Large power consumer load probability prediction model construction method and system, electronic equipment and storage medium

Abstract

The invention relates to the technical field of power load prediction, and discloses a power large consumer load probability prediction model construction method and system, and the method comprises the steps: obtaining production data, weather data and historical load data; performing data mapping on the obtained data to obtain numerical information, and preprocessing the numerical information to obtain final numerical data; performing data analysis on the final numerical data to obtain correlation, stability and periodic information of different data types; classifying historical load data in the final numerical data according to correlation and stability information; inputting the classified historical load data into respective prediction models for training and evaluation to obtain optimized prediction models of different load types; and adjusting the production plan through the output value of the optimized prediction model. By constructing the customized load prediction model, the prediction result is more accurate, and multi-type load prediction information under different confidence intervals is provided to perform more comprehensive risk management.

Description

Technical Field

[0001] The present invention relates to the technical field of power load forecasting, and in particular to a method and system for constructing a load probability forecasting model for large power users. Background Art

[0002] With the continuous reform and improvement of the power market, more and more large power users are participating in power trading. Flexible electricity price policies and fluctuating market prices have brought about the possibility of significantly reducing electricity costs for large power users. For example, enterprises with high power consumption, such as the steel and electrolytic aluminum industries, are highly sensitive to electricity prices. Reasonable power consumption and demand management can significantly reduce the electricity costs of enterprises. Power load forecasting is of great significance in helping enterprises formulate production plans, optimize power procurement strategies and reduce electricity costs.

[0003] However, existing load forecasting methods have some problems in practical applications: (1) Traditional forecasting methods fail to fully consider the combination of electricity load in different industries with specific production plans and equipment characteristics, lack of model customization, resulting in insufficient forecasting accuracy. (2) Existing load forecasts are mostly point forecasts, which cannot provide uncertainty information for load forecasting, which is not conducive to the risk management and trading strategy formulation of enterprises. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for constructing a load probability prediction model for large power users to solve the problem that the existing load forecasting methods fail to fully consider the combination of electricity loads in different industries with specific production plans and equipment characteristics, lack model customization, resulting in insufficient prediction accuracy and inability to provide uncertainty information for load forecasting.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for constructing a load probability prediction model for a large power user, comprising: acquiring production data, weather data, and historical load data; The acquired data is mapped to obtain numerical information and preprocessed to obtain final numerical data; Perform data analysis on the final numerical data to obtain information on the correlation, stationarity, and periodicity of different data types; classifying the historical load data in the final numerical data according to the correlation and stationarity information; The classified historical load data are input into respective prediction models for training and evaluation to obtain optimized prediction models for different load types; The production plan is adjusted according to the output value of the optimized prediction model.

[0007] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, wherein: the production data includes: date type, holiday name, production plan and maintenance plan; The weather data includes: weather history data and weather forecast data.

[0008] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, the acquired data is subjected to data mapping to obtain numerical information, including: The date types in the production data, holiday names, equipment in production in the production plan and maintenance plan, and equipment that is under maintenance or stopped production are mapped to corresponding different numbers.

[0009] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, the load data in the final numerical data is classified according to the correlation and stationarity information, including: Based on the correlation between load data and production plan, the load is divided into production load and life power load; According to the stability value of load data, productive loads are further classified into impact loads and stable loads.

[0010] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, the classified load data is input into respective prediction models for training and evaluation, including: Determine the data window length for input model based on periodic data; Divide the input data into training set, validation set and test set in proportion, and input them into respective load forecasting models for model training; Model training is performed through long short-term memory networks to capture temporal dependencies and nonlinear relationships in load data; The fitting effect of the model is evaluated based on the validation set. If there is no underfitting or overfitting, the prediction accuracy of the model is evaluated based on the test set.

[0011] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, wherein: the model after hyperparameter optimization is evaluated, and the optimization includes optimizing the window length of input training data, the number of hidden layers, the number of hidden layer nodes, and the learning rate; The evaluated optimal model is used to forecast multiple types of loads under different confidence intervals for future time periods.

[0012] As a preferred solution of the method for constructing a load probability prediction model for large power users described in the present invention, wherein: the prediction is performed based on the information coupling relationship between historical load data and production plans within a preset time period, combined with electricity fee information to predict and analyze the production plan; Output forecast analysis results and adjust production plans to optimize resource allocation and production efficiency.

[0013] In a second aspect, the present invention provides a system for constructing a load probability prediction model for a large power user, including: an acquisition module for acquiring production data, weather data, and historical load data; The preprocessing module is used to perform data mapping on the acquired data, obtain numerical information and perform preprocessing to obtain final numerical data; The analysis module is used to perform data analysis on the final numerical data to obtain the correlation, stationarity and periodicity information of different data types; A classification module, used for classifying the historical load data in the final numerical data according to the correlation and stationarity information; The training and evaluation module is used to input the classified historical load data into the respective prediction models for training and evaluation, and obtain the optimized prediction models for different load types; The forecast adjustment module is used to adjust the production plan according to the output value of the optimized forecast model.

[0014] In a third aspect, the present invention provides a computing device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the method for constructing a load probability prediction model for large power users are implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for constructing a load probability prediction model for large power users.

[0016] Compared with the prior art, the present invention has the following beneficial effects: the present invention constructs a customized load forecasting model by comprehensively considering the factory's production plan and equipment characteristics, making the forecast result more accurate and reliable; adopts a probabilistic forecasting method to provide multi-type load forecasting information under different confidence intervals for more comprehensive risk management and decision-making; through in-depth analysis of electricity and demand, dynamically adjusts the production plan according to forecast data and historical data, optimizes resource allocation and improves production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 A schematic diagram of the overall process logic of a method for constructing a load probability prediction model for large power users according to an embodiment of the present invention; Figure 2 A schematic diagram of steel plant load prediction in a method for constructing a load probability prediction model for large power users according to an embodiment of the present invention; Figure 3 A schematic diagram of a module framework of a system for constructing a load probability prediction model for large power users according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0020] Example 1, reference Figure 1-Figure 2 , which is an embodiment of the present invention, provides a method for constructing a load probability prediction model for large power users, comprising: S100: Obtain production data, weather data and historical load data; In the embodiment of the present application, the production data includes: date type, holiday name, production plan and maintenance plan; Weather data includes: weather history data and weather forecast data.

[0021] S200: performing data mapping on the acquired data to obtain numerical information and preprocessing to obtain final numerical data; In the embodiment of the present application, the acquired data is subjected to data mapping to obtain numerical information, including: The date types in the production data, holiday names, equipment in production in the production plan and maintenance plan, and equipment that is under maintenance or stopped production are mapped to corresponding different numbers.

[0022] Exemplarily, data mapping is performed to map two types of character information, date type and holiday name, into numerical information required for model training, and the production plan and maintenance plan are mapped into numerical information, with equipment production being 1 and maintenance or production stop being 0, so that the numerical value is consistent with the degree of impact on the load, so as to facilitate subsequent input into the network for training.

[0023] Specifically, the preprocessing process may include: data cleaning, filling in missing data through forward and backward filling or linear filling to ensure data integrity and continuity; outlier detection through statistical methods or machine learning methods, deleting outliers or replacing outliers to improve data quality; data normalization based on statistical methods to eliminate the impact of different dimensions on model training; and dividing the entire data set according to a certain ratio to construct a training set, a validation set, and a test set respectively.

[0024] For example, the entire data set can be divided into a ratio of 8:1:1 to construct a training set, a validation set, and a test set, respectively.

[0025] S300: Perform data analysis on the final numerical data to obtain information on the correlation, stability and periodicity of different data types; Specifically, correlation can be calculated using the Pearson correlation coefficient or the Spearman rank correlation coefficient; stationarity can be calculated using the unit root test or the autocorrelation function; and periodicity information can be calculated using the Fourier transform or seasonal decomposition.

[0026] S400: classifying the historical load data in the final numerical data according to the correlation and stability information; In the embodiment of the present application, the load data in the final numerical data is classified according to the correlation and stationarity information, including: Based on the correlation between load data and production plan, the load is divided into production load and life power load; According to the stability value of load data, productive loads are further classified into impact loads and stable loads.

[0027] It should be noted that the overall load data is decomposed into different components in order to better understand the components of the load and thus improve the accuracy of the prediction.

[0028] S500: inputting the classified historical load data into respective prediction models for training and evaluation to obtain optimized prediction models for different load types; In the embodiment of the present application, the classified load data is input into respective prediction models for training and evaluation, including: Determine the data window length for input model based on periodic data; Divide the input data into training set, validation set and test set in proportion, and input them into respective load forecasting models for model training; Model training is performed through long short-term memory networks to capture temporal dependencies and nonlinear relationships in load data; The fitting effect of the model is evaluated based on the validation set. If there is no underfitting or overfitting, the prediction accuracy of the model is evaluated based on the test set.

[0029] It should be noted that multiple sub-models are trained using long short-term memory networks to capture the temporal dependencies and complex nonlinear relationships in load data; the model parameter optimization and result visualization module optimizes the model performance through tuning and visualization techniques, and displays the prediction results and model performance.

[0030] For example, the above steps S300-S500 are based on the collected factory load data, taking a steel plant as an example: Including: non-productive load data of residents' lives, rolling mill load, electric furnace load, sintering, oxygen production system and water treatment system load. Different types of loads have different volatility, periodicity and trend. In order to improve the prediction accuracy, different types of loads are modeled and predicted separately.

[0031] The loads are divided into production loads and life power loads according to whether they are related to the production plan; The productive loads can be further classified according to whether the loads are stable or not. Impact loads include the loads of rolling mills, refining furnaces and electric furnaces. Stable loads include the loads of sintering, iron and steel making, and auxiliary production oxygen production systems and water treatment systems. The window length of the model input data is determined according to the periodicity of the data. The three types of loads in the data set are matched with weather data, production plan data and maintenance data and input into their respective load forecasting models for model training and evaluation.

[0032] It should be noted that productive loads are highly correlated with production plans, while non-productive loads are mainly employees’ electricity consumption for daily life, which are highly correlated with the weather. This can help us better understand the components of the loads and thus improve the accuracy of the forecast.

[0033] In the embodiment of the present application, the model after hyperparameter optimization is evaluated, and the optimization includes the optimization of the window length of the input training data, the number of hidden layers, the number of hidden layer nodes and the learning rate; The evaluated optimal model is used to forecast multiple types of loads under different confidence intervals for future time periods.

[0034] S600: Adjust the production plan according to the output value of the optimized prediction model.

[0035] In the embodiment of the present application, the prediction is performed by combining the electricity fee information with the information coupling relationship between the historical load data and the production plan within a preset time period to perform a prediction analysis on the production plan; Output forecast analysis results and adjust production plans to optimize resource allocation and production efficiency.

[0036] Through in-depth analysis of forecasts and historical data, we help companies formulate reasonable production plans, optimize resource allocation, and improve production efficiency.

[0037] The model prediction results are as follows Figure 2 As shown in the figure, the model has high prediction accuracy and higher application value compared with traditional prediction methods.

[0038] Embodiment 2, the above is a schematic scheme of a method for constructing a load probability prediction model for large power users in this embodiment. It should be noted that the technical scheme of the system for constructing a load probability prediction model for large power users and the technical scheme of the method for constructing a load probability prediction model for large power users described above belong to the same concept. For details not described in detail in the technical scheme of the system for constructing a load probability prediction model for large power users in this embodiment, please refer to the description of the technical scheme of the method for constructing a load probability prediction model for large power users described above.

[0039] like Figure 3 As shown, the load probability prediction model construction system for large power users in this embodiment includes: An acquisition module 100 is used to acquire production data, weather data and historical load data; The preprocessing module 200 is used to perform data mapping on the acquired data to obtain numerical information and perform preprocessing to obtain final numerical data; The analysis module 300 is used to perform data analysis on the final numerical data to obtain the correlation, stability and periodicity information of different data types; A classification module 400, for classifying the historical load data in the final numerical data according to the correlation and stationarity information; The training and evaluation module 500 is used to input the classified historical load data into respective prediction models for training and evaluation, and obtain the optimized prediction models for different load types; The prediction adjustment module 600 is used to adjust the production plan according to the output value of the optimized prediction model.

[0040] This embodiment further provides a computing device, which is applicable to the case of constructing a load probability prediction model for large power users, and includes: Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the method for constructing a load probability prediction model for large power users as proposed in the above embodiment.

[0041] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for constructing a load probability prediction model for large power users proposed in the above embodiment is implemented.

[0042] The storage medium proposed in this embodiment and the method for constructing a load probability prediction model for large power users proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0043] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (FLASH), a hard disk or an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0044] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for constructing a load probability prediction model for large power users, characterized in that: include: Obtain production data, weather data, and historical load data; The acquired data is mapped to obtain numerical information and preprocessed to obtain final numerical data; Perform data analysis on the final numerical data to obtain information on the correlation, stationarity, and periodicity of different data types; classifying the historical load data in the final numerical data according to the correlation and stationarity information; The classified historical load data are input into respective prediction models for training and evaluation to obtain optimized prediction models for different load types; The production plan is adjusted according to the output value of the optimized prediction model.

2. The method for constructing a load probability prediction model for large power users according to claim 1, characterized in that: The production data includes: date type, holiday name, production plan and maintenance plan; The weather data includes: weather history data and weather forecast data.

3. The method for constructing a load probability prediction model for large power users according to claim 1 or 2, characterized in that: The acquired data is mapped to obtain numerical information, including: The date types in the production data, holiday names, equipment in production in the production plan and maintenance plan, and equipment that is under maintenance or stopped production are mapped to corresponding different numbers.

4. The method for constructing a load probability prediction model for large power users according to claim 3, characterized in that: Classifying the load data in the final numerical data according to the correlation and stationary information includes: Based on the correlation between load data and production plan, the load is divided into production load and life power load; According to the stability value of load data, productive loads are further classified into impact loads and stable loads.

5. The method for constructing a load probability prediction model for large power users according to any one of claims 1, 2 or 4, characterized in that: The classified load data is input into the respective prediction models for training and evaluation, including: Determine the data window length for input model based on periodic data; Divide the input data into training set, validation set and test set in proportion, and input them into respective load forecasting models for model training; Model training is performed through long short-term memory networks to capture temporal dependencies and nonlinear relationships in load data; The fitting effect of the model is evaluated based on the validation set. If there is no underfitting or overfitting, the prediction accuracy of the model is evaluated based on the test set.

6. The method for constructing a load probability prediction model for large power users according to claim 4, characterized in that: Evaluate the model after hyperparameter optimization, wherein the optimization includes the optimization of the window length of the input training data, the number of hidden layers, the number of hidden layer nodes, and the learning rate; The evaluated optimal model is used to forecast multiple types of loads under different confidence intervals for future time periods.

7. The method for constructing a load probability prediction model for large power users according to any one of claims 1, 2, 4 or 6, characterized in that: The prediction is performed by combining electricity fee information with the information coupling relationship between the historical load data and the production plan within a preset time period to predict and analyze the production plan; Output forecast analysis results and adjust production plans to optimize resource allocation and production efficiency.

8. A system based on a method for constructing a load probability prediction model for large power users, characterized in that: include: An acquisition module (100), used to acquire production data, weather data and historical load data; A preprocessing module (200) is used to perform data mapping on the acquired data to obtain numerical information and perform preprocessing to obtain final numerical data; An analysis module (300) is used to perform data analysis on the final numerical data to obtain information on the correlation, stability and periodicity of different data types; A classification module (400), used for classifying the historical load data in the final numerical data according to the correlation and stationarity information; A training and evaluation module (500) is used to input the classified historical load data into respective prediction models for training and evaluation, so as to obtain optimized prediction models for different load types; The forecast adjustment module (600) is used to adjust the production plan according to the output value of the optimized forecast model.

9. An electronic device, characterized in that: include: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the method for constructing a load probability prediction model for large power users as described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium, characterized in that: Computer executable instructions are stored, and when the computer executable instructions are executed by a processor, the steps of the method for constructing a load probability prediction model for large power users as described in any one of claims 1-7 are implemented.

Images