Supply and demand prediction method and device for electricity price period, server and storage medium
By using the graph neural network model to combine meteorological data and actual supply and demand data, a supply and demand prediction method for electricity price period was established, which solved the problem of low accuracy of medium and long-term electricity price prediction, and improved the accuracy and practicality of prediction.
Patent Information
- Application Number
- CN202411412640.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-11
- Publication Date
- 2025-02-07
AI Technical Summary
In the prior art, the accuracy of medium- and long-term electricity price prediction is low, and the traditional methods ignore the impact of meteorological conditions on the electric sound field and demand, limiting the accuracy and practicality of the prediction.
By obtaining meteorological data from multiple sources, long, medium and short meteorological areas, establishing topological relationship diagrams and weights, inputting graph neural network models, predicting meteorological data, and establishing a prediction model based on actual supply and demand data to predict supply and demand during electricity price periods.
It improves the accuracy of medium- and long-term electricity price prediction, reflects the correlation between short-term weather and location, enhances the ability to extract characteristics changes in long-term weather impacts, and meets the adjustment of electricity consumption needs.
Smart Images
Figure CN119398818A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of supply and demand forecasting, and in particular to a method, device, server and storage medium for forecasting supply and demand during electricity price periods. Background Art
[0002] As an important price mechanism, medium- and long-term electricity prices are frequently used in the process of medium- and long-term market construction. By accurately predicting medium- and long-term electricity prices and assisting the strategy personnel on the power generation side and the power sales side to reasonably sign medium- and long-term contracts, power generation and power sales companies can reasonably arbitrage in medium- and long-term power transactions and increase trading revenues. This not only promotes the consumption of new energy, but also makes the market economy of electricity prices reasonably transmitted, and guides users to use more electricity in the valley section and less electricity in the peak section. It is of certain importance to shaving peaks and filling valleys, and minimizes the cost of the whole society. Accurate medium- and long-term electricity price forecasts are of great significance for power market participants to formulate strategies, optimize resource allocation and manage risks. Existing electricity price forecasting methods are mostly focused on short-term forecasts, and there are relatively few studies on the forecast of medium- and long-term electricity price trends, and the accuracy and practicality need to be improved. In the experience of power trading in many provinces, the accuracy of medium- and long-term electricity price forecasts is very important for assisting medium- and long-term trading strategies, and to a certain extent, it is more important than spot electricity price forecasts.
[0003] The accuracy of medium- and long-term electricity price forecasts is affected by many factors and is extremely difficult to predict accurately. Among the factors affecting medium- and long-term electricity prices, the supply and demand relationship is an important factor. Traditional supply and demand relationship forecasting methods are mainly based on historical electricity price data and simple statistical models. These methods often ignore the significant impact of meteorological conditions on the power field and demand, limiting the accuracy and practicality of the forecast. Summary of the invention
[0004] The embodiments of the present invention provide a method, device, server and storage medium for predicting supply and demand of electricity price periods, so as to solve the technical problem of system security risks caused by the inability to flexibly and dynamically set system authority settings in the prior art.
[0005] In a first aspect, an embodiment of the present invention provides a method for predicting supply and demand during an electricity price period, comprising:
[0006] Obtain multi-source long, medium and short meteorological regional meteorological data for each preset point;
[0007] Establish an initial topological relationship diagram based on the positional relationship between each point;
[0008] Establishing initial weights between points according to the positional relationship between the points to form a training topological relationship diagram; setting scalar weights according to the relevance of the meteorological data of each point;
[0009] Input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, wherein the meteorological data includes: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium-term frequency of 6 hours and a frequency of 3 hours;
[0010] According to the scalar weight, the degree of interaction between the features extracted from each point is set;
[0011] Receiving weather forecast results for multiple points output by the trained graph neural network model;
[0012] The actual supply and demand data released by the provincial dispatch and the meteorological forecast results for multiple points will be used. With the actual supply and demand data released by the provincial dispatch as the sub-items, a forecasting model will be established to predict the supply and demand for the period where electricity prices need to be predicted.
[0013] In a second aspect, an embodiment of the present invention further provides a supply and demand forecasting device for an electricity price period, comprising:
[0014] An acquisition module is used to obtain multi-source long, medium and short meteorological regional meteorological data for each preset point;
[0015] Establishing a module for establishing an initial topological relationship diagram according to the positional relationship between each point;
[0016] An initial weight establishment module is used to establish initial weights between points according to the positional relationship between the points to form a training topological relationship diagram;
[0017] A scalar weight setting module, used to set a scalar weight according to the relevance of the meteorological data of each point;
[0018] An input module is used to input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, wherein the meteorological data includes: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium-term frequency of 6 hours and a frequency of 3 hours;
[0019] A degree setting module, used to set the degree of interaction between the features extracted from each point according to the scalar weight;
[0020] A weather forecast module, used to receive weather forecast results of multiple points output by the trained graph neural network model;
[0021] The supply and demand forecasting module is used to build a forecasting model based on the actual supply and demand data released by the provincial dispatching agency and the meteorological forecast results of multiple locations, and to predict the supply and demand during the period when electricity prices need to be predicted.
[0022] In a third aspect, an embodiment of the present invention further provides a server, the server comprising:
[0023] one or more processors;
[0024] a storage device for storing one or more programs,
[0025] When the one or more programs are executed by the one or more processors, the one or more processors implement the supply and demand forecasting method for the electricity price period as described in any one of claims 1-7.
[0026] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the method for supply and demand forecasting for electricity price periods as provided in the above embodiments.
[0027] The supply and demand forecasting method, device, server and storage medium for electricity price time periods provided by the embodiments of the present invention obtain multi-source long, medium and short meteorological regional meteorological data for each preset point; establish an initial topological relationship diagram for the positional relationship between each point; establish initial weights between the points according to the positional relationship between the points to form a training topological relationship diagram; set scalar weights according to the degree of relevance of the meteorological data of each point; input the training topological relationship diagram and the meteorological data corresponding to each point into a trained graph neural network model, the meteorological data including: short-term 15-minute frequency, medium- and long-term 6-hour, 3-hour frequency wind speed, wind direction, temperature, humidity, pressure and total irradiance; set the degree of interaction between the features extracted from each point according to the scalar weights; receive meteorological forecast results for multiple points output by the trained graph neural network model; use the actual supply and demand data released by the provincial dispatching system and the meteorological forecast results for multiple points, and establish a forecasting model with the goal of sub-items of the actual supply and demand data released by the provincial dispatching system to forecast the supply and demand for the electricity price time period that needs to be predicted. By using the relationship between each point to set up the corresponding topological relationship diagram, and using the topological relationship diagram to input the graph neural network model, and setting weights based on the position relationship, the correlation between short-term weather and location is reflected, and the features are interacted to varying degrees according to the correlation between long-term meteorology, so as to better extract the characteristic changes of long-term weather affecting different locations, so as to further improve the accuracy of short-term and medium-term weather forecasts. The forecast results are combined with the actual supply and demand data released by the provincial dispatch, and the forecast model is used to predict supply and demand. According to the predicted supply and demand, the power consumption or the power supply of steam turbine units not affected by the weather can be adjusted to meet the power consumption requirements. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:
[0029] Figure 1It is a flow chart of a method for predicting supply and demand during electricity price periods provided in the first embodiment of the present invention;
[0030] Figure 2 It is a flow chart of a method for predicting supply and demand during electricity price periods provided in the second embodiment of the present invention;
[0031] Figure 3 It is a flow chart of a method for predicting supply and demand during electricity price periods provided in Embodiment 3 of the present invention;
[0032] Figure 4 It is a schematic diagram of the structure of the supply and demand forecasting device for the electricity price period provided by the fourth embodiment of the present invention;
[0033] Figure 5 It is a schematic diagram of the structure of the server provided in Embodiment 5 of the present invention. DETAILED DESCRIPTION
[0034] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.
[0035] Embodiment 1
[0036] Figure 1 : is a flow chart of a method for predicting supply and demand during an electricity price period provided in the first embodiment of the present invention. This embodiment is applicable to the case where accurate policy making is performed on the supply and demand situation during an electricity price period. The method can be executed by a device for predicting supply and demand during an electricity price period, and specifically includes the following steps:
[0037] Step 110, obtaining multi-source long, medium and short meteorological regional meteorological data for each preset point.
[0038] Weather forecast is an important indicator in supply and demand forecasting. Weather not only affects the amount of renewable energy power generation, but also affects changes in demand. Current traditional meteorological models and time series statistical forecasting models based on historical data as input lack dynamic adaptability. Meteorological conditions are a dynamically changing system, and existing models may have difficulty adapting to rapidly changing meteorological conditions, especially extreme meteorological events.
[0039] In addition, the existing meteorological models become relatively flat and non-volatile in the mid- to late-term, especially after more than 10 days. Stable and neutral weather forecast results often fail to reflect changes in extreme weather.
[0040] Therefore, in this embodiment, the weather forecast model needs to be optimized. At the same time, the corresponding short-term meteorological data, especially the meteorological data of the new energy site points, are added. Exemplarily, the multi-source long, medium and short meteorological regional meteorological data of each preset point can be obtained through historical records.
[0041] Step 120: Establish an initial topological relationship diagram based on the positional relationship between each point.
[0042] In order to cooperate with the weather forecast model, in this embodiment, an initial topological relationship diagram is first established for the positional relationship between each point. Considering that short-term weather usually affects areas of different sizes, an initial topological relationship diagram can be established based on the positional relationship.
[0043] Step 130, establishing initial weights between the points according to the positional relationship between the points to form a training topological relationship graph.
[0044] For example, initial weights may be formed according to positional relationships, especially height and horizontal distance, and a training topological relationship graph may be formed according to the initial weights.
[0045] Step 140: setting a scalar weight according to the relevance of the meteorological data at each point.
[0046] For example, different meteorological conditions affect different areas, so each point has different responses to the same meteorological condition. Therefore, the impact on different points can be determined based on historical meteorological conditions, and the scalar weight can be set using the impact degree.
[0047] Step 150, input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, the meteorological data including: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium- to long-term frequency of 6 hours and a frequency of 3 hours.
[0048] The short-term and medium-term meteorological data of each point and the topological map formed by each point are used to make predictions using the trained graph neural network model. In this embodiment, the graph neural network model is continuously trained according to the initial topological map to find different boundaries of the initial topological map, and then continuously divide the areas affected by different meteorological factors.
[0049] Step 160: setting the degree of interaction between the features extracted from each point according to the scalar weight.
[0050] In a graph neural network, each node can establish weighted connections with its neighboring nodes. These weights can be defined based on geographic distance, similarity, or other physical properties, so as to better simulate actual meteorological phenomena during information transmission. This weighted connection enables effective information exchange between remote nodes (grid points), i.e., "remote interaction". Using a multi-node message passing structure, nodes can exchange and interact with each other in multiple steps. This structure allows the model to build weighted connections between nodes, so that the information of remote nodes can interact and fuse through the message passing mechanism during the prediction process.
[0051] The transmission mechanism allows each node to receive information from its neighbor nodes, update its own state, and pass the updated information to other nodes. This process is repeated in multiple time steps, allowing grid points to complete long-range interactions within multiple steps. The advantage of this mechanism is that it can effectively integrate information from multiple sources and capture complex spatial correlations and temporal dynamics. It is particularly suitable for short-term weather forecasts. The degree of interaction can be guided by using scalar weights to better capture the temporal dynamics of the weather.
[0052] Step 170, receiving the meteorological forecast results of multiple points output by the trained graph neural network model, taking the actual supply and demand data released by the provincial dispatch and the meteorological forecast results of multiple points as the goal, establishing a prediction model with the actual supply and demand data released by the provincial dispatch as the sub-item, and predicting the supply and demand of the period where electricity price needs to be predicted.
[0053] The trained graph neural network model is used to obtain the meteorological forecast results of each output point. Combined with the actual supply and demand data released by the provincial dispatch, it can include: total wind power output, total photovoltaic output, actual external transmission / receiving, actual provincial load, hydropower output, thermal power output, and take the actual supply and demand data released by the provincial dispatch as the target, respectively, to establish a prediction model to predict the supply and demand of the period of electricity price forecast, including total wind power output, total photovoltaic output, actual external transmission, actual provincial load, hydropower output, thermal power output and other variables. For example, random forest model, ARIMA and other time series models can be used.
[0054] This embodiment obtains multi-source long, medium and short meteorological regional meteorological data for each preset point; establishes an initial topological relationship diagram for the positional relationship between each point; establishes initial weights between the points according to the positional relationship between the points to form a training topological relationship diagram; sets scalar weights according to the degree of relevance of the meteorological data of each point; inputs the training topological relationship diagram and the meteorological data corresponding to each point into a trained graph neural network model, the meteorological data including: short-term 15-minute frequency, medium- and long-term 6-hour, 3-hour frequency wind speed, wind direction, temperature, humidity, pressure and total irradiance; sets the degree of interaction between the features extracted from each point according to the scalar weights; receives meteorological forecast results for multiple points output by the trained graph neural network model; uses the actual supply and demand data released by the provincial dispatching and the meteorological forecast results for multiple points, and takes the actual supply and demand data released by the provincial dispatching as the sub-item to establish a forecasting model to predict the supply and demand for the period of electricity price forecasting. By using the relationship between each point to set up the corresponding topological relationship diagram, and using the topological relationship diagram to input the graph neural network model, and setting weights based on the position relationship, the correlation between short-term weather and location is reflected, and the features are interacted to varying degrees according to the correlation between long-term meteorology, so as to better extract the characteristic changes of long-term weather affecting different locations, so as to further improve the accuracy of short-term and medium-term weather forecasts. The forecast results are combined with the actual supply and demand data released by the provincial dispatch, and the forecast model is used to predict supply and demand. According to the predicted supply and demand, the power consumption or the power supply of thermal steam turbine units not affected by the weather can be adjusted to meet the power consumption requirements.
[0055] Embodiment 2
[0056] Figure 2 It is a flow chart of the method for forecasting supply and demand during electricity price periods provided by the second embodiment of the present invention. This embodiment is optimized based on the above embodiment, and the method may further add the following steps: check whether there are missing values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data; check whether there are abnormal values in the existing historical transaction data, regional forecast meteorological data, and provincial forecast supply and demand data; delete the missing values and abnormal values. The historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data after deleting the missing values and abnormal values are normalized. The normalized data and the predicted supply and demand data are input into the trained electricity price forecast GBDT model to obtain the medium- and long-term predicted electricity price.
[0057] See also Figure 2 , the method for predicting supply and demand during the electricity price period comprises:
[0058] Step 210, obtain multi-source long, medium and short meteorological regional meteorological data for each preset point; establish an initial topological relationship diagram based on the positional relationship between each point; establish initial weights between the points based on the positional relationship between the points to form a training topological relationship diagram.
[0059] Step 220, setting a scalar weight according to the relevance of the meteorological data of each point; inputting the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, and setting the degree of interaction between the features extracted from each point according to the scalar weight; receiving the meteorological forecast results of multiple points output by the trained graph neural network model; using the actual supply and demand data released by the provincial dispatch and the meteorological forecast results of multiple points, and taking the actual supply and demand data released by the provincial dispatch as the sub-item, establishing a prediction model to predict the supply and demand of the period where electricity price prediction is required.
[0060] Step 230, check whether there are missing values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data; check whether there are abnormal values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data; delete the missing values and abnormal values.
[0061] It is necessary to clean a large amount of historical transaction data, regional forecast meteorological data, and provincial forecast supply and demand data. The historical transaction data mainly uses the actual day-ahead electricity price as the target value. The forecast data includes daily forecast values for the province's load, new energy output, non-market unit output, provincial wind speed, and provincial irradiance for the next 45 days. The difference between actual supply and demand and forecast supply and demand is introduced as a feature in feature processing. On the forecast day, various spontaneous forecast supply and demand are used instead of actual supply and demand. All feature variables are one data per hour, 24 o'clock every day. Data cleaning includes checking whether the data is missing or has abnormal values, such as output forecasts of negative values, prices exceeding the upper and lower limits set by each region, etc. In such cases, the entire data is deleted.
[0062] Step 240, normalize the historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data after deleting missing values and outliers, and input the normalized data and forecast supply and demand data into the trained electricity price forecasting GBDT model to obtain the medium- and long-term forecast electricity price.
[0063] The above features need to be normalized before entering the GBDT model. For example, z-score normalization can be used. This normalization method is less affected by outlier data and does not depend on the boundary value of the data.
[0064] The z-score normalization data is normalized as follows:
[0065]
[0066] in x is the original data, u is the mean of the data, and σ is the variance of the data.
[0067] Substitute the historical data after data cleaning and normalization into the GBDT model and train it with the minimum electricity price as the goal. The GBDT model may include: neural network, support vector machine (SVM), random forest, etc. When predicting, the forecast day data, including the daily forecast values of the province's load, new energy output, non-market unit output, province's wind speed, province's irradiance, etc. for the next 45 days announced by the provincial dispatch and spontaneous forecasts, as well as new features of the new structure, are brought into the medium- and long-term electricity price forecast model to obtain the day-ahead electricity price on the forecast day.
[0068] This embodiment adds the following steps: check whether there are missing values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data; check whether there are abnormal values in the existing historical transaction data, regional forecast meteorological data, and provincial forecast supply and demand data; delete the missing values and abnormal values. The historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data after deleting the missing values and abnormal values are normalized. The normalized data and the predicted supply and demand data are input into the trained electricity price prediction GBDT model to obtain the medium- and long-term predicted electricity price. Various data can be pre-processed, and the medium- and long-term electricity prices can be accurately predicted in combination with the supply and demand forecast data provided in the above embodiments.
[0069] Embodiment 3
[0070] Figure 3 It is a flow chart of the supply and demand forecasting method for electricity price periods provided in Example 3 of the present invention. This embodiment is optimized based on the above embodiment, and the method can also add the following steps: obtaining extreme-related meteorological data from different sources, inputting the extreme weather forecasting large model, and obtaining output results; using the output results of the extreme weather forecasting large model as the input of the GBDT model to realize a revised forecast of medium- and long-term electricity prices.
[0071] See also Figure 3 The method for predicting supply and demand during the electricity price period includes:
[0072] Step 310, obtain multi-source long, medium and short meteorological regional meteorological data for each preset point; establish an initial topological relationship diagram based on the positional relationship between each point; establish initial weights between the points based on the positional relationship between the points to form a training topological relationship diagram.
[0073] Step 320, setting a scalar weight according to the relevance of the meteorological data of each point; inputting the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, and setting the degree of interaction between the features extracted from each point according to the scalar weight; receiving the meteorological forecast results of multiple points output by the trained graph neural network model; using the actual supply and demand data released by the provincial dispatch and the meteorological forecast results of multiple points, and taking the actual supply and demand data released by the provincial dispatch as the sub-item, establishing a prediction model to predict the supply and demand of the period where electricity price prediction is required.
[0074] Step 330, check whether there are missing values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data; check whether there are abnormal values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data; delete the missing values and abnormal values.
[0075] Step 340, normalize the historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data after deleting missing values and outliers, and input the normalized data and forecast supply and demand data into the trained electricity price forecasting GBDT model to obtain the medium- and long-term forecast electricity price.
[0076] Step 350, extreme-related meteorological data from different sources are obtained and input into the extreme weather prediction model; the output results of the extreme weather prediction model are used as the input of the GBDT model to achieve a revised prediction of the medium- and long-term electricity price.
[0077] In this embodiment, the extreme weather predicted by the large model can be used as input to revise the medium- and long-term electricity prices. Specifically, when a typhoon is predicted and most wind turbines are not shut down, it is expected that there will be more industrial and commercial shutdowns and production stoppages. The medium- and long-term electricity prices for the typhoon period can be lowered and revised according to local historical conditions; when extreme weather periods such as continuous high temperature heat waves, snow and ice are predicted, the medium- and long-term electricity prices for the corresponding period can be raised and revised according to local historical conditions.
[0078] Exemplarily, it may include: generating a corresponding historical data sequence based on the historical data of the electricity market during each extreme weather period, and normalizing the historical data sequence; calculating the Euclidean distance between the data sequence corresponding to the future medium- and long-term forecast electricity price and the normalized historical data sequence; sorting the historical data sequence according to the Euclidean distance, and taking the average electricity price in the historical data sequence as the revised forecast electricity price; comparing the revised forecast electricity price with the medium- and long-term forecast electricity price, and when they are consistent with the direction of artificial experience, taking the average of the revised forecast electricity price and the medium- and long-term forecast electricity price as the final electricity price forecast value.
[0079] A total of m sample data of the electricity market history data of similar extreme weather periods in the region in the past three years can be collected, including historical day-ahead electricity prices, historical spontaneous forecast supply and demand, provincial forecast supply and demand, and the difference between actual supply and demand and provincial forecast supply and demand (the actual supply and demand on the future forecast day is replaced by the spontaneous forecast supply and demand). The supply and demand data include the province's load, new energy output, photovoltaic output, wind power output, non-market unit output, province-wide wind speed, province-wide irradiance, etc. The day-ahead electricity price is recorded as y, and the other independent variables are recorded as x1, ....x n , and the medium- and long-term meteorological data produced by the large meteorological model, the impact on supply and demand can be brought into the model of spontaneous supply and demand forecasting to produce the independent variable value x at each time point in the future period affected by extreme weather 1t , ....x nt .
[0080] All independent variables are processed using the z-score normalization method to obtain the normalized variables w1, ....w n , and the independent variable value w at each time point in the future period affected by extreme weather 1t , ....w nt , we calculate the Euclidean distance d1, ....d of each point in the electricity market historical data sample at the future time point t relative to similar extreme weather periods m , sort the distances from small to large, and get sd1, ....sd m .
[0081] Set a threshold k and take the y value of the top k historical samples in the sd sequence (y1, ....y k ) is averaged to obtain the revised forecast electricity price at time t in the future affected by extreme weather. This revised electricity price is compared with the original medium- and long-term forecast electricity price. If it is consistent with the direction of manual experience, the average of the revised forecast electricity price and the original medium- and long-term forecast electricity price is taken as the final electricity price forecast value. If it is inconsistent with the direction of manual experience, the point is not revised. According to experience, manual experience is accurate in judging the direction, and the original medium- and long-term forecast electricity price can be used as a benchmark value, which has better robustness and accuracy after being averaged with the revised forecast electricity price.
[0082] Exemplarily, the above method is used to implement:
[0083]
[0084] When there are various extreme weather periods in the future, the above process will be repeated to revise the medium- and long-term electricity prices at all points in the future extreme weather periods.
[0085] This embodiment adds the following steps: obtain extreme-related meteorological data from different sources, input them into the extreme weather prediction model, and obtain output results; use the output results of the extreme weather prediction model as the input of the GBDT model to achieve a revised prediction of medium- and long-term electricity prices. By introducing the extreme weather forecast model, the accuracy of the electricity price forecast during extreme weather periods is further enhanced.
[0086] Embodiment 4
[0087] Figure 4 is a schematic diagram of the structure of the supply and demand forecasting device for the electricity price period provided by the fourth embodiment of the present invention, see Figure 4 , the supply and demand forecasting device for the electricity price period comprises:
[0088] An acquisition module 410 is used to acquire multi-source long, medium and short meteorological regional meteorological data for each preset point;
[0089] Establishing module 420, used to establish an initial topological relationship diagram according to the position relationship between each point;
[0090] An initial weight establishment module 430 is used to establish initial weights between points according to the positional relationship between the points to form a training topological relationship diagram;
[0091] A scalar weight setting module 440, configured to set a scalar weight according to the relevance of the meteorological data of each point;
[0092] An input module 450 is used to input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, wherein the meteorological data includes: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium-term frequency of 6 hours and a frequency of 3 hours;
[0093] A degree setting module 460, for setting the degree of interaction between the features extracted from each point according to the scalar weight;
[0094] A weather forecast result prediction module 470 is used to receive weather forecast results of multiple points output by the trained graph neural network model;
[0095] The supply and demand forecasting module 480 is used to establish a forecasting model based on the actual supply and demand data released by the provincial dispatch and the meteorological forecast results of multiple points, and to forecast the supply and demand in the period where the electricity price needs to be predicted, with the actual supply and demand data released by the provincial dispatch as the target.
[0096] The supply and demand forecasting device for electricity price time periods provided in this embodiment obtains multi-source long, medium and short meteorological regional meteorological data for each preset point; establishes an initial topological relationship diagram for the positional relationship between each point; establishes initial weights between the points according to the positional relationship between the points to form a training topological relationship diagram; sets scalar weights according to the degree of relevance of the meteorological data of each point; inputs the training topological relationship diagram and the meteorological data corresponding to each point into a trained graph neural network model, the meteorological data including: short-term 15-minute frequency, medium- and long-term 6-hour, 3-hour frequency wind speed, wind direction, temperature, humidity, pressure and total irradiance; sets the degree of interaction between the features extracted from each point according to the scalar weights; receives meteorological forecast results for multiple points output by the trained graph neural network model; uses the actual supply and demand data released by the provincial dispatching system and the meteorological forecast results for multiple points, and establishes a forecasting model with the goal of sub-items of the actual supply and demand data released by the provincial dispatching system to forecast the supply and demand for the electricity price period that needs to be predicted. By using the relationship between each point to set up the corresponding topological relationship diagram, and using the topological relationship diagram to input the graph neural network model, and setting weights based on the position relationship, the correlation between short-term weather and location is reflected, and the features are interacted to varying degrees according to the correlation between long-term meteorology, so as to better extract the characteristic changes of long-term weather affecting different locations, so as to further improve the accuracy of short-term and medium-term weather forecasts. The forecast results are combined with the actual supply and demand data released by the provincial dispatch, and the forecast model is used to predict supply and demand. According to the predicted supply and demand, the power consumption or the power supply of steam turbine units not affected by the weather can be adjusted to meet the power consumption requirements.
[0097] Based on the above embodiments, the provincial regulation publishes actual supply and demand data, including:
[0098] Total wind power output, total photovoltaic power output, actual power transmission / reception, actual provincial load, hydropower output and thermal power output.
[0099] Based on the above embodiments, the device further includes:
[0100] The first checking module is used to check whether there are missing values in the existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data;
[0101] The second inspection module is used to check whether there are abnormal values in the existing historical transaction data, regional forecast meteorological data, and provincial forecast supply and demand data;
[0102] The deletion module is used to delete the missing values and abnormal values.
[0103] Based on the above embodiments, the device further includes:
[0104] The normalization processing module is used to normalize the historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data after deleting missing values and outliers.
[0105] Based on the above embodiments, the device further includes:
[0106] The electricity price prediction module is used to input the normalized data and the predicted supply and demand data into the trained electricity price prediction GBDT model to obtain the medium- and long-term predicted electricity price.
[0107] Based on the above embodiments, the device further includes:
[0108] The output result acquisition module is used to obtain extreme-related meteorological data from different sources, input them into the extreme weather prediction model, and obtain output results;
[0109] The revised prediction module is used to use the output results of the extreme weather prediction model as the input of the GBDT model to realize revised prediction of medium- and long-term electricity prices.
[0110] Based on the above embodiments, the output result obtaining module includes:
[0111] A sequence generation unit, used to generate a corresponding historical data sequence based on each piece of historical data of the power market during an extreme weather period, and to perform normalization processing on the historical data sequence;
[0112] A calculation unit, used to calculate the Euclidean distance between the data sequence corresponding to the future medium- and long-term predicted electricity price and the normalized historical data sequence;
[0113] A sorting unit, used for sorting the historical data sequence according to the Euclidean distance, and taking the average electricity price in the historical data sequence as the revised predicted electricity price;
[0114] The comparison unit is used to compare the revised predicted electricity price with the medium- and long-term predicted electricity price, and when they are consistent with the direction of artificial experience, take the average value of the revised predicted electricity price and the medium- and long-term predicted electricity price as the final electricity price prediction value.
[0115] The supply and demand forecasting device for electricity price time periods provided in the embodiments of the present invention can execute the supply and demand forecasting method for electricity price time periods provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0116] Embodiment 5
[0117] Figure 5 A schematic diagram of the structure of a server provided in Embodiment 5 of the present invention. Figure 5 A block diagram of an exemplary server 12 suitable for use in implementing embodiments of the present invention is shown. Figure 5 The server 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.
[0118] like Figure 5 As shown, the server 12 is in the form of a general purpose computing device. The components of the server 12 may include, but are not limited to: one or more processors or processing units 16, server memory 28, and a bus 18 connecting different server components (including server memory 28 and processing unit 16).
[0119] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.
[0120] The server 12 typically includes a variety of computer server readable media. These media can be any available media that can be accessed by the server 12, including volatile and non-volatile media, removable and non-removable media.
[0121] Server memory 28 may include computer server readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. Server 12 may further include other removable / non-removable, volatile / non-volatile computer server storage media. By way of example only, storage server 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 5 not shown, usually called a "hard drive"). Although Figure 5 Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The server memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.
[0122] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the server memory 28, such program modules 42 including, but not limited to, operating a server, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.
[0123] The server 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the server 12, and / or may communicate with any device that enables the server 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. In addition, the server 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the server 12 via a bus 18. It should be understood that, although not shown in the figure, the server 12 may be assembled to use other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID servers, tape drives, and data backup storage servers, etc.
[0124] The processing unit 16 executes various functional applications and data processing by running the programs stored in the server memory 28, such as implementing the supply and demand forecasting method for the electricity price period provided in the embodiment of the present invention.
[0125] Embodiment 6
[0126] Embodiment 6 of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the supply and demand forecasting method for the electricity price period as described in any one of the above embodiments.
[0127] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.
[0128] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0129] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
[0130] Computer program code for performing the operations of the present invention may be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or device. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).
[0131] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims
1. A method for predicting supply and demand during electricity price periods, characterized in that: include: Obtain multi-source long, medium and short meteorological regional meteorological data for each preset point; Establish an initial topological relationship diagram based on the positional relationship between each point; Establish initial weights between points according to the positional relationship between points to form a training topological relationship diagram; Setting a scalar weight according to the relevance of the meteorological data at each point; Input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, wherein the meteorological data includes: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium-term frequency of 6 hours and a frequency of 3 hours; According to the scalar weight, the degree of interaction between the features extracted from each point is set; Receiving weather forecast results for multiple points output by the trained graph neural network model; The actual supply and demand data released by the provincial dispatch and the meteorological forecast results for multiple points will be used. With the actual supply and demand data released by the provincial dispatch as the sub-items, a forecasting model will be established to predict the supply and demand for the period where electricity prices need to be predicted.
2. The method according to claim 1, characterized in that The provincial survey releases actual supply and demand data, including: Total wind power output, total photovoltaic power output, actual power transmission / reception, actual provincial load, hydropower output and thermal power output.
3. The method according to claim 1, characterized in that The method further comprises: Check whether there are missing values in existing historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data; Check whether there are abnormal values in the existing historical transaction data, regional forecast weather data, and provincial forecast supply and demand data; The missing values and outliers are removed.
4. The method according to claim 1, characterized in that: The method further comprises: The historical transaction data, regional forecast meteorological data, provincial forecast supply and demand data, and historical electricity price data are normalized after removing missing values and outliers.
5. The method according to claim 4, characterized in that The method further comprises: The normalized data and predicted supply and demand data are input into the trained electricity price prediction GBDT model to obtain the medium- and long-term predicted electricity price.
6. The method according to claim 5, characterized in that The method further comprises: Obtain extreme-related meteorological data from different sources, input into the extreme weather prediction model, and obtain output results; The output results of the extreme weather prediction model are used as the input of the GBDT model to make revised predictions on medium- and long-term electricity prices.
7. The method according to claim 1, characterized in that The extreme weather related meteorological data obtained from different sources are input into the extreme weather prediction model to obtain output results, including: Generate a corresponding historical data sequence based on the historical data of the power market during each extreme weather period, and normalize the historical data sequence; Calculate the Euclidean distance between the data series corresponding to the future medium- and long-term forecast electricity price and the normalized historical data series; Sort the historical data sequence according to the Euclidean distance, and use the average electricity price in the historical data sequence as the revised predicted electricity price; The revised predicted electricity price is compared with the medium- and long-term predicted electricity price. When they are consistent with the direction of manual experience, the average value of the revised predicted electricity price and the medium- and long-term predicted electricity price is used as the final electricity price prediction value.
8. A supply and demand forecasting device for electricity price period, characterized in that: include: An acquisition module is used to obtain multi-source long, medium and short meteorological regional meteorological data for each preset point; Establishing a module for establishing an initial topological relationship diagram according to the positional relationship between each point; An initial weight establishment module is used to establish initial weights between points according to the positional relationship between the points to form a training topological relationship diagram; A scalar weight setting module, used to set a scalar weight according to the relevance of the meteorological data of each point; An input module is used to input the training topological relationship diagram and the meteorological data corresponding to each point into the trained graph neural network model, wherein the meteorological data includes: wind speed, wind direction, temperature, humidity, pressure and total irradiance with a short-term frequency of 15 minutes, a medium-term frequency of 6 hours and a frequency of 3 hours; A degree setting module, used to set the degree of interaction between the features extracted from each point according to the scalar weight; A weather forecast result prediction module, used to receive weather forecast results of multiple points output by the trained graph neural network model; The supply and demand forecasting module is used to build a forecasting model based on the actual supply and demand data released by the provincial dispatching agency and the meteorological forecast results of multiple locations, and to predict the supply and demand during the period when electricity prices need to be predicted.
9. A server, characterized in that: The server comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the supply and demand forecasting method for the electricity price period as described in any one of claims 1-7.
10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the supply and demand forecasting method for the electricity price period as described in any one of claims 1-8 when executed by a computer processor.