Power system power load prediction method, system, equipment, medium and product
The key influence features are screened through the support vector machine recursive feature elimination method, and a variety of neural network models are used to predict power loads. Finally, the hierarchical analysis method comprehensive prediction results are comprehensively used to solve the problem of large power load prediction errors in the existing technology, and the accuracy and efficiency of prediction are improved.
Patent Information
- Application Number
- CN202411816680.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-11
- Publication Date
- 2025-04-08
AI Technical Summary
The existing power load prediction methods for power systems are difficult to accurately consider a variety of influencing factors, resulting in large prediction errors, affecting the safe, stable and efficient operation of the power system.
The recursive feature elimination method of the support vector machine is used to select the feature set of influence features and filter out the key influence features; then the convolutional neural network, recurrent neural network and support vector machine are used to construct power load prediction models respectively; finally, the prediction results of multiple models are comprehensively considered through the hierarchical analysis method to improve the accuracy of power load prediction.
By comprehensively considering various characteristics such as calendar, meteorology, economic and social activities and generator sets, the accuracy and efficiency of power load prediction are improved, helping the power system optimize resource allocation, reduce operating costs, and ensure a balance between power supply and demand.
Smart Images
Figure CN119787312A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of power load forecasting, and in particular to a method, system, equipment, medium and product for forecasting power load in a power system. Background Art
[0002] With the acceleration of global economic growth and energy transformation, electricity demand continues to grow and volatility increases, especially with the widespread access to renewable energy, making power load forecasting the key to ensuring the safe, stable and efficient operation of the power system. Through accurate prediction of future power load, the power system can reasonably arrange power generation plans, timely discover potential load fluctuation risks, prepare countermeasures in advance, avoid power shortages or surpluses, and ensure the balance between power supply and demand. With the advancement of science and technology, such as the application of big data, cloud computing and artificial intelligence, power load forecasting has provided unprecedented accuracy and efficiency, making the forecast results an important basis for power system scheduling, planning, market transactions and energy conservation and emission reduction. Through accurate load forecasting, the power system can optimize resource allocation, improve operating efficiency, reduce operating costs, promote the healthy development of the power market, promote the realization of energy conservation and emission reduction and sustainable development goals, and ultimately improve user satisfaction and service quality, laying a solid foundation for building a more intelligent, green and efficient power system. Therefore, it is of great significance to study the power load forecasting method of the power system.
[0003] In the actual operation of the power system, the calendar, weather, economic and social activities, and the characteristics of the generator sets will have an impact on the power load of the power system: (1) The impact of the calendar on the power load. For example, during holidays, the power load will drop significantly due to the reduction of industrial production and commercial activities. (2) The impact of weather on the power load. For example, the drop in winter temperature will lead to an increase in heating load. The rise in summer temperature will greatly increase the use of refrigeration equipment such as air conditioners and refrigerators, which will lead to an increase in cooling load. Extreme weather events such as cold waves, blizzards, and typhoons will have a serious impact on the power supply, causing damage to the power grid and equipment failure. (3) The impact of economic and social activities on the power load. For example, the rapid development of high-energy-consuming industries will significantly increase the power load. With the rapid development of the economy and the improvement of people's living standards, the demand for electricity is growing. (4) The impact of the characteristics of the generator set on the power load. For example, the power generation capacity of the generator set directly determines the power supply capacity of the power system. When the unit fails or is under maintenance, the power supply capacity will be affected, which will lead to power load fluctuations.
[0004] At present, the power system load forecasting methods mainly include the following: (1) Time series analysis method, which uses time series models (such as ARIMA model (Autoregressive Integrated Moving Average Model)) to make forecasts based on historical load data. The model is simple and mainly suitable for short-term forecasts, but it is difficult to consider the changes in long-term influencing factors and does not fully consider external factors. (2) Regression analysis method, which can comprehensively consider the impact of various factors on load, including meteorology, economy, etc., but the model construction is relatively complex and requires professional statistics and modeling knowledge. (3) Grey system theory method, which is suitable for situations with fewer data samples and has low requirements on data distribution and regularity, but the forecast error for power load with fluctuating changes may be large. (4) Artificial neural network method, which can imitate the human brain for intelligent processing, has adaptive functions for a large number of non-structural and non-deterministic laws, and can achieve high prediction accuracy, but the model construction and training process are complex and require a long computing time. Summary of the invention
[0005] The purpose of this application is to provide a method, system, device, medium and product for predicting power load in an electric power system to improve the accuracy of power load prediction.
[0006] To achieve the above objectives, this application provides the following solutions:
[0007] In a first aspect, the present application provides a method for predicting power load in a power system, comprising:
[0008] Acquire key influencing features of the power system to be predicted at the prediction time; the key influencing features are obtained by selecting the influencing feature set by using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features, and generator set features;
[0009] According to the key influencing features, a first power load forecasting model is used to determine the first power load of the power system to be predicted at the prediction time; wherein the first power load forecasting model is obtained by training a convolutional neural network using a training data set; the training data set includes the key influencing features of the training power system at historical moments and the corresponding power load labels;
[0010] According to the key influencing features, a second power load forecasting model is used to determine a second power load of the power system to be forecasted at the forecasting time; wherein the second power load forecasting model is obtained by training a recurrent neural network using a training data set;
[0011] According to the key influencing features, a third power load forecasting model is used to determine a third power load of the power system to be predicted at the prediction time; wherein the third power load forecasting model is obtained by training a support vector machine using a training data set;
[0012] The power load of the power system to be predicted at the prediction time is determined by using the hierarchical analysis method according to the first power load, the second power load and the third power load.
[0013] Optionally, a support vector machine recursive feature elimination method is used to perform feature selection on the influencing feature set to obtain key influencing features, specifically including:
[0014] Acquire the calendar characteristics, the meteorological characteristics, the economic and social activity characteristics and the generator set characteristics; the calendar characteristics include year, month, day of the week and date; the meteorological characteristics include temperature, precipitation, wind speed and sunshine intensity; the economic and social activity characteristics include economic growth rate, population size and the proportion of each of the three major industries; the generator set characteristics include the operating output of the generator set;
[0015] Performing linear normalization processing on the calendar characteristics, the meteorological characteristics, the economic and social activity characteristics, and the generator set characteristics to obtain normalized calendar characteristics, normalized meteorological characteristics, normalized economic and social activity characteristics, and normalized generator set characteristics;
[0016] Determine the feature weight of each feature by using the support vector machine recursive feature elimination method;
[0017] Based on the feature weights, the features are sorted in descending order to obtain a feature sorting set;
[0018] The features in the feature ranking set that are greater than the set feature weight are taken as key influencing features.
[0019] Optionally, determining the power load of the power system to be predicted at the prediction time by using a hierarchy analysis method according to the first power load, the second power load and the third power load specifically includes:
[0020] Determine a first power load weight, a second power load weight, and a third power load weight by using an analytic hierarchy process according to the first power load, the second power load, and the third power load;
[0021] The power load of the power system to be predicted at the prediction time is determined based on the first power load, the first power load weight, the second power load, the second power load weight, the third power load and the third power load weight.
[0022] Optionally, determining the power load of the power system to be predicted at the prediction time according to the first power load, the first power load weight, the second power load, the second power load weight, the third power load and the third power load weight specifically includes:
[0023] Using the formula P pre =λ1P CNN +λ2P RNN +λ3P SVM Determine the power load of the power system to be predicted at the prediction time; where P pre is the power load of the power system to be predicted at the prediction time; CNN is the first power load; λ1 is the first power load weight; P RNN is the second power load; λ2 is the second power load weight; P SVM is the third power load; λ3 is the third power load weight.
[0024] Optionally, the convolutional neural network is trained using the training data set to obtain a first power load prediction model, specifically including:
[0025] Get the training dataset;
[0026] The convolutional neural network is trained with the key influencing features of the training power system at historical moments as input and the corresponding power load labels as output to obtain the first power load forecasting model.
[0027] Optionally, obtain a training data set, specifically including:
[0028] Obtaining the impact feature set and corresponding power load data of the training power system at historical moments;
[0029] Performing linear normalization processing on the impact feature set to obtain a normalized impact feature set of the training power system at a historical moment;
[0030] The support vector machine recursive feature elimination method is used to select the normalized influence feature set of the training power system at historical moments, and the key influence features of the training power system at historical moments are obtained;
[0031] A training dataset is constructed based on the key influencing features of the training power system at historical moments and the corresponding power load data.
[0032] In a second aspect, the present application provides a power system power load forecasting system, comprising:
[0033] A feature acquisition module is used to acquire key influencing features of the power system to be predicted at the prediction time; the key influencing features are obtained by selecting the influencing feature set using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features, and generator set features;
[0034] A first power load determination module is used to determine the first power load of the power system to be predicted at the prediction time according to the key influencing features and using a first power load prediction model; wherein the first power load prediction model is obtained by training a convolutional neural network using a training data set; the training data set includes key influencing features of the training power system at historical moments and corresponding power load labels;
[0035] A second power load determination module is used to determine the second power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using a second power load prediction model; wherein the second power load prediction model is obtained by training a recurrent neural network using a training data set;
[0036] A third power load determination module is used to determine the third power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using a third power load prediction model; wherein the third power load prediction model is obtained by training a support vector machine using a training data set;
[0037] The analysis module is used to determine the power load of the power system to be predicted at the prediction time by using the hierarchical analysis method according to the first power load, the second power load and the third power load.
[0038] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for predicting power load in an electric power system.
[0039] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for predicting power load in a power system.
[0040] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for predicting power load in a power system.
[0041] According to the specific embodiments provided in this application, this application has the following technical effects:
[0042] The present application provides a method, system, device, medium and product for predicting power load of an electric power system, which obtains key influencing features of the power system to be predicted at the prediction time; according to the key influencing features, a first power load prediction model is used to determine the first power load of the power system to be predicted at the prediction time; according to the key influencing features, a second power load prediction model is used to determine the second power load of the power system to be predicted at the prediction time; according to the key influencing features, a third power load prediction model is used to determine the third power load of the power system to be predicted at the prediction time; wherein the first power load prediction model, the second power load prediction model and the third power load prediction model are respectively obtained by training a convolutional neural network, a recurrent neural network and a support vector machine using training data sets; according to the first power load, the second power load and the third power load, a hierarchical analysis method is used to determine the power load of the power system to be predicted at the prediction time. This application takes into account multiple features such as calendar, meteorology, economic and social activities, and generator sets, and uses the support vector machine recursive feature elimination method for feature selection to screen out the most relevant influencing features as subsequent input features of the model. Based on the hierarchical analysis method, it comprehensively considers the results of various artificial neural networks and support vector machine power load forecasting models to improve the accuracy of power load forecasting. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0044] Figure 1 A schematic diagram of a flow chart of a method for predicting power load in a power system provided in one embodiment of the present application;
[0045] Figure 2 A configuration flow chart of the power system power load forecasting method of the present application in practical application;
[0046] Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0048] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0049] The present application proposes a method for predicting power load in a power system in response to the problem of power load prediction under the influence of calendar, meteorological conditions, economic and social activities, and characteristics of generator sets during the operation of the power system.
[0050] In an exemplary embodiment, Figure 1 As shown, a method for predicting power load in a power system is provided, comprising the following steps:
[0051] S1: Obtain key influencing features of the power system to be predicted at the prediction time; the key influencing features are obtained by selecting the influencing feature set using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features, and generator set features.
[0052] As an optional implementation method, a support vector machine recursive feature elimination method is used to perform feature selection on the influencing feature set to obtain key influencing features, specifically including:
[0053] The calendar characteristics, the meteorological characteristics, the economic and social activity characteristics and the generator set characteristics are obtained; the calendar characteristics include year, month, day of the week and date; the meteorological characteristics include temperature, precipitation, wind speed and sunshine intensity; the economic and social activity characteristics include economic growth rate, population size and the proportion of each of the three major industries; the generator set characteristics include the operating output of the generator set.
[0054] The calendar characteristics, the meteorological characteristics, the economic and social activity characteristics and the generator set characteristics are linearly normalized to obtain normalized calendar characteristics, normalized meteorological characteristics, normalized economic and social activity characteristics and normalized generator set characteristics.
[0055] The support vector machine recursive feature elimination method is used to determine the feature weight of each feature.
[0056] Based on the feature weights, the features are sorted in descending order to obtain a feature sorting set.
[0057] The features in the feature ranking set that are greater than the set feature weight are taken as key influencing features.
[0058] S2: According to the key influencing features, use the first power load forecasting model to determine the first power load of the power system to be predicted at the prediction time; wherein, the first power load forecasting model is obtained by training a convolutional neural network using a training data set; the training data set includes the key influencing features of the training power system at historical moments and the corresponding power load labels.
[0059] As an optional implementation, a convolutional neural network is trained using a training data set to obtain a first power load prediction model, specifically including:
[0060] Get the training dataset.
[0061] The convolutional neural network is trained with the key influencing features of the training power system at historical moments as input and the corresponding power load labels as output to obtain the first power load forecasting model.
[0062] As an optional implementation, obtaining a training data set specifically includes:
[0063] Obtain the impact feature set and corresponding power load data of the training power system at historical moments.
[0064] The impact feature set is linearly normalized to obtain a normalized impact feature set of the training power system at a historical moment.
[0065] The support vector machine recursive feature elimination method is used to perform feature selection on the normalized influencing feature set of the training power system at historical moments, and the key influencing features of the training power system at historical moments are obtained.
[0066] A training dataset is constructed based on the key influencing features of the training power system at historical moments and the corresponding power load data.
[0067] S3: According to the key influencing features, the second power load forecasting model is used to determine the second power load of the power system to be predicted at the prediction time; wherein the second power load forecasting model is obtained by training the recurrent neural network using the training data set. Specifically, the key influencing features of the training power system at the historical moment are used as input, and the corresponding power load labels are used as output, and the recurrent neural network is trained to obtain the second power load forecasting model.
[0068] S4: According to the key influencing features, the third power load forecasting model is used to determine the third power load of the power system to be predicted at the prediction time; wherein the third power load forecasting model is obtained by training a support vector machine using a training data set. Specifically, the key influencing features of the training power system at historical moments are used as input, and the corresponding power load labels are used as output, and the support vector machine is trained to obtain the third power load forecasting model.
[0069] S5: Determine the power load of the power system to be predicted at the prediction time by using the analytic hierarchy process according to the first power load, the second power load and the third power load.
[0070] As an optional implementation, S5 specifically includes:
[0071] S51: Determine a first power load weight, a second power load weight, and a third power load weight using an analytic hierarchy process according to the first power load, the second power load, and the third power load.
[0072] S52: Determine the power load of the power system to be predicted at the prediction time based on the first power load, the first power load weight, the second power load, the second power load weight, the third power load and the third power load weight.
[0073] As an optional implementation, S52 specifically includes:
[0074] Using the formula P pre =λ1P CNN +λ2P RNN +λ3P SVM Determine the power load of the power system to be predicted at the prediction time; where P pre is the power load of the power system to be predicted at the prediction time; CNN is the first power load; λ1 is the first power load weight; P RNN is the second power load; λ2 is the second power load weight; P SVM is the third power load; λ3 is the third power load weight.
[0075] In practical applications, in order to consider the impact of more key influencing features on the power load of the power system, and to comprehensively consider the prediction of each model, a more flexible, efficient and accurate prediction and estimation effect can be achieved. Therefore, this application considers multiple influencing features such as calendar, meteorology, economic and social activities and generator sets, and uses the support vector machine recursive feature elimination (Support Vector Machine Recursive Feature Elimination, SVM-RFE) method for feature selection, and screens out key influencing features as input features of the subsequent artificial neural network model. Based on the final selected key influencing features and output values (historical power load data), the power load forecasting models based on convolutional neural networks (Convolutional Neural Networks, CNN), recurrent neural networks (Recurrent Neural Network, RNN) and support vector machines (Support Vector Machine, SVM) are trained respectively. According to the forecast demand, the key influencing features finally selected in the required forecast area are input to obtain the forecast results of each model, and the hierarchical analysis method is used to realize the power load forecasting of the required forecast area based on ensemble learning, and a power system power load forecasting method is proposed to achieve efficient and scientific solution to the power system regional power load forecasting problem.
[0076] like Figure 2 As shown, the specific steps of the power system power load forecasting method of the present application are as follows:
[0077] (1) Obtain the historical power load data of the area to be predicted, which will be used as the output value of the subsequent model training data set.
[0078] (2) Obtain the characteristic values (influencing feature set) of the historical power load data at the corresponding time, where calendar features include year, month, day of the week, and date; meteorological features include temperature, precipitation, wind speed, and sunshine intensity; economic and social activity characteristics include economic growth rate, population size, and the proportion of each of the three major industries; and generator set characteristics include generator set operating output. These features will be used as the proposed input features in the subsequent model training data set.
[0079] (3) Data preprocessing and normalization: For each of the above eigenvalues, a linear normalization method is used to linearly scale the data to the interval [0, 1]:
[0080]
[0081] In the formula, x is the original data in the eigenvalue, x min and x max are the minimum and maximum values in the data set, respectively, and x' is the normalized data.
[0082] (4) Using the support vector machine recursive feature elimination (SVM-RFE) model, the feature weights of each type of eigenvalue are calculated, and the features are sorted from large to small to form a feature sorting set R.
[0083] (5) Features with smaller weights (less than the set feature weights) are deleted from the feature sorting set R, and finally feature values with a predetermined number of features being N are selected.
[0084] (6) The selected input features are used as the input features of the convolutional neural network (CNN), and the power load values corresponding to the training data set are used as the output, so as to train the power load prediction model based on the convolutional neural network (CNN).
[0085] (7) The selected input features are used as input features of the recurrent neural network (RNN), and the power load values corresponding to the training data set are used as output quantities to train the power load prediction model based on the recurrent neural network (RNN).
[0086] (8) The selected input features are used as input features of the support vector machine (SVM), and the power load values corresponding to the training data set are used as output quantities to train the power load prediction model based on the support vector machine (SVM).
[0087] (9) According to the forecast requirements, the input features of the final selected forecast area are input to the convolutional neural network CNN, the recurrent neural network RNN and the support vector machine SVM respectively, and the output power load values are P CNN , P RNN and P SVM .
[0088] (10) The predicted values are obtained by inputting the input features of the power load to be predicted for each training model. The weight distribution calculation of the above machine learning methods is performed based on ensemble learning using the analytic hierarchy process. The weights corresponding to the three machine learning methods are λ1, λ2, and λ3 (satisfying λ1+λ2+λ3=1). The predicted power load power P of the area is pre for:
[0089] P pre =λ1P CNN +λ2P RNN +λ3P SVM .
[0090] The power system power load forecasting method proposed in this application takes into account multiple characteristics such as calendar, meteorology, economic and social activities, and generator sets, and uses the support vector machine recursive feature elimination model for feature selection to screen out the most relevant influencing features as subsequent input features of the model. This method is simple, convenient, and comprehensive in considering features.
[0091] The power system power load forecasting method proposed in this application is based on the hierarchical analysis method and comprehensively considers the results of power load forecasting models of various artificial neural network methods and support vector machines. This method can integrate the prediction effects of various models and consider factors more fully and comprehensively.
[0092] Based on the same inventive concept, the embodiment of the present application also provides a power system power load forecasting system for implementing the above-mentioned power system power load forecasting method. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the power system power load forecasting system embodiment provided below can refer to the limitations of the dual-channel EEG fatigue detection method above, and will not be repeated here.
[0093] In an exemplary embodiment, a power system power load forecasting system is provided, comprising:
[0094] The feature acquisition module is used to obtain the key influencing features of the power system to be predicted at the prediction time; the key influencing features are obtained by selecting the influencing feature set using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features and generator set features.
[0095] The first power load determination module is used to determine the first power load of the power system to be predicted at the prediction time according to the key influencing features and using the first power load prediction model; wherein the first power load prediction model is obtained by training a convolutional neural network using a training data set; the training data set includes the key influencing features of the training power system at historical moments and the corresponding power load labels.
[0096] The second power load determination module is used to determine the second power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using the second power load prediction model; wherein the second power load prediction model is obtained by training a recurrent neural network using a training data set.
[0097] The third power load determination module is used to determine the third power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using the third power load prediction model; wherein the third power load prediction model is obtained by training a support vector machine using a training data set.
[0098] The analysis module is used to determine the power load of the power system to be predicted at the prediction time by using the hierarchical analysis method according to the first power load, the second power load and the third power load.
[0099] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above-mentioned power system power load forecasting method when executing the computer program.
[0100] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the above-mentioned power system power load forecasting method when executed by a processor.
[0101] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the above-mentioned power system power load forecasting method when executed by a processor.
[0102] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting power load in a power system is implemented.
[0103] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0104] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0105] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0106] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0107] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0108] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for predicting power load in a power system, characterized in that: include: Obtain the key influencing features of the power system to be predicted at the prediction time; The key influencing features are obtained by selecting the influencing feature set by using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features and generator set features; According to the key influencing features, a first power load forecasting model is used to determine the first power load of the power system to be predicted at the prediction time; wherein the first power load forecasting model is obtained by training a convolutional neural network using a training data set; the training data set includes the key influencing features of the training power system at historical moments and the corresponding power load labels; According to the key influencing features, a second power load forecasting model is used to determine a second power load of the power system to be forecasted at the forecasting time; wherein the second power load forecasting model is obtained by training a recurrent neural network using a training data set; According to the key influencing features, a third power load forecasting model is used to determine a third power load of the power system to be predicted at the prediction time; wherein the third power load forecasting model is obtained by training a support vector machine using a training data set; The power load of the power system to be predicted at the prediction time is determined by using the hierarchical analysis method according to the first power load, the second power load and the third power load.
2. The method for predicting power load in a power system according to claim 1, characterized in that: The support vector machine recursive feature elimination method is used to select the influencing feature set and obtain the key influencing features, including: Acquire the calendar characteristics, the meteorological characteristics, the economic and social activity characteristics and the generator set characteristics; the calendar characteristics include year, month, day of the week and date; the meteorological characteristics include temperature, precipitation, wind speed and sunshine intensity; the economic and social activity characteristics include economic growth rate, population size and the proportion of each of the three major industries; the generator set characteristics include the operating output of the generator set; Performing linear normalization processing on the calendar characteristics, the meteorological characteristics, the economic and social activity characteristics, and the generator set characteristics to obtain normalized calendar characteristics, normalized meteorological characteristics, normalized economic and social activity characteristics, and normalized generator set characteristics; Determine the feature weight of each feature by using the support vector machine recursive feature elimination method; Based on the feature weights, the features are sorted in descending order to obtain a feature sorting set; The features in the feature ranking set that are greater than the set feature weight are taken as key influencing features.
3. The method for predicting power load in a power system according to claim 1, characterized in that: According to the first power load, the second power load and the third power load, the power load of the power system to be predicted at the prediction time is determined by using the hierarchical analysis method, specifically including: Determine a first power load weight, a second power load weight, and a third power load weight by using an analytic hierarchy process according to the first power load, the second power load, and the third power load; The power load of the power system to be predicted at the prediction time is determined based on the first power load, the first power load weight, the second power load, the second power load weight, the third power load and the third power load weight.
4. The method for predicting power load in a power system according to claim 3, characterized in that: Determining the power load of the power system to be predicted at the prediction time according to the first power load, the first power load weight, the second power load, the second power load weight, the third power load, and the third power load weight specifically includes: Using the formula P pre =λ1P CNN +λ2P RNN +λ3P SVM Determine the power load of the power system to be predicted at the prediction time; where P pre is the power load of the power system to be predicted at the prediction time; CNN is the first power load; λ1 is the first power load weight; P RNN is the second power load; λ2 is the second power load weight; P SVM is the third power load; λ3 is the third power load weight.
5. The method for predicting power load in a power system according to claim 1, characterized in that: The convolutional neural network is trained using the training data set to obtain the first power load forecasting model, which specifically includes: Get the training dataset; The convolutional neural network is trained with the key influencing features of the training power system at historical moments as input and the corresponding power load labels as output to obtain the first power load forecasting model.
6. The method for predicting power load in a power system according to claim 5, characterized in that: Obtain the training data set, including: Obtaining the impact feature set and corresponding power load data of the training power system at historical moments; Performing linear normalization processing on the impact feature set to obtain a normalized impact feature set of the training power system at a historical moment; The support vector machine recursive feature elimination method is used to select the normalized influence feature set of the training power system at historical moments, and the key influence features of the training power system at historical moments are obtained; A training dataset is constructed based on the key influencing features of the training power system at historical moments and the corresponding power load data.
7. A power system power load forecasting system, characterized in that: include: A feature acquisition module is used to acquire key influencing features of the power system to be predicted at the prediction time; The key influencing features are obtained by selecting the influencing feature set by using the support vector machine recursive feature elimination method; the influencing feature set includes calendar features, meteorological features, economic and social activity features and generator set features; A first power load determination module is used to determine the first power load of the power system to be predicted at the prediction time according to the key influencing features and using a first power load prediction model; wherein the first power load prediction model is obtained by training a convolutional neural network using a training data set; the training data set includes key influencing features of the training power system at historical moments and corresponding power load labels; A second power load determination module is used to determine the second power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using a second power load prediction model; wherein the second power load prediction model is obtained by training a recurrent neural network using a training data set; A third power load determination module is used to determine the third power load of the power system to be predicted at the prediction time according to the key influencing characteristics and using a third power load prediction model; wherein the third power load prediction model is obtained by training a support vector machine using a training data set; The analysis module is used to determine the power load of the power system to be predicted at the prediction time by using the hierarchical analysis method according to the first power load, the second power load and the third power load.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the power load forecasting method for a power system according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting electric power load in an electric power system according to any one of claims 1 to 6 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting electric power load in an electric power system according to any one of claims 1 to 6 is implemented.