price stability, including coal price, natural gas price, economic activity index, temperature, thermal power generation, and electricity market maturity; The economic activity index uses monthly GDP data, which is converted from quarterly GDP data and the industrial value growth rate index; the maturity of the electricity market is represented by the ratio of the number of provincial power generation enterprises to the number of electricity sales enterprises.

3. The electricity user load forecasting method considering carbon trading according to claim 1, characterized in that, In step S2, the nonlinear panel regression model considering carbon trading specifically refers to: (2) In the formula, i represents the province and t represents the time. , , , , These are all regression parameters, which can be obtained by performing parameter regression estimation on the model using the least squares method. It is the carbon trading price. This represents the electricity trading price, which in the actual forecasting process is the daily electricity price to be forecasted. This represents exogenous control variables, including the prices of coal and natural gas. These are demand variables among the exogenous control variables, including the economic activity index and temperature. The supply variable representing the exogenous control variable is thermal power generation. This indicates the maturity of the electricity market in the corresponding province. It is the fitted variable, which usually fluctuates randomly within a certain range; This model reflects that power generation companies transmit the impact of carbon prices to electricity prices in a non-linear manner, that is, there is a non-linear relationship between carbon prices and electricity prices, and the current electricity trading price can be calculated based on the carbon trading price.

4. The electricity user load forecasting method considering carbon trading according to claim 1, characterized in that, In step S3, the specific process of obtaining the carbon price of the predicted day based on the predicted daily electricity price is as follows: The predicted daily electricity price is obtained from the local power trading center, and a nonlinear panel regression model considering carbon trading is used for inverse solution. Based on the predicted daily electricity price and the control variable data affecting the stability of the electricity price, the predicted daily carbon price is obtained.

5. The electricity user load forecasting method considering carbon trading according to claim 1, characterized in that, In step S4, the normalization process is as follows: Each type of data is divided into three parts: training set, validation set, and test set. Normalization is performed using the training set as the standard, retaining the maximum and minimum values ​​in the training set. The maximum and minimum values ​​in the training set are then used to normalize the validation set and test set. The data with uniform dimensions obtained after normalization calculation can be expressed as: ,in, This indicates retrieving data for day j, where n represents the total number of days of historical data to be retrieved.

6. A software system for predicting electricity user load considering carbon trading, used to execute any one of the electricity user load prediction methods considering carbon trading according to claims 1-5, characterized in that, Specifically, it includes: The data acquisition module is used to acquire historical electricity load data of electricity users, the daily electricity price to be predicted, and the core explanatory and control variables that affect the stability of electricity prices. The data association module is used to associate electricity prices with various variables that affect the stability of electricity prices, establish and calculate the parameters of a nonlinear panel regression model that considers carbon trading, and facilitate the prediction of carbon prices by utilizing the nonlinear relationship between electricity prices and carbon prices. The carbon price prediction module is used to predict carbon prices. Based on the predicted daily carbon price obtained from the power trading center, the predicted daily carbon price is obtained through a nonlinear panel regression model that takes into account the carbon trading price. The data processing module is used to normalize electricity price, carbon price, and historical load data of electricity users, so as to facilitate training and improving the LSTM electricity user load prediction model. The model building module is used to build an LSTM load forecasting model optimized based on the PPE algorithm. The load forecasting module is used to obtain the load forecast results of electricity users on the day to be forecasted and to visualize the results.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code for executing the electricity user load forecasting method considering carbon trading as described in any one of claims 1-5.

Images