Load data prediction method and device, computer readable storage medium, processor and computer program product

Abstract

The invention discloses a load data prediction method and device, a computer readable storage medium, a processor and a computer program product. The method comprises the following steps: determining a time period to be predicted, the time period to be predicted comprising at least one sub-time period; a meteorological characteristic parameter set, day types corresponding to the sub-periods and a historical load data set corresponding to the historical periods of the region where the power system is located in the to-be-predicted period are obtained, the meteorological characteristic parameter set comprises at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; and based on the load demand data corresponding to the day type, the meteorological characteristic parameter set and the historical load data set, predicting to obtain predicted load data of the power system in the sub-period. The technical problem of low accuracy of load data prediction is solved.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular, to a method, device, computer-readable storage medium, processor, and computer program product for predicting load data. Background Art

[0002] Currently, with the wide access of renewable energy and the popularization of electric vehicles, the prediction of load data in power systems has become increasingly complex.

[0003] In related technologies, for the prediction of load data, methods such as regression analysis, time series, and exponential smoothing are usually used to predict load data. However, the above methods are difficult to adapt to the non-linear characteristics of the load data sequence, and there is a technical problem of low accuracy in load data prediction.

[0004] For the above problems, no effective solution has been proposed yet. Summary of the Invention

[0005] Embodiments of the present invention provide a method, device, computer-readable storage medium, processor, and computer program product for predicting load data, so as to at least solve the technical problem of low accuracy in load data prediction.

[0006] According to one aspect of the embodiments of the present invention, a method for predicting load data is provided. The method may include: determining a time period to be predicted, where the time period to be predicted includes at least one sub-time period; obtaining a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-time period, and a historical load data set corresponding to the historical time period in the time period to be predicted, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-time period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-time period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0007] Optionally, predicting the predicted load data of the power system in the sub-time period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set includes: screening at least one target meteorological characteristic from the set of meteorological characteristic parameters, where the influence of the target meteorological characteristic on the predicted load data is greater than the influence of other meteorological characteristics in the set of meteorological characteristic parameters on the predicted load data; predicting the predicted load data based on the load demand data, at least one target meteorological characteristic, and the historical load data set.

[0008] Optionally, at least one target meteorological feature is screened out from the meteorological feature parameter set, including: determining the target meteorological feature type corresponding to the region; determining, from the meteorological feature parameter set, the meteorological features whose meteorological feature type is the target meteorological feature type as the target meteorological features.

[0009] Optionally, based on the load demand data, at least one target meteorological feature, and the historical load data set, predicted load data is obtained, including: inputting the load demand data, at least one target meteorological feature, and the historical load data set into a prediction model to obtain the predicted load data, where the prediction model is trained based on the historical load data set.

[0010] Optionally, the method may further include: obtaining the historical load data set and the historical meteorological feature parameter set of the region; using the historical load data set and the historical meteorological feature parameter set as training data, and training the sub-prediction model according to the preset hyperparameters to obtain the prediction model.

[0011] Optionally, using the historical load data set and the historical meteorological feature parameter set as training data includes: respectively performing interpolation processing on the historical load data set and the historical meteorological feature parameter set; using the interpolated historical load data set and the interpolated historical meteorological feature parameter set as training data.

[0012] According to another aspect of the embodiments of the present invention, a prediction device for load data is further provided. The device may include: a determination unit for determining a to-be-predicted period, where the to-be-predicted period includes at least one sub-period; an acquisition unit for acquiring, during the to-be-predicted period, the meteorological feature parameter set of the region where the power system is located, the day type corresponding to the sub-period, and the historical load data set corresponding to the historical period, where the meteorological feature parameter set includes at least one meteorological feature corresponding to at least one sub-period, and different day types correspond to different load demand data; a prediction unit for predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the meteorological feature parameter set, and the historical load data set.

[0013] According to another aspect of the embodiments of the present invention, a computer-readable storage medium is further provided. The computer-readable storage medium includes a stored program, where, when the program runs, it controls the device where the computer-readable storage medium is located to execute the prediction method for load data according to the embodiments of the present invention.

[0014] According to another aspect of the embodiments of the present invention, a processor is further provided. The processor is used to run a program, where, when the program is run by the processor, it executes the prediction method for load data according to the embodiments of the present invention.

[0015] According to another aspect of the embodiments of the present invention, there is also provided a computer program product, which includes computer instructions. When the computer instructions are executed by a processor, the prediction method of load data in the embodiments of the present invention is implemented.

[0016] In the embodiments of the present invention, a to-be-predicted time period is determined, where the to-be-predicted time period includes at least one sub-time period; meteorological characteristic parameter sets of the area where the power system is located, the day type corresponding to the sub-time period, and the historical load data sets corresponding to the historical time periods are obtained during the to-be-predicted time period. The meteorological characteristic parameter sets include at least one meteorological characteristic corresponding to at least one sub-time period, and different day types correspond to different load demand data; based on the load demand data corresponding to the day type, the meteorological characteristic parameter sets, and the historical load data sets, the predicted load data of the power system in the sub-time period is predicted. That is to say, in the embodiments of the present invention, considering that different day types correspond to different load demand data, based on the load demand data corresponding to the day type, the meteorological characteristic parameter sets, and the historical load data sets, the power system is predicted to obtain the predicted load data in the sub-time period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0018] Figure 1 is a flowchart of a method for predicting load data according to an embodiment of the present invention;

[0019] Figure 2 is a flowchart of a short-term load prediction method for optimizing hyperparameters of an ETSformer model using a COA algorithm according to an embodiment of the present invention;

[0020] Figure 3 is a schematic diagram of the correlation between different types of features according to an embodiment of the present invention;

[0021] Figure 4 is a schematic diagram of an optimization iteration curve according to an embodiment of the present invention;

[0022] Figure 5 is a schematic diagram of a load data prediction device according to an embodiment of the present invention;

[0023] Figure 6 is a structural block diagram of a computer terminal according to an embodiment of the present invention;

[0024] Figure 7A block diagram of an electronic device for a method of predicting load data according to an embodiment of the present application. Detailed implementation manners

[0025] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data used can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order different from those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment 1

[0028] According to an embodiment of the present invention, an embodiment of a method for predicting load data is provided. The steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And, although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0029] In this embodiment, a method for predicting load data is proposed. This method takes into account that the load demand data corresponding to different day types are different, and based on the load demand data corresponding to the day type, the meteorological characteristic parameter set, and the historical load data set, the power system is predicted to obtain the predicted load data in the sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0030] Figure 1 A flowchart of a method for predicting load data according to an embodiment of the present invention. As Figure 1 shown, the method may include the following steps:

[0031] Step S102: Determine the time period to be predicted, where the time period to be predicted includes at least one sub-time period.

[0032] In the technical solution provided in step S102 of the present invention, the time period to be predicted can be the time period for which load data prediction is required, which can be a certain day or a certain moment. The sub-time period can be a certain day in a month, a certain month in a year, or a certain moment in a day. It should be noted that this is only an example, and no specific restrictions are imposed on the lengths of the time period to be predicted and the sub-time period.

[0033] Optionally, the time period to be predicted can be a future time window to be concerned about in power load prediction, and can include one or more sub-time periods. For example, if the prediction resolution is 15 minutes, then the time period to be predicted within a day can be divided into 96 sub-time periods (24 hours × 4 15-minute segments). Among them, the prediction resolution can range from dozens of minutes to several days.

[0034] For example, the time period to be predicted is determined by means of inputting instructions or selection operations, etc. For example, through a touch operation, the corresponding time period is selected on the display, and the time period to be predicted is determined according to the selected content.

[0035] Step S104: Obtain the meteorological characteristic parameter set of the area where the power system is located, the day type corresponding to the sub-time period, and the historical load data set corresponding to the historical time period within the time period to be predicted, where the meteorological characteristic parameter set includes at least one meteorological characteristic corresponding to at least one sub-time period, and different day types correspond to different load demand data.

[0036] In the technical solution provided in step S104 of the present invention, the day type can be the date type corresponding to the time period to be predicted, for example, it can include types such as weekdays, rest days, and holidays. The historical time period can be a certain time period before the time period to be predicted. The meteorological characteristic parameter set can include at least one meteorological characteristic corresponding to at least one sub-time period, and the meteorological characteristic can be data such as temperature, humidity, rainfall, solar intensity, etc. It should be noted that this is only an example, and no specific restrictions are imposed on the type of meteorological characteristic. The load demand data can be used to characterize the demand situation of the load data. It should be noted that this is only an example, and no specific restrictions are imposed on the types of the day type, the meteorological characteristic type, and the load demand data type.

[0037] Optionally, after determining the time period to be predicted, the meteorological characteristic parameter set corresponding to at least one sub-moment within the time period to be predicted can be obtained in advance through the Internet, satellite testing, etc., to obtain the meteorological characteristic parameter set corresponding to the time period to be predicted. Determine the day type corresponding to the sub-time period according to the meteorological characteristic parameter corresponding to the time period to be predicted, for example, determine whether the sub-time period is a holiday, a weekday, or a weekend. And obtain the historical load data set corresponding to the historical time period.

[0038] Optionally, the above date type can be determined based on calendar rules. For example, working days, rest days, holidays, etc. There is no specific limitation on the date type here.

[0039] Optionally, after determining the time period to be predicted, meteorological data (i.e., meteorological characteristics) of the region where the power system is located can be obtained. The meteorological characteristics can include data such as temperature, humidity, solar irradiance, rainfall, and wind speed. Sampling can be performed every 15 minutes to construct a set of meteorological characteristic parameters. The set of meteorological characteristic parameters will include the meteorological characteristics corresponding to each sub-period, and these meteorological characteristics have a significant impact on the load data of the power system. At the same time, the date type corresponding to each sub-period can be determined. The date type can refer to working days, weekends, holidays, or special event days, etc., because the load patterns of the power system may be different on different types of days. For example, on weekends or holidays, the load curve may be flatter, while on working days, there may be obvious peak and valley changes. Among them, the date type can be processed by one-hot encoding to capture the periodic characteristics of the load data. Further, a historical load data set can be collected. The historical load data set can contain power load records over a past period of time, covering a long enough time period to reflect the seasonal and intra-day change patterns of the load. The historical load data can be used to train the model so that the model can learn the laws of power load changes, and the historical load data can be input into the prediction model to predict the predicted load data of the power system corresponding to the sub-period using the prediction model.

[0040] For example, it is determined that the time period to be predicted is between February 1st and February 10th. The set of meteorological characteristic parameters of the region where the power system is located in each sub-period from February 1st to February 10th can be obtained. And the historical load data set corresponding to the power system before February 1st. And the date type corresponding to each sub-period from February 1st to February 10th is determined.

[0041] Step S106: Based on the load demand data corresponding to the date type, the set of meteorological characteristic parameters, and the historical load data set, predict the predicted load data of the power system in the sub-period.

[0042] In the technical solution provided in step S106 of the present invention above, based on the date type, the load demand data corresponding to the sub-period is determined. Based on the load demand data, the set of meteorological characteristic parameters, and the historical load data set, the predicted load data of the power system corresponding to the sub-period can be predicted through a prediction model or a mathematical model. For example, it can be determined that the predicted load data set corresponding to March 2nd is 30 kW.

[0043] Optionally, through a prediction model, using a historical load dataset, a set of meteorological characteristic parameters, and load demand data corresponding to the day type, the load demand data of the power system can be predicted. Among them, the above historical load dataset can be used to capture the time series characteristics of the load data of the power system. For example, the trend and periodicity of the load data. The above set of meteorological characteristic parameters can be used to provide information related to weather conditions, and there is a correlation between this information and the load data. The load demand data corresponding to the day type can be used to provide typical patterns of the load under different day types. By the day type, the accuracy of predicting the load data on specific types of days can be improved.

[0044] Furthermore, input the above load demand data, set of meteorological characteristic parameters, and historical load dataset into a prediction model (such as, the ETSformer model) to predict the predicted load data, where the above period to be predicted can be a certain period in the future.

[0045] To further understand the above steps, a prediction scenario can be assumed: predicting the short-term power load of a certain regional power grid in Beijing on the National Day on October 1, 2023. The whole day can be divided into 96 15-minute sub-periods, from sub-period 1 (00:00 - 00:15) to sub-period 96 (23:45 - 24:00), to complete the load prediction for the whole day. It can be determined that the period to be predicted is the whole day of October 1, 2023, including 96 15-minute sub-periods. A set of meteorological characteristic parameters of the area where the power system is located during the period to be predicted can be obtained. This set of meteorological characteristic parameters can include: Temperature: On the National Day, the predicted daytime temperature is 20 degrees Celsius (°C), and the temperature drops to 15 °C at night; Humidity: The humidity throughout the day is expected to be around 50%; Rainfall: No rain is expected, and the rainfall is 0, and Wind speed: The average wind speed throughout the day is 5 meters per second (m / s). It is determined that the day type on the National Day is a holiday, which is different from ordinary working days, and the power load may be significantly different. Therefore, it can be determined that the day type of the sub-period is a holiday. Finally, based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load dataset, the predicted load data of the power system in the sub-period can be predicted.

[0046] Optionally, the above predicted load data may be as follows: from sub-period 1 (00:00 - 00:15) to sub-period 24 (06:00 - 06:15), the load prediction value may be 2,500 megawatts (MW for short), because it is late at night and the load demand is usually low; from sub-period 25 (06:15 - 06:30) to sub-period 48 (12:00 - 12:15), as the day arrives, the temperature rises, and the solar irradiance increases, the load prediction value may rise to 3,000 MW; from sub-period 49 (12:15 - 12:30) to sub-period 72 (18:00 - 18:15), the load prediction value may reach a peak of approximately 3,500 MW; from sub-period 73 (18:15 - 18:30) to sub-period 96 (23:45 - 24:00), as night falls, the load demand gradually decreases, and the prediction value may drop below 3,000 MW. It should be noted that the above predicted load data is only for illustrative purposes and is not specifically limited here.

[0047] Through steps S102 and S106 of the present invention, the period to be predicted is determined, where the period to be predicted includes at least one sub-period; the meteorological characteristic parameter set of the area where the power system is located, the day type corresponding to the sub-period, and the historical load data set corresponding to the historical period are obtained during the period to be predicted, where the meteorological characteristic parameter set includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; based on the load demand data corresponding to the day type, the meteorological characteristic parameter set, and the historical load data set, the predicted load data of the power system in the sub-period is predicted. That is to say, in the embodiment of the present invention, considering that different day types correspond to different load demand data, based on the load demand data corresponding to the day type, the meteorological characteristic parameter set, and the historical load data set, the power system is predicted to obtain the predicted load data in the sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0048] The above method of this embodiment will be further introduced below.

[0049] As an optional implementation manner, step S106, based on the load demand data corresponding to the day type, the meteorological characteristic parameter set, and the historical load data set, predicting the predicted load data of the power system in the sub-period includes: screening out at least one target meteorological characteristic from the meteorological characteristic parameter set, where the influence of the target meteorological characteristic on the predicted load data is greater than the influence of other meteorological characteristics in the meteorological characteristic parameter set on the predicted load data; predicting the predicted load data based on the load demand data, at least one target meteorological characteristic, and the historical load data set.

[0050] In this embodiment, after obtaining the meteorological feature parameter set, to improve the prediction accuracy, at least one target meteorological feature can be selected from the meteorological feature parameter set. The influence of this target meteorological feature on predicting the load data is greater than that of other meteorological features in the meteorological feature parameter set on predicting the load data. Further, based on the load demand data, at least one target meteorological feature, and the historical load data set, the predicted load data can be obtained.

[0051] Optionally, by using a machine learning algorithm (such as, the xgboost algorithm) for feature screening, at least one target meteorological feature can be selected from the meteorological feature parameter set.

[0052] Optionally, through a machine learning algorithm, the type of target meteorological feature corresponding to this area can be determined in advance. After obtaining the meteorological feature parameter set, at least one target meteorological feature can be selected from the meteorological feature parameter set according to the target feature type.

[0053] As an alternative implementation, selecting at least one target meteorological feature from the meteorological feature parameter set includes: determining the type of target meteorological feature corresponding to the area; and determining, from the meteorological feature parameter set, the meteorological features whose meteorological feature types are the target meteorological feature type as the target meteorological features.

[0054] In this embodiment, by processing the historical load data set and the historical meteorological feature parameter set of the areas where different power systems are located through a machine learning algorithm, the type of target meteorological feature corresponding to each area can be determined. After obtaining the meteorological feature parameter set, according to the area where the power system to be predicted is located, the type of target meteorological feature corresponding to this area can be determined, and then, according to this target meteorological feature type, the meteorological features whose meteorological feature types are the target meteorological feature type are determined as the target meteorological features from the meteorological feature parameter set.

[0055] Optionally, the xgboost algorithm is used to calculate the correlation between features, quantitatively analyze the specific influence of meteorological data on the load sequence, so as to screen out features with higher correlation and filter out features with weak correlation with the load, in order to improve the prediction accuracy.

[0056] For example, to obtain a historical load data set and a historical meteorological feature parameter set, the historical load data set and the historical meteorological feature parameter set can be input into the xgboost feature analysis module. By deeply analyzing the correlation between different historical meteorological features, this module determines the feature type corresponding to the target meteorological feature. After obtaining the meteorological feature parameter set, the xgboost feature analysis module can be used to obtain the meteorological feature corresponding to the pre-determined target meteorological feature type from the meteorological feature parameter set, and determine this meteorological feature as the target meteorological feature to obtain at least one target meteorological feature. Through the above steps, the interference caused by features with low correlation to the prediction process is effectively avoided, thus ensuring the prediction accuracy of the model.

[0057] As an optional implementation manner, based on the load demand data, at least one target meteorological feature, and the historical load data set, predicting the predicted load data includes: inputting the load demand data, at least one target meteorological feature, and the historical load data set into a prediction model to obtain the predicted load data, where the prediction model is trained based on the historical load data set.

[0058] In this embodiment, the load demand data, at least one target meteorological feature, and the historical load data set can be processed through the prediction model to obtain the predicted load data. Among them, the above prediction model can be a neural network model (such as, the ETSformer model), and can be trained based on the historical load data set. It should be noted that only examples are given here, and the type of the prediction model is not specifically limited.

[0059] Optionally, select the features that have the greatest impact on the predicted load data from all the collected meteorological feature parameter sets, which are usually called "target meteorological features". The weights of these features in the prediction model will be higher, and their contributions to the final prediction result will also be greater. Use the selected meteorological features (target meteorological features), historical load data, and the load demand pattern corresponding to the day type, input them into the trained prediction model, and perform the power load prediction for sub-periods.

[0060] For example, assume that the analysis results show that temperature and solar irradiance are the "target meteorological features" most closely related to the change in power load. Then, the data of these target meteorological features can be combined with the historical load dataset and the load demand data of a specific day type (such as National Day) and input into the ETSformer model for training. During the training process, the sub-prediction model will automatically learn the complex relationship between these meteorological features and the load, especially how temperature and solar irradiance affect power demand. Finally, the temperature and solar irradiance data for the future prediction period can be input into the trained ETSformer model (i.e., the prediction model), and the predicted load data for the corresponding sub-period can be obtained.

[0061] Optionally, through the above steps, the prediction model can combine key meteorological information with historical load data and the demand patterns of specific day types to perform more accurate load forecasting.

[0062] As an alternative implementation, the method may further include: obtaining a historical load dataset and a set of historical meteorological feature parameters for a region; using the historical load dataset and the set of historical meteorological feature parameters as training data, and training the sub-prediction model according to preset hyperparameters to obtain a prediction model.

[0063] In this embodiment, to obtain the historical load dataset of the power system and the set of historical meteorological feature parameters of the region, the historical load dataset and the set of historical meteorological feature parameters can be used as training data, and the sub-prediction model can be trained according to preset hyperparameters to obtain a prediction model. Among them, the above hyperparameters can be parameters optimized using the Particle Swarm Optimization Algorithm (abbreviated as COA).

[0064] Optionally, inputting the filtered set of historical meteorological feature parameters and the historical load dataset into the sub-prediction model for training may include the following steps: the input representation of the filtered features can be edited and used as the input to the encoder and decoder. At the same time, the hyperparameters of the ETSformer model (i.e., the sub-prediction model before training) are set, and the filtered features are input into the ETSformer model for training. Further, the hyperparameters can be optimized based on the COA algorithm. Among them, during the prediction stage of the ETSformer model, to ensure that each model can adopt the most appropriate hyperparameters, the COA algorithm can be used to optimize the hyperparameters. Through the optimized hyperparameters, the prediction model can successfully complete the prediction task and output the load prediction results of each component.

[0065] In this embodiment, the test set data can also be input into the trained prediction model, and the output of the test data can be calculated. The output result can be objectively evaluated through evaluation metrics, and based on the evaluation metrics, it can be determined whether it is necessary to further train the trained prediction model.

[0066] Optionally, the above evaluation metrics can be error evaluation metrics, which can include Mean Absolute Error (MAE for short), Root Mean Square Error (RMSE for short), and Mean Absolute Percentage Error (MAPE for short), etc. It should be noted that this is only an example here, and the type of evaluation metrics is not specifically limited.

[0067] As an alternative implementation, using the historical load data set and the historical meteorological feature parameter set as training data includes: performing interpolation processing on the historical load data set and the historical meteorological feature parameter set respectively; using the interpolated historical load data set and the interpolated historical meteorological feature parameter set as training data.

[0068] In this embodiment, before training the sub-prediction model using the historical load data set and the historical meteorological feature parameter set, preprocessing will be performed on the historical load data set and the historical meteorological feature parameter set. This preprocessing can include interpolation processing. It should be noted that this is only an example here, and the preprocessing method is not specifically limited.

[0069] Optionally, when obtaining the historical load data set and the historical meteorological feature parameter set, considering that abnormal or missing situations may occur during the collection of power load and weather data, in order to reduce the potential impact of these data problems on prediction accuracy, preprocessing can be performed on the historical load data set and the historical meteorological feature parameter set. During the preprocessing process, the linear interpolation method can be selected, and this method can be executed according to the following formula:

[0070] y = y0 + [(x - x0)y1 - (x - x0)y0] / (x1 - x0)

[0071] Among them, the above y can be used to represent the missing value, and x can be used to represent the abscissa of the missing value. x0, x1, y0, and y1 can be used to represent the data in the historical load data set or the data in the historical meteorological feature parameter set.

[0072] Optionally, after interpolating the historical load data set and the historical meteorological feature parameter set, the interpolated data set can be divided to obtain a divided data set, and the divided data set can be normalized. For example, in order to build a model, the data set can be divided into a training set and a test set according to a ratio of 7:3 respectively. In order to eliminate the differences between different dimensions, accelerate the convergence speed of the model, and improve the training accuracy, the data set can be normalized. The normalization process can follow the following formula:

[0073]

[0074] In the formula, Xmax can be used to represent the maximum value of the sample data, and Xmin can be used to represent the minimum value of the sample data.

[0075] During the data collection process, due to the existence of phenomena such as missed collection, miscollection, and false alarms, these problems may have an adverse impact on the training of the load prediction model. Therefore, preprocessing the original data is a necessary step, and the preprocessing can include data cleaning, missing value processing, and data normalization transformation. Further, according to the input and output characteristics of the ETSformer model, a suitable time series data set can be constructed. The data after completing the above preprocessing steps will be used as the training data and test data of the power prediction model.

[0076] In the embodiment of the present invention, considering that the load demand data corresponding to different day types are different, based on the load demand data corresponding to the day type, the meteorological feature parameter set, and the historical load data set, the power system is predicted to obtain the predicted load data in the sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0077] Embodiment 2

[0078] The technical solution of the embodiment of the present invention will be illustrated below in conjunction with the preferred implementation manners.

[0079] At present, power system load forecasting is an important part of power generation planning and also the basis for the economic operation of the power system. Its accuracy has a crucial impact on the operation, maintenance, and planning of the entire power system. With the access of diverse loads such as large-scale distributed power sources and electric vehicle charging piles, the volatility, time-variation, and randomness of the load have increased significantly, which undoubtedly increases the difficulty of load forecasting. Short-term load forecasting (STLF) plays a theoretical basis role in achieving power supply-demand balance, ensuring power supply security, and optimizing power dispatching. Accurate load forecasting can not only ensure the balance of power supply and demand, effectively avoid problems such as power supply shortages and peak shaving power curtailment, but also ensure the safe and economic operation of the regional power grid, thereby improving the utilization efficiency of new energy in the power system. However, traditional load forecasting methods, such as regression analysis method, time series method, exponential smoothing method, Kalman filtering method, etc., are difficult to adapt to the non-linear characteristics of power load series due to their low forecasting accuracy.

[0080] To overcome the above problems, the load data is usually predicted through the following several methods: First, a short-term load forecasting method based on a hybrid model of convolutional neural network and long short-term memory network; Second, a forecasting model of long short-term memory network and xgboost forecasting model is established, and the error reciprocal method is used to combine the two for forecasting; Third, a gated recurrent unit network is introduced to process historical load sequences with temporal characteristics, model and learn the internal dynamic change rules of load data, and its output result is then fused with other external influencing factors (such as weather, day type, etc.) as new input features, and a deep neural network is used for processing to finally complete load forecasting; Fourth, a convolutional neural network based on the attention mechanism is used to complete the short-term power load forecasting; Fifth, based on the probabilistic power load forecasting framework, combined with clustering-based quantile-long short-term memory network learning to improve the accuracy and robustness of short-term power forecasting; Sixth, an attention-based encoder-decoder model, namely empirical mode decomposition-attention long short-term memory.

[0081] However, the above methods are difficult to adapt to the non-linear characteristics of power load series due to their low forecasting accuracy. And the above methods mainly focus on single-step load forecasting, while in practical applications, the value of multi-step load forecasting is particularly prominent, especially in aspects such as power market bidding and spot electricity price calculation. In the field of time series forecasting, the most significant problem lies in the high computational complexity and large memory consumption required for processing long sequences, and there are technical problems with low accuracy in load data forecasting.

[0082] To solve the above problems, in this embodiment, a short-term load forecasting method for hyperparameter optimization of the ETSformer model using the COA algorithm is proposed. First, the xgboost algorithm is used to calculate the correlation between features, quantitatively analyze the specific impact of meteorological data on the load sequence, so as to screen out features with higher correlation and filter out features with weak correlation with the load, in order to improve the prediction accuracy. Secondly, the screened feature data is input into the ETSformer model for training and prediction, and by introducing the Exponential Smoothing Attention (ESA) mechanism and the Frequency Attention (FA) mechanism, efficient and accurate short-term load forecasting is achieved. Finally, the COA algorithm is used to optimize the hyperparameters of the ETSformer model to further improve the prediction accuracy of the model.

[0083] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; the magnitudes of the numbers in this embodiment are only for illustrative purposes and are not specifically limited here. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention, which is not specifically limited here.

[0084] Figure 2 is a flowchart of a short-term load forecasting method for hyperparameter optimization of the ETSformer model using the COA algorithm according to an embodiment of the present invention, as Figure 2 shown, the method may include the following steps:

[0085] Step S202, load the historical meteorological feature parameter set and the historical load data set.

[0086] In this embodiment, the historical load data set of the power system and the historical meteorological feature parameter set of the area where the power system is located can be obtained.

[0087] Step S204, data preprocessing.

[0088] In this embodiment, considering that abnormal or missing situations may occur during the collection of power load and weather data, in order to reduce the potential impact of these data problems on the prediction accuracy, the original historical meteorological feature parameter set and historical load data set can be preprocessed first. The linear interpolation method can be used to preprocess the above data sets.

[0089] Step S206, data normalization processing.

[0090] In this embodiment, in order to eliminate the differences between different dimensions, accelerate the convergence speed of the model and improve the training accuracy, the data set can be normalized.

[0091] Step S208, dataset division.

[0092] In this embodiment, the normalized dataset can be divided into a training set and a test set according to a ratio of 7:3.

[0093] Step S210, feature selection.

[0094] In this embodiment, the historical meteorological feature parameter set and the historical load dataset after the above preprocessing can be input into the XGBoost feature analysis module. By deeply analyzing the correlation between different feature labels, key features can be extracted, which can be used to construct a new feature input set. And based on the key features, the target meteorological feature type corresponding to the area where the power system is located can be determined. Through the above steps, the interference caused by features with low correlation to the prediction process can be effectively avoided, thus ensuring the prediction accuracy of the model.

[0095] Step S212, model training.

[0096] In this embodiment, the filtered historical meteorological feature parameter set and the historical load dataset can be input into the ETSformer model for training to obtain a prediction model.

[0097] Optionally, the input representation of the filtered features can be edited and used as the input of the encoder and decoder. At the same time, the hyperparameters of the ETSformer model can be set, and the filtered features can be input into the ETSformer model for training.

[0098] Step S214, hyperparameter optimization of the model.

[0099] In this embodiment, based on the COA algorithm, the hyperparameters of the ETSformer model can be optimized.

[0100] Optionally, in the prediction stage of the ETSformer model, to ensure that each model can adopt the most suitable hyperparameters, the COA algorithm can be used to optimize the hyperparameters. Through the optimized hyperparameters, each model successfully completed the prediction task and output the load prediction results of each component.

[0101] Step S216, prediction evaluation.

[0102] In this embodiment, the test set data can be input into the trained model, the output of the test data can be calculated, and the prediction results can be objectively evaluated through evaluation metrics.

[0103] In this embodiment, historical load data, meteorological characteristic parameters, and calendar rule data are selected as the input features of the model. Among them, the calendar rule can be used to determine the day type, which can refer to differentiating different day types during prediction. For example, weekdays, rest days, and holidays belong to different day types, etc.

[0104] Optionally, the above historical load data can cover the load sequence in the 1 hour before the prediction time point and can be used to capture the short-term change trend of the load of the power system; the meteorological characteristic parameters can include key meteorological indicators such as temperature, humidity, irradiance, and rainfall, which can be used to evaluate the potential impact of weather factors on load changes. At the same time, in view of the periodic characteristics shown by the load data within a day, within a week, and between holidays and non-holidays, this embodiment can also perform one-hot encoding processing on calendar rules such as hours, weeks, and holidays. Table 1 is the characteristic information table corresponding to the calendar rules. As shown in Table 1: By processing the following features, the model can better understand the change law of the power load and make more accurate predictions. For example, since the temperature feature has a high correlation with the power load, it is given a higher weight, while the one-hot encoded calendar feature helps the model identify the impact of different days within a week and different hours within a day on the load.

[0105] Table 1 Characteristic information table corresponding to calendar rules

[0106]

[0107] Furthermore, this embodiment also uses the XGBoost algorithm to deeply analyze the correlation between different categories of features and represents it through a correlation heat map. Figure 3 It is a schematic diagram of the correlation between different categories of features according to an embodiment of the present invention. As Figure 3 shown, both the weather information and the historical load data show a relatively large correlation. For example, among the weather information data, the temperature feature stands out with its significant contribution degree and becomes the primary factor affecting load prediction; following closely are rainfall and irradiance. Since rainfall and irradiance are highly correlated with people's daily activities, the importance of these two indicators cannot be ignored. In contrast, although the humidity feature contributes to the prediction result, its influence degree is relatively small.

[0108] In addition, the calendar feature also has a significant impact on the prediction model. Among them, holidays and weekends, as special time nodes, have a particularly prominent contribution degree and significantly affect the result of load prediction. On the contrary, looking at other calendar functions, such as ordinary weekdays, etc., their contribution to the prediction is relatively weak and can almost be ignored.

[0109] Optionally, based on the in-depth analysis of the feature contribution degree mentioned above, a highly targeted feature selection strategy is adopted when constructing the prediction model. Key features (i.e., at least one target meteorological feature) that have a significant impact on the prediction result can be screened out as inputs, so as to optimize the operation efficiency of the model, reduce the computational burden of the model, and simultaneously improve the accuracy and reliability of the prediction.

[0110] In this embodiment, for the hyperparameter optimization of the short-term load prediction model, Figure 4 is a schematic diagram of an optimized iteration curve according to an embodiment of the present invention. As Figure 4 shown, the COA algorithm can be used to optimize key parameters such as the number of encoder layers, the number of decoder stacks, the model dimension, the feed-forward layer dimension, the number of heads in the multi-head exponential smoothing attention mechanism, the kernel size of the input embedding, the number of frequencies (K), the backtracking window size, and the learning rate. The parameter settings that perform best on the validation set can be selected. It can be clearly observed from the optimization curve graph that the COA algorithm shows a fast convergence speed and successfully finds the optimal combination of model hyperparameters when approaching 70 iterations.

[0111] Optionally, Table 2 is the parameter setting table of the prediction model, as shown in Table 2.

[0112] Table 2 Parameter Setting Table of the Prediction Model

[0113]

[0114]

[0115] Optionally, to further verify and evaluate the effectiveness of the prediction model proposed in this paper, multiple typical days are selected and compared with four different baseline models. These four baseline models are Baseline Model 1 (ARIMA), Baseline Model 2 (BiLSTM), Baseline Model 3 (Transformer), and Baseline Model 4 (Informer). A time segment with a length of 96 steps (equivalent to 24 hours) is selected for the comparison test. Based on the data set corresponding to this time segment, the model proposed in this paper and these four baseline models are respectively used to perform rolling predictions on it, and the predicted power data is recorded. The predicted data of each model at each step for a sampling point is selected, so as to obtain the predicted power curve graph.

[0116] According to the test results, first, in terms of the volatility of the power prediction value, the load change between adjacent two sampling points of the model proposed in this embodiment is significantly smaller than that of the other four baseline models, showing better smoothness. Second, when observing the prediction and actual power curve graphs during the load ramp-up stage from the 15th step to the 45th step of the sampling points, the deviation of the load ramp-up prediction value of the model proposed in this embodiment is smaller. When the actual load reaches the output peak, the model can also correct it in time with a smaller deviation, which indicates that the model trained by the above method in this embodiment has higher prediction accuracy.

[0117] Optionally, to more quantitatively demonstrate this advantage, Table 3 is a comparison table of prediction accuracies, which details the comparison results of the prediction model trained by the above method and other baseline prediction models in multiple error evaluation metrics. It can be clearly seen from the data in the table that the prediction model in this embodiment performs optimally in all evaluation metrics, with the highest prediction accuracy and the lowest error, significantly superior to other comparison models. Especially compared with the ARIMA model, the key indicators such as MAE, RMSE, and MAPE are reduced by 44.6, 58.85, and 10.91% respectively. This data difference intuitively reflects the excellent performance of the model in this paper in reducing prediction errors and improving prediction accuracy.

[0118] Table 3 Comparison Table of Prediction Accuracies

[0119]

[0120]

[0121] In this embodiment, by using the XGBoost algorithm to calculate feature correlations, the influence of meteorological data on the load sequence is quantitatively analyzed, so as to obtain features with higher correlations, and key features that have a significant impact on the prediction results are selected as inputs. This step optimizes the operation efficiency of the model, reduces the computational burden of the model, and at the same time improves the accuracy and reliability of the prediction. The screened features are input into the ETSformer for training and prediction, and by introducing the exponential smoothing attention (ESA) mechanism and the frequency attention (FA) mechanism, efficient and accurate short-term load prediction is achieved. Using the COA algorithm to optimize the hyperparameters of the ETFformer model can further improve the prediction accuracy of the model.

[0122] In the embodiment of the present invention, considering that the load demand data corresponding to different day types are different, based on the load demand data corresponding to the day type, the meteorological feature parameter set, and the historical load data set, the power system is predicted to obtain the predicted load data in the sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0123] Embodiment 3

[0124] According to an embodiment of the present invention, there is also provided a prediction device for load data. It should be noted that the prediction device for load data in this embodiment can be used to execute the prediction method for load data in Embodiment 1 of the present invention.

[0125] Figure 5 is a schematic diagram of a prediction device for load data according to an embodiment of the present invention. As Figure 5 shown, the prediction device 50 for load data may include: a determination unit 502, an acquisition unit 504, and a prediction unit 506.

[0126] The determination unit 502 is configured to determine a period to be predicted, where the period to be predicted includes at least one sub-period.

[0127] The acquisition unit 504 is configured to acquire, during the period to be predicted, a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-period, and a set of historical load data corresponding to the historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data.

[0128] The prediction unit 506 is configured to predict the predicted load data of the power system during the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the set of historical load data.

[0129] The prediction device for load data in this embodiment determines, through the determination unit, the period to be predicted, where the period to be predicted includes at least one sub-period; acquires, through the acquisition unit, during the period to be predicted, a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-period, and a set of historical load data corresponding to the historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; and predicts, through the prediction unit, the predicted load data of the power system during the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the set of historical load data, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0130] Embodiment 4

[0131] An embodiment of the present invention may provide a computer terminal, and this computer terminal may be any one of the computer terminal devices in a computer terminal group. Optionally, in this embodiment, the above computer terminal may also be replaced with a terminal device such as a mobile terminal.

[0132] Optionally, in this embodiment, the above computer terminal may be at least one network device among multiple network devices of a computer network.

[0133] In this embodiment, the above computer terminal may execute the program code of the following steps in the load data prediction method: determining a period to be predicted, where the period to be predicted includes at least one sub-period; obtaining, during the period to be predicted, a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-period, and the historical load data set corresponding to the historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0134] Optionally, Figure 6 is a structural block diagram of a computer terminal according to an embodiment of the present invention, as Figure 6 shown, the computer terminal 608 may include: one or more (only one is shown in the figure) processors 602, a memory 604, and a transmission device 606.

[0135] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the load data prediction method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, implements the above load data prediction method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory may further include a memory remotely set relative to the processor, and these remote memories can be connected to the computer terminal 608 through a network. Examples of the above network include but are not limited to the Internet, enterprise intranet, local area network, mobile communication network, and their combinations.

[0136] The processor can call the information and application programs stored in the memory through the transmission device to execute the following steps: determining a period to be predicted, where the period to be predicted includes at least one sub-period; obtaining, during the period to be predicted, a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-period, and the historical load data set corresponding to the historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0137] Those of ordinary skill in the art can understand, Figure 6The structure shown is only illustrative. The computer terminal 608 can also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a personal digital assistant, and terminal devices such as Mobile Internet Devices (abbreviated as MID), PAD, etc. Figure 6 It does not limit the structure of the above computer terminal 608. For example, the computer terminal 608 may further include more or fewer components (such as a network interface, a display device, etc.) than those shown Figure 6 in the figure, or have a different configuration from that shown Figure 6 in the figure.

[0138] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a read-only memory (abbreviated as ROM), a random access memory (abbreviated as RAM), a magnetic disk or an optical disc, etc.

[0139] Embodiment 5

[0140] According to an embodiment of the present invention, there is also provided a computer-readable storage medium, which includes a stored program, wherein the program executes the load data prediction method in Embodiment 1.

[0141] Optionally, in this embodiment, the above computer-readable storage medium may be located in any one of the computer terminals in a computer terminal group in a computer network, or in any one of the mobile terminals in a mobile terminal group.

[0142] Optionally, in this embodiment, the computer-readable storage medium is set to store program codes for performing the following steps: determining a period to be predicted, where the period to be predicted includes at least one sub-period; obtaining, in the period to be predicted, a set of meteorological characteristic parameters of the area where the power system is located, the day type corresponding to the sub-period, and a historical load data set corresponding to a historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0143] Optionally, the above computer-readable storage medium may also execute program code for the following steps: filtering at least one target meteorological feature from a set of meteorological feature parameters, where the influence of the target meteorological feature on the predicted load data is greater than the influence of other meteorological features in the set of meteorological feature parameters on the predicted load data; predicting the predicted load data based on the load demand data, at least one target meteorological feature, and a historical load data set.

[0144] Optionally, the above computer-readable storage medium may also execute program code for the following steps: determining a target meteorological feature type corresponding to a region; determining, from the set of meteorological feature parameters, the meteorological features whose meteorological feature type is the target meteorological feature type as the target meteorological features.

[0145] Optionally, the above computer-readable storage medium may also execute program code for the following steps: inputting the load demand data, at least one target meteorological feature, and a historical load data set into a prediction model to obtain the predicted load data, where the prediction model is trained based on the historical load data set.

[0146] Optionally, the above computer-readable storage medium may also execute program code for the following steps: obtaining a historical load data set and a historical meteorological feature parameter set of a region; using the historical load data set and the historical meteorological feature parameter set as training data, and training a sub-prediction model according to preset hyperparameters to obtain a prediction model.

[0147] Optionally, the above computer-readable storage medium may also execute program code for the following steps: performing interpolation processing on the historical load data set and the historical meteorological feature parameter set respectively; using the interpolated historical load data set and the interpolated historical meteorological feature parameter set as training data.

[0148] In this embodiment, considering that the load demand data corresponding to different day types are different, predicting the power system based on the load demand data corresponding to the day type, the set of meteorological feature parameters, and the historical load data set to obtain the predicted load data in a sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0149] Embodiment 6

[0150] According to an embodiment of the present invention, there is also provided a processor for running a program, where when the program is run by the processor, the prediction method of load data in Embodiment 1 is executed.

[0151] Optionally, in this embodiment, the above computer terminal may be located in at least one network device among multiple network devices of a computer network.

[0152] In this embodiment, the above computer terminal may execute the program code of the following steps in the multilingual translation method: determining a to-be-predicted period, where the to-be-predicted period includes at least one sub-period; obtaining, in the to-be-predicted period, a set of meteorological characteristic parameters of the region where the power system is located, the day type corresponding to the sub-period, and a historical load data set corresponding to a historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0153] The memory may be used to store software programs and modules, such as program instructions / modules corresponding to the multilingual translation method and device in the embodiments of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, implements the above multilingual translation method. The memory may include a high-speed random access memory, and may further include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory may further include a memory remotely provided with respect to the processor, and these remote memories may be connected to the computer terminal through a network. Examples of the above network include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0154] The processor may call the information and application programs stored in the memory through a transmission device to execute the following steps: determining a to-be-predicted period, where the to-be-predicted period includes at least one sub-period; obtaining, in the to-be-predicted period, a set of meteorological characteristic parameters of the region where the power system is located, the day type corresponding to the sub-period, and a historical load data set corresponding to a historical period, where the set of meteorological characteristic parameters includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; predicting the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the set of meteorological characteristic parameters, and the historical load data set.

[0155] Optionally, the above processor may further execute the program code of the following steps: screening out at least one target meteorological characteristic from the set of meteorological characteristic parameters, where the influence of the target meteorological characteristic on the predicted load data is greater than the influence of other meteorological characteristics in the set of meteorological characteristic parameters on the predicted load data; predicting the predicted load data based on the load demand data, at least one target meteorological characteristic, and the historical load data set.

[0156] Optionally, the above processor may further execute the program code of the following steps: determining a target meteorological characteristic type corresponding to the region; determining, from the set of meteorological characteristic parameters, the meteorological characteristics whose meteorological characteristic type is the target meteorological characteristic type as the target meteorological characteristics.

[0157] Optionally, the above-mentioned processor may also execute the program code of the following steps: input the load demand data, at least one target meteorological feature, and the historical load data set into a prediction model to obtain predicted load data, where the prediction model is trained based on the historical load data set.

[0158] Optionally, the above-mentioned processor may also execute the program code of the following steps: obtain the historical load data set and the historical meteorological feature parameter set of the region; use the historical load data set and the historical meteorological feature parameter set as training data, and train the sub-prediction model according to the preset hyperparameters to obtain the prediction model.

[0159] Optionally, the above-mentioned processor may also execute the program code of the following steps: perform interpolation processing on the historical load data set and the historical meteorological feature parameter set respectively; use the interpolated historical load data set and the interpolated historical meteorological feature parameter set as training data.

[0160] By adopting the embodiment of the present invention, considering that the load demand data corresponding to different day types are different, based on the load demand data corresponding to the day type, the meteorological feature parameter set, and the historical load data set, the power system is predicted to obtain the predicted load data in the sub-period, thereby achieving the technical effect of improving the accuracy of load data prediction and solving the technical problem of low accuracy of load data prediction.

[0161] Embodiment 7

[0162] According to the embodiment of the present invention, there is also provided a computer program product, which includes computer instructions, where when the computer instructions are executed by a processor, the prediction method of the load data in Embodiment 1 is implemented.

[0163] Embodiment 9

[0164] An embodiment of the present application may provide an electronic device, which may include a memory and a processor.

[0165] Figure 7 It is a block diagram of an electronic device for a prediction method of load data according to an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described herein and / or claimed.

[0166] As Figure 7 shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory 702 or a computer program loaded from a storage unit 708 into a random access memory 703. In the RAM 703, various programs and data required for the operation of device 700 can also be stored. The computing unit 701, ROM 702, and RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0167] Multiple components in device 700 are connected to the I / O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disc, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows device 700 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0168] The computing unit 701 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 701 executes the various methods and processes described above, such as the data verification method. For example, in some embodiments, the data verification method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed onto device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the data verification method described above can be executed. Alternatively, in other embodiments, the computing unit 701 can be configured to execute the data verification method in any other appropriate manner (e.g., by means of firmware).

[0169] According to an embodiment of the present application, a method for predicting load data is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0170] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), system on a chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a dedicated or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0171] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to the processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program codes can be executed entirely on the machine, partially on the machine, executed partially on the machine as an independent software package and partially on a remote machine, or executed entirely on a remote machine or server.

[0172] In the context of this application, a computer-readable storage medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of the computer-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0173] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or an LCD (liquid crystal display, monitor)); and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0174] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0175] A computer system may include a client and a server. The client and the server are generally far from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0176] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0177] In the above embodiments of the present invention, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0178] In several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are merely illustrative. For example, the division of units can be a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of units or modules can be in electrical or other forms.

[0179] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0180] In addition, the functional units in various embodiments of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0181] When the integrated unit is implemented in the form of a software functional unit 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 causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, read-only memories, random access memories, mobile hard disks, magnetic disks, or optical discs.

[0182] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for predicting load data, characterized in that: include: Determining a time period to be predicted, wherein the time period to be predicted includes at least one sub-time period; Acquire a meteorological characteristic parameter set of the area where the power system is located in the forecast period, a day type corresponding to the sub-period, and a historical load data set corresponding to the historical period, wherein the meteorological characteristic parameter set includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; Based on the load demand data corresponding to the day type, the meteorological characteristic parameter set and the historical load data set, the predicted load data of the power system in the sub-period is predicted.

2. The method according to claim 1, characterized in that: The predicting and obtaining the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the meteorological characteristic parameter set and the historical load data set includes: Selecting at least one target meteorological feature from the meteorological feature parameter set, wherein the impact of the target meteorological feature on the predicted load data is greater than the impact of other meteorological features in the meteorological feature parameter set on the predicted load data; The predicted load data is predicted based on the load demand data, at least one of the target meteorological characteristics, and the historical load data set.

3. The method according to claim 2, characterized in that The step of selecting at least one target meteorological feature from the meteorological feature parameter set includes: Determining the target meteorological feature type corresponding to the area; From the meteorological feature parameter set, the meteorological feature whose meteorological feature type is the target meteorological feature type is determined as the target meteorological feature.

4. The method according to claim 2, characterized in that: The predicting and obtaining the predicted load data based on the load demand data, at least one of the target meteorological characteristics, and the historical load data set includes: The load demand data, at least one of the target meteorological characteristics, and the historical load data set are input into a prediction model to obtain the predicted load data, wherein the prediction model is trained based on the historical load data set.

5. The method according to claim 4, characterized in that The method further comprises: Acquire the historical load data set and the historical meteorological characteristic parameter set of the region; The historical load data set and the historical meteorological characteristic parameter set are used as training data, and the sub-prediction model is trained according to pre-set hyperparameters to obtain the prediction model.

6. The method according to claim 5, characterized in that The using the historical load data set and the historical meteorological characteristic parameter set as training data includes: performing interpolation processing on the historical load data set and the historical meteorological characteristic parameter set respectively; The historical load data set after interpolation processing and the historical meteorological characteristic parameter set after interpolation processing are used as the training data.

7. A load data prediction device, characterized in that: include: A determination unit, configured to determine a time period to be predicted, wherein the time period to be predicted includes at least one sub-time period; an acquisition unit, configured to acquire, in the period to be predicted, a meteorological characteristic parameter set of the area where the power system is located, a day type corresponding to the sub-period, and a historical load data set corresponding to the historical period, wherein the meteorological characteristic parameter set includes at least one meteorological characteristic corresponding to at least one sub-period, and different day types correspond to different load demand data; A prediction unit is used to predict the predicted load data of the power system in the sub-period based on the load demand data corresponding to the day type, the meteorological characteristic parameter set and the historical load data set.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 6 when being run by the processor.

10. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the method according to any one of claims 1 to 6.

Images