Power load prediction method and system suitable for multiple nodes

Abstract

The invention relates to a power load prediction method and system suitable for multiple nodes, load prediction under the condition of multiple influence factors is realized by considering various influence factors influencing power load prediction and load prediction models under multiple types of nodes, different load prediction models are adopted at the same time, and according to the method, the load prediction efficiency is greatly improved. The method can predict and combine various types of scenes, so that the method can predict the future load more accurately and reliably, and the method is simple, reliable, scientific and practical.

Description

Technical Field

[0001] The present invention belongs to the field of power load analysis methods, and relates to a power load prediction method and system, in particular to a power load prediction method and system applicable to multiple nodes. Background Art

[0002] Accurate power load forecasting is the key to power system dispatching, real-time control, and operation planning, and is the basis for maintaining the balance of power system supply and demand. With the substantial increase in the total power load and the continuous changes in the power grid structure, system analysis involving power flow analysis and emergency power supply restoration requires more detailed load data support, which requires multi-node load forecasting for power systems with rapidly changing structural scales; if the power load changes at multiple nodes of the power system are not understood, various dynamic and static analyses of the system cannot be performed to determine the quality and stability of energy supply, such as angle stability and voltage stability. In fact, multi-node load forecasting is of great significance to the optimization of power system operation, and it is the basis for refined system control and emergency handling.

[0003] Existing load forecasting methods are more focused on single-node load forecasting, ignoring the temporal and spatial correlation between its upper nodes and it, resulting in low accuracy of multi-node power load forecasting. Currently, most of the multi-node power load forecasting problems are based on a single forecasting task. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and provide a method and system for predicting power loads suitable for multiple nodes. By considering various influencing factors affecting power load prediction and load prediction models under multiple types of nodes, load prediction under multiple influencing factors is realized. At the same time, different load prediction models are used to predict and combine multiple types of nodes, so that the method of the present invention can predict future loads more accurately and reliably, and the method of the present invention is simple, reliable, scientific and practical.

[0005] The present invention solves the technical problem by the following technical solutions:

[0006] A method for predicting power loads for multiple nodes, characterized in that it comprises the following steps:

[0007] S1: Select the area to be tested;

[0008] S2: Obtain the load forecasting influencing factors in the area to be forecasted, sort them, and construct a power load data set;

[0009] S3: Based on the load forecast influencing factors and ranking results obtained in S2, obtain historical data of load forecast influencing factors in the area to be forecasted;

[0010] S4: Based on the historical data of load forecast influencing factors in the area to be forecasted obtained in step S3, a preliminary forecaster for medium- and long-term power load is constructed based on the learning model, and a neural network model for the preliminary forecaster for power load is created;

[0011] S5: using the power load data set in step S2, training and testing the medium- and long-term power load preliminary predictor constructed in step S4, thereby obtaining a power load forecasting model and a power load forecasting error database;

[0012] S6: using an algorithm to perform probability modeling on the power load forecast error database obtained in step S5 to obtain a cumulative probability distribution of the power load forecast error, thereby further obtaining a power load forecast error interval;

[0013] S7: Inputting the power data of the standard time period into the power load forecasting model to obtain the power load forecasting result within the standard time period.

[0014] As a preferred technical solution of the present invention, the load forecast influencing factors for obtaining the forecast area in step S2 include weather, environment, holidays, usage habits and materials used.

[0015] As a preferred technical solution of the present invention, in step S2, the method for sorting the load forecast influencing factors in the prediction area is: according to the influence of the load forecast influencing factors on the historical load values ​​of the prediction area, the size of each load forecast influencing factor is calculated and sorted.

[0016] As a preferred technical solution of the present invention, the least square method is used to calculate the size of each load forecast influencing factor.

[0017] As a preferred technical solution of the present invention, in step S2, the power load data set S is represented by {T1, T2,…, Ti, Tn, Ts}, where n is the number of nodes to be predicted in the power system, Ti represents the load sample data set of node i, and Ts represents the load sample data set of the upper node.

[0018] As a preferred technical solution of the present invention, in S3, it is necessary to pre-process the historical data of load forecasting influencing factors in the area to be forecasted, so as to eliminate erroneous data in the historical data and supplement the missing data in the historical data.

[0019] As a preferred technical solution of the present invention, a grey theory prediction model is used to pre-process the historical data of load prediction influencing factors in the area to be predicted that is required to be obtained in S3.

[0020] As a preferred technical solution of the present invention, the learning model in step S4 includes one or more groups of learning models among XGBoost, ATaBoost, and random forest.

[0021] As a preferred technical solution of the present invention, the algorithm in step S6 adopts an algorithm for establishing a radial basis function neural network model; using a Gaussian function as the basis function of the radial basis function neural network model, or using a k-nearest neighbor clustering algorithm to determine one or more groups in the hidden layer structure of the radial basis function neural network model as a modeling algorithm.

[0022] A power load forecasting system suitable for multiple nodes, characterized by comprising an influencing factor acquisition module, an influencing factor historical data acquisition module, a power load preliminary predictor neural network model creation module, a power load forecasting model and forecasting error database acquisition module, a power load forecasting error interval acquisition module and a power load result forecasting module,

[0023] The influencing factor acquisition module is used to acquire the load prediction influencing factors in the area to be predicted, sort them, and construct a power load data set;

[0024] The influencing factor historical data acquisition module is used to acquire the historical data of the load forecasting influencing factors in the area to be forecasted based on the acquired load forecasting influencing factors and the sorting results;

[0025] The power load preliminary predictor neural network model creation module is used to construct a medium- and long-term power load preliminary predictor based on the learning model according to the historical data of load prediction influencing factors in the area to be predicted, and to create a power load preliminary predictor neural network model;

[0026] The power load forecasting model and forecasting error database acquisition module is used to train and test the constructed medium- and long-term power load preliminary forecaster with the power load data set, thereby obtaining the power load forecasting model and the power load forecasting error database;

[0027] The power load forecast error interval acquisition module is used to perform probability modeling on the obtained power load forecast error database to obtain the cumulative probability distribution of the power load forecast error, thereby further obtaining the power load forecast error interval;

[0028] The power load result prediction module is used to input the power data of the standard time period into the power load prediction model to obtain the power load prediction result within the standard time period.

[0029] The advantages and beneficial effects of the present invention are:

[0030] The present invention is applicable to a method and system for predicting electric loads at multiple nodes. By considering various factors affecting electric load prediction and load prediction models at multiple types of nodes, load prediction under multiple influencing factors is achieved. Different load prediction models are adopted at the same time. According to the present method, multiple types of scenarios can also be predicted and combined. Thus, the present invention can predict future loads more accurately and reliably. Moreover, the present invention is simple, reliable, scientific and practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0032] The present invention is further described in detail below through specific examples. The following examples are only illustrative and not restrictive, and the protection scope of the present invention cannot be limited thereto.

[0033] Example 1

[0034] like Figure 1 As shown, the present invention provides a method for predicting power loads applicable to multiple nodes, the innovation of which is that it comprises the following steps:

[0035] S1: Select the area to be tested;

[0036] S2: Obtain and sort the load forecasting influencing factors in the area to be forecasted, and construct a power load data set;

[0037] Obtain the load forecast influencing factors in the forecast area, specifically using the remaining influencing factors such as weather, environment, holidays, usage habits and used materials as load forecast influencing factors.

[0038] The load forecast influencing factors are ranked, specifically, according to the influence degree of the load forecast influencing factors on the historical load values ​​of the forecast area, the size of each load forecast influencing factor is calculated and ranked.

[0039] The multi-node load sample data set S is represented as {T1, T2, …, Ti, Tn, Ts}, where n is the number of nodes to be predicted in the power system, Ti represents the load sample data set of node i, and Ts represents the load sample data set of the upper node. In S2, the acquired historical data needs to be preprocessed to eliminate erroneous data in the historical data and supplement the missing data in the historical data.

[0040] S3: Based on the load forecast influencing factors and ranking results obtained in S2, historical data of load forecast influencing factors in the area to be forecasted are obtained; the grey theory forecasting model is used to pre-process the historical data, and the least squares method can be used to calculate the size of each load forecast influencing factor.

[0041] S4: Based on the historical data of load forecasting influencing factors in the area to be forecasted obtained in step S3, a preliminary predictor for medium- and long-term power load is constructed based on the learning model, and a neural network model for the preliminary predictor for power load is created; the learning model can adopt one or more groups of learning models selected from XGBoost, ATaBoost, and random forest.

[0042] S5: using the power load data set in step S2, training and testing the medium- and long-term power load preliminary predictor constructed in step S4, thereby obtaining a power load forecasting model and a power load forecasting error database;

[0043] S6: using an algorithm to perform probability modeling on the power load forecast error library obtained in step S5 to obtain a cumulative probability distribution of the power load forecast error, thereby further obtaining a power load forecast error interval;

[0044] The algorithm can be used to establish a radial basis function neural network model; use Gaussian function as the basis function of the radial basis function neural network model, or use the k-nearest neighbor clustering algorithm to determine one or more groups in the hidden layer structure of the radial basis function neural network model as a modeling algorithm.

[0045] S7: Inputting the power data of the standard time period into the power load forecasting model to obtain a target power load forecast value.

[0046] Determine whether the prediction of the LSTM neural network prediction model is accurate based on the root mean square error, mean absolute percentage error and prediction accuracy of the test sample set,

[0047] Wherein, the root mean square error MAPE is:

[0048]

[0049] Wherein, n is the number of test samples in the test sample set, Yact(i) is the actual value of the power load corresponding to the i-th test sample, and Ypred(i) is the predicted value of the power load corresponding to the i-th test sample.

[0050] The mean absolute percentage error RMSE is:

[0051]

[0052] Among them, RMSE is the prediction error, β is the adjustment coefficient of the cost, α is the discount coefficient, Pi is the real-time price at the i-th time point, P is the historical average price, Ti is the predicted value of the power load at the i-th time point, and Ti′ is the real-time value of the power load at the i-th time point;

[0053] The prediction accuracy FA is:

[0054]

[0055] Wherein, n is the number of test samples in the test sample set, Yact(i) is the actual value of the power load corresponding to the i-th test sample, and Ypred(i) is the predicted value of the power load corresponding to the i-th test sample.

[0056] The present invention realizes load prediction under multiple influencing factors by considering various influencing factors affecting power load prediction and load prediction models under multiple types of nodes. At the same time, different load prediction models are adopted. According to the present method, multiple types of scenarios can also be predicted and combined, so that the method of the present invention can predict future loads more accurately and reliably. Moreover, the method of the present invention is simple, reliable, scientific and practical.

[0057] Compared with the existing multi-node power forecasting method, this method has good market prospects. This embodiment compares a large amount of data between the existing method and the method of the present invention: it includes the following steps: S1: select the area to be tested; S2: obtain the power load data set in the area to be predicted; S3: use the power load data set in step S2 to train and test the medium- and long-term power load preliminary predictor constructed in step 2, so as to obtain a power load forecasting model; S4: input the power data of the standard time period into the power load forecasting model to obtain the power load forecasting result within the standard time period.

[0058] When the existing method is used, since it does not take into account the historical data of the load forecasting influencing factors in the forecasting area, the final forecasting result is quite different from the actual result, and the data is not accurate enough. If used, it will have an impact on actual production and life.

[0059] Embodiment 2: The method of the present invention can also calculate the total future charge under different usage scenarios by inputting corresponding environmental data or other historical data according to different usage scenarios. The specific steps are as follows:

[0060] S1: Establish corresponding power load forecasting models for each industry;

[0061] S2: Calculate the power load of each industry in the year to be predicted according to the power load prediction model corresponding to each industry;

[0062] S3: establishing corresponding power load forecasting models for the industries respectively according to the historical data of the necessary factors corresponding to the industries and the historical data of the power load corresponding to the industries;

[0063] S4: Add up the electricity loads of the various industries in the year to be predicted to obtain the total electricity load for the year to be predicted.

[0064] This embodiment 2 meets the use requirements of embodiment 1 in various scenarios. It only needs to consider various influencing factors that affect the power load prediction, as well as the load prediction models under multiple types of scenarios, to achieve load prediction under multiple influencing factors. At the same time, different load prediction models are used to predict and combine multiple types of scenarios and multiple types of nodes, so that the method of the present invention can predict future loads more accurately and reliably.

[0065] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0066] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0067] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting power loads applicable to multiple nodes, characterized in that: The steps include: S1: Select the area to be tested; S2: Obtain the load forecasting influencing factors in the area to be forecasted, sort them, and construct a power load data set; S3: Based on the load forecast influencing factors and ranking results obtained in S2, obtain historical data of load forecast influencing factors in the area to be forecasted; S4: Based on the historical data of load forecast influencing factors in the area to be forecasted obtained in step S3, a preliminary forecaster for medium- and long-term power load is constructed based on the learning model, and a neural network model for the preliminary forecaster for power load is created; S5: using the power load data set in step S2, training and testing the medium- and long-term power load preliminary predictor constructed in step S4, thereby obtaining a power load forecasting model and a power load forecasting error database; S6: using an algorithm to perform probability modeling on the power load forecast error database obtained in step S5 to obtain a cumulative probability distribution of the power load forecast error, thereby further obtaining a power load forecast error interval; S7: Inputting the power data of the standard time period into the power load forecasting model to obtain the power load forecasting result within the standard time period.

2. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: The load forecast influencing factors in the prediction area obtained in step S2 include weather, environment, holidays, usage habits and materials used.

3. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: In step S2, the method for sorting the load forecast influencing factors in the forecast area is: according to the influence degree of the load forecast influencing factors on the historical load values ​​of the forecast area, the size of each load forecast influencing factor is calculated and sorted.

4. The method for predicting power loads applicable to multiple nodes according to claim 3, characterized in that: The least squares method is used to calculate the size of each load forecast influencing factor.

5. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: In step S2, the power load data set S is represented by {T1, T2, ..., Ti, Tn, Ts}, where n is the number of nodes to be predicted in the power system, Ti represents the load sample data set of node i, and Ts represents the load sample data set of the upper node.

6. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: In S3, it is necessary to pre-process the historical data of the load forecast influencing factors in the area to be forecasted, so as to eliminate erroneous data in the historical data and supplement the missing data in the historical data.

7. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: The grey theory prediction model is used to pre-process the historical data of load prediction influencing factors in the area to be predicted that need to be obtained in S3.

8. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: The learning model in step S4 includes one or more groups of learning models among XGBoost, ATaBoost, and random forest.

9. The method for predicting power loads applicable to multiple nodes according to claim 1, characterized in that: The algorithm in step S6 adopts the algorithm of establishing a radial basis function neural network model; using Gaussian function as the basis function of the radial basis function neural network model, or using k-nearest neighbor clustering algorithm to determine one or more groups of hidden layer structures of the radial basis function neural network model as the modeling algorithm.

10. A power load forecasting system applicable to multiple nodes, characterized in that: It includes an influencing factor acquisition module, an influencing factor historical data acquisition module, a power load preliminary predictor neural network model creation module, a power load forecasting model and forecasting error database acquisition module, a power load forecasting error interval acquisition module and a power load result forecasting module. The influencing factor acquisition module is used to acquire the load prediction influencing factors in the area to be predicted, sort them, and construct a power load data set; The influencing factor historical data acquisition module is used to acquire the historical data of the load forecasting influencing factors in the area to be forecasted based on the acquired load forecasting influencing factors and the sorting results; The power load preliminary predictor neural network model creation module is used to construct a medium- and long-term power load preliminary predictor based on the learning model according to the historical data of load prediction influencing factors in the area to be predicted, and to create a power load preliminary predictor neural network model; The power load forecasting model and forecasting error database acquisition module is used to train and test the constructed medium- and long-term power load preliminary forecaster with the power load data set, thereby obtaining the power load forecasting model and the power load forecasting error database; The power load forecast error interval acquisition module is used to perform probability modeling on the obtained power load forecast error database to obtain the cumulative probability distribution of the power load forecast error, thereby further obtaining the power load forecast error interval; The power load result prediction module is used to input the power data of the standard time period into the power load prediction model to obtain the power load prediction result within the standard time period.

Images